Pattern Separation Underpins Expectation-Modulated Memory.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Memory Research

Background:

• The hippocampus is crucial for memory encoding and retrieval through pattern separation and completion.
• The influence of top-down processes, like expectation, on these hippocampal computations is not fully understood.

Purpose of the Study:

• To investigate how expectation modulates hippocampal pattern separation.
• To determine if expectation violation specifically enhances memory for similar items.

Main Methods:

• A behavioral experiment with 29 participants performing a recognition task with expected and unexpected cues.
• A computational model of the hippocampus simulating the behavioral task.
• Representational similarity analysis to examine neural representations in hippocampal subfields.

Main Results:

• Memory accuracy improved for highly similar foils that were unexpected during encoding.
• No significant differences were observed for targets or low-similarity foils based on expectation.
• Computational models showed increased representational dissimilarity in DG and CA3 for unexpected items, particularly with moderate to high input similarity.

Conclusions:

• Expectation violation enhances pattern separation in the dentate gyrus (DG) and CA3.
• This mechanism selectively improves memory disambiguation for similar, unexpected information.
• These findings highlight the role of pattern separation in memory for unexpected events.