Believing for a Reason is (at Least) Nearly Self-Intimating

Abstract

This paper concerns a specific epistemic feature of believing for a reason (e.g., believing that it will rain on the basis of the grey clouds outside). It has commonly been assumed that our access to such facts about ourselves is akin in all relevant respects to our access to why other people hold their beliefs. Further, discussion of self-intimation—that we are necessarily in a position to know when we are in certain conditions—has centred largely around mental states. In contrast to both assumptions, this paper argues that believing for a reason is (at least) very nearly self-intimating: necessarily, if a subject believes that q for the reason that p, then, provided relevant conceptual and rational capacities, she is in a position to form a justified true belief that she believes that q for the reason that p. We should think this on the basis of the role that believing for a reason plays from the subjects’ perspective, and in particular, the way in which it intellegises one’s belief.

This paper argues that necessarily, if a subject believes that q for the reason that p, then, provided relevant conceptual and rational capacities, she is in a position to form a justified true belief that she believes that q for the reason that p. For brevity, I will henceforth say that she is in a position to self-ascribe her reason. E.g., necessarily, if Sally believes that it will rain for the reason that there are grey clouds, Sally will be in a position to self-ascribe believing that it will rain for the reason that there are grey clouds. If we take self-intimating conditions to be those we are in a position to know, this paper argues that believing for a reason is very nearly self-intimating (which isn’t to preclude that further arguments can show that it is in fact self-intimating). I argue for this by considering the role of believing for a reason from the subject’s perspective, and in particular, the way in which it intellegises one’s belief.

I proceed as follows. Section 1 introduces the topic at hand. Section 2 uses what I call the argument from intelligibility to argue that necessarily, believing for a reason is at least nearly self-intimating, and I elaborate on this thesis in Sect. 3. Section 4 then addresses three objections and Sect. 5 notes several consequences. On surprising upshot is that believing for a reason potentially bears a tighter necessary connection to self-ascription than belief itself does.

Introducing Our Topics

This paper concerns a particular epistemic feature of believing for a reason. I’ll start by saying something about the two main elements in my picture—believing for a reason and self-ascription—and why a claim of this sort may be surprising and significant to many. I will focus on the case of reasons for belief for simplicity, although I take it to be a broader point concerning motivating reasons simpliciter.

First, what do I mean by ‘believing for a reason?’ The following two examples illustrate this phenomena:

• (1) Sally believes that it will rain for the reason that there are grey clouds.

• (2) Nour believes that Smith is a bad politician for the reason that Smith has bad economic policies.

We can distinguish these from other ways in which we talk of ‘a reason’:

• (3) A reason why Sally believes that it will rain is that her light receptors processed perceptual inputs so as to result in a state of belief.

• (4) A reason why Gael believes that the bridge is dangerous is that he is afraid of heights.

• (5) A reason for believing that it will rain is that there are grey clouds.

Examples (1) and (2) cite the reasons for which the subject holds her belief. We can contrast the statements in (1) and (2) with those in (3) and (4), which reference purely causal factors such as low-level processes, biases, and the like. This is important, because a consequence of my view will be that if a subject (granting relevant capacities) isn’t in a position to learn of some factor involved in forming/sustaining her belief, then it is only a reason in this purely explanatory sense and cannot be the reason for which she believes. Further, we can distinguish the first two claims from that in (5). (5) Says something about the normative reasons for the belief—in non-philosophical parlance, it states the good reasons. But just because p is a good reason, it does not mean that p is the subject’s motivating reason.¹ To talk of the subject’s reason for her belief is to say something about the basing relation: for S to have a reason for her belief involves basing the belief on a reason. (Paradigmatically, this will be another belief, but could also be, say, an experience). For example, Sally bases her belief that it will rain on her belief that there are grey clouds.²

There are different accounts of the basing relation. To an extent, this debate is orthogonal to my argument here, but bearing it in mind will prove useful at various junctures. Philosophers usually appeal one of two broad kinds of requirement, or indeed both—doxastic and causal. A doxastic condition requires that for S’s belief that q to be based on her reason that p, she must believe that p is a good reason for believing that q. We can also call this the ‘taking constraint’: the subject must take p to be a good reason for believing that q.³ Indeed, some have argued that it is both necessary and sufficient for believing that q on the basis of p that one believes that p is a good reason for believing that q (see, e.g., Setiya, 2013; Leite, 2004, 2008). In contrast, we might also identify a different kind of putative constraint on what it is to believe for a reason: a causal or disposositional constraint. First, causal accounts argue that ‘for a belief to be based on a reason, the reason must cause the belief in an appropriate way’ (Korcz, 2015). For example, under at least some versions of the account, we would say that a subject bases her belief that q on the reason that p iff it originally caused her to have the belief, even if she has now forgotten that reason. Or, more plausibly, we would appeal to what sustains her belief now.⁴ While some authors take just one of the above conditions to be necessary and sufficient for basing, we might also think that neither suffices, and that both are necessary. As such, we might say that basing requires the subject to believe that p is a reason for believing that q and also that the belief that p must cause the belief that q in the right way (see e.g., Longino, 1978; Audi, 1993; Ye, 2019). I happen to think that both conditions are necessary, but although my loyalties occasionally will become visible, the paper does not rely on this.

I will argue that believing for reason bears a specific epistemic feature—that necessarily, provided one has the relevant concepts, one will be in a position to form a justified true belief that one believes for that reason. This is very close to self-intimation, and it’s worth noting some different strengths of this sort of claim [these distinctions are delineated in Cassam (2014: 42)].⁵ In its extreme iteration, a self-intimation thesis claims that necessarily, if a subject bears a particular feature, she will know that bears it. In contrast, what I call strong self-intimation claims that there are features that a subject cannot fail to be in a position to know that she bears. And weak self-intimation claims that there are features that a subject cannot rationally fail to be in a position to know that she bears. I will specifically argue that our motivating reasons strongly self-intimate in a qualified sense, where I have in mind the more circumscribed ability to form the relevant justified true belief.

Discussions of self-intimation generally centre around certain bodily and mental states. For example, in his discussion of self-intimation theses, Cassam just raises the possibility that ‘sensations, beliefs, and other attitudes’ bear this feature (2014: 42). And Williamson’s (2002) famous criticism of the possibility for self-intimation (and consequent literature engaging with him) focused on whether mental states self-intimate, and in particular, feelings of pain and heat.⁶ That is to say, the question of whether our reasons for belief also bear features of this sort has been relatively overlooked in the central discussions.⁷

Why might this be?⁸ I suspect this absence results from the more general assumption that self-knowledge of why we have our attitudes is akin in all relevant respects to our knowledge of why others hold theirs. This thought holds traction even among those who think that self-knowledge of mental states such as beliefs and experiences is in various ways special compared to others’ knowledge of them.⁹ For fuller discussion on all this, see (Keeling, 2021). But briefly, note for example, that Gertler’s introduction to self-knowledge (Gertler, 2011: 72–75) quickly dismisses the possibility that we might have a distinctive access to the reasons for which we believe, even if we have distinctive access to our beliefs themselves. As such, for example, the thought is that we must rely on inference to learn of our’s and others’ reasons alike.

There seem to be two main reasons why we might doubt that our access to our reasons is in any way distinctive (both of which we find in Gertler (2011: 72–5)). First, subjects at least sometimes engage in confabulation, that is, instances whereby they mistakenly self-ascribe motivating reasons after the fact. E.g., as Gertler discusses, Nisbett and Wilson (1977: 243–4, 249) famously got subjects to state which of four pairs of identical stockings they thought were the ‘best quality’. In tending to pick the pair placed to the right of the table, subjects were influenced by the position. Yet instead, they attributed their preference to supposed features of that particular pair, such as their ‘sheerness’. Perhaps we might worry that this propensity for error precludes anything approaching a self-intimation thesis. Second, if we think of basing as a causal relation, we might worry that subjects could only learn of this relation inferentially, i.e. using the exact same method they would use to ascribe it to others. And further, it might seem implausible that a causal relation would bear any sort of necessary connection to the capacity to self-ascribe this relation. I will revisit both worries in Sect. 3 having introduced the positive case for thinking that in fact, believing for a reason (at least) very nearly self-intimates. I therefore hope to show that we bear a tighter epistemic relation to our reasons than one might have thought.

Believing for a Reason as (at Least) Nearly Self-Intimating: The Argument from Intelligibility

My claim is that if one believes for a reason, provided one has relevant rational and conceptual capacities, necessarily, one is in a position to self-ascribe that condition. The capacities in question are those involved in critical reasoning and higher-order thought, and the grasp of concepts of such as belief and reason. In other words, they are the general cognitive and conceptual capacities possessed by most adult humans. As such, I can allow that some subjects including non-human animals and infants may be said to believe for reasons without being in a position to self-ascribe them. Because my scope is limited in this way, I will proceed by reflecting on the role of basing in subjects with these capacities, i.e., most adult humans. The subjects in the following examples are intended to be of that sort, as is the subject ‘S’ in the below argument.

I will call the following the ‘Intelligisation Argument’; the remainder of this section discusses the premises in turn.¹⁰

PREMISE ONE: Necessarily, if S has a motivating reason that p for believing that q, p makes believing that q intelligible to S.

PREMISE TWO: For a motivating reason that p to make believing that q intelligible to S, S must be in a position to self-ascribe the motivating reason that p.

CONCLUSION: Necessarily, if S has a motivating reason that p for believing that q, then (given S’s relevant capacities and conceptual repertoire), S is in a position to self-ascribe the motivating reason that p.

Premise One

P1 identifies a feature of believing for a reason which I will term intelligibility or rational explicability. I’ll first pinpoint it, before then suggesting why this might be so.

It will help to start by considering the case of acting for reasons. Going shopping for the reason that I need hummus intelligises the action to me in that renders it rationally explicable to me.¹¹ It is not that, as far as I am concerned, I happen to move in that direction as a result of unfathomable spasms (even if I also happen to think that visiting the shop is a good idea because of my hummus shortage). Valaris helpfully contrasts intelligible actions with those of blindsighters who act in response to stimuli while being completely unaware of this: ‘their success comes as a surprise to the subjects; they merely find themselves extending their arm in the right direction and pre-shaping their hand for the right sort of grip; they can make no sense of why they act in these ways’ (2013: 11). Here we can also note Anscombe’s (2000 [1957]) discussion of the question ‘why?’. When asked why I am going to the shop, I can cite my lack of hummus. I do not just treat this as a potential justification, but also as an explanation of my action. The thought, then, is that something similar also pertains regarding reasons for belief.¹²

To start with some illustrative examples, take again Sally who believes that it will rain for the reason that there are grey clouds in the sky. Contrast this with cases in which one does not believe for a reason such as the following example drawn from BonJour (1980). Norman possesses a reliable sixth sense. By stipulation, this ‘sixth sense’ requires no evidence or perceptual experience in order to issue true beliefs concerning various matters. It is not that Norman, say, forms the beliefs on the basis of premonitionary dreams or gut feelings. Norman, though, has no knowledge about his ability—perhaps he has only just acquired it. Say for instance, that his sixth sense has given rise to the true belief that the president is in New York City. Regardless of whether or not his belief is justified, so the story goes, it is not based on reasons. Also, take an alternative case in which the subject even more clearly has an alienated belief.¹³ Suppose Carl is standing front of a table and has OCD which causes him to believe that everything in his environment is contaminated, including that the table facing him is dirty. In comparison to Norman’s belief that the president is in New York City, it comes as no surprise to Sally that she believes that it will rain. She does not find herself inexplicably believing it as if by chance.¹⁴ It is not that Sally finds herself with the belief out of the blue. And unlike Carl, neither does she experience it as ‘forced’ on her. Rather, it is rationally explicable to Sally that she has that belief—she believes it on the basis of the grey clouds. We can note both a negative and positive claim here: she isn’t unaware of why she has the belief, and further, she seems to be aware of why she has it.

At this point, it is helpful to note a related feature that we might also attribute to believing for a reason, or at the very least, motivating reasons insofar as they play a role in responsible belief. I will call this ‘rationalisation’. Indeed, we might use this sort of observation to motivate a doxastic constraint on believing for a reason (in the context of responsible belief see Leite (2008), and reasoning, see Thomson (1965)). To borrow Quinn’s (1993) language in discussing reasons for action, believing for a reason means that holding the belief is ‘sensible’ and ‘makes sense’ to the subject. By ‘rationalise’, I mean to say that motivating reasons make a belief rational and sensible to her as far as she is concerned: the belief will be rational in a subjective sense, although not necessarily in an objective one. And indeed we can note that it makes sense to Sally to believe that it will rain in light of the grey clouds. This is unlike Norman’s belief, for example, as emphasised by BonJour’s own gloss: despite being formed by a reliable mechanism, ‘from his subjective perspective, it is an accident that the belief is true’ (1980: 63).

But there is more to believing for a reason than the role it plays in rationalising the belief. We can clarify the way in which Norman and Carl are unusual by considering two related subjects:

CLARA: Clara reads the newspaper in the morning. The headline states that the president is in town. Clara believes that this article provides good reason to believe that the president is in town, but the president’s immanent presence doesn’t ‘sink in.’ She doesn’t judge that the president is in town, and neither does she act accordingly (e.g., she stays at home despite loving the president). It looks, then, that she does not actually believe that the president is in town. Say, though, that in the afternoon, she develops a reliable sixth sense like Norman’s which does lead her to believe that the president is in town. Clara doesn’t have any evidence to suggest that she has this sixth sense.

Suppose you ask Clara why she has the belief, or she asks herself (e.g., perhaps her friend accuses her of just believing things at random!). Like Norman, she is dumbfounded and flummoxed in response, and just draws a blank. She can’t even demonstratively gesture at a gut feeling (‘I believe the president is in town because I have this feeling that he is here’). Clara has nothing to say, even to herself, and certainly can’t evaluate any factors that might be motivating her. As far as Clara is concerned, the belief has just come out of nowhere.

OSCAR: Oscar has OCD and, like Carl, always believes that tables are dirty. He is well aware of his OCD and visits a therapist. Say that the table actually is clearly dirty—there are lots of crumbs. Oscar therefore happens to take there to be good reason for believing that the table is dirty in this case. Perhaps also his belief about the crumbs just so happens to play a causal role in sustaining his belief for once—unbeknownst to him, some medication has taken effect earlier than expected and his OCD has slightly abated. Still, as far as he is concerned this match is accidental (e.g., the phenomenology is the same as with an OCD belief, and there’s no defeating evidence suggesting that on this rare occasion his belief is based on the reason). Suppose we ask him why he has the belief. He wouldn’t be dumbfounded, but as far as he is concerned, the belief resulted from his OCD: ‘I believe it because of the OCD—the belief has been forced on me’.

Clara and Oscar differ from Norman and Carl, but nevertheless bear certain similarities. Clara and Oscar satisfy the doxastic requirement, and Oscar also satisfies the causal requirement. But like Norman, Clara is completely unaware of where her belief came from—her belief that the president is in town happened upon her out of nowhere as the result of a sixth sense she had no idea about. And as far as Oscar is concerned, as with Carl, his belief that the table is dirty is out of his control and forced on him by OCD. As a result, Clara and Oscar intuitively still seem not to believe for a reason.

We might also put this sort of point in a different way. One claim is that in believing that q for a reason, S is aware of what, in her mind, propositionally justifies believing that q, i.e., what would justify her belief that q if she were to form it. This is partly what distinguishes Sally’s case from that of Norman. But it seems that S is also aware of what, in her mind, is her doxastic justification, i.e., what justifies her belief which she does in fact hold. (This is not to say that the belief actually will be justified.) This distinguishes Sally’s case from Norman’s and Carl’s, and also from Clara and Oscar. Clara and Oscar are clearly aware of the propositional justification for their respective beliefs: the newspaper article and the crumbs. But they have no awareness at all about the doxastic justification, i.e., what actually justifies the belief itself. But when Sally believes that it will rain, she has some sort of awareness of both. Sally is aware that the grey clouds justify believing that it will rain, and that the grey clouds justify her belief itself, i.e., she is aware that the grey clouds forms part of her belief’s evidential base.

At this point, I should note that P1 attempts to identify a necessary feature of belief, and indeed, I find it hard to think of any factors as constituting a subject’s reason if they find themselves as flumouxed as Norman. The following discussion will also motivate taking intelligibility to be a necessary feature of believing for a reason, given the role that it plays in responsible belief.

Why would we expect believing for a reason to intelligise belief in this way? First, note here that my task is a little tougher than it could be. I’m not just interested in beliefs that have been recently formed through explicit inference, but rather, all beliefs held for reasons. Jenkins (2018) rightly uses the notion of intelligibility to motivate a self-knowledge constraint on inference in contrast to causal-doxastic alternatives, but it also bears fruit in the context of believing for a reason simpliciter. That inference requires the capacity for self-knowledge and/or self-ascription in some form seems less contentious, especially since Jenkins plausibly already takes self-consciousness to be part of the explanandum in this case. Inference, or at least a significant sub-type, is already an event at the level of consciousness from which point it seems easier to appeal to self-consciousness. And insofar as inference is a mental action, we can already appeal more directly to the way in which way rational actions are arguably intelligible.¹⁵ But I am interested in the broader phenomenon of forming and holding beliefs for reasons. It has required more work to show the relevant similarities in the case of holding beliefs for reasons.

In order to understand the importance of intelligisation, it is useful to bear in mind the qualification to my central claim. Here I want to argue that necessarily one is in a position to self-ascribe one’s reasons provided the relevant rational and conceptual capacities possessed by most adult humans. Regardless of whether other subjects can believe for a reason, believing for a reason seems apt to play a particular role for subjects with those capacities. (I also happen to think that it would have to play this role for other subjects but that it could be satisfied in ways that fall short of belief. I want to leave that complication aside here.) Namely, basing plays a key role in forming and holding beliefs responsibly, obtaining inferential justification, and critical reasoning. For example, Shoemaker (1994) proposes something similar regarding belief and the more complicated role it plays depending on one’s conceptual capacities, and also the rational expectation for self-knowledge that arises as a result.

There are several ways in which intelligibility might be necessary for responsible belief, and different readers might prefer different strategies.

One option is to suggest that insofar as believing for a reason is to believe in some way responsibly, the subject herself must see the belief as having positive epistemic status. This strategy mirrors one used by some when arguing that at the very least responsible belief requires satisfying the doxastic condition (e.g. Leite, 2008). It can’t be a ‘lucky coincidence’ from her perspective if the belief is true (Leite, 2008: 424). We might say therefore that believing for a reason at least in this context must rationalise the subject’s belief, and that accordingly, she must believe that the reason supports this belief. But this forms part of a broader phenomenon encompassing another positive epistemic status. To believe responsibly, it shouldn’t just matter to the subject whether or not the belief is true, but also whether it is well-formed and/or knowledgeable. That is, doxastic justification bears importance and not simply propositional justification. So insofar as motivating reasons play this role in responsible belief formation it seems natural to say that reasons intelligise the belief as well as rationalise it—in making clear from the subject’s perspective their own basis and not just possible grounds. This strategy may well appeal to those who think that believing for a reason, or at least believing responsibly, requires believing that it is a good reason.

Second, perhaps we might deny that being a responsible believer, and being responsible regarding a belief, requires seeing the belief as possessing positive status. One might think that Sally can responsibly believe that it will rain on the basis of her horoscope despite believing that this is a bad reason. This possibility closely mirrors the debate concerning whether being responsible and non-alienated with respect to one’s belief that p requires endorsing (or being disposed to endorse) p. Here some have argued that it does not; what is important is that we can check and assess our beliefs and will revise them if needed. Here it’s helpful to turn to Burge’s conception of critical reasoning (see also Leite, 2018 and for useful discussion, Gertler, 2011). Burge writes that ‘we are epistemically responsible only because we are capable of reviewing our reasons and reasoning. And we are paradigmatically responsible for our reasons when we check and review them in the course of critical reasoning’ (Burge, 1996: 111). In critical reasoning, ‘one not only reasons. One recognizes reasons as reasons. One evaluates, checks, weighs, criticizes, supplements one's reasons and reasoning’ (1996: 98). This ability to check is what matters, not whether at that moment one is disposed to endorse the belief. Relatedly, one might also deny that in order for motivating reasons to play the required role in responsible belief we need to see the reasons as good reasons. Perhaps what matters is that we can assess these reasons and revise the beliefs if needed. In this way, one might pull intelligibility apart from rationalisation.¹⁶

But we already see something significant for our purposes in Burge’s discussion (although the appeal to knowledge outstrips what I am aiming for at this stage). For us to check our reasons, ‘one must know what one’s reasons, thoughts, and reasoning are’ (1996: 111), or at least, do so most of the time. So we might say that, insofar as a subject believes for a reason, this reason must make the belief intelligible to her so that she can evaluate her grounds and revise her belief if she concludes that they are in fact bad. We can draw the relevant difference between Sally’s horoscope-based belief and Norman and Clara’s beliefs in the following way. If the question arises, Sally can recognise why she holds the belief and evaluate its basis (and all going well, she will revise her belief in accordance with her assessment). This capacity is necessary if believing for a reason is to play this role in critical reasoning. It isn’t just that Sally is in a position to provide putative reasons for the belief (which in this case she takes to be bad). In critical reasoning, the subject cares about the epistemic status of her beliefs generally. Note that, as before, what matters to the subject is not just whether beliefs are propositionally justified, but also whether they are well-grounded and doxastic justified. Clara still cannot say why she holds the belief, even though she can provide possible justifications. And neither is it simply that Sally recognises that she believes that p; she is able to grasp the relation in order to fully assess it as her reason. To turn now to Carl, Carl can offer an explanation of his belief, albeit incorrect—that he holds it as a result of OCD. And yet, purely causal explanations aren’t what is at stake in the context of critical reasoning, but instead one’s motivating reasons for the belief (or at least, such explanations are relevant insomuch as they suggest that the subject doesn’t actually hold the belief on the basis of reasons). So we might say that believing for a reason, at least in the sense I have in mind, must intelligise the belief in order to play the required roles in critical reasoning.

Premise Two

Ensuring that basing has this feature seems to require necessarily that subjects can self-ascribe their reasons given the correct resources. After all, a belief’s intelligibility appears to inherently involve self-awareness in some form—it is to be aware of what (at least partially) rationally explains one’s belief. But why specifically appeal to self-ascription which, as I have been using the term, is to have a justified true belief? I.e., why not simply a true belief, or another sort of state entirely, like an experience?¹⁷

Perhaps there will be situations where something less than belief would suffice, and indeed, we don’t form higher-order beliefs like this very often. But I here want to emphasise that we still need to be in a position to form a belief. The point is that when we do try to do so—e.g., if we want to assess the epistemic status of our belief or our reasoning ability—we are able to, and are not left dumbfounded.¹⁸ Infants and non-human animals, on the other hand, wouldn’t even try. Further, the belief would have to be true for the basing relation to hold because otherwise it would be false that the subject has that reason.

And the belief also needs to have a degree of epistemic umph to satisfy the requirement that believing for a reason intelligises the belief, and to ensure that it does not come as a surprise that one has it. This wouldn’t be the case if, were the subject to form it, the further belief would also seem like a surprise and poorly formed.¹⁹ E.g., suppose Sally believes that it will rain, and comes to believe that she believes that it will rain because of the grey clouds, but this higher-order belief is just a guess. As such, she’s not much better off from her perspective than if she was dumbfounded, even if the guess happens to be true.

As a result, then, I think that necessarily, given relevant capacities, if one believes that q for the reason p, they are in a position to self-ascribe this. One consequence worth noting is that this thesis may well lead us to deny that adult humans with the relevant capacities count as believing for a reason in a particular instance if they fail to meet the self-ascription constraint, even if the relation bears the same characteristics of say, a cat’s believing for a reason. But this seems plausible if we think that motivating reasons are apt to play this role in responsible belief. We raise the bar and expect more from those depending on their capacities. When Milton the cat’s beliefs happen upon him, he isn’t surprised—because that’s just how things go as far as he is concerned. His beliefs aren’t intelligised to him when he believes for a reason, but neither do his beliefs fail to be intelligised when he does not. Yet Norman’s and Clara’s beliefs fail to be intelligised in a problematic way and which is inimical to responsible belief. Most adult humans, in having certain capacities, are now responsible for their beliefs and it is now embolden on them to discharge this responsibility [see Wright (2001: 70–71)].

Believing for a Reason as (at Least) Nearly Self-Intimating

This section will spell out the self-ascription thesis in more detail. To recall:

If a subject believes that q for the reason that p, necessarily, provided relevant conceptual and rational capacities, she is in a position to form a justified true belief that she believes that q for the reason that p.²⁰

I.e., if S is not in a position to form the justified true belief that she believes that q for the reason that p, then it cannot be the case that she believes that q for the reason that p. Specifically, the subject must be in a position to form the justified true belief that she herself believes that q for the reason that p: compare the difference, as Perry (1979) notes, between complaining that the person spilling sugar is very annoying and then finally realising that I am spilling sugar.

I will now elaborate on the account and revisit the two initial worries we started with. The following Sect. 4 will consider three further objections.

Details of the Proposal

Roughly, I take ‘being in a position’ to self-ascribe the reason as being placed such that all the subject needs to form a justified true belief is to employ the relevant method of belief-formation in a responsible way. That is, she can use a reliable method, and has access to the relevant grounds even if she does not currently have them. And, following Williamson’s discussion of being in a position to know, we can add require that ‘[n]o obstacle must block one's path’ (2002: 96) to self-ascribing the reason; at least, there cannot be a systematic and permanent obstacle. As such, I wouldn’t say that a subject who has repressed her motivations in the sort of way Freud discusses is in a position to self-ascribe them, or that she believes on the basis of reasons. In these cases, the putative reason would seem to have completely ceased to play the required role in intelligising the belief.²¹ I must leave the details to further discussions on the topic, but I trust that the precise formulation will not pose problems for my account.

We might construe the method by which subjects can self-ascribe the reason in various ways, but here I will follow my (2021, 2019) account of the ‘Reasons Transparency Method’ (RTM). In those papers, I formulated an account of self-knowledge of reasons which extends the transparency method as oft discussed regarding self-knowledge of belief. Philosophers frequently observe that subjects can learn whether they believe that q simply by considering whether q is true (see Boyle, 2011a, 2011b, 2019; Byrne, 2005, 2011; Evans, 1982; Fernández, 2013; Moran, 2001). E.g., if S wants to know whether she believes that it will rain, she will consider whether it will indeed rain. And, the resulting warrant, so it is thought, is non-inferential in the sense that S’ justification is not transmitted from her justification for believing supporting propositions. What about the case of reasons? Subjects learn why they (currently) have a belief by considering what the good reasons are for having that belief. For example, subjects treat the question of ‘why do I believe that it will rain?’ as transparent to the question ‘what are the normative reasons for believing that it will rain?’.²² Subjects then conclude that the grey clouds is a normative reason for having the attitude at hand and can reply that the grey clouds is their motivating reason for holding it. And I take it that the resulting self-ascription is, as with belief, also warranted non-inferentially.²³ It’s important that this is only one possible way of accounting for it, since this fits most naturally if we also accept the doxastic constraint. Indeed, another way to argue for a self-ascription thesis is to accept the doxastic constraint and also this particular account of self-knowledge.²⁴ But my claim regarding self-ascription itself can be motivated independently regardless of our ultimate account of the underlying metaphysics. If one wishes, one could adopt a different account of how the self-ascriptions are warranted, e.g. appealing to entitlements along the lines of Burge (1996).

It's also here worth noting that one might think that some form of self-awareness is a necessary constraint on believing for a reason simpliciter. Neta has recently argued that basing is constituted by ‘a de se, object-involving representation of [the subject’s own] own exercise of a disposition to [form the relevant belief] when [the reason is present], under the aspect ex post justifying, and that disposition-exercise is individuated by that very representation’ (2019a, 213).²⁵ I haven’t said anything about whether self-awareness is part of basing itself, given my qualifications about relevant conceptual capacities. That said, my thesis would help lend support to such an account.

Revisiting the Initial Concerns

Recall that I mentioned two initial doubts about this sort of proposal, and in general, about taking our epistemic access to our reasons for belief to be at all special (see also Keeling, 2021).

First, we might worry about the potential for confabulation and making mistakes. Yet, in spelling out the self-ascription constraint, I should note what I am not saying. I do not think that our self-ascriptions are infallible. While I maintain that we cannot have a motivating reason that we are not in a position to self-ascribe, I am happy to accept that we can mistakenly ascribe reasons that we do not have.

For what it’s worth, though, I don’t think that subjects would mistakenly ascribe reasons as frequently as we might worry. First, it's worth emphasising that motivating reasons can be very general. For example, a subject might base their belief that the vase is valuable simply on the belief that the vase looks valuable, and not any particular features that make it look valuable. Even if the subject doesn’t base her belief on a very specific consideration, she will still have a motivating reason in this instance. So, a subject gets something right if she ascribes this general reason. And second, subjects will often have a motivating reason even if other factors also explain why they have that reason. For example, say I believe that Fellini dislikes me for the reason that he left my lap (in this scenario, ‘Fellini’ is a small grey cat). Perhaps I only hold the belief for that reason because I’m generally suspicious and prone to think that everyone dislikes me. But still, Fellini’s departure is my motivating reason for the belief. If he hadn’t left my lap, I wouldn’t have formed that belief (at least not then). So, when I believe that I believe that Fellini dislikes me for the reason that he left my lap, I still correctly ascribe a motivating reason.

Second, we might worry about how a causal account of epistemic basing would mesh with such a proposal. One option would be to abandon causalism, perhaps in favour of a purely doxastic account (e.g. Setiya, 2013).²⁶ Yet I do want to say that basing is a causal relation, but also that certain causal relations can plausibly be self-intimating. I should again emphasise that I do not take the relevant self-ascriptions to be infallible: I allow that the causal chain might break down/ be deviant without us being in a position to form true beliefs about this. And I am only claiming that one, special, type of cause self-intimates. I can therefore even say that one belief can still cause another in the absence of being in a position to self-ascribe it—the relation just wouldn’t be of this distinct kind. And it’s plausible that reasons would constitute a special kind of cause and thus bear such constraints. After all, reasons play distinct roles, such as rationalising, intelligising, and justifying our beliefs.

Objections

I will now consider three objections. They can be traced back to the sorts of worries facing doxastic accounts of basing and internalist accounts of knowledge and justification. In employing related strategies to these views, I face similar concerns. But notably, my account can address these more easily than classic internalist positions, owing to my narrower scope. I can allow that subjects can know that p without being in a position to say why they believe/know that p; I just deny that such subjects hold that belief for a reason.

Over-Intellectualising Basing

We might think that the self-ascription thesis over-intellectualises believing for a reason and is over-demanding. After all, it is implausible that for every belief we form on the basis of reasons, we self-ascribe what those reasons are.

But I am only claiming that subjects will necessarily be in a position to self-ascribe their motivating reasons in virtue of having them, and even that requirement depends on one possessing the requisite concepts. So, I allow for cases in which the subject forms a belief in the quick un-self-reflective way that is commonplace. Such a subject wouldn’t have thought about why she holds that belief until she is asked. But still, she is able to form the justified true belief that she holds the lower-order belief for a given reason.

The Risk of Regress

We might worry that the regress objection to doxastic accounts of basing also applies to the self-ascription constraint.²⁷ I have argued that for certain subjects, necessarily, if S believes that q for the reason that p, she is in a position to form the justified true belief that she believes that q for the reason that p. So, S must have propositional justification for believing that she believes that q for the reason that p. And, we might worry, S would have this propositional justification in virtue of possessing a further normative reason on which she would base the self-ascriptive belief. It therefore looks like my proposal requires that S also have justification for believing that r is her reason for believing that [p is her reason for believing that q]. And so on, and so forth.

We might stem the regress in various ways.²⁸ One option is to deny that the self-ascriptive belief—the subject’s belief that they believe that q for the reason that p—is based on reasons. That self-knowledge is not based on reasons is a popular thought. For example, we might follow Moran (2001), and think that the transparency method gives rise to warranted self-ascriptions as a result of a rational entitlement to use this method. Alternatively, even if we hold that self-knowledge is based on reasons (e.g., Keeling, 2021), there will most likely be a regress stopper further down the chain. Again I should emphasise that all we need is to accept that there can be some justified beliefs that aren’t based on reasons of any sort.²⁹ At any rate, we now find ourselves with the broader questions facing epistemologists concerning the structure of knowledge; this paper just aims to introduce and motivate the self-ascription thesis, so I hope to leave the broader issues it might raise to future work.

Forgotten Reasons

One might object that forgotten reasons present a case in which subjects believe that q for the reason that p without being in a position to self-ascribe this:

FORGOTTEN REASONS: Lizzy believes that epistemic externalism is false. She originally formed the belief on the basis of an a priori argument. She has, though, now forgotten that argument. As a result, when I ask Lizzy for the argument why she believes that epistemic externalism is false, she replies that she doesn’t know.

But I deny that Lizzy has a motivating reason she isn’t in a position to self-ascribe, since, I contest, the a priori argument is no longer her reason for believing that externalism is false. It once was, but not anymore. If she bases her belief on anything, it may well be a consideration such as that she remembers that the belief is true or that there is good evidence that it is true (even though she cannot remember precisely what that evidence is).³⁰

My opponent might formulate a variation, though, in light of this response.

SLIPPED REASONS: Suki reads an article listing four facts concerning a politician, Smith. Suki then forms the belief that Smith is awful on the basis of all four of these facts. Later, I learn of Suki’s belief and ask her why she believes that Smith is awful. One of the considerations has slipped Suki’s mind. As such, at this moment, Suki only self-ascribes three of her four motivating reasons. She replies with: ‘Smith is racist, has bad foreign policy, is sexist, and ... what was that last one again?!’ Later in the day it dawns on Suki: ‘ah yes, Smith is generally incompetent!’. While Suki hadn’t forgotten the reason, it had slipped her mind when I asked her why he believes that Smith is awful.

SLIPPED REASONS require a different response to FORGOTTEN REASONS. After all, it’s implausible to deny that Smith's general incompetence is one of the reasons for which Suki believes that Smith is awful. Basing relations would hardly disappear and reappear in such a spontaneous way. In this case, I hold that Suki is still in a position to form the justified true belief that she believes that Smith is awful for the reason that Smith is incompetent. It is just that she has a momentary difficulty in doing so—there is no systematic blockage of the sort that we might see in repression cases. Suki can in general still use a method like RTM to learn of her reason, and she has accessible grounds. And she will come to self-ascribe the reason given a little more time or a slight reminder. This is like how one can be in a position to form the justified belief that there is a pomegranate on the table by looking, even if the term has slipped one’s mind: ‘there’s fruit on the table … you know, the one with the big seeds, what’s it called again? … this is so annoying—I know this!! …… Ah, yes, it’s a pomegranate!!’.

Upshots and Consequences

I can now start drawing the paper to a close by highlighting two upshots/ broader points concerning the discussion so far.

• i. I have argued that believing for a reason very nearly self-intimates and that epistemic basing requires being in a position to form a justified true belief about one’s motivating reasons; what about self-intimation proper? I happen to also endorse this stronger claim, i.e., that the justified true beliefs must also be reliably formed, safe, etc. Whether one can use the intelligibility argument to ground this further claim will depend on one’s underlying picture of why basing requires intelligibility. Of the two options canvassed earlier, the first would not require the subject to know her reasons for them to intelligise her belief, but perhaps one could construct a Burgean argument for this further claim. And at the very least, the discussion in this paper lends it further support.

• ii. It’s worth highlighting an intriguing consequence of the self-ascription thesis. I’ve argued that believing for a reason is nearly self-intimating in a strong sense. I.e., provided certain general capacities, subjects will be in a position to self-ascribe the reasons for which they believe. If such subjects are not in such a position to self-ascribe a particular influence, then it cannot be the reason for which they believe. And yet we may well think that belief itself lacks this characteristic, and that we can fail to be in a position to self-ascribe our beliefs even granted certain general capacities. Even if we think that belief self-intimates in some way, a plausible position holds that S cannot rationally believe that q without being in a position to know that she believes that q—the belief would be irrational in this case (e.g. Burge, 1996; Moran, 2001). But on this account, S could still be credited with a belief in the relevant cases. This is a way of allowing for the possibility of repressed or alienated beliefs that subjects might only uncover through years of therapy. Perhaps we might disagree, and perhaps maintain that subjects with the relevant capacities are in a position to learn of their beliefs simpliciter.³¹ But interestingly, it is at least possible to hold that, while a self-ascription thesis like the one I am proposing is limited to rational belief, no such restriction applies to the reasons for which one believe. I have claimed that in subjects with the relevant capacities, necessarily, if they have a motivating reason they will be in a position to self-ascribe it.

There is a lot more to say here, and will relate to precise details such as our preferred account of basing. But briefly, this upshot should be plausible because motivating reasons and attitudes are different kinds of things. After all, they occupy distinct roles. Beliefs guide action, and a state can guide action in accordance with q even if the subject is unable to learn of this belief.³² It is possible to believe irresponsibly and irrationally. And yet, I have suggested that epistemic basing is a specific kind of causal relation which intelligises (and indeed potentially also rationalises) one’s beliefs. And believing for a reason is to believe responsibly and rationally at least to some extent.

As such, it is not just that we have a distinctive epistemic relation to the reasons for which we believe; further, it is potentially even tighter than the relation to our beliefs themselves.

Conclusion

I have argued that, necessarily, if one believes for a reason, provided the relevant concepts, one will be in a position to form a justified true belief that one believes for that reason. In other words, I have argued that believing for a reason is at least very nearly self-intimating. This is to identify one way in which our epistemic relation to our own reasons is special, and is therefore a far cry from the popular picture according to which it simply resembles third-personal access.³³