COMMENTARY Homeostatic Regulation of Memory Systems and Adaptive Decisions

ABSTRACT: While it is clear that many brain areas process mnemonic information, understanding how their interactions result in continuously adaptive behaviors has been a challenge. A homeostatic-regulated predic-tion model of memory is presented that considers the existence of a single memory system that is based on a multilevel coordinated and integrated network (from cells to neural systems) that determines the extent to which events and outcomes occur as predicted. The “multiple memory systems of the brain ” have in common output that signals errors in the prediction of events and=or their outcomes, although these signals differ in terms of what the error signal represents (e.g., hippocampus: context prediction errors vs. midbrain=striatum: reward prediction errors). The prefrontal cor-tex likely plays a pivotal role in the coordination of prediction analysis within and across prediction brain areas. By virtue of its widespread con-trol and influence, and intrinsic working memory mechanisms. Thus, the prefrontal cortex supports the flexible processing needed to generate adaptive behaviors and predict future outcomes. It is proposed that pre-frontal cortex continually and automatically produces adaptive responses according to homeostatic regulatory principles: prefrontal cortex may serve as a controller that is intrinsically driven to maintain in prediction areas an experience-dependent firing rate set point that ensures adaptive temporally and spatially resolved neural responses to future prediction errors. This same drive by prefrontal cortex may also restore set point firing rates after deviations (i.e. prediction errors) are detected. In this way, pre-frontal cortex contributes to reducing uncertainty in prediction systems. An emergent outcome of this homeostatic view may be the flexible and adaptive control that prefrontal cortex is known to implement (i.e. working memory) in the most challenging of situations. Compromise to any of the prediction circuits should result in rigid and suboptimal decision making and memory as seen in addiction and neurological disease. VC 2013 Th