Do procedural and declarative category learning form distinct or shared representations? An fMRI-RSA study

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Dr. Priya Kalra¹ (pkalra7@uwo.ca), Laura Batterink², J. Paul Minda³, Marc Joanisse⁴; ¹University of Western Ontario

Studies on interactions between procedural and declarative learning have focused on largely on competition during encoding, consolidation, or use (retrieval). Less attention has been paid to interactions between the representations created by each system. In a behavioral study, we demonstrated that information from both declarative and procedural learning can contribute to response selection. Participants were instructed to use a completely diagnostic, verbalizable, shape-based rule to categorize exemplars and received feedback after each trial. However, the categories also differed probabilistically in their color distributions. Participants used both color (learned procedurally) and shape (learned declaratively) to categorize exemplars, making faster responses to stimuli that were congruent with training color distributions compared to incongruent (2.98s vs 3.28s, respectively, t = 3.16**). Debriefing confirmed participants were unaware of the color distributions. This result suggests that both the color (procedural) and shape (declarative) information contributed to response selection. We now ask whether procedural and declarative learning contribute to a shared representation of the stimulus space, or whether there are distinct procedurally-learned and declaratively-learned representations that jointly contribute to response selection. In the current study, participants performed the same categorization task while fMRI data was acquired. Preliminary univariate analysis reveals greater activation in a fronto-parietal network (including inferior frontal gyrus, ventro-medial frontal gyrus, and inferior parietal lobule), as well as striatal areas, to incongruent than congruent stimuli. Ongoing fMRI representational similarity analysis will allow us to determine whether information is shared between systems at the level of stimulus space representation, or only later at response selection.

Topic Area: LONG-TERM MEMORY: Skill Learning