Heterogeneity in strategy use during arbitration between experiential and observational learning

Area of Science:

• Cognitive Science
• Neuroscience
• Computational Psychology

Background:

• Navigating social environments requires flexible learning strategies, including experiential and observational learning.
• Individual differences in strategy use and environmental adaptation remain poorly understood.

Purpose of the Study:

• To investigate an arbitration mechanism where prediction errors from each learning strategy influence behavioral weighting.
• To characterize individual differences in learning strategy arbitration using computational methods.

Main Methods:

• Developed an online behavioral task to test a computational model of learning strategy arbitration.
• Analyzed participant data to identify distinct groups based on their learning strategy use.

Main Results:

• A significant portion of participants utilized the proposed prediction error-based arbitration mechanism.
• Identified distinct groups: fixed strategy mixers, single-strategy users, and non-learners.
• Observed differences in behavioral signatures and transdiagnostic symptom dimensions (autism traits, anxiety) across groups.

Conclusions:

• Prediction error-driven arbitration is a key mechanism for flexible social learning.
• Heterogeneous learning strategies exist, with distinct behavioral and clinical correlates.
• Computational approaches applied to large datasets can reveal individual differences in learning.