Selective Memory Operations Shape History-Dependent Population Representations and Choice under Sensory Ambiguity: A Hippocampal-Inspired Dynamic Gating Model

Abstract How does past experience continue to shape behavior? The hippocampal–entorhinal system supports memory-guided behavior, yet how selective memory operations shape population activity and history-dependent choice remains unclear. We introduce a hippocampal-inspired dynamic gating model (HDGM) that integrates predictive recurrence, online associative memory, and feedback-dependent regulation to examine how past experience shapes population representations and subsequent choice under sensory ambiguity. HDGM exhibited distinct history-conditioned population states despite identical current sensory input. Selective disruptions of encoding and associative processing altered subsequent population activity and impaired history-dependent choice. Output-gain modulation increased linear decoding accuracy for task-relevant classes, whereas cross-trial gate reassignment reduced delay-state separation with little change in mean choice performance. Across complementary tasks, population organization, linear information accessibility, and behavioral choice showed distinct responses to selective memory operations. Together, these findings provide a computational framework for investigating how selective memory operations shape the expression of past experience and its relationship to subsequent choice.