Predictions and declarative memory encoding: two fMRI paradigms provide slim pickings for SLIMM.

Memory is often better for both highly unexpected and highly expected information. To explain this U-shape relationship between memory and expectancy, the SLIMM framework (schema-linked interactions between medial prefrontal and medial temporal regions) appeals to two memory systems: one involving the hippocampus and another involving the medial prefrontal cortex (MPFC). Specifically, SLIMM predicts that these two brain regions support opposite ends of the U-shape, with the hippocampus supporting memory for unexpected events and the MPFC underpinning memory for expected events. We tested these predictions in two functional magnetic resonance imaging (fMRI) paradigms: one examined memory for the location of objects within scenes (e.g. a toothbrush in a bathroom, either on a sink, as expected, or on top of a bin, as not expected); the second tested memory for words that completed well known idioms either in an expected or unexpected fashion (e.g. 'Don't count your chickens before they … hatch/cook'). Both paradigms replicated the predicted U-shaped relationship between expectancy and memory performance. However, neither study found the predicted interaction between expectancy and subsequent memory (remembered versus forgotten trials) in terms of fMRI activation at encoding in the a priori defined hippocampal and MPFC regions of interest. We discuss the implications of these findings for SLIMM, and how our results fit within the broader literature on expectancy and memory. This article is part of the theme issue 'The role of hippocampal predictions in cognition: bridging perception and memory'.