Introduction
In this paper I will argue that belief in free will is epistemically innocent. Some authors have argued that even if belief in free will is false, it might be recommended on pragmatic grounds. I suggest another reason why belief in free will might be good. More specifically, I argue that our belief in free will has certain, otherwise unavailable, epistemic benefits. To do so I rely on work done on the epistemic status of beliefs, developed by Lisa Bortolotti and her research team (2020). Of course, if such a belief about free will is true, this obviously makes it a good belief to have. My argument is therefore that even if this belief turns out to be false, it is still a good belief to have, for hitherto unappreciated reasons.1 An important upshot of this account, therefore, is that it provides a novel mechanism for exonerating free will beliefs (if they turn out to be false).
Many contemporary accounts which deal with the “illusion” of freedom seek to describe the pragmatic benefits of belief in free will, such as how it facilitates or grounds our notions of moral responsibility or basic desert2 (Mele 2005; Wegner 2002; Smilansky 2000; Strawson 2010). While these proposals have their place (and use), I will not explicitly engage with them. I aim to establish that our (potentially false) belief in free will is an epistemically innocent one, and that this holds independently of whether we do in fact have free will or not.
Human beings have long been considered the prime example of rationality. However, the empirical literature suggests that we are not as rational as we were traditionally conceived to be (Bortolotti 2015a, 1). This is not to claim that we are at base irrational or insane, but rather to point out that our beliefs are not always guided by reason, and that biases, heuristics, and affect all come to play a role in how we reason. In this paper, therefore, I aim to bring two distinct research projects together for the first time: The literature on the epistemic innocence of beliefs, and the literature on free will.
One of the main meta-justifications for this epistemically-orientated approach is one that informs most of philosophy: A desire to get, at the very least, closer to the truth. While pragmatically belief in free will certainly provides benefits, the epistemic benefits I consider in this paper provide us with an additional, distinct set of reasons for evaluating our belief in free will. In order to show that this is the case regarding belief in free will, I will move beyond purely pragmatic concerns and show how this false belief can lead to epistemic gains for agents who adopt it.3
I will proceed as follows. First, I put forward that I assume some version of ‘illusionism’ about free will. Second, I outline what it means for a belief to be epistemically innocent. Third, I apply this to belief in free will, and show what unique, otherwise unobtainable, epistemic benefits it might accord.
1. The Way Forward
We might not have free will. If this is true, it would be false for us to believe that we have free will. We find support for what has been termed ‘illusionism’ in the philosophical literature on free will (Smilansky 2000; Strawson 2010). Smilansky, for example, explicitly endorses such a view of free will. He uses the term ‘illusion’ because it draws our attention to the “various ways in which false beliefs are held, without complete awareness of their falseness, in the face of stronger epistemic claims to the contrary” (Smilansky 2000, 148). For my purposes, and in what follows, I will bracket the question of what exactly free will is, and whether it really is an ‘illusion’ in the sense above. Instead, I will proceed as if our belief in free will is simply false, and then show how it might be exonerated on epistemic grounds.
There are many ways to understand what ‘free will’ might mean: Is it alternative possibilities for choice and action, freedom from causal determination, the ability to act or refrain from acting at a certain time, or the ability for an agent to act rationally? I cannot settle this debate here, but from these different senses of what free will means, we can get a rough sense of what a ‘belief’ in free will might entail. Such a belief, from the perspective of the agent who adopts it, would have something to do with control, knowledge, and action. In what follows, therefore, it is these general aspects of free will that I focus on in my articulation of the potential epistemic benefits.
So, assuming that our belief in free will is false, one way to exonerate such a faulty belief is by appealing to pragmatic upshots of the belief. The broadly pragmatic benefits of this belief are relatively clear, such as facilitating our ascriptions of moral responsibility, allowing us to view ourselves as ‘in control’ of our actions, etc. The question I am interested in, however, is whether there are also epistemic benefits to this type of belief. While in an ideal world our beliefs would all be supported by and responsive to the available evidence, the limited cognitive capacities we have as agents ultimately leads to us adhering to some poorly supported beliefs, which may be unresponsive to evidence. On the face of it, one might think it rational to dismiss these epistemically dubious claims altogether. Yet in some cases a prima facie epistemically costly cognition4 can in fact lead to positive epistemic outcomes (Bortolotti 2015b). I will argue that this is the case regarding free will. An important upshot of this account is that it provides further reasons for exonerating belief in free will.
2. Epistemic Innocence
In this section I will show that even if our belief in free will is false, it is nonetheless an epistemically innocent belief. This argument steers clear of the metaphysical problems introduced earlier and simultaneously moves beyond the traditional pragmatic focus of other theories in such debates (Dennett 1984). I will endeavour to show that if we carefully consider the circumstances in which particular beliefs (such as our belief in free will) are adopted, we can come to better appreciate not just their psychological but also their epistemic benefits. This strategic approach to our beliefs allows us to guard against the ‘trade-off’ view regarding certain types of false beliefs.
The trade-off view assumes that while certain faulty cognitions may offer psychological or pragmatic benefits, these benefits come with epistemic costs (Letheby 2016, 31). To put it differently, the trade-off view assumes that there are only epistemic costs associated with faulty beliefs, and no associated benefits. These costs are presumed to stem from the irrational nature of the belief, as it might be unresponsive to evidence, implausible and/or not an accurate representation of reality (Bortolotti 2015, 492). However, this trade-off view presents us with an overly simplistic representation of what is going on regarding both the formation and retention of our beliefs.
In many cases there is considerable overlap between both pragmatic and epistemic criteria, as in situations when psychological well-being is positively correlated with increased social engagement, leading to the formation of more true beliefs over time. However, there are also cases in which epistemic and pragmatic considerations can compete with one another, such as in cases where irrational beliefs can be pragmatically beneficial,5 or when rational beliefs are not useful.6 For a belief to be innocent is to suggest that even though it is epistemically irrational, it might nonetheless confer certain benefits which could act as an excuse for holding the belief. Thus, the notion of ‘innocence’ at work here is an application of the “sense of innocence as absence of guilt to the epistemic domain” (Bortolotti 2020, 9).
The type of agents we are necessarily implies that we are limited by certain physical constraints in our ability to coherently form and maintain our beliefs. Our limited cognitive capacity often leads us to adopt poorly supported beliefs, which often act as helpful heuristics as opposed to facilitating proper reasoning. Kahneman (2011) gives the example of what he calls the “affect” heuristic.7 The affect heuristic is a cognitive shortcut which allows agents to efficiently solve problems by relying on their current mood. It allows people to judge the risk or benefits of a specific action by relying on which feelings are associated with that outcome, as opposed to engaging in temporally expensive reasoning. There are cases where this can be useful (better avoid this spider) or misleading (climate change does not produce an affective response in many, and so is thought by some to not be a serious issue).
While it might be reasonable to dismiss these types of ‘epistemically costly’ cognitions8 altogether, there are times when an epistemically costly cognition can enhance our long-term epistemic functionality, such as our ability to form more true beliefs over time (Bortolotti 2016, 888). A classic example cited in the literature is that of BX, a former musician who, after a car accident, became a quadriplegic (Bortolotti 2015b, 492). Before the accident BX was in a healthy relationship, but soon after the incident his partner broke up with him. Following this BX developed the delusional belief that his partner was still with him (known as “reverse-Othello syndrome”). This false belief might have allowed BX to get through the trauma of the accident, and so there is a case to be made that it provided psychological benefits. But there are also epistemic benefits to this false belief, such as BX being more willing and able to engage with his doctors and therefore acquire knowledge about how he might best go about the world post-accident. Bortolotti (2015, 495) argues that these cognitions can be construed as being epistemically innocent.
2.1 Epistemic Status
There are two criteria which are necessary and jointly sufficient for a delusional belief to qualify as epistemically innocent:
(1) Epistemic Benefit: The delusional belief confers a significant epistemic benefit to an agent at the time of its adoption.
(2) No Alternatives: Other beliefs that would confer the same benefit are not available to that agent at that time. (Bortolotti 2015, 496).
For the purposes of this paper, however, I am not necessarily endorsing the view that belief in free will is delusional. Rather, I propose more modestly that it might be a false belief. Therefore, with respect to (1) above, we can replace ‘delusional’ with ‘false’.
But just what exactly counts as an epistemic benefit? There are two main lenses one could use when evaluating the epistemic status of a specific belief. Firstly, one could argue that a belief is epistemically advantageous if it allows for the retention or acquisition of true beliefs over time (a veritist). And secondly, one could argue that a belief is epistemically advantageous if it allows for the promotion of an agent’s intellectual virtues such as intellectual curiosity or honesty (a virtue epistemologist) (Bortolotti 2016, 889). Taken together these attributes constitute the epistemic functionality of an agent, i.e., the ability of the agent to function well epistemically.
In terms of the no alternatives condition, there are three ways in which a cognition may be construed as being unavailable: It may be strictly, motivationally, or explanatorily unavailable (Sullivan-Bissett 2015, 554). A cognition is strictly unavailable when it is based on information that is unavailable to an agent via introspection. An example of this would be an agent who suffers from dementia and as a result of which has severe memory impairment. Such an individual may claim to have been at a theme park in the morning, when the trip actually occurred when they were a teenager. They would be incapable of forming the correct sort of belief regarding the trip because of their memory impairment, and so such a cognition is strictly unavailable. A cognition is motivationally unavailable when it is unavailable due to motivational factors. A common example of this type of cognition involves cases of self-deception. Take the case of the cuckolded husband who falsely believes that his wife is faithful to him. He might have evidence that she is being unfaithful, but his desire to believe that she is not having an affair makes the belief that she is unfaithful motivationally unavailable to him. Lastly, a cognition is explanatorily unavailable when an agent dismisses it due to its perceived high improbability (Sullivan-Bissett 2015, 554). Consider the example of you finding porcupine quills in your garden. You also observe that there are pieces of your favourite tree missing. It is reasonable at this point to conclude that there is a porcupine chewing the bark off your tree. However, you could also believe that there is a magical fairy that drops porcupine quills and cuts bite-like marks out of trees with a hunting knife. This second type of explanation is dismissed due to its implausibility. It is dismissed because of how unreasonable it seems, and so is explanatorily unavailable when compared to other, more plausible, cognitions.
3. Belief in Free Will as Epistemically Innocent
3.1 Epistemic Benefit
In order for belief in free will to be considered epistemically innocent it must be shown that this belief does in fact provide an epistemic benefit to the agent who adopts it. This is not to say that the belief is epistemically good overall or free from epistemic faults. Rather, it is simply to modestly claim that such faulty cognitions can confer some epistemic benefits (such as BX being able to continue interacting with his doctor’s post-accident and therefore being open to the acquisition of more true beliefs over time) (Sullivan-Bissett 2015, 554). I will show how belief in free will helps us maintain a more coherent sense of self, and, secondly, how it can facilitate the process of reason-giving and taking, which could, for example, help us make various implicit biases explicit (with the hope of their eventual correction). Lastly, I will argue that it heightens our sense of ‘perceived control’, contributing further to our epistemic well-being.
The first point to consider is the way in which belief in free will facilitates a more coherent sense of self. The mechanism by which this is done turns on the essential causal opacity of certain folk-psychological concepts, such as our beliefs, desires, etc. Beliefs are molar-level phenomena which might have various correlates at different levels of abstraction (scientific psychology, neuroscience, etc.) (Bortolotti 2010, 2). The implications of this opacity mean that we might be unable to introspect the ‘real causes’ that lead to the retention or adoption of our beliefs.9
However, belief in free will could potentially mitigate this by giving us a plausible ‘just so story’ about what ‘caused’ us to act in this or that way. This illusion of competence adds an important sense of coherence to our sense of self, which may enhance our self-confidence and well-being (Sullivan-Bissett 2015, 548). With a coherent sense of self an individual is better able to engage with the external world, increasing the possible range of affordances available to them (Bortolotti and Miyazono 2015). It seems reasonable to suppose that individuals who are sure of themselves and feel secure in their beliefs will be more likely to engage with their surroundings.
Furthermore, this type of psychological security might allow for individuals to put themselves in new or perhaps uncomfortable situations, which would grant them new experiences and increase the probability that they acquire more knowledge about the world. An individual is far more likely to be willing to engage with their environment and increase the landscape of affordances available to them if they feel sure of their place in the world.10 These affordances provide a scheme by which they can expand their knowledge about the world through the sharing of expectations and conventions (Ramstead, Veissiere, and Kirmayer 2016, 4). Such affordances should not be seen as ‘things’, but rather as possibilities for action. These possibilities for action can be viewed as opportunities for an agent to acquire new beliefs about the world, which will also be subject to feedback. In this recursive way an agent may come to gather a significant amount of information about the world around them, leading to an increase in overall epistemic functionality. The agent both increases their epistemic virtues by engaging in reason giving and revising certain problematic beliefs and can acquire and retain more true beliefs about the world.
Relatedly, we do not always have direct access to the underlying causes of our actions, and so when we are questioned as to why we performed a certain act we often to respond with post-hoc rationalizations (Sapolsky 2018, 401). An easy example of this type of explanation (unfortunately) occurs in the self-reports of some explicitly egalitarian individuals. When assessing application documents of students applying for a lab position, faculty members consistently rated male applicants CVs as better suited to the job than their female counterparts (Sullivan-Bissett 2015). The confabulation comes out when the applicants’ supporting documents are controlled for: There is no other distinguishing factor between applicants besides their gender, and so it is clear that this gender-bias was causally efficacious in the faculty member’s decisions. However, the reasons given were that the male candidates were “more competent and hireable” (Sullivan-Bissett 2015). These reasons can be seen as confabulatory as they do not express the ‘true’ rationale for behaviour.
Therefore, when we engage in the exercise of deliberation, we are almost inevitably engaging in imperfect reasoning strategies, as we cannot have all of the required evidence to make perfect decisions (Bortolotti 2015a, 18).11 Despite the epistemically faulty nature of our decision-making procedure, however, the process of reason giving itself can confer epistemic benefits (Bortolotti 2009). The way this comes about is that through reason giving we might come to better understand ourselves, and, significantly, we may have to make explicit any implicit biases12 that we have. Once explicit, these commitments can be challenged and revised if shown to be false when faced with evidence to the contrary. When conversing with others, we often engage in practices in which we question the intentions that they may have when performing or not performing certain acts (‘why don’t you donate some of your salary to charity?’, ‘why do you still eat factory farmed meat?’, etc.). Through debate and dialogue with one another we can progressively adopt more epistemically sound beliefs. In order to do this, however, our initial, potentially false, belief that we are free is presupposed. In this way belief in free will, belief that we really are the ‘willers’ of our actions, opens up the possibility of ‘peer-review’ for our beliefs.
Additionally, epistemic gains afforded by this potentially faulty cognition are linked to the enhanced sense of psychological well-being associated with “perceived control” (Wegner 2012). People who believe themselves to be the causes of events are more likely to be psychologically healthy. An example of this was uncovered in a study that investigated the coping mechanisms displayed by people who had recently been involved in paralyzing accidents (Bulman and Wortman 1977). In brief, the study found that those who attributed the cause of events as external (“someone else did it”, “it was random”) struggled to cope as well as those who characterized the events internally (“I was responsible”). These feelings of control are therefore positively correlated with psychological well-being, which in turn is correlated with a willingness to engage with one’s environment and peers.13
We might also think that social interaction encourages us to make explicit our beliefs and facilitates a process of interpersonal hypothesis testing. This interpersonal hypothesis testing is an inherently social phenomenon, insofar as it requires others to listen and potentially respond to what we are saying. It is natural for us to want to be liked and admired by our peers, and so it might be plausible to think that such social interaction might motivate us to do what we think is morally right, increasing our desire to discover something like moral truths. By communicating with others, we become accountable to them and ourselves. We might state our wishes and desires or express our values. Having expressed these there is a pressure to actually follow through: If I claim to be charitable, I had better express this virtue when the situation demands. Of course, there is the very real worry that our social groups might also encourage the formation of epistemically harmful beliefs, but this would not be true in all cases. More importantly for my purposes, however, it is crucial to note that this mechanism enables individuals to gradually acquire a greater number of accurate beliefs, which constitutes an epistemic benefit.
Once again, by making our problematic beliefs explicit we open ourselves up to the opportunity of being proven wrong: Others who might know better than us can correct our faulty beliefs and we can then make an attempt at improving the veracity of our cognitions. In other words, while it might be the case that psychological well-being facilitates this process, there is nonetheless an epistemic component to this type of cognition, however minimal it might be. This is not to claim that the epistemic benefits outlined here are stellar. Rather, what I am modestly suggesting is that a simple psychological account of this belief may miss the greater epistemic picture.
3.2 No Alternatives
The second criterion required for a belief to be considered epistemically innocent is the No Alternatives condition. I will suggest that any alternative to the belief in free will is explanatorily unavailable (Sullivan-Bissett 2015, 554). It is important to note, however, that even if my argument here is unsuccessful, it does not take away from the epistemic benefits outlined above. Should the No Alternatives condition not hold, belief in free will might not be epistemically innocent, but this would not force us to conclude (based on what I have argued above) that it has no epistemic benefits at all. Additionally, the No Alternatives condition refers to specific agents and their beliefs. Thus, it is not a general claim that no belief other than belief in free will is explanatorily unavailable for all agents. Rather, it is about particular agents and their beliefs. In order for this condition to obtain, therefore, it should be the case that from the perspective of the agent no other belief is available that confers the same epistemic benefit.
A belief is explanatorily unavailable when it is “dismissed [by the subject] due to its apparent implausibility” (Sullivan-Bissett 2015, 554). These are cases in which an agent may have certain beliefs about experiences they have had, and where alternative accounts that explain the belief or cognition strike the agent in question as implausible or insufficient (such as the porcupine example introduced earlier). Consider again our feeling of ourselves as unified agents with free will. This feeling, and our subsequent belief in its truth, might be false. However, when we introspect, we are constantly confronted with the fact that we continue to feel and experience ourselves as free. In other words, no other explanation for the way we feel about the actions we perform is available. Of course, different individuals may report different degrees of freedom: The point is that, for some, such a feeling might be stronger, and so other explanations for their actions may be unavailable.
It is for this reason that I believe that no other belief is explanatorily available. The justification for this claim relies on the fact that there is a close relationship between our perception of each other as free and our ascriptions of moral responsibility. We tend to hold others responsible for their actions because we believe them to be in control of what they are doing, and it is this sense of responsibility that is fundamental to the successful functioning of society. The perception that we are in control of our actions makes us responsible in this morally credible sense.
Before concluding it is important to note a comparison between us believing in free will versus us believing in our lack of freedom.14 There is difference between us having free will being true in fact and us believing it to be true (Duus-Otterström 2008, 223). If it is true, this does not mean that people in general or policymakers specifically will radically change the way they act. The real worries arise when we believe that we are not free. Now it is beyond the scope of this paper to go into detail with respect to this claim. Suffice to say that it is not at all obvious that us not being free would be that disastrous (Pereboom 2003, 2014). However, I do think there are serious consequences to such a belief that may have a negative effect on overall epistemic functionality. Notions such as basic desert would have no justification, as nobody would ever, strictly speaking, deserve anything, as who they are and what they do would not be a product of their will. Such responsibility ascriptions are important to the functioning of society more generally, but also for us as individuals. Believing ourselves to be free prompts us to take seriously the fact that we are, in some sense, in control of our actions and can be held accountable for them.
Such accountability leads us to want to be better, which causes other epistemic gains, as noted above. We acknowledge that we are responsible for our actions and seek to act in ways which are morally appropriate. This encourages our pursuit of what is truly morally correct, as we strive to improve as moral agents. Being better moral agents implies that we are better informed about what to do, and thus have a larger reservoir of information when it comes to making morally laden decisions. It is in this sense that such a belief in free will may have positive epistemic consequences which would not be possible in its absence: we believe we are free, facilitating justified ascriptions of responsibility, which enables us to be better informed moral agents. It therefore seems desirable for us to believe we are free.
4. Concluding Remarks
In this paper, I have argued that a false belief in free will might be justified on epistemic grounds. There are cases in which it is possible that the adoption of a false belief can prevent an epistemic harm from occurring, and in such a case we may say that the belief is innocent. I have argued that free will is one such epistemically innocent belief. This belief, while not epistemically good overall if it turns out to be an illusion, was found to offer clear epistemic benefits to the individual, such as a more coherent sense of self and the acquisition and retention of true beliefs over time.
Furthermore, it was found that no alternatives, other than belief in free will, are explanatorily available which confer the same epistemic benefits. This is perhaps the most obvious weak point of the paper, as there are many people who do not in fact believe in free will, and so this belief does indeed seem to be available. Moreover, it seems as though one could not believe in free will and yet gain the epistemic benefits I’ve described above. My response to this charge, linked to what I have already said in Section 3.2 above, is that perhaps such a belief is only available to some agents. That is, for some, belief in free will is the only means to these epistemic benefits, whereas for others this might not be the case. What exactly might explain this difference is beyond the scope of this paper. Importantly, however, this point is predicted on the fact that the No Alternatives condition is about particular agents and their beliefs, and not about beliefs in general.
To reiterate, this is not to say that this belief is epistemically good overall, but rather to claim that there are at the very least some epistemic gains to be had. Consequently, in conjunction with the many pragmatic benefits of this belief that have been the focus in much of the literature (such as an enhanced sense of moral responsibility), there might also be further epistemic benefits that have yet to be explored. The implications, therefore, for future investigations into the philosophy of free will are that we should consider whether we have been too narrow in our pragmatic defences of free will, and that we should also be sensitive to epistemic considerations.
Acknowledgments
I would like to thank my former supervisor, Dr. Tanya de-Villiers Botha, who played an important role in shaping the ideas presented here. Two anonymous reviewers of this journal provided insightful comments, which greatly improved the quality of this manuscript. I am also thankful to Stephanie Meek, who proofread the paper and made perceptive suggestions.