Virtual fMRI brown bag : August 14
Please join us for a talk given by Anna Leshinskaya, a postdoctoral fellow in Charan Ranganath's Dynamic Memory Lab at UC - Davis.
How learned relations influence neural responses to objects and events
Our knowledge of everyday objects goes beyond recognizing their visible physical features; it also includes more 'functional' knowledge such as that plants need water and that telephones allow communication. In my work, I argue that we learn properties like these by inferring relations among events, and I investigate the neural mechanisms that allow us to acquire such relational knowledge and represent it in long term memory. In Experiment 1, I query the nature of tool-selective neural responses that have been long demonstrated in lateral temporo-occipital cortex (LOTC). I find that tool-selective LOTC, which responds preferentially to real-world tools (e.g., hammers, saws vs clocks, shoes), also responds to novel objects that cause effects on the environment, but not to novel objects that do not cause effects (rather, react to them). This finding suggests that the structure of event relations associated with an object can influence the response of category-selective neural areas to those objects. Thus, event relations are an important a factor in the neural organization of object knowledge. In Experiment 2, I ask how the brain balances the need to represent the visual features of stimuli vs their predictive relations -- factors that can often be at odds. For example, we can recall that a specific switch is associated with a lamp turning on, but simultaneously know that the lamp does not resemble the switch (whereas it does resemble unassociated lamps). I find that these two functions--predictive memory and visual similarity--trade off in neural space, relying on reliably dissociated neural areas in the temporal lobe. If time permits, I will show some preliminary work on the role of medial temporal areas in the build-up of predictive relational knowledge.