Extracting patterns across memories to guide decisions (360G-Wellcome-204277_Z_16_Z)

How do we use past experience to guide behaviour and how does this process break down in specific patient populations? I will develop a novel experimental-computational tool to understand how individuals transform memories of previous events into generalised models of the world that allow them to behave appropriately in the future. For past experience to guide behaviour, we need to extract generalities or patterns across a set of events. Pattern extraction allows us to predict what will happen in novel but related situations. I will develop a new experimental technique in healthy humans (based on rodent research), where participants are required to extract patterns across a set of past events in order to make accurate predictions about future events. I will build computational models to predict and explain behaviour at the level of individual people. The models will assess what pattern a participant has extracted, and how this differs from the pattern that would optimise future predictions. Critically, when an individual is not behaving appropriately, the experimental-computational technique will reveal what is impaired and why. In the future, this technique will be used as a microscope to pinpoint impairments in memory-guided decision-making in patient populations.