1 Introduction

Our instinctive psychological ontology is realist: in our sober, i.e. not philosophical, moments we hardly doubt that we have mental states. If looked at a bit more philosophically, however, we do not say very much by saying this, and what we say is far from clear. ‘Not very much’ because we do not really say what this thing our realism is about, and we do not say clearly in what sense we are talking about realism. Dummett (1991: 4) thinks that the debates about realism are unequivocal from at least the realist angle:

These are problems about whether or not we should take a realist attitude to this or that class of entity. In any one instance, realism is a definite doctrine. Its denial, by contrast may take any one of numerous possible forms, each of which is a variety of anti-realism concerning the given subject matter: the colourless term ‘anti-realism’ is apt as a signal in that it denotes not a specific philosophical doctrine but the rejection of a doctrine.

Millikan (1986: 194) is even stricter when she claims that there is only one possible form of both realism and anti-realism. In this paper I intend to show that mental realism is far less definite than we might think from reading Dummett or Millikan: in this context there are at least two possible and distinct realist positions. And whilst this paper focuses on mental realism, its lessons can be generalised to the realism/anti-realism debate.

The two senses of realism can be distinguished thus: in one sense, ‘realism’ is an ontological commitment to a class of entities. According to this sense of mental realism the list of existing entities contains mental states, properties, objects, attributes, substances, etc.—depending on our substantive metaphysics. In the other sense, ‘realism’ may mean the realist interpretation of a discourse. In the present case this is the view that folk psychology is a fact-stating discourse. Being an interpretation, this kind of realism, which can also be labelled as factualism, does not necessarily entail any attitude towards the existence of mental entities.

The difference here is between two views of what the realist attitude is about: it may be about how things are in some region of the world, or about how to interpret discourses on a purported region of the world. This brings about important differences in the content of the respective realist positions. They are not mutually exclusive, but they do not come hand in hand. As we will see, a realist interpretation of folk psychology may end up in realism about mental entities, if it is combined with the conviction that folk psychology gives a true description of the mental region of the world. But the two kinds of realism are, nevertheless, independent—it is possible to be a realist about mental entities without interpreting folk psychology realistically. And vice versa: it is possible to interpret folk psychology in a realist way without being a realist about mental entities.

2 Non-Differentiating Positions

To be a common-sense mental realist is to think of mental states as existing in accordance with a literal reading of folk psychology. This commitment is fairly widespread among philosophers of mind too. When they talk about mental realism, philosophers are frequently committed the claim that mental states have representational and/or causal properties. To think that this is part of mental realism is to overlook the important distinction I introduced above. Therefore I call these positions non-differentiating: they do not help us to see the difference between two kinds of mental realism.

According to Kim (e.g. 1995: 202; 1998: 79, 90), to acknowledge mental properties as active parts of the causal structure of the world, i.e. to accept them as causes and effects, is essential in mental realism. This approach seems to echo a causal identity criterion of properties (cf. Shoemaker 1980): properties are individuated by their causal powers, thus the existence of a property can be granted only if it contributes to events. This can be read in two ways, at least. On the one hand, it can be read as a verificationist criterion stipulating that properties exist only when some causal power can be discovered. This ontology may be appealing: it prevents us from postulating mystical properties, and entitles empirical research to reveal what properties there are in the world. But this is a shortcoming at the same time, as it allows us to acknowledge only those properties that are at least in principle accessible from a human standpoint—and it is far from obvious that there are only properties and powers of this kind in the world. Human cognition is limited by human cognitive capacities, and thus not even the fiction of an ideal science can save this reading as long as it remains human science.

On the other hand, the criterion can be read without the shortcomings of a verificationist reading. One can say that the concept of ‘property’ entails that of ‘causal power’. If property is analysed thus, it is a conceptual necessity that, if there are mental properties, they must be causally efficacious. In this case the burden is on the analysis of ‘property’ and not on empirical research into causal relations. Against this background it seems to follow that if we want to be mental realists, we have to acknowledge the existence of mental properties that are causally efficacious by conceptual necessity. By this analysis we implicitly attach a substantive metaphysics to the common-sense view of the discourse which holds that mental states are properties of agents—in some robust sense to be specified by an ontology of properties that takes ‘causal power’ to be central in the analysis of ‘property’.

But in fact, as we shall see soon, this commitment should not be central in any form of mental realism. It puts unnecessary restrictions on it: it makes it necessary to see causal powers working behind ascriptions of mental states. This excludes epiphenomenalist positions, for which mental states are not causally efficacious but inert shadows of causally efficacious processes of a different kind. Epiphenomenalists thus take folk psychology to describe a region of the world, but maintain that its causal idiom is merely apparent; there are no real causal processes underlying it. Thus, despite interpreting folk psychology as a descriptive discourse, they deny the causal efficacy of mental states. And this does not seem to be incoherent: the question of causal efficacy belongs not to the interpretation of the discourse, but to a metaphysics pertaining to the nature of mental states.

This metaphysics is logically secondary: it can come only after we are committed to a realist interpretation of the discourse. We can discuss the causal efficacy of mental states only if we agree first that folk psychology describes some region of the world, with which causal powers may or may not be associated. One has to be a realist about the interpretation of folk psychology prior to asking if mental states postulated by it are causally efficacious. Therefore it is not appropriate to deny that epiphenomenalists are realists merely because they reject causal powers—and here the label ‘realist’ may concern either mental entities or the interpretation of folk psychology. We can say at most, if we accept the causal analysis of ‘property’, that epiphenomenalists do not think of mental entities as properties, but as something else, depending on the ontological categories that they accept.

Let us take one step further. Fodor (1985) gives a definition of realism about propositional attitudes that consists of two components. For Fodor a realist thinks that (a) propositional attitudes are causally efficacious as specified by the laws or platitudes of folk psychology, and (b) they are semantically evaluable. The first may be partly familiar from Kim, but it contains a novelty, namely that it requires causal efficacy in accordance with the platitudes of folk psychology. The problem in this respect is that Fodor here seems to rely on a nomological understanding of causation, which may be appealing, but it is not clear why mental realism is supposed to rely on it. Even if we accept that a commitment to causal efficacy is part of mental realism, it does not follow that it should rely on the nomological model with folk psychological generalisations playing the role of laws. Alternative theories of causation might do as well.Footnote 1

The first component of Fodor’s definition is motivated by an influential account of folk psychology which poses special problems for him. Fodor here seems to share the view that folk psychology is a theory consisting of law-like empirical generalisations. Why should realism about propositional attitudes be influenced by this view? Our psychological practice can be conceived as being based not on some theory, but instead on our empathic abilities, thanks to which we can put ourselves into the position of others in our imagination, simulate their mental states and draw conclusions concerning possible behaviour. Thus causal efficacy belongs to mental states not as specified by folk psychological theory but by our decision-making mechanisms: while applying psychological concepts we do not deploy a theory, but a heuristics, thus it makes sense to say that causal laws are not part of mental realism. By accepting this our realist commitments are not threatened, we do not have to doubt that the mental region of the world is the way folk psychology describes it—merely that it relies on a theory consisting of law-like empirical generalisations.

Fodor’s second criterion is problematic because it is tautological. If there are propositional attitudes then they are semantically evaluable by conceptual necessity, as propositions are semantically evaluable. And as it does not make sense to say that there are semantically non-evaluable propositions (these would not be propositions at all), nor does it make sense to say that there are semantically non-evaluable propositional attitudes.

This problem has further implications. The criterion is not merely empty; it can come close to a definition of mental realism only if we accept that folk psychology represents the bulk of our mental states as having propositional content. But it makes sense to say that although we talk in terms of propositional attitudes, in reality our mental states are not like that; they may be, as Dennett (e.g. 1981: 53) has it, like the centre of gravity of a triangle, i.e. calculation-bound abstract entities:

The center of gravity is just an abstractum. It is just a fictional object. But when I say it is a fictional object, I do not mean to disparage it; it is a wonderful fictional object, and it has a perfectly legitimate place within serious, sober, echt physical science. (Dennett 1992: 104)

Attitudes exist, as Dennett (e.g. 1981: 53; Dennett 1987b: 37) himself is willing to admit, but they do not exist the way folk psychology represents them. His realism is not compromised by this, as he acknowledges mental entities as components of the world; he just denies that folk psychology gives a literally true description of the mental region—it gives a true description, but our mental states are not what they seem to be at first glance.

Clearly, Fodor’s and Kim’s causal criterion is part of the common-sense view, although Fodor’s law-like generalisations may not be. It is also clear that semantic evaluation is also part of it. But these components are logically independent. We can interpret folk psychology in a realist way without accepting either a causal or a semantic criterion. For Fodor, being a realist about folk psychology is to accept it as literally true, and for him the consequences of giving up this kind of realism are intolerable (Fodor 1989: 77). But this could hardly be the key to the realist interpretation of folk psychology, as we have just seen: epiphenomenalists can sensibly deny the causal criterion while holding that they are still realist about folk psychology, their realism being merely of a different kind. Dennett can say something similar while denying the criterion of semantic evaluation. What is then to be realist about folk psychology?

3 Realism About Folk Psychology

It is possible to give a definition of realism about folk psychology that might be acceptable to those mentioned above. Kim’s and Fodor’s realism can be taken to commit us to the existence of a specific property. If we take folk-psychological predicates to express specific properties, then we are realists about folk psychology in the spirit of the common-sense view of the discourse. But this seems to be too restrictive; realism does not require this much. It requires only that there be some properties of the agent that can make folk-psychological propositions true. In Dennett’s case one can say that there are properties of a triangle, allowing for the calculation of its centre of gravity, which can make the calculation true. Similarly, attitude ascriptions can be calculated and can be true: mental states are calculated by reference to patterns of the agent’s behaviour.

These abstract properties are not part of the world in the same way that observable physical properties are, but this is not even necessary for a realist interpretation of the discourse about them. As John Heil (1999: 200) points out correctly:

Realism about a given predicate, ‘Φ’, requires only that ‘Φ’ applies truly to objects in virtue of properties actually possessed by those objects. Realism does not require that ‘Φ’ designate a property shared by every object to which it truly applies.Footnote 2

The case is similar with epiphenomenalism: either we allow causally inert properties, or, if we insist on the causal analysis of ‘property’, we say that mental state ascriptions describe causally inert epiphenomena of causally efficacious processes. Thus, the common ground can be put this way: mental state ascriptions can be true in virtue of the agent’s real properties—without further restrictions on the number or nature of the properties in question.

This said, we have not yet agreed on the meaning of realism about folk psychology. The central claim of realism about any kind of discourse is frequently put forward as a conjunction of two theses (cf. Wright 2002: 207):

  1. 1.

    semantic thesis: the statements in the discourse have a content representing aspects of the real world;

  2. 2.

    metaphysical thesis: the real world is furnished with facts that these statements can represent.

This is the core of the positions discussed above: folk psychology is a descriptive discourse representing some existing aspects of the world.

It seems possible, however, to give a realist interpretation of a class of statements without accepting that there are facts to be suitably represented in the discourse—i.e. it is possible to deny the metaphysical thesis while maintaining a realist interpretation of the discourse. This is to say that given the nature of the discourse it conforms to the semantic claim (i.e. it aims at a true representation of some aspects of the world), but the real world does not contain facts that the statements could represent. Realism construed in this way is an attitude concerning the proper interpretation of the discourse—a view about its nature, and not about the metaphysics of the facts (allegedly) represented in it. It is thus the semantic thesis only that is essential for a realist interpretation of any kind of discourse, and not its conjunction of the metaphysical theses.

For instance, if we consider that statements in the discourse about witches aspire to describe some aspects of the world, we may well believe that it fails in this aspiration—thus we interpret the discourse in a realist way, but take an anti-realist stance towards its metaphysics. Realist interpretation does not require us to believe that the discourse is true, and that the world is furnished with facts that the discourse is fitted to represent: it is enough to believe that the role of the discourse is to represent these putative facts. In the case of folk psychology, eliminativists (Churchland 1981) hold a position like this. They think that folk psychology is, by its nature, a fact-stating discourse, an explanatory theory; but there are no facts in the world that could make its statements true, so folk psychology is false. Eliminativists reconstruct folk psychology as if it was an empirical theory, a case par excellence of fact-stating discourses, and argue that it fails to live up to the standards.

As Bas van Fraassen (1980: 11) aptly puts it: a fundamentalist theist, an atheist, and an agnostic can agree on how to interpret sentences about God or the angels, but they cannot agree with liberal theologians who hold that these sentences should not be interpreted literally. If this casting is carried over to the present context, then Fodor plays the role of a fundamentalist theist, and eliminativists play that of the atheist: they believe folk psychology to be fact-stating in a literal sense. The liberal theologian is Dennett, who rejects the literal reading, but maintains that the discourse is fact-stating in an abstract-calculative sense.

For van Fraassen interpreting a discourse as fact-stating does not amount to realism, as for him it also entails a commitment to truth. I would rather say that in the previous passages we met the realist interpretation of folk psychology in three versions. The criterion of acceptance of a statement follows from the interpretation of the discourse to which it belongs—in the case of descriptive discourses this criterion is truth. Therefore the interpretation of a discourse is logically prior to the criterion of acceptance for statements in the discourse, and following our interpretation we can also undertake or withdraw from ontological commitments to the entities postulated in it. The criterion of acceptance depends on how we look at the discourse in which the statement is made. We accept statements like ‘Sherlock Holmes enjoyed opium’ under very different circumstances if Conan Doyle’s stories are interpreted as historical chronicles and not as short stories, and our ontological commitments to the entities postulated in them may be adjusted accordingly.

These three versions of realist interpretation agree that the essential property of folk psychological ascriptions is their truth conditions picking out states of affairs in the world. They differ with respect to the prospects of fulfilling these conditions and the nature of the facts that could fulfil them. This is why we can talk about three versions of realist interpretation. The first of these, Fodor’s, is the strongest one. It maintains that folk psychology serves to describe the mental constituents of the world as they are, and it succeeds. The second version, Dennett’s, also maintains that folk psychology aims to describe some facts that are accessible from a specific calculative stance, and that it succeeds; but its truth-makers do not reside on the level of the properties ascribed, but on a sub-personal level. The third version, the eliminativist one, holds that although folk psychology is a descriptive theory, its function is to state facts, and in this it fails systematically. They are all realists about folk psychology because they share the view that the discourse is used for stating facts about the real world, only they judge its success, or the way it succeeds, differently.

One final question must now be answered in order to understand realism about folk psychology: what does it mean to say that folk psychology is a fact-stating discourse? This is to say that folk psychology serves the purposes of true description of some aspects of the world, and if it succeeds then its truth is grounded in how things are in the world. This is shared by all three versions: there are functions specifying how folk-psychological propositions should map semantically onto the world. These propositions have correspondence truth conditions as their truth is grounded in how things stand. Thus, following Devitt (1991: 29), we can list the conditions on which the truth of these propositions depends. If interpreted in a realist way a folk-psychological proposition is true if and only if:

  1. 1.

    the sentence expressing the proposition is well-formed,

  2. 2.

    the entities postulated in it exist,

  3. 3.

    there are proper semantic relations between the discourse and the world.

The realist interpretations can be grouped from this angle too. The first condition is typically not challenged: the syntax of folk-psychological propositions does not pose a problem. Fodor’s realism holds that folk-psychological propositions can describe some region of the world and their structure is a good guide to knowing how things stand in that region. For Dennett, the structure of propositions is not a good guide, thus he envisions the relevant semantic relations differently, but believes that these relations hold. Eliminativists argue that the semantics of folk psychology is that of a fact-stating discourse, its propositions should semantically map onto entities that make its law-like generalisations true, but deny that these entities are there in the world.

To sum up this section: realism about a given discourse is a position about its nature—its function or role in our form of life, logic, Wittgensteinian grammar. A discourse can be interpreted in various ways. Realist interpretations answer the question about the relation of world and discourse by saying that it is of a fact-stating character, i.e. its role is to give a true description, which is acceptable by discourse-independent facts, of a region of the world. If we share this conviction, we give a realist interpretation of folk psychology. This is not too restrictive, as the fact-stating character can be reconstructed in various ways. If we give up this conviction, then we deny that the discourse is descriptive of some region of the world. Thus we do not expect the discourse to be fitted to represent facts.

4 Realism About Mental Entities

As I indicated in the introduction of this paper, realism about mental entities and about folk psychology are two distinct positions. The latter is an interpretative stance concerning the nature of the discourse. Folk psychology is a discourse; realism about it cannot be but a way of interpreting it—it is thus a semantic doctrine about the content of psychological propositions. Realism about mental entities is a metaphysical doctrine about the nature of the world, and as such it is independent of semantic considerations.

Realism about mental entities is an ontological commitment to a region of entities from which neither semantic nor substantive metaphysical considerations follow. This position can be put very succinctly as “Mental entities exist.” What they are, what their nature is, how they work, etc. is not a matter of being realist about them, not a matter of ontological commitment, but of an empirical hypothesis or speculative metaphysics, depending on one’s cognitive taste, which is connected to realism only in a contingent way. Thus the thesis can be expressed more informatively as “Mental entities exist independently of what we think about them.”

As this kind of mental realism says nothing about the nature of the entities acknowledged as existing, it can be represented in various forms from dualism to functionalism, or even to computational or connectionist approaches—i.e. to approaches incompatible with the common-sense view of folk psychology. It seems to hold that propositional attitudes are intentional inner states; computational theories deny that they are intentional but admit that they are inner states; Cartesian dualists admit that they are intentional but deny that they are inner states (i.e. within the body). These theories can be understood as metaphysical or scientific “research programmes” and realism about mental entities belongs to their “hard core” (cf. Lakatos 1968).

It is obvious at first sight that realism about mental entities is a philosophically very thin doctrine: it is easier to tell what it does not mean than to tell what it does. One could almost say that it is empty: it says only that there exists something, but as it contains no explanation of what ‘mental entity’ means, we have not the slightest idea of what this thing, of which we say that it exists, might be. This realism, it may seem, is precisely of the kind for which, as Goodman (1978: 20) might say, it is “not worth fighting for.” Taking a closer look, however, reveals implications that give it a real philosophical value.

For instance, it does not require us to be able to represent the entities of the putative region. Quite possibly, we may never achieve full knowledge of them, e.g. because they are entities more complex than a single theory could handle, or because our limited cognitive capacities inhibit access to some relevant facts about them. Given that, there is no guarantee that our folk psychology contains even only a part of what can be known about the mental. But taking an ontological commitment to mental entities urges us to inquire into the question of what they might be—that is why it belongs to the hard core of various research programmes on the mental: realism about mental entities is logically prior to any kind of research into their nature.

This kind of realism does not treat folk-psychological concepts as constitutive in the meaning of ‘mental’. It does not, of course, exclude the mental being what is represented by folk psychology, but this may not be due to conceptual necessity. It allows for an understanding of the mental from outside folk psychology: e.g. cognitive science may have the last word on what mental entities actually are. It is probably empirical research, and not conceptual analysis or speculative metaphysics, that tells us how things are in the world.

Accordingly, realism about mental entities may be taken to be the position starting from which empirical investigation and theory construction become feasible. From this perspective it makes sense to say with Lycan (1988: 31 f.) that mental entities belong to natural kinds that we do not have sound enough knowledge of, and whose nature is supposed to be revealed by further empirical research. Folk psychology may provide initial contingent, and thus possibly false, descriptions of entities to start with, but it has no special importance beyond that. In this case the conceptual resources of folk psychology are not privileged in the explanation of human behaviour and cognitive functioning. If we go this way, our folk-psychological vocabulary can face three possible fates. Empirical research may vindicate the common-sense view of folk psychology; it may revise it, thereby giving psychological concepts a technical content; or it may introduce novel technical concepts more suited to the purpose of further empirical research.

The last two options are not only alien to the common-sense of the discourse but also to our folk-psychological practice. From this angle, it does not seem sensible to say that the concept ‘mental’ can be cut out of the texture of psychological concepts because by doing so we detach it from common usage and start talking about the mental not in its everyday meaning but in a technical sense—we no longer speak the language of folk psychology but of something else. A view like this is shared by those who believe that folk-psychological concepts are constitutive of the ‘mental’: if we do not talk this way we do not talk about the mental (cf. Pettit and McDowell 1986: 3). Giving it up entails a revision of our anthropology based on folk psychology to a significant degree. For instance, we should withdraw from the deeply-rooted idea that due to privileged access we are the best placed to know how things stand in our own mental world.

But this may be tolerated, and seems less irritating if one considers the historical character of folk psychology. Our ideas about the mental, pace Churchland (1981), underwent substantial change over the centuries from Homer to Jane Austen and onwards to Freud. In Homer there is no real mental life: Achilles’s thought takes place in public, personality, thought and action are not separate categories, the idea of a private inner world as we know it today is absent (cf. Havelock 1963: 199).Footnote 3 In comparison with this, our image of the mental is substantially different. If we accept that Homer applies psychological concepts in a way typical of his age, then either there was then no mental life in which the idea of a privileged access could make sense and it appeared later, or we do not have it yet. This suggests that the question about the nature of the mental invites an a posteriori answer, the first step towards which is an ontological commitment to mental entities.

It is thus clear that realism about mental entities is not an empty position free from consequences. Making an ontological commitment creates the possibility of searching for empirical findings about how things are with the mental, i.e. to begin scientific investigations. This kind of realism is independent of the interpretation of folk psychology. A realist interpretation, being committed to the fact-stating character of the discourse, creates a competitive tension between scientific and folk psychology as two incompatible descriptions of the same phenomena. Realism about mental entities suggests the possibility that folk psychology may not be an ideal tool for satisfying our epistemic curiosity, as here it has to compete with empirical research, and its resources may not fit the task. But, being untouched by questions of interpretation, it also affords the chance to separate empirical research and folk psychology—if the latter is not interpreted as a fact-stating discourse.

5 The Realism-Antirealism Debate

Having drawn the distinction in this way between two kinds of mental realism, the usual content of realist and antirealist positions is reworked to some extent. If it is related to realism about mental entities, Millikan’s dictum quoted at the beginning of this paper seems right: there is only one possible way in which both realism and antirealism can be put. Realism and antirealism are both definite doctrines: they make or refuse ontological commitment respectively, and there is no third option. One could perhaps say that it is possible to suspend judgement about whether to make an ontological commitment or not, but this is neither realism nor antirealism: this is agnosticism which is not part of the debate. It seems to belong more to the problem of how ontological commitment emerges: the question why or why not someone believes in, or suspends judgement about, the existence of witches or mental states is not a philosophical one; it belongs more to sociology or psychology. In the case of interpretation neither realist nor antirealist positions are definite. Although realists are committed to the fact-stating character of a given discourse, while antirealists deny it, both options can take various forms, since discourses can be interpreted as fact-stating, and it can be denied that they are, in a variety of forms.

The only claim that realist interpretations of discourses have in common is that the truth of its propositions is systematically connected to relevant facts. Nonetheless, it is frequently doubted that realism entails commitment to the fact-stating character of a discourse. In the present context, Daly’s (2005) argument can be considered as saying that a mental realist can think of the world as containing mental entities but lacking truth-making entities for folk-psychological propositions. Positions like this accept realism about mental entities while denying realism about folk psychology. But if we take psychological propositions as representing how things are in the represented region of the world, then their truth depends obviously on how things are in that region—i.e. a realist interpretation links the truth of the propositions in question to relevant facts. The case is different with realism about entities which does not entail any semantic commitment, not even that their representation is possible at all.

The realism-antirealism debate usually focuses on whether or not semantic and epistemological considerations are primary in questions of existence (cf. Devitt 1991; Dummett 1991). In this respect, realism about entities is close to Devitt’s (1991: 3 f.) position that decision in questions of metaphysics should precede semantic and epistemological questions; thus a commitment to the existence of some entities does not bear on the meaning and truth of propositions. When viewed from the angle of realist interpretation, realism does not imply holding true (or mostly true) the propositions of the discourse interpreted in a realist way. A realist interpretation is compatible with the view that the propositions in a discourse are false—this is the view of eliminativism in the philosophy of mind, and error-theory in ethics (cf. Mackie 1977: 48 ff.).

Drawing a distinction between two kinds of realism is always possible if it makes sense to say that the acceptance of a given proposition does not depend on how things stand in the world putatively represented by the discourse to which it belongs. This is a position frequently argued for in ethics and aesthetics, and I shall suggest that it is possible to argue this way about folk psychology too. Furthermore, it is possible to make this case concerning science. Given the present distinction, everyone interprets science realistically as long as they accept that it is a fact-stating discourse. Thus constructive empiricism (cf. van Fraassen 1980: 11 ff.) can also be read as a kind of realism, which does not question that the propositions of the discourse aim at describing the world. It only says that in some part of the discourse, i.e. the one about unobservable entities, there are standards of acceptance that are different from those in another part which is about observable entities. In the former part the basis of acceptance is empirical adequacy, the fact that our theories about unobservables explain observable phenomena. Propositions about unobservables are systematically connected to observations which are the basis of acceptance of these propositions. So, scientific discourse has epistemic virtues and constructive empiricism shows that the acceptance of statements about unobservables is based not on accessing the relevant facts but on calculating them on the basis of observable phenomena.

The difference between constructive empiricism and the views van Fraassen thinks are realist is not due to their different views about the nature of the discourse but about the nature of the facts on which acceptance is based. Whereas, for a constructive empiricist, sentences can be literally true, their acceptance does not hang on the truth of the proposition they express, but on another pragmatically equivalent proposition. Acceptance depends not on facts about unobservables, but on a set of facts, or on the truth of a complex proposition, about observables. Therefore, the acceptance of sentences about the unobservable domain remains systematically connected to truths about observables, but they state different facts.

Antirealism in this context would be a position that science is not fact-stating. It could hold, for instance, that its aims are not epistemic but ideological: it serves the purposes of disguising and maintaining certain forms of social power. Habermas (1970), for example, sketches a picture of science as representing itself as objective knowledge standing above ideologies. Thus science itself becomes an ideology in terms of which an increasing number of problems require technical solutions. In doing so, it makes practical social problems less visible and thus serves particular class interests. This process is enhanced by mass media, educational institutions, labour associations, and channels of communication—agencies distorted by science that together enable the ruling classes to maintain their social power. This is an antirealist interpretation of the scientific discourse which denies that it aims at describing facts in the world and determines its function by a reconstruction of its social role which is alien to its realist interpretations. But even if this view is accepted, one can still maintain that the entities postulated by science exist— only science is ill-suited to tell truths about them.

6 Folk Psychology and Scientific Psychology

It is frequently taken for granted that both folk psychology and cognitive science aim at the prediction and explanation of behaviour, so they are about the same thing (cf. Stich 1978). Therefore they have to be reconciled with one another, or one of them should be eliminated for the sake of the other. But it makes sense to believe that this is a false dilemma. Separating realism about folk psychology and about mental entities gives us a chance for separating the fields of inquiry where they are in charge, which promotes peaceful coexistence. Besides, it thus becomes possible to investigate the features of folk psychological features and to revise its common-sense view if necessary, without bearing on scientific psychology at all.

Basically, ascriptions of mental states are made in two contexts: on the one hand, in the explanation and prediction of behaviour, i.e. in epistemic contexts; on the other, in making moral judgements, i.e. in evaluative contexts. Giving up the idea that folk psychology is in some sense a fact-stating discourse amounts to withdrawing from its epistemic uses, and resolving the competitive tension between folk psychology and scientific psychology in the explanation of mental entities and of behaviour. If mental entities are treated as belonging to natural kinds, then research on them does not presuppose or require anything from the conceptual resources of folk psychology. Once viewed from this angle, scientific psychology focuses with its technical concepts on explanation and prediction, while folk psychology turns out to be the discourse offering concepts necessary for understanding and evaluating persons in the social world. Thus we give up the idea of folk-psychological explanation and prediction of behaviour, but retain the practice of folk-psychological interpretation—if it is reconstructed in a way so as not to rely on the alleged epistemic uses of the discourse.

A promising way of doing this is to take a fictionalist stance towards folk psychology, modelled on moral fictionalism (cf. e.g. Joyce 2001; Kalderon 2005), of which I can only give a sketchy outline here. The central claim of fictionalist poisitions consists in the conjunction of the following three theses:

  1. (a)

    Folk psychological sentences express propositions.

  2. (b)

    As such they are truth evaluable.

  3. (c)

    Truth value is not their essential property.

Now, (a) distinguishes fictionalism from expressivism, and up to (b) it goes hand in hand with error theory (or eliminativism), but with (c) it diverges from it. The fictionalist account thus holds that folk psychology provides representations that may not represent anything, and whose virtues are not epistemic.

From this angle, folk psychology can be taken to be an everyday hermeneutic device facilitating non-epistemic understanding. Representing the springs of behaviour by well-known concepts of folk psychology amounts to representing it as coherent and meaningful, as behaviour that makes sense. This kind of understanding can be seen on par with, e.g. understanding extreme bad weather as a result of the activity of witches: it can provide affective resolution in those cases where unfamiliar events take place, but without providing knowledge about it. Also, by representing behaviour with folk psychological concepts it becomes morally evaluable. Behaviour is thus represented as an action of a person, and not as a simple event in an impersonal causal chain. Moral evaluation needs this kind of representation, otherwise behaviour cannot be seen as a person’s action, only as an event happening to her. Due to the evaluative language of motivations and intentions, folk psychological representations can be used to express and influence moral sentiments.

A good model of how folk psychological sentences can do that could start from Davidson’s (1978) account of metaphors (putting aside the question if it is correct as a theory of metaphors), which Kalderon (2005: 127 ff.), too, uses for making a fictionalist sense of moral sentences. According to this view, moral sentences are quasi-assertions that do not assert a proposition at all but convey some noncognitive attitude. Prima facie, folk psychological sentences, just like moral sentences, seem to express fact-stating propositions. Still, for the fictionalist, it is not their semantic content that is relevant for an understanding of what role they really play, but a kind of non-semantic, affective content. This content is not entirely independent from the semantics of the sentences, as the conventions of folk psychological discourse (generally known as the platitudes of folk psychology) stabilise the use of its expressions. For the fictionalist, folk psychology is a convention-based expressive code with purely social uses, and not a metarepresentational device that could tell truths about our cognitive architecture.

If understood in this way, neither the hermeneutic nor the evaluative uses of folk psychological representations presuppose their truth, only their ability to express and cause the relevant sentiments. On the face of it, they are representations, but their use does not rely on the truth of their representational content. Although they may not be true representations, they still can play an important role in social interaction: due to their affective content they can motivate behaviour, but not as a result of calculating with propositional attitude ascriptions, but by conveying sentiments. In this sense, folk psychology is a useful ideology that introduces us as coherent and moral agents, and makes the social world liveable—but does not serve the purposes of the true representation of the causes of behaviour.