Poster D98, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
Neuroimaging context-dependent perceptual classification
Matthew Crossley1, Jessica Roeder2, Lauren Vucovich2, F. Gregory Ashby2; 1SRI International, 2UC Santa Barbara
Context-dependent learning (CDL) allows an agent to learn and engage different response policies given identical stimuli in different contexts. CDL is essential for survival, and many neurological disorders and maladaptive behaviors can be traced to dysfunction within CDL systems. Here we aim to understand how neural circuits drive CDL. Towards this goal, we measured fMRI BOLD signal in participants as they performed a context-dependent perceptual classification task thought to rely on procedural learning and memory systems. Throughout the entire experiment, participants used correct / incorrect feedback provided after every response to learn to classify circular sine wave gratings through trial-and-error. During the first two scanning blocks, participants first learned one category (i.e., stimuli-response mappings), and during the final four blocks of scanning 50% of these mappings were reversed. No warning was given prior to this change. A GLM analysis identified a single cluster spanning parts of posterior parietal and inferotemporal cortex that was significantly more active during blocks 3 and 4 than it was during blocks 1 and 2, and 5 and 6. Thus, this area may be important for recognition of the context change that happened between blocks 2 and 3. We will also report a variety of other analyses that investigate the functional connectivity of this area with whole brain networks, and attempt to link activity in this area to the behavioral profile of individual participants.
Topic Area: LONG-TERM MEMORY: Skill learning