Will, Anterior Cingulate Cortex, and Addiction

See allHide authors and affiliations

Science  31 May 2002:
Vol. 296, Issue 5573, pp. 1623-1624
DOI: 10.1126/science.1072997

To survive, maximize benefit, and minimize harm, individuals need to select actions that optimize attainment of adaptive goals. Often, and particularly in cases where a long and possibly uncertain sequence of actions is required to attain a goal, it is necessary to exert “willed” control over selection of the appropriate behaviors (1). New clues about the neural mechanisms mediating this type of behavioral control are described by Shidara and colleagues on page 1709 of this issue (2). Using visually cued multitrial reward schedules in monkeys, they identify a group of neurons in the anterior cingulate cortex (ACC) of monkey brain that encode reward expectancy and may be crucial for boosting actions that will improve the odds of attaining the reward. Delineating such neural mechanisms will help us to understand the determinants of both normal and pathological patterns of action in humans.

The ACC is one of numerous interconnected brain structures that mediate willed control of actions. More specifically, the ACC contributes to the generation of emotional states (3) and to the executive control of the influence of those states on behavior selection (37). Consistent with these observations, electrophysiological studies of the ACC (primarily Brodmann area 24C) show that single neurons in this region respond to and differentiate between (i) primary appetitive and aversive events such as food and shock, (ii) rewards of varying magnitude, and (iii) stimuli and actions that differentially predict appetitive and aversive events (811). These data demonstrate that neural signals in the ACC encode motivational aspects of events along a good-bad (and perhaps a better-worse) continuum.

Although consistent with human studies of the ACC, electrophysiological recording from animal brains demonstrates that ACC activity may also encode the degree of reward expectancy. Shidara and colleagues have conducted a test of this hypothesis. In their study, monkeys received a reward contingent on completing a schedule of visual color discrimination trials. The schedule included a variable number of trials. At the onset of each trial, animals were presented with additional light cues. Increasing brightness of these cues signaled advancement through the schedule and hence an increasing likelihood that the animals would ultimately attain the reward. The light cues were the only information available about the schedule. The monkeys made progressively fewer errors as the rewarded trial approached, showing that the cue regulated the behavior of the animals. A subset of ACC neurons fired more rapidly in response to these predictive cues. Importantly, the neural responses were graded, such that the magnitude of the change in firing increased according to cues that signaled a progressively greater likelihood that reward would ultimately be attained. These changes in neural activity completely disappeared when the relation between the cue and the reward was randomized. The differential response of individual ACC neurons to identical cues presented during the predictive and randomized sessions suggested that graded changes in firing during the predictive sessions encode the degree of reward anticipation. Significantly, this neural encoding of anticipation occurred at a time when it could influence future actions and in conjunction with cues that did in fact modulate behavior.

The degree of reward expectation can impact emotion and willed control of behavior. For example, and perhaps relevant to the Shidara et al. study, increases in reward expectancy can facilitate persistence in a course of action despite the interim failure to receive a reward. The cue-related firing patterns that Shidara et al. describe in the ACC (1214) imply that the degree of reward expectancy is part of the reward-related information that determines the ACC's contribution to executive control of emotion and behavior. The degree of reward expectancy is influenced by multiple factors, including the probability and imminence of a reward (12, 13, 15). It will be instructive to determine whether the ACC is differentially influenced by one or more of these factors, or is instead sensitive to the overall level of reward anticipation (16). There is reason to suspect that the latter is true. The ACC is engaged during many types of tasks involving processing of stimulus information in relation to an array of motivational events and actions (47). These observations suggest a global and integrated contribution of the ACC to motivational influences on behavior.

Consistent with this role, abnormalities in the ACC (and other frontal regions such as the orbitofrontal cortex) have been implicated in a range of disorders involving disturbances in emotions and actions, including obsessive-compulsive disorder, post-traumatic stress disorder, depression, and mania (47, 1722). Although not commonly cited in the literature, researchers including Shidara et al. have suggested that abnormalities in the ACC and interconnected regions contribute to drug addiction. Addicted individuals exhibit symptoms associated with insults to the ACC (and related regions), including anhedonia (absence of pleasure) and an inability to make adaptive decisions regarding future actions (21, 22). In fact, a hallmark of drug addiction is compulsive and uncontrollable drug use, despite knowledge of adverse consequences (23). Consistent with this symptom profile, neuroimaging studies show hypoactivity and low cell density in the ACC (see the figure) and associated structures of addicted individuals (21, 22, 24, 25). It is thus possible that addicted individuals, like others who suffer from hypoactivity in these brain regions, are unable to experience normal affective responses to future events or to exert willed control (1) over actions that maximize benefit and minimize harm.

When rewards go awry.

The ACC (Brodmann areas 32 and 24) of cocaine-addicted patients has striking defects in both blood flow (perfusion) (A) and gray matter density (B) compared with controls (23, 24). Resting hypoperfusion was assessed was accessed by positron emission tomography with 15O water as the blood-flow tracer; regions of reduced gray matter density were analyzed by voxel-based morphometry (27).


The convergence of addiction research and clinical and functional studies of brain regions such as the ACC has implications for the treatment of addicts and for related public health policies (26). The merging of these research areas also illuminates new lines of investigation that may enhance our understanding of both adaptive and pathological regulation of actions. For example, in light of the Shidara et al. data, it will be important to test whether addicted individuals suffer from multiple deficits in emotion and executive control of actions.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article