The scope and ambition of Kurzban et al.'s model is commendable. I believe it advances understanding of mental fatigue and task performance and integrates hitherto disparate literatures on mental fatigue, self-control, and vigilance. My comments focus on areas that remain to be elucidated in the model, which I hope will further the debate on the link between mental fatigue and task performance. I will confine my comments to three main areas: (1) automaticity and conscious awareness, (2) the role of individual differences, and (3) resource depletion models of self-control.

First, I felt it was unclear as to the extent to which the processes outlined in the model were automatic and outside the subjective experience of the individual, or whether they were, at least in part, driven by deliberative decision-making. This opens the question as to which kind of cognitive system(s) controls the opportunity-cost computations. The implication is that the computations occur outside conscious awareness and the phenomenology of subjective fatigue is a by-product signalling the effort involved and regulating task performance accordingly. However, the worked example involving “next-best” alternatives to the task at hand (e.g., daydreaming, using a smartphone) implies some conscious awareness of these as viable alternatives. The problem here is the extent to which individuals will have a clear representation of these alternatives if we were, for example, to apply the model to understand persistence and fatigue on tasks presented in laboratory environments where other alternatives are relatively limited (other than the “background” alternative of daydreaming). I found Kurzban et al.'s account relatively silent on this matter, and there were occasions where the authors' narrative implied deliberative decision-making processes (e.g., “We can think of this participant as having a choice [emphasis added] between performing those calculations or, alternately, daydreaming”; sect. 2.4.1, para. 1).

Another illustration lies in the use of the Stroop task as an analogy for the proposed opportunity-cost computations. The performance decrements experienced on incongruent Stroop tasks are due to competition between the visual and word-naming systems leading to a response-inhibiting processing “bottleneck.” This is an automatic process; individuals have no subjective awareness of the interference or control over whether or not their visual system reads the presented words. So, while this competition in processing systems may be the cause of subjective fatigue, it is independent of, and different from, the opportunity cost decision-making process involved in whether to persist with the task or select an alternative. So I think this analogy is problematic, in that it does not elucidate the extent to which the individual consciously deliberates over decisions to persist with the “best” task or allocate resources elsewhere (or not at all), or whether decisions on the devotion of processing capacity are automatic and outside the individual's awareness. Perhaps Kurzban et al.'s account needs to identify the extent to which the computational processes are accessible to the individual. Dual-systems models of social cognition describing the relative contribution of deliberative (reflective) and automatic (impulsive) processing may provide a possible framework (Strack & Deutsch Reference Strack and Deutsch2004).

Second, Kurzban et al.'s account also does not incorporate individual differences. A hallmark of social-cognitive models is the assumption that individuals process information in identical ways. This is not the case if one takes into account individual differences that affect cognitive processing. For example, there is research demonstrating that individual differences in trait self-control moderate effects of mental effort on computationally-demanding tasks (Hagger et al. Reference Hagger, Wood, Stiff and Chatzisarantis2010a). How could the opportunity cost model explain individuals differences in processing bias brought about by such traits? Could it be that traits bias individuals' tendency to interpret the opportunity costs of their responses relative to the next most desirable action, similar to the way experimenter presence is outlined in the model (cf. Figure 3 of the target article)? It would be interesting to incorporate this into the model.

Finally, despite an array of examples in support of the model, including alternative explanations of limited-resource models of self-control, a question remains as to whether all experiences of mental fatigue can be attributed to changes in opportunity cost over time. Kurzban et al. state that “mental ‘resources’ are finite, dynamic, and divisible, … rather than finite and depletable over time” (sect. 2.4.2, para. 1, emphasis theirs). Does this mean that in the absence of “next-best” alternate tasks for which perceived opportunity costs do not exceed those of the “best” task, performance on tasks will be consistent and indefinite? In such cases the marginal utility of the current task would consistently exceed that of the marginal value of the “next-best” task, leading to a decision not to divide processing capacity across the tasks. This seems implausible given research on vigilance tasks that consistently demonstrates fatigue and performance decrements over time. Kurzban et al. concede that “to the extent that there are no offsetting benefits . . . the relationship between perceived costs and benefits can become less favorable over time” (sect. 2.3.2, para. 4), suggesting that, given sufficient time, processing resources will inevitably be allocated elsewhere. However, no alternative explanation is provided in the model to explain fatigue in the absence of a competing alternative task that “wins out” in the decision-making process over the “best” task in opportunity-cost computations.

So is there still room for a “resource” account that provides additional limits on the extent to which processing capacity can be allocated over time? Kurzban et al. point out recent research that has challenged the limited-resource approach. These include studies demonstrating that beliefs (e.g., Job et al. Reference Job, Dweck and Walton2010) and motivational incentives (e.g., Muraven & Slessareva Reference Muraven and Slessareva2003) mitigate ego-depletion, as well as conceptual (Kurzban Reference Kurzban2010a) and empirical (Hagger & Chatzisarantis Reference Hagger and Chatzisarantis2013) accounts that raise doubts over glucose as a physiological analog for the resource. These issues notwithstanding, recent evidence suggests that self-control performance is impaired in the presence of beliefs about resources and motivation, provided the level of depletion is sufficiently extensive (Vohs et al. Reference Vohs, Baumeister and Schmeichel2012). Further, Kurzban et al. acknowledge that further candidate physiological analogs may exist for the limited resource, but conceptual and empirical verification are needed. The ego-depletion literature is problematic; but it seems that in light of new evidence, and the possibility that the opportunity costs model may not provide a comprehensive explanation for mental fatigue, future research should aim to reconcile these differences, perhaps through theoretical integration (Hagger Reference Hagger2009).