2. The simple arbitrariness objection to permissivism

The permissivist thinks fully rational people who have the same evidence can sometimes disagree. So, some total bodies of evidence and some propositions admit of multiple, maximally-good-from-the-standpoint-of-epistemic-rationality, responses (e.g., belief, suspension, 0.6 credence, 0.8 credence, etc.). But the permissivist also thinks these people can pick particular responses, or take up particular attitudes, and be fully committed to those. According to the permissivist, even when a 0.6 credence is no better supported than, say, a 0.4 credence, someone could be fully rational in holding 0.6. That is, someone could sincerely and completely inhabit a 0.6 credence, rejecting the (equally recommended by evidence) 0.4 option.

Here's the worry: such picking and committing to one among maximally good options must be ‘arbitrary’, or at least guided by non-epistemic reasons. For if multiple attitudes are maximally recommended on the basis of epistemic considerations, then a person's having any particular rational attitude rather than another must depend on arbitrary (epistemically arational) features of her assessment of the evidence. But it is implausible – the thought goes – that fully, epistemically rational responses could be so arbitrary or influenced by non-epistemic reasons. We oughtn't, from the standpoint of epistemic rationality, fully inhabit particular attitudes no better than other alternatives. Surely, instead of being 0.6, we ought to be in some more suspended, neutral state recognizing the no-worse merit of being 0.4. But to recommend some unique such suspension state is to give up permissivism.

More precisely, this objection incorporates three premises, which seem jointly inconsistent.

Arbitrariness Triad #1

P1 and P2 apparently entail that agents can “pick” any particular rationally recommended attitude from among a set of such attitudes, though it is no better than alternatives in that set, and thereby come to hold a rational attitude. But P3 claims that a rational attitude cannot have been merely as good as alternatives. It must have been strictly better.

How should a permissivist respond? An epistemic existentialist will reject P2. But I acknowledge that, initially, P2 may seem to be the least objectionable of these premises.

A perhaps more attractive route would be to give up P3. After all, this sounds potentially question-begging against the permissivist. P3 says that a rational attitude must be strictly better than all other possible attitudes toward the proposition in question. Perhaps rationally recommended attitudes – i.e., attitudes that could be rational for an agent, given her evidence – may be merely maximally good with respect to whatever properties conduce to rationality. But an actual, rationally held attitude must have had a maximum of such properties, ex ante. This is not only a strong claim but also perhaps a strange one. Why should we think that epistemically rational attitudes must be better than alternatives, while practically rational actions clearly needn't be? Clearly, Buridan's ass can rationally eat from either bale of hay.Footnote ¹¹

Interestingly, however, many actual defenses of permissivism rely instead on the disambiguation of P1. We can distinguish what's called intrapersonal permissivism from a weaker form, interpersonal permissivism. Consider:

P1_{Intrapersonal}, together with P2, does apparently entail that at least some agent can “pick” any of multiple rationally recommended attitudes, though they are each no better than alternatives, and thereby come to hold a rational attitude. But P1_{Interpersonal} does not have this implication. If we assume that a permissivist is committed only to interpersonal permissivism (as indeed, is the seemingly more popular position), then she needn't license any individual's picking among multiple, maximally good attitudes. Agents can simply “pick” the attitude that's uniquely rationally recommended for them and thereby come to hold a rational attitude. Merely interpersonal permissivists, then, seem to have a straightforward way to wriggle out of this first and simplest version of the arbitrariness objection.Footnote ¹²

However, two problems linger.

3. Lingering problem #1: intrapersonal permissivism

The first problem is that, arguably, subjectivism pushes us not just to interpersonal permissivism but to intrapersonal permissivism.

Why? Subjectivists are committed to thinking that rational attitudes are determined by agents’ frameworks. (Many would specify: rational attitudes are determined by agents’ legitimate or permissible frameworks. Even subjectivists may want objective constraints on the kinds of frameworks that are rationality-conferring.Footnote ¹³) But since these frameworks are grounded in an agent's psychology,Footnote ¹⁴ there seems to be no reason to suppose they are complete (in the sense of weighing in on all possible evidential scenarios and propositions) and free from internal tension.

Say an agent finds herself in a novel evidential scenario. Perhaps she is confronted for the first time with conflicting testimony from two sources she simply recognizes as “really trustworthy”, or perhaps she is just considering a proposition very foreign to everyday affairs, where her evidence is of little weight.Footnote ¹⁵ She may be genuinely torn and at a loss as to what her evidence supports, even according to her own best lights. Her phenomenology may be of trying on different ways of thinking about the issue, different policies to use to settle the matter, without any option seeming superior.

The subjectivist's normative elevation of agents’ psychologically grounded frameworks, together with the potential for tension or incompleteness in the recommendations of psychologically grounded frameworks, should push the subjectivist toward intrapersonal permissivism. For if a person's framework is genuinely unable to decide between, e.g., suspension and belief, it seems the subjectivist cannot say either is uniquely rational for that person.

This argument is clearly too quick; I discuss these issues in greater detail elsewhere.Footnote ¹⁶ My claim here is weak: there seems be an argument that subjectivists should not stop at interpersonal permissivism but rather should endorse the stronger, intrapersonal version of P1. And if this is so, the standard permissivist response to the arbitrariness objection will not work.

4. Lingering problem #2: framework arbitrariness

Even if one maintains a merely interpersonal version of permissivism, however, the standard permissivist response can seem only to have pushed the worry “back a step” (Feldman Reference Feldman and Antony2007: 206). The standard permissivist response, again, says that we each simply ought to have the (unique!) attitude recommended by our individual frameworks. But whose framework is better, from the standpoint of rationality? If there is no good answer to this question, as permissivists claim there is not, then the very frameworks with which we find ourselves might seem problematically arbitrary. White (Reference White2005: 452) seems to have something similar in mind, where in the following quote he uses “epistemic standards” much as I use “framework”:

We might formalize this higher-order worry in a parallel structure, with three seemingly jointly inconsistent theses.

Arbitrariness Triad #2Footnote ¹⁷

How might the permissivist respond? First, notice that, whereas the original arbitrariness worry (Arbitrariness Triad #1) relied on the impossibility of having a rational attitude no better than alternatives (P3), this version relies on the supposed impossibility of having rationality-conferring frameworks one is not committed to seeing as better than alternatives (P6). In part because of this feature, perhaps the permissivist will protest that P6 is less plausible than P3. Do we really think that one must take one's framework to be best? After all, the permissivist may at this point wish to appeal to the inevitability of relying on one's own framework, regardless of how arbitrary we may admit it to be on reflection.Footnote ¹⁸ Perhaps, contrary to P6, we can employ our frameworks simply because we're stuck with them, while acknowledging that there's no reason for preferring our own to others.

But even if we grant the permissivist the relevant sort of inevitability – even if we grant that we are “stuck” with our frameworks – this seems less than satisfactory; necessity is cold comfort. Indeed, this response seems merely to deny that we have pills that could swap out legitimate frameworks for other legitimate frameworks – not that we would do just as well by taking them were they available.

A different response, one given in Simpson (Reference Simpson2016), again relies on disambiguating between intra- and inter-personal forms of a kind of permissivism – this time, a higher-order permissivism about frameworks. Simpson argues that an agent may see her framework as best for her, given her cognitive abilities, while acknowledging that different frameworks may be better for others, given their different cognitive abilities. So, on this view, P4 is only true in a qualified, interpersonal sense:

Given this version of P4, Simpson could reinterpret P6 intrapersonally and avoid tension:

Sure, the thought would go, other people's frameworks might be just as good for them as mine is for me, but mine really is best for me, given my cognitive abilities. This is an interesting proposal, and it does seem to fit the bill of explaining the non-arbitrariness of an agent's framework. Simpson can say that an agent has (or perhaps should have) the particular framework she does because that is the framework best suited to her cognitive abilities.

My principal worry for Simpson's approach, which echoes section 3 above, is that we should doubt that an agent with certain cognitive abilities will always be best served by one, unique framework. For at least some agents, multiple frameworks – pretty clearly, multiple precisifications or refinements of those agents’ standing frameworks – may do maximally well, even once we hold fixed their cognitive abilities.Footnote ¹⁹

To summarize: permissivists have a standard answer to the arbitrariness objection (Abritrariness Triad #1) appealing to subjective frameworks, but that standard response is inadequate if a subjectivist embraces intrapersonal permissivism (as I briefly suggested she should, in section 3). Moreover, we may worry that the standard response just pushes the problem back a step (Arbitrariness Triad #2).

Truly attractive versions of permissivism would not only satisfactorily answer the simple arbitrariness objection in a way that is compatible with intrapersonal permissivism but also somehow make it no arbitrary matter which framework an individual has.

5. An epistemic existentialist response to the simple arbitrariness objection

Let us revisit the premises of the simple arbitrariness objection, this time using the problematic, P1_{Intrapersonal} explicitly, in light of section 3:

The typical permissivist response, i.e. disambiguating the original P1 and committing only to the interpersonal version of permissivism, is no longer available. Let's consider the responses that remain. First, we might reconsider P3. If P3 is true, then agents in the kinds of confusing cases we imagined in section 3 (call these intrapersonally permissive cases) cannot hold rational attitudes simply by taking up some particular rationally recommended attitude. For P3 requires that rational attitudes be ex ante preferable to all other attitudes. And agents in intrapersonally permissive cases face only attitudes that are no better than some alternatives.

One might reject P3 on this basis. One could argue that, since (i) agents in intrapersonally permissive cases can presumably come to hold rational attitudes,Footnote ²⁰ and (ii) (arbitrarily) picking among attitudes unpreferenced by reason would seem to be the only way that an agent could arrive at a particular attitude in an intrapersonally permissive case, such picking must not prohibit rationality. P3 must be too strong.Footnote ²¹

However, first, rejecting P3 should be considered a substantial cost. The appeal of P3 seems to be a tacit point of agreement between uniqueness theorists and permissivists. The standard permissivist way of avoiding the arbitrariness objection isn't to deny that it even matters whether rational attitudes are strictly better than alternatives. Rather, permissivists often simply distance themselves from P1_{Intrapersonal} as opposed to P1_{Interpersonal}. Rejecting P3 seems to amount to accepting a deep arbitrariness about the particular attitudes we hold, even when we are rational.

Second – as I shall argue in the remainder of this section – (ii) is false. Arbitrary attitude picking isn't the only possible response to intrapersonally permissive cases. I suggest the following, alternative possibility: agents’ frameworks may change in the fact of intrapersonally permissive cases – they may be extended or precisified – such that they subsequently recommend a particular, unique attitude. If we accept this possibility, then we may say that when such changes occur (and only when they occur), agents may rationally adopt an attitude. So, I claim, P2 is false. It is not the case that if an agent simply adopts any particular rationally recommended attitude (with the appropriate basing relation), she thereby comes to hold a rational attitude. Rather, agents may only take up a rationally recommended attitude and come to hold a rational attitude if (or once) the attitude is uniquely rationally recommended.

Frameworks are, after all, glossed as the guiding epistemic principles, values, or expectations to which we are committed. Such commitments seem to be the kind of thing that can change – indeed, the kind of thing we, as agents, can shape over time. I will say a bit more in the next subsections about why we might positively expect to be able to alter our frameworks, but first I want to make clear the resolution of the arbitrariness worry that this possibility enables.

This solution is compatible with permissivism because the epistemic existentialist thinks agents really do face situations in which multiple attitudes are rationally recommended, i.e. multiple attitudes really could be rational – if the agent were to adopt subtly different and more determinate frameworks that recommended those attitudes uniquely. However, the solution relies on there being an important distinction between cases in which some attitude is recommended alongside other attitudes, and cases in which that attitude is uniquely recommended. Only in the latter case, the epistemic existentialist claims, can an agent actually rationally adopt the attitude in question. This appeal to the importance of unique recommendation allows the epistemic existentialist's solution to the arbitrariness objection to also accept deep intuitions underlying uniqueness. In this way, the solution may hope for broad appeal.

Yet it also requires an unorthodox view of frameworks. One thesis to which this resolution is already obviously committed:

The other two definitional claims of epistemic existentialism will be needed, in answering the framework arbitrariness objection, Arbitrariness Triad #2.

6. An interlude: Source Voluntarism

One of those claims however, Source Voluntarism, requires some unpacking before we can put it to use. Recall the remaining tenets of epistemic existentialism:

Here I will clarify what this “voluntarist fashion” is meant to signify.

First, it is not a reference to anything like “doxastic voluntarism”, or the claim that we can believe at will. Source Voluntarism is entirely orthogonal to doxastic voluntarism and so is not hostage to the latter's (im)plausibility.

Rather, I mean to invoke the use of voluntarism common among certain ethicists and theorists of practical normativity. In particular, I am interested in the version of voluntarism developed by Ruth Chang (Reference Chang2002, Reference Chang2013a, Reference Chang and Shafer-Landau2013b), which she calls hybrid voluntarism. Chang's hybrid voluntarism is a view about the “source” Footnote ²² of a reason's normative oomph, or what it is in virtue of which something counts as a (normative) reason.

To understand the question that voluntarism attempts to answer, start with the thought that reasons/rationality are normative: if you have reason (or most reason) to do/believe/have/want something, then there is some sense in which you ought to do/believe/have/want it. Now where does that ought, that normative pressure, come from?

You might think it comes from us, in the sense of coming from our (idealized) desires and dispositions (call this view “source internalism”). Or you might think it comes from outside of us, from objective values or norms or brute reason facts (call this view “source externalism”). Voluntarists, on the other hand, think (at least some of) this normative pressure comes from the will. The will and willing are, of course, variously understood. Voluntarists don't have to think this normative pressure always comes from our consciously deciding or intending things. As Chang (Reference Chang2013a: 169) puts it, willing is sometimes “understood not as a conscious, deliberate decision to do something but as the activity of (rational) agency as such”.

Kant famously thought all normativity and all reasons ultimately derive their normativity from the will, from what rational agents are constrained to will.Footnote ²³ But Chang paints a different, more complex picture. After critiquing standard versions of source internalism, source externalism, and voluntarism, she proposes her own “hybrid voluntarism”:

For Chang, then, some of our reasons are given – they are reasons in virtue of the kinds of explanations that source internalists or source externalists might give – and some are based in our wills or agency. She claims these two kinds of reasons interact in a highly structured way:

For Chang, then, will-based or voluntarist reasons “show up” only in hard cases, where all one's given reasons fail to settle an option as best. Such cases might include, e.g., the choice to pursue a career as a lawyer vs. a philosopher (Chang Reference Chang2002: 668); the opportunity to commit in a partner relationship (Chang Reference Chang and Shafer-Landau2013b: 105); or the choice between a cup of tea and a cup of coffee (Chang Reference Chang2002: 669). If we fill in the details suitably, it does seem an agent's standing reasons and preferences might not settle which option is better. But in such situations, according to Chang, one's committing to an option – one's putting one's very agency behind the considerations that make that option more valuable – gives one most reason to choose it.

At one point, perhaps, I had no more reason to be a philosopher than to be a lawyer. But now I've decided to put my will behind, to champion, or to be for, the kinds of freedom and intellectual goods that accompany philosopher-hood. Being a philosopher is right for me. Similarly, at one point I might not have committed to my particular partner; I might plausibly have committed to someone else or not committed at all. But now I do have reason to be in a committed relationship with my particular partner – and to care for him in the particular ways I ought – in virtue of my commitment.Footnote ²⁴

Now Chang's view is far from uncontroversial, and I can't possibly evaluate its prospects thoroughly here. I am trying merely to explain what will-based or voluntarist reasons are, sketch how they might interact with given reasons, and make the briefest gesture at why we should believe they exist. My interest lies in the return to epistemology and Source Voluntarism – what does this claim amount to? It says that at least some reasons for having particular frameworks have their source in the will. In particular, if we ever commit to a particular legitimate framework – e.g., a particular extension or refinement of our own standing framework – we can give ourselves extra reason to have that framework, directly in virtue of that commitment. We can give ourselves extra reason to be, e.g., a more cautious or trusting intellectual agent, by willing or committing to be such.

Now, one might be tempted to extend this even further and claim that all reasons for having particular frameworks are voluntarist. Source Voluntarism is not so ambitious. The more ambitious, explicitly non-hybrid view would face challenges that are familiar from the practical realm. Surely the only reason for me to have my particular framework isn't just that I'm committed to it. (My commitment to my partner isn't my only reason for loving him, either – he's a great person!) Presumably there are some “given” reasons for having certain frameworks – the legitimate ones – rather than others. One has reason not to have a framework, e.g., that licenses the gambler's fallacy, or that celebrates the kind of solipsistic insistence on perceptual evidence that can undergird flat earth conspiracies. One presumably even has some given reasons stemming from one's standing framework; one may have reason to adopt extensions or refinements of that framework, rather than wholesale revisions.

Compatibly with all this, one's given reasons for having particular frameworks may run out. They may not settle which particular framework one should have going forward. And – at least when they do run out in this way – the voluntarist claims we can nonetheless have most reason to have some particular framework, through committing to it.

Source Voluntarism, as I'll explain below, is a crucial part of the epistemic existentialist's response to the second and more challenging level of the arbitrariness objection to permissivism.

7. An epistemic existentialist response to the framework arbitrariness objection

Recall that the epistemic existentialist's response to the simple arbitrariness objection simply required accepting something like Changing Frameworks. In principle, one could accept Changing Frameworks, together with any of a number of stories about how exactly such changes in frameworks occur. Perhaps in intrapersonally permissive cases, our frameworks spontaneously morph to make unique recommendations in ways that have nothing to do with agency, commitment, or choice.Footnote ²⁵ However, in part for the sake of a better resolution to the full scope of the arbitrariness objection (including Arbitrariness Triad #2), I claim that the epistemic existentialist should also accept Agency in Change and Source Voluntarism.

Via Agency in Change, the epistemic existentialist claims that framework alterations of the sort that enable individuals to be rational in the face of intrapersonally permissive cases can be due to the agent herself, her commitment. My framework needn't just morph spontaneously to recommend an attitude uniquely in such a case. Rather, I myself can come down on how best to interpret and evaluate tricky or novel evidence. I can, for example, place a bit more weight on this person's testimony rather than that person's (or weight them equally, or adopt a policy of deferring to some further third party, etc.). I can decide that on some particular kind of question avoiding error matters more than believing truly, or maintaining systematic understanding is not worth some particular sacrifice of expected accuracy. These words: “come down”, “choose”, and “decide”, are admittedly psychologically unrealistic exaggerations. These ‘choices’ needn't rise to the level of consciousness. But they may be attributable doings nonetheless – doings that show up immediately in the particular judgment at hand and subsequently in our enacting these epistemic policies going forward.Footnote ²⁶ Each time I face an intrapersonally permissive case I choose – or else abnegate my ability to choose – a more specific set of epistemic commitments.

This triggers the important consequence claimed in Source Voluntarism. Source Voluntarism, again, claims that such choices or commitments actually generate voluntarist reason to prefer the particular framework chosen. Sure, there is still some sense in which multiple competing frameworks are no better than each other. To use the terminology reviewed above, multiple frameworks may be no better than alternatives with respect to one's given reasons. An agent herself may perhaps recognize this. However, once one actually commits to a particular framework, one has some additional reason to have it, in virtue of that very commitment.

We are now ready to return to Arbitrariness Triad #2:

In response, the epistemic existentialist permissivist will insist on a disambiguation of P4:

Moreover, while she will accept P4_{Given reasons}, she will reject P4_{All reasons}. P4_{Given reasons} is compatible with P5 and P6, since we should take “preferability” in P6 to draw on all reasons, both given and voluntarist. The epistemic existentialist claims that particular, legitimate frameworks can be preferred over others in virtue of having been chosen.

In the following section, I take up the main objection to this epistemic existentialist resolution to the arbitrariness problem: namely, that it too simply pushes the worry “back a step”. How exactly do agents choose which frameworks to commit to? Won't such choices themselves ultimately be arbitrary?

8. Arbitrariness in committing to frameworks?

Let's consider the position of a person whose current framework (F1) is currently amenable to change in two different ways that would result in either of two different frameworks (F2, F3). I am assuming that this “amenability” entails that this person has no more given reason to have F2 than F3, and vice versa. If it were “already”, for given reasons, better for her to have F2 than F3, for example, then F3 would not be on the table.

So, there is certainly a way of characterizing “arbitrariness” in which this person's choice is doomed to be arbitrary: we might just define any choice of framework as arbitrary that is not dictated by given reasons.

But this is not a good characterization of the kind of arbitrariness that is to be avoided at the level of choosing frameworks. First, it is not at all clear why we should insist, for non-arbitrariness, on a framework's being privileged by given reasons in particular. That is, if we accept that there are such things as voluntarist or will-based reasons, why wouldn't a framework's being privileged by those be sufficient to make it non-arbitrarily chosen?

Second, choices among frameworks should not be as easily labeled as problematically arbitrary as choices among doxastic attitudes. The arbitrariness objection outlined in sections 2 and 4 posits two levels of arbitrariness to be avoided. Arbitrariness Triad #1 concerns arbitrariness in the particular attitudes we hold, when we hold epistemically rational attitudes, and Arbitrariness Triad #2 concerns arbitrariness in the particular frameworks we have when we have legitimate frameworks. I have accepted that the only attractive ways out of Arbitrariness Triad #1 will retain P3. I.e., evidence together with a legitimate framework must fully determine a particular rationally recommended attitude, if an agent is to adopt it and count as rational. But there is no plausible analogue of P3 at the level of Arbitrariness Triad #2. There is no prohibition on adopting a policy or set of commitments that is no better than some alternatives, because reasons bearing on which policy to adopt are not sensitive only to strictly “epistemic” reasons.

Epistemic reasons are commonly understood simply as those that bear on (probable) truth of propositions or what one should believe. Talking about “epistemic reasons” for having/adopting frameworks, then, must invoke an extended sense of the term. Granted – sometimes such an extended sense of the term is used, to suggest that there are epistemic reasons to organize scientific organizations in certain inquiry-conducive ways or to institute various true-belief-promoting social policies.Footnote ²⁷ Perhaps in this extended sense we do have epistemic reasons to have legitimate, rather than illegitimate, frameworks. We have epistemic reasons to have frameworks that sufficiently respect or promote true belief and knowledge, for creatures like us. But once we are working with this extended sense of epistemic reasons, they are clearly also amenable to weighing together with other kinds of reasons – moral reasons, economic reasons – to determine the best organization or policy overall. They seem far less austerely exclusive than epistemic reasons for belief are typically thought to be.Footnote ²⁸ As with other questions that are (partly) practical, we should be perfectly able to make rational choices in the face of ties or parity.

Now we might like the epistemic values or whatever epistemic reasons there are (in an extended sense) to uniquely fix the frameworks we should have at all times, such that no other reasons would be needed and no choices would remain. But this seems not to be the case. Rather, just as there seem to be lots of permissible sets of practical values and preferences, there seem to be a number of permissible ways of being committed to interpreting evidence. Rational people who love the truth can become slightly more or less trusting, more or less moved by the values of believing truly vs. avoiding error, etc.

So what is a better characterization of the arbitrariness to be avoided in one's framework, in Arbitrariness Triad #2? I've suggested it isn't merely the “arbitrariness” of not being fully determined by given reasons or epistemic reasons. I think rather the claim to be avoided is this: there is no reason at all an agent has the particular framework she has; she just happens to have it. This is the kind of arbitrariness that is tacitly accepted in standard forms of subjectivist permissivism. This is the kind of arbitrariness that makes it seem we should be indifferent between employing our own (legitimate) framework and taking a pill to change to a different (legitimate) one.

And this is the kind of arbitrariness the epistemic existentialist avoids. According to the epistemic existentialist, agents have the frameworks they have because of the choices they've made about who to be – about the values, methods, and expectations they shall employ as inquirers and opinion-formers. An agent's framework is thus hers in a deeper sense than merely being the framework she happens to hold or the framework associated with her. One's framework at any point in time is partly a function of one's choices; it is an intellectual aspect of the self one is shaping.

Such choices about who-to-be are in general difficult to understand and to model, for they are choices that have the potential to generate the very preferences and values that – in neat, tidy decision-making scenarios – guide decision.Footnote ²⁹ And yet, somehow, we often take such choices seriously. This potential for seriousness is a further sense in which the epistemic existentialist can claim choice among legitimate frameworks is “non-arbitrary”. We can commit to particular ways of judging deliberately, and not by tossing coins. Thus, as reluctant as we may be to do so, carving out space for importantly non-arbitrary choice among options unpreferenced by given reasons (e.g., F2 and F3) seems to be a necessary condition on carving out space for beings like us.

In short, while the epistemic existentialist can admit that picking randomly from among rationally recommended doxastic attitudes would be worryingly arbitrary, she should insist that becoming the sort of person one is – including the sort of intellectual agent one is – is a different matter. Though we may not always have given reasons to go on, we can make choices about who to be in a perfectly good way – with some appreciation of their gravity and in good faith.