In one line of research, lab co-director Todd Braver and collegues have suggested that PFC actively maintains representations of context information in working memory (Braver, Cohen and Servan-Schreiber, 1995; Cohen, Braver and O’Reilly, 1996; O’Reilly, Braver, and Cohen, 1999). These context representations serve as a “top-down” source of excitatory activation that biases the local competitive interactions between mutually incompatible thoughts or actions (e.g., between two alternative response options or between two alternative perceptual interpretations of a stimulus). These internally maintained context representations are especially critical in situations in which a dominant, but inappropriate, response must be inhibited (e.g., the Stroop task).
Context representations can be thought of as being similar to goal representations (e.g., task instructions), but may also represent more specific information (e.g., the information provided by a previous stimulus). This theory has motivated a series of functional neuroimaging studies (using fMRI methods). These studies have demonstrated a consistent pattern of PFC activity during performance of working memory tasks that suggests a role in actively maintaining context (Cohen et al., 1994; Cohen et al., 1997; Braver et al., 1997; Barch, Braver et al., 1997).
Currently, the CCP Lab has been testing how PFC activation is affected by the specific mode of cognitive control that is dominant (proactive vs. reactive, seeCognitive Control). For example, A recent study has suggested that control mode significantly affects PFC activity dynamics during task-switching and working memory paradigms (Speer, Jacoby and Braver, submitted; Braver, Reynolds, and Donaldson, submitted).
Another current line of research examines PFC organization. In particular, we hypothesize that frequently updated information may be represented in particular regions of PFC (e.g., posterior, dorsolateral areas), while information that must be maintained over longer of spans and/or protected from intermittent updating and interference, is represented elsewhere (e.g., anteriorly, in frontopolar regions). An implication of this hypothesis is that some PFC areas (e.g., posterior) will represent more specific forms of information (e.g., stimulus features or particular responses), while other areas (e.g., anterior) will represent more abstract information (e.g., featural dimensions, or more complex rules). This is because representations that must endure are more likely to be abstract (e.g., task instructions or high level goals), as compared to those updated more frequently (e.g., subroutines, or specific stimuli appearing in a given trial. Initial support for this hypothesis has come from a neuroimaging study, in which anterior PFC activity was found selectively during performance of a task which requires the maintenance of higher-order task goals while simultaneously carrying out subgoal processing (Braver & Bongiolatti, 2002).