Symmetrical "super learning": Enhancing causal learning using a bidirectional probabilistic outcome

Area of Science:

• Cognitive Psychology
• Learning Science

Background:

• Cues in learning environments can interfere with or enhance each other.
• Prior research primarily examined cues for binary outcomes, creating asymmetry between excitatory and inhibitory learning.

Purpose of the Study:

• To investigate symmetrical enhancement effects in learning about generative and preventative causal cues.
• To explore bidirectional learning using a novel experimental paradigm with opposite polarity events.

Main Methods:

• Conducted three experiments with 133 human participants using predictive learning tasks.
• Exposed participants to probabilistic cues predicting incremental positive (generative) and negative (preventative) outcomes.
• Analyzed explicit predictive ratings to assess learning.

Main Results:

• Found evidence for symmetrical learning for both generative and preventative cues.
• Demonstrated enhanced learning effects in opposite directions for both cue types.
• Results align with theories of competitive learning and contrasting probability estimates.

Conclusions:

• Learning about generative and preventative cues is symmetrical and enhanced.
• The findings challenge the perceptual and motivational asymmetry in previous excitatory and inhibitory learning models.
• Introduces the concept of 'super learning' for bidirectional causal cue acquisition.