2025/02/16 by de Cheveigné, Alain
#FOS: Biological sciences #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Machine Learning (cs.LG) #Neurons and Cognition (q-bio.NC)
paper · doi:10.48550/arxiv.2502.11105
A rational framework is proposed to explain how we accommodate unbounded sensory input within bounded memory. Memory is stored as statistics organized into structures that are repeatedly summarized and compressed to make room for new input. Repeated summarization requires an intensive ongoing process guided by heuristics that help optimize the memory for future needs. Sensory input is rapidly encoded as simple statistics that are progressively elaborated into more abstract constructs. This framework differs from previous accounts of memory by reliance on statistics as a representation of memory, the use of heuristics to guide the choice of statistics at each summarization step, and the hypothesis of a process that is complex and expensive. The framework is intended as an aid to make sense of our extensive knowledge of memory, and bring us closer to an understanding of memory in functional and mechanistic terms.