Epistemic Reasons Are Not Normative Reasons for Belief

Abstract

In this paper, I argue against the view that epistemic reasons are normative reasons for belief. I begin by responding to some of the most widespread arguments in favor of the normativity of epistemic reasons before advancing two arguments against this thesis. The first is supported by an analysis of what it means to “have” some evidence for p. The second is supported by the claim that beliefs, if they are to be considered as states, cannot have epistemic reasons as normative reasons.

1 Introduction

The standard approach to framing the debate regarding what we ought to believe in the ethics of belief is to oppose both evidentialism and pragmatism. Evidentialists defend the idea that only epistemic reasons¹ can be proper normative reasons for belief, and that practical reasons for belief are not really normative reasons for belief. Pragmatists hold the position that practical reasons—reasons pertaining to prudential or moral considerations for belief—can be normative reasons for belief in their own right. While this dispute has been useful in framing current debates, a distinct mapping of the ethics of belief should include more views.²

With respect to what is considered a normative reason for belief, there are five views one can hold, the first being that there are no normative reasons for belief and, as such, one might be qualified as an error theorist with respect to our normative judgements about belief. Evidentialists hold either that beliefs can be justified by both epistemic and practical reasons, resulting in different types of justification, or that epistemic reasons are reasons for belief while denying that practical reasons can constitute reasons for belief. Evidentialism would then regroup two distinct flavors of evidentialism: incommensurability evidentialism and totality evidentialism.³ I shall term a “classical pragmatist”⁴ someone who holds that both epistemic and practical reasons can be normative reasons for belief and that we can reach an all things considered judgment regarding what to believe after weighting on both sets of reasons. I write “classical pragmatism” because I wish to contrast this position with the position that only practical reasons are normative reasons for belief. Let us title this last position totality pragmatism.

My goal, in this paper, is to rebuke both the evidentialist’s and the classical pragmatist’s position by attacking the thesis that epistemic reasons are normative reasons for belief. I will argue that if a normative reason for believing p is one that justifies praising or blaming the agent for believing p, then epistemic reasons are not normative reasons for belief but are instead explanatory reasons for belief. I do not intend to defend the remaining positions, i.e., the error theorist’s position or totality pragmatism in this paper. I shall leave the task of delivering a positive account of each viewpoint for another time. I will, however, ask the reader to let me presume the falsehood of the error theory for the sake of the argument. In doing so, I invite my reader to presume, like me, that we are usually correct in offering praise or assigning blame to agents for their beliefs and that these reactive attitudes are indicative of our intuitions regarding what we ought to belief.⁵ Blame, in this sense, will be the major touchstone when expounding my arguments about what people should believe,⁶ and I will further suppose for the remainder of the paper that doxastic agency and responsibility for our beliefs requires a certain kind of control on our beliefs, and that an agent lacking any sort of control on a belief will consequently not be a proper object of blame or praise.

In the following pages, I will attempt to answer some of the strongest arguments according to which we ought to believe what the evidence favors. Subsequently, I will offer two arguments against the thesis that epistemic reasons are normative reasons for belief. The first argument is supported by an analysis of what it means to “have” some evidence for p. The second is supported by the claim that beliefs, if they are to be considered as states, are consequently not normative. In that sense, this second argument is a variation on the trend to frame the question of belief normativity in terms of states.

2 Some Arguments for the Normativity of Epistemic Reasons

It is commonly acknowledged that epistemic reasons are normative reasons for belief. When someone fails to believe what is clearly supported by epistemic reasons, we tend to blame them for failing to do so. In the literature stemming from the ethics of belief, when positive arguments are made in favor of the normative character of epistemic reasons, it is usually in the context of comparing epistemic reasons to practical reasons for belief, especially when those different sets of reasons favor incompatible beliefs. A favorable way to engage this debate for the benefit of my proposal would be to examine what it means to comply with the evidence at hand. I believe that one of the strongest arguments in favor of my proposal that only practical reasons for belief are normative reasons for belief is based partly on the notion that belief formation is not something the agent does, or something he controls (at least, not directly). This puts me at odds with doxastic voluntarism, which is the view that there is no significant difference between forming a belief and other actions that an agent might undertake (Steup, 2012; Marušić, 2011).⁷ Usually, the most common way to defend this position is by arguing that the act of judging is an action that mediates belief formation (Cassam, 2010; Gerken, 2014; McHugh, 2015; Shoemaker, 2009; Soteriou, 2005). This is explicitly defended by Quassim Cassam (2010: 82–83) in the following passage: “Judging is a mental action […]. Suppose that I am presented with a sound and valid argument for some proposition P. I go through the argument and conclude that P. Concluding that P is just judging that P, so here we have a case in which the formation of belief is mediated by judgment.” For the doxastic voluntarist, we reflect on a proposition after evaluating considering the evidence, which leads to forming a certain belief about that proposition. Mathias Steup (2011: 551) gives the following example to illustrate this process:

“Suppose that, having returned from a trip and taken a shuttle to the airport parking garage, I am now where I thought I left my car. To my surprise, it is no longer there. I wonder whether it has been stolen. There is, of course, the possibility that I don’t accurately remember where I parked it. So, I retrieve the paper slip which states the exact location of my parking spot. According to the slip, I am at the right spot. Considering my evidence — the parking slip and the absence of my car — I conclude that it was stolen.”

In this example, Steup claims that after examining his parking slip, he judges that he is at the right spot. The belief that he is at the right spot is mediated by him considering the evidence and judging that this is sufficient. I wish to dispute the idea that judging, inasmuch as it can be considered an action the agent undertakes, is motivated by epistemic reasons. What happens, from a phenomenological standpoint, when I judge that the car was stolen? It would appear, from a phenomenological standpoint, that judging is a two-step process which includes, firstly, bringing the evidence one has to one’s attention and, secondly, forming the relevant doxastic attitude. Of the two, only the former is under the agent’s control. If the latter step was under the agent’s control, it should be possible for the agent to decide that the evidence presented is not sufficient to form a belief. This is clearly not the case. Let’s take an example were there are no If I notice a cup of coffee on a table, I cannot decide that the evidence before me is not sufficient to form the belief that there is a cup of coffee on the table. Rather, I will be compelled by the evidence, and will irresistibly form the belief in question. Furthermore, if we were able to decide to form a belief, or abstain to form one, in light of the evidence presented, the decision should be motivated by a certain reason. This reason should be epistemic in nature, or else we are led to hold the position that what we ought to believe ultimately depends on practical reasons. But if it is so, new problems spring out of the earth. Let us suppose that you must decide whether to form a belief based on the evidence that you brought to your attention.

If this decision is not made based on a practical reason, it must otherwise be reached on the basis of an epistemic reason. However, should the decision that you have sufficient evidence to form a belief is itself based on an epistemic reason, you will need to determine if this second epistemic reason is sufficient to settle the matter at hand. This will most likely require more epistemic reasons, leading to infinite regression; otherwise, the agent will need to base his decision on a practical reason or give up on the decision-making process, thus removing the second step of judging from his control.⁸

Returning to the act of judging, it seems fruitful to analyze it as a two-step process; the first one consists in bringing the evidence I have to my full attention, while the second remains passively convinced (or left unconvinced) by the epistemic reasons I examine. Only the first step is under my control and it should be plainly clear that practical reasons motivate it. When we reflect on Steup’s example, we note that the act of recalling the parking spot and placing the evidence to the forefront of his attention is motivated by practical reasons. This suggests that the act of judging, so long as it involves our agency, has more to do with practical reasons than epistemic reasons. Hence, even if judging is an action the agent might undertake, we should carefully distinguish the steps of that action that are not under the agent’s control. Unsurprisingly, what falls short of being under the agent’s control also involves the agent’s epistemic reasons.

It might be argued that even if epistemic reasons are not mobilized in the belief-forming process that involves our agency, we may nonetheless hold that we ought to believe according to our epistemic reasons (Feldman, 1988, 2008; Wedgwood, 2002, 2013; Kvanvig, 2003; Lynch, 2004a, 2004b; Shah & Velleman, 2005; Shah, 2006; Vahid, 2006, 2010; Horwich, 2010). One of the most compelling arguments to support this position without committing to any claims about our agency is the reasoning put forth by Pamela Hieronymi (2006, 2008) in which she discusses the role of constitutive reasons.⁹ When talking about constitutive reasons (reasons that bear on the question of whether p), we distinguish them from extrinsic reasons which bear on whether it is desirable to believe p. It is significant to note that, on its own, this distinction says nothing about what we ought to believe: constitutive reasons indicate what is the case, which is not sufficient to indicate what one ought to believe. The gap is then bridged with Hieronymi’s following argument:

“Any activity or attitude for which one is answerable – for which one can be asked one’s reasons – will be reasonably understood as (or as a result of) the settling of some question (or set of questions) on which such reasons bear. For example, one can be asked for one’s reasons for believing p, and believing p can be understood as settling (or having settled) for oneself (positively) the question of whether p – the question on which one’s reasons for believing p will bear.”¹⁰

As we usually justify holding of some belief by citing our constitutive (epistemic) reasons for that belief, it follows that we ought to believe what our epistemic reasons indicate.

As I will further discuss, compliance with our epistemic reasons is not something over which we have control and thus, it is not something we “ought” to follow in any relevant normative sense. Hieronymi’s argument is ingenious in that regard as it offers an explanation as to why we ought to believe what is indicated by our epistemic reasons by discussing answerability rather than agency. Nevertheless, I wish to make a case for doubting Hieronymi’s argument. While it is usually true that we invoke evidence when we explain why we hold a particular belief, it is not always the case. We sometimes justify holding a belief by citing practical reasons. This is most often the case when we give reasons for someone else’s belief. Imagine you are talking with your friend Claude about Antoine, a friend you have in common. Antoine has been working for several years to complete his PhD, which, of course, is a difficult task. Despite numerous problems, Antoine is confident he will earn his PhD in no more than a year. Baffled, you ask Claude why Antoine believes this. Claude responds by saying that Antoine believes he will earn his PhD in a year because it is a reassuring idea to entertain. “Believing that he will finish his PhD in a year calms him and helps him continue with his work. For this reason, he is not dragged down by the recurrent challenges of his program. This is the reason why he believes he will finish in the next year.” Now it appears to me that Claude gives a perfectly acceptable reason for Antoine’s belief, even if this reason is practical in nature. It is important to note that Antoine may hold a true belief, perhaps epistemically justified by some epistemic reason in some cases, even if the reason Claude gives for Antoine’s belief is practical. While I do not think this is damning Hieronymi’s argument, more should be said about the differences between the reasons invoked to answer for a belief held in a first-person point of view or third-person point of view.

Yet a stronger counter-argument can be made against Hieronymi’s reasoning. Recall that for Hieronymi, an activity, or an attitude for which one is answerable is understood as (or resulting from) the resolution of some question on which some relevant reasons bear. However, it is important to note that believing may be viewed as nothing more than the final result of an inquiry, the accumulation of evidence then presented to the agent’s attention. In that sense, the activity for which the agent is answerable with respect to his belief may be the inquiry and the review of the evidence he considered. The reasons for believing a certain proposition p would then not be the reasons establishing that p is the case, but rather the reasons for accumulating the evidence and the reasons for considering the evidence one holds. Naturally, such reasons will be practical rather than epistemic reasons. This alternate explanation to what is answerable with respect to belief is coherent with the idea that belief is an attitude one acquires after a process. The result of this process may be criticized so long as we can hold the agent answerable for the process that preceded this result. This suggests that Hieronymi’s account does not clearly rule out practical reasons as genuine reasons for belief, and even make them the kind of reasons that makes us answerable to the activity of forming a particular belief. Furthermore, Hieronymi’s account, given that it focuses entirely on epistemic reasons, cannot help us distinguish a situation in which the agent is responsible for her belief from a situation where the agent is not responsible for her belief. If the agent is hypnotized into believing that Canada is still a colony, she is most likely not answerable for her belief.¹¹ But why is it so? Hieronymi’s account says nothing on this subject, nor was there any pretension to offer any explanation of the sort. Nonetheless, it is interesting to note that once we consider belief as the result of gathering and exposing ourselves to what provokes the acquisition of belief, the mystery then unfolds. The person who believes that Canada is a colony will be answerable for her belief in the event she has some control over the process of belief formation. For instance, she will not be answerable for her belief if she was hypnotized against her will. She will be answerable in the case where she was lax when gathering evidence purporting Canada’s colonial status or if she actively sought to acquire the belief via hypnosis. Again, agency seems to lurk in the background when we talk about responsibility for our beliefs.

The last argument I wish to examine in favor of the notion that we ought to believe what the evidence indicates is what is occasionally called the motivational argument. This argument is rather simple. There are only two possible types of reasons for belief: practical reasons and epistemic reasons. Of these two types of reasons, only the epistemic reasons can (consciously) motivate the formation of a belief. In contrast, practical reasons for belief cannot do this. If I were to offer you a million dollars to persuade you to believe that the moon is made of cheese, you will not be able to form this belief on the basis of the financial incentive I offer. Given that practical reasons cannot motivate the formation of a belief, we ought not to form beliefs on the basis of practical reasons. Hence, the only reasons for which an agent ought to form a belief are epistemic reasons.

The most convincing thought experiment in favor of the motivational argument must be that of Thomas Kelly (2003: 626), which I would like to quote here in its entirety:

“Not only are there (many) subjects with respect to which I have no preference for having true beliefs, there are also subjects with respect to which I would prefer to have no beliefs at all to having true beliefs. Thus, I tend to see newly released movies after many of my friends. During the interval of time which is bounded on one side by my friends’ viewing of the movie and bounded on the other side by my viewing the movie, I often make a conscious, deliberate effort to avoid finding out how the movie ends—since doing so might very well interfere with my enjoyment when I do see it. (When conversations about the movie begin in my presence, I either excuse myself or, reminding the discussants that I have yet to see the movie, implore them not to “give away” the ending, and so on.) That is, I quite deliberately take steps to avoid acquiring information about the movie. Sometimes these efforts are successful, sometimes they are not. When they are unsuccessful—as when someone inconsiderately blurts out the ending in my presence—it does not follow that I have no epistemic reasons to believe the propositions which he asserts. Indeed, with respect to the question of which epistemic reasons I possess, there is no difference between this case and a case in which I ask the individual to tell me the ending because I do have some goal which would be better achieved by my believing the relevant truths. The fact that in the one case I do have a goal which is better achieved by my believing the relevant truths, but in the other I have no such goal—indeed, I have goals which would be hindered or frustrated by my believing these truths—makes no difference to my epistemic reasons.”

Now, it is important to note that Kelly was using the motivational argument to argue against the instrumental view of epistemic rationality; the view that epistemic rationality is a subset of instrumental rationality (defended by Giere, 1989; Kitcher, 1992; Laudan, 1990). Like Kelly, I do not believe that epistemic rationality is a subset of instrumental rationality, and I also believe that in the case presented, the agent is not to blame for acquiring the belief that the movie ends in a particular manner, even if this goes against his objectives and practical reasons for belief. I think, however, that there is much to say against the motivational argument.

My chief complaint against the motivational argument would certainly be that there is a strange magic trick unfolding in which we seem to shift from a descriptive to a normative claim when talking about the power of epistemic reasons. Recall: what justifies the idea that we ought to believe according to our epistemic reasons is the fact that only epistemic reasons can motivate the formation of a belief. This is a suspicious claim. To be sure, there are cases in which the fact that only one means exists to reach a certain state of affairs is a normative reason to use that means, given that we should reach the state in question. Let us suppose that I aim to visit a friend and the only means to reach this goal is to take the car, then it seems I have a normative reason to take the car to visit my friend. But, as Kelly showed with his example, our cognitive goals do not impact the rationality of belief formation. Now, it seems rather clear to me that the moviegoer in Kelly’s example does not have the obligation to acquire the belief that the movie ends as the other person said it does. Certainly, he may be forced to form the belief that it ends in such manner due to the strength of his epistemic reasons. However, this does not mean he has the obligation to, or that his epistemic reasons are normative reasons to believe that. In order to demonstrate this, let us imagine for a second that, by some miracle, our moviegoer does not acquire the belief that the movie ends as the person told him it does, despite the strength of the evidence presented. I do not think he would be blamed if this was the case. It does not follow, from the mere fact that we only form beliefs by virtue of epistemic reasons, that we ought to believe—in a normative sense—what our epistemic reasons indicate.

Now, this notion that agents are not actually obliged to follow a normative rule that they really cannot break has been disputed (Ogien, 2001). For instance, Richard Feldman (2000: 678) wrote about belief formation: “It seems to me reasonable to say that when only one attitude is permitted, then one has an epistemic obligation to have that attitude.”

In a sense, this is a very reasonable standpoint. When discussing applied ethics, if only one action is permitted, then it seems that this action is, in fact, obligatory. It is important to note, however, that this notion of permissibility is considerably different from what Feldman and I have in mind. To illustrate this, let us suppose that I am pondering whether or not to destroy the International Space Station. As it stands, it is impossible for me to destroy the International Space Station. Given the circumstances, it would be surprising to say that I am under the obligation to refrain from destroying the International Space Station. This is suggested by the fact that I do not deserve any praise for fulfilling the moral obligation of not destroying the International Space Station when I simply do not do so due to a lack of opportunity. Furthermore, arguing that we have an obligation to follow a rule that we have no choice but to follow opens the door to a myriad of moral obligations, such as the constant obligation to abstain from hurting people we cannot hurt and to avoid lying to people we will never meet.

Permissibility is not the same as possibility. From the mere psychological fact that we can only (consciously) form doxastic attitudes by virtue of our epistemic reasons, it does not follow that we have the normative obligation to do so. This is one reason why the motivational argument fails, the other reason being that it misconstrues the relation between belief formation and practical reasons for belief.

3 Some Arguments Against the Normativity of Epistemic Reasons

In the previous section, I’ve discussed some arguments in favor of the thesis that we ought to believe according to our epistemic reasons. In this section, I wish to present two arguments in favor of the thesis that epistemic reasons are not normative reasons. At this juncture, it would be useful to state what I think are epistemic reasons.

When we say that p is a reason for X, we usually mean one of two things. We often use “reason” in a normative way that can justify praise, or blame, or establish a basis for using deontic terms such as obligation or permissibility. If not, we usually use the term “reason” in a non-normative way, as we attempt to explain how a certain state of affairs came to be. For instance, when I say “The reason why the Titanic sank was because it hit an iceberg,” I am not saying anything normative. I am merely explaining why the Titanic sank. This is what I believe happens with epistemic reasons: they provide an explanation as to why we believe what we believe because we are psychologically geared to respond to them by forming doxastic attitudes. Similarly, when we say that we ought to believe according to the evidence, we are merely making a prediction as to the correct functioning of our psychological apparatus. Again, speaking of epistemic reasons in the context of making a prediction or giving an explanation is not the same as talking of reasons in a normative sense, where we usher in justification for blame and praise.

Prior to making my two arguments for the thesis that epistemic reasons are not normative reasons, I want to say a few words about what it means to have some evidence for p. Keith DeRose (2000: 700) once highlighted that the expression “having an epistemic reason for p” has at least three distinct interpretations:

• 1. “Having an epistemic reason for p” as denoting the fact of having epistemic reasons presented to one’s attention.

• 2. “Having an epistemic reason for p” as denoting the fact of having epistemic reasons that are not presented to one’s attention, but are mentally available.

• 3. “Having an epistemic reason for p” as denoting the fact of having epistemic reasons that are not mentally available, but easily available through some physical actions.¹²

For the sake of brevity, let us call the first interpretation as directly having an epistemic reason for p, and both the second and third interpretations as indirectly having an epistemic reason for p. This seems warranted, as the agent having some evidence understood in the second and third interpretations must take action in a physical or mental way to bring her evidence to the forefront of her mind and make the evidence causally effective in the formation of the belief (in short, the evidence one has “indirectly” cannot directly cause the formation of belief).

Now, on to the first argument:

• 1. Epistemic reasons are either directly or indirectly possessed by agents.

• 2. For any normative reason r, it should be possible for an agent to not follow r and be held responsible for not following r.

• 3. It is not possible for an agent to consciously not follow an epistemic reason r the agent indirectly has and be held responsible for not following it.

• 4. If an agent cannot consciously X based on some reason s, then s is not a normative reason to X. (From premises 2 and 3)

• 5. An agent cannot consciously form a belief p at time t based on an epistemic reason r that the agent has indirectly at time t.

• 6. Epistemic reasons that an agent indirectly possess are not normative reasons for belief. (From premises 4 and 5)

• 7. It is not possible for an agent to not follow an epistemic reason r the agent directly has and be held responsible for not following it.

• 8. Epistemic reasons are not normative reasons for belief. (From premises 1, 6, and 7)

So let’s summarize the argument before defending its premises: an epistemic reason is either directly or indirectly held by agents. If it is indirectly held by an agent, that agent cannot be held responsible for the belief formation that such reason induces. When an epistemic reason is directly held by an agent, that agent is not responsible for whether that epistemic reason causes the formation of a belief. If blame or praise cannot be attributed to an agent by virtue of following (or not following) the epistemic reasons held by that agent, by it directly or indirectly, then epistemic reasons are not normative reasons; they do not justify normative attitudes like praise or blame.

Let us discuss the premises, beginning with premise 7. Previously, I provided grounds to doubt that agency is involved in the formation of belief when one’s epistemic reasons are under one’s attention when I discussed Steup’s position. The case presented by Kelly also shows that we will not blame someone for forming a belief that goes against her goals once that person is unfortunate enough to be accidentally presented with relevant evidence. Conversely, if someone, when presented with strong evidence, fails to form a belief by virtue of some psychological defence mechanism that furthers that person’s goal, then that person should not be blamed for failing to form the belief in question. The following is an example:

Terrible news has befallen Thomas. While watching the news, he learns that the plane that his son most likely took has crashed. There are no survivors. The plane’s number matches that on his son’s ticket, the destination was the right one, and Thomas has no reason to believe that his son missed his flight. Still, despite the evidence that his son has perished, Thomas cannot form the belief that his son died.

I will presume that most people will agree that Thomas is not to blame for failing to form the belief that his son is dead, even if the evidence for that proposition is strong. This is so because once the agent is fully aware of the evidence, he or she has no control over the effects that the evidence will have on the process of belief formation. In short, the reason why the case of the moviegoer and the case of Thomas both present us with blameless agents is that they lack any form of control over the formation of belief once the evidence is presented to them. At the conclusion of both thought experiments, both agents are confronted with the evidence and they are unable to control how they will react to this evidence regardless of their respective goals. My defence of premise 2 rests on what I believe to be a very natural conception of normative reasons: an essential property of normative reason is that it should be possible for an agent to not follow what his or her normative reason dictates; furthermore, not following said normative reason should justify an attitude of blame towards the agent. In the case of epistemic reasons, we are never able to resist their power except in situations which absolve us from any blame whatsoever. This may, in fact, be my strongest claim against the normativity of epistemic reasons.

Let us examine premise 5 before turning our attention to the premise 3. If the agent has some epistemic reason, but he or she has not yet considered that reason, then said reason cannot consciously form a belief, at least, not until the agent brings it to his attention (I will address the topic of unconscious belief formation in a moment). It is important to note that whether or not the agent should bring the evidence to the forefront of his attention depends on practical reasons. Indeed, if the agent indirectly possesses the evidence, then the agent will be required to undertake some action to bring it to his attention, be it physically or mentally. If no action is taken, then the agent will not be able to (consciously) form the relevant belief. Following the Kantian adage that obligation implies possibility, it would appear that the agent cannot be held responsible for a belief that he cannot form. Being responsible for a belief will then be contingent on the agent being able to act in such a way that the evidence he indirectly has will be presented to his attention. Carrying out a difficult inquiry, collecting the evidence in an exhaustive manner, and drawing the relevant conclusions from said evidence all require some mental efforts that the agent will make not on the basis of some epistemic reason he has, but rather on the practical reasons he has about forming the relevant belief. Normativity of belief then, so long as we consider epistemic reasons to be indirectly possessed by the agent, is less about what the epistemic reasons indicate than what the practical reasons invite us to do with the epistemic reasons we have.

Now, some may say that we are responsible for beliefs that we unconsciously form when they result from wishful thinking. This gives me the opportunity to defend premise 3 of our first argument. While technically correct, this statement is somewhat imprecise. If you unconsciously form a belief, you are not responsible for the formation of that belief (the psychological process is, by definition, out of your hands). However, you may very well be responsible for not revising that belief if you discover that it is based on wishful thinking. So, whether or not you are responsible for an unconsciously formed belief will most likely depend on how long you have had the belief and at what point it is reasonable to expect an agent to re-evaluate a belief formed in such a way. If I unconsciously formed a belief based on wishful thinking five minutes ago, I may not yet be responsible for that belief, for it would demand that I have the opportunity to critically examine my own belief, and five minutes feels way too short a time frame for that. It is important to note that my revising this belief will be predicated on practical reasons to do so. If I have a belief that affects no one in any way, I am most likely not required to revise said belief. If some important consequences result from my having this belief, I may be required to revise it rather quickly. Again, the normativity of belief formation and revision seems to crumble into practical considerations.

Let us now turn to a “quick and dirty” second argument that aims at establishing that epistemic reasons are not normative reasons for belief. This argument relies what we previously said regarding equating direct belief formation with the act of judging.

• 1. If it is not possible for S to act on the basis of r in a way that allows S to reach a certain state e, then r cannot be a normative reason for agent S to be in a certain state e.

• 2. Believing is a state.

• 3. If it is not possible for S to act on the basis of r in a way that allows S to believe that p, then r cannot be a normative reason for agent S to believe that p.

• 4. It is not possible to do something that leads to the belief that p on the basis of some epistemic reasons indicating that p.

• 5. Epistemic reasons indicating that p are not normative reasons for believing that p.

This argument is rather straightforward and pairs with similar pragmatist arguments defining beliefs as states (Rinard, 2015). I do not think premise 2 requires a thorough defence; rather, I shall focus on the first and third premises.

“Acts are things onedoes, but there is no answer to the question “What are you now doing?” which goes, “I am now believing that it will rain” […]”.¹³ I think Searle was right about this. Beliefs are things that happen to agents, rather than something that agents do. To appreciate this, let us examine the distinction between believing and accepting. Accepting is a technical term extremely close to that of belief, in the sense that it is the attitude of holding a certain proposition as true, and furthermore, it can motivate actions in the same way beliefs do (Van Fraassen, 1980; Cohen, 1989; Velleman, 2000; Zalabardo, 2010; Shepherd 2018). Jonathan Cohen (1989: 368) gives an excellent definition of the term:

“[To] accept that p is to have or to adopt a policy of deeming, positing, or postulating that p – that is, of going along with that proposition (either for the long term or for immediate purpose only) as a premise in some or all contexts for one’s own and other’s proofs, argumentations, inferences, deliberations, etc. Whether or not one assents and whether or not one feels it to be true that p. Accepting is thus a mental act (as what was ‘judgement’ often used to be), or a pattern, system, or policy or mental action, rather than a speech-act. What a person accepts may in practice be reflected in how he or she speaks or behaves, but it need not be. […] Acceptance implies commitment to a policy of premising that p.”

So, what separates believing from accepting? I want to argue that it is their phenomenological characters that separate them. What does it mean, from a phenomenological standpoint, to believe a proposition? Nothing more than a feeling of trust, the confidence that the proposition is true.¹⁴ You do not have to have this feeling of confidence to accept a certain proposition, but it is impossible to believe a proposition and not feel confident to some degree that the proposition you believe is true. This explains why the first premise is so compelling: feeling confident about something is not something an agent does, it is something that happens to an agent. The agent who feels confident that a certain proposition is true—in short, who believes the proposition—is in a certain state that came upon him.

What can an agent do to enter a state of belief? Gathering and reviewing the evidence and keeping his biases in check are all actions performed by the agent, but they are not initiated on the basis of epistemic reasons. An epistemic reason is simply a fact (or a proposition, depending on the philosophical theory you favor) which indicates that a certain proposition is more likely to be true than false. But the fact that a certain proposition p is more likely to be true than false is not a normative reason to do anything.¹⁵ The fact, for instance, that the picture on the wall is most likely to be that of my grandmother is not in itself a reason to do anything. Of course, taking note of this fact might cause the formation of a belief, but this belief formation is not something that I do, even if it is caused by some epistemic reason I might have. Actions that lead to the formation of belief, such as taking a look at the picture on the wall, often have no normative reasons to support them. But when there are, those reasons are practical and not epistemic reasons. Some might be tempted to fall back on a position according to which belief is mediated the act of judging. Unfortunately, we have previously seen that this option will simply not do. Initially, we stated that judging is a two-step process and that the step involving epistemic reasons is not one the agent does. The argument then reiterates this idea that the direct formation of belief is both something influenced by epistemic reasons and something that is entirely out of the agent’s hands.¹⁶