Apr 06

CNS 2016 – Dissociating Strategies in 2D Cat Learning.


Human category learning is driven by at least two qualitatively distinct mechanisms. One is based on pre- decisional stimulus-response mappings which integrate information (II) across stimulus dimensions, are learned gradually through reinforcement, and are associated with implicit learning. This is contrasted by rule-based (RB) category learning that is tied to declarative memory. Category structures elicit II mechanisms by requiring integrating two stimulus dimensions, however, little research has carefully examined participant reports to verify that learning has occurred outside of awareness. Using a novel post-learning interview based on techniques from studies of implicit learning, we present a verbalizability measure for category knowledge that assesses conscious expression of this knowledge. Participants are asked about their strategic approach and indirectly probed for reportable heuristics used to judge category membership. They are then asked to annotate a diagram of the two-dimensional category space to probe for conscious, non-verbal rule knowledge. These reports suggest that a notable subset of participants in II category learning conditions actually explicitly discover a rule for using both stimulus dimensions to predict category membership. These participants exhibit performance that matches simple computational models (e.g., decision-bound theory; DBT) of II choice behavior even though performance is likely driven by conscious rules and MTL-dependent memory. Thus, prior dissociations between category systems underestimates category learning differences because a subset of putatively II learners may be utilizing rule-based strategies. By incorporating assessments of conscious knowledge, future work can more accurately identify specific characteristics of implicit visual category learning by excluding these explicitly-driven participants.

Reuveni II Verbalizability_CNS2016_Final

PDF version:

Reuveni II Verbalizability_CNS2016_Final

Leave a Reply