Sex differences in implicit punishment sensitivity: Evidence from two cognitive paradigms

Abstract

Men, relative to women, are more likely to engage in disinhibited behaviors – such as excessive alcohol consumption or violent crimes – that implicate lower levels of punishment sensitivity. The goal of the present two studies (N = 207) was to implicitly model punishment sensitivity in terms consistent with a process-based view of this construct. Study 1 found that women slowed down following error feedback in a cognitive task to a greater extent than men. Study 2 found that women, but not men, altered their predictions following error feedback in a purported precognition task. Results are discussed in relation to theories of sex differences, punishment sensitivity processes, and proneness to disinhibited behaviors.

Introduction

Sex differences in cognition and behavior were largely ignored prior to a classic review by Maccoby and Jacklin (1974). Since then, an increased focus on sex differences has occurred. It has since become apparent that one of the most prominent sex differences involves men’s greater tendency to engage in disinhibited behaviors of multiple types. Men commit 90% of homicides (Daly & Wilson, 1988) and commit the vast majority of other serious crimes as well (Bennett, Farrington, & Huesmann, 2004). Men are more likely to use and abuse alcohol and other drugs of a psychoactive nature (Walitzer & Dearing, 2006). Moreover, men desire and seek to instantiate sexual relations with barely known partners to a much greater extent than women (Buss & Schmitt, 1993). What appears to emerge from multiple literatures, then, is a robust sex difference in proneness toward disinhibited behavior, which appears more characteristic of men.

Gray, 1987, Gray, 1990 theory of disinhibited behavior has been especially influential to the personality and clinical literatures (e.g., Smillie, Pickering, & Jackson, 2006). Gray suggested that disinhibited behavior primarily, though not exclusively, involves low levels of punishment sensitivity – that is, a relative insensitivity to feedback indicating that behavioral tendencies are problematic (Fowles, 2006). Although this model has typically emphasized learning processes, it is also clear that reactivity to punishment feedback, in more proximal terms, is relevant to the theory (Patterson & Newman, 1993). Indeed, there are multiple sources of support for the idea that disinhibited individuals are less likely to alter their behaviors following punishment signals from the environment (Blair, 2004), though such results typically involve clinical populations.

From a self-regulation perspective, sensitivity to punishment signals is crucial. Self-regulation is conceptualized in terms of over-riding behavioral tendencies that are problematic and error-prone (Baumeister, Heatherton, & Tice, 1994). To the extent that individuals are punishment-insensitive, then, they are likely to repeat behaviors that have been punished in the past (Patterson & Newman, 1993). In more proximal terms, such individuals are likely to be non-responsive to error feedback and to fail to recruit regions of the frontal lobes responsible for mitigating problematic processing routines (Holroyd & Coles, 2002). For example, Kerns et al. (2004) found that individuals whose brains were more sensitive to error-proneness in a cognitive task were more capable of self-regulating problematic outcomes on the next trial. In sum, sensitivity to punishment feedback can be viewed as somewhat critical for modifying behavioral tendencies that are error-prone and disinhibited (Barkley, 1997).

Men’s greater proneness to disinhibited behavior of multiple types (reviewed above) might be understood in terms of their lesser tendencies to adjust behavior following punishment feedback. For example, problematic alcohol use engenders negative consequences related to lower levels of achievement, hangovers, missed meetings, and so on. The fact that men are more prone to binge drinking, alcohol dependence, and indeed to alcoholism (Walitzer and Dearing, 2006, Wilsnack et al., 2002) can therefore be plausibly interpreted as suggestive that heavy-drinking men (relative to women) continue their drinking habits despite evidence of their problematic nature.

Following Gray, 1987, Gray, 1990 influential theory, at least two prominent self-report measures of punishment sensitivity have been developed (Carver and White, 1994, Torrubia et al., 2001). It is now apparent that women score higher than men on such scales (Jorm et al., 1999, Li et al., 2007). However, Gray’s (1987) theory is a processing theory that would benefit from process-based assessments rather than self-reported ones. This is particularly true because reactivity to punishment feedback has been linked to brain structures such as the amygdala that mediate unconscious threat registration processes, which would not be introspectively available to the individual self-reporting such tendencies (e.g., LeDoux, 1996).

We therefore assessed potential sex differences in punishment sensitivity in implicit cognitive terms. In Study 1, we modeled such processes in terms of tendencies to slow down following error feedback. Although tendencies to slow down following error feedback have long been observed in normative terms (Rabbitt, 1968), there are preciously few studies linking individual differences in such tendencies to personality variables posited to involve lesser levels of punishment sensitivity (for a recent exception, see Wilkowski & Robinson, 2008).

A different implicit assessment paradigm was used in Study 2. A recent development in the decision-making literature suggests that punishment sensitivity processes can be modeled in terms of altering choices that have been previously punished (Bechara, Damasio, Tranel, & Damasio, 1997). In Study 2, we created a paradigm of this type in which tendencies to alter behavioral predictions following error feedback were measured. Studies 1 and 2 therefore cognitively model punishment sensitivity in terms of both RT (Study 1) and choice behaviors (Study 2). Convergence of results across these very different paradigms would therefore be particularly impressive.

We suggest that the implicit cognitive paradigms developed here may have general value to the personality-processing literature. Of primary importance, though, was to determine whether there is a sex difference in implicit punishment sensitivity that could perhaps account for men’s molar tendencies toward disinhibited behavior. If so, men should slow down following error feedback to a lesser extent than women in the Study 1 task. Further, men should be less likely to alter behavioral predictions following error feedback in the Study 2 task.