By Alex Gordienko in Research — Nov 15, 2026

Curriculum Effects in Associative Inference: How Interleaving Episodes Promotes Integrated Representations and Blurs Memory

My nanosymposium talk at SfN 2026 in Washington, D.C., on Sunday, Nov 15 at 2:00 PM in WCC Room 145.

Abstract

Memory theorists have debated to what extent and under what conditions memories of distinct experiences become integrated. Further, when integration does occur, theories differ as to whether this integration comes at the expense of fidelity of the memories for the individual experiences. Prior work has found that interleaved learning of object pairs (AB and BC) leads to integrated representations of distinct items (A and C) akin to those of a canonical neural network model employing distributed representations and exposed to information in interleaved order. We have proposed in the C-HORSE model of the hippocampus that this kind of integration may be happening along the monosynaptic pathway, in contrast to the trisynaptic pathway which maintains pattern separated representations that are curriculum agnostic. Such models would also predict a lower fidelity representation of the constituent individual memories relative to a model specialized for episodic encoding.

Here we sought to replicate the prior findings and test whether constituent memories do show this blurring under conditions in which episodic retrieval is unlikely, using a paradigm where memory for perceptual details of the direct memories can be easily assessed. In a behavioral experiment, 34 participants learned chains of associate pairs composed of novel objects that shared overlapping items but with altered visual features. Specifically, participants learned that object A was related to B and that B’ was related to C, where B and B’ were the same object but viewed from a different angle. Eight pairs were learned in an interleaved curriculum and eight in a blocked curriculum. Memory for direct pairs (AB), adjacent pairs (AB’), and indirect pairs (AC) was then tested in a speeded recognition task and two non-speeded explicit inference tasks.

We found that interleaved learning led to integrated representations of indirect associates (AC), replicating existing findings. Furthermore, we found that increased false alarms for unseen but perceptually similar associates (AB’) in the interleaved condition suggested generalized memory for associated objects, providing additional support that interleaving promotes integration. These behavioral findings are consistent with accounts of memory integration based on interleaved learning in neural network models employing distributed representations. Forthcoming fMRI work aims to illuminate the neural underpinnings of these curriculum effects and characterize the nature of these integrated representations both immediately following learning and after a one-week delay.

Alex Gordienko & Anna C. Schapiro, Department of Psychology, University of Pennsylvania. Presented in nanosymposium NANO15, Behavioral and Neural Insights Into Representational Learning. Read the abstract on the Neuroscience 2026 meeting planner.

This work was funded by National Institutes of Health grant R01 MH129436.

Curriculum Effects in Associative Inference: How Interleaving Episodes Promotes Integrated Representations and Blurs Memory

Abstract

More from the lab at SfN 2026

The emergence of memory systems through meta-learning

A mixture-of-experts model of hippocampal-mPFC interactions