Understanding how individuals respond to mental challenges and why mental effort evokes fatigue and aversion remains a challenge for cognitive science. In the 1960s, attentional resource theories were proposed to account for dual-task interference and linked resources to physiology through the use of measures such as pupillometry (Kahneman & Beatty Reference Kahneman and Beatty1966). These theories fell into disfavor in the 1980s when recognized as largely circular and unable to provide testable hypotheses (Navon Reference Navon1984). Although resource theories continued to play a role in applied psychology (e.g., Wickens Reference Wickens, Parasuraman and Davies1984), cognitive researchers focused on structural explanations for dual-task interference (e.g., Pashler Reference Pashler1994) and largely ignored the subjective aspects of mental effort. At the same time, social psychologists began to develop resource theories to describe “ego depletion” effects on self-control, which ultimately led to the notion that glucose serves as a physical resource for mental effort (Baumeister et al. Reference Baumeister, Bratslavsky, Muraven and Tice1998). However, the physical-resource theory has also turned out to be problematic (Kurzban Reference Kurzban2010a).

Kurzban et al.'s account of subjective effort as an adaptive signal of the opportunity cost of using limited executive control mechanisms offers a new way forward for understanding the psychological and neural mechanisms underlying mental effort. Importantly, the framework proposed in their account does not require depletion of a single resource (physical or attentional) to explain performance declines and subjective effort. The conflicting evidence for a single physical resource, most notably glucose, and consistent neurophysiological evidence for estimation of value and cost in prefrontal networks makes an opportunity cost model of effort particularly compelling. A critical challenge for this account of effort is to formally express how signals for value and cost interact, particularly in choosing to adaptively persist or withdraw effortful behavior. While Kurzban et al. focus on the role of dopamine, we propose that a successful normative account of effort persistence and aversion will require consideration of other brainstem neurotransmitter systems.

Recent proposals of the function of brainstem neurotransmitter systems advocate for their role in signaling useful decision variables. Leading examples include Niv's (Reference Niv2007) proposal that tonic dopamine in the striatum signals average reward rate and Yu and Dayan's (Reference Yu and Dayan2005) proposal that norepinephrine and acetylcholine signal different estimates of uncertainty. Behavioral and neural evidence supports the ability of interconnected brainstem nuclei and executive structures to influence decision-making processes (Aston-Jones & Cohen Reference Aston-Jones and Cohen2005; Kurniawan et al. Reference Kurniawan, Guitart-Masip and Dolan2011). In particular, the pattern of activity and connections of the locus coeruleus-norepinephrine (LC-NE) system suggests a causal role in effortful behavior (Aston-Jones & Cohen Reference Aston-Jones and Cohen2005).

An organism that utilizes an opportunity cost model of effort expenditure requires the ability to rapidly adjust task engagement in response to information from the environment and internal homeostasis monitors. The locus coeruleus (LC) receives input from the anterior cingulate cortex and orbital frontal cortex, structures implicated in the evaluation of cost/benefit trade-offs and valuation, as well as arousal-related inputs from the autonomic nervous system (Aston-Jones & Cohen Reference Aston-Jones and Cohen2005). Additionally, human and animal studies demonstrate that prefrontal networks are sensitive to norepinephrine concentration (Robbins & Arnsten Reference Robbins and Arnsten2009), with optimal levels necessary for successful task performance. This literature advocates for continuous feedback between the LC and cortical structures estimating the utility of maintaining the current effort-allocation policy. Critically, top-down cortical signals and peripheral autonomic input may shift the activity of the LC-NE system in a temporally relevant manner (Aston-Jones & Cohen Reference Aston-Jones and Cohen2005). For example, projections from the anterior cingulate cortex may shift the firing rate of the noradrenergic neuron population, in turn altering the level of norepinephrine in the cortex, which decreases stability of the current effort policy and promotes disengagement and selection of a new action plan (Aston-Jones & Cohen Reference Aston-Jones and Cohen2005; Sara & Bouret Reference Sara and Bouret2012).

A system that adaptively shifts among action contingencies, as proposed in prominent theories of LC-NE system function, is central to an opportunity cost model of effort. Theories of LC-NE function broadly conceptualize its activity as shifting the balance of exploratory versus exploitative behavior or mediating a global signal to reset brain networks involved in action selection (Aston-Jones & Cohen; Sara & Bouret Reference Sara and Bouret2012). Examining LC-NE system activity in humans is difficult, due to the small size of the nucleus, its brainstem location and the feasibility of assessing cortical levels of norepinephrine in vivo. However, several functional neuroimaging studies have described patterns of activity in LC. An early study described patterns of activation in a putative LC region and right lateralized prefrontal regions that appear to respond parametrically to task difficulty (Raizada & Poldrack, Reference Raizada and Poldrack2007). Although consistent with a connection between LC and lateral prefrontal self-control networks, the study lacked the spatial specificity necessary to attribute a specific role to the LC. A subsequent study claimed to pharmacologically modulate LC activity (Minzenberg et al. Reference Minzenberg, Watrous, Yoon, Ursu and Carter2008) but faced similar scrutiny about the precision of LC localization (Astafiev et al. Reference Astafiev, Snyder, Shulman and Corbetta2010). Recently, a group applied improved brainstem spatial alignment to conclude that activity in LC correlates with unexpected uncertainty in a decision-making task (Payzan-LeNestour et al. Reference Payzan-LeNestour, Dunne, Bossaerts and O'Doherty2013), consistent with a theoretical model (Yu & Dayan Reference Yu and Dayan2005).

Assessing the LC-NE system in humans remains a challenge, but recent studies point to a possible alternative solution. Several groups have demonstrated the utility of peripheral neurophysiological measurements, notably changes in pupil diameter, as an index of LC-NE system activity. As classically described by Kahneman (Reference Kahneman1973) and revived by Jepma and Nieuwenhuis (Reference Jepma and Nieuwenhuis2011), Nassar et al. (Reference Nassar, Rumsey, Wilson, Parikh, Heasly and Gold2012), Eldar et al. (Reference Eldar, Cohen and Niv2013) and others, changes in pupil size appear linked to the noradrenergic arousal system and related to decision variables such as novelty and uncertainty that are useful for a system estimating opportunity costs to control effort-allocation policy. As Kurzban and colleagues note, a normative account of effort will benefit from unification of executive and self-control literature. We propose that validation of peripheral measurements of LC-NE activity and their integration with effortful tasks constitutes a worthwhile approach to test Kurzban et al.'s opportunity cost model. Evaluation of LC-NE activity in effort contingency, trade-off, and performance tasks will provide key evidence to support or refute particular mechanisms by which valuation and control systems interact to shift behavior in accordance with an opportunity cost model. Together with parallel investigations of other neuromodulatory systems, this work will provide the quantitative framework that a normative model of effort requires.