1 Introduction

Things such as sleep and comas raise serious challenges for phenomenologists, because those occurrences rely so heavily on first-personal experience—and one simply cannot reflect on oneself during these unconscious states. To address such issues, Husserl describes the loss of consciousness as “the relaxation of the will” (Willensentspannung) (Hua XXXIX, 2008, p. 591). Maine de Biran (1841) proposes a similar view, arguing that sleep consists in “the suspension of the will.” Brian O’Shaughnessy (2000) and Matthew Soteriou (2013) have more recently suggested that mental actions constitute wakeful consciousness. Indeed, the idea that agency is constitutive of phenomenal consciousness even finds application in current clinical practice. Among patients with disorders of consciousness, those in “Minimally Conscious States” (MCS) can follow commands and react to emotional stimuli. They behave “purposefully” (Leon-Carrion et al., 2008). Those in “Unresponsive Wakeful States” (UWS)Footnote 1 can open their eyes and have sleep/wake cycles, but they behave only reflexively (Giacino, 2005; Giacino et al., 2002).

Jacob Hohwy (2009) raises a concern about these proposals. If one tracks phenomenal consciousness by agency, “now the mark of consciousness is voluntary control rather than what philosophers refer to as phenomenality […] Finding the neural correlates of volition, the NCV, is a valid and important but different project from finding the NCC [neural correlates of consciousness]” (p. 434). Hohwy’s criticism would apply if agency and phenomenal consciousness had only a loose connection, like that between the wind blowing and leaves falling. In principle at least, one event in such a situation can occur without the other. They are merely “nomologically” connected by a causal law. If the relationship between agency and phenomenal consciousness is tighter than that, however—if it is “metaphysically necessary” (Ruben, 2018, p. 118), or in other words, if agency is constitutive of phenomenal consciousness just as extension is constitutive of color—then one could legitimately track phenomenal consciousness by agency.

This article takes on just this central question: to what extent and how is agency constitutive of phenomenal consciousness?Footnote 2

Here, it is necessary to distinguish two kinds of constitutiveness, which I call “conceptual” and “instantial.” Conceptual constitutiveness is the stronger of the two notions. This constitutiveness means that one concept contains another as an indispensable part of itself. For example, the notion of spatial properties is conceptually constitutive of that of physical objects, because the latter notion is an indispensable constituent of the former. This reasoning also applies to Ruben (2018)’s example of trying or power exertion and action, because the notion of action already contains the fact that an agent makes an effort to bring it about. Color and extension, in contrast, as abstracted from their instantiations, do not contain one another; but, whenever color is instantiated, so must be extension. In other words, if a notion y could be instantiated only in relation to another notion x, then notion x is instantially constitutive of y. Here, the property relational to x is an indispensable constituent of y with respect to y’s instantiation. As such, instantial constitutiveness is the weaker of the two notions. Two notions in event causation, such as “sunrise” and “wake up,” do not in general satisfy the conditions even for instantial constitutiveness, because one instance can in principle occur without the other. Hence, instantial constitutiveness is weaker than its conceptual counterpart, but is still a stronger notion than that of event causation. In fact, Husserl’s definition of “founded by μ” as “in need of a complement from μ” (ergänzungsbedürftig durch μ) (Hua XIX/1, 1913, Teil III, § 14) addresses instantial, but not conceptual, constitutiveness—spatial properties are not complementary for physical objects, while the complementarity relation holds for color and extension with respect to their individual instantiations.

The goal of this article is to argue that agency is instantially constitutive, but not conceptually constitutive, of phenomenal consciousness; phenomenal consciousness may be instantiated only if it stands in some essential relation to agency. In other words, the role it plays for agency is indispensable in every instance of phenomenal consciousness.Footnote 3

The article first offers a critical review of the proposals already mentioned in their capacity as answers to the central question. Section 2 begins with Husserl’s “relaxation of the will.” It shows that, while his approach is unaffected by problems found in other authors, it is so general that it cannot guide an informative answer. It says nothing about why and how agency is responsible for phenomenal consciousness.

Section 3 turns to Maine de Biran. For him, sleep means that the will no longer exerts an effort on any organs, leaving them to work by their internal forces. This argument is speculative, however, and relies on a confusion between habitual actions and mere reflexes. But it still makes an important point: the will is presented as the “hyperorganic” force bringing “organic” forces to a higher degree of consciousness. It is agency, in other words, that raises lower-level consciousness to a higher level.

Section 4 then considers the work of O’Shaughnessy and Soteriou. Their arguments divide into the Argument from Synthesis and the Argument from Self-Consciousness. The first holds that some kind of synthesis is necessary for wakeful consciousness; mental action is responsible for that synthesis. The second claims that wakeful consciousness is always self-consciousness, where the “self” here is understood as an agent. By applying Husserl’s distinction between actuality (Wirklichkeit) and phantasy, this section claims that the Argument from Synthesis does not correctly specify what kind of synthesis makes an essential contribution to wakefulness and phenomenal consciousness. Similarly, the argument does not explain why that kind of synthesis must take the form of mental action. On the other hand, Soteriou’s Argument from Self-Consciousness wrongly assumes that the past self, by placing constraints on the current self, must be an agent. It may instead be a merely receptive self.

The following section, Sect. 5, attempts to recover the virtues of these two arguments by distinguishing between different notions of the self and by developing an account of action based on agent causation. On that account, an agent exerts his power for certain reasons, but factors that are guiding the agentive power exertion, in order to qualify as reasons, have to be understood as such. Phenomenal consciousness, this section proposes, is what makes that process possible. It does so by enabling either an understanding of potential reasons for the agent to exert his power, or the agentive power exertion itself. In the revised version of the Argument from Synthesis, “synthesis” is taken as the connection between reason and agentive power exertion, and in this context the self is either the “receptive” self, confronted with reasons, or the agent. These points help to make sense of agency’s role as constitutive of phenomenal consciousness. The results of this section offer a partial solution to panpsychism’s Combination Problem (Chalmers, 2016), in its attempt to solve the hard problem of consciousness along the lines of the Integrated Information Theory (Oizumi et al., 2014; Tononi, 2017; Tononi et al., 2016). How can micro-consciousness contribute to macro-consciousness? The answer may be in recognizing agency as the means for granting the transition of micro-consciousness to macro-consciousness, by integrating reasons and power exertion.

Section 6 discusses the rationale behind treating clinical patients exhibiting purposeful behavior as “minimally conscious.” Assessing these patients is difficult: are they truly conscious, or conscious only in a diminished sense? Or is it that, because they cannot report their experiences, their behavior should be described as that of a “zombie” (Naccache, 2018)? If the impossibility of self-reflection forces a first-personal approach to confront a limit here, then a physiological approach faces the same limit. Inferring patients’ experiential states based on neuroimaging results would be illegitimate if one held a generally skeptical attitude toward the second-person understanding. I argue instead that there is justification for attributing phenomenal consciousness to patients with purposeful behavior, like command following, so long as the command builds a potential reason for agentive power exertion.

Such is the plan of the paper. Before continuing, however, a clarification on terminology.

First, this article speaks deliberately of pushing the first-personal approach to its limit, rather than pushing phenomenology to its limit, because phenomenology does not reduce to any particular approach. This is the view of Michel Henry (1990), who argues that it is the task and not the method that defines phenomenology. According to Husserl in First Philosophy, transcendental phenomenology should be the “absolute, universal Geisteswissenschaft” that attempts to understand every meaning through intentional correlation (Hua VIII, 1996, p. 276, p. 287). In theory, one could apply all available methods to accomplish this task, such as first-personal reflection, second-personal empathy, investigation into transcendental facts and the eidetic variation (Hua XXXV, 2002, p. 111, p. 344). Hence, the fact that humans cannot self-reflect during dreamless sleep does not imply that such an occurrence is beyond the scope of phenomenology.

Thus, the term “phenomenology” in this article should not be conflated with phenomenal consciousness. The latter refers to subjective states of experiencing, while the former denotes a task with its available methodology.

The second clarification is that, for terms such as “consciousness” and “will,” one should not assume that different authors use the words equivalently.

As I shall show later, Husserl would take both dreaming and wakefulness as conscious states, and for him only a dreamless sleep would imply total loss of consciousness. Maine de Biran denies the existence of dreamless sleep, in contrast, and so takes consciousness as equivalent to the wakeful state. O’Shaughnessy and Soteriou speak similarly. The use of “consciousness” here is closest to Husserl’s meaning. Indeed, one can study the neural correlates of consciousness in two ways. One is “content-based,” and considers what is required for a specific content to be conscious by using experimental paradigms such as binocular rivalry. The other “state-based” way contrasts the overall conscious state with non-conscious ones, such as deep sleep and coma (Hohwy, 2009). In referring to agency as constitutive of phenomenal consciousness, the arguments here are not suggesting that every bit of experience is subject to agentive control. Rather, the idea is that the overall state of consciousness requires agency.

Maine de Biran would reserve terms such as “will,” “volition,” and “agency” for deliberation, decision-making, and the resulting execution, as he takes all habitual behavior to be non-agentive. O’Shaughnessy and Soteriou differ here and would claim that thinking, even when one has not decided to do so, also qualifies as a mental action. For Husserl, the question is complicated, because his understanding of the will undergoes an extensive transition. In the 1908–1914 Lectures on Ethics, the will means either “resolution-will” (Entschlusswille), which works in deliberation, or “action-will” (Handlungswille), which carries out the result of deliberation (Hua XXVIII, 1988, pp. 108–109). Husserl confines himself to examples of bodily action (Ibid., p. 107, p. 110), and demands that the action’s aim be limited and thus realizable (Ibid., p. 104). Later, however, he broadens his understanding of the will in the following directions:

  1. (1)

    In Active Synthesis, Husserl argues that “the will is not a particular kind of consciousness” (eine eigene Weise des Bewusstseins), but rather a “form of activity” that “can appear everywhere” (Hua XXXI, 2000, pp. 9–10).Footnote 4 Around the same time, in 1919, he also claims that “tendency and self-striving are modes of consciousness, modes of every one consciousness” (Hua XLIII/III, 2020, p. 118). In other words, Husserl says that the will should not be taken as a particular kind of consciousness juxtaposed with cognition and emotion. Rather, the will is the ubiquitous “form” or “mode” for cognition and emotion, which are the “material.”

  2. (2)

    Husserl’s theory starts including not only bodily actions but also mental ones (Ibid., p. 5, p. 54, p. 107, p. 128; K II 4/10–12).

  3. (3)

    Husserl, especially in the Kaizo-articles, would allow infinite or idealized aims into his notion of the will (Hua XXVII, 1989, pp. 36–38).

  4. (4)

    Moreover, later developments in Husserl’s thought no longer limit the notion of action to the execution of decisions. He instead speaks of “drive-actions” (Triebhandlungen) and “mechanical actions,” such as habitually riding a bike (Hua XLIII/III, 2020, p. 81, p. 245). He uses terms such as “will-passivity,” “passive will,” and “drive-will” to capture those phenomena (Ibid., p. 415, pp. 418–419). In other words, actions are not limited to the “full-blooded” type (Velleman, 1996), where they are controlled by reflective states. Habitual actions such as brushing one’s teeth are actions and not just mere reflexes.

  5. (5)

    Finally, the tendency from one experiential state to another, thought of as a kind of striving, has intentionality. According to Husserl, though, this intentionality is non-objectifying, and one must distinguish between the objectifying intentionality of an act and the intentionality of such strivings (Ibid., pp. 309–310).

Two caveats apply to this discussion. First, Husserl is not always consistent with his own terms; one must avoid simply labeling as “earlier” and “later” the two Husserls who endorse the narrower and broader notions of the will. Husserl is experimenting with different theoretical options, and it may be that he maintains the narrower definition of the will in one place while considering the broader one a mere metaphor (Ibid., p. 381); even then he can reaffirm in another place that those strivings and aims in cognition and emotion are true strivings and aims, rather than metaphors (Ibid., p. 333). A more appropriate characterization might be the “more conservative” and “more liberal” Husserl. In any case, the lesson from the broader notion is not that we should abandon the traditional trichotomy of cognition, emotion, and will. The broader notion instead just challenges the intellectualist hierarchy of these three—that the will is grounded by cognition, while cognition can exist without the will—and seeks a restructuring.

This paper endorses the broader notion of the will and defines agency as an agent’s capacity to perform an action. This is not equivalent to the sense of agency, which is the awareness of one’s control over an action. According to the popular comparator model, the sense of agency is a result, rather than a condition, of action performance (Blakemore et al., 2002; Synofzik et al., 2008). Here, it will be left open as to whether this is always the case. At the very least, one must not confuse agency in this article with the sense of agency.

2 Husserl: loss of consciousness as the relaxation of the will

Husserl attempts to describe the state of dreamless sleep with the experience of falling asleep. When falling asleep, “more and more I stop (einstellen) my activity.” One “reposes” one’s interest, “letting it sink.” One does not “play with thoughts” unless they help to reduce one’s “participation” by increasing fatigue. For example, one thinks of a complicated mathematical problem to make oneself sleepy, and although “there is still a remaining activity,” this is “in a specific mode,” namely “in all [my] life of willing, [I] let the interest, the self-participation totally sink” (total in allem Wollen in eins das Interesse, die Ich-Beteiligung sinken Lassens). This mode does not just modify the energy of this specific interest. It also adjusts “the entire will-life in one” (das gesamte Willensleben in eins). When one falls asleep, “now no interest is exercising influence in actual willing and doing, now I am striving for nothing (will ich auf nichts hinaus), [and] I am not at all [living] in a [state of] striving, a desiring, a wanting-to-realize (Verwirklichenwollen).” One could also say that “the willing ceases more and more” (Das Wollen hört immer mehr auf) (Hua XXXIX, 2008, pp. 589–590).

This “letting it go” (fahrenlassen), as Husserl puts it, means that “the willing self becomes passive in a certain sense.” It differs from setting the interest aside in exchange for something else, such as playing a video game rather than performing an occupational pursuit. It is not about letting go of one particular interest for another, but rather involves a state characterized by becoming a passive self in every bit of experience.

The “sinking” of the will into such a passive mode accompanies the reduction of an object’s attractiveness, which may otherwise provide a reason for one to act. For example, a delicious dish may attract one’s attention when awake, but one is less moved to enjoy it when falling asleep. Hence, “in letting go, I also let go the affecting entity.” “If I relax myself,” Husserl says, “if I am in the mode of totally relaxed interest, then the affections also lose their correlative tension of appeal” (Hua XXXIX, 2008, p. 591).

Letting the will sink and go (Sinken- und Fahrenlassen) is a state where one is “still wakeful but falling asleep,” with sleep as the limit of this transitional process. The limit means the total “relaxation of the will” (Willensentspannung) on the one hand and, correlatively, the total “relaxation of affection” (Affektionsentspannung) on the other (Hua XXXIX, 2008, p. 591).

But sleep is not a homogeneous state, as Husserl distinguishes dreaming from dreamless sleep. Only dreamless sleep represents the “utmost limit,” which implies “no more sinking, but [is] the sunken state, no more letting go, releasing from every grip (aus jedem Griff Entlassen), but the state of having been released (Entlassen-haben)” (Hua XXIX, 1993, p. 337). If Husserl is right, then dreaming does not imply a total absence of agency, as one would not have yet reached the utmost limit. Agency remains during dreams and terminates at the stage of dreamless sleep. Husserl likens death to dreamless sleep: “isn’t death still the brother of sleep? Isn’t death, viewed from the inside, also a letting-go of the world […]?” (Hua XXIX, 1993, p. 338). Death is an “involuntary letting-go, involuntary passive alteration of interests,” which is just like a dreamless sleep, where interests are totally diminished (Hua XLII, 2013, p. 21).

According to the above passages, the loss of phenomenal consciousness in general for Husserl is a relaxation of the will. He identifies dreamless sleep as an utmost limit and separates that state from dreaming, presenting death as dreamless sleep. This approach differs from that of authors such as Maine de Biran, O’Shaughnessy, and Soteriou, who restrict the role of agency to wakeful consciousness, and explain dreams as the absence of agency. Arguably, though, dreams are phenomenally conscious. In other words, Husserl’s thesis is more radical than those authors: for him, agency is constitutive of phenomenal consciousness itself, rather than of wakeful consciousness. It is the more radical thesis that this article defends.

Husserl does not specify the notion of “will” here. As we shall see, Maine de Biran, in contrast, restricted the notion to “full-blooded” actions (Velleman, 1996). This understanding excluded habitual actions, which is a problematic position because habitual actions should be distinguished from mere reflexes. O’Shaughnessy and Soteriou reserve agency’s contribution to wakefulness to mental actions. As Husserl does not limit himself in this way, his view does not face the challenges these other authors do. But the generality has a price. Husserl’s account is in fact too general, as it cannot explain how agency plays a constitutive role in phenomenal consciousness; such is the case for Husserl’s published manuscripts, at least. His work sheds no light on what was later called the “hard problem” of consciousness (Chalmers, 1996): how does phenomenal consciousness arise from physiological mechanisms? Sect. 5 will address this weakness in Husserl’s view, along with its solution.

One might worry that I have underestimated the differences between sleep, coma, and death, given that in the texts cited above Husserl characterizes death as “the brother of sleep” and explains it similarly in terms of “letting-go of interests” (Hua XXIX, 1993, p. 338; Hua XLII, 2013, p. 21). Now, that may be Husserl’s underestimation, but it is not the position taken here. Section 6 explains that many things commonly taken as comatic states actually involve very different cases, such as MCS, UWS, actual coma, and brain death. And all these are to be distinguished from locked-in syndrome. In addition, electroencephalogram (EEG) recordings show that sleep has different stages. These differences are not clear in Husserl’s account, probably because of the dominant role first-personal reflection plays in his discussions and because of the neuroscientific and technological limitations of his time. Nevertheless, I still endorse one point in Husserl’s treatment: that these phenomena, despite their differences, all reveal a diminished degree of consciousness, and the diminishment goes hand in hand with a decline in agency. It is therefore intriguing to consider agency as the factor responsible for an individual’s overall state of consciousness. The problem is how to do this, given that Husserl has no ready answer.

3 Maine de Biran: sleep as the suspension of the will

Although Maine de Biran (1841) offers a notion similar to Husserl’s relaxation of the will, no evidence exists to suggest that Husserl connected his view to Maine de Biran’s. In his article “New Considerations on Sleep, Dreams and Somnambulism,” Maine de Biran argues against the view that sleep is the sum of particular periods of sleep of each organ (p. 216), and so explains sleep in purely physiological terms. He intends to unify physiological and metaphysical views, taking the latter as intending to “determine the status of the soul or the thinking principle in sleep,” while the former aim at “the status of the body or the functions of the principle of life.” (p. 218). He argues that.

sleep consists in nothing other than the suspension of the state of effort (le suspension de l’etat de d’effort), namely, of the action which is present in a will or the motoric force upon the organs that are submitted to it. (p. 221)

This view requires comprehending Maine de Biran’s conception of the “nature of human beings.” He believed that humans comprised two principles: one “subordinated to the vitality of the organs and the animal sensibility,” and the other “liberated from this principle to a certain degree and obeys the hyperorganic laws” (p. 295).Footnote 5 The first principle rules the internal forces of organs, which “offer the will such a degree of inertia or resistance, without which the feeling of effort would never occur, and without which, consequently, the will […] would have no time to grow” (pp. 229–230). The will is the “hyperorganic” force acting on the organs, but that action cannot occur without organic forces reacting to the will. “Without this reaction, on which the feeling of the resistance and of the inertia of the body depends, the effort of the will or of myself (moi) could not commence,” and the organs would “no longer inform the soul about its existence” (p. 225). According to Maine de Biran, sleep means that the hyperorganic principle has been suspended and that only the organic principle is operating (p. 295).

Maine de Biran classifies the factors that may provoke sleep into active and passive (p. 222). Passive factors include drinking alcohol, poisoning, and diseases. Active factors involve the will (p. 228). According to Maine de Biran, “natural sleep results immediately from the fatigue of prolonged wakefulness” (p. 226). He illustrates the point with descriptions like the following—“through the contraction or accumulation of motoric forces in those particular organs,” prolonged effort.

augments their irritability or their contractibility itself, in such a manner that there would be no place for the reproduction of the effort, and the will, retreating gradually from its action […] finally loses the possibility to exercise itself or to restart new contractions in the organs. (p. 230)

Because the will has been acting on the organs for a long time, the accumulated forces “render them independent of action from the motoric center or the instruction of the will” (p. 231). The will accordingly loses its influence on the brain, which was at that time the organ considered closest to the soul, and further on the motoric organs as well (pp. 228–229).

Maine de Biran offers similar explanations for habits and somnambulism. He claims that through practice, a movement would “accomplish itself in the limbs, without there being the will or the self who directs [the movement].” Likewise, specific movements in somnambulism stem not from the will acting but from the absence thereof. The body moves on its own (pp. 290–292).

Maine de Biran’s approach differs from Husserl’s, who argues that only dreamless sleep, as the utmost limit of will relaxation, can lack agency entirely. Maine de Biran, in contrast, insists that the will is absent in dreaming and characterizes dreams as impressions of visual objects occurring “despite my will or without its assistance.” For Maine de Biran, “the entire active modes” are “the attributes of myself” (p. 250). He also believes that dreams cannot involve attention, which may be considered a kind of mental action. He has two reasons for this view. First, dreams are often bizarre or extravagant—one can dream of talking with a dead person while not dreaming of that person as dead (p. 239). Second, memory requires attention, but because it is impossible to remember what one dreams (p. 243), how could one talk about the properties of dreams if one cannot remember them at all (p. 245)? Maine de Biran says that we may encounter certain circumstances in our waking life that could “awaken immediately the associated ideas that have composed the entire dream” (p. 246). He also claims that “the entire sleep should be filled with dreams” and that, whenever one remembers a dream, the recollection occurs only because sleep has not been perfect (p. 245). If sleep is deep enough, he says, one will have no memory of it.

The first observation to make about Maine de Biran’s perspective is that he supports only a weak version of agency’s constitutive role for consciousness. He does not argue that a suspended will is responsible for the loss of phenomenal consciousness, as Husserl did; he limits himself instead to distinguishing wakeful from non-wakeful consciousness. This distinction results, in part, from his denial of dreamless sleep. For Maine de Biran, the will is utterly absent in dreams, but he holds that consciousness persists throughout wakefulness and sleep, albeit in different forms. There is consequently no point in speaking about the loss of phenomenal consciousness.

It must be admitted that Maine de Biran’s position is highly speculative. How can one determine whether someone in a deep sleep is dreaming? An external observer cannot access that information, but it may be impossible for individuals to access their own recollections as well. If a sufficiently deep sleep would not allow any memory, as Maine de Biran says, then at what point does consciousness vanish? In a coma? With anesthesia, or death?

Nor is Maine de Biran’s argument to exclude agency from dreaming convincing. The analogy of dreams with impressions of visual objects is misleading. A distinctive feature of some mental actions is that they do not intervene on our mental states. Some such actions are means to endow an intentional object with a certain meaning. Husserl, for instance, seems to regard emotion and cognition as having a striving character (Hua XLIII/III, 2020, p. 152, pp. 148–164, p. 333, pp. 482–484). This claim may illuminate the understanding of particular mental actions. To perceive a flower is to be aware of it, but its existence must be validated. One corroborates that existence by getting the flower “verified as an existent flower.” One strives for that aim, but doing so does not alter the flower’s properties. The perceptual experience of a visual object is therefore not completely receptive. Likewise, during dreams it may be that striving occurs in a similar way, as when one strives to know what is going on when a suit-wearing man in one’s dream shoots another person of whom one is also dreaming. This phenomenon is to be qualified as mental work.

Maine de Biran also erred in excluding attention from dreaming. That dreams are bizarre is no evidence against the presence of attention, unless one also assumes that attention involves rationality and common sense. But a person can pay attention to the bizarre images of a drug-induced hallucination. A similar point holds for Maine de Biran’s idea about memory and dream. One’s waking circumstances may cause the recollection of a dream, but how would that be possible if the dream were not remembered at all during sleep? If memory requires attention, as Maine de Biran assumes—which is itself doubtful—then there is no reason to believe it is not functioning in dream as well.

Furthermore, Maine de Biran’s explanation for habits and somnambulism rests on misconceiving habitual action. “Full-blooded” actions, monitored and controlled by a higher, reflective mental state (Velleman, 1996), are not the only kinds of actions. Husserl calls that kind of control “active motivation” or “reflective self-determination” (Hua XXXVII, 2004, pp. 107–110, pp. 169–166, p. 239, p. 332); for example, executing my new year’s resolution is a full-blooded action, controlled by the reflective state of planning. Beyond these there are also “passive actions,” such as the habitual behavior of brushing one’s teeth. These are not mere occurrences (Hua XLIII/III, 2020, pp. 10–12, p. 81). Husserl uses terms such as “passive activity” and “will-passivity” to describe these actions (Hua XLIII/III, 2020, pp. 418–420, p. 486). Passive actions differ from mere reflexes. Yanking my hand from a fire before I become aware of having done so is a reflex, but action is always performed for a reason. Even for passive actions, this reason is conscious.

Effort, which manifests the exertion of agentive power, is not completely absent in passive actions. Here, one must distinguish between the effort to “find a way” to do something, and the effort to overcome one’s intrinsic resistance to a task. Suppose one is learning how to ride a bike. One must figure out how to exert one’s effort fluently. But even when the skill is learned, the action is not entirely effortless. Wishing to ride a bike does not entail one’s riding, and the difference is that one must exert power to overcome the intrinsic resistance to the task. Maine de Biran confines agency to full-blooded actions, and thus conflates habitual action with reflex.

There is something salvageable from Maine de Biran’s account. For him, wakefulness means that the will acts upon the organs, while the organs themselves, independently of the will, are responsible for consciousness in dreaming. If wakefulness is a higher level of consciousness, does the will raise the lower level of consciousness to the higher? If so, then by replacing the notion of wakeful consciousness with that of phenomenal consciousness in general, one arrives at the thesis that agency raises the lower level to the higher level of phenomenal consciousness. This view is viable as long as one discards Maine de Biran’s model of mind and body, and the crucial insight behind the thesis underlies the discussion in Sect. 5.

Maine de Biran says that the hyperorganic will acts on the organic forces during wakefulness, and he identifies the self with the agent exerting effort. In these ideas, he anticipates the Arguments from Synthesis and Self-Consciousness, which are the topics of the following section.

4 O’Shaughnessy and Soteriou: mental actions are constitutive of wakefulness

Husserl does not specify what kind of agency constitutes phenomenal consciousness. O’Shaughnessy (2000) and Soteriou (2013) do. O’Shaughnessy argues that bodily action is not constitutive of wakeful awareness, because we can be awake even when “bodily will” is totally paralyzed. Hence “the contribution to [wakeful] consciousness is solely made by mental will” (p. 87).Footnote 6

It is not possible here to provide an exhaustive textual analysis for each of O’Shaughnessy’s and Soteriou’s arguments. This section treats the most plausible ones and summarizes them in two lines of reasoning.

4.1 Argument from synthesis

The first line of reasoning is called the Argument from Synthesis:

  1. (1)

    Wakeful consciousness requires a kind of synthesis.

  2. (2)

    Mental action is responsible for this kind of synthesis.

  3. (3)

     Hence, mental agency is constitutive of wakeful consciousness, if mental agency is considered the capacity to perform mental actions.

What kind of synthesis does the argument involve? Husserl’s distinction between actuality (Wirklichkeit) and phantasy can help to answer this question. The actual world is always a unity, but different phantasy worlds, like those of Jane Eyre and Gone with the Wind, can be disjointed (zusammenhangslos), unless we wish to merge them. A tree destroyed in Jane Eyre’s world, for example, has no effect on a counterpart tree in Gone with the Wind’s world. “The fulfillments and disappointments of intentions which are constitutive for one phantasy world can never reach in the intentions that are constitutive for another phantasy world” (Hua XXXIII, 2004, p. 336). The actual world has a unique timeline, but each phantasy world has its own (Hua XXXIII, 2004, p. 358). Those worlds could merge, according to Husserl, but to merge them would be unnecessary (Hua XXIII, 1980, p. 561). The extreme case of disjointed worlds occurs in dreaming, where people make bizarre conceptual and spatial connections that neuroscientists describe as “binding errors” (Revonsuo & Tarkko, 2002).

In general, phantasy differs from dreaming: one can imagine things while awake and be aware of oneself as imagining. The self who imagines a dog is a self who is aware of the imagining self visualizing a non-existent dog, and is not an imagined self perceiving an unreal dog. In other words, wakeful phantasy always involves being aware of the contrast between the imagining and the imagined self. In a so-called absorbed dream, on the other hand, one is not aware of that contrast, and does not become aware of the fact of one’s dreaming without waking up from the dream. Phenomenologists have used different terms to capture this state of affairs. Eugen Fink (1966), for example, proposes that “dream is nothing else than an absorbed phantasy,” and that what distinguishes dreaming from wakeful phantasy is that “the most extreme absorption of the dreaming self is the condition for the dream” (p. 63, emphasis in original). Theodor Conrad (1968), another student of Husserl, expresses a similar idea. He says the imagined world is “there” for the wakefully imagining self, while this “there” becomes a “here” for the dreamer. In particular, a dream experience is “one-leveled” (auf einer Ebene), without the contrast or distance between the dreaming and the dreamed self, or between the “here” and “there,” so long as the subject is absorbed in the dream. On the contrary, a wakeful imagination like a daydream is always “two-leveled” (zwei-ebenig) (p. 61, p. 66). More recently, Nicolas de Warren (2012) argues that the dreamed ego is awake during an absorbed dream, in the sense that it is pre-reflectively aware of itself. But the dreaming ego remains asleep (p. 468). The difference between an absorbed dream and a wakeful imagination is that “the experience of oneself at a distance […] is lacking” (p. 471). That distance is just the contrast between the dreaming self and the dreamed one. Contrary to Zippel (2016), who claims these authors distance themselves from Husserl in stressing the dream’s presentness, it is clear there is no conflict here. Authors such as Fink note the lack of contrast between the dreaming self and the dreamed one during absorbed dreams, while Husserl’s characterization of dream as reproduction is justified when this contrast appears when one is awake. The accounts complement rather than contradict each other.

Distinguishing between dream and wakeful imagination does not exclude intermediary cases, such as the lucid dream. A lucid dream is dreaming with the “secondary consciousness,” or awareness, of one’s state of dreaming (Voss & Hobson, 2015). Lucidity is on a continuum (La Berge & DeGracia, 2000; Voss et al., 2013), and one could take the ends of the continuum as absorbed dreaming and wakeful imagination. Because neither O’Shaughnessy nor Soteriou focus on lucid dreams, the discussion here concentrates on the absorbed dream and just calls it “dream.”

To summarize: if a synthesis is required for wakefulness, then it must be a process unifying all actual objects in one actual world, while contrasting the imagining self with the imagined one, according to Husserl’s distinction. The question then becomes whether the synthesis must occur as mental action. The following passages expand on this issue.

O’Shaughnessy (2000) has a similar view regarding unification. During wakefulness, one “sites its objects in one real world” while one “aim[s] at one real world” that is “consistent” (p. 124), where all actual objects are embedded (p. 155). The consciousness in a hypnotic trance, by contrast, is “not connective” (pp. 122–123). These ideas thus illustrate the unifying aspect of the synthesis. But does it require mental action?

O’Shaughnessy believes that it does, because “it is certain that intentional internal action is an experience which internally fuses the necessary three temporal axes,” which are past, now, and future (2000, p. 92). However, why must the temporal synthesis take the form of mental action? An action does have special temporal properties (Hua XLIII/III, 2020, pp. 27–28, p. 118, p. 353) compared with a mere occurrence, but even those occurrences have a temporal dimension “automatically.” The synthesis of retention, protention, and the original impression need not be contributed by mental action. O’Shaughnessy realizes this in explaining that retention of the past “is not merely a necessity for intentional action” (O’Shaughnessy, 2000, p. 204), but also applies to all experiences in general, including passive ones. But he offers no further clarification. He repeats that the “intentional project is a synthesizing force” (O’Shaughnessy, 2000, p. 204, emphasis added). So, is mental agency the synthesizing force? Synthesis does not entail action, after all. O’Shaughnessy (2000) himself notes this paradoxical fact later, in discussing synthesis in visual perception. He writes that “we make no appeal to a synthesizing agency” (p. 616).

Now, what about the other aspect of the synthesis in wakeful consciousness, or the contrast of the imagining self with the imagined self? Imagining during a waking state involves knowing that one is imagining. That knowledge is not possible during an absorbed dream. Soteriou (2013) argues that the knowledge, when available, comes from mental agency—one can access it because imagination is a mental action. A dream, while a form of imagination, lacks such knowledge. Dreaming is the mere occurrence of images, sensations, thoughts, and emotions, rather than being an action. In daydreaming, on the other hand, one may not always control every part of the experience, but one still remains aware of being disconnected from reality, and so retains the capacity to intervene (p. 322, pp. 327–329). This capacity explains the contrast between the imagining and the imagined selves, because the contrast enables that specific knowledge. But the contrast may not necessarily result from, or even be possible because of, mental agency. During a dream, one may not be aware that one is dreaming, for example, but if one suddenly wakes, then one realizes that it was a dream. In this case, one has not exerted agentive power to enable the variation from one state to the other, and the contrast therefore does not result from the fact that one is capable of actively bringing about that variation either. It merely occurs. Any argument that mental agency justifies this process fails.

O’Shaughnessy (2000) tries to explain this knowledge from another perspective. He argues that one has no “insight” into the “principle of progression” in a dream (pp. 215–216). When awake, one understands why this particular experience follows another, and O’Shaughnessy (p. 222) calls that understanding “intelligibility” or “pellucidity.” In a dream, however, one may have no idea why a particular occurrence follows another—why, for example, an image of a monster may be followed by one of a soccer game.

O’Shaughnessy explains pellucidity by mental agency. “Only a mind steering its own cognitive path through a wider cognitive scene […] can introduce pellucidity into the flow of experience” (2000, p. 227). This steering process explains why reasons are antecedents of action (p. 228) and distinguishes the will from mere forces (p. 227), “through enabling reason to determine” how the stream of consciousness advances. It therefore makes possible that “the mental will imports internal intelligibility” (p. 226).

The trouble for O’Shaughnessy’s view is daydreaming. When daydreaming, one seems to just let the imagination run on its own, and yet daydreaming is still a wakeful state. O’Shaughnessy insists that in shifting from one daydream scene to another, “at each point behind the scene lies an intention,” and “it is these intentions which determine the immediately succeeding phases of the process” (p. 224). But why assume that there must be some intention behind it? O’Shaughnessy argues that we have a “foreknowledge” of what will be experienced next in a daydream (p. 218), but how can that be for every case of daydreaming? One can have the experience of imagining a fictional soccer game, the dinner to be attended tonight, and then a film star. In this sequence one’s imagination runs wild and one may have no idea what scenario will follow another. One retains the agentive capacity to become immersed in daydreaming or to come out of that state, which to a certain extent may explain “pellucidity” in daydreaming, but perhaps O’Shaughnessy exaggerates our agentive capacity in daydreaming.

Finally, one must not neglect the possibility of some mental agency being preserved in a dream. If one dreams of being shot by a mysterious man, might not one strive, even while dreaming, to uncover the shooter? Moreover, it is not uncommon for people to report solving mathematical problems or puzzles in dreams. In short, mental agency may not be exclusive to wakefulness.

4.2 Argument from self-consciousness

The second argument is the Argument from Self-Consciousness:

  1. (1)

    Wakeful consciousness is always self-consciousness.

  2. (2)

    The self referred to in self-consciousness is an agent.

  3. (3)

    Hence, mental agency is constitutive of wakeful consciousness if mental agency is considered an agent’s capacity to perform mental actions.

The above argument is reconstructed from Soteriou (2013). Soteriou uses as examples the suspension of judgment and inferences. As a phenomenologist, if one brackets the world’s existence, one cannot then use it when reasoning about something else. In other words, a constraint from the previous self, who set the world aside, must later be imposed on the current self. That constraint applies, though, only if the current self acknowledges the previous self’s authority. In this way, the current self is always conscious of the previous self during thought. Soteriou concludes that “agency is implicated in the bracketing of one’s belief,” because “when one brackets one’s belief that p, one imposes a constraint on one’s reasoning by reasoning in recognition of it” (p. 267). Soteriou continues: “When one acts in recognition of a self-imposed constraint,” one is also “the source of the constraints,” and Soteriou takes this source to be an agent imposing a constraint (p. 304). In other words, the process of imposing restrictions on one’s present self by recognizing the prior selves’ authority reveals the previous self to be an agent.

The target in this section is premise (2) in the argument, which confuses different notions of the self and what it means to be an agent. Section 5.1 tackles this confusion. One need not be an agent to suspend judgment or to think about Euclid’s axioms, because those thoughts might merely come to one’s mind. In this case, the self to whom the judgment-suspension or the axiom-thoughts merely occur functions as a receptive self. The current self’s recognition of this authority, however, does not mean that the previous self is agentively imposing a constraint on the current one. For a phenomenologist, to bracket the world is one’s own decision, which makes it a full-blooded action. The previous self is an agent, but that fact does not comprise a general condition; it does not apply, for example, when judgment-suspension merely comes to one’s mind. Soteriou could reply to this objection by restricting mental agency to full-blooded actions, but he would then be making the same mistake as Maine de Biran (cf. Section 3).

To conclude this section—like Maine de Biran (1841) but unlike Husserl, O’Shaughnessy (2000) and Soteriou (2013) argue for the constitutive role of agency for wakeful consciousness rather than for phenomenal consciousness in general. They differ from Maine de Biran in attributing that role to mental agency, while Maine de Biran himself offers a model where the will interacts with bodily organs. In their present forms, the Arguments from Synthesis and Self-Consciousness both fail. One reason for their failure is the limitation to wakeful consciousness. Mental actions can take place in dreams, and we can also be aware of ourselves while dreaming—during a dream one can be aware of fighting a monster, even though the self that is aware in that moment is not the self of wakeful life.

Although Husserl did not explain how agency is constitutive of phenomenal consciousness, his intuition on the matter does not seem misguided. While all of the proposals from Maine de Biran, O’Shaughnessy, and Soteriou face problems, perhaps they could inspire more concrete ways to understand this constitutive role. The next section attempts to save the Arguments from Synthesis and Self-Consciousness by replacing wakeful consciousness with phenomenal consciousness. The key is determining the relevant sort of synthesis and self that agency is responsible for. Maine de Biran’s insight—that agency raises lower-level consciousness to a higher level—may shed light on the hard problem of consciousness.

5 Rescuing the arguments from synthesis and self-consciousness

Revised versions of the Arguments from Synthesis and Self-Consciousness might go as follows:

The original arguments

The revised arguments

Synthesis

(1) Wakeful consciousness requires a kind of synthesis

(1) The overall state of phenomenal consciousness requires a kind of synthesis

(2) Mental action is responsible for that kind of synthesis

(2) Agency is essential for that kind of synthesis

(3) Hence, mental agency is constitutive of wakeful consciousness if mental agency is considered the capacity to perform mental actions

(3) Hence, agency is constitutive of the overall state of phenomenal consciousness

Self-Consciousness

(1) Wakeful consciousness is always self-consciousness

(1) The overall state of phenomenal consciousness essentially involves self-consciousness

(2) The self referred to in self-consciousness is an agent

(2) Agency is essential for that self-consciousness

(3) Hence, mental agency is constitutive of wakeful consciousness if mental agency is considered an agent’s capacity to perform mental actions

(3) Hence, agency is constitutive of the overall state of phenomenal consciousness

The original and revised versions have four differences. First, I side with Husserl in endorsing the presence of agency in dreams. It is an agency that diminishes in dreamless sleep. I thus replace wakeful consciousness in the original arguments with one’s overall state of phenomenal consciousness. Second, the original Argument from Self-Consciousness identifies agency’s constitutive role with the thesis that the self (in the relevant self-consciousness) must be an agent. By contrast, the revised version can uphold agency’s constitutive role for phenomenal consciousness without that identification. The self in the self-consciousness may be an agent or non-agent, but even if it is the latter, that non-agent can still carry out its function essentially for the agent. In this way, agency would still be essential for self-consciousness. The last two differences concern the determinations of synthesis and self-consciousness. In Husserlian terms, O’Shaughnessy and Soteriou take the relevant synthesis for wakefulness as unifying all actual objects into one actual world, and as contrasting the imagining self with the imagined one. These authors also assume the relevant self-consciousness to be the consciousness of an agent. The previous section exposed some problems for their assumptions—the revised versions of the arguments replace wakeful consciousness with the overall state of phenomenal consciousness, such that the original characterizations of synthesis and self-consciousness no longer work. We therefore need new determinations for synthesis and self-consciousness, ones that can be essential for the overall state of phenomenal consciousness.

To specify the synthesis and self-concept involved in the arguments, the next part of the article distinguishes between different notions of the self to develop an account of action based on agent causation. This account uncovers how agency is constitutive of phenomenal consciousness, and how it might lead to implications for a panpsychist solution to the hard problem.

5.1 Different notions of the self

According to Soteriou’s (2013) reasoning (cf. Section 4.2.), the notion of the self is often taken to be a primitive, and it gets used interchangeably with the notion of “agent.” But this use is misleading and debatable at best. Notions such as “self”—and, correlatively, “mine”—are too thick to be primitive and seem to be grounded by more primary notions (Textor, 2018).

The first notion of the self to be distinguished here comes from the views of David Lewis (1986). He argues that actuality is indexical, in the sense that the real world is “actual” because it is our world, the world opposed to us. “Us” refers to the possible selves facing this actual world, and possible worlds are merely possible because they are not opposed to us. Similarly, an experience may be considered actual because one lives in it. The self at play here is an indexical self. The indexical self makes actuality distinguishable from possibilities, but does not intervene in actuality itself. It thus differs from an agent.

A second notion of the self would be the axiological self, or one capable of recognizing valuable properties. Examples come from medical science: depersonalized patients could feel their experience and body as “present,” but not as theirs (Billon, 2017), because the experience and the body do not “reach the soul” (Janet, 1928, p. 91). This phenomenon is usually explained as “de-affectualization” (Medford, 2012). In de-affectualization, the indexical self remains intact, because the experience and body are present for the patient, but the axiological self is impaired. The Greek word for “valuable” (ἄξιος) relates etymologically to “weight” and “volume,” and the German word “Anliegen” translates as “matter of concern” and literally means that what is valuable “lies close to the heart”; thus, the axiological self can be understood as the place where properties of value lie. The experiences of depersonalized patients cannot “lie on” that self in a normal way. The properties are “weightless.”

The third notion of the self is to be characterized as the teleological self. This self can focus on an end, and can desire it without necessarily acting to obtain it. The teleological self can also merely wish for something.

An agent is distinct from all three notions of the self. An agent is not just indexical, because it intervenes in actuality. An agent is not just axiological or teleological, because it exerts power to bring about a valuable goal.

There is one caveat to add: all the selves described are “immersed” and “occurrent” (Marcel, 2003). They represent what we experience pre-reflectively in a given moment and are not persistent substances, as most agent causalists tend to assume (Lowe, 2008; O’Conner, 2002).

5.2 A phenomenological agent causalist account of action

The goal in this section is to develop an account of action based on agent causation and first-personal experience. Although Husserl does not use the term “agent causation” in published manuscripts, he does define the action of the self as “an event it produces in the manner of doing” (Hua XLIII/III, 2020, p. 139). If the self—rather than an event in the self, like desire—causes the action, then that state of affairs implies agent causation. In an unpublished manuscript, he uses the term “self-causality” (Ich-Kausalität), which he takes to be a notion of causation distinct from that involved in event causation (Unpublished manuscript, 1916–1917, A VI 2/15b, 23b, 29a). This view echoes the notion of agent causation in current action theory.

Husserl argues that, when a person performs a bodily action, that person does not observe the experience of the action and the action as such to then connect both aspects, as might happen when one associates sunlight with a tree’s growth (Hua XLIII/III, 2020, p. 48). In event or substance causation, as in psychophysical causation, “each mind is separated from all physical things, and is related to its body and indirectly to other things only through causal relations” (Hua Mat IX, 2012, p. 230). How is this separateness to be understood? The argument here suggests the following interpretation: the cause and result in event causation can, in principle at least, take place without each other. According to David-Hillel Ruben (2018, pp. 118–120), if sunlight causes a tree to grow, then—as Hume claims—one can always conceive of the one event occurring without the other. For Husserl, by contrast, the agent is not something “separated from the environment, but in unity with it”: the agent “pours itself out” (sich ergießt) over the environment (Hua Mat IX, 2012, p. 229). Because one has direct control over the body in bodily action, as when pulling a trigger—an action through which a change occurs in the environment, like killing someone—this view must first be attributed to the relationship between the agent and the agent’s action. These two relate more closely than do two distinct events or substances. This very closeness means that an agent who disconnects from their actions, or actions where an agent is absent, are inconceivable by definition. This is in contrast to what occurs with cause and effect in event causation.

How close is the connection, then? Closer than that between two different events or substances—but not too close. Marcel (2003) argues that the agent is “immersed” during an action. This event occurs here and now, for example, when one plays the piano. “Playing the piano” is the action’s whatness, while “here and now” is the spatial–temporal position where the action takes place. One’s fingers are part of the action’s whatness, but the agent is neither identical to that whatness nor a part of it. At the same time, the agent is so distinct from the action that the connection between them can be taken as causation (Marcel, 2003). Although some scholars do claim that an agent’s “bringing about” the action does not imply causation (Lowe, 2008), they do not explain the phenomenological evidence that action is an intervention.

The way Husserl characterizes the will can explain this closeness. The will “is not a special manner of consciousness, but a special and further activity form, which can appear everywhere […]” (Hua XXXI, 2000, pp. 9–10). He also says that “tendency and self-striving are modes of consciousness, modes of everyone’s consciousness” (Hua XLIII/III, 2020, p. 118). Here, the will is not treated as a conscious event, parallel to emotion and cognition; it is instead a possible, ubiquitous mode of events that turns those events into actions. Ruben (2018, pp. 118–120) argues that the relationship between “trying” and action is closer than that in event causation, and that notion of “trying” could perhaps be identified with will’s Husserlian action-making mode.

Various languages use similar verbs to describe an event’s occurrence: “take place” in English, “avoir lieu” in French, “stattfinden” in German. An event’s occurrence would mean that its “whatness,” like moving one’s arms, finds a spatiotemporal position. But the will’s mode would not refer to the whatness as it finds that position, but to how the position is found: the agent is the one who brings about the event. If the relationship between the agent and the action is closer than that between two events, but not close enough to be part of the action’s whatness, then the agent’s place lies between the whatness and the spatiotemporal position. One might say it is immersed in this gap. The agent brings the whatness into position by overcoming resistance in favor of other possibilities, as when one’s arms are in pain and the pain prevents moving them; but the agent can also accomplish this by means of luck, as when a nurse helps to move the arms.

5.3 Phenomenal consciousness: the relevant kind of synthesis and self

The previous arguments have made it possible to explain how agency could be essentially responsible for both the relevant kind of synthesis and the self as indispensable for phenomenal consciousness.

The first thing to note is that agents do not exert their power arbitrarily, but do so only based on reasons. As Anscombe observed, agents have “a distinct kind of knowledge of their intentional actions,” which is a “practical knowledge: direct, non-observational knowledge of what they are doing and why” (Asma, 2021). At the same time, it is this practical knowledge that causes the action: it is “the cause of what it understands” (Anscombe, 1976, p. 87, as cited in Asma, 2021). This thought is helpful here. “The capacity to act for reasons,” as Anscombe understands it—if taken as “a capacity for non-perceptual, non-inferential knowledge of what one is doing and why” (Setiya, 2011, p. 192)—gives insight into how agency constitutes phenomenal consciousness. If one identifies that non-observational knowledge with phenomenal consciousness, then it is possible to assign it an essential role: everything one is conscious of becomes either a potential reason for a possible agentive power exertion (“why”), or a power exertion itself (“what one is doing”).Footnote 7 Hence, the relevant synthesis for the conscious state could be that between the reason for power exertion and the exertion itself. Before exploring that point, though, there are a few other concerns to address.

There has long been debate about the notion of a reason in action theory. Reasons tend to divide into “normative” or “justifying” ones and “explanatory” or “motivating” ones, but this distinction, according to Maria Alvarez (2010), is too crude—the reason one cites in explaining a girl’s crying, for example, which is that she is sentimental, need not be the reason that in fact motivates her to cry (p. 37). Instead, certain context must be distinguished when we speak of the reason for an action (p. 36, reconstructed by the author):

  1. (1)

    Justification beforehand. Before eating candy, I deliberate and convince myself that it would not harm my health.

  2. (2)

    Justification afterward. The judge asks me why I shot the victim, and my lawyer answers that the victim had a weapon in his hand and so I shot for self-defense, even if my actual motivation was revenge.

  3. (3)

    For both (1) and (2), it is common to say that there is a reason for someone to act, and that practice must not be confused with the reason for which one actually acts (Alvarez, 2010, p. 64).

  4. (4)

    Actual performance of an action. I hear a noise, and for that reason I close the window. This is the reason that in fact motivates one to act. This is the sense in which the current article uses “reason.”

  5. (5)

    Explanation. We could ask the question “why is he marrying the old lady,” and the answer may go in two directions. We could say that he wants to inherit her money when she dies, regardless of his actual motivation. In that case, the explanatory context overlaps with (2), the context of justification afterward. But we could also question his actual motivation, which would overlap with the reason in context (3). In this latter case, we use the locution “the reason for which one acts,” while the language “the reason why one acts” covers both cases.

Differences in context, however, do not exclude overlap in roles (Alvarez, 2010, p. 34, p. 37).Footnote 8 The noise for which one actually closes the window also justifies one’s effort in performing the action, although this is a weak sense of justification (Stout, 2005, p. 15, p. 69). The reason only makes the effort “pro tanto right,” but not absolutely so (Alvarez, 2010, pp. 12–15). For example, the noise may be a civil defense siren, and one’s best action would then be to run, rather than to close the window.

Justification illustrates one aspect of the connection between reason and power exertion. The other aspect is specification: the reason specifies how agentive power is exerted. According to Erasmus Mayr (2011), power exertion and reason “settle” different questions. If one asks why a certain action is taking place, then the answer might be that an agent is performing it. If one further questions why the agent is performing this action—like why he is closing the window, rather than running away—then the answer may be his ignorance of the noise’s meaning (p. 232). The agentive power exertion is not “automatically competing with other determining factors”; rather, it even requires “the presence of some further factors” (pp. 249–250), and one of those factors is the reason. Reason “structures” the power exertion by rendering the power exerted “in certain ways in specific situations” (p. 295). It specifies, for instance, that the power is exerted for running, rather than for closing the window.

Contrary to both Alvarez (2010, p. 38, p. 140) and Anscombe, who seem to take reasons as implying rationality and practical reasoning (Kalis & Ometto, 2021), reasons are here taken to be whatever intelligibly guides the agentive power exertion. This view rejects the Alvarez and Anscombe position because of the following considerations. If one requires reasons for action to be rational, then one must distinguish between the authentic reason and what merely appears to be a reason. The traffic noise that is in fact a civil siren would not qualify as a reason, because one’s belief about the traffic noise is wrong and thus irrational (Alvarez, 2010, p. 141). But this point would confuse the context of justification afterward with that of actual performance. Even if one’s belief about the traffic noise were irrational, the noise is in fact what initiates one’s actions intelligibly. It is thus the reason for which one actually does act. The perspective taken in this article refuses to restrict reason for action to practical reasoning, as otherwise it would be difficult to explain both habitual actions and animal actions.

The definition of reason given above, on the other hand, can explain why habitual actions do qualify as actions. When one brushes one’s teeth out of habit, one can also tacitly anticipate repeating the same action the following day. Repeating similar actions in similar contexts is one of the reasons guiding habitual actions. From that thinking, reasons also distinguish habitual actions from mere reflexes, as when one’s eyes close instantly before a dazzling light appears, which is something one does unconsciously. Practical reasoning is unnecessary in those cases, and Alvarez (2010) would be able to make only a limited response. She would say that those action-guiding factors are reasons, but only inasmuch as they could be justified by practical reasoning. But, then, she would also have to say that a dog acts for no reason when searching for a bone, so long as one does not ascribe to it the capacity for practical reasoning (p. 98). But that is counterintuitive.Footnote 9

Some could worry that this general definition of reason might conflate reason with cause, on the one hand, and on the other, motivation in Husserl’s thought with causation. To reply, it is necessary to deny the idea that motivation is not causal in any sense: “this subject exerts influence causally, but in the sense of motivation” (dieses Subjekt wirkt kausal, aber im Sinn der Motivation) (Unpublished manuscript, 1915–1921, K II 4/195b). The problem is not whether motivation is causal, but how it is causal.

Husserl’s second meaning for motivational causality, which I also endorse, is to take the notion as part of agent causation. What motivates us (das Motivierende) need not be an experience; rather, it is “the environmentally posited as such” (das umweltlich Gesetzte als solches), or the “noema,” such as a “favorite dish” (K II 44/44a–b). The motivation is neither event causation nor what Husserl calls “objective causality,” but is instead “the relation between the self-subject and ‘the’ object,” “as one is conscious of it in each intentional experience” (so wie er in dem jeweiligen intentionalen Erlebnis bewusst ist). Thus, “the object as understood, consciously characterized in this or that way” (der Gegenstand als so und so aufgefasster, bewusstseinsmäßig charakterisierter), “motivates, determines the subject” and “is the starting point of tendencies.” These “tendencies” are identical with motivations, because Husserl repeats that “motivations are […] what could be found at the object, what could come from (ausgehen von) the object” (A VI 2/15b).

There are two conclusions to draw from this citation. First, Husserl endorses, as I do, the view of so-called reason externalism, which takes reasons to be what one is conscious of, rather than as conscious mental states or events, like belief and desire (Lowe, 2008, p. 175, p. 181). Hence, the argument here takes reasons to include the aim, one’s abilities, the environmental affordances, and the previous phases of the action. Motivation cannot then be an instance of event causation but is instead a part of agent causation. Motivation is how the reason specifies and justifies the agentive power exertion, as discussed above. This point resolves any concern that the definition of reason given here may conflate motivation with event causation.

The second conclusion is Husserl’s emphasis on the role of consciousness in motivation. Not every bit of information guiding our actions is conscious. Visual processing during the famous Ebbinghaus illusion may suggest that unconscious processing, or vision-for-action, may sometimes guide action (Goodale, 2011). In addition, patients suffering from Alice in Wonderland syndrome can exhibit motor skills, such as walking normally, even though they have distorted proprioception and perceive their bodies as abnormally immense (Pitron & de Vignemont, 2017). An action-guiding factor qualifies as a reason only if it is “consciously characterized,” according to Husserl. Once again we see the difference between a reason and a mere cause.

To summarize, agentive power exertion always connects to a reason, and this by virtue of phenomenal consciousness. The relation to phenomenal consciousness is thus an indispensable part of every instance of agency, which is the agent’s capacity to exert power for a reason. In this sense, it can be said that phenomenal consciousness is constitutive of agency.

Could it also be, however, that agency plays such an essential role for phenomenal consciousness that it is constitutive of that consciousness? The argument here is that it does—everything one is conscious of may become a reason for exerting possible agentive power. Becoming aware that a laptop is available gives a potential reason to use it. Anger can turn into a reason to control oneself and one’s emotions. The awareness that one has exerted agentive power can even be a reason for further power exertion.Footnote 10 For instance, one can accelerate upon realizing that one is not walking fast enough to arrive in time. In Stout’s terms, a conscious mental state is “bound up” (2005, pp. 48–49) with a reason for action. It is not possible to identify a conscious mental state without also identifying a potential reason for agentive power exertion. To identify the perception of a fierce dog is to, simultaneously, identify the dog as a potential reason to flee.

If everything one is conscious of is a potential reason for agentive power exertion, then the argument seems to have arrived at the indispensable synthesis for the overall state of phenomenal consciousness: the synthesis between reasons and agentive power exertion. Agency, taken as the capacity to exert power for a reason, is so essential for this synthesis that it ends up constitutive of phenomenal consciousness. Phenomenal consciousness is possible only if it plays a certain role for agency. Revising the Argument from Synthesis in this way recovers its explanatory value.

In contrast to the original Argument from Synthesis, the selves confronted with reasons are “non-agents,” because they do not intervene in the world by themselves (cf. Section 5.1.); however, those selves always relate to the agent. One knows that one can brush one’s teeth, and that ability is relative to the indexical self. One understands the value of daily teeth brushing, and one can feel a desire to repeat the action. In these other experiences, the corresponding selves are the axiological and teleological selves. These selves are not the agent as defined here, but they play a role in the agent’s action. Conversely, one brushes one’s teeth only in connection with the guidance of these “non-agents.” The synthesis between an agent’s reasons and their agentive power exertion also implies a synthesis between the agent and the corresponding “non-agents.” As a result, if some kind of self-consciousness is essential for phenomenal consciousness, then the relevant sort of self in this self-consciousness is to be determined as follows: the selves indispensable for the overall state of phenomenal consciousness are either an agent or “non-agents” that carry out their functions for the agent.Footnote 11 Agency—understood as an agent’s capacity to perform actions for reasons given by “non-agents”—is then constitutive of phenomenal consciousness. This reasoning justifies a revised Argument from Self-Consciousness.

Agency’s constitutive role for the overall state of phenomenal consciousnessFootnote 12 explains why Husserl’s views on sleep as will-relaxation and as powerless affection are not independent of each other. Husserl claims that affection “concerns (betreffen) the self” during sleep, and that it is powerless (kraftlos) in the sense that the self is not “attracted” and “awakened” to the extent that it must reply (antworten) (Hua XLII, 2013, pp. 34–35, cf. Section 2, where Husserl says that an entity’s attractiveness diminishes in correlation with the relaxation of the will). A “powerless affection” for a delicious dish does not qualify as a reason for possible agentive power exertion, which would be the agent’s “reply.” If there cannot be agentive power exertion, here owing to the will’s relaxation during sleep, there cannot be consciousness of a potential reason for power exertion either.

5.4 Agency raises consciousness to a higher level

Although current phenomenologists and neuroscientists still study Husserl’s and Maine de Biran’s general theories of the will (Spano, 2021; Lafarge & Franck, 2009; Demanet et al., 2013), one may wonder whether their views on sleep in terms of agency have only historical interest. This section argues for their contemporary value in solving the hard problem of consciousness. According to David Chalmers (1996), the hard problem of consciousness is that, although it is clear that phenomenal consciousness is correlated with certain physiological processes, it is not clear why the correlation exists in the first place. Panpsychism has become influential against this background, because it is difficult to understand how purely physiological processes can give rise to conscious experience if one does not also assume that the processes themselves are somehow already experiential. But, by itself, panpsychism does not solve the hard problem. Rather, it merely gives the problem a new form, the so-called Combination Problem (Chalmers, 2016): how can micro-experiences give rise to macro-experiences?

A popular account along these lines is the Integrated Information Theory (IIT) (Oizumi et al., 2014; Tononi, 2017; Tononi et al., 2016). The theory has many mathematical details, but this section treats only the theory’s core ideas and challenges. Two of the theory’s axioms are “Composition” and “Integration,” which say that every experience is structured (composition) and irreducible to its independent parts (integration). That is, a macro-experience is composed of many micro-experiences, and the macro-experience is irreducible to its components because the components embed in a structure that both differentiates and integrates them simultaneously. Thus, IIT stresses differentiation and integration for phenomenal consciousness. If the micro-experiences are more differentiated while also being more integrated, the result would be a higher degree of macro-consciousness. Micro-experiences that are too homogeneous or too disconnected reduce the degree of consciousness. Many transcranial magnetic stimulation and electroencephalography findings support this (Casarotto et al., 2016; Koch et al., 2016).

The argument here leaves open the question of whether to endorse panpsychism; either way, the role of differentiation and integration for phenomenal consciousness is worth considering. Yet, one may still ask proponents of IIT how exactly the passage from sub-personal micro-consciousness to personal macro-consciousness is possible. IIT says that a single neuron has consciousness to the degree that it has a structure irreducible to its components. Call this consciousness “level 1” consciousness. Multiple neurons may then form a group having its own irreducible structure, and that would be level 2 consciousness. At some point the scaling reaches the final sub-personal level, p – 1, just before the personal-level p. That transition, rather than the one from level 1 to level 2, is what demands our attention in Chalmer’s formulation of the hard problem. Humans, as conscious individuals, can have experience only at the personal level. The p – 1 level sub-personal experiences are still non-experiential for conscious humans, and it is not clear what could bridge the gap between them. If IIT intends to address this problem, it must determine what kind of differentiation and integration enable the p – 1 to p transition.

The value of the proposal in this paper here becomes evident. According to Maine de Biran (cf. Section 3), agency raises the sub-personal to the personal level of consciousness, and the IIT framework explains how this is possible. Section 5.3 argued that the synthesis between reason and agentive power exertion is essential for the overall state of phenomenal consciousness. This synthesis can also be reformulated as that between the agent and the non-agents carrying out their functions for the agent. This very synthesis is the integration and differentiation that raise sub-personal-level consciousness to the personal level. Reasons are differentiated from agentive power exertion, because one may have a reason to reduce one’s sugar consumption and yet make no effort to do so. But reasons and the exertion of agentive power are also integrated—at least so long as a reason is the reason for some agentive power exertion, and at least so long as power exertion, when agentive and not reflexive, occurs for some reason.

With IIT’s terminology, one may say that at the p – 1 level, one sub-personal consciousness is responsible for factors guiding subjective behavior, while another is responsible for the power exertion bringing about that behavior. If the two remain isolated, then the action-guiding factors do not qualify as reasons, nor does the power exertion qualify as agentive. On the other hand, if the two sub-personal experiences integrate, then the result would be personal-level consciousness—which would be either the consciousness of the reason for power exertion, or the consciousness of the agentive power exertion itself. Still, certain factors guiding the power exertion may remain isolated from the integration, and that isolation would explain why some visual and bodily information that plays a guiding role for action is unconscious, at least sometimes (cf. Section 5.3). The power exertion isolated from this integration is non-agentive and so merely reflexive, as in the power exerted to withdraw one’s hand from a fire.

To conclude, the idea that agency is constitutive of phenomenal consciousness advances discussion on the hard problem of consciousness, even if it cannot solve each difficulty.

6 Minimally conscious states

Agency’s constitutive role in phenomenally conscious states can also justify certain clinical practices. In patients suffering from a disorder of consciousness or an overall diminished conscious state, multiple conditions can be distinguished: brain death, coma, UWS/Vegetative State, and MCS. Brain death patients cannot breathe on their own and need mechanical ventilation (Jellinger, 2010). Coma patients may breathe unsupported but cannot open their eyes and so have no sleep/wake cycle, nor do they show any voluntary behavior. The term “Unresponsive Wakeful State” refers to patients who show spontaneous eye-opening but otherwise behave only reflexively—hence the name “wakeful.” In contrast, those in a MCS can follow commands, though inconsistently, and may react somewhat appropriately to certain stimuli, like crying in response to something bad but not to something neutral (Giacino, 2005; Giacino et al., 2002). MCSs can be further subdivided into MCS + and MCS–. Patients in MCS + can follow simple commands and intelligible verbalization, whereas those in MCS– can only visually pursue objects, point to noxious stimuli, and hold an object still (Bruno et al., 2011; Hirschberg & Giacino, 2011; Thibaut et al., 2020). All these conditions must be distinguished from locked-in syndrome. Locked-in patients can perform few if any bodily actions, but their brain metabolic connectivity is at normal levels (Laureys, 2016; Vanhaudenhuyse et al., 2010). This normal connectivity suggests that locked-in patients do not suffer from a diminished degree of consciousness; their deficiency appears to lie in motoric capacity.

The approach developed in this article explains why it is reasonable to attribute phenomenal consciousness to MCS patients but not to UWS patients. When an experimenter requests MCS patients to open their eyes, it is reasonable to expect that, if the patients are phenomenally conscious, the request would give them a potential reason to exert agentive power. A visual object shown to these patients, or an emotional stimulus, could also give a potential reason. Upon observing the corresponding behavior, it would be interpreted as “purposeful” (Leon-Carrion et al., 2008), because the patients appear to perform it for a reason. If the thesis developed here is correct—that everything one is conscious of either is a potential reason for possible agentive power exertion, or the power exertion itself—then purposeful behavior would be evidence for phenomenally conscious states. In contrast, the fact that a UWS patient cannot follow a command or pursue a visual stimulus suggests that verbal commands or visual stimuli cannot provide a reason to that patient for an agentive power exertion. Even if a UWS patient could exert power by closing her eyes in bright light, the power exerted in that case would be reflexive, rather than agentive.

Purposeful behavior in MCS also occurs outside bodily action, in the form of mental action. In Owen et al. (2006), a patient’s functional magnetic resonance imaging scan showed reduced brain activity in a quiet state compared with normal individuals. That result distinguishes the condition from locked-in syndrome. However, when researchers instructed the patient to imagine playing tennis and then to mentally visit all the rooms in her house, the brain activity showed patterns indistinguishable from those gathered from healthy controls. This experiment suggests that the patient could follow a command and perform the imagination, which would be a mental action. That action would indicate phenomenal consciousness and corroborate the thesis of this article. Owen et al., however, classify the patient as vegetative owing to a lack of purposeful bodily behavior. The arguments given here suggest that this classification is incorrect, because bodily actions are not the only kind of action human agents are able to perform.

Interpreting MCS this way is also compatible with the attempt to help solve the hard problem of consciousness, as discussed in Sect. 5.4. Potential agentive power exertion, such as that derived from command following, requires connecting different neural networks for language comprehension, memory, and motor execution (Thibaut et al., 2020). Fingelkurts et al. (2012) argue that the connectivity among these neural assemblies was smallest in UWS, intermediate in MCS, and greatest in normal subjects. The intermediate degree of MCS may result from a “fluctuating” (Giacino, 2005) agency. Articulated connections between reasons and agentive power exertions may not always be possible in MCS, so that a patient may be conscious only occasionally.

Skepticism toward the notion of a MCS may take two forms. Naccache (2018) argues that patients in MCS are not really conscious. They are in cortically mediated states, but cortically generated behavior may be unconscious. In other words, patients in MCS may be “zombies.” Overgaard (2009), in contrast, questioned whether MCS is minimal. How is it possible to ensure that patients in UWS are phenomenally unconscious? UWS subjects, such as those under general anesthesia or in a deep sleep, may well be conscious but later simply fail to remember the state they were in. Mashour and LaRock (2008) call this the “inverse zombie problem.”

This multi-faceted skepticism poses a challenge to the second-person understanding. From that perspective, patients are treated as subjects rather than as mere physical systems. Researchers examining them try to “read their minds.” In other situations, though, researchers may evaluate other subjects from the third-person point of view, treating them as “dead matter” rather than as living subjects. Naccache and Overgaard doubted whether the presence or absence of phenomenal consciousness could be inferred from observing patients’ behavior. Naccache suggests instead that any inference should come from brain-imaging results—but those results can be associated with phenomenal consciousness only if one already knows whether the subject is phenomenally conscious. Physiological data give information about phenomenal consciousness, but do so only with first-personal experiences or some sort of second-personal understanding. One theory may consider the cortex responsible for phenomenal consciousness, for example, while another attributes it to the brain stem. These associations are possible only in limited circumstances—as when a patient with a lesion or otherwise diminished brain activity reports being unconscious of certain content (e.g., in the experimental framework of binocular rivalry), or when an experimenter judges on the basis of behavior that a subject is in a comatic state. One could infer whether a subject is conscious from brain-imaging results, but it is the first-personal reflection and second-personal understanding that lay the foundation for this inference, and not the other way around. When first-personal reflection is impossible, one must in the end rely on the experimenter’s empathy. If states such as MCS and deep sleep challenge the first-personal approach, the same challenge applies to purely physiological approaches. The second-person understanding may not always be reliable, but there is no hope of advancing the understanding of these phenomena without it.

7 Conclusion

To what extent and how does agency play a constitutive role in phenomenal consciousness? This paper answered these questions by offering a critical review of different proposals while retaining their important insights. Husserl correctly notes the essential role of agency for the overall state of consciousness, but his view does not specify the relevant notion of agency and so remains an empty framework. Maine de Biran offers a more informative model, suggesting that agency raises lower-level consciousness to a higher level. But Maine de Biran’s proposal is speculative and does not apply outside wakeful consciousness. To treat the constitutive role of agency, O’Shaughnessy and Soteriou propose the Arguments from Synthesis and Self-Consciousness. But those arguments misidentify the relevant kind of synthesis and self-consciousness.

To recover insights from these proposals, this article revised the two arguments by distinguishing various notions of the self and by developing an agent causalist account of action. The summary of this work is that agency is constitutive of the overall state of phenomenal consciousness, because everything of which one is conscious is either a potential reason for a possible agentive power exertion, or is the power exertion itself. The self referred to in the self-consciousness essential for phenomenal consciousness is either an agent, or a non-agent confronted with reasons for action that thus carries out its function for the agent.

These revised arguments rescued the insights of O’Shaughnessy and Soteriou. They also explained why Husserl’s two accounts of sleep—relaxation of the will, and powerless affection—are in fact one. More importantly, and within the context of panpsychist solutions to the hard problem of consciousness, this reasoning developed Maine de Biran’s suggestion that agency raises the lower level of consciousness to a higher level. In views on consciousness like the IIT, a crucial open question is how personal-level consciousness arises from sub-personal consciousness. The answer is that the micro-consciousness responsible for guiding subjective behavior, and the micro-consciousness responsible for the power exertion bringing about the behavior, remain isolated at the sub-personal level. Integrating these two micro-consciousnesses—whereby the guiding factors become reasons for agentive power exertion—gives rise to personal-level consciousness. Finally, the account given here also explained why patients exhibiting purposeful behavior can be justifiably characterized as minimally conscious, so long as the experimental stimuli serve as reasons for the patients’ mental or bodily actions.