Causal inference and cognitive-behavioral integration deficits drive stable variation in human punishment sensitivity

Area of Science:

• Cognitive Neuroscience
• Decision Science
• Behavioral Economics

Background:

• Adaptive decision-making relies on integrating punishment feedback to modify behavior.
• Individual differences in learning from punishment and avoiding harmful actions are not well understood.
• Understanding these differences is crucial for addressing persistent harmful behaviors.

Purpose of the Study:

• To identify distinct phenotypes of human punishment learning.
• To uncover the underlying cognitive mechanisms driving these phenotypes.
• To examine the longitudinal impact of these phenotypes on behavior.

Main Methods:

• A conditioned punishment task was administered to a diverse international sample (N=267).
• Phenotypes were identified based on responses to direct punishment and informational interventions.
• Cognitive mechanisms were assessed through analysis of causal inference and knowledge integration.

Main Results:

• Three distinct phenotypes were identified: Sensitive, Unaware, and Compulsive.
• Sensitive individuals adapted to punishment; Unaware individuals required informational intervention; Compulsive individuals persisted in harmful actions.
• Phenotypes were linked to causal inference deficits and integration failures, predicting outcomes six months later.

Conclusions:

• Individual variation in punishment learning stems from specific cognitive deficits.
• Targeted interventions can address causal inference deficits but not integration failures.
• This framework aids understanding of persistent harmful behavior and informs tailored interventions.