abstract Andrea Greve

Andrea Greve (University of Cambridge, U.K.)

Successes and failures in learning of novel associations.

There is compelling evidence that the Hippocampus (HC) promotes learning of novel associations. Current memory theories posit that detailed episodic information is rapidly acquired by the Hippocampus and gradually integrated into long-term memories via the neocortex. Contrary to this notion, a recent study by Sharon et al. (2011) reported that four amnesic patients, who suffered from profound hippocampal damage, maintained the capacity to learn novel word-picture association when trained with a paradigm that encourages ‘fast mapping’ (FM). Hence, this FM technique might offer an alternative route for learning novel information in populations experiencing memory problems. We’ve examined this prospect in healthy ageing and I’ll present behavioural and volumetric MRI data from young and older participants who completed the FM procedure. I’ll discuss how the obtained pattern of results might substantiate competing memory theories and introduce the recently proposed Predictive Interactive Multiple Memory Systems (PIMMS) framework. PIMMS suggests that unexpected events (violations of prior predictions) play a key role in memory encoding. Finally, I will present a series of behavioural investigations which support the PIMMS account by demonstrating that prediction errors (PE) enhance learning of novel associations and conclude my talk by considering potential alternative explanations.