1 Introduction

Before Benjamin Libet conducted his groundbreaking experiments, which show that conscious intentions and voluntary actions are preceded by a build-up of unconscious activity in the brain (Libet, 1985, see also, e.g., Filevich et al., 2013; Haggard & Eimer, 1999; Soon et al., 2008), the relationship between free will and consciousness has not been a central topic of discussion in the philosophical literature on free will. And even though few scholars believe that Libet-style experiments have shown that free will is an illusion, his work has challenged them to reflect on whether and to what extent conscious states and processes play a role in what we do and decide. Many philosophers and psychologists have argued that even if the conscious intentions studied in Libet-style experiments are not the causes of what agents do, that does not imply that consciousness is a mere epiphenomenon (see, e.g., Baumeister & Masicampo 2010; Baumeister et al., 2011; Baumeister et al., 2018; Mele, 2009; Pacherie, 2014; Sinnott-Armstrong, 2011; Slors, 2015, 2019; Walter, 2014; Zhu, 2003).Footnote 1 Even if consciousness would be redundant when it comes to button presses or finger movements, it is crucial for decision-making and planning for the future, for example.

Less attention has been paid, however, to whether this response captures the sense in which consciousness is relevant for free will. My main aim in this paper is to investigate the relationship between the two. I start by critically reflecting on the current discussion, and argue that by merely arguing that conscious states and processes sometimes do play a causal role, scholars seem to accept two assumptions that they should reject: (1) that the relationship between free will and consciousness is best characterized in terms of conscious states and/or processes being part of the causal chain leading up to the action, and (2) that the third-person perspective is a suitable means to capturing the relationship between free will and consciousness (Sect. 2). After that, I provide an alternative proposal of how free will and consciousness may be related, in which an agent’s self-understanding of what she is doing and why, while acting, takes center stage. In order to capture the relationship between the two, the first-person perspective should be investigated instead of explained away (Sect. 3).Footnote 2

2 Free will, causal chains, and consciously formed intentions

In response to the epiphenomenalist threat Libet-style experiments pose, a distinction is often made between proximal intentions, intentions about something to do straightaway (Mele, 1992, p. 72), and distal intentions, intentions for the non-immediate future (Mele, 1992, p. 137). Even if it would turn out that proximal intentions, the objects of study in Libet-style experiments, do not cause what we do, we have no good reason to think that conscious distal intentions are causally inefficacious as well. Consciousness “may be ill suited for direct control of physical behavior” (Baumeister et al., 2011, p. 352), but it is indispensable when we plan trips, decide on which house to buy, or write an academic paper. An example of conscious distal intentions that do seem to play a causal role would be Gollwitzer’s (1999) implementation intentions. Implementation intentions are not just plans to act in a certain way, but include the situation in which the action should be executed, i.e., “Whenever situation x arises, I will initiate the goal-directed response y!” (Gollwitzer, 1999, p. 493). As Mele (2009, pp. 138–140) argues, it is very unlikely that when the agent’s response in situation x is indeed y, consciousness did not play a causal role (see also Baumeister et al., 2011; Pacherie, 2014; Schlosser, 2014). Similarly, it is very unlikely that subjects in Libet-style experiments did not have a causally efficacious conscious distal intention to follow the instructions (e.g., Mele 2009, p. 46; Sinnott-Armstrong 2011, p. 239).

But does this response to Libet-style experiments capture the relationship between free will and consciousness? Even though philosophers disagree substantially on the nature of free will and under which conditions agents act freely, philosophers do tend to maintain that free will has to do with whether the agent is in control over her actions (see, e.g., O’Connor & Franklin, 2018). For the agent to be in control, in turn, two conditions are generally taken to be important (even though not all philosophers think they are both necessary): (1) the ability to do otherwise and (2) self-determination (ibid.). The reason why Libet-style experiments have attracted so much attention is that they suggest that instead of your conscious intentions and, by extension, you, your brain decides what your next move is going to be. If it would turn out that conscious intentions do not cause what we do, it would suggest that we do not determine for ourselves what we are going to do.Footnote 3 This relationship between conscious intentions and self-determination is reflected in the causal theory of action, according to which bodily movements are controlled by the agent if and only if they are caused, in the right way, by (conscious) intentions (e.g., Bratman 1987; Mele, 1992).Footnote 4 For an agent to determine for herself what she does she should act and, with that, her intentions should, at a minimum, play a causal role. Libet-style experiments suggest that our brain instead of our conscious intentions decide what we do, because of which we would not be acting and would not determine for ourselves what to do.

Even though this theory of action is explicitly defended by many philosophers and seems to be implicitly adopted by many psychologists and neuroscientists, for example by Libet, it is not without its problems. One major problem is the possibility of deviant causal chains. Davidson (1973, p. 79) gives the following example: a climber may form the intention to loosen his hold of the rope in order to rid himself of the weight of another climber, but the intention may unnerve him so that he loosens the hold accidentally.Footnote 5 His intention caused him to loosen the hold, but yet it what he did was an accident, not an action. This shows that, at least, the causal theory of action is incomplete: apparently not all bodily movements that are caused by intentions are actions and, with that, determined by the agent herself. Most, if not all, defenders of the causal theory of action acknowledge that causal deviance is a problem that needs a solution. Several solutions have been offered (e.g., Aguilar 2012; Mele, 1992; Peacocke, 1979; Schlosser, 2007; Shepherd, 2021), but none of these solutions are taken to be successful (e.g., Asma, 2022; Mayr 2011, p. 128; Schlosser 2007, p. 191; Steward 2012, pp. 57–58). That means that causally efficacious intentions, whether they are conscious or not, are not sufficient for control over action and, with that, self-determination, and it is not clear whether within this metaphysical framework a solution can be found.

What is more, Schlosser (2012) is not convinced by this response to Libet-style experiments. He argues that even though it may seem likely that conscious distal intentions, for example implementation intentions, are causally efficacious, it still may be the case that the conscious agent is a mere spectator of a process that is already underway and sufficient for the behavior to take place. Unconscious precursors of conscious distal intentions may not take the form of readiness potentials like in Libet’s experiments, but if they are related and specific to the ensuing movements and have all the ingredients to produce the conscious distal intention, the challenge to free will (and self-determination) remains the same: consciousness may still be passive and only accompany a process that is already underway (p. 370). Indeed, research suggests that distal intentions have neural precursors as well, for example in the case of consumer choices (e.g., Knutson et al., 2007; Tusche et al., 2010). The claim that it is ‘very likely’ that conscious intentions do play a causal role is far from conclusive; other interpretations in which the conscious intention is not causally efficacious are also possible (Schlosser, 2012, p. 371). We cannot be certain that consciousness causally contributed.Footnote 6

How can these two challenges be addressed? In which cases can we be sure that conscious intentions are not merely accompanying a causal chain that is already underway, and how can the problem of deviant causal chains be solved? In which cases is the conscious agent necessarily active and in control over her actions? Perhaps we need to take the distinction between proximal and distal intentions one step further, by appealing to different ways in which agents can come to intend something: we can either passively acquire an intention or actively form one (Mele, 1992, p. 184). When I cycle to work, for example, I passively acquire the intention to stop when the traffic light is red. I already know that a red light means stop, and therefore I do not have to reflect or deliberate about what to do. In these cases, intentions are “acquired independently of an action (like deciding) of intention formation” (Mele, 1992, p. 231). In contrast, I actively form an intention if there is a need to make up my mind, if I need to resolve uncertainty about what to do (Mele, 2003, pp. 209–210; 2009, p. 8). When I notice I have a flat tire, for example, I need to deliberate about whether to repair it, take public transport, or work from home.

Slors (2015) points out, however, that this distinction would not solve Schlosser’s (2012) challenge. Even if the agent had to make up her mind, that does not imply that the conscious agent is necessarily active. Agents can resolve uncertainty about what to do quite passively. Slors (2015) gives the example of choosing between vanilla or strawberry ice cream. As he argues, there are no other reasons to choose one over the other besides “becoming aware of my preference,” and because of that “[i]t may well be the case, and very probably is the case, that some unconscious urge to opt for vanilla was already present before my conscious ‘deliberation’” (Slors, 2015, p. 109). Again, we may simply have an agent consciously registering a causal process that suffices for the movement to occur (p. 98). In order for consciousness to contribute, then, we have to look for conscious intentions that “do not exist prior to conscious reflection and would not have existed without conscious reflection” (Slors, 2015, p. 100). Here Slors draws on the commonly accepted claim that an important function of consciousness is the integration of information (e.g., Baumeister & Masicampo 2010; Morsella, 2005). As Mascolo & Kallio (2019, p. 450) argue, consciousness plays a role if demands cannot be resolved by existing endogenous sources, for example an existing preference for vanilla, but pose novel challenges. In those situations, conscious reflection and integration of information about the state of affairs, your preferences, and beliefs, for example, are necessary in order to decide (Slors, 2015, p. 102). Slors (pp. 100–101) uses the example of planning a flight: we need to integrate information about our preferences, the time we have to be at a certain location, the flight schedule, etc. to form an intention. Conscious integration of information is necessary for the intention to be formed, and the agent could not have merely registered an intention or inclination to act in a certain way that is sufficient for the action to occur.

Consciously formed intentions, then, are not targeted by Schlosser’s (2012) challenge: these intentions would not have been there if it wasn’t for consciousness, and because of that, when we act from them, consciousness necessarily contributed. Examples of consciously formed intentions show that the threat of experiments like Libet’s is limited. What is more, the proposal is in line with a prominent model of intentional action, complementary to the causal theory of action, which is reflected in Bratman’s (1987) belief-desire-intention model: typically, agents weigh reasons for and against a certain action, form an intention on the basis of their deliberation, and subsequently act accordingly (see Mele 1992 for a similar proposal). An agent who acts for reasons does not passively acquire an intention, but consciously forms it by weighing reasons for and against acting as such. As a result, this agent determines for herself what to do. Indeed, Baumeister and colleagues (2011; 2018) have argued that consciousness is crucial for deciding on and planning sophisticated actions, and take the ability to act on the basis of conscious reflection as central to free will (Baumeister et al., 2011, p. 4). Relatedly, many philosophers maintain that free will has to do with our ability to act for reasons (see, e.g., O’Connor & Franklin, 2018; Schlosser 2014, p. 251; Slors 2019, p. 3; van Miltenburg & Ometto 2020; Walter, 2014; Wolf, 1993).

But does conscious formation of intentions in fact capture the sense in which consciousness is important for self-determination and, with that, for free will?Footnote 7 In the remainder of this section, I will argue that it does not. First of all, conscious formation of the intention does not solve the problem of deviant causal chains. The problem of deviant causal chains I addressed occurs between the intention and action: the intention of the climber caused his bodily movement and yet the bodily movement was not an action. Because of that, the problem of deviant causal chains will not be solved by distinguishing between different ways in which the intention has been formed; the problem does not depend on whether the climber passively acquired or consciously formed the intention to let go of the rope. In both scenarios, the agent may have lacked control over what she did. Conscious formation of intentions, then, is not sufficient for self-determination and free will.

What is more, there is no good reason to think that conscious formation is necessary for self-determination either. As we have seen, agents consciously form an intention when they are uncertain about what to do and have to integrate information to resolve the uncertainty.Footnote 8 But why would my action be determined by myself if I was uncertain about what to do, and not if I wasn’t? We have no good reason to think that my bike ride to work is only self-determined the first time, after I consciously decided to take the bike instead of the metro, but not the fifth time, when I already made up my mind about how to travel to work. It would imply that we only determine for ourselves what to do if we are uncertain, which seems to be an unwelcome and counterintuitive result. Within the metaphysical framework of the causal theory of action, for example, there is no good reason to think that passively acquired intentions amount to anything less than full-blown intentional actions: I still intentionally and consciously take the bike, even if I did not have to make up my mind about what to do. Relatedly, many philosophers have argued that we can act for reasons in the absence of conscious deliberation (see, e.g., Arpaly & Schroeder 2012; Audi, 2017; Kalis & Ometto, 2019; Railton, 2009; Sauer, 2012; Schlosser, 2014). Slors (2019) acknowledges this as well when he says that “[o]n most views, this does not mean that the relevant choices and actions must be based on conscious deliberation about pros and cons” (p. 2).Footnote 9 When I stop for a red traffic light, I act for a reason, even though my action is not preceded by conscious deliberation.

Furthermore, even if we would think that acting intentionally and for reasons is insufficient for self-determination and free will, it is still not clear how conscious formation of intentions would make the relevant difference. Some scholars emphasize that for an action to be determined by the agent, it needs to have a certain quality (see the influential views of Frankfurt 1971 and Wolf 1993 for example): the reasons and actions have to be good, either from the perspective of the agent or in an ethical sense. That does not mean, however, that philosophers like Frankfurt (1971) and Wolf (1993) think that conscious formation of intentions is necessary for free will.Footnote 10 We should keep in mind that conscious formation of intentions fulfills a function. It is a capacity, it can help us to make good (long-term) decisions (think of Gollwitzer’s (1999) implementation intentions), when we do not yet know what to do and information needs to be integrated. In practice, we would not manage to make good decisions and act for good reasons if we would never consciously deliberate about what to do, but that does not mean that actions that are the result of consciously formed intentions by definition have a better quality. We could do a good thing for good reasons in the absence of conscious deliberation, and we could consciously deliberate and our decision could still be bad..

Moreover, we could argue that conscious formation of intentions is necessary for the other central condition of free will: the ability to do otherwise. Schlosser’s (2012) challenge could be interpreted as such: if a conscious intention is merely part of a causal chain, the agent just happens to act that way and could not have done otherwise than she did. But the fact that intentions have causes, in the sense that we would not have them if these conditions were not in place, does not imply that the subjects could not do otherwise.Footnote 11 If I for example decide to bring my umbrella because of the rain – the rain caused me to passively acquire this intention – that does not mean that I could not have gone outside without an umbrella. In such a situation, I do not feel the need to deliberate or reflect on my initial tendency to take my umbrella, simply because it is the perfectly sane thing to do. From that it doesn’t follow that other actions weren’t open to me. I simply do not question the rationality of my initial tendency. What matters is whether I have the possibility to deliberate and reflect, not whether I in fact do. In fact, as I argued in a different paper (Asma, 2017), Libet-style experiments are set up in such a way that subjects have no reason to disrupt the causal chain that is already underway. Acting in line with the build-up of brain activity that is related and specific to the ensuing movements (Schlosser, 2012, p. 370) makes perfect sense for these subjects. Subjects in Libet-style experiments may still have the ability to do otherwise, they simply had no reason to do so.

Finally, Slors (2015) suggests that conscious processing is a means of “laying all the options on the table” (p. 102), which allows decision-making mechanisms to access all the relevant information, “so that we can compare, deliberate, and finally decide and form an intention” (Slors, 2019, p. 5). But even though it sometimes may be good to lay the options on the table, I do not see how this would make an action more or less self-initiated (or self-determined) either. Sometimes, we simply no longer have to lay all the options on the table; we already know what the best course of action is. Actions for which we haven’t laid all the options on the table can be done for reasons, for good reasons, and/or endorsed by the acting agent. In his (2019), Slors also states that if agents have to integrate information, their decision “is sufficiently biography- and character-dependent to yield different, idiosyncratic, personalized responses from different people to the same inputs to count as ‘self-initiated’” (p. 3, see also p. 8). But why would my passively acquired choice for chocolate mint ice cream not be personal or character-dependent (p. 4)? Why would I have to consciously deliberate and/or select a taste that is idiosyncratic for it to be my self-initiated choice?

To conclude, we have no good reason to think that conscious formation of intentions is necessary or sufficient for self-determination. Pointing out that some intentions are consciously formed may be a convincing way to reject epiphenomenalism, but it does not capture the relationship between self-determination and consciousness. The question I have been interested in in this paper, then, has not been answered by paying attention to consciously formed intentions.

How should we continue from here? Within the metaphysical framework of the causal theory of action, which is adopted by most scholars in the field, self-determination depends on the causal chain leading up to the action. The crucial question is whether a conscious intention is part of this chain, or whether it is merely accompanying a causal chain that is already underway and sufficient for the action to occur. Even though we may have the impression that our conscious intention is part of the causal chain leading up to the action, Libet-style experiments suggest that in reality all our actions have unconscious causes. That is what the scholars I have discussed been concerned with: they aim to argue that conscious intentions are part of the causal chain and, preferably, that this not only likely but certain (see Schlosser 2012; Slors, 2015; 2019). That means that the framework of the causal theory of action, according towhich whether the agent determined for herself what to do depends on the causal chain preceding the action, plays a decisive role in whether Libet-style experiments are interpreted as undermining self-determination, but also whether studies like these suggest that the first-person perspective is unreliable. After all, the perspective of the acting agent is taken to be mistaken, because she experiences her conscious intention to determine what she does, while in reality an unconscious causal chain explains it. If, in contrast, in turns out that action and self-determination cannot be captured in terms of the causal chain leading up to the action, preceding unconscious causes would not undermine self-determination, and neither would they show that the first-person perspective is illusory when it comes to action, self-determination, and free will.

The claim that the causal role of conscious intentions should be certain instead of merely likely is also relevant here. The issue Schlosser (2012) and Slors (2015; 2019) are concerned with is how we can be certain that the agent acted, and determined for herself what she did. But who should be certain that we have a case of self-determination? In line with Libet-style experiments, the idea seems to be that the observer, for example the neuroscientist, should be certain. And from such an external perspective, a causal chain in which intentions are passively acquired seems to have little to do with self-determination: the bodily movement appears to simply be the result of a causal process that started unconsciously or externally. In order to be certain that consciousness did make a difference, then, an observer needs to be able to establish, from her perspective, that something new was created by the conscious agent, something that wasn’t there before. But this line of reasoning, as we have seen, leads us to adopt a role of consciousness that may be certain, but doesn’t capture how consciousness and self-determination are in fact related. What is more, if the first-person perspective is taken to be unreliable because of the framework of the causal theory of action, and this framework is problematic, perhaps the first-person perspective is an important source of information for understanding the relationship between self-determination and consciousness?

3 An alternative perspective on self-determination and conscious intentions

On the basis of the previous section, we can gather three insights. Firstly, consciously formed intentions are not necessary for self-determination or free will. They are not required for acting intentionally, for (good) reasons, or having the ability to do otherwise. Secondly, for an agent to have control over what she does, a (conscious) intention being part of the causal chain leading up to the action is not sufficient. Even if an agent consciously formed the intention, what she did could still have merely happen to her, and we would not have a case of self-determination. Thirdly, the emphasis on consciously formed intentions – the conscious agent ‘disrupting’ the causal chain – seems to be the result of adopting a third-person perspective on self-determination; only if something new is created we can be certain consciousness played a causal role. But what if this external perspective is not the best way to capture the relationship between self-determination and consciousness?

Even though defenders of the causal theory of action take it that the problem of deviant causal chains can be solved, it should be clear that, as such, something is missing. Even if a conscious intention causes the action, the agent may still not determine for herself what she did – think of the climber accidentally letting go of the rope. One possibility would be to explore the proposals of how to extend the causal theory of action. I take it, however, that it would be worthwhile to explore a different perspective on the relationship between self-determination and consciousness. That is my goal in this final section of the paper. As became clear in the previous section, self-determination is tightly connected to action and, because of that, to intentions and reasons; accordingly, I will explore the relationship between acting for reasons and consciousness in more detail. I will argue that even though acting for reasons does not depend on conscious deliberation preceding the action, consciousness does play a crucial role while we are acting for reasons. Acting for reasons and self-determination, I propose, are constituted by understanding what you are doing and the reasons for which you are doing it.

I start by addressing another commonly made distinction in relation to Libet-style experiments, which is the difference between choosing and picking (see, e.g., Schlosser 2014). Schlosser (2014) argues that the subjects in Libet-style experiments can only pick. The subjects have no preference for moving their wrists later or sooner, move their wrists or refrain from doing so, or to press the left or right button; they have to choose among options that are indistinguishable. Because of that, Schlosser (2014) argues, Libet is not investigating free will proper, but the liberty of indifference, “an insignificant and uninteresting kind of freedom” (p. 251). But what exactly is the difference between the liberty of indifference or picking on the one hand, and free will proper or choosing on the other? Ullmann-Margalit & Morgenbesser (1977) analyzed this distinction in detail, and state:

we speak of choosing among alternatives when the act of taking (doing) one of them is determined by the differences in one’s preferences over them. When preferences are completely symmetrical, where one is strictly indifferent with regard to the alternatives, we shall refer to the act of taking (doing) one of them as an act of picking. (p. 757)

Schlosser’s (2014) distinction is in line with this proposal. When the options are completely indistinguishable for the chooser, she does not prefer one option over the other, and therefore sees no reason to choose one over the other (Schlosser, 2014, p. 251). In such a situation, for example in Libet-style experiments, the agent is merely picking and we have an example of the freedom of indifference.

I think, however, that it is important to further clarify what exactly the difference between free will proper and freedom of indifference is. Whether the options are indistinguishable, whether the agent prefers one option over the other, and whether she has reason to choose one over the other are different aspects of a choice situation, and they can come apart. Think of the following two examples: I can choose between two bags of candy with the same content, but have, for marketing purposes, different colors. The options are distinguishable and let’s assume that I prefer the purple one over the green one. Compare this to a case where I have the choice between two jobs, one is more ambitious and challenging, while the other is more familiar and less stressful. These options are also distinguishable, but let’s assume, for the sake of argument, that I do not have a clear preference for one or the other. I could see myself having a busy job in which I challenge myself, but I could also imagine that it would be nice if work took up less energy. Now, even though I have a preference for the purple bag of candy, buying that one still seems to me to be a case of liberty of indifference. In contrast, even though I do not have a preference for one job over the other, it seems to me that deciding between the two amounts to a proper choice.

In virtue of what then, is choosing the purple bag liberty of indifference while deciding in favor of one job over the other free will proper? What matters are reasons. We could of course say that the fact that I prefer the purple bag over the green one gives me reason to choose the purple one and, of course, in a minimal sense, that may be true. But this choice does not have any impact; both for me personally and normatively, it is irrelevant which one I choose (or pick). In the case of choosing a job, however, the options amount to different lives. The job I choose has implications for my mental and physical health, my relationships with family and friends, my self-image, etc. In making a certain decision, I at the same time decide what kind of life I want to live and, by extension, what matters to me. This, I maintain, is what makes choosing between jobs a genuine choice. My claim thus is that choosing is not merely about whether I prefer one option over the other as Ullmann-Margalit & Morgenbesser (1977, p. 758) and Schlosser (2014, p. 251) maintain, but whether the options are genuinely different options, from the perspective of the agent. What captures the distinction between picking and choosing is whether the options, to a certain extent, matter from the perspective of the agent.

By emphasizing this aspect of the difference between choosing and free will proper on the one hand and picking and liberty of indifference, I am not arguing against the views of Ullmann-Margalit & Morgenbesser (1977) and Schlosser (2014). I am merely highlighting that this aspect, whether the options make a difference for the agent, is more important than the others. In fact, Ullmann-Margalit & Morgenbesser (1977, p. 777) support my proposal when they say that “the importance, as well as a certain degree of irreversibility of the selection, and the individuality of the alternatives seem to suggest that these are as close as one can get to paradigm cases of choosing” (pp. 777–778, emphasis added). Similarly, Schlosser (2014) states that the choices in Libet-style experiments “are devoid of any normative significance and personal value” (p. 252).

In my view, the difference between their view and mine has to do with our respective perspectives on reasons for action. On the one hand, we can think of reasons as causes. When we act for reasons, our beliefs, desires, preferences, etc. cause our actions. On such a view, the question is whether we in fact have desires or preferences that speak in favor of one option over the other (think back of Bratman’s (1987) belief-desire-intention model). If we do not, we cannot act for reasons, as for example Schlosser (2014) maintains. Crucially, this perspective on acting for reasons also suffers from the problem of deviant causal chains; even if my action is caused by a reason I may not act for that reason (Asma, 2022). In recent years, following Anscombe (1963), a different perspective on acting for reasons has gained popularity (e.g., Singh 2020; Thompson, 2008; Wiland, 2012). These scholars maintain that the reason for which we act is the more general action of which the specific action is part. For example, the reason why I cut an onion is that I am making soup, and the reason for which I choose to buy a house in a particular city is that I aim to spend the rest, or at least a substantial proportion, of my life in that city.

This alternative way of looking at reasons captures the distinction between picking and choosing: in light of my larger action, it matters whether I am cutting an onion instead of an apple, or whether I buy a house in my favorite city or a village close by. What doesn’t matter is which onion I cut or whether I buy the fifth or the sixth house in the same street. That is, of course, to the extent that both onions and both houses fit my action at a higher level of description equally well; if I only need a small amount of onion, it makes sense to choose the smallest one, and if one house has a larger garden and I enjoy gardening I should choose that house. In such a view, the reasons for which we act make the specific action, in some sense and from the perspective of the agent, good. When I am making soup, it is a good idea to cut an onion, and if I want to spend the rest of my life in a certain city, it is a good decision to buy a house there (see, e.g., Frey 2013, p. 56). In line with that, I propose that we choose and have a case of free will proper when the options amount to different actions at a higher level of description, and we pick and have liberty of indifference when both options amount to the same action at a higher level of description, or there simply is no higher level of description. The question thus is whether the options pertain to different actions and, with that, whether they pertain to different reasons and intentions. From this perspective on intentional action and acting for reasons, further intentions are kinds of reasons (Anscombe, 1963, p. 21): I cut the onion with the further intention to make soup, and I buy the house with the further intention to spend my life in that city.

According to Anscombe (1963), this inherent means-end structure of actions is constituted by our practical knowledge of that means-end structure (see also, e.g., Singh 2020). When we act, we do not just know what we are doing – cutting an onion for example – we also know that we are doing one thing in order to do something else. And we do not just know this structure, we know it non-observationally. I do not have to perceive the knife and the onion and reflect on my bodily movements in order to know that I am cutting an onion in order to make soup.Footnote 12 When it comes to my actions and the reasons for which I act, I can directly answer the question ‘Why?’. This is so, because my practical knowledge of cutting an onion and cooking soup is the cause of what it understands (Anscombe, 1963, p. 87).

How are these reflections relevant for the relationship between free will and consciousness? First of all, one implication is that when we act, we do so consciously. After all, intentional action is characterized by the fact that we can answer the question ‘Why?’ directly, and being able to report is taken to be a defining feature of consciousness (e.g., Mele 2009, p. 21). One way to spell this out would be to point out, in line with the integration function of consciousness discussed previously, that in order to act we have to integrate information. In order to cut the onion successfully, I have to integrate information about the location and shape of the onion, the knife, the size I want the pieces of onion to be, etc. (see also Mascolo & Kallio 2019, p. 450).

I do not think this is the main sense in which consciousness is important for free will, however. Even though I may have to integrate information to cut the onion and consciousness may play a role, it does not seem to me that I am consciously deliberating my way through the action. Indeed, as Frey (2013, p. 56) points out, Anscombe (1963) is not committed to such a claim. In fact, Anscombe (1963) states it “has an absurd appearance when practical reasonings […] are set out in full” (p. 79). What is more, this sort of perceptual consciousness does not seem to capture the relationship between self-determination and consciousness. After all, we could describe all these processes in terms of third-person physical mechanisms (Gallagher, 2006, p. 113), and some or perhaps many of these processes may take place unconsciously (see, e.g., Shepherd 2015 for discussion). Rather, as Gallagher maintains, what is crucial is that intentional action at “the highest pragmatic level of description” (p. 119) involves not just consciousness of what our body is doing or the environment in which we act, but that “[i]n conscious deliberation of the sort found in situated reflection certain things in the environment begin to matter to the agent,” and “[m]eaning and interpretation come into the picture” (p. 121).

Gallagher (2006) does not spell out in much detail what this exactly amounts to, how meaning and interpretation are causally efficacious when it comes to intentional actions, and how conscious reflection makes it so that things begin to matter to the agent. In my view, on the basis of the view of acting for reasons that I set out, we can say more about this. At the highest level of description, most of the time, we do not just walk to the train station, open the front door, or type a certain word. Our intentional actions typically go beyond the physical movements of our body: in walking to the train station, I am at the same time going on holiday, by opening the front door of the house I am in the process of deciding to buy a certain house, and I type this word in order to make an argument. This, I maintain, captures an important sense in which self-determination and consciousness are related: if I know what my specific action means to me at a more general level of description and this consciousness – practical knowledge – is the cause of my specific intentional actions, I act for a reason and I determine for myself what to do. I do not need to consciously form an intention or consciously deliberate before or during the action; I need to know what I am doing and in light of which action at a higher description I am doing it (see also Asma, 2022).

This proposal does not suffer from deviant causal chains: the climber’s letting go of the rope is not caused by his practical knowledge of letting go of the rope, and because of that is not an action. What is more, this account of the relationship between self-determination and consciousness explains why, in contrast to the emphasis on consciously formed intentions, more consciousness implies more self-determination: the better I understand what I am doing on a higher level of description, the more control I can exercise not only over my actions, but over the direction my life is taking (see Walter 2014). The better I grasp what it means to lead a good life, which can be seen as the action at the highest level of description and the ultimate reason (see Frey 2019, p. 1144), the better I know which specific actions I need to perform in order to act for this ultimate reason. Distal intentions like buying a house or writing an academic paper, then, are relevant for free will neither because they are about the nonimmediate future, nor because they are generally consciously formed, but because they tend to be formulated at a higher level of description and make it possible for us to transcend the mere physical movements of our body and oversee what, given our conception of the good life, the best line of action is in a certain situation.Footnote 13 In virtue of having an intention at a higher level of description, we can organize our lives accordingly. An agent that can form the distal intention to be compassionate, to be a good partner, or to be healthy, and knows how to translate this general intention to concrete, specific actions, exercises more control over her life. This agent is acting for reasons properly. Of course, without doubt, a lot of preparatory work is done in terms of conscious deliberation and reflection, but when it comes to concrete actions, we can act for reasons and determine for ourselves what to do in their absence. What matters for self-determination is not primarily whether we have conscious awareness or exercise control over the processes preceding our actions, but whether we understand what our intentions and actions amount to, whether they are rational given our conception of the good life.

What, then, does this tell us about Libet-style experiments? In line with Schlosser (2014), I agree that the subjects can only pick one of the options, because on the higher level at which their action can be described, participating in the experiment, it does not matter what they do. Whether they move earlier or later, or push the left or right button, at a higher level it is irrelevant, because they still act for the same reason and realize the same action at the higher level of description. I also agree with Schlosser (2014) that this does not mean that these subjects do not act out of free will, but mainly that the kind of freedom that is tested in Libet’s study is “insignificant and uninteresting” (p. 251). The more the options amount to different actions at a higher level of description and actually make a difference to the life of the agent, the more we find ourselves in the domain of free will proper.

An important implication of my proposal is that in order to find out whether the agent acts for reasons and which reasons she acts for – and with that, to what extent she determines for herself what to do –, we need to take the perspective of the acting agent seriously. Libet observed that the subjects’ bodily movements were preceded by an unconscious build-up of brain activity, but most likely these subjects were acting for a different reason altogether: they moved their body in order to participate in the experiment correctly. In other words: if we only observe agents from the outside, it is much more difficult to distinguish between mere causal chains and actions that are done for reasons. As soon as we take the perspective of the acting agent seriously, we recognize that the causal chains Libet identified take place within a larger intentional action the agent is consciously performing.