Hippocampal and medial prefrontal cortices encode structural task representations following progressive and interleaved training schedules

Memory generalisations may be underpinned by either encoding- or retrieval-based generalisation mechanisms and different training schedules may bias some learners to favour one of these mechanisms over the other. We used a transitive inference task to investigate whether generalisation is influenced by progressive vs randomly interleaved training, and overnight consolidation. On consecutive days, participants learnt pairwise discriminations from two transitive hierarchies before being tested during fMRI. Inference performance was consistently better following progressive training, and for pairs further apart in the transitive hierarchy. BOLD pattern similarity correlated with hierarchical distances in the left hippocampus (HIP) and medial prefrontal cortex (MPFC) following both training schedules. These results are consistent with the use of structural representations that directly encode hierarchical relationships between task features. However, such effects were only observed in the MPFC for recently learnt relationships. Furthermore, the MPFC appeared to maintain structural representations in participants who performed at chance on the inference task. We conclude that humans preferentially employ encoding-based mechanisms to store map-like relational codes that can be used for memory generalisation. These codes are expressed in the HIP and MPFC following both progressive and interleaved training but are not sufficient for accurate inference.Author summary: Integrating information across distinct situations allows both humans and non-human animals to solve novel problems. For instance, by observing that topaz is hard enough to scratch quartz, and that quartz is hard enough to scratch gypsum, one can infer that topaz must be harder than gypsum—even if these materials have never been seen together. This type of generalisation (transitive inference) can be achieved by combing different pieces of information either, 1) when an inference is actually needed (retrieval-based generalisation), or 2) when new information is first encountered (encoding-based generalisation). We predicted that the use of these generalisation mechanisms depends on the order in which information is presented and whether that information was learnt before an overnight rest. Contrary to our predictions, behavioural and neuroimaging analyses of a transitive inference task in humans showed convergent evidence for encoding-based generalisations in all conditions. While these conditions had a large impact on inferential ability, we found that brain regions involved in memory invariably learnt inferred relationships between items that had not been seen together. Strikingly, this appeared to be the case even when participants were unable to make accurate inferences.