McKay & Dennett (M&D) ask if there are adaptive misbeliefs. The positive illusions they discuss in section 13 of the target article are plausible candidates for propositional attitudes that are adaptive irrespective of their truth, but I want to question whether they are really beliefs. The authors adopt a broad definition of belief as a functional state that “implements or embodies” (sect. 1, para. 1) the endorsement of a state of affairs as actual. However, this glosses over a distinction often made between belief and acceptance (e.g., Bratman Reference Bratman1992; Cohen Reference Cohen1989; Reference Cohen1992; Engel Reference Engel1998; Reference Engel2000; Frankish Reference Frankish2004; Stalnaker Reference Stalnaker1984), and it may be that some putative adaptive misbeliefs are better classified as judicious pragmatic acceptances.

The belief/acceptance distinction is drawn in slightly different ways by different writers, but a central claim is that belief is involuntary and acceptance voluntary. To accept a proposition is to adopt a policy of treating it as true (taking it as a premise) for the purposes of reasoning and decision making. Now acceptance can be motivated by epistemic reasons, and when it is it can be regarded as a form of belief. (I would argue that all-or-nothing belief, mentioned by the authors in note 2, is a truth-directed form of acceptance; see Frankish Reference Frankish2004; Reference Frankish, Huber and Schmidt-Petri2009.) However, we can also accept things for non-epistemic reasons – ethical, professional, religious, and so on. For example, loyalty may require a person to accept that their friend is telling the truth, and professional ethics may oblige a lawyer to accept that their client is innocent, even if they do not believe these things (Cohen Reference Cohen1992). Acceptance can also be prudential, designed to simplify complex deliberations or handle error-management considerations of the sort discussed in section 9 of the target article (Bratman Reference Bratman1992). (M&D suggest that such considerations need not motivate belief, but only a cautious action policy. The present suggestion, however, is that they may prompt the formation of a deliberative policy, which constitutes a type of propositional attitude.) I shall refer to acceptance that is motivated by non-epistemic concerns as pragmatic acceptance.

With the notion of pragmatic acceptance in place, what should we say about the unrealistically positive self-appraisals identified by Taylor and her colleagues (e.g., Taylor Reference Taylor1989; Taylor & Brown Reference Taylor and Brown1988)? Are these genuine misbeliefs or pragmatic acceptances, motivated perhaps by a sense of their therapeutic value or a desire to maintain a comforting self-image? The distinction between belief and acceptance is often overlooked, so it is not enough to note that these attitudes are typically classified as beliefs. Nor would it be sufficient to detect the influence of pragmatic motives in their formation, since these may be operative in both cases – illicitly in one case, legitimately in the other. In short, how can we tell the difference between beneficial misbelief and judicious pragmatic acceptance?

One way is by considering subjects' attitudes to their self-appraisals. In particular, do they feel they have control over these judgements and do they think it is legitimate to allow non-epistemic factors to influence them? If so, this would suggest that their attitude is one of pragmatic acceptance rather than belief. There is some evidence that this is the case. Everyday wisdom says it is beneficial to adopt a positive outlook – to think positively, be optimistic, and have confidence in oneself – and we often strive to take this advice to heart. Moreover, we do so without feeling that we are thereby violating epistemic norms, even if we have no evidence to support the views adopted. This is not decisive, however. Our control here may be only indirect, and some self-deception may be involved.

A second consideration is the deliberative context in which our positive illusions are active. Pragmatic acceptance, unlike belief, is context-dependent. More specifically, our beliefs (including truth-directed acceptances) guide us in an open-ended range of deliberations, including, crucially, ones where we want to be guided only by the truth – truth-critical deliberations (Frankish Reference Frankish2004). Our beliefs are our best bets at truth, and they are what we rely on when we want to rely on the truth. Pragmatic acceptances, on the other hand, are operative only in contexts where non-epistemic values, such as loyalty or professional ethics, matter to us more than truth, or where we prefer to err on the side of caution. (Note that this means that pragmatic acceptance requires the ability to classify deliberations as truth-critical or not, and hence requires metacognitive abilities.)

The question, then, is whether we rely on our optimistic self-appraisals in truth-critical contexts. Is there evidence for this? It might be replied that people report their self-appraisals with apparent sincerity, and that sincere reports are the product of truth-critical deliberation. This is too swift, however. If positive self-appraisals have considerable therapeutic value, then that would be a reason for us to treat deliberations about ourselves as not truth-critical, unless there is a lot at stake. Moreover, this may go for what we tell ourselves as much as for what we tell others; there need not be any conscious insincerity involved.

Experiment should help here. For example, we might ask subjects to form assessments of their own abilities and attributes, offering varying rewards for accuracy. (It would not matter if accuracy could not easily be determined, provided subjects thought it could.) If a subject revised or abandoned an assessment as the rewards – and thus the truth-criticality of the context – increased, this would suggest it was an object of pragmatic acceptance rather than belief.Footnote ¹ Some pragmatic acceptances, however, will be hard to detect. Truth-criticality is determined by the subject's priorities; and, in general, the stronger a person's pragmatic reasons for accepting a certain claim, the harder it will be to create conditions under which they will treat deliberations involving it as truth-critical. Indeed, at the extreme, they may treat none as such, rendering their attitude functionally equivalent to belief.

Despite these practical difficulties in applying the distinction between belief and pragmatic acceptance, it is important to keep the distinction in mind when theorizing about adaptive misrepresentations. For one thing, it suggests there are distinct routes to the formation of such attitudes – one involving the overriding of barriers to the influence of motivational processes on belief formation (the breaking of what M&D call “doxastic shear pins”), the other involving mechanisms of pragmatic acceptance in which such barriers are not present. There may also be differences in the psychological and physiological effects of optimistic self-appraisals depending on whether they are pragmatically accepted or genuinely believed. This again may be a matter for experiment.