Attention, predictive learning, and the inverse base-rate effect: evidence from event-related potentials

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Learning and Memory

Background:

• The inverse base-rate effect (IBRE) is a non-rational bias in predictive learning where individuals make decisions contrary to base rates.
• Existing theories suggest error-driven attention drives the IBRE, with shared cues attracting more attention.

Purpose of the Study:

• To conduct the first electrophysiological investigation of the inverse base-rate effect (IBRE).
• To test the prediction that error-driven attention theories explain the IBRE by examining brain potentials.
• To differentiate attentional explanations from alternative accounts like novelty or response accuracy.

Main Methods:

• Electrophysiological recordings (Selection Negativity and Selection Positivity) were used to measure brain activity during a predictive learning task.
• Participants learned symptom-disease associations with varying base rates.
• Control conditions with matched frequencies but no shared symptoms were included.

Main Results:

• Brain potentials associated with visual attention (posterior Selection Negativity, anterior Selection Positivity) were observed for the more frequently predicted symptom (C) compared to the less frequently predicted symptom (B).
• These attentional effects and the IBRE were absent in control conditions without shared symptoms.
• Response accuracy was higher for B than for C, dissociating accuracy from attentional allocation.

Conclusions:

• Electrophysiological evidence supports error-driven attention theories in explaining the inverse base-rate effect.
• Attentional allocation, specifically to shared cues, is a key mechanism underlying the IBRE.
• The findings distinguish attentional mechanisms from other potential explanations for this learning bias.