Computational Models Suggest that Human Memory Judgments Exhibit Interference due to the Use of Distributed Representations

Episodic memory is a core function that allows us to remember the events of our lives. Given that many events in our life contain overlapping elements (e.g., similar people and places), it is critical to understand how well we can remember the specific events of our lives vs. how susceptible we are to interference between similar memories. Several prominent theories converged on the notion that pattern separation in the hippocampus causes it to play a greater role in processes such as recollection, associative memory, and memory for specific details, while distributed representations in the neocortex cause it to play a stronger role in domain-specific memory. We propose that studying memory performance on tasks with targets and similar lures provides a critical testbed for comparing the extent to which human memory is driven by hippocampal pattern separation vs. more distributed representations (e.g., neocortex) vs. a blend thereof. We generated predictions from several computational models and then tested these predictions in a large sample of human participants. We found a linear relationship between memory performance and target-lure pattern similarity within a neural network simulation of area IT, an object-processing region of the brain. We also observed strong effects of test format on performance and consistent relationships between test formats. Altogether, our results were better accounted for by distributed memory models than the pattern-separated representations of the hippocampus; therefore, our results provide important insight into prominent memory theories by suggesting that memory performance is primarily driven by distributed representations (e.g., neocortex).