Reasons, Rationalization, and Rationality

Abstract

In this paper, I provide an answer to the question “what is it for a reason to be the reason for which a belief is held?” After arguing against the causal account of the reason-for-which connection, I present what I call the rationalization account, according to which a reason R a subject S has for a belief P is the reason for which S holds P just in case R is the premise in S’s rationalization for P, where the argument from R to P becomes S’s rationalization in virtue of her endorsing it. In order to bring explicitly into view the version of the rationalization account I aim to argue for, I draw two distinctions, one between occurrent and dispositional endorsement and the other between personal and public endorsement. I show that the version of the rationalization account thus clarified receives intuitive support from various cases and survives some formidable objections that might be tempting to level against it.

1 Introduction

Epistemic rationality requires having good epistemic reasons¹: a subject’s belief can only be rational if the subject has good reasons for that belief. However, having good reasons is not sufficient for rationality.² What must be added to “good reasons” in order to attain rationality? A generic answer is this: a subject’s belief is rational just in case the good reasons the subject has for that belief are adequately “operative”: the belief must be “built on” those good reasons. A subject’s belief is rational just in case the reason for which the subject holds that belief is the same as the good reason the subject has for that belief. The central question that will concern us in this paper is this: what is it for a reason to be the reason for which a belief is held?

The rationality of a given belief requires two things (and only two things), viz. that the subject have good reasons for the belief and that there be a (reason-for-which) connection between those good reasons and the belief. I take it as definitive of the relevant notion of “the reason for which” that the following conditionals express conceptual (or “non-negotiable”) truths:

• (A) If a given belief of a certain subject is rational, then the good reason the subject has for it is the same as the reason for which the subject holds it.

• (B) If the good reason the subject has for a given belief of a certain subject is the same as the reason for which the subject holds it, then the belief is rational.

Combined with our intuitive judgments in various scenarios about the rationality of beliefs, (A) and (B) enable us to say something substantive about the reason-for-which connection. Suppose that we have a toy theory, T, about the reason-for-which connection, according to which the reason for which a subject holds a given belief is the reason she has been having for the longest amount of time. Is T true? No. Why? Because the following cases are intuitively conceivable:

Case 1: Jack believes, and has a good reason to believe, that his mother does not love him as much as she loves his sister (his reason being that his mother treats his sister with much more care than she does treat him with). Furthermore, the good reason he has for this belief is the one he has been having for the longest amount of time (he acquired the good reason when he was a little kid and has been having it since then). However, Jack has the belief in question because he believes that the tea leaves say so. Intuitively, Jack’s belief is not rational. (However, assuming that because captures the reason-for-which connection, then given (B), T entails that it is rational.)

Case 2: Jack believes that his mother does not love him as much as she loves his sister because he believes that his mother said so a few days ago (and he knows that his mother is mercilessly sincere when it comes to family issues). Intuitively, Jack’s belief is rational. Nevertheless, he acquired the good reason he has for the belief quite recently and therefore it is not a reason he has been having for the longest amount of time. (However, assuming that because captures the reason-for-which connection, then, given (A), T entails that the good reason Jack has for the belief is the reason he has been having for the longest amount of time.)

In Case 1, Jack’s belief is not rational, though T entails that it is rational. In Case 2, the good reason Jack has for his belief is not a reason he has been having for the longest amount of time, though T entails that it is. T gives a wrong verdict in either case, and hence it is false.

A crucial move in the description of Cases 1 and 2 is the use of “because”, which affects our intuitive judgments about the rationality of Jack’s belief. The “because”, as it is meant in these cases, denotes the reason-for-which connection this paper aims to account for. (And, the fact that we don’t feel any uneasiness when we say in Case 1, for instance, that “Jack has the belief in question because he believes that the tea leaves say so” signals a fact that we knew all along, that T is a non-starter.) What is the sense of the “because” in Cases 1 and 2? In other words, how are we to account for the “because” that denotes the reason-for-which connection?

There are basically two different answers to these questions. First, the “because” that denotes the reason-for-which connection is the “because” of causation. It is the “because” that we typically mean when we say such things as “He believes that God exists because he was brought up by his parents as a Catholic” and “The tap is leaking because its washer is broken”. When one says “P because R”, intending to mean the “because” of causation, one typically assumes that the audience takes it for granted that P and one puts forward a purported explanation of why P is the case. The intention here is not to convince the audience that P but explain why P. Second, the “because” that denotes the reason-for-which connection is the “because” of argumentation, the “because” that connects the premises of an argument to its conclusion.³ It is the “because” that we typically mean when we put forward such arguments as “Jack was here yesterday because I saw him here yesterday” or “The tap is leaking because there is a sound of dripping water coming from the bathroom”. It is clear that neither of these sentences is to be conceived as intending to make a causal claim (that is, it is clear, for instance, that the fact that I saw Jack here yesterday is not causally responsible for the fact that he was here yesterday). When one says “P because R”, intending to mean the “because” of argumentation, one typically assumes that the audience needs (or demands) a reason to think that P and, accordingly, one puts forward a reason to think that P. The intention here is to convince the audience that P.

A point of clarification about the view that the reason-for-which connection is the same as the “because” of argumentation is in order. The question that concerns us is what it is that makes a reason that one has the reason for which one believes something. In order for the “because” of argumentation to be a suitable candidate for capturing the reason-for-which connection, the argument expressed by a sentence of the form “P because R” must be related to the subject in question in the right way: otherwise, R cannot be the reason for which that subject believes P, whatever other features the argument in the abstract might have. So, the view that the reason-for-which connection is the same as the “because” of argumentation is to be conceived as the view that the reason-for-which connection is the same as the “because” of an argument the subject is somehow related to (more on this later).⁴

The reason-for-which connection that I intend to give an account of is a connection that is non-accidentally tied to the rationality of beliefs. It is captured in its essentials by conditionals (A) and (B) above. The reason-for-which connection is typically intended to be expressed by using “because”, for instance while saying things like “Jack believes that his mother does not love him as much as she loves his sister because he believes that the tea leaves say so”, the implication being that the reason for which Jack has that belief regarding his mother’s belief is that (he believes that) the tea leaves say so. However, it would be hasty to conclude, from the fact that the reason-for-which connection is typically denoted by using “because”, that it is a causal connection since there is another “because”, the “because” of argumentation, representing an alternative thesis that cannot be rejected out of hand.

Is the reason-for-which connection denoted by the “because” of causation or the “because” of argumentation? When, for instance, we make such claims as “Jack believes that his mother does not love him as much as she loves his sister because he believes that the tea leaves say so”, which grounds our judgment that Jack’s belief regarding his mother’s love is not rational, do we mean the “because” of causation or the “because” of argumentation? Shall we read this “because” claim, by which we judge the rationality of Jack’s belief regarding his mother’s love, as the claim that Jack’s belief that the tea leaves say so causes (or causally sustains) his belief regarding his mother’s love or as the claim that (the content of) Jack’s belief that the tea leaves say so is the premise in an argument endorsed by Jack for (the content of) his belief regarding his mother’s love?

2 The Causal Account: the “Because” of Causation

According to the causal account, a reason R the subject S has for a belief P is the reason for which S holds P just in case R causes (or causally sustains) P.^5,6 Given (B), the causal account entails that if the good reason a subject has for a belief causes the belief, then the belief is rational. And, given (A), the causal account entails that if a given belief of a subject is rational, then the good reason the subject has for that belief causes the belief.

The causal account gives a straightforward answer to the question of what it means for a reason to be adequately “operative” (“it is to be causally operative in belief formation”) and it also sits well with a sense of “the reason for which” according to which the phrase can be plausibly interpreted as indicating causation (e.g., “There has been a delay, the reason for which is that there was a fire alarm”).

The question is whether the causal sense of “the reason for which” is the sense that is characterized by (A) and (B). To answer this question, let us take a look at a well-known problem the causal account faces, viz. the problem of deviant causal chains. I may believe that there is hot coffee in the cup, and I may also believe that I have just poured hot coffee into the cup. The latter belief is a good reason for, and let’s suppose, causes the former one. Still, my belief that there is hot coffee in the cup may fail to be rational. Consider for instance the following scenario. The belief that I have just poured hot coffee into the cup makes me so nervous that I tremble and pour coffee on my favorite t-shirt. The emotional and physical pain I thereby feel puts me into mental shock, which in turn causes me to have such absurd beliefs as “I am the President of Canada” along with “There is hot coffee in the cup”. In this case, it intuitively appears that my belief that there is hot coffee in the cup is not rational, despite the fact that the good reason I have for it causes it. However, the causal account entails that it is rational and hence it is false.

A standard move by the proponents of the causal account against this sort of counterexample is to claim that the causal chain from the (initial) “pouring” belief to the (final, target) “in-the-cup” belief is deviant (i.e., deviates from what it is “supposed” to be), and the challenge – the problem – is then to account for the difference between deviant and non-deviant causal chains in a principled way without trivializing the causal account. What I call the (causal) non-deviance account holds that the reason-for-which connection is a non-deviant causal connection, and the challenge is to account for the alleged “non-deviance” in a non-trivial way.

The trivialization challenge for the non-deviance account is an old chestnut, explicitly recognized by various philosophers⁷; and I think it cannot be met, so the non-deviance account is either false or trivial. Suppose that a given particular chain is regarded as deviant by the non-deviance account, and we raise the sensible question: in virtue of what is that causal chain deviant? What is it that renders that causal chain deviant rather than non-deviant? I hold that no answer to this question that does not ultimately appeal to the rationality of the target belief is plausible. However, an answer that appeals to the rationality of the target belief clearly trivializes the non-deviance account by rendering it circular because the non-deviance account is supposed to explain rationality in terms of the non-deviance of causal chains and not vice versa.

Let us now see how the non-deviance account might give an account of “non-deviance” by taking a look at the counterexample presented above against the causal account. The causal route specified there starts from the “pouring” belief, which then causes a non-cognitive phenomenal state (that is, pain), and ends with the final, “in-the-cup” belief. This suggests what I call the cognitive (causal) non-deviance account: a causal chain from a reason R to a belief P is non-deviant just in case all the links in that chain are nothing but cognitive (belief-like) states.⁸ However, this does not work. Suppose that we have a causal chain starting from a good reason R for a belief that P, which involves nothing but belief states, and ending at P. Now, we can readily imagine the following scenario: R causes a belief that Q, which in turn causes P, where P is not rational. For instance, due to some radical momentary brain dysfunction, my belief that I have just poured hot coffee into the cup may cause a belief that aliens are trying to abduct me, which in turn causes a belief that there is hot coffee in the cup. The initial belief is a good reason for, and causes, the final one, and all the links in the chain are cognitive; however, the final belief is not rational and the causal chain must then be deviant. However, the cognitive non-deviance account entails that it is non-deviant and hence is false.

A moment’s reflection on the counterexample against the cognitive non-deviance account might suggest an obvious revision. The causal route from the reason (the “pouring” belief) to the final belief (the “in-the-cup” belief) is mediated by a different belief (the “abduction” belief), one that is epistemically irrelevant and hence does not stand in any support relation to the other two beliefs. The revision suggested is the elimination of mediation. The direct (cognitive causal) non-deviance account holds that a causal chain from a reason R to a belief P is non-deviant just in case R directly causes P. However, this does not work either. Suppose that, due to some radical momentary brain dysfunction, my belief that I have just poured hot coffee into the cup directly causes such oddities as a belief that aliens are trying to abduct me and a belief that the best Kung Fu fighter in the entire human history is Napoleon as well as a belief that there is hot coffee in the cup. In this case, the “pouring” belief is a good reason for, and directly causes, among other things, the “in-the-cup” belief; but again, the latter belief is not rational and hence the causal chain must then be deviant. However, the direct non-deviance account entails that it is non-deviant and hence is false.

The counterexample against the direct non-deviance account might pave the way for yet another version of the non-deviance account. The “pouring” belief directly causes not only the “in-the-cup” belief but also the “abduction” and “Napoleon” beliefs. A quick fix is, it might be tempting to claim, the elimination of the (direct) causation of epistemically irrelevant beliefs. The unilateral (direct cognitive causal) non-deviance account holds that a causal chain from R to P is non-deviant just in case R directly causes P and R does not (directly) cause any other belief that is epistemically irrelevant. However, this again does not work. Suppose that I am hit by a hammer in the head, as a result of which the “pouring” belief directly causes the “in-the-cup” belief: what makes the “pouring” belief directly cause the “in-the-cup” belief is my being hit by a hammer. If it were not for the hit, there would be no causal interaction between the two beliefs; and if I were hit by a wrench rather than a hammer, the “pouring” belief would cause an epistemically irrelevant belief (say, my “Kant” belief that Kant is the greatest modern philosopher). What is more, if the hit were to land slightly to the left, an epistemically irrelevant belief (say, my “Kant” belief) would directly cause the “in-the-cup” belief. In this case, the “pouring” belief is a good reason for, and directly causes, the “in-the-cup” belief, and it does not cause any other belief that is epistemically irrelevant; but again, intuitively, the latter belief is not rational and hence the causal chain must then be deviant. However, the unilateral non-deviance account entails that it is non-deviant and hence is false.

None of the versions above of the non-deviance account works, and there is indeed a principled reason to think that there are no versions of the non-deviance account that can be made to work. And, the reason is that what qualifies a causal chain as deviant is its deviance from what it is supposed to be, which can be made sense of only by an appeal to our judgments regarding the rationality of beliefs and a fortiori our judgments regarding whether the reason-for-which connection obtains. There is no way to tell whether a given causal chain is deviant only on the basis of inspecting the chain and without being in a position to tell whether the belief formed as a result is rational or not. The sole arbiter regarding the deviance in question is our judgments about rationality, while the non-deviance account must distinguish deviant from non-deviant causal chains without making an appeal to those judgments – this is a recipe for a hopeless mission.

The problem for the non-deviance account is that a causal chain belongs to the natural order while the notion of deviance involved in the notion of a deviant causal chain is a normative one, one that is essentially tied to our judgments regarding the rationality of a given belief (and a fortiori our judgments regarding whether the reason-for-which connection holds). There is and can be nothing deviant, in and of itself, in a causal chain qua being causal, something which simply obtains as a part of the natural order. A deviant causal chain is, trivially, a chain that is causal and deviant, but it is not a deviant causal chain because it is “deviantly causal”. There are deviant causal chains but there is no such thing as deviant causation: the “deviance” in question qualifies the chain but does not do so in virtue of qualifying causality. What makes a deviant causal chain deviant is simply the fact that the “output” belief (the end result) of that chain is not rational despite the fact that it is caused by a good reason for it, the fact that the reason for which the output belief is held is not the good reason the subject has for it. Since only an appeal to the normative order can tell us which of those chains in the natural order are deviant, the non-deviance account can be true only at the cost of being trivial.

The causal account is false and the non-deviance account faces the trivialization challenge. The causal account holds that the reason-for-which connection is a causal connection, and given (B), it entails a false sufficiency claim (viz. if the good reason a subject has for a belief causes the belief, then the belief is rational). The non-deviance account holds that the reason-for-which connection is a non-deviant causal connection, and given (B), it entails a qualified sufficiency claim, viz. that if the good reason a subject has for belief non-deviantly causes the belief, then the belief is rational. Without a grip on the notion of non-deviance, the non-deviance account and the accompanying sufficiency claim are empty; and the trivialization challenge is to give an informative account of non-deviance, one that does not appeal to the rationality of ensuing beliefs. I have presented three different versions of the non-deviance account and have shown that all the three corresponding sufficiency claims are false. And, I have also provided a principled reason to think that the trivialization challenge is insurmountable.

The causal account claims, roughly, that causation (more specifically, that a reason for a belief causes the belief) is both sufficient and necessary for (the obtaining of) the reason-for-which connection (between the two). And, roughly again, (A) claims that the reason-for-which connection (between a good reason for a belief and the belief) is necessary for (the) rationality (of the belief) and (B) claims that it is sufficient. The causal account, combined with (A), entails that causation is necessary for rationality, and combined with (B), it entails that it is sufficient. What has given rise to the problem of deviant causal chains for the causal account is the latter entailment, that causation is sufficient for rationality. The causal account needs to avoid that problematic entailment and rest satisfied with a commitment to a necessity claim, viz. that causation is necessary for rationality. And, if causation is only necessary (but not sufficient) for the reason-for-which connection, then given (A), causation is necessary for rationality, but that it is sufficient for rationality does not follow, even with (B). So, the causal theorist, attempting to save from the causal account whatever she can save, might retreat from the causal account to a necessary condition claim and hold that causation is not the whole answer but is at least a part of the whole answer. What I call the soft causal account claims that causation is necessary but not sufficient for the reason-for-which connection.

Is the soft causal account true? My answer is no, and my argument will be indirect. Suppose that we have an adequate account of the reason-for-which connection that has the following features: (i) it provides a sufficiency condition for the reason-for-which connection and (ii) whether that sufficiency condition holds is independent of whether the causal connection deemed to be necessary by the soft causal account holds. That would have shown that the causal connection deemed to be necessary by the soft causal account is not necessary and thus that the soft causal account is false. In the next two sections, I will argue that there is an adequate account of the reason-for-which connection that satisfies (i) and (ii).

3 The Rationalization Account: the “Because” of Argumentation

In this section, I will present and specify a version of a time-honored account of rationality (and the reason-for-which connection), an account which ties rationality to rationalization, to what the subject does or is capable of doing, cognitively speaking. According to the rationalization account, the reason-for-which connection is denoted by the “because” of argumentation and not by the “because” of causation. More specifically, a reason R a subject S has for a belief P is the reason for which S believes P, the rationalization account holds, just in case R is the premise of S’s rationalization for P.⁹ Given (B), the rationalization account entails that if the good reason a subject has for a belief of that subject is the premise of that subject’s rationalization for that belief, then that belief is rational. And, given (A), the rationalization account entails that if a belief of a given subject is rational, then the good reason the subject has for the belief is the premise of that subject’s rationalization for it.

By “rationalization,” I mean an “articulation of an argument in favor of a belief.” Some clarifications are in order. First, rationalization does not require “good” argument but simply argument. Just as there are “bad” arguments, there are also “bad” rationalizations. Second, I take it that arguments are abstract structures composed of propositions and their relations, and as such they do not belong to any concrete subject; but a rationalization is, as I use the term, always a subject’s rationalization. Arguments that are not “articulated” by any subject are not rationalizations. One way to put it is this: arguments are potential rationalizations, and when they are articulated by a subject in favor of a belief, they become (actual) rationalizations. Third, by “articulation”, I do not merely mean entertaining but also endorsing. Arguments that are merely entertained but not endorsed by a given subject are not articulated by that subject according to my quasi-technical use of the term.

According to the rationalization account, the fact that the reason R the subject S has for P is the premise in her rationalization for P suffices for R to be the reason for which P is held by S. I now want to argue that whether this sufficiency condition obtains is independent of whether the causal connection deemed to be necessary by the soft causal account obtains. That is to say, if the fact that the reason R the subject S has for P is the premise in her rationalization for P suffices for R to be the reason for which P is held by S, then that R causes S’s belief that P is not necessary for R to be the reason for which P is held by S. Assume that the antecedent of this conditional is true. In that case, that R causes P is necessary for R to be the reason for which P is held, only if that R causes P is necessary for R to be the premise in the subject’s rationalization for P. However, it is clear that that R causes P is not necessary for R to be the premise in the subject’s rationalization for P. That is, R might be the premise in the subject’s rationalization for P, without R causing the subject’s belief that P. Consider the following case. John believes that he will be promoted in the next year, the premise in his rationalization of this belief is that he has shown full dedication to the company, but his “promotion” belief is not caused by his “dedication” belief but is rather caused, perhaps unknown to him, by his belief that the boss thinks he is a good golf partner. If this case is possible, as it appears to be, then if the rationalization account is right in claiming that rationalization is sufficient for the reason-for-which connection, then the soft causal account is false.¹⁰

The question is whether the rationalization account is right in claiming that rationalization is sufficient for the reason-for-which connection. An adequate answer to this question requires a closer look at the rationalization account. In the rest of this section, I will further specify the rationalization account by drawing two distinctions regarding the notion of endorsement, one between occurrent and dispositional endorsement and the other between personal and public endorsement.

3.1 Occurrent versus Dispositional Endorsement

Here is a paradigmatic case of what I take to be a (subject’s) rationalization (and of what I take to be a subject’s endorsing an argument). Suppose I believe that P, and you ask me now why I believe that P. In response to your query, I say “P because R”, and I am sincere when I say that: I believe that R and I take R to be a good reason for P.¹¹ In this case, P because R is my rationalization for my belief that P – I endorse the argument.¹²

There is a question that needs to be addressed here regarding the very idea of endorsing an argument. Suppose that t refers to the time when the question why I believe that P is raised, and t + refers to the time when I sincerely offer “P because R” as my answer to that question. At t + , I endorse the argument P because R. The question, then, is this: might it be the case that before t + (at t, for instance, a time when I haven’t yet sincerely offered “P because R” as my answer), I still endorse the argument P because R?

My answer is yes, and it is so for the reasons similar to those why it might be the case that I now believe that P while I am not, at this moment, sincerely affirming that P. If at t you ask me, say, whether Paris is the capital of France, and at t + I sincerely affirm, then I believe at t + that Paris is the capital of France. However, before my affirmation (at t, for instance), do I still believe that Paris is the capital of France? There is a good reason to say “yes” because that I believe, before my affirmation, that Paris is the capital of France explains my affirmation: that I sincerely affirm at t + that Paris is the capital of France is explained by, and therefore supports, that I believe, even before that time (at t, for instance), that Paris is the capital of France. As is well-known, this demands a distinction between occurrent and dispositional beliefs: I might now believe that Paris is the capital of France, even though I am not now consciously affirming its content; my current belief in question is not occurrent but dispositional. The relationship between occurrent and dispositional belief is not entirely uncontroversial,¹³ but something like the following seems to be widely taken for granted: a given subject has a dispositional belief that P (or dispositionally believes that P) if the subject has a disposition to have an occurrent belief that P (or a disposition to occurrently believe that P) upon considering a question such as “Is P the case?” or “Do you believe that P?” Sincere affirmative answer to a question like “Is P the case?” is indicative of an occurrent belief that P, and that the subject comes to have an occurrent belief that P, upon considering the question, is explained by the fact that, even before the subject comes to have an occurrent belief that P, she has a dispositional belief that P.

Just as we make a distinction between occurrent and dispositional belief, we can also make a corresponding distinction between occurrently and dispositionally endorsing an argument, and for similar reasons. What is it that explains the fact that at t + I sincerely offer “P because R” as my answer to the question raised at t? One plausible explanation is this: I endorse, not only at t + but also at t, the argument P because R. That I sincerely offer “P because R” as my answer at t + is explained by, and therefore supports, that I endorse at t the argument P because R. That is, I might now be endorsing the argument P because R, even though I am not now consciously offering it at my rationalization for my belief that P; my current endorsement is not occurrent but dispositional.

Once we make a distinction between occurrent and dispositional endorsement, it is clear that I might now be endorsing an argument while not occurrently endorsing it – that is, I might be dispositionally endorsing it. However, there is a further question that needs to be addressed at this point: is the fact that at t + I sincerely offer “P because R” as my answer to the question raised at t why I believe that P a conclusive reason for thinking that I (dispositionally) endorse at t the argument P because R? The ground for attributing a dispositional endorsement of that argument to me is that such an attribution explains why I sincerely offer it in response to a query regarding my belief in its conclusion. If so, then if there are some alternative plausible explanations of why I sincerely offer the argument, then it might be the case that despite the fact that at t + I offer “P because R” as my answer to the question raised at t, I do not (dispositionally) endorse at t the argument P because R. Might there be alternative plausible explanations of why I sincerely offer a particular argument at t + in response to a query at t regarding my belief in its conclusion?

Before answering this question, let me briefly present Audi’s (1982) answer to a similar question regarding the relation between believing and affirming. Audi observes that the following thesis regarding that relation is “widely accepted” (1982, p. 115):

For any person, S, any proposition, P, and any time, t, if, upon being asked, at t, whether P is the case, S would [at t+] sincerely…answer in the affirmative, then, at t, S believes P. (1982, p. 116)

However, Audi notes that this thesis neglects an important distinction, the distinction between (i) dispositionally believing that P and (ii) being disposed to dispositionally believing that P upon considering it. Audi writes:

Suppose that over lunch S is excitedly telling a story. It might be that while he does not in any sense believe (or disbelieve) he is talking too loudly, he would realize and assent to this if he simply entertained the proposition. What such cases show is that we need to distinguish between dispositionally believing and a disposition to believe. (1982, p. 117)

Given this distinction, Audi persuasively argues, the “widely accepted” thesis cited above about the relation between believing and affirming must be rejected:

Affirmation may indicate the formation of a [dispositional] belief, as well as a [dispositional] belief already formed. (1982, p. 120)

The distinction Audi draws between having dispositional beliefs and having dispositions to have dispositional beliefs is plausible; and, once it is granted, it is clear that the fact that sincerely affirming, at a time, a proposition does not by itself indicate a dispositional belief held before that time. The fact that at t + I sincerely affirm (and thereby have the occurrent belief) that P has two competing explanations: either, before t + , I have a dispositional belief that P, or, before t + , I have a disposition to have a dispositional belief that P, upon considering whether P.

If a subject has at a time a disposition to have an occurrent belief that P, upon considering whether P, then the subject has at that time a dispositional belief that P. However, if a subject has at a time a disposition to have a dispositional belief that P, upon considering whether P, the subject does not have at that time a dispositional belief that P: having a disposition to X is not X-ing. There being a disposition to affirm that P indicates the presence of a dispositional belief that P, but there being merely a disposition to have a disposition to affirm that P indicates the absence of a dispositional belief that P. So, sincere affirmation may either indicate an antecedent dispositional belief or indicate merely an antecedent disposition to have a dispositional belief (and hence the absence of an antecedent dispositional belief).

Now, in light of Audi’s original distinction presented above, we can draw a corresponding distinction between dispositionally endorsing an argument and having a disposition to dispositionally endorse an argument, upon considering a relevant question. The fact that I, at a time, sincerely offer “P because R” as my answer to the question why I believe that P does not by itself indicate that I, before that time, dispositionally endorse P because R as my rationalization. This is so because the fact that I, at a time, sincerely offer “P because R” has two competing explanations: either before that time, I have a disposition to occurrently endorse P because R, or before that time, I have a disposition to have a disposition to occurrently endorse P because R, upon considering a relevant question. So, sincere offering of an argument may either indicate an antecedent dispositional endorsement or indicate merely an antecedent disposition to have a dispositional endorsement of it (and hence the absence of an antecedent dispositional endorsement).

How can we tell whether, at a certain time, a given subject dispositionally endorses an argument or merely have a disposition to dispositionally endorse it, upon considering a relevant question? I don’t think we can draw a precise line here, but we can offer some defeasible criteria for preferring one option over the other. One criterion appeals to considerations as to whether the subject in question goes through a process of explicit, conscious deliberation in her attempt to answer the relevant question. Consider the following cases:

Case 3: You ask me why I believe that P. The question immediately activates my antecedent disposition to occurrently believe that R, accompanied with a strong (but perhaps resistible) ‘impression’ that R provides adequate support for P. And, without going through any conscious deliberative process, and as automatic and fast as these things can be, I say “P because R” in response.

Case 4: You ask me why I believe that P. The question does not immediately activate any antecedent disposition to have some relevant occurrent belief. Rather, I go through a conscious deliberative process, entertaining various alternative reasons such as R, Q and S, that might count for P, neither of which initially strikes me as (or comes with an ‘impression’ of) providing adequate support for a belief that P; and, after a considerable amount of reflection and effort on the issue, I finally come to say “P because R”, as an expression of my “all things considered” endorsement, in response.

It is plausible to say that, in Case 3, I dispositionally endorse P because R even before the relevant question is raised, and that in Case 4, I merely have a disposition to dispositionally endorse P because R before the relevant question is raised. This suggests that absence of conscious deliberation indicates antecedent dispositional endorsement, and lengthy conscious deliberation merely indicates antecedent disposition to have a dispositional endorsement.

Now, consider the following case:

Case 5: You ask me why I believe that P. As in Case 3, the question immediately activates my antecedent disposition to occurrently believe that R, accompanied with a strong (but resistible) ‘impression’ that R provides adequate support for P. However, in this case, for one reason or another – I might for instance be in a skeptical mood –, I don’t automatically say “P because R” in response. Rather, as in case 4, I go through a conscious deliberative process, entertaining various reasons (alternative to R) such as T and S that might count for P but neither of which initially strikes me as (or comes with an ‘impression’ of) providing adequate support for a belief that P; and, after a considerable amount of reflection and effort on the issue, I finally judge that T but not R provides adequate support for P and come to say “P because T”, as an expression of my “all things considered” endorsement, in response.

Case 5 has some features in common both with Case 3 and with Case 4, and therefore might plausibly be thought of as an “in-between” case. The intuitive thing to say about Case 5 is that before the question why I believe that P is raised, I dispositionally endorse P because R but do not dispositionally endorse P because T (while it might be true that I have a disposition to dispositionally endorse it). Absent conscious deliberation, I would offer “P because R” as my response in view of the fact that R immediately strikes me as providing adequate support for P, which provides the basis for attributing to me a dispositional endorsement of P because R. Enter conscious deliberation, and I come to say “P because T”, which provides the basis for attributing to me, at most, a disposition to dispositionally endorse P because T.

In light of the foregoing interpretation of these cases, I offer the following criteria to distinguish between a dispositional endorsement and a disposition to have a dispositional endorsement:

(C1) If it is now true that if asked why S believes that P, S would, absent conscious deliberation, offer “P because R” as her answer, then S dispositionally endorses P because R now.¹⁴

(C2) If it is now true that if asked why S believes that P, S would offer “P because R”, only after (lengthy) conscious deliberation, as her answer, then S has merely a disposition to dispositionally endorse P because R now.

Here is then where we have arrived. The rationalization account claims that a reason R a subject S has for a belief P is the reason for which S holds P just in case R is the premise of S’s rationalization for P, and R is a premise of S’s rationalization for P just in case S endorses an argument from R to P. There are two ways in which a given subject might endorse an argument – either occurrently or dispositionally; and, if the subject now merely has a disposition to dispositionally endorse an argument, then she does not endorse it now in any sense. The question arises how we can mark the distinction between dispositionally endorsing an argument and having a disposition to dispositionally endorse an argument. I have suggested, as a response, a phenomenological criterion that appeals to the presence or absence of some pertinent conscious deliberation once the relevant question is considered. The upshot is this: a given subject endorses an argument from R to P just in case she either occurrently endorses it or, if asked why believes that P, would, absent conscious deliberation, offer “P because R” as her answer.

3.2 Personal versus Public Endorsement

A subject’s sincere endorsement of an argument is not the only way in which that argument may become her rationalization. Suppose that I say “P because R”, but I’m insincere when I say that: either I don’t believe that R or I don’t take R to be a good reason for P. Under these circumstances, is P because R still my rationalization for my belief that P? It is clear that P because R is a rationalization I offer for my belief that P, so there is a clear sense in which I endorse the argument P because R, which means that there is also a clear sense in which P because R is my rationalization. However, suppose also that if I were sincere in my response to your query, I would say “P because Q” rather than “P because R”. If this is so, then there is also a clear sense in which I don’t endorse the argument P because R but endorse the argument P because Q, and which means that there is also a clear sense in which P because Q is, and P because R is not, my rationalization.

That the argument I offer to someone else might be different from the argument I offer, as it were, to myself suggests that we need to make a distinction between two different ways of endorsing an argument and therefore to make a distinction between two different ways in which an argument might be a subject’s rationalization. A subject publicly endorses an argument if that is the argument the subject offers to someone else, and that in such a case, the argument is the subject’s public rationalization. And, a subject personally endorses an argument if that is an argument the subject offers, as it were, to herself, and in such a case, the argument is the subject’s personal rationalization.¹⁵ It is clear that a subject’s personal rationalization might be the same as or different from her public rationalization; and, in case they are the same, the subject is sincere.

The distinction between personal and public rationalization gives rise to the distinction between personal and public rationalization accounts. The former claims that the-reason-for-which connection has to do with personal rationalization, and the latter claims that it has to do with public rationalization. More specifically, we have the following:

PeR: A reason a subject has for a belief is the reason for which the subject holds it just in case the reason is the premise in the subject’s personal rationalization for the belief.

PuR: A reason a subject has for a belief is the reason for which the subject holds it just in case the reason is the premise in the subject’s public rationalization for the belief.

In case the subject’s personal and public rationalizations are the same, she is sincere. So, in order to decide between personal and public rationalization accounts, we need to take a look at those cases in which a subject is insincere, cases in which her personal and public rationalizations differ, since in those cases, these two accounts give different verdicts about the rationality of a given belief. Now, consider the following case:

Case 6: Susan has excellent reasons to think that John is guilty, and those reasons are the premises of her public rationalization for her belief that John is guilty. However, Susan does not take those excellent reasons that she offers as her public rationalization as what they are, and she instead thinks that they are weak. She believes that John is guilty rather because he looks suspicious: that John looks suspicious is the premise in her personal rationalization for her belief that John is guilty.

Intuitively, Susan’s belief that John is guilty is not rational but might appear to be rational to “an outsider”: a person who judges that Susan’s belief is rational on the basis of the reasons she publicly offers would be mistaken. However, given (B), the public rationalization account entails that Susan’s belief in question is rational (and does not merely appear to be so): the excellent reasons Susan has for the belief are the premises of her public rationalization for it. The personal rationalization account, on the other hand, entails, given (A), that Susan’s belief is not rational: the good reasons Susan has for the belief are not (though, some others, the bad ones are) the premises of her personal rationalization for it. So, Case 6 supports the personal rationalization account while undermining the public rationalization account.

Let us take a closer look at what exactly Case 6 supports and undermines. Combined with (A) and (B), the public rationalization account entails the following theses:

PuR₁: If the good reason a subject has for a belief is a premise of that subject’s public rationalization for that belief, then that belief is rational.

PuR₂: If a given belief of a certain subject is rational, then the good reason the subject has for it is a premise of that subject’s public rationalization for that belief.

PuR₁ entails that in Case 6, Susan’s belief that John is guilty is rational. However, intuitively, Susan’s belief is not rational. So, PuR₁ is false. However, Case 6 does not show that PuR₂ is false.

Furthermore, combined with (A) and (B), the personal rationalization account entails the following theses:

PeR₁: If the good reason a subject has for a belief is a premise of that subject’s personal rationalization for that belief, then that belief is rational.

PeR₂: If a given belief of a certain subject is rational, then the good reason the subject has for it is a premise of that subject’s personal rationalization for that belief.

PeR₂ entails that in Case 6, Susan’s belief that John is guilty is not rational. And, intuitively, Susan’s belief that John is guilty is not rational. So, PeR₂ is supported by Case 6. However, Case 6 does not support PeR₁.

So, Case 6 supports PeR₂ and undermines PuR₁, but it does not support or undermine PeR₁ or PuR₂. Now, it is easy to imagine a case that supports PeR₁ and undermines PuR₂:

Case 7: Susan has excellent reasons to think that John is guilty, and those reasons are the premises of her personal rationalization for her belief that John is guilty. However, when asked why she believes that John is guilty, the only reason that she offers is that John looks suspicious (because, say, she feels tired and has no intention to list all those excellent reasons she has): that John looks suspicious is the premise in her public rationalization for her belief that John is guilty.

Intuitively, Susan’s belief that John is guilty is rational but might appear not to be rational to an outsider: a person who judges that Susan’s belief is not rational on the basis of the reason she offers would be mistaken. However, PuR₂ entails that Susan’s belief is not rational (and does not merely appear not to be rational). PeR₁, on the other hand, entails that Susan’s belief is rational. So, Case 7 supports PeR₁ and undermines PuR₂.

Furthermore, Case 7 also serves as a counter-example to the soft causal account, the thesis that causation is necessary for the (obtaining of the) reason-for-which connection. If, in Case 7, Susan’s belief that John is guilty is rational, as it seems to be the case, then given (A), the good reason she has for that belief is the reason for which she has the belief. However, as it is described, causation plays, in Case 7, no role (and hence no necessary role) in its being the case that the good reason Susan has for her belief is the reason for which she has the belief. So, Case 7 describes a scenario in which the reason-for-which relation between a belief and a reason obtains irrespective of whether there is a causal relation between the two.¹⁶

The rationalization account I defend in this paper is the personal rationalization account. According to the personal rationalization account, a reason R a subject S has for a belief P is the reason for which S holds P just in case R is a premise of S’s personal rationalization for P. And, in order for R to be a premise of the subject’s personal rationalization for P, the argument from R to P must be personally endorsed by the subject, where the endorsement in question might be either occurrent or dispositional. The personal rationalization account is supported by such cases as 6 and 7. Are there any good reasons to think that it is false? I argue in the next section that the answer is no.¹⁷

4 Objections Answered

There are two basic different kinds of objection that might be raised against the personal rationalization account, because that account holds, bluntly put, two main theses: rationalization is necessary and sufficient for the (obtaining of the) reason-for-which connection. Call the former “the Necessity Thesis” and the latter “the Sufficiency Thesis”. Furthermore, given (A) and (B), the rationalization account entails two further theses, namely, PeR₁ and PeR₂. Given (A) and (B), the rationalization account is false if either PeR₁ or PeR₂ is false. There are various reasons to think that either the Sufficiency Thesis (and PeR₁) or the Necessity Thesis (and PeR₂) is false, but they are not convincing – or so I will argue.

4.1 Objections against the Sufficiency Claim and PeR₁

I will now present four objections against the Sufficiency Thesis or PeR₁, each of which is followed by a reply. As we have seen in Sect. 3, if rationalization is sufficient for the reason-for-which connection, that is if the Sufficiency Thesis is true, then the soft causal account is false. So, defending the Sufficiency Thesis against the following objections contributes to undermining the soft causal account.

Objection 1

Consider the following scenario. Mary believes that Jack is cheating on her, and has two different reasons for this. One is (R₁) that her loyal friend Susan told her that Jack was hanging out with a different girl last night, laughing and having fun, and the other is (R₂) that Jack has been receiving some persistent unknown calls from “a stranger” lately. R₁ is a premise in one of her rationalizations for her belief, and R₂ is a premise in another. The rationalization account entails that both R₁ and R₂ are the reasons for which Mary holds the belief. However, it is surely possible that despite the fact both serve as premises in Mary’s rationalizations, only one of them – suppose R₂ – is the reason for which Mary believes that Jack is cheating.¹⁸

Reply

This objection is fallacious: either it begs the question against the rationalization account or it rests on equivocation. The rationalization account entails that given Mary’s epistemic life, the possibility that only one of those two reasons is the reason for which Mary believes that Jack is cheating is eliminated: simply claiming that it is possible begs the question. On the other hand, there is a sense of “the reason for which” in which, given Mary’s epistemic life, it is possible that only one of those reasons is the reason for which Mary believes that Jack is cheating – for instance, the sense in which only of those reasons causes the belief. However, the point is that the reason-for-which connection characterized by (A) and (B) need not be the same as the reason-for-which connection specified by causal relations.

Objection 2

The rationalization account assimilates rationality (and “real reasons”) to rationalizability (and “rationalizations” respectively), and this is a mistake because there are beliefs that are merely rationalized but not rational. To see this, let us first assume, plausibly, that the reason that motivates one to believe a proposition – one’s real reason for believing that proposition – is the same as the reason for which one believes that proposition. Now consider the following scenario. You ask me why I believe (H) that a teacher ought to be harsh on her students, and I say “Good education demands strict discipline”, something which I do not believe but say nonetheless to convince you. In this case, (R₁) that good education demands strict discipline is merely my rationalization, and it is not the reason that motivates me to believe H, it is not my real reason. Rather, my motivating (real) reason is (R₂) that students tend to mock those teachers that are not harsh on them. However, the rationalization account entails that given that R₁ is the premise that occurs in my rationalization, R₁ is my real reason, and also that assuming that R₁ is a good reason for H, H is rational, irrespective of whether R₂ is a good reason for H or not. However, both entailments are false: R₂ is the reason for which I believe H, and H is not rational if R₂ is not a good reason for H.¹⁹

Reply

It is true that the rationalization account assimilates rationality to (some kind of) rationalizability, but it does not follow that it is incapable of distinguishing rational from merely rationalized beliefs. It is capable of doing so by an appeal to the distinction between personal and public rationalization. According to the rationalization account, merely rationalized beliefs are those beliefs for which the subject provides a public rationalization that is not her personal rationalization: if a subject publicly defends her belief that P by reference to R, while she either does not believe R or does not take R as a good reason for P, then her belief that P is merely rationalized and, assuming she has no other reasons for P, is not rational. The example mentioned in the objection above fits nicely into this picture: the subject’s belief in question that a teacher ought to be harsh on her students is merely rationalized because the public rationalization that the subject provides for that belief is not her personal rationalization. Furthermore, according to the rationalization account, the reason for which a subject believes a proposition – her real reason – is the reason that serves as a premise in her personal rationalization for that proposition. So, it entails the correct result that R₂ is the subject’s real reason.

Objection 3

The rationalization account assimilates rationality to personal rationalizability, and also assimilates the reason for which one believes a proposition to the reason that serves as a premise in her personal rationalization. However, this is a mistake because the following scenario is possible. I now have the belief (G) that global warming is real, and I have had this belief since the time I watched a documentary about global warming four years ago. I also have a good reason R₁ for this belief (e.g., that atmospheric carbon dioxide concentrations are about 35% higher than before the industrial revolution). However, the reason for which I believe that global warming is real is not R₁ but R₂, which is itself a bad reason for the belief (e.g., that summers feel hotter to me now than they felt to me when I was a little kid), despite the fact that the reason I would sincerely offer, if asked why I believe that global warming is real, is R₁: once the question is raised, I would come to appreciate, after some deliberation, the force of R₁, and offer it (instead of R₂) as my reason. However, all the same, at this very moment when the question has not yet been raised, R₁ is not the reason for which I believe that global warming is real, though the rationalization account entails that it is.²⁰

Reply

It is true that the rationalization account assimilates the reason for which one believes a proposition to the reason that serves as a premise in her personal rationalization, but it does not have such unwelcomed consequences as those the objection claims it has. And, this is so because it does not simply identify a subject’s personal rationalization with the rationalization she would sincerely offer upon considering a relevant question. To see this, recall the distinction drawn above between dispositionally endorsing an argument and having a disposition to dispositionally endorse an argument. A subject (personally) dispositionally endorses an argument if it is now true that that is the argument she would, absent conscious deliberation, (sincerely) offer, upon considering a relevant question. And, a subject has a disposition to dispositionally endorse an argument if it is now true that that is the argument she would, only after (lengthy) conscious deliberation, (sincerely) offer, upon considering a relevant question. In the example given in Objection 3, the subject in question would come to appreciate the force of R₁ through some deliberation. According to the rationalization account, then, there is some ground for claiming only that the subject has a disposition to dispositionally endorse an argument from R₁ to G but not that the subject (personally) dispositionally endorses that argument. So, the rationalization account delivers the correct result that as of now, when the relevant question has not yet been raised, R₁ is not, but another reason such as R₂ might be, the reason for which the subject believes G.

Objection 4

The rationalization account entails that if there are two different subjects S₁ and S₂ that have the same belief that P such that these subjects have the same good reason R for P and they have the same personal rationalization from R to P, then both S₁′s belief and S₂′s belief are rational. Now suppose that R causes S₁′s belief that P but it does not cause S₂′s belief that P. Suppose further that S₂′s belief that P is caused by a mental state that is not a good reason for the belief that P, say, S₂′s desire that P. Is it not then the case that the rational status of S₁′s belief that P differs from the rational status of S₂′s belief that P, despite the fact that they both satisfy the two conditions deemed to be sufficient by the rationalization account?²¹

Reply

I don’t think that this question, despite its rhetorical force, must be answered affirmatively. The fact that the cause of S₂′s belief that P is her desire that P might tell something about S₂′s cognitive traits in general but it is not relevant to an assessment of the rational status of her particular belief that P. The fact that S₂′s belief that P is caused by a mental state that is not a good reason for it might be indicative of the fact that S₂ has a general disposition to form irrational beliefs: if, for instance, S₂′s beliefs are caused in general by her desires (her belief that P being a case in point), and if those beliefs lack adequate support, then S₂ has a general tendency to form irrational beliefs. But nothing in the case described above suggests that the antecedents of this conditional are satisfied. Furthermore, even if we assume, for the sake of the argument, that S₂ has a general tendency to form irrational beliefs, that does not imply anything by itself about the rational status of her particular belief that P. The fact that one has a rational belief is consistent with the fact that one has a general tendency to form irrational beliefs. A person might have a general tendency to take a beating in a fight but in this case, he did not.

If the fact that S₂′s belief that P is caused by her desire that P were to detract from the rationality of her belief that P, then S₂′s learning about the causal history of her belief would be expected to have a certain effect, i.e., that her belief that P would be less rational after the learning takes place than it was before. This is so because if there is a factor that counts against the rationality of a belief (if there is a “rationality detractor” for the belief), and if the subject of that belief becomes aware of the fact that there is such a factor, then such an awareness makes an epistemic difference: it is now even less rational for the subject to hold the belief than it was before. Suppose that I have the occurrent belief that there is a pink elephant before me now, and my evidence for the belief is the experience I am having of there being a pink elephant before me now. Suppose also that I have the dispositional but not occurrent belief that I have taken some hallucinatory drugs this morning. Even though the latter belief is not right “before my mind”, it seems plausible to suggest that it detracts from the rationality of my occurrent belief about the pink elephant: that I have such a dispositional belief makes my occurrent belief less rational than it would otherwise be. However, suppose now that the dispositional belief I have becomes occurrent: now, there are two occurrent beliefs right before my mind, one about the pink elephant and the other about the hallucinatory drugs. If, under these circumstances, I continue to hold the belief about the pink elephant, then that belief, I claim, is even less rational than it was before. The subject’s conscious awareness of the rationality detractor makes a difference, one that is unfavorable to the rational status of the belief. So, if, in the scenario above, being caused by a desire that P is a rationality detractor for S₂′s belief that P, then, if S₂ learns that her belief that P is caused by such a desire, then S₂′s belief that P is less rational than it was before. However, this is not the case: if S₂ learns that her belief that P is caused by her desire that P, then her belief that P is as rational as it was before. Learning that her belief that P is caused by her desire that P might come as a surprise for S₂, and it may even lead her to raise doubts about her purported knowledge regarding the causal transactions between her mental states. But, as long as she has the good reason R for the belief that P and R is the premise in her rationalization for the belief that P, it is as rational for her to hold the belief that P as it was before. Upon learning the cause of her belief that P, S₂ may say something like “That’s interesting!”, and then simply shrug her shoulders and continue to believe that P as rationally as before. The rational thing for S₂ to do is still to believe that P, and it is as rational as before.

I have considered four objections against the Sufficiency Thesis or PeR₁ and shown that each receives a plausible reply. And, given the intuitive support the Sufficiency Thesis and PeR₁ receive from such cases as Case 7 in the previous section, I conclude that they are true (and therefore that the soft causal account is false).

4.2 Objections against the Necessity Thesis and PeR₂

I will now defend the Necessity Thesis and PeR₂ against two possible objections.

Objection 1

Rationalization is a “higher-level” cognitive achievement: it requires taking something as a reason for something else and having the ability to offer an argument from what is taken as a reason to what it is taken as a reason for. It is clear that human infants and non-human animals cannot have rationalizations for their beliefs, while it is also clear that some of their beliefs are rational. So, contra the Necessity Thesis, rationalization is not necessary for rationality.

Reply

Consider two subjects, S₁ and S₂, and suppose that both believe that P, and that while S₁ has a good reason R for the belief that P, S₂ does not have any. And, suppose further that the reason for which S₁ believes that P is not R but is either a bad reason for P or mere hunch. Now, there is a clear sense in which S₁′s belief that P is epistemically superior to S₂′s belief that P: S₁ has a good reason for P, and S₂ has none. And, this is so despite the fact that the good reason R S₁ has is not the reason for which S₁ believes that P. The epistemic superiority here entitles us to say this: S₁′s belief that P is more rational than S₂′s belief that P.

Now, take such cognitive beings as human infants that do not have those higher-level cognitive abilities required by the rationalization account. I grant that a human infant might have a rational belief in P in the sense that she might have a good reason for that belief, by virtue of which her belief that P might be more rational than someone else’s belief that P. However, note that the notion of rationality the rationalization account attempts to explain is more demanding than the one that licenses an attribution of rationality on the mere basis of having good reasons: it requires not only having good reasons for a belief but also those reasons’ being the reason for which that belief is held.

This leaves the proponent of the objection with a dilemma: the notion of rationality that she deploys in her claim that some of the beliefs human infants have are rational is either less demanding (hence different) from or the same as the notion the rationalization account attempts to explain. If the former, then the rationalization account is safe because what it is concerned with is different from what the objection claims it excludes. And, if the latter, then it is not clear that human infants have rational beliefs in that more demanding sense and a mere assertion that they have rational beliefs in that sense simply begs the question against the rationalization account.

Objection 2

Suppose that John is taking an oral exam on the history of the Ottoman Empire, and he is all too nervous because he knows that if he fails, his mother, an expert in the Ottoman history, will be devastated. The examiner asks the (first) question, “When did the first war between Russia and the Ottoman Empire start?”, and John feels lucky because he has a vivid memory impression of his history teacher telling in class the (correct) answer to this question, “1676”. John gives the answer as he (thinks he) remembers it, and then the examiner asks an unexpected (second) question, “Why do you think so?” Being nervous and prepared to answer only history questions, John becomes cognitively paralyzed by the examiner’s query about his epistemic credentials and cannot find a satisfactory answer to her “why” question, he can’t even remember at that very moment that he had a pertinent vivid memory impression just a moment ago. In this case, John fails to rationalize (i.e., fails to articulate an argument for) his belief about the first Russo-Turkish war in response to a query. Despite this failure, however, John’s belief appears to be rational. So, PeR₂ is false.

Reply

There are two major assumptions in this objection. The first one is that since, in the scenario described, John fails to rationalize his belief about the first Russo-Turkish war when the relevant query is made, he does not dispositionally endorse any relevant argument before the query is made. And, the second one is that since John does not dispositionally endorse any relevant argument, the rationalization account entails that John’s belief about the first Russo-Turkish war is not rational because, the objection assumes, whatever good reason John might have about that war is a reason for which John has that belief, according to that account, only if that good reason is a premise in an argument that he dispositionally endorses.

Both assumptions are suspect, but since showing that one of them is false is enough to undermine the objection, I will target only one of them, viz. the second one. Let me grant, for the sake of the argument, that John does not dispositionally endorse any relevant argument. But the rationalization account does not entail that John’s belief in question is not rational, and this is because that account does not exclude the possibility of (occurrently) endorsing an argument without dispositionally endorsing it. The question, then, is whether John occurrently endorses (at an appropriate time) any relevant argument for his belief in question, and there is a good reason to say “yes”. To see this, note, first, that, as the case is described, when the examiner raises the question about the time of the Russo-Turkish war, it does not appear to John as if he is simply guessing, that he is without any reason to think that the answer he gives is correct. On the contrary, his vivid memory impression is his (good) reason for his belief: if it were not for that memory impression, John’s belief would not be rational. And, secondly and more importantly, not only does John have that reason for his belief, but he also appreciates its force, that it is a good reason for his belief (that’s why he feels lucky when the examiner raises the first question): John takes his memory impression as a good reason for his belief. So, John occurrently endorses the argument from the reason he has (i.e. his vivid memory impression) to that belief at the time when he answers the examiner’s first question (which is about the date of the first Russo-Turkish war).

Of course, that there is a time when John occurrently endorsed an argument from a good reason for the belief in question to that belief is not sufficient for John’s belief to be rational now: the endorsement in question must occur at an appropriate time. Suppose for instance that John occurrently endorsed an argument from a good reason for a belief to that belief a long time ago (say, five years ago) (and he neither occurrently nor dispositionally endorses any such argument for his belief now). This is surely not sufficient to render John’s belief rational now. However, if John is now occurrently endorsing a relevant argument, the rationalization account holds, his belief in the conclusion of that argument is rational now.

In the case described, John occurrently endorses a relevant argument for his belief about the date of the first Russo-Turkish war at the time when he answers the examiner’s first question but not at the time when he becomes paralyzed by her second question (which is about the epistemic credentials of Jack’s belief about the date of the first Russo-Turkish war). The question is whether the rationalization account as such is committed to holding that at that time when he becomes thus paralyzed, Jack’s belief about the date of the first Russo-Turkish war is not rational, which we intuitively grant is rational. And, I want argue that the answer is no. The rationalization account might hold, plausibly I think, that if a subject occurrently endorses an argument from a good reason R she has to her belief that P at t, and if t + is a time after, but relatively close to, t, then the subject’s belief that P is not only rational at t but is also rational at t + . The rationality granted to the subject’s belief in virtue of her occurrent endorsement at t of such an argument has “a life span” that is not limited to the moment t but extends from t to a time afterwards (and plausibly including t +). So, even if at t + , the subject does not occurrently or dispositionally endorse any argument from R to P, her belief in question might still be rational in virtue of her occurrent endorsement of such argument at a prior but relatively close time t. We can think of the status “rationality” as something earned and attached to a belief thanks to having occurrently endorsed a (good) argument in its favor and as something, once earned, that keeps being attached to the belief for some time. This is just like what happens when one gets a title in sports (or in general). Thanks to winning a world heavy-weight boxing championship title match, for instance, a boxer earns the title “the world heavy-weight boxing champion” and he keeps that title for some time (e.g., up until the next title match he gets defeated).

So, the rationalization account can reply to the objection at hand as follows. After the examiner raises her first question, John occurrently endorses a (good) argument for his belief about the date of the first Russo-Turkish war. John’s failure to articulate the good reason he has for the belief in response to the examiner’s second question simply signals a fact which we have already granted, viz. that he does not occurrently or dispositionally endorse any argument for his belief at that moment. However, the belief still retains the earned title “rational” given that the title itself has a life of its own, which surely extends, given the shortness of the distance, from the time when John occurrently endorses the relevant argument to the time when he becomes paralyzed by second question. So, the rationalization account can accommodate the intuitive claim that John’s belief is rational.²²

I have considered two objections against the Necessity Thesis or PeR₂ and have shown that each receives a plausible reply. And, given the intuitive support the Necessity Thesis and PeR₁ receive from such cases as Case 6 in the previous section, I conclude that they are true. Combined with the conclusion of the previous subsection, the upshot is that the personal rationalization account is true.

5 Conclusion

My main aim in this paper has been to provide an adequate account of the reason-for-which connection, which is to be understood as a connection that is non-accidentally tied to the rationality of beliefs. In this regard, I have argued for two main theses: firstly, neither the causal account, nor the non-deviance account, nor the soft causal account is true; and secondly, the personal rationalization account is true, according to which a reason R a subject S has for a belief P is the reason for which S holds P just in case R is a premise of S’s personal rationalization for P.

As for the first thesis, I have more specifically argued that the problem of deviant causal chains shows that the causal account is false and also that the non-deviance account faces the trivialization challenge, which I have maintained is insurmountable. As for the second thesis, I have more specifically argued that in order for a reason R to be a premise of the subject’s personal rationalization for P, the argument from R to P must be personally (and not merely publicly) endorsed by the subject, where the endorsement in question might be either occurrent or dispositional. After showing that the personal rationalization account receives intuitive support from a number of cases, I have also argued that it survives some formidable objections that might be raised against it. The moral is that the reason-for-which connection is a connection that obtains between a reason R and a belief P just in case the subject personally endorses an argument from R to P, which entails, as a corollary, that the soft causal account is false.