The Simulated Self-Fiction Reading and Narrative Identity

How do participating in a work of fiction and imagining a fictional world intertwine with the reader’s life? I develop an account that explores the relation between fiction reading and the reader’s narrative identity. Starting with an investigation of Paul Ricoeur’s account of narrative identity and of Kendall Walton’s account of the nature of representations, I develop my own model of fiction reading. My account is based on two starting assumptions: first, that human beings are entangled in stories, and second, that emotions are complex and have a narrative structure. I argue that during the reading process, the fiction reader creates her own narratives which contain not only the story provided by the work of fiction, but also event sequences from their own experiential memories. I investigate the creation and the influence of self-conscious emotions which are generated during the reading process, especially when a reader identifies with a fictional character from within. I consider how and why these fiction-induced emotions can continue after the reading is finished, and how these emotions can motivate the reader to engage in self-reflection and to refigure her self-narrative. My account examines a new topic: the interactive influence of fiction reading and the fiction reader’s narrative identity.

Introduction What does it mean to read, and to engage with, a work of fiction? The reading experience often makes us readers feel absorbed in the fictional events, as if we were transported into a world apart from our daily lives. The deepest absorption occurs when we manage to empathize with fictional characters. Many readers value fiction reading in particular because it offers them the possibility of escaping from their usual troubles and imagining a different life, experienced through a fictional person in a story. In fiction, there are no constraints as to what can happen in the story, with the exception of those imposed by the work of fiction itself. We can slay a dragon to save a princess, gain intimate knowledge of the prostitutes of Montparnasse, or ride to the South Seas along an iron equator belt on a horse on roller-skates. We do not need to explain or justify what we imagine when we are engaged with a work of fiction. Compared to the risks we run in real life, the risks of reading fiction lie in different areas, and they appear to be less pressing. Whereas real life has to be continually negotiated with the world and society, imagining fiction needs only to be negotiated with our self, our hopes and fears and our moral values, and we can do that at our leisure. Reading can be uplifting and pleasant, or disturbing, or emotion-provoking. Reading can arouse strong feelings, especially when we feel for a fictional character from within. We rave against repression together with Jane Eyre, bursting with outrage against her cousin who plays her nasty tricks and gets away with it. We feel the dagger in our heart when Scarlett O'Hara finally realizes that she loves Rhett Butler, and is rejected by him. We feel the desperation of not being able to clear ourselves from unjust suspicion with Oliver Twist who realises that his benefactor must take him to be a thief.
What exactly happens to us when are deeply engaged with a work of fiction, and how does our reading experience relate to our real life, in the real world? We would not be able to experience emotions in the fictional world unless there was some connection between the two worlds, between what we perceive and feel in the fictional world and what we have experienced in the real world in the past, what made us who we are. This connection can only be brought about by us, by the reader herself. Our lived self is engaged with a work of fictioneven if we leave our daily concerns outside, there is something of our self which we bring into the reading process. What is the relation between our own life story and the fictional story we enter? This question has not yet been fully explored, and in my account of fiction reading, I use a new approach to gain deeper insights into this issue.
My account is based on the assumption that, as Wilhelm Schapp puts it, we human beings are always entangled in stories. 1 Stories put us into relationships with others, and these relationships can be the cause of emotions. This does not only happen during our daily lives, with real people, but also during fiction reading, with fictional people. The fiction reader tends to create different kinds of narrative which are more or less loosely based on the work of fiction. In these narratives, both the reader herself and the fictional character are somehow entangled in the story. I am particularly interested in the nature of these entanglements: how they emerge, who is in the lead, how they are driven, and what consequences they have for the reader. Hence, narratives play an important part in my inquiry.
I argue that, typically, the fiction reader does not get entangled only in one narrative, but in several narratives. In each of these narratives, there is a different kind of relationship between the reader and the fictional character she is interested in. In my model of fiction reading, I categorize three different kinds of narrative that the reader develops during the reading process. For each of these kinds of narrative, I explore how emotions can be evoked, how these emotions can develop, and how they can influence the reader's self and her self-narrative.
There have already been endeavours to explore the connections between fiction reading and the reader's self, 2 but they have not used the concept of narrative to illuminate the degree and the possible development of the reader's involvement with a work of fiction. My contribution to already existing accounts of fiction reading is built upon understanding the connection between the reader's narrative identity and the narratives which emerge during the reading process.
As a basis for my investigations, I bring together the works of two very different writers on narrative: Paul Ricoeur and Kendall Walton. I create my own model of fiction reading, drawing on, but also extending, both of these accounts. Paul Ricoeur's account of personal identity differs from other accounts by introducing a hermeneutic approach to examining a person's self. It is part of Ricoeur's account of narrative identity that we engage with fiction to develop our identities, and it is part of Walton's account that our engaging with fiction has an impact on our lives. As well as showing how each of these accounts can help to fill a significant gap in the other, I shall argue that important lacunae remain with respect to the influence of fiction reading on the reader's self. I fill the lacunae in their accounts with my own model of fiction reading.
Part I of this thesis provides an introduction to and a discussion of Ricoeur's account of narrative identity. Contemporary philosophy provides theories that attempt to explain how personal identity is constituted, and how it persists over time. Among these models, accounts of narrative identity have been gaining importance in recent discussion. 3 In his model of narrative identity, Ricoeur holds that a person represents her self by a narrative that she weaves, and continuously refigures, across the events of her life. In this process, a person has two roles: she is the author of her own life in as far as she integrates the contingencies of the past into her on-going story, and she is also the protagonist in her own narrative. Ricoeur says that fiction reading plays an important part in the refiguration process of our self-narrative. By identifying with characters in fictional stories, we test and evaluate various ways of interacting with the world in our imagination and apply some of them to our self-narrative. Thus, reading fiction helps to shape our narrative identity.
Regarding the reading process as such, however, Ricoeur is content to describe the reader as engaging in thought experiments which canafter reflectioninfluence the reader's self-narrative. He does not investigate how the reader can develop real-life emotions during the reading process, or how reading-induced emotions can be instrumental in leading the reader towards self-reflection. These are significant gaps in his model, and filling them will give a fuller insight into the process of self-reflection and the development of self-narrative.
Part II gives an overview of Walton's account of engaging with a work of fiction. In aesthetics, Walton cast a new light on the role of the reader and her emotions by developing a theory about the nature of representations. He interprets the reader's engagement with a work of fiction as participating in a game of make-believe. Hence, Walton's account has the potential to help to fill the gap that I have identified in Ricoeur's account. However, Walton's account is also incomplete in two respects. According to Walton, the reader accepts the rules of the fictional world prescribed by the author, and makes-believe that the characters in the novel exist, and that he feels for them 'from the inside'. Walton himself lists among the things that are still left to do to complement his model 'the specific ways in which such participatory experiences contribute to our lives'. 4 Ricoeur's account of narrative identity can contribute to fill this gap in Walton's theory.
The additional gap I see in Walton's account is that when he is considering the affective impact of fictions, he focusses exclusively on episodic experiences rather than long-term changes. For Walton, the reader's engagement with a fictional work leads to physiological-psychological states that can be identified as the affective part of an emotion. Walton does not go beyond the immediate emotional effects of fiction-reading.
I argue that the reading experience, in connection with accompanying imaginings, often triggers a chain of self-conscious emotions in the reader, even if the starting point was only a fiction-induced affective state. In my view, this is an important link between the reading experience and the reader's subsequent self-reflection. In order to account for this, I will develop a new account of fiction-reading.
In Part III, the main body of my work, I lay out my own account and show how it fills the gaps in both Ricoeur's and Walton's account of the reader's role in fiction reading. I do this by examining four questions: 1) What happens to the reader when she imagines a character(f) 5 from within?
2) Which kinds of narrative are created or influenced during fiction reading?
3) How can emotions be evoked during fiction reading? 4) If fiction reading leads to self-examination, does this also direct the reader towards a more realistic self-narrative? Current accounts of fiction reading focus primarily on the question of whether fiction reading has an influence on the reader's future actions. The attention paid to this issue is important from an ethical and educational point of view, but it tends to neglect the intricate fabric of emotions that is triggered within the reader whilst they are reading.
This web of emotions is not measurable from the outside, by observation of the reader's actions, but only from inside, from the reader's self and the story which weaves her life together with the world she lives in. The outcome of my investigations will be a deeper understanding of the effects of fiction reading on a person's narrative identity, and thus, on her self and her self-unity.

A Plan of the Thesis
The thesis is structured in the following way: Part I focusses on Ricoeur's model of narrative identity. In chapter 1, I give an overview of his account. An important feature of Ricoeur's account is the definition of two aspects of personal identity which are not reducible to each other: the object (idem) and the lived self (ipse). In chapter 2, I compare Ricoeur's model to other accounts of personal identity (psychological continuity accounts, bodily continuity accounts, other narrative identity accounts). This chapter includes a critical evaluation of the accounts of Derek Parfit, Eric Olson, Marya Schechtman, Kim Atkins, Christine Korsgaard, Peter Goldie, and Galen Strawson. I explain why Ricoeur's model gives a richer account of personal identity than these other models.
In Part II, I present Kendall Walton's account of fiction reading. In chapter 3, I give an overview of his model of fiction reading as a game of make-believe, focussing on the topic of reader participation in fiction. I also discuss critical objections to this account by Noel Carroll and John Gibson. I show that Walton's account of the reader's playing a game is compatible with Ricoeur's account of narrative identity. I also identify a gap in Walton's account, which I am going to fill by the account I offer in Part III.
The main body of my own account is contained in Part III.
In chapter 4, I give a high-level outline of my complete model of fiction reading in connection with narrative identity.
As the investigation of emotions is important for my account, I outline the notion of emotions that I adopt in chapter 5. I assume that emotions are intentional and structured, and that they develop in a way that takes narrative form.
In chapter 6, I investigate the reader's involvement in a work of fiction during her actual reading experience. I distinguish between three levels of participation in fiction reading, and I investigate what happens to the reader when she identifies with a character(f) from the inside. I define three types of narrative the reader can be involved in while engaging with a work of fiction: the other-narrative, the self-in-other-narrative, and the other-in-self-narrative. This is an important step for the development of my account: in the following chapter I claim that these three types of narrative differ regarding the chain of emotions they can invoke in the reader.
In chapter 7, I pick up the notion of the three types of narrative which the reader can weave during fiction reading, and I argue that each type of narrative can trigger a complex structure of emotions. However, there are typical emotion chains for each of these narrative types. I say that self-conscious emotions play an important part in this, and I investigate the nature of reading-induced self-reflection. I argue that self-reflection is usually caused by self-conscious emotions, and that the nature of these emotions governs the depth and the direction of self-reflection.
In chapter 8, I compare my account of fiction reading with Ricoeur's account. I argue that although Ricoeur does think that the reader's emotions form an important factor in the reading process, he stops exploring them too early. Ricoeur says that catharsis frees the reader from emotions, thus enabling her to learn from fiction. I challenge this view, pointing out that the emotions generated by fiction reading are too complex to be represented in this simple way. In addition to this, I argue that Ricoeur's account of self-reflection is too optimistic. In my view, self-reflection does not always, as Ricoeur claims, lead to a more realistic self-narrative and an improved self-unity.
In chapter 9, I compare my account of fiction reading with Walton's account. I argue that Walton does not pay enough attention to the imaginings the reader has alongside the imaginings prescribed by the text. These subsidiary imaginings link the reader's experiential memories and their own mini-stories to the fictional story. I argue that the subsidiary imaginings play a large role in the development of the self-in-otherand the other-in-self-narrative, and that they often spawn chains of self-conscious emotions. Walton does not investigate how the emotions which are generated during fiction reading persist and develop within or beyond the reading process. Hence, he misses the opportunity to explore reading-induced self-reflection and its possible consequences for the reader's self-narrative and self-unity.
In the summary of Part III, I recapitulate my account of the influence of fiction reading on the reader's narrative identity. As a main point, I stress the importance of investigating the different kinds of narrative the reader weaves around the fictional text.
These narratives can be sources of self-conscious emotions which often persevere after the reading process, urging the reader to reflect on her self-narrative, and sometimes to refigure it.

PART I: NARRATIVE IDENTITY 1 Ricoeur's account of narrative identity
In this chapter, I am going to outline Paul Ricoeur's account of narrative identity as he presents it. His work follows the tradition of so-called Continental Philosophy, which takes a different approach to questions of personal identity than the Lockean approach standardly taken within what is commonly understood as Analytic Philosophy. Ricoeur searches for the 'hermeneutics of self', 6 combining the tradition of Husserlian phenomenology with philosophical hermeneutics. 7 In his main work on this topic, The Self as Another (1992), Ricoeur is primarily concerned with the central dilemma from which human beings suffer. As we are embodied, and experience the world around us through our body, we are subject to the restraints to which the physical world submits us. But we also see ourselves as agents and feel responsible for our actions. We are both objects and agents in our lives. This tension is often illustrated by the example of a person grasping her own hand. She experiences touching her hand as if it was an object, and at the same time she has the subjective feeling of being touched.
I perceive my own body as both a subject and an object. Because of this ambiguity, my self cannot be completely reduced to a physical object, and my perception of myself is active as well as passive. 8 The irreducibility of these two aspects of personal identity plays a pivotal role in Ricoeur's account of narrative identity.
In sections 1.1 to 1.6 of this chapter, I am going to show how Ricoeur derives his two-level view of personal identity (objective and subjective) from factors that influence the human situation: perception, embodiment, action, and responsibility. As a next step, in sections 1.7 and 1.8, I am going to outline Ricoeur's view of mimesis as story-telling, and of the links between fiction and self-narrative. Finally, in sections 1.9 and 1.10, I shall set out where Ricoeur sees connections between narrative unity and the unity of one's own life. 6 Ricoeur 1992, 4. 7 I shall explain this in detail in the next section. 8 Atkins 2005, chapter 4.

Ricoeur and Hermeneutics
As a first step towards presenting Ricoeur's account of narrative identity, I shall give a brief summary of the philosophical framework on which he builds his model. He embeds his theory in two interconnected fields of research: phenomenology and hermeneutics. Being embodied in the world, he says, gives us sensations which provide an inseparable link between body and mind. Every sensation which we experience has a horizon of its own, extending over space and time. The structure of such an experience can be read and interpreted using the tools of hermeneutics. As Ricoeur's model of narrative identity hinges on the link between the body and the experienced self, I shall outline the main points of phenomenological hermeneutics in this section.

Phenomenology
Phenomenology as a philosophical discipline investigates the structure of different kinds of experiences. It deals with what experiences 'feel like' in our acts of consciousness. For philosophers such as Husserl, Heidegger and Merleau-Ponty 9 , phenomenology does not restrict itself to investigating sensory experiences such as seeing, touching or hearing, but also includes the way in which we experience memory, emotion, volition, embodied action, social activity, and the self. The structure of these experiences requires intentionalitythe experience is always directed towards an object of our lived world. Husserl holds that this directedness is possible only through the mediation of particular concepts or images. 10 Husserl's method of reduction or epoché (bracketing) stipulates that all naturalistic assumptions about the world must be put 'out of play' so that the phenomenologist can attend to the phenomenological mode of presentation of objects without importing naive presuppositions about their attributes or their existence. 11 In the phenomenological mode of presentation, the focus is on my experience of the perceived object, and specifically the content or meaning of the object in my experience. The phenomenological presentation of objects depends on instruments that mediate between 9 Merleau-Ponty 2005 10 Smith 2011. 11 Husserl 1997, 91. into the future, so that I will be able to perceive them during that future time period as well if I choose, and I often expect that I may have different sensory experiences of them in the future.
All individual experiences have a subjective 'horizon' which reflects the relationship between the subject and the object. These experiences change with the subject-object relationship they are depending on and with the alteration of this relationship itself as it unfolds over time. They constitute what Husserl calls an 'intentional horizon' 16 of reference. The intentional horizon refers to the possibilities of consciousness within the subject-object relationship. This means that I do not only perceive the object as presented to me by my senses, but I am also aware of other aspects of it that my experience horizon of this object already includes as possible. If I have a perceptual experience of a flower in a meadow, for instance, I am immediately aware of its shape and colour, maybe even its smell. But I also anticipate that if I step forward and grasp the stem, I will feel its texture, and the reluctance of its roots in the earth when I try to pull it out. The flower, the object of my perception, already transcends the information currently observable by methe horizon of my flowerperception goes beyond my sensual observation.
The importance of the human body for phenomenological experiences, and the investigation of the different time horizons which can be distinguished in connection with personal identity, are the two key themes in Ricoeur's theory of narrative identity.

Idem and Ipse
Ricoeur's account of personal identity differs from other accounts of personal identity in one central and very important respect: he examines personal identity on two levels, using two different categories. On the metaphysical level, personal identity is defined by sameness (Latin: idem), and this concerns the person as a spatial-temporal object.
The idem provides the answer to the question 'What am I?'. The idem relates to all attributes of a person that tend to be permanent over a long-time horizon, including for instance the body, the genetic code, and the character(p). Character(p), as Ricoeur uses it in this context, defines the set of lasting dispositions by which a person can be 16 Beyer 2007, chapter 4. recognized. 17 Although character(p) is subject to (usually slow) changes over time, Ricoeur sees it as part of the idem because it concerns the objective aspect of personal identity. Persons can be identified from the outside by their character(p): we feel justified to make statements like: 'he is reliable' or 'she is unforgiving'.
On a phenomenological level, personal identity is represented by the self (Latin: ipse). The ipse is the experienced or lived identity of a person, the feature that distinguishes the 'I' from the 'other'. One of its most important aspects is that it is constitutive of self-constancy, the perseverance of faithfulness to a promise. A prerequisite for the ipse is the ability of a person to make plans and enter obligations towards herself and others for the future. The ipse covers all areas of a person's life where the personal pronoun 'mine' can be applied in the sense of 'what I feel responsible for'. It answers the question 'Who am I?'. 18 Ricoeur describes the difference between idem and ipse as the 'difference between a substantial or formal identification and a narrative identity '. 19 In his development of a hermeneutics of self, Ricoeur examines what it means to belong to one's own world and one's own body, as they are experienced. Ricoeur describes the involved processes in hermeneutic terms, exploring how the lived embodied self is affected by constant interaction with the world: by feeling, acting, suffering, reflecting, and taking responsibility for her own actions. In Ricoeurian terminology, the interaction of the self with the world can be described as 'a discourse of appropriation' that involves the acceptance of one's world and one's body as one's own. In contrast to this, the typical analytic approach is concerned with the objective world and the objective body in what Ricoeur calls a 'discourse of distance'. 20 Ricoeur claims that the analytic examination of questions concerning personal identity is not sufficient to grasp all relevant aspects of the self.
Ricoeur also makes a point of distinguishing his own account of selfhood from the existential proposition of the Cogito that states 'I think, therefore I am'. In particular, he claims that Descartes' approach of extreme doubt only provides a proof of the 'I' which is atemporal and must thus be reasserted from moment to moment. 21 He says that Descartes proves a purely formal 'I' as a subject which cannot answer the 20 question 'what am I?' as an object. According to Ricoeur, Descartes does not consider human embodiment as an aspect of personal identity, but as an unimportant contingency. Ricoeur says that if you view the body as nothing but an object, you lose important characteristics of consciousness. Consciousness is not contained within itself, but directed towards the outside world in two ways: through voluntary acts, willed by the self and carried out by the body, and through involuntary acts performed by the body. 22 Here, he adopts Sigmund Freud's notion of the unconscious mind, saying: 'The philosopher will not attribute thought to the unconscious, but he will admit the dependence of thought, which is consciousness itself, on a hidden psychic dynamism whose dramas often erupt at the very heart of consciousness and remove an entire region of intentions and actions from its control.' 23 Although Ricoeur investigated Freud's views on psychoanalysis in great depth, he does not take this to be a basis for his account of personal identity as narrative identity. Rather, the point that he takes from Freud here is the view of the subconscious as the seat of repressed emotions. 24 Ricoeur also examines the other extreme, the 'shattered Cogito' philosophy proclaimed by Nietzsche who says that truth can never be determined as Nature 'threw away the key'humans do not possess, and cannot develop, the tools necessary for unravelling Nature's secrets. 25 This can be roughly paraphrased as 'I may think that I think, but this does not establish any truth'. Ricoeur comes to the conclusion that this anti-Cogito does not, as it claims, prove Descartes' argument wrong but instead destroys the original question ('Can I know that I exist?') to which the Cogito claims to give an absolute answer. 26 If I can never know the truth about anything, asking whether I know that I exist is just a silly question. Nietzsche holds that there is no distinction between lies and truth, that there is only illusion (Nietzsche calls this 'Verstellung'), because Nature deprived man of the powers to decipher illusions. 27 For Ricoeur, this 22 Ricoeur 1966, 6. 23 Ricoeur 1966, 400. 24 Ricoeur's views on Freud are still subject to ongoing discussions in philosophy, see for instance Carney, Eoin, 'Technique and Understanding -Paul Ricoeur on Freud and the Analytic Experience' Études Ricoeuriennes / Ricoeur Studies, Vol 7, No 1 (2016), pp. 87-102. I shall not include these discussion in the scope of my work. 25 Ricoeur 1992, 11. 26 Ricoeur 1992, 14. 27 'Wahrheiten sind Illusionen, von denen man vergessen hat, daß sie welche sind, Metaphern, die abgenutzt und sinnlich kraftlos geworden sind' (‚Truths are illusions whose illusionary nature has been forgotten, metaphors which have become shopworn and sensually weak'), Friedrich Wilhelm Nietzsche: Über Wahrheit und Lüge im außermoralischen Sinne, 1873 -Kapitel 1, http://gutenberg.spiegel.de/buch/-3243/1, my own translation. claim is a paradox in itself: either Nietzsche's theory exempts itself from illusion, or it is itself a subject to it. Nietzsche claims to utter a truth, while saying that the truth can never be determined. Ricoeur describes this is as special case of the liar's paradox. 28 Thus, Ricoeur distances himself from both the Cogito and the anti-Cogito, but he does not intend to posit the self directly. Although he uses analytic methods in order to examine the validity of his intuitions, his hermeneutic research is different from the deductive style of analytic philosophy, as we shall see. He develops his account by investigating three areas: the linguistic use of the term 'I', the relation between personal identity and action, and the role of personal responsibility within the concept of personal identity.
To summarize this section: Ricoeur's distinction between idem (sameness) and ipse (the lived self) is a central part of his account. The following sections will show how these two aspects of personal identitywhich cannot be reduced to each otherdefine a person, and how they are able to characterize and clarify changes of personal identity over time.

The Self and Language
Reference to the self begins with language. Ricoeur starts his exploration of the self on a linguistic level by examining how we identify ourselves. In a descriptive speech utterance that uses 'I' (for instance 'I see') there is a logical gap between the referential function that picks out the 'I' as a public entity that can be identified by others, 29 and the reflexive function that indicates the 'I' as a private entity that the utterer is conscious of, while speaking to a 'you'. An utterance which has a purely referential function is, for example, 'I was born on 9 th May, 1953'. The utterance 'I feel embarrassed', on the other hand, is reflexive.
It is an important point for Ricoeur that the object the referential function points at, the 'I', is irreducible to any other kind of particular, such as, for instance, a brain, or a psychological history. The referential function indicates the 'I' as a 'basic particular'. Ricoeur follows Peter Strawson here who says that basic particulars belong to a distinguishable class of particulars that can be identified and re-identified without reference to particulars of other kinds. 30 In Ricoeur's words, 'Physical bodies and the persons we ourselves are constitute, in this masterful strategy, such basic particulars in the sense that nothing at all can be identified unless it ultimately refers to one or the other of these two kinds of particulars.' 31 When an utterer uses this referential function, she denotes herself as 'I'that is, the person who is speaking. The complete signification of the 'I' in the objectified view, as uttered by the person herself, is achieved by the person's capacity for self-designation.
Thus, when talking about herself, the utterer can use either (or both) of two viewpoints: the 'I' as the person whom the utterer refers to when talking about herself as an entity, and the 'I' of the utterer as she experiences herself.
Ricoeur's distinction between the referential and reflexive functions of 'I' acknowledges the puzzling fact that the 'I' plays two roles: the one that speaks and the one of whom the utterer is speaking, the objectified aspect of the 'I'. When an utterer puts herself into a relation to the person spoken to, she exchanges information with a 'you', and this information can have different kinds of intentional object. For instance, if someone hurts me, and I say to him: 'I hate you', I as the utterer give information about my psychological state to my tormentor as the person spoken to. Typical (though not all) utterances of this kind concern the utterer's feelings, like 'I admire you' or 'I share your grief'. The objectified 'I' fades into the background because the utterer is only viewing herself from within. I do not give the person spoken to any information about myself except my psychological states as they relate to him or her, and the fact that I have these psychological states.
Regarding this reflexive function of the 'I', Ricoeur distinguishes between constative and performative speech. 32 If the utterer makes a constative statement about herself, she describes a situation, an action, or a psychological state. In a performative statement about herself, however, the utterer also performs the action she describes.
Here, the utterance does not refer to anything but this performance. A promise is an example of performative speech: it is a binding contract between utterer and person addressed. When an 'I' promises something to a 'you' in an utterance, this speech is already the actual promise. 33 If I say to my neighbour: 'I promise I will look after your children during the summer holidays', I inform my neighbour of my promise, and 30 Strawson 1959, 38. 31 Ricoeur 1992, 31. 32 Here, Ricoeur refers to Austin 1962. 33 Ricoeur 1992, 42. simultaneously I take over the obligation to actually perform the baby-sitting for her. As shown in the previous section, entering obligations for the future is an important power of the self, the ipse. In performative speech, the language use links the ipse to her social nexus.
So far, Ricoeur's approach has shown that an utterer can refer to herself in two ways: in an objectified way, and in a reflexive way. She can also refer to a commitment she makes, through the utterance of a promise. But how are these three types of reference connected?
The referential and the reflexive approach to language theory can converge, Ricoeur claims, if the 'I' who is speaking to a 'you' and the 'him/her' of whom the utterer is speaking are assimilated in a 'unique type of objectification '. 34 This happens in utterances that contain both the referential and the reflexive viewpoint on the 'I' on the part of the utterer. For instance, if a father says to his son: 'Your father will always support you', he refers to himself objectively (speaking of himself, as seen from the outside, in his role as a father) and reflexively (speaking of his personal commitment).
Thus, the idem and the ipse work together in performative speech: By joint reference to the objective person and her subjective aims, they create a personal promise to the 'you' to whom it is uttered, and they constitute a personal obligation for the utterer.

The Self and Action
In the previous section, we saw how the objectified 'I' and the subjective 'I' can be represented in speech. Ricoeur's concept of action supplements his model of personal identity by showing the connection between the lived body that can be identified physically (the idem) and the sphere of mineness that I am conscious of (the ipse). 35 A person interacts with the world through her body, and her actions are visible from the outside, by others who attribute intentions and causes to them, as well as from the inside, by the person who plans and experiences them. But what exactly defines an action, what is its time horizon, and how is the agent involved in an action? 34 Ricoeur 1992, 52. 35 Harrison-Barbet 2008, 5. For Ricoeur, it is important to get a clear analytic understanding of the terms 'event' and 'action' to complement the intuitive interpretation obtained by phenomenological experience. Ricoeur differentiates between event and action by saying that an event isin generalsomething which happens (connected to the question 'what?'), whereas an action is something for which a person is responsible (connected to the questions 'who? ' and 'why?'). The question 'who?' is the most important point for him. He sets out his own account against Donald Davidson's theory of action to highlight why he thinks that this question is essential for an account of action. His comparison with Davidson's account is not, however, intended to cover all of Davidson's arguments. Ricoeur wants to drive home one particular point: the importance of personal responsibility, as seen from the inside of the agent, in contrast to an objective accountability which could be proved in court. 36 In Ricoeur's view, Davidson is interested in the wrong question, 'what?', while neglecting the question 'who?'. Although it is debatable whether this does justice to Davidson's theory,37 I am going to outline Ricoeur's arguments here as they give a valuable insight into his own account.
Davidson holds that human action can be explained by applying the same concept of causality that we use to explain physical processes. Ricoeur interprets Davidson's account as taking an exclusively external view. This is where Ricoeur pins his criticism: he says that it is not enough to get an objective view on an action to fully understand it: an action needs to be viewed from the perspective of the agent's self. In Ricoeur's view, Davidson neglects this aspect of human action. Davidson says that the agent's intention can be derived from outside observation; Ricoeur says that this is not the case, not even when all externally measurable facts about the agent are known.
Davidson distinguishes actions from mere happenings by claiming that an agent has an intention to perform an action, whereas no intention is needed for a happening. He says that an action can be explained by a 'primary reason', which implies a pro attitude (a desire or goal) towards some not-yet-existing state of affairs, and a belief that the action is of the kind that brings about this state of affairs. 38 Thus, the 'what?' of the action is contained in the question 'why?', and, according to Ricoeur,36 He distances himself from Locke, who said: '"Person" a forensic term. Person, as I take it, is the name for this self. Wherever a man finds what he calls himself, there, I think, another may say is the same person.' (Locke 1999, Book II Chapter XXVI). 37 For a wider discussion, see for instance Child 2006 andvan den Hengel 1996. 38 Davidson 1963, 686. such an explanation does not require an examination of how the consciousness of the agent changed over time (from the first motivation to act via the actual action to the final completion of the action's consequences), or of the impact the action has, has had, or will have, on the agent.
Ricoeur claims that Davidson misses a crucial point in his account of action.
Davidson considers the intention with which an action is done, but not the temporal development of the act of intending. He examines acts which have already been finished, looking at them in retrospect. 39 For him, an intention can be either correctly or incorrectly identified; the naming of the intention has a truth value which is atemporal. If the action is only analysed after it has been completed (this is what Ricoeur criticizes in Davidson's model), the planned effect of the agent's intention is substituted by the actual effect, and it cannot be determined how the causal intentions of the action's beginning, performance and completion were developing. 42 Although Davidson also sees the importance of determining the agent's intention, he looks only for the trigger which starts the action. Any intentions the agent may have had before or afterwards are not part of his model. For instance, the sentence 'the man desired to rise in his girlfriend's estimation and believed that giving her lilies would achieve this' has a truth value in Davidson's model: either it is true or not true that the man wanted to rise in his girlfriend's estimation when he acted to give her lilies. For Ricoeur, this sentence would not be sufficient to characterize the action: he would only see the explanation as complete if we knew whether the man always intended to rise in his girlfriend's estimationor whether he had any other intentions -, whether he learns something from the experience, and so on.
Hence, Ricoeur has two issues with Davidson's account: first, the fact that Davidson does not distinguish between the experienced and the observed action, and second, that Davidson only examines the intention of an act in retrospect, after the action was performed. The causes of some kinds of actions can be typified, for instance mechanical actions (like pushing or pulling a tractor), leading to general rules of causation. But for the explanation of actions that have a series of reasons that shift while the action is being performed one needs information about the action as it is experienced. Only the actor herself can say what her intention is. Stating this knowledge which is practical knowledge, knowledge without observationconstitutes an attestation of the agent to the act which acknowledges responsibility and confirms the confidence of the actor in her ability to follow her intention. This can only be given by the actor herself, in an utterance regarding both her accountability and her intention. For this kind of utterance, truth values cannot be determined, only degrees of veracity. 43 She may be lying, or she may be telling the truth about the responsibility she has committed herself to.
Put in a nutshell, Ricoeur criticizes Davidson's account because its focus is analytic, not hermeneutiche is accusing Davidson of asking the wrong question, while using Davidson's account as backdrop for expounding his own model. Ricoeur claims that Davidson neglects the moral imputation of an action to its agent and, especially, its influence on the development of the agent's self. 44 Ricoeur raises the question whether Davidson's ontology of events was not 'condemned to conceal the problematic of the agent as the possessor of his or her action'. 45 As we shall see in the next section, Ricoeur proceeds to re-examine the question of the relation of action to the agent with a special focus on the question 'who?', which is a central point in his account.

Self-Ascription
We are only able to interact consciously with the world we live in on a social level, Ricoeur claims, because we have the ability to ascribe our actions and our sufferings to our self, to an agent we identify as.
Ricoeur combines three claims from Strawson's account of personhood to define the term 'ascription'. 46 The first claim is that there are two kinds of predicate in connection with a person: material and psychological. These kinds of predicate cannot be reduced to each other. The second claim is that we attribute both material and psychological attributes to the same entity, the person. For instance, if we say that Donald Trump is a rude misogynist, we are referring to a single entity, the person Donald Trump. The third claim is that mental predicates such as intentions and motives can be attributed directly to a person. I can attribute intentions and motives to myself as my subject-attributes, and to someone else as their supposed subject-attributes. This process of attribution shows that I perceive myself to be a person who can be distinguished from other persons.
But -picking up Ricoeur's point from the previous sectionthe ascription of an action to an agent is only successful if it is linked to the agent's power to act. Having the power to act in a certain way means that I know what I want to change, what I want the outcome to be, and that I am confident that I have the physical, psychological and intellectual strength to perform the action. Ricoeur calls this 'a basic capacity' because it links the agent's intention with the real world. As Ricoeur puts it, a person's power to act is 'the basic capacity to make one of the things he or she knows how to do (knows he or she is able to do) coincide with an initial system-state, thereby determining the system's conditions of closure '. 47 Every action requires initiative, 'an intervention of the agent of action in the course of the world, an intervention which effectively causes changes in the world '. 48 Initiative is enabled by the power to act. It is, however, not the only causal influence on action. Initiative can be seen as both disjunctive and conjunctive regarding the agent's lived world: at one extreme, the original causality of the agent is antagonistic to other modes of causality, which are outside the agent's influence; at the other extreme, there is a synergy effect between the original agent's causality and other forms of causality, so that the forms of causality outside the agent's influence actually further the agent's action. For instance, if I want to pick up my dog, and he runs away from me, his behaviour is antagonistic to my initiative; if he jumps up to me, on the other hand, his action is synergetic to mine. The agent's initiative depends on a distinctive causal 28 capacity that cannot be reduced 49 to any other sort of causality, but can only manifest itself in conjunction with the other causalities of the agent's world. In this way, Ricoeur explains the relation we have to the world we are linked to through our embodiment: on the level of the lived self (the ipse), we ascribe motives and actions to ourselves, and on the level of our object identity (the idem), we perform actions in the world, either helped or hindered by its contingencies.
Establishing that the self is able to ascribe actions to herself is an important building block in Ricoeur's model of the self in its twofold role: as embodied, and as a living entity who can take responsibility for her own actions.

Time Horizons of Idem and Ipse
How is self-ascription, as described in the previous section, linked to Ricoeur's overall concept of personal identity? Self-ascription belongs to the concept of the lived self, the ipse, but taking responsibility also has consequences for objectified part of the I, the part by which a person can be identified from the outside. Are the promises I make consistent with my character? The answer to this question may depend on the time of asking.
For Ricoeur (in contrast to Descartes, as we saw in section 1.2) the question of identity is tied to that of temporality. He points out that 'the person of whom we are speaking and the agent on whom the action depends have a history, are their own history '. 50 The two aspects of personal identity that he defines, idem (sameness) and ipse (self), each have their own time horizon which is subject to change. In a constant dialectic 51 between the idem and the ipse, these two horizons can overlap, or they can become distinct. For instance, a young man in a rural community might promise his father to work on the farm permanently after his studies, driven to this promise by his happily settled ways in the country: he likes to be responsible for the livestock, he likes to be respected for his physical strength, and he believes in stability. At this point in time, the time horizon of his character(p) and the time horizon of his promise overlap. 49 Ricoeur uses the term 'reduce' in the non-eliminative explanatory sense. 50 Ricoeur 1992, 113. 51 I am following Ricoeur's understanding of the term 'dialectic': a struggle between two positions, in the course of which each is becoming clearer, urging for a choice to end the conflict.
During his studies, however, he encounters new ideas and gets used to a new lifestyle.
He now believes that it is more useful to spend his life in study, and he enjoys library research. His character(p) is no longer the same after many years of study, whereas his promise is still valid. 52 The time horizon of his (former) character(p) has ended, but the time horizon of his promise has not.

Identity as sameness over time
Ricoeur criticizes analytic philosophers for conceiving of personal identity exclusively in terms of the identity of the idem.
Traditionally, the question of permanence in time starts with two concepts: first, with the definition of numerical identity as the relation that holds between two relata when they are the self-same entity, that is, when the terms designating them have the same reference, and second, the definition of qualitative identity as the relation that holds between two relata that have properties in common. 53 The concepts of numerical and qualitative identity are not reducible to each other. Ricoeur says that to get an understanding of personal identity over time, these definitions need to be supplemented by 'uninterrupted continuity' 54 between the different stages of the individual whom we consider to be the same person. If the changes to a person are small between time tn and time tn+1, it seems safe to assume identity of a person at time tn and time tn+1. On the other hand, if we view a person at the time of their birth and in middle age, the difference seems so great that there appears to be a threat to identity. 55 To dissipate this threat, Ricoeur says, it is necessary to conceive a principle of permanence in time which is based on resemblance and uninterrupted continuity. Even if the threat to identity appears to be small, if the development of a person can be reconstructed over a long time period by following the gradual changes, a structure needs to be identified which shows that all of these changes have been happening to the same person. Hence, Ricoeur sees the necessity for a 'relational invariant' 56 , such as the structure of a tool that stays the same even if its parts are replaced. 52 As explained in section 1.2, Ricoeur counts the character(p) as part of the idem because it is one of the characteristics that make people recognizable from the outside. However, like the body, it is subject to slow changes over time. 53 See http://www.dictionary.com/browse/numerical-identity. 54 Ricoeur 1992, 117. 55 For a detailed overview of these accounts, please see section 2.3 Psychological Connectedness Accounts. 56 Ricoeur 1992, 117. Regarding identity as sameness, he suggests the genetic code as a good candidate to measure the identity of a biological organism. 57 But this criterion cannot apply to the identity of the self, which is the 'I' as it is experienced by the person herself, through which she enters obligations to the world she lives in. So Ricoeur looks for an identity criterion that is connected to the question 'who?' without reducing it to the question 'what?'. 58 He distinguishes between two models of permanence in time: character(p) and keeping one's word. The latter is a commitment towards a person, or persons, or a community.
In Ricoeur's terms, character(p) 'designates the set of lasting dispositions by which a person is recognized'. 59 A disposition develops out of two activities: the formation of habits and the acquisition of identifications.
Habits develop over time and can have different status: they can be in the process of being newly formed, or they can be in an already acquired, 'sedimented' state. The sedimentation of habits has a tendency to cover innovations which occurred earlier. As habits turn into lasting dispositions, they constitute character traits. They provide the character(p) with a history. If the sedimentation is strong, and the lived habits match the character traits, the idem overlaps the ipse, and they operate in harmony.
Acquired identifications are the value predicates that a person takes over from the community she lives in. For instance, she may be known in her village as the parson's daughter, the brilliant brain, the quiet one, or the village witch. If she takes over the social values of her community, she enters the community identity to a certain extent. As a result, she can be identified by these common values. The community values are values that concern things beyond our own survival; they can, for instance, be common ideals or common customs which strengthen the bond of the community. If a person identifies with such community values, this incorporates an element of loyalty into her character(p) and turns it towards fidelity, which is a defining element of the ipse-self. When these identifications turn into lasting dispositions, they also become character(p) traits. This is where the two poles of identity are in harmony with one another: the committed self and the settled character(p) share the same goals and values.
If, however, either the self or the character(p) undergo any changes due to other influences, this harmony can be destroyed. Character(p) has a history which is formed by two activities: the dialectic of innovation and sedimentation that takes place in habit acquisition, and the dialectic of the world's otherness and my internalization of its values that makes up the process of identification. 60

Identity as self-constancy over time
The other model of permanence in time mentioned above concerns faithfulness in keeping one's word, or self-constancy. Self-constancy is an achievement of the self which belongs to the dimension of 'who?'. 61 If I give a promise to someone, I become accountable for it, and thus the action of promising has an ethical aspect. Word-keeping develops its own time horizon from the words spoken as well as the trust the other person puts in my faithfulness. For instance, if a colleague tells me that someone spoke badly of her, and I tell her: 'I'll back you up', we both understand that this concerns her current work situation, and my conversations with her boss in the coming weeks.
Neither of us expects me to be true to this promise in ten years' time.
Holding one's promise within the pledged time horizon can be seen as a challenge against time: that I will honour my promise even if I should change my views on it, or my inclinations. 62 For instance, I might suddenly see my colleague as a plagiarist of other people's ideas, thus not wanting to speak up for her any longer.

Two time horizons
If there is great permanence of character(p) over time, this means that the idem and the ipse overlap almost completely -the faithfulness of keeping one's word is always aligned with character(p). At the other extreme, a permanence of promise-keeping (selfconstancy)while the character(p) undergoes some changeindicates that there is a large gap between the self and the character(p). This shows that none of these polar notions can be reduced to the other. 63 In the extreme case of faithful promise-keeping my character(p) (my idem) has changed, but my promise-keeping (my ipse) remains unchanged.
An example from fiction can illustrate this: in Graham Greene's The End of the Affair 64 , Sarah makes a solemn promise to God that she will break off her affair if her lover's life is spared in the bomb attack. He survives, and as she is a devout believer, she keeps her word. For the rest of her life, she is miserable because her promise prevents her from being with her true love. Love has become more important to her than before, changing her outlook on life. Her promise-keeping self and her loving character(p) are far apart.
Viewing the sameness of character(p) and the constancy of the self in opposition raises the question of how these can be viewed as part of whole, in a model of personal identity.
Ricoeur stresses the necessity of such a model by examining the paradoxes of personal identity and the attempts made by other philosophers to overcome these paradoxes, in particular by Locke,Hume and Parfit. 65 He claims that none of them give a complete account of personal identity because they do not distinguish between idem and ipse. 66 By conflating these two aspects of personal identity, they fail to recognize the enduring dialectic between them.

Narrative and Mimesis
If there is, over time, a continuing struggle between a person's lived self and her objectified identity, how is this established? Ricoeur says that all events in a person's life are woven together in a fabric of a story, a self-narrative. To illustrate how and why this happens, he distinguishes three levels of narrative activity. In this section, I am going to show how Ricoeur links the concepts of idem and ipse to the concept of selfnarrative through application of these different narrative activities.
As we have already seen in section 1.2, idem and ipse each have their own time horizon, which may or may not overlap. The development of idem and ipse over time is an important point in Ricoeur's account. The changes in a person's character(p) and in a 64 Greene 1979. 65 These accounts, Ricoeur's criticism of them, and state-of-the-art evaluations of these and other accounts of personal identity will be discussed in detail in chapter 2. 66 Ricoeur 1992, 126. person's actions and commitment can be measured from the outside in chronological time: for instance, I failed my driving test in 1995, and I promised to marry my (then) boyfriend in 1998. But when a person weaves her own life together in a self-narrative, events and actions no longer follow the chronological sequence: It seems a long time ago now that I made that promise to that man, but I feel the shame of failing the test as if it had been yesterday.
How, then, does narrative relate to time?
In Ricoeur's words, there is a correlation 'between the activity of narrating a story and the temporal character of human experience'. 67 He says that 'time becomes human to the extent that it is articulated through a narrative mode, and narrative attains its full meaning when it becomes a condition of temporal existence '. 68 To show how different levels of temporality come into being, Ricoeur defines three kinds of narrative activity: mimesis1 (prefiguration), mimesis2 (configuration) and mimesis3 (refiguration). As I shall explain, all of these three kinds of mimesis are involved in the weaving of a self-narrative.
Of the three kinds of mimesis, mimesis2 provides the central concept. The composition of the plot is the main instrument in the building of narrative. The narrator chooses and orders events and actions and puts them into a time frame. 69 During plotting, she imitates and modifies these events and actions, using existing conceptions.
The inspiration for the organization of events can varytheir choice and sequence can be driven by the character(p) of the protagonist, or it can follow a pattern which is embedded in the society's culture. For instance, the plot of Henry James' Portrait of a Lady 70 is mainly structured by the psychological development of Isabel Archer, whereas Emile Zola's Germinal 71 is built on the awareness of social problems in the nineteenth century.
In his definition of mimesis1, mimesis2 and mimesis3, Ricoeur shows that emplotment (the process of selecting and ordering imitated actions) mediates between practical human experience and the application of narrative to human life. This is especially important for the configuration of a self-narrative: understanding a fictional narrative gives a person the opportunity to gain new insights into the world she lives in. 67 Ricoeur 1984, 52. 68 Ricoeur 1984, 52. 69 , 33. 70 James 2009. 71 Zola 2008. Ricoeur analyses the competences a person needs to have in order to participate in a narrative on these levels. 72 To make the point clearer, I shall distinguish between three different roles here: The fiction reader, the fiction writer and the self-narrator. After a brief explanation of mimesis1, mimesis2 and mimesis3, I shall summarize the specific activities required for each of these roles.

Mimesis1
Mimesis1 is the activity of understanding the world of action and its structures. In order to be able to compose or understand a plot, a person needs to be familiar with the symbolic, temporal and structural characteristics of action.
Regarding the symbolic characteristics of mimesis1, a person must be able to distinguish action from mere physical movement. This entails identifying the agent and recognizing the agent's motives and social situation. 73 Many actions can be understood because they follow a symbolic convention that is specifically designed and acknowledged for the context in which the actions are performed. For instance, raising one's arm means voting for something in an assembly, but in a busy street it means hailing a taxi. Thus, the use of symbols makes an action readable to those who belong to a society in which these symbols have a commonly acknowledged meaning. The same applies to the symbols which represent social or ethical norms: a debutante curtsies when being presented to the Queen as an acknowledgment of the Queen's royal status, and in the nineteenth century it was customary for gentlemen to raise their hats to acquaintances in the streets as a mark of respect.
The second area we must be familiar with to understand a plot is the changeability of time as we experience it. The way we perceive the temporal properties of action is determined by the fact that we notice time differently when we are preoccupied with doing something. Past events and future plans are drawn into the experienced present when we are engaged with an action. Ricoeur adopts Heidegger's term 'within-time-ness' ('Innerzeitigkeit') for the time in which we ordinarily act. 74 72 At this point of his inquiry, Ricoeur does not distinguish between the competences needed by an author of literary fiction, a reader of literary fiction, or a person in the normal process of building a selfnarrative. 73 Ricoeur 1984, 55. 74 Ricoeur 1984, 61. Finally, we must know about the literary structures in which a plot is typically represented. In addition to the practical understanding of action and symbols, the author or reader needs to understand the rules of composition that enable the ordering of events into 'the total action constitutive of the whole story'. 75 These rules of composition form part of the society's cultural background. As Ricoeur puts it, 'To understand a story is to understand both the language of "doing something" and the cultural tradition from which proceeds the typology of plots'. 76 He says that 'knowing how to do something constitutes a 'repertory' common to the writer and the reader'. 77 The most widely used convention is that of leaving out events which have no relevance for the story. If it is not important for the work of fiction that a character eats, drinks, or uses the toilet, the writer does not describe these actions, and the reader assumes that they happen in the background.
The choice of subject also depends on a common cultural background of the writer and the reader. All societies have treasure-troves of orally transmitted myths and stories which are about gods, spirits, heroes and victims who behave in ways which are typical for a specific culture. For instance, many communities in the South Sea Islands have traditional mythical tales which deal with the dead, and the importance of preserving their honour and keeping in contact with them.
All of these pre-narrative structures (symbols, temporal experience, and the organization of action and plot,) give human life the quality of a story not yet toldthe meaning has yet to be drawn from all of these distinct elements by weaving them into a story. Understanding pre-narrative structures also encourages the creation of new fictional narratives and helps to understand the ones already told. Literature would not be comprehensible if it did not make use of existing figures of human action. 78 How could we understand the tale of Cinderella if we did not know that sweeping the chimney is seen as low and dirty work, and that marrying a prince is considered the height of social climbing? 75 Ricoeur 1984, 56. 76 , 57. 77 Ricoeur 1991b Ricoeur 1984, 64.

Mimesis2
Mimesis2 is the activity of configuring a plot into a narrative. It is the central part of two processes: the weaving of a person's self-narrative, and the creation of a fictional narrative. 79 Building on the pre-narrative structures of mimesis1, mimesis2 organizes events into a narrative whole. In order to achieve this, the configuration has to mediate between different elements of the narrative. This works in three steps: unifying separate events into a story, providing them with relationships, and ordering the events over time.
The first step is to find connections between the individual events of the story and the story as a whole. This can be viewed from two angles: as a unifying operation, when events are knit together into a story, or as an abstraction, when the story as whole is derived from a number of separate events. 80 The mediation consists of the struggle between these two reciprocal relations: the influence of events on the story as a whole, and the influence of the story idea on the treatment of the events. As a result of this mediation, a configuration is created out of simple succession.
For the second step of mediation it is necessary to find the most important factors of the story, and to create or identify relations between them. These factors can be agents, sufferers, goals, means, social interactions, external events, or turns of fortune.
The third step of plot mediation is to order the events of the story over narrative time. Two dimensions of narrative time are combined: the chronological time slices in which single episodes take place are consolidated into a time sequence which gives sense to the story as a whole. This process can include conflation or extension of the time intervals and leaps between different points of chronological time. There is an additional requirement for the configuration of narrative time: a reader follows a story in expectation of an ending. The ending of the story needs to establish a point of view that makes all of these time configurations acceptable to the reader in retrospect. 81 In addition to these three functions of mediation, plotting also has to make the connection between the theme and the representation intelligible to the reader. The story 79 It also comes into play with the creation of non-fictional narratives, for instance of historical novels, biographies, or autobiographies. The distinction between fiction and non-fiction is, however, not part of this work. 80 Ricoeur 1984, 65. 81 Ricoeur 1984, 67. becomes understandable to the reader because she is able to apply the schemata which are constituted in the cultural tradition of her society. For example, if the narrative starts with the words 'Once upon a time ...' the reader knows that she can expect a fairy tale with typical elements such as stereotyped characters, repetitions, and a moral lesson.
These schemata themselves are constantly subject to gradual updates when cultural concepts change over time. 82 The factors which define mimesis2 are not only relevant to the plotting of fiction. They also come into play during the emplotment of a person's self-narrative which extracts a personal configuration from a succession of life events, as I shall show in the section 'Emplotment and One's Own Story'.

Mimesis3
Mimesis3 is the activity of the reader who creates her own version of the fictional story by reading it. The reader follows the set of instructions given by the text and applies them to the story according to her own understanding and experience of the world. This is possible because the reader recognizes rules and paradigms which are known to her from mimesis1. Based on these, she is able to fill the lacunae of the story, and finish the configuration of the story in her own way. Thus, she completes the plotting process the author set up in mimesis2. 83 Mimesis3 finally applies the narratives which were plotted in mimesis2 to the time of action and suffering, 84 providing an 'intersection of the world of a text and the world of a reader'. 85 The reader engages with the fictional narrative and the character(f) on the basis of her own knowledge and experience.
Actualizing the story by following it is, however, not the only thing the reader does. When a reader engages with a work of fiction, she relinquishes the cares of her daily life, her self, and commits herself to the role of the reader as suggested by the text.
She appropriates the text in the sense of letting her ego go and opening herself to the revelatory powers of the text. Hence, the reader is able to overcome 'what could be 82 Ricoeur 1984, 68. 83 Ricoeur 1984, 76. 84 Ricoeur differentiates between action and suffering: an agent is an active participant in an action, a sufferer is a passive participant in an action. A character(f) can be an actor or a sufferer in a fictional story, and be understood by the reader, based on her own experience. A self-narrator can be an actor or a sufferer in her self-narrative. 85 Ricoeur 1984, 71. called 'the narcissm of the reader': to find only oneself in a text, to impose and rediscover oneself.' 86 He draws a parallel here between the Freudian effect of selfanalysis and the relinquishing of the reader's ego in a fictional text.
By entering the realm of a fictional world, the reader gains a new perspective on the world, which enables her to perceive situations through other eyes, and subsequently to reflect on her own life. I shall discuss this process in detail in section 7.2.

The circle of mimesis
Within the threefold mimesis, mimesis2 is the main force which is responsible for the building of narratives. Mimesis2, the process of emplotment, 'opens up the kingdom of the as if', 87 and in mimesis3, the reader creates her own rendering of the story, with the help of the pre-understanding gathered in mimesis1. What does this mean in connection with one's own life story? A fictional story is plotted by the author. A life story is a narrative which is derived from the events of a person's own life. How do these two kinds of story relate to each other? Ricoeur claims that we continually weave the story of our lives into a self-narrative of which we are the co-authors regarding its meaning. 88 In the process of doing this, the three levels of mimesis form an ongoing cycle, assuring the continuity between the potential story and the actual story for which a person feels responsible. 89 The following matrix gives a brief overview of the involvement of the fiction writer, the fiction reader and the self-narrator in the activities mimesis1, mimesis2 and mimesis3:  86 Ricoeur 1991a, 95. 87 Ricoeur 1984, 64. 88 Ricoeur 1992, 162. 89 Ricoeur 1984, 75.  In the matrix, there is an arrow from the 'refiguring fiction' activity of the fiction reader to the 'plotting varieties of self-narrative' activity of the self-narrator. This connection between these two activities is the main topic of my own account, outlined in part III. I am going to claim that the fiction reader's reading experience canunder certain conditionsinfluence her self-narrative.
But first, in the next section, I shall explore the general concept of the selfnarrator.

Emplotment and One's Own Story
In Ricoeur's model, a person's narrative identity relies on two processes: emplotment and the personalization of the plot.
In the previous section, I discussed Ricoeur's definition of plot as the organization of events and actions which are included in the story. What does this mean regarding our own life, and our self-narrative? When things happen to us, we need to make sense of them; we feel the urge to link them to possible causes, and to anticipate possible developments. However, our affairs often do not work out as planned or expected. As a result, we have to re-assess the value and the importance of past events to accommodate new developments. For instance, if I buy a valuable-looking antique gold ring at a jumble sale for a good price, I will afterwards proudly recall howover time -I built up the expertise to recognize the value of this ring. When I later show the ring to an acclaimed expert, and he proves to me that the ring is a cheap forgery, I will re-organize my self-narrative to assemble evidence of how I have often been too gullible in the past, and how people have warned me about that.

Concordance and discordance
Ricoeur claims that maintaining the connectedness of life as we experience it is an ongoing project that is configured by opposing forces. Our desire for an ordering principle (concordance) of the events in our life is confronted with the reversals of fortune that threaten to transform the ordering principle we have previously accepted (discordance).  91 We plot our self-narrative to make sense to our self. Thus, we become both co-author and reflexive character(f) of our own story.

Emplotment in fiction
Ricoeur draws a parallel between self-narrative and fiction reading: he says that when we engage with a story that links a character(f) to a plot, the events in that story cannot be viewed as impersonal because one invariably sees them as relevant for that character(f). 92 Emplotment does not only order the elements of the story, but it also attributes actions to the character(f), thus giving her a role in the story that may be active or passive. Ricoeur says that we understand the relation between the character(f) and the story told because we possess a 'narrative intelligence', a practical intelligence based on the stories that are intrinsic to our culture. By applying narrative intelligence to the story told, we attribute possible, actual, or completed actions to an agent who plays a role in the story. 93 Thus, a variety of different roles can be constructed. Ricoeur distinguishes between two extremes, the role of the agent who initiates a process of change or conservation, and the role of the sufferer who is affected by this process. In a complete narrative, action is interaction between different characters(f), and interaction is a rivalry between different projects of the characters(f) which might in turn oppose or promote each other. 94 The basic moral problem told in the narrative is based on the appreciation of the dissymmetry between the role of the agent and the role of the sufferer, and this moral problem drives the action. 90 Ricoeur 1992, 141. 91 Ricoeur does not introduce this issue at this point, but takes it up later, to be discussed in section 2.4. 92 Ricoeur 1992, 142. 93 Ricoeur 1992, 144. 94 Ricoeur 1992, 145. In a work of fiction, the author and the readers have a common understanding of how a story develops: the connection between action and character(f) causes an ongoing conflict that is internal to the character(f). The character(f), searching for concordance, draws her sense of herself as a singular person from the unity of her life as she perceives it. When unexpected events happen, this unity is in danger of being disrupted. In order to balance the impact of such intrusions, she has to integrate these events into the history of her life, thus changing contingencies into facts in her story.
Ricoeur says that the identity of the 'emplotted' character(f) can only be explained in relation to this dialectic: 95 the construction of the narrative constructs the identity of the character(f). Here is an example from literature: in The End of the Affair, Sarah holds on to a shaky concordance in the beginning. She is true to her passionate nature by having an affair with the man she loves, but she still feels guilty about betraying her husband.
Then a bomb falls, and she fears her lover has been killed. This leads her to make a promise to God that does not only contradict her inclinations, but forces her to change her life completely. The effects of the bomb have created discordance in her narrative identity.
How does the principle of emplotment apply to self-narrative?
Ricoeur says that the plotting our self-narrative is similar to plotting a fictional narrative. Self-narrative inverts the effect that contingency has in real life: an event that might have happened differently, or not at all, assumes the nature of a necessity. It ceases to be a random occurrence as soon as it is reviewed in retrospect, after it has been transfigured into a given event that fits in with the story. 96 In real life, a person is pressed to make decisions about her plans and her acceptance of obligations when unexpected events upset the planned course of her life. When such conflicts between anticipated events and real events occur, this can lead to a dialectic of the character(p) (idem) and self-constancy (the ipse). If, for instance, I am offered the job of my dreams, but find that this would prevent me from caring for an aged relative for which I pledged myself, my promise and my urge for self-realization create a conflict which can only be resolved by re-ordering my self-narrative. I need to re-evaluate the weight of my original promise in retrospect, making it either more or less binding, and I need to re-tell 95 Ricoeur 1992, 147. 96 Ricoeur 1992, 142. the story of my travel towards self-realization, as well as re-order my plans for the future. In this way, I have to refigure my whole self-narrative, as it presents itself to me now.
There are, of course, also other differences between plotting our self-narrative and plotting a fictional narrative. We can stop weaving the fictional narrative whenever we want, leaving the character(f) in their current dilemma until we choose to resume our work. In our daily encounters with the world and our social nexus, however, we need to regain our equilibrium quickly in the event of disruptions if we want to keep on functioning on an acceptable level. This means that in the event of severe conflicts, we often find it necessary to re-tell or re-order our own story in order to maintain a sense of self-unity. However, as Ricoeur says, the parallels between engaging with fictionplotting and self-narrating are impressive.

The help of fiction for self-narrative
Our self-narrative does not fall into a clear pattern automatically; on the contrary, it undergoes a variety of imaginative variations that we try out in the process of reinterpreting our own life history. During this process, different scenarios are put to the test. Ricoeur points out that fiction provides a vast treasure trove of stories that enables the author/hero of a self-narrative to imaginatively try out possible courses of action: actions which are in line with stability of character(p), actions which are in line with self-constancy, and any number of variations between these extreme poles. 97  the missing aspect would be the man's indoctrination as a religious fanatic). Practices can be seen as second-order units: 105 for instance, a profession like farming entails subordinate activities like feeding the poultry or starting a tractor, which are not practices in themselves. Taken out of context, the action of throwing seeds on to the ground does not make sense, but in the context of a farmer feeding free-range chickens it has a meaning. The rule that throwing seeds into a chicken yard 'counts' as feeding chickens, and thus, an act of farming, is an example of what Ricoeur calls a 'constitutive rule' of a practice: 'The rule, all by itself, gives the gesture its meaning' 106 .
In addition to that, a constitutive rule stresses the interactive nature of most practices. Practices typically involve actions of others, which makes them part of a larger social interaction. If the interaction is external, it can take the form of a conflict, or a collaboration, or a competition. For a chicken farmer, for instance, there could be a fight for big customers with other farmers, or a cooperation with others to buy commonly used machinery.
The interaction can also be internal, for instance in learning a profession.
Usually, professions are taught from person to person, they are based on tradition, and they go through a process of innovation and adaptation.
Summing up his definition of practices, Ricoeur concludes that the organization of practices contributes to giving a pre-narrative quality to real life scenarios, already providing roles for the agent who performs these practices. Narrative gives 'the polemical form of a competition of the narrative programs of different persons to the forms of human interaction that practices are based on.' 107 This means that the practices a person is involved in already form the grounds of her self-narrative.

Narrative Unity
In the previous section, we saw  The reading process is based on a 'pact of trust' 113 between the author and the reader: engaging with a work of fiction, the reader grants the author the right to show the character(f)'s mind from the inside, and to make an assessment of this character(f) and the actions and events in the work. During the reader's assessment of the character(f), the conflict between the author's strategy of seduction and the reader's strategy of suspicion comes into play.
The real author transforms herself into the implied author, and in this role, she tempts the reader to adopt the specific values which underlie the presentation of the fictional text. 114 The reader, on the other hand, refigures the fictional story during the reading process, fills the gaps with her own knowledge of the world, and decides whether the characters(f) and the fictional events are compatible with her own expectations. As Ricoeur says, the modification of the reader's expectations 'consists in traveling the length of the text, in allowing all the modifications to "sink" into memory, while compacting them, and in opening ourselves up to new expectations entailing new modifications'. 115 Thus, the reader creates her own fictional world and her own understanding of the character(f)'s life plan.
Reading fiction can only have a positive influence on the reader's narrative unity of life, however, if it eventually leads to self-examination and self-adaption. 116 There will be a full discussion of this point in Part III, where I present my own account of fiction reading in relation to narrative identity.

Self-narrative and authorship
An objection to the concept of narrative unity is that, obviously, we are not literally authors of our own lives. 117 Our self-narrative is constrained by many things over which we have little or no influence, such as our genetic code, our family, the date and time of our birth, illnesses, or natural catastrophes. 118 Ricoeur defuses this criticism by saying that we have three roles regarding the narrative of our own lives: we are narrator, protagonist, and co-author as to its meaning.

Narrative Identity and Ethics
In the previous section, I discussed Ricoeur's claim that humans are able to view their lives not only in narrative form, but even as a unified narrative. But how is such a narrative connected to the ipse, the lived self? Ricoeur says that we need to investigate 120 Sometimes, the reader gets the impression that the author chose this ending to get rid of the protagonist, not knowing what else to do with him, like in the ironic ending of the first version of 'Der grüne Heinrich' by Gottfried Keller: 'und es ist auf seinem Grab ein recht frisches und grünes Gras gewachsen' ('and the grass on his grave grew fresh and green', my own translation). 121 Sterne 1967, 35. 122 Ricoeur 1992. 123 Ricoeur 1992, 162. 124 Ricoeur 1993, 163. what humans want, and he claims that a person's ethical aims are linked to her life narrative, thus completing the unification.
He picks up his earlier statement that when the self appropriates an action to herself as the agent, she also imputes its moral consequences to herself. What part does self-narrative play within the process of moral self-assessment and moral imputation, and how does moral assessment of actions unfold?
What is good for us?

Ethical aims and fiction reading
The importance of our ethical aims becomes apparent when we read a fictional narrative: we invariably engage in evaluating the actions and characters in the story, approving or disapproving of them from an ethical point of view. Fiction provides us with a variety of thought experiments regarding possible actions and interactions. We judge these actions to be good or bad by applying standards to them. These standards represent the ethical aims that we have integrated into our own self-narrative. By applying our ethical aims and our practices to the action models outlined in fictional narratives, we come across warnings or incentives that assist us in examining our own aims and life plans. Thus, reading fiction can help us to form a unified self-narrative, which can be a good thing. However, fiction reading can also have other effects, and a unified self-narrative needs to meet certain prerequisites to be a good thing. I shall discuss this in detail in the presentation of my own account in part III.
In this chapter, I showed how Ricoeur's account of narrative identity is based on his two-level view of personal identity: the objective (the idem) and the subjective (the ipse) are both necessary to represent a person, and these two views cannot be reduced to each other. The idem and the ipse are linked on various levels: in language, in action, and in self-ascription. These links are incorporated in our self-narrative, which has two main functions: it works as a practical identity which helps us to deal with the conflicts in our interaction with the world and our social nexus, and it provides our life with a sense of unity. An inquiry into personal identity usually starts with the determination of what a person is. However, I shall not go into a detailed discussion here, but will assume the generally held view that a person is a being capable of self-reflective consciousness and agency. 132 The question of personal identity over time contains many puzzles. After the determination or assumption of what constitutes a person, most research into the persistence of person over time engages with the definition of numerical identity as a next step. 133 Numerical identity is the relation that holds between two entities when they are the self-same entity, that is, when the terms designating them have the same reference at all points in time and space. The relationship of numerical identity is reflexive, commutative and transitive. On this definition, identity claims must always be either true or false.
Trying to apply this definition to persons, some problems become apparent immediately. Persons change over time, both physically and psychologically, so that a person x at time tn may look and act differently from person x at time tn+1 even though she claims to be the same person and bears the same name. If the time difference is big, for instance between the teenager x and the old-age pensioner x, it may be very difficult to find even resemblance, let alone qualitative identity. The properties of a person can change dramatically over time. Even if this can be traced back to gradual small alterations, the overall change can be so severe that an additional link seems to be necessary, a bracket that links the different stages of a person's history.
Three different strands of research have been developed in quest of an explanation of personal identity over time: the bodily continuity accounts (BCAs), the psychological connectedness accounts (PCAs), and accounts of narrative identity (NIAs). I am going to present the main arguments of the most influential contemporary representatives for each of these strands and confront their claims briefly with Ricoeur's account. following four examples of 'threatened' personal identity (cases where the link between the various stages of a person's history appears to be broken):

Examples of personal identity in question
Case 1: There is factual evidence (fingerprints and DNA found on the murder weapon) that I killed a man ten years ago, but I do not have any recollection of having done this. So I insist that I am innocent. Am I right?
Case 2: I wake up in the body of a giant bug (like Gregor Samsa in Franz Kafka's short story Metamorphosis 134 ). Am I still myself?
Case 3: I am going to be executed next week, but a friend of mine will create a complete clone of myself (body and the complete set of brain states at the time of cloning) just before the guards take me away. The clone will escape and live on. Should I be concerned about my execution?
Case 4: My husband claims that I am now a completely different person from the woman he married, and consequently files for a divorce. Am I the woman whom he promised to love forever at our wedding?

Bodily Continuity Accounts (BCAs)
BCAs claim that where x is a person at time t1, and y is a person at any time tn, y is the same person as x iff y's body is biologically continuous with x's body. For these accounts, a person's identity over time consists in the identity of her body. 135 Olson's 'Human Animal' account is a currently popular variant of the BCA.
Olson claims that people are human animals, that 'each of us is numerically identical with an animal'. 136

Overview
Psychological Connectedness Accounts (PCAs) define criteria for the persistence of a person over time that consider the continuity of psychological states. They claim that where x at time t1 and y at any time tn are persons, y is the same person as x iff y's psychological states are continuous with x's psychological states, provided that these psychological states have developed from one another either directly (psychological connectedness), or through overlapping chains of direct psychological connections (psychological continuity).
These accounts have evolved from John Locke's claim that personal identity entails the consciousness (through memory) of having existed as the same being at different times. 142 Psychological connectedness accounts have standardly focussed on 140 Ricoeur 1992, 126. 141 Ricoeur 1992, 116. 142 'Personal identity. This being premised, to find wherein personal identity consists, we must consider what person stands for;which, I think, is a thinking intelligent being, that has reason and reflection, and can consider itself as itself, the same thinking thing, in different times and places', Locke 1999, Book II Chapter XXVII section 9.
the existence of memories, and I am going to assume that standard version here. One objection against this criterion of psychological connectedness has been raised: if the concept of our memory entails that I can only recall my own actions and experiences, then the continuity of memory presupposes personal identity.
To cope with this problem, philosophers defined the term 'quasi-memory' or 'q-memory': Having a quasi-memory of a past experience entails that I seem to remember having that experience, that somebody had this experience, and that my apparent memory is causally dependent on that past experience in the right kind of way. 143 Under these conditions, psychological connectedness is preserved. Ordinary memories can thus be understood as a subclass of quasi-memories. If we assume that I can also have a q-memory of an event that was actually performed by a different person, the continuity of memory does not presuppose personal identity.
Derek Parfit's view is a close cousin to PCAs that has raised much controversy. Parfit argues that although I am intuitively afraid to die, the existence of my replica has the same quality as my ordinary survival. My replica is psychologically continuous with me as I will be when I press the button. She believes she is me. Ifas Parfit claimsa person is nothing over and above psychological continuity, nothing is lost by my death if my replica is created as an immediate successor.
Parfit would find the following answers in the example cases from section 2.1: In case 1, he would look for evidence / witnesses to establish whether there is psychological continuity between my person at the time of the murder I am accused of, and the person I am now. My current lack of memory that I committed the murder does not contradict Relation Rmy brain states now might be psychologically connected to my brain states at some time between the murder and now, when I still remembered having committed it. If the conditions for relation R are fulfilled, I am wrong in proclaiming my innocence.
Case 2 is more difficult. In a sudden metamorphosis like this, Relation R would require that the brain states I had before the transformation should exist unchanged after the accomplished change. If a giant bug is conceivable, is it also conceivable that this giant bug is capable of producing psychological states that have the same structure and quality as the human psychological states I had before? Parfit is fond of science fiction examples, so he could perhaps assume that the bug was created by aliens who were able to copy my previous brain to something that processes information in the same way, especially as my immediate memories appear to be intact. Otherwise the transformation of human brain states to a bug brain would have been like trying to run the newest Windows version on a1980s Atari computer. Assuming that the bug's brain is able to process information in the same way as my previous human brain, Parfit  Parfit's claim that everything about a person's persistence over time can be described in an impersonal way, without referring to the notion of a self.

Reductionism and what really matters
According to Parfit, a person's existence consists in nothing but the existence of a brain and a body, and a sequence of interrelated physical and mental events. Ricoeur argues that this definition disregards the intuitive concept of 'mineness'. Although Parfit constantly uses personal pronouns like 'mine' or 'our' (see for example 'Why Our Identity Is Not What Matters' 153 ), he dismisses this concept as nothing more than a practical linguistic usage. He does not attach any importance to the process of selfascription in a person's social relationships.
Ricoeur disagrees with Parfit's dismissal of mineness as nothing more than a linguistic custom. As Ricoeur points out, mineness is an important issue for a person in daily practical life as it generates a feeling of self-constitution. This is what our experience in daily life tells us: no matter whether something is a physical object, or a belief, or a personal attribute -it feels more valuable to us if we see it as 'mine'. An interesting and more recent expression of this is the 'Jesus and Mo' cartoon in which Mo and Jesus agree that if our beliefs are based on bullshit, then so are our identities. Relation R consists of person stages at isolated points in time, linked by psychological continuity in a non-branching form. For Parfit, Relation R is the only thing that matters in survival. If this were the case, how could it be possible to impute responsibility for a promise that was given by a person stage at time t0 to a person stage at any time tn? 157 Viewed from the outside, this does not present a problem, but Relation R does not explain why a person should be in an inner conflict if she does not keep her promise. The event-ontology of Relation R does not support the self-accountability of an agent for a promise made, or an action performed, in the past.
Hence, Ricoeur claims that Parfit does not ask the right kind of question when he investigates the conditions for survival. For Ricoeur, self-accountability is an important aspect of personal identity which cannot be reduced to anything else.
Parfit does integrate ethical considerations in his account, especially discussions about the rationality of self-interest. He says, however, that it is credible to be more interested in the quality of future experiences than in the question whose experiences they are going to be. 158 Regarding past acts and commitments, he holds that if the psychological connections between a criminal now and the same criminal at the time of her crime are weak, she is less culpable than she would be if they were strong, and a similar principle should be applied to commitments. 159 For Parfit, personal responsibility is proportional to psychological connectedness. As his approach does not consider self-attestation and self-constancy, it leaves out thoughts about ethical subjectivity. 160  Reductionist claims about our concern about our own future: the 'Extreme Claim' states that there is no reason for our concern about our future at all, and the 'Moderate Claim' says that we are justified in caring about our survival. 163

Reductionism and the Impersonal Description claim
The second area of Ricoeur's disagreement with Parfit concerns Parfit's claim that the identity of a person over time is determined by facts that can be described in an impersonal way, without referring to the notion of a person.
Ricoeur sees the root of the problem in Parfit's use of terminology. Parfit defines Relation R as a series of connected mental events, and Ricoeur claims that the word 'event' as a term of reference eliminates mineness. Mineness does not fit into the ontology of eventsit describes the relation of one's lived self to one's own actions, one's own character, and one's own body. Ricoeur sees Parfit's insistence on the term 'mental event' as a further indicator that Parfit fails to make the necessary differentiation between the ipse and the idem. 167 Christine Korsgaard supports Ricoeur's criticism. She stresses that before examining whether a person is a subject of experiences, or just a stream of experiences that can be described without reference to a person, it is important to investigate the underlying assumptions first. Korsgaard says that Parfit's account is based on viewing a person as a location of experiences, and a person's actions as experiences. 168 Parfit begins by claiming that 'a person is what has experiences, or is the subject of experiences', 169 and then, as Korsgaard puts it, he takes away the subject. 170 Korsgaard's approach is similar to Ricoeur's in that she also takes a practical view on personhood. In her own account, Korsgaard distinguishes between internal (authorial) and external sources of change. Following the Kantian view that the conception of ourselves as agents is the basis for the standpoint of practical reason, 171 she claims that internal changes preserve authorial connectedness and the continuity of the self, whereas external changes break authorial connectedness and change the self.
She holds that a person is an agent who has to plan her actions, even the most trivial ones, some way into the future. In order to make the necessary choices, she needs to identify herself as a person who is constructing a specific life plan. To be what she is now, she has to identify with her future. 172 She views herself as the author of her actions and the leader of her life. 173 Korsgaard says that Parfit's Reductionist account disregards the authorial feature of our sense of identity. 174 Regarding the issue of impersonal description, Ricoeur raises the question whether subjectivity can be reduced at all. For him, the self depends on being embodied and interacting with the world. To illustrate his point, he points out the similarities between real life and realistic literary fiction. From a hermeneutical point of view, selfnarrative entails a process of ongoing interpretation and is thus closely related to reading realistic literary fiction. One of the reasons why we like to engage in realistic literary fiction is that the protagonists experience something similar to what we ourselves go through: they interact with the world they live in through sensations and actions of the body. It is an integral part of the story that they perceive their own body as a dimension of themselves. Literary narrative presupposes 'our corporeal anchoring in the world'. 175 The imaginative variations that literary fiction weaves around the protagonists' embodiment are also variations on their self, on their lived existence that depends on relations with the world. This illustrates that subjectivity is an essential part of our interaction with the world.
Ricoeur claims that it is of insurmountable importance for us that we live in our own body, calling the corporeal and terrestrial condition of the self the 'existential invariant'. He criticizes Parfit's thought experiments becausein opposition to realistic literary fictionthey render our corporal anchoring to the world contingent and so turn the invariant into a variable in an imaginary construction. 176 In Parfit's account, brain states are taken to be the equivalent of the person, and they undergo various technological applications: blueprinting, cloning, transplantation, bisection, reduplication. Ricoeur holds that these science-fiction variations only deal with identity in the sense of sameness, whereas the variations of literary fiction apply to selfhood in its tension with sameness.
Atkins also criticises Parfit's denial that personal identity is in any way related to the conception of one's experiences as being one's own. She says that Parfit ignores an important feature of self-consciousness. 177 Atkins endorses Kant's third paralogism here: According to Kant, it is unsuitable to apply a model of permanence in time that we use to understand empirical objects to understand the persistence of subjects as well.
The numerical identity of a subject cannot be proven, as it can for objects, because a subject is not of the same epistemological status as an object. 178 Atkins agrees with Ricoeur that the continuity of personal identity over time requires an integrated process of self-ascription that must be effected by its subject. It is not enough if others believe in withouthowever -aiming to provide a conclusive objection against Parfit's account.
In the next section, I shall give an overview of other accounts of narrative identity, and I shall explain why I chose Ricoeur's accountand not one of the othersas a foundation for my own account of the relationship between fiction reading and narrative identity.

Narrative Identity Accounts (NIAs)
Narrative Identity Accounts (NIAs) claim that a person shapes her sense of herself by weaving the events of her life together in an ongoing autobiographical narrative. 180 In doing this, a person takes the role of author, protagonist, and narrator in parallel. As we have seen, Ricoeur's account is a kind of NIA. There is, however, a fundamental difference between his account and most other NIAs: Ricoeur views personal identity from two perspectives, the question 'what?' and the question 'who', which cannot be reduced to each other. We shall see how this distinction gives Ricoeur's account more explanatory power than most other NIAs can offer. But first, let us have a look at the general approach of NIAs. These accounts claim that narrative is an aspect of personal identity: what does 'narrative' mean in this context, and how can narrative be connected to personal identity?
Generally, a narrative is understood to be a representation of a sequence of events, which shows them as standing in some causal or meaningful connection. 181 The recounted events should be organized in the form of a plot, laying out the events, the actions of a protagonist and the actions of other persons involved in the context of the story world. The plot has to arrange all of these element in such a way that one can both ask and answer the questions 'who? ', 'what?', 'when?', 'where?', 'how?', and 'why?'. 182 All NIAs agree that a self-narrative is a story, with 'I' as the main character, the narrator, and the co-author of the story.
Weaving one's life events together into a self-narrative may appear to be a leisurely, reflective activity, and this impression has given narrative identity accounts an air of the ivory tower in public reception. This is, however, not what most of these accounts say. They are more concerned with the needs of action and with ethical concerns 183 than with reflection. Practical identity provides the guidelines that enable a person to act and make decisions in her daily life. NIAs claim that the continuous process of maintaining a self-narrative develops a set of rules that serves as a basis for moral choices and actions.
The main point that NIA theorists argue about with their critics is how a selfnarrative presents itself. Is it explicitly communicated to oneself, in moments of selfreflection, or told to others, in scenes of more or less truthful self-portrayal? Is it a running commentary in one's head, or is there anything that gives it stability? Is there more than one self-narrative at a time, and, if there is, which, if any of them, constitutes the self? I shall discuss these questions in the subsequent sections, in particular in section 2.4.3.
Some NIAs are descriptive, some normative. Descriptive accounts (NIDAs) claim that self-narrative is inherent in human nature: all human beings make sense of their lives by putting the events of their lives, their memories and their expectations into narrative formalthough the theories differ on whether this generates a complete autobiographical self-narrative, or only selected strands of one's life-story. Ricoeur's account is one of the most influential NIDAs in philosophy, although other models have been developing independently of his.
Based on the general claim that humans develop their own stories about themselves, NIDA theorists take differing views of the importance of self-narrative.
Some put forward the strong claim that self-narrative is constitutive of personal identity, or of the self. 184 Others endorse the weaker statement that self-narrative does not constitute anything, but is involved in creating a practical identity; it functions to combine the different aspects of an embodied human life. I am going to discuss the importance of self-narrative in section 2.4.1.
Normative accounts (NINAs) claim that the process of self-narrative is conducive towards leading a good life, which is generally understood to be a unified life. There is disagreement between the different theories about whether this is always the case, and under which circumstances building a self-narrative can be dangerous. I am going to evaluate these arguments in section 2.4.2. Ricoeur's account, which is descriptive, also has a normative aspecthe says that we cannot help developing a selfnarrative, but that it would be better for us if we concentrated more on self-reflection in order to refigure our self-narrative towards an improved unity of life. 184 Different meanings of the term 'self' will be discussed in connection with the specific accounts that have their own definitions for this.

NIDA: do we narrate our lives?
In this section, I shall deal with three questions: 1) What is self-narrative actually like?, 2) Must self-narrative be based on experiential memories?, 3) Does self-narrative have to be realistic and explainable?
What is self-narrative actually like?
Regarding the first question, let us start with what descriptive narrative identity accounts claim. NIDAs are based on two assumptions: first, that we have 'story-telling brains' that enable us to put events and persons into a coherent narrative, and second, that we are forced to use this capability in order to perceive ourselves as subjects that persist over time, and thus see ourselves as unified agents who need to make decisions about their current life and future plans. For some NIDAs, the emphasis is on the practical concern raised by a person's continuous need to act, others focus on the concept of personal continuity over time. All of them stress that self-narrativewhether we actively engage with it or notgives us an understanding of ourselves, and of what we want and value.
NIDAs do not imply that we consciously experience our lives as coherent narratives as we go through life-they say only that self-narrative is an ongoing process that we human beings perform. A narrative self-conception establishes itself in patterns of one's actions, one's commitments, one's dispositions, and one's relationships. 185 Ricoeur sees narrative identity as a necessary mediator between the actions of a person and her ethical values. The main feature that sets his account apart from others is his differentiation of the idem and the ipse. As we saw earlier, the dialectic between idem and ipse can lead to personal development.
His model can be classified as a weak NIDA. He holds that a person develops her ability to weave a self-narrative as her interaction with the world moves from the simple articulation of desires and wants to the realization that she is a person among other persons. 186 This insight leads her from simple desire gratification to the bigger goal of leading a life that feels continuous to her, in which she can develop self-esteem as a basis for interacting with others. To underpin her feeling of continuity, she tries to entwine her past experiences and her expectations for the future into a unified narrative that contains the contingent events that happened in her life and integrates them into her own story. Her own point of view introduces causal connectedness in the narrative, and this gives her reasons to make plans for the future which then become part of her selfnarrative.
Ricoeur sees self-narrative as instrumental in achieving a unified sense of self.
Having a unified sense of self helps us to develop self-esteem. This makes his account a NINA as well as a NIDA: on the one hand, he depicts the struggle for self-esteem as an automatic procedure, but on the other hand, self-esteem is also a moral judgment of one's self. Ricoeur  by a different interpretation of the world in which I move. So the conflict drives me to a dialectic between idem (sameness, which includes character) and ipse (the self that has entered obligations towards its social world) during which the self-narrative is reflected, reviewed and reconfigured whenever self-esteem is in jeopardy. 187 Not all changes of the self-narrative are due to conflict. Gradual changes of the ipse lead to changed habits and dispositions, which will eventually 'sediment' into the story, and thus into the core of its protagonist, the idem. In this way, practices of the self can gradually settle into the character(p).
Ricoeur does not claim that the creation of self-narrative is always a conscious procedure, nor that a person always tells this narrative to herself or to others. Rather, the narrative serves two purposes: it is part of the process of personal development, and it provides values and rules for the person's interaction with her social environment. In his view, narrative identity is a practical identity.
Atkins' account of practical identity is similar to that of Ricoeur These three perspectives are unified by the subject's self-narrative: 191 based on the learning process from second-person relationships, the subject perceives herself as the protagonist of her own story, and her sensory experiences and social encounters become events in her autobiographical narrative. This view is similar to Ricoeur's explanation of the development of self-narrative, but Atkins goes a step further in stressing the importance of bodily contact in early childhood that constitutes the secondperson view. As Ricoeur would express it, the ipse-identity, the 'self', is involved in a dialectic not only with the idem-identity, but also with the 'other than self'. 192 Atkins adopts and expands Ricoeur's view here.
Atkins holds that the capacity for narrative is inherent in the human brain: narrative is imitation of action, and a person's brain structures allow her from a very early age to learn to identify others by imitating them. As the ability to interact with others develops in a social context which is structured by a value system, a person as an agent is a subject of imputation whose actions can be judged by two sets of values: by her own standards and by the standards of her community. 193 When a conflict occurs between one's actions and either one's own values or the values of the community, this leads to self-reflection as a necessary step to resolve the problem. For Atkins, being a self is basically an activity. This activity consists of self-constitution and selfunderstanding, where self-understanding is articulated by narrative. So for both Ricoeur and Atkins, it is important that at least some of the process of self-configuration is reflective. They both say that when we create a self-narrative, we follow the aim of a unified life, and they also claim that in doing so, we develop a practical wisdom in a 'dialogical process of critical solicitude '. 194 In other words, our practical wisdom enables us to articulate the reasons for our actions. This is an important aspect of Ricoeur's account which I am going to discuss later in connection with self-reflection and fiction reading.
Summing up the discussion of the first question (what is a practical identity?), the accounts of Ricoeur and Atkins present a common framework for the claim that a person needs a practical identity which is built up during the course of her life.

Must self-narrative be based on experiential memories?
The second question (must self-narrative be based on experiential memories?) comes up in Peter Goldie's account of narrative identity. Goldie's NIDA aims to investigate 'the narrative sense of self', the way that human beings think about their lives in narrative terms. His account differs from Ricoeur's and Atkins' accounts in that he does not discuss the aspect of personal identity as sameness, 195 and he is not interested in the role of embodiment in the development of the self. In his view, 'having the ability to think about our past and future is part of what it is to be human '. 196 He uses the term 'narrative sense of self', meaning 'a way of thinking of oneself, and of others, in narrative thinking '. 197 A narrative sense of self does not require deep narrative coherence in the content, the only required coherence is that it reflects one's own life. 198 We use this ability in our social interactions because we have a narrative sense not only of ourselves, but also of other persons. This definition is very similar to Ricoeur's view that all persons weave a self-narrative, butin contrast to Ricoeur -Goldie does not think that embodiment plays an important role in the configuration of self-narrative.
Goldie combines his weak claim with the strong assertion that one does not need to have experiential memories in order to be able to think narratively of one's past. 199  But as soon as I get into an ethical struggle which requires an urgent decision regarding social interactions, experiential memories with their emotion-laden stories can make me 'go the extra mile', which a mere recollection of facts would not accomplish. Humans are continually involved in situations where they have to decide and to act, and as I showed earlier in this section, self-narrative is required at moments of conflict. So the question is: does it make a difference whether we resort to our self-narrative at a moment when urgent action is required, instead reflecting on our sense of self at a quiet moment? I say that there can be a difference when the conflict arises in contact with other persons: Our reaction to others often depends on how we can compare their situation with our own, and during this comparison, only experiential memories have enough psychological weight with us to play the required role.
Here is an example to illustrate this. Imagine the following situation: you suffer from loss of experiential memory. A particularly unprepossessing homeless person stumbles on an icy sidewalk and falls down in front of you, vomiting profusely, and you see that she is seriously injured, her arms and face are bleeding. You have a disposition to be helpful, though you know from your self-narrative that you are not self-sacrificing.
So you ring the ambulance and wait until they arrive, then you leave. If you had not lost your experiential memory, however, you might have re-experienced the memory of the time when you yourself fell onto an icy surface, how hard and cold it was and how the fresh wounds burnt. In this case, you might have overcome you disgust and done something to ease the unfortunate person's discomfort, for instance putting a shawl under her face or dragging her away from the dirt. The experienced pain is part of your narrative sense of self.
I hold that the loss of experiential memory makes you react in accordance with your character, whereas only the re-living of experiences makes it possible for you to react with sympathy, as a fellow human being who 'goes the extra mile', because you are familiar with the phenomenological sensations and the feelings of the other through your experiential memories. You 'relive' them in your imagination prior to your decision on how to act. 200 This example shows that although it is possible for you to narrate the events of your life without experiential memories, you need to have experiential memories (which you remember because of the emotional impact of the stories in which they are embedded) in order to have a narrative sense of self that allows you to have social interactions which go beyond mechanical reactions.

Does self-narrative have to be realistic and explainable?
For the treatment of the third question, I introduce Marya Schechtman's account of narrative identity. It adds a further assertion to the NIAs I have presented so far, therefore I discuss it in this separate section.
For Schechtman, self-narrative constitutes the moral self, and this puts two constraints on the quality of the narrative: the reality constraint and the articulation constraint. As Schechtman's account is both a NIDA and a NINA, I discuss it in both sections. I am including what she terms 'the reality constraint' and 'the articulation constraint' here because they contain conditions under which self-narrative could actually fail to be established.
The reality constraint requires that the self-narrative is consistent with basic facts that we know about persons and about the world we live in. 201 This is a stronger claim than that of Ricoeur, who merely stipulates that a self-narrative must be built on the principles of realistic fiction in order to form a basis for action. Atkins also endorses the reality constraint, but in a weaker form than Schechtman. For her, self-narrative must not be based on 'factual errors of a major kind', 202 for instance about places and events, because 'such errors would disable the semantic web that links 'who' to 'where' and 'when''. 203 Atkins claims that our narrative identity is subject to defeat if our selfview differs greatly from the view others have of us. In contrast to this, Schechtman says that such a situation will not lead to an overthrow of the self-view: even if a person's self-narrative is built on major factual errors, it will still overlap with a possible realistic narrative of this person sufficiently to enable self-constitution. 204  This constraint does indeed seem necessary, provided that one concedes that it is not necessary to answer these questions verbally. It should be enough to have a sense of self-ascription.
In the preceding discussion, I have outlined the ways in which a narrative identity can be seen as a practical identity. In addition, I have presented an example that shows the importance of experiential memory for the narrative sense of self. As a last 204 Schechtman 1996, 127. 205 Regarding moral issues, it is of course a different question whether it is possible to lead a fulfilled life, a 'good life', that is built on wrong premises. But such a discussion, fascinating as it would be, would go beyond the scope of my work. 206 I would like to stress that practical identity is under discussion here, not metaphysical identity. 207 Schechtman 2007, 165. step, I challenged the reality constraint on self-narrative, showing that only a very weak constraint is neededi.e. one that rules out only extreme cases of factual error -, and suggesting that it should be renamed a 'social acceptance constraint'.
Do we have to relate to our past in a positive way in order to live a unified life?
As a next step, I am going to discuss Schechtman's claim that narrative identity only gives us a sense of self if it is based on past experiences for which we still feel a positive affective connection. is in the quality of personal survival. She says that subtle survival implies personal continuation on a practical level that gives cause for morality, self-interested concern, autonomy, and authenticity. 209 New decisions are informed by recognition and pull of characteristics which belong to a past self that we still feel positively connected with.
Schechtman agrees with Ricoeur and Atkins that the appropriation of actions as one's own requires self-narrative, but she has an issue with theories of narrative identity in general. She says that usually, theories of narrative identity have one fault: even psychological changes that seem identity-threatening can be told in a coherent selfnarrative, and the narrative glosses over the fundamental change without being able to account for it. 210 These theories of narrative identity, she says, hide losses of subtle survival, and she claims that her own theory mends this fault.
To illustrate her own theory, she uses an example that Derek Parfit created in a different context: 211 A young Russian socialist is going to inherit a large estate within a few years, and he plans to give it to the peasants who live there. Fearing that the irresistible enticement of money will weaken his current ideals in the future, he signs a document that binds him to give away the land as soon as he inherits, and which can only be revoked with his wife's consent. He then asks her never to give that consent because the person who would ask her for it would only be 'his corrupted later self'. 212 For Schechtman, this example shows that it is important for a person's continued identity that no radical psychological changes occur. In her terminology, the corrupted Russian in the future will be only a case of 'basic survival' because there is no empathic access between his corrupted self and his idealistic past self. There would, however, be 'subtle survival' if he were to hold on to his ideals, as in that case the empathic access to his past self would allow him to base his actions and decisions on his past learnings and experiences.
Schechtman criticises other narrative theories for making this kind of transition appear plausible too easily: the young Russian would be able to imagine many possible narratives that give good reasons for his gradual transition from an idealist to a capitalist. The change would be appear to be continuous, although the overall effect would be drastic.
She uses another example to demonstrate the difference between basic and subtle survival: a party girl marries, gets a career and grows up into a serious matron.
When serious matron looks back on her wild youth, she cannot evoke the emotions and predilections she had at that time. Schechtman claims that in this case, the change is identity-threatening because subtle survival is not maintained. NIDA followers would hold that the changes from giddy girl to responsible matron are woven into a coherent self-narrative, but Schechtman says that this is not enough to preserve identity. 213 Here, we only have basic survival. If, however, the party girl grows into a less serious matron, she recalls how exciting it felt to go to parties. She feels wistful when she watches party-goers although now she no longer wants to join them because she is wiser now, If he sees himself as someone who had better not stick his nose out too far, knowing from past experience that he has always hated pain more than anything else, he might think up an excuse not to go to school the next day. If his image of himself is that of a hero in the making who has history of successful fighting, he might take boxing lessons and pay the bully back in kind himself after more training. If he thinks he has been developing a Machiavelli-like cleverness, he might persuade his big brother to beat up the bully the next day. This everyday example shows two things: first, that even people who are not set on making ethical plans for their lives do make plans for the future which are based on the way they see their own life and their self-narrative, and second, that the future self involved in these plans does not seem remote to them. The direction of our concern for the futureof what exactly we want to achieve for ourselvesis far from being hard-wired, as Strawson claims. 221 Strawson says that our concern for our well-being is based on our instinct for self-preservationhe does not, however, consider that our models of well-being are more complex than that, and thus subject to change, depending on the development of our life-story. Our concern for the future appears to depend on the experience and self-estimation of us as persons, which is part of the story of our lives.
A third objection to Strawson's theory is that his distinction between diachronic and episodic persons is not sharp enough to support his conclusion. On the one hand, he claims that the diachronic and the episodic style of self-experience are fundamentally different, but on the other hand he says that one is not necessarily narrative theorists do not say that we try to live our lives as if we were characters(f) in a novel, and they do not say that we are consciously aware of weaving a self-narrative.

NINA: does narrative provide a unity to one's life?
Following the discussion of descriptive accounts of narrative identity (NIDA) in the previous section, I am now going to give an overview of normative accounts of narrative identity (NINAs). Most NINAs make claims about whether or not building a narrative identity, or developing a sense of narrative identity, can help a person to lead a fuller and more satisfactory life. On top of the descriptive claim that self-narrative is something that 'we do', they say that it makes a difference whether or not the secondary reflection involved in self-narrative is actively supported. All NINAs agree that it is beneficial for a person to have as much personal unity as possible in order to lead a meaningful life, but they disagree on how this could be achieved in the context of selfnarrative. There are two particular bones of contention between NINA supporters: first, whether it is necessary for personal unity that one's self-narrative is based on facts, and second, whether it is necessary to endorse one's favoured memories and character traits in order to maintain personal unity.

Does our self-narrative need to be based on facts?
For normative accounts, this question is slightly different from the problem of the reality constraint discussed in the previous section. Here, the requirement that Atkins postulated for the descriptive account of narrative identity 227that self-narrative fails if it is based on grave factual errorscomes in again, this time as a normative prerequisite for a good and unified life.
This view is endorsed by Ricoeur, Schechtman, Atkins, and Goldie, in different degrees. Ricoeur and Schechtman both say that one's self-narrative needs to stand the test of action, and that one's major factual errors about one's self can be corrected by interaction with one's social nexus and subsequent self-reflection.
Atkins partially agrees with this, saying that 'all elements of a narrative identity must stand in relations of mutual implication and explanation', but she does not make assumptions about how major factual errors about one's self could be corrected. 228 She also says that delusional mental states can prevent a coherent narrative identity. 229 Ricoeur does not particularly mention delusional cases, but the following example shows that this case needs to be considered.
In Patricia Highsmith's short story 'The Heroine': 230 a young woman works as a nurse for a family with two children, and she is disproportionally grateful for having the job. In her own self-reflective view, she is so unworthy of her position that she decides she has to do a heroic deed in order to deserve it. So she sets fire to the house with the intention of saving the children at the very last moment. She makes no special preparations for the rescue because she is convinced that her passionate longing is enough to ensure success. The story of the heroine ends when the fire becomes threatening, so we do not know whether she manages to rescue the children. If she does not, and survives, the terrible outcome might cure her self-delusion and lead her to a better life. However, if she does save the children, and survives, this success will probably strengthen her delusion. This example shows a flaw in the accounts of Ricoeur and Schechtman. They say that one's self-narrative will lead to self-reflection in case of conflict -I am not going to argue with this statement -, but they do not investigate what happens if one's self-narrative is never put to the test of action. I think they should add this possibility to their accounts, and concede that in such a case, a self-narrative may persist even if it contains basic factual errors.
Goldie is also much stricter on the subject of self-delusion than Ricoeur and Schechtman. He insists that our narrative sense of self must be based on factsif it is not, it is not a sense of self, because it does not refer to us. 231 To illustrate why a true narrative sense of self is important for a good life, he points out four 'fictionalizing tendencies' that endanger the creation of a realistic self-narrative. 232 The first one he identifies is that we plot our lives, meaning that we tend to forget that we are only the co-authors of the representation of our lives, not only dependent on our own plans but also on other agents, and on other contingencies beyond our control. 233 The danger in this particular tendency lies in making unrealistic plans for the future. Nevertheless, it is not quite clear why Goldie sees this as problematic for one's sense of self. If a person has no confidence in her authorial power over her own life at all, personal development is impossible. On the other hand, Goldie's concern makes sense where extreme cases like Highsmith's heroine are concerned. The heroine imagines that she will be able to carry the children out of a burning house, and it is clear that she endangers them as well as herself by acting on the plot of her fictional self-narrative.
The second fictionalizing tendency is detecting agency in the world where it is not. 234 In works of fiction, events typically do not happen by accident, but because some action caused them, thus adding more coherence to the story. In real life, there is usually less coherence, and if a person tries to create it artificially by imagining action, her story moves away from reality and is no longer a reliable basis for planning. That is a convincing argument. However, it is not clear why this fictionalizing tendency should always lead to unrealistic plans for the future. There may be cases when this tendency actually helps a person to carry on with her life after a tragedy; for instance, a mother whose child dies in infancy might find the thought 'The angels took her away to Heaven' so consoling that she can overcome her grief.
Goldie's third fictionalizing tendency concerns the desire for narrative closure.
He claims that fictional persons have their end in them in the beginning, 235 whereas real lives do not. Persons who suffer pain or loss are often tempted to find an answer to the question 'why did this happen to me', expecting the sequence of events in their lives to be tied up just as neatly as a piece of erotetic fictional narrative that provides answers to the question it raises. This fictionalizing tendency hinders the people who give way to it from accepting their pain and grief and going on with their lives. I endorse this view.
In his fourth fictionalizing tendency, Goldie describes the danger of imposing genre on one's life. 236 Genre assigns certain attributes to a main character, thus flattening the character to a cliché: the heroes are more heroic than in real life, the children more angelic, the clowns funnier. If a person sees herself in a particular role taken from genre literature, she might feel compelled to stick to this role, making decisions which she would not have made otherwise. Goldie claims that this has a bad side and a good side to it. If someone sees herself type-cast as a victim, her actions may be unnecessarily defeatist, and she will never be able to change because the outcome will always confirm her delusion. As a positive example, Goldie cites cases where a person deliberately overestimates her powers in order to deal with a situation, in order to be effectively either verified or falsified in the attempt. If the situation does not develop in the positive sense that the genre role seemed to demand, the person will be able to reflect on why it went wrong and how she should try to change to do better in life. If Highsmith's heroine were to survive after not having saved the children, she might get rid of her delusion and do some real good for other children. As Goldie says, 'seeing oneself in a certain light can help one to stick to one's resolutions, and this can be for the good' 237 . But it is not easy to say under which circumstances the good will be possible. In the case of Highsmith's heroine, the price for her own good would be the lives of two children, and it seems cynical to call it 'good' under these circumstances.

NIA: summing up
In the sections on NIAs, I outlined the most important accounts of narrative identity, both descriptive and normative.
In the section about descriptive accounts, I presented Ricoeur's account as one of the most sophisticated models of narrative identity as a practical identity. His definition of the dialectic between idem (sameness) and ipse (self) accounts for the ongoing struggle of the embodied person to maintain personal unity in her interactions with her social world. After discussing the question of whether narrative identity is a practical identity in connection with the accounts of Ricoeur and Atkins, I showed that there are convincing arguments for the claim that a person has a practical identity which is being developed and re-configured during the course of her life, and that this practical identity is a narrative identity. I criticized the reality-constraint on self-narrative by showing that only a weak version is applicable to strong NIDAs, and that weak NIDAs do not require a reality constraint.
Regarding the role of experiential memory, I contradicted Goldie by demonstrating that you can narrate your life without experiential memories, but you cannot have a full narrative sense of self without them.
Schechtman claims that narrative identity can only provide us with a rich sense of self as far as we still share the values and emotions of our past selves. I came to the conclusion that empathic access needs to cover negative as well as positive strong emotional engagement with one's past.
As a last step in the discussion of descriptive accounts of narrative identity, I In the section on normative accounts of narrative identity, I discussed the accounts of Ricoeur, Atkins and Goldie, with a special emphasis on the problems of self-delusion and 'fictionalising tendencies'. I argued that ongoing self-reflection involved in narrative self-constitution is a necessary, but not a sufficient condition for a person to lead a more unified and fulfilled life.

Overview
In his work Mimesis as Make-Believe, Walton presents a theory of representation that is based on an analogy. He holds that engaging with a work of art is analogous to a child's game of make-believe. I shall discuss this analogy in detail because I shall draw on it in The analogy of fiction reading with a child's game of make-believe, Walton claims, leads to valuable insights into what it means to engage with a work of fiction.
He discovers important parallels: for instance, the use of imagination is required for both activities. He also holds that a child's practice of make-believe continues later in life in interaction with representational works of art. 239 I am now going to outline the main points of his account.
Imagine two children playing a game of 'explorers': in turns, each child draws a line as far away from herself as possible, and the other child has to jump onto this line.
If she does not get both feet on to the line, she falls into an abyss and loses the game.
To play this game successfully, the children need to agree about three aspects of the game: a fictional world, a set of relevant rules, and the props that can be used. In The props that can be used are: a stick that represents a torch to highlight a rock, lines that represent solid rocky ledges, and the unlined ground that represents an abyss.
It is an important feature of games of make-believe that the players themselves are props in the game. In my example, they take the role of explorers and assign attributes to this role. They imagine, de se, that they are skilled mountaineers and can jump across chasms because of their training and experience.
Where, then, are the parallels to fiction reading? When I read the first chapter In my children's games example, it is true in the game world of 'explorers' that each player is an experienced and competitive mountaineer. The proposition that each player is an expert mountaineer is fictional, and the fact that it is fictional is a fictional truth. 243 There is, however, a big difference in the ways that fictional truths are generated in games of make-believe and in engagement with a work of fiction: In a game of make-believe, the players are free to choose their props and the rules under which they operate, as long as they agree about the rules and the props among themselves.
In fiction reading, the chosen work of fiction will act as a prop for the reader's imaginings when she participates in it. In contrast to games of make-believe, this prop already prescribes many of the features that the reader's imagination will work on.
These features include the work's genre, aspects of the fictional world, characteristics of fictional persons and, in most cases, a fictional time setting. Among these, the work's genre is the most important. For instance, The End of the Affair is a novel set in London during the Second World War. A reader who takes it to be a guide to London restaurants, or a science fiction novel about a war with the Martians, does not take part in the game which the novel requires her to imagine. If the reader is completely unaware of the social and historical background of the work of fiction and the intentions of the author, she is not able to engage with the game of make-believe suggested by the work.
Walton says that a fiction reader plays an 'authorized game of make-believe' if her imaginings are in line with the relevant rules given by the work of fiction. If her imaginings run contrary to these rules, she plays an unauthorized game. 244

Generating fictional truths
The generation of fictional truths from a work of fiction depends on the 'principles of generation' that are valid in the 'the relevant social context', 245 and on the reader's grasp of them. Walton points out that a reader can obtain fictional truths directly (primary) or indirectly (implied). 246 In The End of the Affair, it is an implied fictional truth that Maurice's room has a window, although this is not mentioned explicitlythe reader is supposed to imagine this as given because in the novel, it is fictional that a bomb destroyed a tree that had formerly blocked the light in his room. In a work of fiction, the directly generated fictional truths are usually in the minority; the reader is required to fill the gaps of the described fictional world from the implications suggested by the work of fiction.
Is it possible to find a full definition of which principles of generation for implied fictional truths there are, and how they work? Walton investigates two widelyused approaches: the Reality Principle (RP) and the Mutual Belief Principle (MBP).
Walton gives this definition of the RP: 'If p1, …, pn are the propositions whose functionality a representation generates directly, another proposition, q, is fictional in it if, and only if, were it the case that p1, …, pn it would be the case that q' 247 .
If a reader follows the RP in order to fill the gaps in the work of fiction with implied fictional truths, she pretends that everything that is primarily fictional in the work of fiction has implications which can be followed through her own knowledge of the real world, and can thus be verified and filled out with more realistic detail. 248 On the one hand, RP does a good job of explaining why we assume that a society, so the reader following the MBP will accept that proposition q can be derived from proposition p. She will imagine the strange gravitational experiences of the travellers and try to fill out further details that go along with them. If she manages to accept it as fictional, she might also imagine it, accompanied by a feeling of guilt. 252 This strategy is common in sensational fiction or comics, when the author omits the description of a particularly gruesome scene and leaves it to the reader's imagination to picture it, e. g. the way a murder was committed. As Scott McCloud says: 'To kill a man between panels is to condemn him to a thousand deaths.' (McCloud 1993, 69). 253 Walton 1990, 184. 254 Gendler 2010, 180. 255 Walton 2008 As a third possibility, she might accept it as fictional but refuse to imagine it.
This leaves her two courses of action: either she could reject the story completely, or she could follow a convention which holds that we are not obliged to imagine things that are morally repulsive to us, 256 and decide to skip imagining this particular bit of the story.

Fictional worlds
The discussion about the generation of fictional truths highlights an important aspect of fiction reading: it is necessary to distinguish between the fictional world suggested by the work of fiction, the 'work world', and the world imagined by the reader in her engagement with the work of fiction, the 'game world'. 257 The game world contains many fictional truths that were prompted by the reader's own associations and may thus be quite different from the author's work, or the fictional world that is manifested in the work of fiction according to contemporary literary judgment. Every reader, and every author, creates her own appreciator's world when engaging with a work of fiction. The work world, on the other hand, contains the fictional truths that an ideal reader of the work of fiction would generate. 258 If the distinction between the work world and the appreciator's world were not made, it would not be possible to account for the wide range of different responses that readers have to works of fiction. 259

Participating in a work of fiction
As mentioned above, Walton calls the engagement with the fictional world suggested by the work of fiction the 'authorized game' if it follows the rules of generation for fictional truths that are generally accepted for this kind of work. 260 The reader can deviate from the authorized game in two different ways: she can play a game that is not 256 Walton 1990, 155. Modern-day schools and publishers have found a third solution: Passages in works of fiction that might be offensive to politically correct or supposedly vulnerable readers are deleted or rewritten, to the effect that the original work of fiction is lost to the modern reader. 257 Walton 1990, 58. 258 Walton 1990, 59. A discussion of how an ideal reader of a work of fiction could be defined would be interesting, but it is beyond the scope of this project. 259 Walton 1990, 59. 260 Walton 1990, 398. authorized, or she can play the authorized game, but have either imaginings in addition to the game, or fail to imagine some key propositions authorized by the game. If a reader engages with The End of the Affair under the assumption that it is a guide to London restaurants, she plays an unauthorized game. If, on the other hand, she reads it as a psychological novel but during the reading process also indulges in ponderings about her own failures as a Catholic, she plays the authorized game, although the additional imaginingsalbeit triggered by the novel and the protagonists' conflicts with their faithare not part of this game. 261 The phenomenon of imaginings that occur alongside the authorized game often arises when the reader is involved in psychological participation, imagining a character(f) from inside. What happens to a reader when she is 'caught up in a story'? 261 I am going to distinguish between authorized imaginings (work world) and non-authorized imaginings that are nonetheless tied to the work (game world). There is a fuzzy area where non-authorized imaginings are only associations triggered by, but not tied to the work; I am not going to investigate this in my thesis. 262 Walton 1990, 205. 263 Walton 1990, 273.

Quasi-emotions
Walton says that it is fictional that we feel for fictional characters. We have the impression that we care about characters(f), but it is only fictional that we do. We do, however, have real emotions that are triggered by fictionally caring. To clarify his view, he coins the term 'quasi-emotion'. For him, a quasi-emotion contains all the elements of an emotion except the cognitive element and the motivational element. 264 Emotions generated by real-life situations are based on a belief that something is the case which prompts not only a quasi-emotion but also certain motivations. Emotions generated by engagement with a work of fiction, on the other hand, are based on make-believe that something is the case which prompts a quasi-emotion. Hence, the quasi-emotions generated by make-believe in fiction reading do not prompt us to action, the feelings and sensations may have a different intensity, and their duration can be different, either longer or shorter.
Here is an example of quasi-worry in fiction reading to clarify this distinction: In

Criticism of Walton's model and open points
In this section, I am going to address two specific questions that have been raised by own observations of the world to fill out the picture. This simulation process is best described by Walton's concept of role-playing because this indicates that by participating in a work of fiction the reader simulates a person in a certain situation.
This simulation process is not explained by Carroll's notion of entertaining a proposition as unasserted.
The problem with Carroll's account is that it precludes a deeper discussion of the role of the reader. This is because Carroll takes fictional propositions as merely unasserted, as opposed to simulated. If fictional propositions were nothing more than unasserted propositions, it would remain mysterious why they can provoke an emotional reaction in the reader. I am going to discuss this in detail in Part III.

Does the concept of make-believe explain why we value fiction reading?
The second objection to Walton's model is that the notion of fiction reading as makebelieve does not cover everything that we think is important for us when we read fiction. Gibson claims that a make-believe theory represents fiction as nothing more than 'fodder for fantasy'. 277 If we see fiction like that, Gibson

This is an interesting objection because it leads into the discussion of what
'making-believe' actually entails, especially in fiction reading. As a counter-argument to Gibson, I hold that he underestimates the complexity of the make-believe process.
Imagining something means simulating a situation in which a proposition is true although we do not believe it is true. In order to be able to simulate such a situation, however, we need a backdrop of other propositions which we do believe to be true: that London existed in the 1940s, that it was bombed during the Second World War, that lovers are often jealous, that religious belief can make people do strange things, and so on. Make-believe has to operate on elements of our beliefs, and this is why it is interesting to us. If someone told us to imagine that crylophants are crylomaric, for instance, we would not be able to do it because we have no beliefs connected with anything that might be represented by the words 'crylophant' or 'crylomaric'.
Making-believe is not equivalent to idle flights of fancy that are triggered by chance. In Part III, I am going to argue that in order to 'make herself believe' that something is fictional a reader has to actively search for associations within her memories of experiences and images of the world.
Another objection to Gibson' This is why I hold that Gibson is mistaken in his claims: He under-estimates the make-believe process, and he mischaracterises the nature of the knowledge we gain from fiction. In my view, we can advance our knowledge from fiction reading in two ways: by gaining possible insights which we need to verify with reliable sources, and by actively reflecting on human behaviour.
In this section, I addressed two questions that were raised by Walton's critics, both connected with the role of the reader in Walton's account of fiction reading as a game of make-believe.
I discussed Carroll's claim that the analogy to games of make-believe does not reflect the individual experience of a reader, and found that Carroll does not attach enough importance to the role of the reader in the process. In my view, making-believe a fictional story involves the reader's own experiential memories, and thus, the simulation of the story goes a long way beyond Carroll's notion of entertaining a proposition.
I also investigated Gibson's assertion that the make-believe analogy for fiction reading does not cover the recognition of reality in literature, and came to the conclusion that Gibson's understanding of the process of make-believe falls short of what is actually involved. Making-believe a fictional story is not arbitrarythe reader has to find associations for the work of fiction within her own life story and experience.
Here to a definition of emotion as a complex process that has a narrative structure. In chapter 5, I shall outline the account of emotion that I adopt.
I shall present my own account in chapters 6 and 7, leaving comparisons and contrasts until chapter 8.
In chapter 6, I shall investigate the actual reading process. Section 6.1 deals with the degrees of involvement the reader can attain during this activity. I distinguish between three levels of participation: 1) reading 'from the outside' with aesthetic appreciation, 2) participation that includes sympathy with a character(f), 3) participation that includes identifying and empathizing with a character(f) from within. I say that it is possible for reader participation to occur on all three levels.
I am not making a general claim about reader participation. Every reader is different, and some readers may never experience sympathy or empathy with a character(f). The aim of my investigation is to provide a model that accommodates the effects of fiction reading that have been observed in socio-psychological and cognitive research. My assumption is that these apply in some, though not all, cases.
In section 6.2, I define three different kinds of narrative that the fiction reader can develop during the reading process: the other-narrative, the self-in-other-narrative, and the other-in-self-narrative. I shall argue that the generation of these narratives is brought about by a psychological mechanism which I call the 'doppelgänger function'.
I am going to explain the working of the doppelgänger function with some examples from fiction.
My main focus during this and the following discussions is going to be on the reader who identifies with a character(f) from within.
In chapter 7, I am going to investigate the long-lasting effects that fiction reading can have on the reader. What can cause the reader to continue her engagement with a fictional story after the reading process has been completed? I am going to argue that a chain of emotions can evoke both spontaneous and purposeful self-reflection, and that self-reflection can trigger refiguration of the reader's self-narrative.
Finally, I shall summarize the differences between my own account and the two accounts that constitute its basis. In chapter 8, I am going to highlight two gaps in Ricoeur's account of the connection between fiction reading and narrative identity, and show how my account helps to fill them. In addition to that, I shall discuss how 282 See Goldie 2000, 11. 283 There are different views about the duration of an emotion or an emotional episode, but this is just a terminological issue that makes no difference to my argument. There is a prevalent agreement among philosophers that emotions either are or occur as part of a broad, complex response. For a detailed discussion and a literature overview, see Price 2012, 58-60. 284 I shall discuss this in detail in in the course of this chapter.
In the following discussion, I am going to use emotion-related terms in the following way: • When I use the term 'emotion' what I have in mind is the typically complex process that constitutes an emotional episode.
• An emotional episode is a continued emotional response, triggered by an emotional reaction, and united by being about the same theme.
• The target of an emotion is a special case of an intentional object, namely the thing, person or event that an emotion is directed to. For example, if I am afraid of being killed by a bull that is charging towards me, the target of my fear is the bull.
• The focus of an emotion is the background object or state of affairs that constitutes its main underlying concern. 285 In the case of the charging bull, the focus of my emotion is my health.
• An emotional disposition is the tendency to react with a certain emotion under certain conditions.
• An emotional reaction is a short-term emotional response, which might last for a few seconds or a few minutes.
• An emotional experience is the combined phenomenological feeling (including bodily sensations as well as psychological feelings) generated in the course of an emotional reaction or an emotional episode. 286

Emotion as a complex process
In a multifaceted structure, an emotion merges evaluation, psychological and bodily feelings and responses into an intertwined phenomenological experience which feeds upon itself. The target of the emotion becomes the focus of increased attention. As The duration of an emotional experience is largely dependent on the type of emotion, in various ways. For instance, a self-directed emotional experience is likely to have more facets and a more intricate structure than an other-directed emotional experience because it has more and richer material to feed upon. A negative emotional experience typically has a stronger impact than a positive emotional experience; empirical evidence has shown that the focus of a negative emotion is usually narrower, and thus deeper, than that of a positive emotion, provided that these emotions have the same intensity. 289 I shall return to these thoughts in chapter 7.  Regarding the reader's activities during the reading process, I shall define the concepts of other-narrative, self-narrative, self-in-other-narrative and other-in-selfnarrative in chapter 6, and discuss the connections between them using representative examples from literature.

The intentionality of an emotion
An emotion is characteristically directed towards a target. In most cases, it is easy to identify the target of an emotion. If I see a bull charging towards me from a short distance, my initial intuitive evaluation identifies the most important part of the situation: I am in danger, and my emotional reaction is fear.
The target of this fear is the bull which is about to kill me within the next few seconds, and I am urged to immediate action, in this case to run for cover. This is accompanied retrospect that the cause of the emotion was based on false beliefs; and third, finding in retrospect that although the cause of the emotion was appropriate for an emotion of this kind, the intensity of the resulting emotion was out of proportion. I am going to present an example for the first of these cases, as I am going to refer to it in chapter 6 when I investigate the reader's role when engaged with a work of fiction. 292 Goldie 2000, 181. 293 Regarding the problem of "feeling" for a fictional character, see Walton 1990, 196,   In such a case, phenomenology (in this case the angry feeling) is borrowed, or rather shared, between different narratives. We know what triggered them, and we know which target belongs to which emotion, but we are no longer consciously aware of it while engaged in the experience. Blustering against my boss who makes critical remarks about my work seems natural at the time. Afterwards, however, I realize that I became over-excited, and I start to wonder why. If the merging of an existing emotional disposition, or a previous emotion, with a current emotion is strongbecause the previous emotion was strongthis enhances the subsequent need for reflection and 'untangling' the stories, thereby re-assessing them.
I am going to discuss this confusion of feeling towards an object's characteristics and feeling towards an object in the following chapters about fiction reading. In particular, I am going to examine the merging of emotions which are directed at a character(f) and emotions which are directed at the reader's self.

The development of an emotional experience
When we have an emotional experience, our feelings are directed towards the target of the emotion. This is, however, not a one-sided transfer: as a consequence of the feeling, our perception of the object becomes more discerning in a way that is likely to enhance its emotional impact. The longer the emotional experience lasts, the more intense is it likely to become. The emotion tends to return to its object, feeding both on the object and on itself. During this 'feeding' process, we are stimulated to gather information about the object that is in keeping with the nature of the emotion as it started out. For instance, if a boy falls in love with a girl, he is bound to notice all of her features that can enhance love and admiration, like beautiful hair, bright eyes, or a gentle nature. His emotion motivates him to discover more about her, and he uses his imagination as a tool to find out more possible aspects of her. Finally, the information-feeding becomes an

Levels of participation
In

Why focus on fictional characters?
Why can the attitude of the reader towards a character(f) be important for the analysis of the reading process? One might object to this choice of focus by pointing out that people sometimes read fiction because they appreciate the aesthetic pleasure they get out of a finely crafted text, or because they want to be distracted by a thrilling plot, or to 295 Williams 2012. 296 Walton 1990, 273. delve into a 'period atmosphere'. 297 Fiction reading is not just about meeting fictional people, it is more complex than that. If the reading process is to be successful, the language and the imagery of the text have to capture the attention of the reader; the focus of the reader on a character(f) is not necessary for a successful reading experience. There are works of fiction which do not even contain fictional persons that are represented in any depth, as for instance Milorad Pavic's novel Dictionary of the Khazars 298 . Still, it is possible for us to engage with such a work of fiction in a way that we find satisfactory. So why insist on the importance of characters(f)?
Human beings characteristically look out for an 'other' in any area of life of which they become aware. As discussed in section 2.4 in connection with narrative identity accounts, a convincing explanation for this kind of behaviour is the typical development of selfhood in a young child: a human being is able to recognize herself as distinct from others when she has experienced a relationship with an 'other' who is important to her. 299 This taught pattern of defining oneself by looking at others motivates a fiction reader to scan the work for an entity with human features whom she can either identify with or distance herself from. The fictional entity does not need to be represented as human being; it may have humanoid features in the guise of an animal, a plant, a machine, or an extra-terrestrial.
In Dictionary of the Khazars, for example, there are no prominent characters(f).
But even in such a case, the reader will discover the attributes of a particular person.
Taking in the style and the perspectives which the author used in the novel, and interpreting it in her own way, the reader speculates about the author's character and personal life. The reader will usually be tempted to do this because the urge to look for an 'other' is strong enough to motivate her. The template on which the reader works to find this 'other' is the implied author. As Ricoeur says in his account of fiction reading, there is typically a dialectic between the imaginings mandated by the work of fiction and the expectations of the reader where the 'implied author takes the initiative in the show of strength underlying the relation between writing and reading'. 300 In every work of fiction, the implied author is defined by the choices made and representations created by the real author; the implied author is designed by the real 297 In Samuel Johnson's opinion, 'It is difficult to enumerate the several motives which procure to books the honour of perusal: spite, vanity, and curiosity, hope and fear, love and hatred, every passion which incites to any other action, serves at one time or other to stimulate a reader.' (Johnson 1754). 298 Pavic 1989. 299 Atkins 2008, 46. 300 Ricoeur 1985, 160. author, regardless of whether the real author intended this or not. As W. C. Booth puts it, 'though the author can to some extent choose his disguises, he can never choose to disappear'. 301 The reader, however, creates her own version of this implied author -I shall call the reader's version the 'inferred author' from now onby applying her own knowledge of the world and her preferences to the text. During the reading process, the reader discovers the inferred author to have certain attitudes and preferences, for instance an ironic stance towards gender questions, which the reader might endorse or reject. The reader interprets the inferred author's point of view and, from the snippets of information given, immediately extrapolates these data into a guess about what kind of person this inferred author might be, and what her own attitudes towards this hypothetical person might be. It is important to note that the inferred author is not part of the fictional storyhe is, however, part of the story the reader makes up in parallel to following the fictional story. One might call him a meta-fictional charactera character(mf). 302 The reader interacts with him as the story develops, agreeing or disagreeing, being disgusted with him or awed by his powers. In general, the reader tries to identify with the person she takes the inferred author to be. The reason for this is that by reading the book, the fiction reader has already given the inferred author partial control over the direction of her imagination, thus creating a relationship of trust from the reader's side. I agree here with Gregory Currie who says that if the reader adopts the framework and point of view provided by the author, the reader has an experience of sharing that depends on psychologically important mechanisms of joint attention and imitation. 303

Why distinguish between three levels of participation?
My interest in the reader's relation with characters(f) makes it useful to classify the reader's engagement according to her degree of involvement with a character(f). Here, my approach differs slightly from Walton's: His emphasis is on the difference between appreciation of a work of fiction with participation and appreciation without participation. He holds that participation involves taking part in the game of make- 301 Booth 1983, 20. 302 It is possible to imagine other characters(mf) besides the inferred authorfor instance the inferred librarian who chose the book for the lending library, or the inferred original purchaser in case of a secondhand book (most fascinating when they leave marks or comments in the book). Investigating this is, however, outside the scope of my work. 303 Currie 2010, 87/88. believe which the work of fiction mandates, where the reader becomes a reflexive prop who generates fictional truths about herself and gets caught up in the work of fiction. In appreciation without participation, he says, the reader is usually reminded of the falsity of fictional propositions and thus unable to take part in the game and imagine it from inside. 304 I agree with his view of fiction reading as a game of make-believe, but I prefer to take a wider view of participation than Walton does. As an inferred author can always be discovered in a work of fiction (regardless of whether the actual author intended this or not), I hold that participation takes place on two levels: first, the reader negotiates with the inferred author about how and to what extent she is going to accept the given text, and second, the reader enters the fictional world offered in the text. Even what Walton calls 'appreciation without participation' has a playful aspect which involves an interaction between the reader and the inferred author. The reader creates a story of her own in which she has an ongoing dialogue with the inferred author, and on this level the inferred author becomes a character(mf) with whom the reader tries to identify during her appreciation of a work of fiction. This is why I think that all of my three levels of reader involvement (appreciative, sympathetic, or empathetic, from a first person point of view) can be termed 'participation'.
In order to participate in a work of fiction, a reader has to use her imagination. I participate in the work of fiction on more than one level simultaneously, on the level of empathy as well as of appreciation. Goldie's concept of 'in-his-shoes imagining' is an attempt to span several levels with one definition. He says that for 'in-his-shoes imagining', one needs to merge two perspectives: the internal, to imagine the other person's goals and fears, and the external, to judge the differences of characterisation between the other and one's own self. 309 Thus, his definition can cover the scale of feelings for a character(f) from appreciation to empathy. This is an interesting approach, but I do not adopt it because I shall argue that the level of empathy is most likely to trigger strong emotions in the reader, and therefore, I am going to investigate it 308 Austen 2008,1. 309 Goldie 2000, 199. separately. In my view, these levels of participation should not be conflated; they can exist independently of each other. All possible reactions (with the exception of an immediate closing of the book after the first sentence) show that reader interacts with the text, taking on the challenges the inferred author created, and building expectations based on both the story and the reader's own life experiences. When the reading process starts, the world of the text and the world of the reader intersect, even if the reader keeps a distance from the text by remaining aware of its artificiality. The reader may view the text as contrived, but the inferred author has a real presence that cannot be denied.

Sympathy
Sympathy is different from appreciation because it requires that I take a personal interest not only in the text, but also in a character(f)'s plight. It takes three steps to develop sympathy towards a character(f): first, just as in appreciation, I need to understand the meaning of the text regarding the character's situation in the story (as far as the narrator has unveiled it to me). wrong that has finally been righted. As this dramatic feature is strongly emphasized in the text, the reader is stuck with the image that forced itself on her, and there is no motivation for her to probe into the character(f)'s feelings from the inside. The buttonpushing effect is so strong that it even tends to blind the reader to subtleties of the text. 311 Dickens 2008b. As if one had inadvertently looked into a dazzling light, the outlines of the obvious drama seem to linger for a while, obscuring everything else. Depending on the level of reflection the reader likes to adopt, she may either continue to enjoy her feeling of happy fulfilment, or she may be ashamed of her sentimental outburst (and subsequently become sentimental again on a more subtle level when she relishes the funny antics of the characters(f)).
At the upper end of sympathy-induced emotions, we typically find more subtle feelings like compassion or sorrow. In Our Mutual Friend, an example of an emotion that requires some reflection is that evoked by the relationship between Riah and Fledgeby. The reader feels pity for Riah because he is oppressed and unjustly accused of being greedy and cruel. This is not an emotion that hits the reader immediately; at first, one is confused as to why Riah bears this treatment so patiently, without protesting against it. But then, one separates the mystery of Riah's character from his piteous situation and feels for him, from the outside, because he has the role of the unjustified sufferer.
Sympathy towards characters ( there has to be an initial spark of interest. As soon as the reader is motivated to simulate a character(f)'s inner life, she starts to fill in the gaps in the picture with 315 Actually, such an appeal may have the opposite effect: mentally agreeing with the author's appeal can make the reader feel they have already done something worthwhile, thus being freed of the obligation to do more. 316 I shall investigate this more closely in chapter 7. 317 As Ricoeur puts it, 'To follow a story is to actualize it by reading it.' (Ricoeur 1984, 76). 318 Feagin 1988, 494.   forms of empathy and identification, but that one of these types of engagement is always distinguishable as the driving force. I am going to refer to this in the next section when I investigate the reader's role in the fictional world.

Ricoeur's account of participation in the reading process
Ricoeur distinguishes between two levels of participation in fiction reading: aesthetic enjoyment and identification with a character(f) from within. He does not use the term 'level', though; presumably because it would suggest a stability within the reading process which he denies. For Ricoeur, the reader's engagement with the text is an ongoing struggle between the inferred author and the reader, between the strategy of seduction on the inferred author's side and the strategy of suspicion on the reader's side. 328 He describes fiction as an instrument we use in order to understand the connection between action and agent, as 'an immense laboratory for thought experiments in which this connection is submitted to an endless number of imaginative variations.' 329 He does not see these imaginative variations as an impersonal game, but as a form of deep reader involvement in the work of fiction: 'The reader belongs to both the experiential horizon of the work of fiction imaginatively, and to the horizon of his action concretely '. 330 There is, however, one area which Ricoeur does not investigate in detail: the significance of the reader's emotional engagement during her fiction-related imaginings.
I am going address this gap in chapter 8. In the following section, I shall explore the role of the reader who feels empathy with a character(f) in more detail. In particular, I am going to examine the kinds of narrative the reader develops while empathizing with a character(f). I am going to argue that the role of the reader can be best understood by examining the emotions that stimulate the reader into creating layers of narrative which have different connections with the reader's self.

The reader in two worldsreal and fictional
In I hold that they are important because any of them, or any combination of them, can feature as protagonists in the narratives the reader weaves during fiction reading with empathy. Four kinds of narrative might become active in the fiction reading process: self-narrative, other-narrative, self-in-other-narrative, and other-in-selfnarrative. Working out the differences between them, I shall argue that self-narrative is largely suspended during fiction reading. We relinquish our ego and our self-narrative and enter the story of a character(f), the other-narrative. Typically, the other-narrative triggers a self-in-other-narrative when the reader starts to imagine the character(f) from within. However, when the reader feels that imagining the character(f)'s actions pushes her out the realm of her own ethical aims, the self-in-other-narrative eventually turns into an other-in-self-narrative. There is a shifting of protagonists from the character(f) to the reader and vice versa that continues throughout the reading process. As I shall explain, this transferral of roles is managed by the doppelgänger function that accompanies the on-going dialectic between the text (represented by the character(f)) and the reader. Thus, the self-in-other-narrative and the other-in-self-narrative can be viewed as not fundamentally different, but as different outcomes of the doppelgänger function.

Different beings
Before going into the details of the doppelgänger function, I would like to introduce the beings that can be involved in the reader's engagement with a work of fiction.
The idem 332 of the reader contributes to the reading process in a variety of ways.
It includes the capacity to manage the physical part (for instance holding the book and reading the words), the reader's character, habits and preferences ( The reader's self (or ipse 334 ) is the lived identity of the reader in the real world.
During the reading process, the self is mainly restricted to two phenomenological activities: the experience of imagining the fictional world and the characters(f), and the experience of the emotions that are prompted by the reading.
In defining a character (

Other-narrative
The other-narrative is the story of the character(f) as set up in the work of fiction and understood by the reader. My assumption is that every work of fiction contains an entity which can be interpreted as a character(f) (it could be a fictional person, a fictional animal, a fictional machine, or even a fictional landscape), and that a narrative involving the character(f) as an agent (or a sufferer) is constructed in the work of fiction, some of it explicitly, and a large part of it implicitly.
As discussed in section 6.1, the engagement with a work of fiction may also induce the reader to imagine a character(mf), usually the inferred author. As the character(mf) is not part of the fictional narrative itself, the reader does not have an actual story to feed her imagination on, only a bundle of inferred attitudes, preferences, or passions. For this reason, I hold that although the reader can also build an other-narrative for a character(mf), this narrative will probably be too flimsy to trigger identification from within, and thus cannot be subject to the doppelgänger function.
Hence, my following discussion only concerns the other-narrative for a character(f).
As many philosophers have successfully argued, 339 the story contained in a work of fiction needs the reception by a reader to be fully configured. Ricoeur says 'To follow a story is to actualize it by reading it'. 340 Each reader creates her own version of the plot and of the characters(f) involved, adding her own knowledge and experience of the world while missing meanings and associations in the text which are foreign to her.
The reading process involves a struggle between the reader and the text, resulting in what Ricoeur describes as the 'three dialectics of reading': 341 First, the reader defends her own expectations against the deceptive strategies used in the text; second, the reader tries to configure the implied but unwritten parts of the text; third, the reader sets the cultural paradigms she is familiar with against those which are either confirmed or challenged by the text. As a result of these activities, the reader continues to build her own narrative of the character(f) which is bound to change as the reading progresses.
Does the reader need to feel sympathy for a character(f) to be able to create this character(f)'s other-narrative? It seems plausible that I need to have a minimal feeling of human companionship with a character(f) to be motivated to follow her story.
Sympathy would include an interest that the character(f) should flourish, but I hold that one can be motivated to engage with a fictional story without wanting the characters ( , 77. 341 Ricoeur 1985. 342 I am not following up this distinction because this would go beyond the scope of my work.

Self-in-other-narrative
When a reader empathizes with a character(f), she enters the fictional story and the fictional world -up to a point. Ricoeur says 'reading itself already is a way of living in the fictitious universe of the work; in this sense we can already say that stories are told but also lived in the imaginary mode'. 343 What exactly does this 'way of living' in the imaginary mode entail; can we define it? Some of the limitations the reader encounters while imagining the character(f) from within are straightforward. First of all, the reader cannot act in the fictional world (although in cliff-hanger scenes, for instance, the reader might imitate the action of the character(f) by grabbing the table or clenching her fist).
Second, and following from this, the reader cannot suffer any real-life consequences from what happens in the fictional world. The climber in the story may fall to his death; although I suffered inside him until he died, I continue to live. In spite of this divergence, some things happen in parallel during the reading process: the fictional climber is released from his life, and I am also released from the fictional climber's life.
The other-narrative of the fictional climber comes to an end simultaneously with my self-in-other-narrative. If the story continues after this, I see the deceased climber from the outside.
This is an indication that there can be many switches between the other-narrative and the self-in-other-narrative. To understand how this develops, and in which respect these two narratives mirror each other, we need to get back to the question of how empathy is generated. The first step towards empathy is typically that the reader takes an interest in the character(f) because the character(f)'s fears and desires are similar to her own. She can proceed to imagine the character(f) from inside if she is able to identify the shape of the character(f)'s conflict 344 as something she knows from her own whereas I am a middle-aged woman sitting in a library in Zürich in 2018. What draws me into him is our common terror of near-drowning and existential annihilation.
If I manage to feel empathy with a character(f), my imagination has already started to integrate me into the fictional story in a meaningful way: I let the part of my self that is affected by the conflict take part in the narrative to such an extent that I can 349 Ricoeur 1988,168. 350 Steinbeck 2001. 351 Steinbeck 2001 Because it is unexpected, and because it interrupts the reading process, the otherin-self-narrative is usually put on hold deliberately by the reader as soon as she becomes aware of this, to be evaluated in a more organized and differently structured way after the reading process.

The doppelgänger-function
The The search for similarities between the character(f) and the reader appears to be a fundamental feature of fiction reading with empathy. This does not seem surprising becauseas already discussed in section 2.4.1it is an integral part of personal development, starting in early childhood, that a person learns to see herself as distinct from others, and to compare herself to others. Because a person's position in her social nexus is important for her self-esteem, a part of her is trained to keep on scanning her environment for other persons to relate herself to, although she may not be conscious of this. 352 Social psychologists have been investigating the question of how and why people perform social comparisons. Recent research has shown that people often compare themselves to others spontaneously and unreflectively, and that this occurs on a regular basis because it is an efficient way of gaining self-knowledge. 353 352 See for instance Corcoran et al. 2011, 119-120 andGarcia et al. 2013, p. 634 for a literature overview on psychological research on the topic of self-evaluation by social comparison. 353 Corcoran / Mussweiler 2010, 79 andTaylor 1981, 195. During the reading process, the reader's spontaneous social comparisonin my terms, the doppelgänger functionis directed towards the characters(f), in particular towards the ones the reader imagines from inside. I call this procedure the doppelgänger function of narrative generation because it becomes stronger and more effective whenever it discovers a large overlap between the character(f)'s attributes and the reader's own. My hypothesis, supported by an inference to the best explanation, is that the doppelgänger function is a general human capability. It works as a monitor, constantly switched on, alerted to find relations between the reader and the character(f).
On a meta-level, it steers my attention, influencing the direction of my imagination.
The minimal likeness I need to discover to empathize with a character(f) is a similarity in a disposition or character trait that is important to me; at the maximum stage, I would find out that the narrative is really about myself. As soon as I have The following diagram summarizes the relation between the reader's self and the fictional other during empathic fiction reading:

Stoner
In this section, I am going to give an example of how the doppelgänger function prompts me as a reader to generate fictional narratives on different levels, with different protagonists. The work of fiction I choose to illustrate a possible reading experience is John William's novel Stoner. I selected this novel because it is widely acknowledged to create a particularly vivid reading experience, evoking readers' emotions, so the general processes I describe here are not likely to be idiosyncratic but shared by many.
The novel's protagonist, William Stoner, is presented to the reader slowly. We have to build the other-narrative (in this case: the Stoner-narrative) step by step. This is in keeping with Stoner's character and development: He is introduced to the reader as an introverted man who takes a long time to confront his own fears and desires, and who is also slow in recognizing the driving emotions of the people close to him. This

Empathy with a fictional other during fiction reading Narrative Relationship
Other-narrative Self-in-other-narrative Other-in-self-narrative Self Other

Perception and identification
Other Other Domination Domination makes it easy for a reader to imagine him from inside: we all have moments of timidity and denial, and Stoner lives and suffers them openly for us.
A particularly moving part of the book is the scene where Stoner furnishes his office, his sanctuary from a difficult world.
The description of Stoner sanding a wooden board in his study is an image of what the book does to the reader: the old caked paint is rubbed off painfully with a rough surface so that the structure below begins to shine. In the same way, the reader begins to appreciate a character(f) who was initially presented as a nondescript, rather awkward man who dies unlamented and unremembered. The sanding of the board is a beautiful image which has many associated meanings, and it leads the reader to appreciate the work of fiction aesthetically.
Simultaneously, this scene enables the reader to identify with Stoner from the inside in more depth. We already feel for him because in his relations with other people in the novel, he is generally treated unfairly. The others misjudge him, or are just not interested in him. We have all been in situations like that, and can thus contribute experiential memories to make Stoner's conflicts more vivid in our imagination. When we read the description of how the wood begins to show its structure under Stoner's workmanship, starting to glow, we are motivated to recall memories of occasions where we made a rough thing shine through the labour of our own handsthis might be the stripping of floorboards, the cleaning of an old painting, anything that involved a similar process of successfully making a physical object more attractive through our own manual labour. These memories bring their own mini-stories into the narrative as we read: for example of how our arms hurt after the room was finished, and none of our friends understood our elation at the job well done; or of how the wood-dust tickled our nose, or how an artistic mural was revealed when we took the paper off the wall. All of these stories run on together with the Stoner-narrative we are building, producing a new self-in-Stoner-narrative. The glow of the remembered experience, enhanced by the Stoner story, is rekindled in our breast. Through our near-doppelgänger Stoner, we are able to relive our pleasant or unpleasant memories, enhanced by new attributes from the fictional scene. Although caught up in the novel, we are happy about pleasing sensual experiences that happened to ourselves, or unhappy about conspiracies that were directed against ourselves. We are caught up in a self-in-Stoner-narrative.
As the sanding-scene continues, we become awed by Stoner's quiet persistence in his work and in his way of consistently 'making' himself into the man he feels he can be, and should be. There is something inspiring about following Stoner's actions: we get the impression that Stoner is a man who will always be true to himself, as best he can.
Influenced by his example, we begin to feel guilty about the petty worries we often feel, or felt at the time of our recalled memories. We felt slighted when our friends did not praise the cleaned paintingand we suspect that Stoner would not have minded, being content with the result of his work. As we introduce an imaginary Stoner into our memory of restoring the painting, we generate a Stoner-in-self-narrative. We are glad to have Stoner as our near-doppelgänger, showing us the way we should go. The Stonerin-self-narrative will not last long, however, as it keeps us from reading on, but for the time being, it generates another emotion in us: the joy of having discovered a fellowfeeling with someone else, feeling 'worthy of being human' based on our success of becoming aware of the importance and difficulty of the human situation.
As the example illustrates, the doppelgänger-function manipulates the way our reading experience develops, and it leads us to create both self-in-other-narratives and other-in-self-narratives as variations of the other-narrative that we need to build in order to understand the story. In the other-in-self-narrative, we simulate the emotions of the fictional 'other', but in addition, we have genuine emotions which are about situations in which we found ourselves in the past, through our recalled experiential memories.
These are often self-conscious emotions like shame or pride. The other-in-self-narrative can lead to a slightly different kind of genuine emotion: to a fleeting elation about a lesson learned or a connection understood, or to a feeling of guilt about an unacknowledged desire to do something that seems wrong to us. I shall explore these emotions in more detail in section 7.1.
This sums up the investigation of a kind of experience the reader can have while the reading still continues. As we saw, the reading experience includes simulated emotions (from 'inside' a character(f), as part of empathizing with this character) as well as genuinely felt emotions (from inside the reader's self).
In the next chapter, I shall investigate how the reading process affects the reader beyond the actual reading process.
In this chapter, I deal with the question of why and how fiction reading can invoke subsequent self-reflection. This is an important part of my inquiry because it addresses one of the gaps in Ricoeur's account. I am going to outline my own views on selfreflection in this chapter, leaving the dispute with Ricoeur's views to chapter 8.
The discussion of the influence of fiction on the reader's mind and morals dates back to Plato, who famously stipulated in Book II of The Republic that only ethically uplifting stories should be told to the guardians, so that they cannot be confused in their moral judgments. 354 In more recent discussions, there are two bones of contention, one of them quite similar to Plato's worry. I am looking at two issues here. First: does fiction reading typically lead to conscious and voluntary self-reflection? Second: is reading-induced self-reflection always good for us?
The first (Platonic) question is this: does fiction cause the reader to reflect consciously on the fictional content, and her own emotional reaction to it, or is there an insidious unconscious influence against which the reader is powerless? Typically, a mixture of both will occur, but which of these influences persists after the reading process? My aim is to find out more about the long-term effects of reading fiction when a reader imagines a character(f) from within, especially its potential of changing the reader's self.
There are a number of articles on empirical psychological research which suggest that readers who enjoy being 'transported' into a fictional story tend to be influenced by the preferences of protagonists whom they like, without being aware of this. 355 These experiments are typically based on the readers' self-assessment immediately after the reading process. As already discussed in section 6.1, their results have strengthened my resolve to investigate the relation between fiction reading and the reader's self during the reading process. However, it remains an open question whether the short-term results gained in this way can be extrapolated into the long term, and the studies themselves state that there is no reliable basis for such an extrapolation. This does represent a challenge to my account; however, it is far from decisive because 354 For instance on Homer's tales about the gods: 'These are the kind of sentiments about the gods which will arouse our anger; and he who utters them will be refused a chorus; neither will we allow teachers to make use of them in the instruction of the young, meaning, as we do, that our guardians, as far as men can be, should be true worshippers of the gods and like them. ' Plato,Republic,II,383c. 355 Green / Dill 2013, 1.
currently, there are no representative empirical studies that examine the longer-term effects that my account is concerned with. So for that reason, I am going to set the question aside for the purposes of this discussion. What I am doing here is building a plausible model that requires empirical testing.
As a second topic, I shall discuss the positive or negative effects of fiction reading that can result from conscious reflection on the reading process. Do stories help me to reflect on my life, and thus find an inner equilibrium that supports me in leading a unified life? Normative accounts of narrative identity (NINAs) say that it does, provided that the reality constraint and the articulation constraint are met. 356 If this were the case, all fiction readers (except the seriously deluded ones) would be happy people who are content with the life they lead. When I look at the discontented lives of some of my fiction-reading acquaintances, it is obvious to me that fiction reading does not always have this effect. It can even have dangerous consequences. As I discussed earlier, Peter Goldie identifies fictionalizing tendencies that can jeopardize a person's self-narrative, and this kind of fictionalizing can also be triggered when a reader engages with a fictional story. 357 I am going to investigate on which of the three narrative levels I defined (other-narrative, self-in-other-narrative, other-in-self-narrative) fictionalizing tendencies can occur.
Other philosophers tend to place more importance on the positive aspects of a reader's self reflection. Paul Ricoeur holds that 'we need the help of fiction to organize life retrospectively, 358 and, in the same vein, Martha Nussbaum says: 'novels can be a school for moral sentiments'. 359 I am going to refer to these thoughts in my discussion of the possible influence of fiction reading on self-reflection. Both Ricoeur and Nussbaum agree that fiction reading can be used as a 'thought laboratory' that allows us try out ways of acting / reacting to certain events in theory without having to take the risk of adverse reactions within the real world, thus inducing us to learn more about ethical behaviour. Jenefer Robinson argues that the emotions we experience during fiction reading lead us to make affective appraisals of situations and characters from a fresh point of view, reflect on these assessments, and thus undergo a 'sentimental education' which makes us more discerning in our perception of other people. 360 356 See discussion in section 2.4.2. 357 See Goldie 2012, 161-172, and the discussion in section 2.4. 358 Ricoeur 1992, 162. 359 Nussbaum 1990, 240. 360 Robinson 2006, 157. Starting with these ideas, I am going to investigate the influences of fiction reading on the reader that go beyond the actual reading process. I shall argue that emotional experiences invoked by fiction reading are just the starting point of a chain of emotional episodes which can culminate in stretches of self-reflection and subsequent gradual adjustments of the reader's self-narrative. In contrast to Nussbaum and Ricoeur, I am going to claim that the success of self-reflection depends on the absence of bias in the self-narrative on which it is built. I define a self-narrative as 'biased' if the role the self plays in her self-narrative diverges from the role the self plays, or is capable of playing, in her real-life social environment. As discussed in section 2.4, I hold that a person can maintain their self-narrative even if it is based on factual errors, except in the most extreme cases. In a biased self-narrative, the reader bases her self-reflection on a weak or a purely imaginary aspect of her personality, thus giving it more than its due importance. I am going to argue that there is at least one drawback to a biased selfnarrative: self-reflection may not lead to an improved unity of self.
My focus will be on three areas: the characterization of reading-invoked emotions as real-life experiences, the nature of self-reflections based on these emotions, and the effect of these reflections on the reader's self-narrative.
As we saw in the previous chapter, fiction reading can induce emotions in the reader at various stages of her engagement with the text. In some cases, these emotions give rise to spontaneous self-reflection which does not stop when the reader puts down the book.
Building on the discussion in section 6.1 on levels of participation, I hold that it is important to distinguish between sympathy and empathy when investigating a fiction reader's emotional responses. When we sympathize with a character(f), we feel compassion from an external point of view. Sympathy typically calls forth real-life emotions directed at the type of person, group, or role that the character(f) stands for. In contrast to this, empathy triggers simulated emotions from 'inside' the character(f) as well as real-life emotions directed towards the reader herself. As discussed in section 6.1, the recall of experiential memories can invoke self-directed emotions in the reader.
During the reading process, sympathy and empathy can fluctuate and interlace, but typically they belong to different types of reader-generated narrativesympathy occurs more frequently during the observance of the other-narrative, whereas empathy is characteristic for the self-in-other-and the other-in-self-narrative. 361 I claim that real-life emotions which are invoked during fiction reading can continue, and continue in parallel, beyond the actual reading process if they are strong enough, and if they of the right kind. My second claim is that if real-life emotions were generated during fiction reading at all, these emotions typically start (or continue) a process of reflection and self-refiguration. Depending on the quality of the reader's selfnarrative, this self-refiguration can either help or hinder the reader in her quest to maintain a unified self.
This chapter outlines the claims of my own account of the relations between narrative identity and fiction reading.

Emotions from fiction reading
In this section, I examine if and how emotions that were triggered by fiction reading can persevere and call forth additional emotions when the reader puts down the book and carries on with her 'real' life in the real word. Again, I would like to stress that I am not making a general claim about all readers and all textsin some readers, fiction reading may never induce emotions.
As shown in the section 6.2, emotions can be invoked at different levels of participation during the reading process. Some types of emotion are more likely to persist after the actual reading process has been finished than others, and I am going to investigate how far they can have an influence on the reader's self-reflection.
I shall distinguish between different types of emotion that can be triggered by fiction reading, putting a special focus on self-conscious emotions and the way they can develop from the other-narrative, the self-in-other-narrative and the other-in-selfnarrative. I hold that all of these emotions are real-life emotions; they are not imagined, but they affect the reader directly.
In the context of fiction-reading, I understand self-conscious real-life emotions to be emotions whose focus is the reader's lived self. Self-conscious emotions come into being because we are aware of other people's reactions to us. Our self-conscious emotions are based on our own judgment of our self, evaluating our situation, our 361 See definitions in section 6.2.
actions and our feelings as either failing to reach a given goal, or exceeding it. 362 The goals that serve as guidelines for such a judgment can be derived from, and motivated by, either the practiced rules of our social nexus, or our own internalised value system. 363 Self-conscious emotions like guilt, shame, embarrassment or pride are related to our self-esteem. As Kristjánsson has argued, self-esteem is not necessarily global, but it can also be 'domain-specific', 364 concerning only certain aspects of the self. When the aspects of our self on which we want to build our self-esteem suddenly appear to diverge from the aspects we defined in our self-narrative, the practical identity we rely on in daily life can get out of balance.
Under which circumstances are such self-conscious real-life emotions induced by fiction reading, and how do they influence my motivation for self-reflection?

Experience of 'self-conscious' emotions
Empirical psychological research has shown that self-conscious emotions like shame, guilt, embarrassment or pride have a particularly high influence on people's motivation to change their lives, or their life story, much higher than non-self-conscious emotions like fear, disgust, joy or sadness. 365 Self-conscious emotions can be distinguished from non-self-conscious emotions as being secondary in the sense that they involve an appraisal of how a situation affects the self.
The influence of self-conscious emotions on a tendency for subsequent selfreflection is widely corroborated by empirical research. As Tracy and Robins point out, self-conscious emotions 'require self-awareness and self-representation' 366 . Selfreflection has already started when these emotions occur, and it does not stop with their emergence. Empirical evidence has shown that people tend to experience self-conscious emotions when they become aware that they have not lived up to their own standards in some way. 367 During these empirical tests, the participants are confronted with real or fictional cases in which people have reason to feel self-conscious emotions, and they are required to imagine themselves in these situations. 362 Lewis 2011, 2. 363 Kristjánsson 2010, 127. 364 Kristjánsson 2010, 112. 365 Lickel et al. 2014. 366  In my investigation, I want to explore under which conditions fiction reading can invoke self-conscious emotions, and subsequently cause a process of self-reflection which may lead to self-revaluation.
In general, self-conscious emotions can arise when the subject either fails to conform to social or personal moral standards, or manages to rise above them. Moral emotions in general can be directed against one's own behaviour as well as against the behaviour of others, so self-conscious emotions form a special class of moral emotions.
Self-conscious emotions typically create a feeling of imbalance in the emoter, which in its turn motivates her to 'set things right' in order to regain her feeling of personal unity and self-esteem, and to fit the new situation into her life-story. With negative emotions like guilt or shame, the 'setting right' may involve getting rid of bad habits or attitudes, whereas with positive emotions like pride, it may involve getting rid of self-imposed restrictions like shyness. This 'setting things right' usually involves refiguring her self-narrative, either for past events or future plans. The directions the 368 Fessler 2007, 187. 369 I assume here that the reader does not use electronic books which collect statistical figures about her reading, or if she does, that she is not aware of that.
refiguring will take depend on the bias (or genre) that accompanies her self-refiguration.
I shall discuss the problem of bias and genre later in section 'Self-examination'.
In the current section, I shall start by examining how, and why, the fiction reading experience can have a different influence on a person's self-conscious emotions than experiences in real life. unrestrained and risk-free use of imagination, free from social judgment and real-life passions; and motivation to imagine a character from inside. I am going to have a look at these particular attributes of fiction reading (detachment, risk-free imagination, motivation to empathize) separately.

Detachment
The first difference that comes to mind is that in real life, self-conscious emotions are often evoked in situations which require swift action and leave little room for selfreflection. If we are caught in an embarrassing situation, we have to defend ourselves as soon as possible; if we receive a prize, we have to step up and deliver a speech. Even if the real-life situation does not necessitate an immediate response, there is usually a sequence of parallel events going on which needs our constant attention, so the opportunity for paying attention to the self-conscious emotion is limited. By the time we come to reflect on it, other emotions might already overlap it. In fiction reading, however, we shut ourselves off from the outside world, motivated to concentrate on a story without stopping in between to attend to other things. As fiction readers with participation, we are 'lost' in the fictional story. Self-conscious emotions have room to grow on us, gaining weight that carries them beyond the reading process into a process of self-reflection that continues afterwards.
It is therefore plausible that self-conscious emotions that are aroused during fiction reading may grow into self-reflection at leisure, in contrast to self-conscious emotions that occur during the wear and tear of our action-driven daily life.
This does not, however, fully answer the question of whether fiction reading has a stronger influence on self-reflection than other activities that can be performed when the subject is at rest. What is the difference, in this respect, between fiction reading and patiently listening to a tale of woe presented by a friend? Here, it is interesting to examine more of the features of fiction reading listed above.

Risk-free imagination
When we read a fictional story, our imagination gathers associations and memories along the way to help fill the gaps in the story as it is told, to 'generate fictional truths'. 370 As examined in section 3.1, the reader has the opportunity to imagine situations and actions which she would not sanction in real life. One of the reasons for this is that imagining an action is different from performing an action. When I imagine doing something, I entertain the thought, I may even picture myself performing the  370 Walton 1990, 184. listen, I am aware of the presence of the other, and the social relationship between us.
Whatever I imagine when I listen to her story will typically be studded with references to our relationship. which is in the process of listening. Thus, my imagination is likely to be more focussed than in fiction reading, but less rich.

Motivation to empathize
There is still one difference to discuss: Is there a difference between fiction reading and non-fiction reading regarding the evocation of self-conscious emotions?
This is not an easy question to answer because there are many cases where fiction and non-fiction overlap. There may be some clear-cut opposites -Nussbaum, for instance, distinguishes between novels and traditional forms of philosophical writing in this context. Novels, she says, 'enlist the reader as a participant by sympathy and compassion', whereas philosophical texts typically 'ask the reader to be wary and sceptical '. 371 In these clear-cut cases, the intention of the work is either to tempt the reader into emotional engagement, or to tempt her to analyze a problem rationally. As we all know, however, even the most 'traditional' philosophical literature contains many instances where the reader is lured into believing something by having her emotions touched through the polemics of the author. The emotional involvement of the reader can be compassion or revulsion. For instance, when Schopenhauer describes the 'common man' as 'factory-made', 372 I feel my blood boiling at this contempt for humankind. 371 Nussbaum 1990, 238. 372 Schopenhauer 1970  one of friends or family members, my emotional involvement with the text will not only be evoked by imaginings, but will cover the area of my actual daily cares as well.
The second aspect is also connected with reader expectations. When we read a work of fiction, we usually assume that the fictional characters were constructed to emphasize typical problems of the human situation. If a work of non-fiction contains characters, we expect them to be drawn from real life, having been shaped by contingencies. Ricoeur picks up this point and says (with Aristotle) that fictional stories 373 Lamarque 2014, 20. 374 The difference between knowing and seeing is certainly one of the influencing factors on empathy in real life, as this example from G. B. Shaw's St. Joan illustrates: 'CAUCHON. Were not the sufferings of our Lord Christ enough for you?' DE STOGUMBER. No. Oh no: not at all. I had seen them in pictures, and read of them in books, and been greatly moved by them, as I thought. But it was no use: it was not our Lord that redeemed me, but a young woman whom I saw actually burned to death. It was dreadful: oh, most dreadful. But it saved me." (Shaw 1924, Epilogue). However, the investigation of the influence of actual seeing on empathy goes beyond the scope of my work.
'reveal universal aspects of the human condition', whereas historical stories depend on anecdotic aspects of a particular life. 375 If we do indeed expect this, we are inclined to take the outlined conflict in a fictional story more seriously. It is debatable in which cases the reader approaches the work of fiction with the explicit wish to learn more about the human situation; but if she does, this point is likely to influence her tendency to experience self-conscious emotions.
Having discussed the influence of fiction reading on the generation of selfconscious emotions in contrast to other means of information-gathering, I will now proceed to investigate how self conscious emotions grow in the three different kinds of narrative that are typically developed during fiction reading with participation. Shame, the self-conscious emotion in my example, is a follow-up emotion, brought about indirectly by engagement with the other-narrative. It is triggered by the experience of sympathy. As Michael Lewis points out, a person can feel shame (or any other self-conscious emotions) only if she possesses the 'mental representation of a "me" or self-reflected awareness', 377 so having this emotion is already the first step of a self-reflective process which may lead to a refiguration of the self-narrative.

'Self-conscious' emotions from the self-in-other-narrative
The self-in-other-narrative involves empathy with a character (f)  accumulate with the present one to such an extent that they overflow the reading process and carry on worrying me afterwards.
The main difference between the self-conscious emotions experienced within the self-in-other-narrative and the self-conscious emotions invoked through the othernarrative is one of distance, and thus intensity: in the other-narrative we only experience the self-conscious emotions on second thoughts, as a consequence of sympathy, but in the self-in-other-narrative our empathy with the character(f) can lead us deeply into ourselves, and into feeling for ourselves.

'Self-conscious' emotions from the other-in-self-narrative
In the other-in-self-narrative, a character ( wake of other fiction-induced emotions, thenas we saw in the previous sectionthey typically lead to the reader's awareness that her self-understanding, woven together as a self-narrative, does not stand the test of the actions / reactions which she imagined during her engagement with the text. This feeling of imbalance leads to a desire to put the self 'back into kilter', to re-establish a feeling of personal unity.
According to Ricoeur, this is done by consciously examining the imagined actions of the fiction-reading process and confronting them with one's self-narrative.
The resulting dialectic between counterfactual actions and the story one has woven about one's own life will, upon self-reflection, lead to a revaluation of the self. During this revaluation process, new ideas and new goals will be introduced, and old ones will be either deepened or disabled. As Ricoeur puts it, there is a 'play of sedimentation and innovation applied to our understanding of ourselves '. 380 This thought suggests that self-reflection is a reliable tool to achieve self-unity and self-respect. Ricoeur holds that fiction reading helps to investigate and try out one's own action in a non-threatening environment, thus receiving a wealth of self-testing material in a non-threatening environment.
Following a similar train of thought, Nussbaum says that novels can distance us 'from blinding personal passions' and cultivate 'those that are more conducive to community', 381 and that novels also enable readers to give way to self-indulgence without harming real people. So, if this is right, reading has two good influences on us: It makes us reconsider our past and present actions from an ethical point of view, and it purges us from anti-social urges.
Suzanne Keen conducted an experiment with her students, giving them three texts that were aimed at the reader's emotions, all of them written in the first person, from the point of view of a down-at-luck orphan: a fraudulent email from Africa asking for a bank transfer, a hand-written letter with an attached report card asking for money, and a fictional text where the orphan character talks about her dreams of becoming a mechanic. A common factor of the students' reactions to these texts was that they thought the fictional text was the most trustworthy. 382 Keen sums up the results of her study like this: 'My research suggests that readers' perception of a text's fictionality plays a role in subsequent empathetic response, by releasing readers from the obligations of self-protection through scepticism and suspicion. Thus they may respond with greater empathy to an unreal situation and characters because of the protective fictionality, but still internalize the experience of empathy with possible later real-world responsiveness to others' needs '. 383 Is fiction reading then always good for us, turning us into more empathetic persons, and restoring our self-esteem? When we look at the people who are part of our social circle, it soon becomes obvious that some avid fiction readers are not as socially capable as such a claim might lead us to expect. This suggests that some further prerequisites that must be met for the chain process that leads from strong selfconscious emotions to self-reflection, and from self-reflection to self-refiguration, to be successful. Another question is: what happens to the reader who only reads fiction in order to be transported into the story and to enjoy the ridedoes self-reflection enter this process at all? Is, in short, reading good for people who do not manage, or do not look for, self-reflection?
In my view, the value of reading for the reader's unity of self depends (among other factors) on the quality of her self-narrative. Even though a self-narrative is typically only accessible to the self-narrator, not to the external world, it can have similar attributes as fictional narratives. As discussed in section 2.4, I adopt Ricoeur's view that we (unconsciously) use our self-narrative as a practical guideline to cope with the necessity to act and react in our daily lives. If we are not very discerning in our observation of the people in our social nexus, our self-narrative can be built on simple (and often idealized) lines. Hence, I find it useful to distinguish between literary fiction and genre fiction even in self-narratives.
There are many common language distinctions between these two, with a huge grey area where they seem to overlap, so I shall use my own definition. I shall take literary fiction to be a fictional narrative that has the main objective of illuminating the human situation, without diminishing this aim by obeying normative generic conventions. In my definition, genre fiction has the main goal of providing stereotyped roles which are easy to recognize, and differentiated insights into the human situation take a subordinate role, only helping to make the story convincing. 384 383 Keen 2006, 220. 384  The objective of genre fiction is to provide the reader with easily recognizable types of character(f) and easily recognizable story patterns, thus achieving some, but not necessarily all of these three goals: giving the reader the pleasure of finding something that is familiar to her; making the reader feel instantly at home in the story; making the reader pleased with herself by echoing and confirming her values. None of these goals is likely to make the reader conscious of a conflict in her life while she is within the reading process, and the main emotion to be expected in the reader is a mild feeling of contentment. 385 Hence, self-reflection is only likely to happen if the reader engages with a work of fiction that represents genuine conflicts of the human situation.
What happens if the reader's self-narrative resembles a piece of genre fiction, consisting of clichéd story lines? If a person manages to keep her self-narrative on simplified lines, without much discernment of social conflicts, she is likely to cherrypick the items of the fictional story that fit her expectations, and ignore potentially difficult story parts. Thus, I hold that self-reflection can only happen, and subsequent self-figuration can only lead to a stable sense of self, if the reader's self-narrative has the nature of literary fiction, as opposed to genre fiction.
A genre-based self-narrative is usually biased, giving undue weight to some characteristics of the self and her life, while ignoring other aspects. I am going to discuss the problem of genre in fiction and in self-narrative in detail in the course of this section.

Self-reflection and unity of self
As discussed in the previous section, self-conscious emotions have a tendency to persist after the reading process because they create a sense of imbalance in us which is perturbing. As we saw in section 1.9, shame, embarrassment, guilt or even pride can make us feel that we are playing the wrong role, in a current situation, or in our selfnarrative, or in both. They force us to recognize a conflict, and they motivate us to reflect on our life and on the way we have presented it to ourselves up to the present. 385 Even being frightened by Stephen King horror novel, or crying over Margaret Mitchell's Gone with the Wind, eventually leaves us with this feeling of contentment, a bit like digesting a spicy but satisfactory meal. What can be objected is that indulging in genre fiction that reinforces existing delusions can make real life unbearable to the reader afterwards (as in Gustave Flaubert's Madame Bovary, for instance). In such a case, however, fiction reading is not likely to trigger self-reflection: the real life appears to be wrong to the reader, not her self-narrative.
self, and our self-narrative? Depending on the strength and the valence of the emotion, there are three possibilities.
The first possibility is that the emotion challenges our ethical values so much that we stop the reflective process as soon as we feel the onslaught. The most likely source for this would be a violent negative emotion generated during the generation of the other-in-self-narrative. Iffor instancereading the wittily written erotic scenes in These three variants of self-reflection have one thing in common: they serve to give the reader a stable sense of self, supported by a believable story, which helps her to tackle her daily conflicts.
Self-reflection after fiction reading does not, however, always behave in this ideal way. What can go wrong? I shall investigate the causes of non-functional selfreflection in the next section.

Self-reflection and bias
It is a well-known truism that self-reflection can only be as good as the mirror we watch ourselves in. How does this translate to fiction reading? What makes self-reflection as a result of fiction-induced emotions realistic and useful, and under which circumstances is self-reflection uncritical and self-serving?
I shall investigate two possible factors which can be responsible for a mirror distortion: the motivation of the reader when she chooses her fictional reading matter, and the quality of the reader's self-narrative.
Motivations for reading a work of fiction are similar for most readers, but differ in strength. Many readers choose a work of fiction in order to be entertained, to get out of their every-day drudgery into an unthreatening fictional world. Whereas some readers want to explore new worlds, others want the fictional world to be source of solace for 388 Robinson 2006, 156. their own imperfect lives, providing them with dreams which are unlikely to come true in real life. If the need for the dream is too strong, the reader will not be able to assimilate the complete work of fiction, but only the parts of it that relate to the dreamstory. This is, for instance, illustrated by the number of romantic comedies that were made on the basis of Jane Austen's sharp social novels; these films typically constitute visible versions of self-serving reading by extricating some sickly romance from the rich satire of Austen's writing.
Readers who engage with a work of fiction in order to fill a self-serving need As another example, a recalcitrant ten-year-old girl might convince herself that her mother's scolding and her schoolfellows' jeering are not to be taken seriously because they just fail to see that she has been sadly neglected and misunderstood by her own family, and that she is entitled to eccentric behaviour and bad temper to make up for ill-treatment. In such a case of biased self-narrative, reading a book like Frances Hodgson Burnett's The Secret Garden 394 might strengthen the girl's misconception of herself. She would empathize with Mary Lennox and look for similarities between Mary and herself, while tending to ignore the differences. As Michael S. Brady says, 'subjects often seek for reasons that confirm, rather than disconfirm, their initial emotional construals', making themselves 'susceptible to a 'confirmatory bias'. 395 The reading process would probably evoke an enhanced self-conscious feeling of pride in the girl, a glorious feeling of being different and hence superior to others, and subsequent selfreflection would increase the biased genre-like character of her self-narrative.
A biased self-narrative will present an image of a unified self to the subject, but in contrast to a reality-based self-narrative, it is not only fragile, but also rigid. It cannot prove or disprove itself by social encounters. In such cases, the bias-driven version of self-reflection will not be likely to amend the real unity of the self, or to allow any additional discernment.
As shown in this section, self-reflection is even possiblein a very limited way if a reader chooses a work of fiction that is suitable for affirmative reading. If, however, the self-narrative of the reader is already biased and unrealistic, any selfreflection after the reading process is likely to sharpen the problem instead of paving the way towards a stable sense of self-unity.
In this chapter, I discussed two issues: first, the possibility that reading-induced emotions persist after the reading process, triggering self-reflection, and second, the question of whether self-reflection is always good for us. I arrived at the conclusion that fiction reading with empathy for a character(f) can invoke self-conscious emotions, and that self-conscious emotions are particularly likely to lead to self-reflection. Regarding the self-reflection process itself, I argued that this can only help us towards an improved unity of self if our self-narrative does not deviate too much from the role we play in our social nexus. 394 Burnett 2018. 395 Brady 2013, 162.

Filling out Ricoeur's account of fiction reading
In the previous two chapters, I developed my own account of how fiction reading and personal identity are interlaced through narrative: first, during the reading process, and second, as a consequence of the reading process.
Here, I am going to highlight the lacunae in Ricoeur's account of fiction reading in connection with narrative identity and explain how these are filled with my own account. As I explained earlier, the lacunae are located in these areas: 1. during fiction reading, in the influence of imaginative variations on the reader's self, in particular on the reader's emotions; and 2. after fiction reading, in the role of reflection during the refiguration of narrative identity.
In the detailed discussion, I am also going to compare relevant parts of Ricoeur's account with Walton's account of engaging with a work of fiction.
Before I start to investigate these points in detail, I would like to give an overview of Ricoeur's main claims regarding the effects of fiction reading on the reader: 396 Reading fictional texts provides the reader with imaginative variations of her self to gain understanding and try out new perspectives on life. The reader who engages with a work of fiction enters the fictional world evoked by this work. Her reading process is accompanied by a dialectic with the inferred author 397 : her own expectations of how the text should be interpreted are continually challenged by the inferred author who wants to seduce the reader towards viewing the world in a specific way. The reader deals with this challenge by appropriating the fictional world in her own way: she merges her own experiential and cultural horizon with the horizon offered by the inferred author, thus experiencing imaginative variations of her self in a low risk environment. To be able to do this, she must first relinquish her self and then playfully adopt a new understanding of the world as offered by the text: 'to understand is not to project oneself into the text; it is to receive an enlarged self from the apprehension of proposed worlds which are the genuine object of interpretation'. 398 Ricoeur calls this activity mimesis3. 399 During mimesis3, the reader experiences passive contemplation (aísthēsis). In connection with 'the readers' tendency to identify with the hero' 400 this can lead to confrontation with unknown norms and, finally, to catharsis. 'Aesthetism frees the reader from everyday concerns, catharsis sets the reader free for new evaluations of reality that will take shape in rereading'. 401 Catharsis typically has three effects on the reader: purging of pent-up or hidden emotions, subsequent clarification of the conflicts in the fictional story, and learning from the newly gained insights.
Learning leads to reflection on the reader's own conflicts and to refiguration of the reader's self-narrative.
I am going to explore the overlappings and the differences between Ricoeur's account and mine in the following areas: Identifying with a character from within often evokes emotions in the reader, and these emotions lead the reader to either take on the part of the character(f) as an extension of herself on a concordant and presumably predictable path (albeit with some new vistas), or to let herself be led astray by an uncannily familiar, but ethically dubious character(f) on a precarious journey on thin ice. These different paths are the self-in-other-narrative and the other-in-self-narrative. 414 In the reader's unconscious mind, an inner monitor, the doppelgänger-function, is always on the alert for indications of which path the reader should take. I take this doppelgänger function to be a socially developed mechanism that runs below the reader's current level of consciousness. 415 Ricoeur's account is much thinner regarding the reader's imaginings. About my example, he would just say that the reader appropriates the fictional world of Stoner and identifies with a character(f). Joyfully identifying with Stoner when he is sanding a wooden board with quiet enjoyment on one extreme, and reluctantly / guiltily travelling with Edith in her angry revenge against her husband on the other, belong to the same kind of activity for him, and thus, he does not investigate them separately. For me, these are two different kinds of story: the first becomes a self-in-other-narrative during reading, whereas the second develops into an other-in-self-narrative. 416 They provide different kinds of reading experience, and they typically generate different kinds of emotion in the reader.
In contrast to Ricoeur, I say that during the reader's own actualization of the fictional story, memory sequences of the reader's own life become part of the reading experience. As discussed in section 6.2, these memory sequences can dominate the reader's imagination of the story, or they can themselves be dominated by the influence of a character(f). These are the two extremes of the doppelgänger-function.
Distinguishing between them enables me to examine whether these two paths lead to different possibilities of self-reflection after the reading process. I am going to say more about this under 2a).

1c) Do my imaginations belong to the fictional story?
Ricoeur says that it is constitutive for the self to be embodied, and thus anchored to the world. In his account of fiction reading, however, he claims that the reader leaves her self behind while appropriating the fictional world. This is a bold move, and it keeps him from investigating the connections between the reader's embodiment and her engagement with a work of fiction.
When Ricoeur talks about the reader's self here, he means the ipse, the lived self, to be distinguished from the idem, the objectified personal identity which includes character(p). Ricoeur acknowledges that the reader is cognizant of the features of her idem during the reading process, and that this influences the way she reads and understands the text: 'Insofar as the body as one's own is a dimension of oneself, the imaginative variations around the corporeal condition are variations of the self and its 416 For a detailed discussion see section 6.2. selfhood'. 417 What Ricoeur underestimates is the connection between the lived self and the reading process.
In my view, it is not accurate to say that the reader relinquishes her self when she enters a fictional world. I agree that she leaves her current concerns behind, but a basic self-awareness remains. A reader is still alert to phenomenological perceptions and able to react to them: if her leg becomes numb, she may change her position, and if she feels peckish, she may nibble a snack while continuing to read. In addition to that, she may be reminded of sensual experiences in her own past when she identifies with a character(f). When she enters the warm reception room together with Stoner, when the heat of the room almost pushes him back, she may recall how strange it can feel to enter a warm room coming in from the cold, and the context in which this happened to herself in the past. Ricoeur does not put enough emphasis on the sensory quality of experiential memories and the emotions which are linked to this. For him, the fiction-induced imaginative variations around the bodily condition merely serve to help the reader to identify with a character(f).
A basic self-awareness also remains on the level of social behaviour. A reader who identifies with Stoner in the scenes where Stoner is treated unfairly by his colleagues will feel the injustice not only for Stoner, but for herself, in all the remembered instances when she was in a comparable situation.
I hold that when a reader engages with a work of fiction, she does not leave her self, but rather her self-narrative behind. She enters the fictional narrative insteadbut she retains access to her experiential memories which may be triggered by imagining a character(f) from within. These experiential memories form part of the reader's idem; the reader can retrieve them because they are rooted in her embodiment. As soon as she has recalled them, they form part of her ipse, her lived self, because they become part of the reading experience.

2a) How is reflection triggered?
As we have seen, some reflection can already be motivated by the reading process itself, but the more interesting question is whether the reading experience can lead to subsequent and deeper reflection, and eventually a refiguration of the reader's selfnarrative. Ricoeur says that provocation from fiction reading induces the reader to reflect on her self-narrative. By 'provocation' he means an unforeseen confrontation with new insights and moral viewpoints. In principle, I agree with him here. However, the provocation process is more complex than Ricoeur allows. A new insight is not itself sufficient to trigger self-reflectionif it were, newspaper-reading would throw us into spells of self-reflection every day. In my view, the emotions which were invoked during fiction reading typically lead to subsequent self-reflection. This a more than a prod. As shown under 1a), fiction readingand, in particular, empathizing with a character(f)can trigger a chain of strong emotions in the reader, many of which are self-conscious. Self-conscious emotions are especially potent in disturbing the reader's peace of mind, making her reconsider the way she has been living. Emotions from fiction reading lead the reader towards reflection in a subtle way, on different levels: towards a situation-based moral decision, or towards a general moral lesson.
Having identified with Edith's revenge while reading Stoner, a reader would typically be torn between two emotions: empathy with Edith, her frustration and her rage about a life that has turned out to be completely empty and useless, on the one hand, and the ethical concern that what Edith does is deeply unfair and hurtful towards her husband, on the other hand. Identifying with Edith, one would slide into an emotional conflict that is not easy to resolve. Therefore, I hold that often, it is the gnawing sensation of one's own inner conflictbrought about by identifying with a character from within, in the other-in-self-narrativewhich induces reflection after the reading process. These examples show that self-reflection typically develops from emotions.
The emotions make us aware of how important the unexpected information from fiction reading is for us and our lives. In an environment which does not call for immediate action, we can reflect on this and think about refiguring our self-narrative. As discussed in section 7.1, I arguecontrary to Ricoeurthat self-reflection can be triggered in different ways by the reader's complex emotions, and that a simple confrontation with new insight is not sufficient to explain the reader's motivation to self-reflect.

2b) Does reflection move my self towards an improved unity of life?
In Ricoeur's view, reflection can lead the reader to change her self and her selfnarrative. A prerequisite for this is that the reader has been able to engage with the work of fiction and identify with a character(f) from within. For reflection to take place, the reader needs to reach a moment of impetus 'when reading becomes a provocation to be and act differently', and, in addition to that, this impetus is transformed to action by the decision 'Here I stand!'. 419 Ricoeur says that reading-induced reflectionif it occurs successfullyleads the reader to make a decision about her ethical responsibility, and thus about changing her commitments (her self) and her plans and her self-assessment (her self-narrative). For him, this reflection naturally leads the reader to better insight, and puts her on the way towards a more realistic self-narrative.
In my view, the problem here lies in Ricoeur's explanation of how the reader finds an appropriate way to identify, and reflect upon, the text's suggestions to be and act differently. Ricoeur thinks that the only struggle the reader has is against the provocation of the text, deciding whether or not she will allow herself to be influenced by it.
I hold that there are further factors which influence the reader's ethical choices.
These factors are not theoretical ethical considerations; they stem from the quality of reader's current self-narrative. If the reader's self-narrative complies with the facts of life as an external observer would perceive it in her social and physical environment, the reader will typically be able to identify with a character(f) in a way which is at least roughly similar to what the text prescribes. In this case, reflection is likely to lead the reader on the road to coping with her life even more successfully. If, on the other hand, the reader has delusions about herself and her life story, identifying with a character(f) may send her even deeper into her misconception about her life. Gustave Flaubert's novel Madame Bovary 420 is a literary example of this. The eponymous heroine likes to read romances about elegant and noble men who lay the world at the feet of the woman whom they love. Identifying with the ardently adored women in such novels encourages Madame Bovary in her delusion that she is destined to live a life of passion and romance. She acts according to this delusion (following her erroneous self-narrative), and this does not help her towards a better unity of life. Instead, it leads her to ruin and suicide. Thuscontrary to Ricoeur -I say that an existing bias towards the unrealistic in a reader's self-narrative can have an influence on the success of her post-reading reflection.

2c) Does reflection refigure my self-narrative?
Ricoeur says that reflection, if successful, is usually followed by a decision to act and be differently, and thus to a refiguration of the reader's self-narrative, and towards an improved unity of life. 421 I agree with this, but I find it important to add that post-reading reflection is not always either successful or unsuccessful. As stated in the previous paragraph, reflection 420 Flaubert 2003. 421 It is theoretically possible that my self-reflection leads to the conclusion that my life, and my self, are already as well-adjusted as they can be. But if this were the case, it does not seem plausible that I had fiction-induced emotions which led me to self-reflection.
can also lead to harmful results if it sets out from a biased self-assessment. In such a case, it can harden existing erroneous convictions about the reader's self and selfnarrative, which is likely to lead to additional clashes and conflicts in her daily life.
So, contrary to Ricoeur, I hold that reflection can only refigure my selfnarrative if my self-narrative does not deviate too much from my role in real life as perceived within my social nexus. If my current self-narrative is delusional, reflection after fiction reading can force this unrealistic narrative even further from any narrative of myself that an objective observer might construct. In this case one could say that my self-narrative is not refigured, but hardened.
In this chapter, I highlighted the differences between Ricoeur's account of how a reader's self-narrative can be influenced by fiction reading, and my own account. This is the summary of my investigations: In particular, I pointed out that the emotions experienced during fiction reading, and the imaginings that occur alongside the 'authorized' game of make-believe, can have a strong influence on the emotional reactions of the reader, and that they may trigger a chain of new self-conscious emotions. This consideration is missing in Ricoeur's account, and hence, he cannot give a convincing explanation of how and why the reader is motivated to reflect on her own life after fiction reading. For Ricoeur, catharsis and subsequent learning explain this motivation; for me, re-living conflicts of one's past is just part of a complex web of self-directed emotions which emerge during the reading process.
I also showed that reading-induced self-reflection can take three forms: it can be successful, unsuccessful, or harmful in developing the reader's self and self-narrative. In

Filling out Walton's account of fiction reading
In Part III of this work, I referred to Walton

1a) What part do subsidiary imaginings play during fiction reading?
Walton says that when a reader identifies with a character(f) from within, she engages in imaginings 'that are not part of the authorized game but occur along with it'. 424 He argues that this is necessary because the reader cannot be completely identical with a character(f), and thus has to use her imagination to fill some gaps in order to imagine being in the character(f)'s shoes. In Walton's view, these imaginings are similar to those we undertake when we try to imagine a real person from within.
I wholly agree with Walton's general concept of subsidiary imaginings.
However, I hold that there remains more to be said about their sources and their consequences to explain the nature of the reader's emotions. Subsidiary imaginings do not just attach themselves to the fictional narrative out of the blue; they are usually based on emotionally charged memories of the reader. In order to imagine a character(f) from inside, to know what it feels like to be him, I need to resort to my own experiential memories. These memories bring my own unresolved emotions from the past with them. Imagining William Stoner at the point of his story where he inadvertently becomes a good teacher stimulates me to recall scenes from my own life when I realized that I was doing something well which I did badly before. I may remember, for instance, how I used to ignore my grandmother's constant complaints about not being able to go out on account of the bad weather until one day I took her for a walk and realized that she just needed someone to go with her and did not want to say so.
Suddenly, I felt I was acting as a caring relative should act. As a result, I felt remorse about not having been attentive enough in the past, and this remorse hits me again during the reading of Stoner.
Usually, the memories we resort to in order to fill the gaps in fictional texts are charged with emotions; this is why they come back so easily to our consciousness. The memories we recall are embedded in their own stories, and these stories can evoke strong emotions in us when we live through them again during fiction reading. Hence, the subsidiary imaginings during fiction reading can have an influence on how emotions are generated in the reader, and whether they subsist after the reading process. 424 Walton 1990, 255. This is an important point for my account because it explains how the reader typically develops a network of self-conscious emotions which can lead the way to selfreflection, and refiguration of the self-narrative.
1b) What are emotions during fiction reading like?
As we saw in section 3.1, Walton makes a clear distinction between the emotions we experience in real life and the emotions we experience during fiction reading when we identify with a character(f This gap in Walton's account has been recognized by Jerrold Levinson. He says that we should supplement Walton's account 'in such a way as to be more adequate to our abiding sense that at least in some cases or interactions, something is really going on emotionally, i.e., there are some real, and not just quasi, or make-believe, emotions involved in the situation as a whole'. 429 Although Levinson appears to equate the terms 'quasi' and 'make-believe' here, which does not reflect Walton's views, the point he makes is true: in addition to emotions that are based on imagining that something is the case, fiction reading can also trigger emotions are based on a belief that something is the case. When an emotion is based on a belief, a quasi-emotion and some motivations are generated. Levinson suggests that there is a leakage between two levels, 'that of imaginative connection with the characters and that of half-remembered, dimly focussed recollections of stored life experiences'. 430 The real-life emotions evoked by fiction reading are based on the beliefs about these stored life experiences, and they prompt a motivation to resolve a conflict which has become apparent during the reading process. Levinson's distinction between two levels delves deeper into the details of the reading experience than Walton's account does. It stresses the influence of the reader's recalled memories on the quality of the reading experience. However, I hold that This is an aspect of fiction reading which is neglected by Walton. He acknowledges that we can have emotions when we engage with a work of fiction, but he does not investigate how this is influenced by our recalled memories, or what the target of these emotions is. In short: Walton says that during fiction reading, we can only have mental states that consist of quasi-emotions plus a make-belief, and that we care about these emotions because they enrich our lives. In my account, I claim that we can have both quasi-emotions towards characters(f) and strong real-life emotions towards ourselves when we engage with a work of fiction. These real-life emotions are not just an appreciation of quasi-emotions, as Walton claims. Often, real-life emotions are caused by recalled experiential memories. They can be intense, and they can persist after fiction reading. Hence, the reader adds more emotional content to the reading process than Walton allows.

2a) How does fiction reading contribute to our lives?
Walton says that the connection between fiction reading and the reader's real life, especially the benefit of participatory experiences, is something that still needs to be investigated in detail, in particular 'the specific ways in which participatory experiences contribute to our lives'. 432 In his search for the benefits of fiction reading he mentions the opportunities for trying out new roles, releasing unacceptable emotions, recognizing repressed feelings, facing disturbing aspects of ourselves, or gaining experience in dealing with specific situations. He wonders whether there is 'an evolutionary explanation of why human beings find experiences that do in fact produce insight exciting or enjoyable'. 433 I concur with Walton's idea that the desire of humans to learn is something that can be investigated further, especially in empirical studies by psychologists or neuroscientists. From the philosophical angle, however, I think that Ricoeur's account of narrative identity provides the most interesting insight into the post-reading effects of fiction reading. I agree with Ricoeur that working out conflicts or trying out new roles are not isolated activities (as Walton represents them), but experiences that are closely woven together with the reader's self. In my own account, I explore these activities more deeply by defining the doppelgänger-function that monitors the other-narrative, letting it swing towards the self-in-other-narrative or the other-in-self-narrative. These narratives represent different kinds of reading experience, evoking in their turn different kinds of emotion which often persist after the reading process.
Thus, the reading experience can form the basis for extended self reflection and the long-term effects on the reader's goals and self-understanding, including a refiguration of her self-narrative. As the self-narrative is a practical identity needed for the interaction of the embodied self with the world and her social environment, 434 a fiction-reading induced refiguration of the self-narrative does indeed contribute to a person's life.
Ricoeur's model of narrative identity has given us the framework for the integration of fiction reading into the reader's lived self and her self-narrative. However, my account is an advance on Ricoeur's in that I explore the details in more depth, distinguishing between different reader narratives and different kinds of emotions. This is at least a partial answer to Walton's open question: I showed that fiction reading can contribute to our lives by changing our self-narrative, including our plans and commitments for the future. I do not attempt, however, to provide any answer to the question whether fiction reading can change the way we act; I leave these investigations to future empirical research by psychologists.  These are the most important points of my discussion: I hold that fiction reading does not only involve quasi-emotions, but also real-life emotions directed towards the reader herself; and I argue that the concept of narrative identity shows how fiction reading can contribute to the reader's own life via refiguration of her self-narrative, which is a practical identity.
In this main part of my work, I produced an account of the influence of fiction reading on the reader's self, which draws but also advances on Ricoeur's account of narrative identity. I investigated the effects of fiction reading both during and after the reading process. My theory covers two areas: the development of the emotional involvement of the fiction reader, and the process of self-reflection and self-refiguration after the reading process. I claim that Ricoeur's account of fiction reading does not fully explain how and why the reader experiences emotions during and after the reading process.
Ricoeur says that learning from fiction leads to self-reflection. This is not plausible: learning increases our knowledge, but it does not motivate us to re-think our lives. The important question is: what can spark the motivation for self-reflection? In my own account of fiction reading, I explore the different kinds of narrative a fiction reader weaves during the reading process, and the kinds of emotion that can be evoked by these narratives.
I hold that emotions often have a complicated structure (which Ricoeur does not recognize), and that they frequently have a narrative quality. In chapter 5, I explained that my view is compatible with any theory which says that an emotion is intentional, complex, and structured.
In chapter 6, I investigated how the reading process deepens into a reading experience when the fiction reader feels sympathy or empathy, caused by her engagement with a character(f). Starting from Ricoeur's interpretation of fiction reading as an 'imaginative variation of our ego', 435 I examined how emotions can be evoked during fiction reading, and what these emotions are about. I showed that the reading process has an experiential quality for the reader, and that this experience in itself can trigger a chain of new emotions in the reader, many of them self-directed.
Regarding the development of self-directed emotions, I took encouragement from the empirical tests performed by psychologists who questioned readers on the emotions that they felt after reading examples from works of fiction. Their test results indicated that in almost all cases, the fiction reader experienced self-conscious emotions after the reading process. Taking up this idea, I suggested that the reader's emotions are not created as a direct result of imagining something fictional, but as part of a subsequent chain of emotions which develops from the engagement of the reader with the fictional story. The full engagement with the story is possible only if the reader fills the gaps with her own experiential memories, which are linked to her personal history in which they originally occurred. Thus, both the experience of reading a work of fiction and the experience of re-living one's own memories are woven into the reader's engagement with the work of fiction.
The thought that the act of reading is an experience in itself, interwoven with the narratives of the reader's experiential memories, is under-represented in the fiction reading models of both Ricoeur and Walton. The addition allows for a deeper analysis of the reader's experience and its time frame. Why are some reading-induced emotions longer-lasting, and more disturbing, than others?
My claim is that identifying with a character ( How do the emotions enter the reading process? Ricoeur and Walton say that the reader creates her own narrative, her own interpretation of the text, and Walton acknowledges that the reader has imaginings that run alongside the mandated narrative of the text. I showed that this distinction is not detailed enough to explain the different chains of reading-induced emotions. In my view, the fiction reader creates three levels of narrative during the reading process: the other-narrative, the self-in-other-narrative, and the other-in-self-narrative. The first of these does not involve any variation of the reader's ego: the othernarrative consists of the story of a character(f) understood by the reader, based on the hints given by the work of fiction. To build the other-narrative, the reader applies her experience of the world and her knowledge of cultural customs to the fictional text, but she does not need to project herself into the story. She appreciates the story, and she may feel sympathy with characters(f).
The self-in-other-narrative and the other-in-self-narrative develop from the other-narrative the reader creates for a character(f). They usually evolve when the fiction reader imagines a character from within. I showed that Ricoeur's 'imaginative variations of our ego' can be explained in more depth by studying the web of stories the reader creates during the reading process. The reader enters the narrative during her empathy with the character(f), bolstering up her imaginings of this character(f) with her own experiential memories. These memories bring their own mini-stories with them, which then become part of the reading experience. By imagining a character(f) from inside, the reader slides into the fictional world, entangled with her experiential memories, and develops emotions about her own self. During this process, it is possible that the reader is either envious or desirous of the character(f)'s actions, or disgusted by them. In the first case, the reader imagines new views and ways of living (which are compliant with her ethical values) through the character(f), and the other-narrative swings towards a self-in-other-narrative. In the second case, the character(f) leads the reader on to paths the reader would not have dared or wanted to tread in real life, thus turning the self-in-other-narrative into an other-in-self-narrative. As the reader can switch between these two self-involving narratives within the same reading experience, I call this process the 'doppelgänger-function'.
The doppelgänger function is constantly on the look-out for similarities between the reader and the character(f), thus influencing the direction of the reader's imagination. I hold that my concept of the doppelgänger function is a fine-tuned tool to distinguish in which directions the imaginative variations of the fiction reader's self may go. As shown with the example of John William's Stoner, the self-in-other end of the doppelgänger function tends to cause emotions about situations in which we found ourselves in the past, whereas the other-in-self end typically leads to either a fleeting elation or a concern about a lesson newly learned, or shame about an unwelcome connection understood. So my distinction of the three types of narratives, monitored by the doppelgänger function, gives an explanation of how different kinds of selfconscious emotion can evolve from fiction reading.
As a next step, in chapter 7, I investigated the effects of fiction-induced emotions on the reader's self-reflection. I showed that self-conscious emotions like shame, guilt, embarrassment or pride are often generated during the reading process with participation, in particular through sympathy or empathy with a character(f), for two main reasons: detachment from the reader's current real-life concerns, and the possibility of risk-free imagination. I argued that these self-conscious emotions can result from the three different kinds of narrative discussed in chapter 6: the othernarrative, the self-in-other-narrative, or the other-in-self-narrative. Reading-induced emotions have an important feature in common: they arein the endmainly about myself, the reader.
Self-conscious emotions already contain a nucleus of self-scrutiny when they are invoked. This is why they have a high influence on a reader's motivation to engage in subsequent self-reflection: they draw the reader's attention towards unresolved conflicts in their self-narrative.
In chapter 8, I discussed the differences between Ricoeur's account of fiction reading and my own. Regarding the reading process itself, I argued that Ricoeur misses two important points: first, his view of catharsis is too simple to represent the complex chain of emotions which are triggered by fiction reading, and second, he does not attach sufficient importance to the impact of experiential memories which the reader recalls during fiction reading. Hence, he has no convincing explanation of how the reader comes to reflect on the fictional worlds she got to know. In my account, I filled these gaps. First, I showed how different kinds of emotions emerge during the three types of narrative the reader creates when engaging with a work of fiction. Second, I explained why many of these emotions are self-conscious, and hence possible starting points for self-doubts.
Regarding the reader's self-reflection after the reading process, I argued that Ricoeur's account seems over-optimistic regarding the benefits of fiction reading.
Ricoeur says that self-reflection will guide the dialectic between counterfactual actions and the story one has woven about one's own life towards a revaluation of the self. His account seems to suggest that fiction reading is always beneficial to us, leading us in the direction of a unified life and better social integration. As I pointed out, this is the ideal path of self-reflection, but self-reflection can work in different ways. It canas Ricoeur suggestschallenge our values, it can challenge our self-narrative, and it can make us more discerning in our social relationships.
If, however, the reader's self-narrative is severely distorted when she engages with a work of fiction, her subsequent self-reflection will be based on wrong premises. I showed that readers with a biased self-narrative are likely to react only to parts of the fictional work which support their erroneous conception of their own life. In such cases, the self-narrative will not have much chance of refiguration. I pointed out that Ricoeur's account does not include this possibility. Ricoeur holds that self-reflection can only lead to an improved self-unity. In my own account, I showed that self-reflection can only do that if it is based on a realistic self-narrative.
In chapter 9, I tackled two points in Walton's account of fiction reading which are open to questions. First, I challenged Walton's view of the reader's emotions during fiction reading. Walton says that during fiction reading, our emotions are only felt within the fiction. I showed that the reading experience and the recall of experiential memories can trigger a chain of real emotions in the reader. In my account, I explored how and when experiential memories can enter the reader's narrative, transform it, and evoke a chain of self-conscious emotions which exist outside the reading experience.
Second, I pointed out that the relationship between fiction-induced emotions and self-reflection can contribute to the reader's life by influencing the way she refigures her self-narrative. Walton says that this point is still to be explored. I showed that selfconscious emotions which evolve during the reading process can persist and start a process of reflection, during which the reader can rewrite her own story and, if it is realistic, gain a fuller unity of life.
Overall, the main benefit of my account of fiction reading is that it accounts for the creation and the influence of self-conscious emotions which are generated during the reading process, and which can continue after reading. My account addresses a new topic: the interactive influence of fiction reading and the fiction reader's narrative identity on each other.
Emotional experience: combined phenomenological feeling (including bodily sensations as well as psychological feelings) caused by an emotional reaction or an emotional episode.
Emotional reaction: short-term emotional response, which might last for a few seconds or a few minutes.
Ethical subjectivity: a person's ability to act according to her personal aims, preferences and values. This ability is shaped and developed in a person's interaction with her social nexus and its prevailing value system.
Event schema: the conceptualization of past experience. In memory, the events are structured into a recognizable script.
Fiction: a story the content of which was not constrained by fidelity to reality.
Idem: one of the two aspects of personal identity: the durable self, the answer to the question 'what?'.

Iff: philosophical term for 'if and only if'.
Inferred author: each reader's own version of the implied author.
Implied author: the author who is implied by the genre and style of the work of fiction, by the choices of theme, point of view, protagonist, story structure, vocabulary and general setting.
Intentional content: content of an intentional state. Can be descriptive or directive, or a combination of both (appraisals).
Intentional state: a mental state that is directed towards an object, or a state of affairs.
Ipse: one of the two aspects of personal identity: the lived self, the answer to the question 'who?'.

Mimesis: Imitation of action.
Mood: an affective state that is more 'global' or 'diffuse' than emotion.
Narrative: a representation or specific manifestation of a story.
NIA: Narrative Identity Account, claims that a person shapes her sense of herself by weaving the events of her life together in an ongoing autobiographical narrative.
NIDA: Narrative Identity Descriptive Account, claims that every person weaves her autobiographical narrative without the need of a specific intention to do so.
NINA: Narrative Identity Normative Account, claims that a person benefits from weaving her autobiographical narrative.
Other-narrative: one of the narratives that can emerge during fiction reading: story that the reader weaves for a character(f).
Other-in-self-narrative: one of the narratives that can emerge during fiction reading: the other (in this case the character(f)) influences the way the self views herself and her own values.
PCA: Psychological Connectedness Account, claims that a person's identity over time consists in connectedness or continuity of her psychological states.
Self: short for "the lived self", the conscious human being in interaction with the physical and social world.
Self-narrative: the ongoing story of the self. The self-narrative provides the practical identity needed for the interaction of the embodied self with the world.
Self-in-other-narrative: one of the narratives that can emerge during fiction reading: the self of the reader enters the story of the fictional 'other' by adopting her emotional experiences.