Research on the dynamics of cognitive effort have been dominated, over recent decades, by accounts centering on the notion of a limited and depletable “resource” (Baumeister et al. Reference Baumeister, Bratslavsky, Muraven and Tice1998; Baumeister et al. Reference Baumeister, Vohs and Tice2007). Quite recently, however, a trend has emerged, away from resource-based theories and toward accounts centering instead on motivation or value (Hagger et al. Reference Hagger, Wood, Stiff and Chatzisarantis2010a; Inzlicht & Schmeichel Reference Inzlicht and Schmeichel2012; Job et al. Reference Job, Dweck and Walton2010). To paraphrase recent work by Inzlicht and Schmeichel (Reference Inzlicht and Schmeichel2012), the question of interest has begun to shift from whether an individual is capable of exerting cognitive effort to whether the individual will choose to do so.

The target article by Kurzban et al. contributes robustly to this motivational turn. To start, the article offers a penetrating and authoritative critique of the resource model, convincingly asserting both its theoretical and empirical liabilities, and clearing the way for a fresh value-based perspective. Of course, to be satisfying, such a perspective must be specific, indicating precisely how value or motivation constrains cognitive effort. Kurzban and colleagues come through on this front as well, offering a formally explicit, testable theory, framed in terms that place it in continuity with a wealth of recent work on value-based decision making.

We agree with Kurzban et al. that it may be fruitful to view cognitive effort as carrying subjective costs. Having said this, however, we also see at least two problems with the specific proposal the authors put forward, which identifies the costs involved with opportunity costs.

The first problem involves the question of sufficiency: One can think of many situations that feature salient opportunity costs, but that seem unlikely to involve any sense of subjective effort. Imagine, for example, sitting in a restaurant with a friend who is enjoying a dish you wish you had ordered. This scenario involves an awareness of opportunity costs, and perhaps an experience of regret, but no obvious role for effort.

A second problem arises from the theory's explanation for so-called resource-depletion effects: the finding that voluntary effort exertion is diminished following bouts of obligatory exertion. According to the opportunity cost model, such declines occur because, over time, the expected utilities of alternative mental activities rise through learning, ultimately triggering a shift in focus. This account relies on the unfounded assumption that initial value estimates for alternative activities will generally display a negative bias, and implausibly predicts that depletion effects should be isolated to novel task circumstances.

In contrast to the opportunity cost model, we favor a simpler hypothesis: Subjective effort reflects an intrinsic cost attaching directly to the exertion of cognitive control.

The idea that cognitive control carries inherent disutility has arisen as a background assumption in numerous literatures over the years. In recent work, we have been able to undergird this idea with some direct empirical support. Using a variety of choice tasks, we have provided evidence that, when all else is held equal, decision-making displays an avoidance of cognitive control demands, and that people will avoid such demands even at the price of delaying the accomplishment of task goals (Kool et al. Reference Kool, McGuire, Rosen and Botvinick2010). Using functional magnetic resonance imaging (fMRI) we have shown that neural responses to monetary rewards are reduced when such rewards are framed as payment for a cognitively demanding task, consistent with the view that cognitive demand registers as costly (Botvinick et al. Reference Botvinick, Huffstetler and McGuire2009). Further fMRI results show that task-induced activity in cortical regions associated with cognitive control predicts later avoidance of the same task (McGuire & Botvinick Reference McGuire and Botvinick2010).

In very recent work, we have provided behavioral evidence that the cost of control is context sensitive: The more control is exercised, the more costly it becomes (Kool & Botvinick, in press). Rather than arising from resource depletion or fatigue, the data suggest that this effect arises from a set of preferences that favor a balance, over time, between cognitive exertion and cognitive disengagement or rest, an idea that originates in labor economics and which has been fruitfully applied to physical effort.

A view of effort based on the intrinsic cost of cognitive control appears to avoid some of the difficulties of the opportunity cost model. The restaurant scenario introduced above is no longer problematic, since it features no demand for cognitive control, and therefore predicts no sense of effort. (However, effort may arise when the dessert menu arrives, as recent findings suggest that the intrinsic cost of control extends to the exertion of self-control; Kool et al. Reference Kool, McGuire, Wang and Botvinick2013.) The intrinsic-cost perspective also fares better with depletion effects, as the context-sensitivity of control costs predicts that the sustained exertion of control will trigger eventual cost-driven disengagement, even in contexts involving no learning (see Kool & Botvinick, in press).

It is worth noting that the predictions of the intrinsic-cost approach may, in certain cases, mimic those of the opportunity cost model. In particular, the availability of appealing alternative activities may increase demands for cognitive control, in order to maintain focus on the central task. In such situations, effort could be defensibly attributed either to intrinsic control costs or to the registration of opportunity costs. Such considerations indicate that some care will be necessary in designing experiments to test between the relevant theories. However, whatever the empirical challenges, it is encouraging to see specific competing motivational accounts for cognitive effort now emerging.