How Does the Brain Keep Information in Mind?

For decades, theoreticians and experimentalists have worked under the assumption that the short-term retention of information depends on elevated activity in the neural elements that represent this information. This was reinforced by the observation of elevated activity in the prefrontal cortex (PFC) during the performance of working memory tasks. Increasingly, however, researchers are realizing that the face-value interpretation of regional patterns of elevated activity, including in the PFC, is constrained, a priori, to the principle that brain functions (like working memory) are organized in a modular manner. This is at odds with our current understanding of the brain as a highly distributed, dynamical system capable of multiplexed processing. An alternative empirical approach, the application of multivariate pattern analysis to neural data sets, has led to two important developments in our conceptualization of working memory. First, it has shown that sustained delay-period activity in the PFC (and elsewhere) often does not correspond to the active retention, per se, of information. Second, it has provided strong empirical evidence for two mechanisms that may underlie our ability to keep in mind small amounts of information with which we can guide our behavior: sensorimotor recruitment, and the temporary activation of long-term memory representations.