Evidence for a virtual human analog of a rodent relational memory task: A study of aging and fMRI in young adults

A radial maze concurrent spatial discrimination learning paradigm consisting of two stages was previously designed to assess the flexibility property of relational memory in mice, as a model of human de- clarative memory. Aged mice and young adult mice with damage to the hippocampus, learned accurately Stage 1 of the task which required them to learn a constant reward location in a specific set of arms (i.e., learning phase). In contrast, they were impaired relative to healthy young adult mice in a second stage when faced with rearrangements of the same arms (i.e., flexibility probes). This mnemonic inflexibility in Stage 2 is thought to derive from insufficient relational processing by the hippocampus dur- ing initial learning (Stage 1) which favors stimulus-response learning, a form of procedural learning. This was proposed as a model of the selective declarative and relational memory decline classically described in elderly people. As a first step to examine the validity of this model, we adapted this protocol to humans using a virtual radial-maze. (1) We showed that performance in the flexibility probes in young and older adults positively correlated with performance in a wayfinding task, suggesting that our par- adigm assesses relational memory. (2) We demonstrated that older healthy participants displayed a deficit in the performance of the flexibility probes (Stage 2), similar to the one previously seen in aged mice. This was associ- ated with a decline in the wayfinding task. (3) Our fMRI data in young adults confirmed that hippocampal activation during early discrimination learning in Stage 1 correlated with memory flexibility in Stage 2, whereas caudate nucleus activation in Stage 1 negatively correlated with subse- quent flexibility. By enabling relational memory assessment in mice and humans, our radial-maze paradigm provides a valuable tool for transla- tional research. V C 2011 Wiley-Liss, Inc.