Instance-based learning: integrating sampling and repeated decisions from experience.

Area of Science:

• Cognitive psychology
• Decision science
• Computational modeling

Background:

• Decisions from experience are studied via sampling and repeated-choice paradigms.
• These paradigms have been investigated independently, often assuming distinct cognitive processes.
• Existing computational models for each paradigm are diverse and specialized.

Purpose of the Study:

• To demonstrate that sampling and repeated-choice paradigms rely on common cognitive processes.
• To propose a unified cognitive model based on instance-based learning theory (IBLT).
• To integrate these paradigms by identifying the stopping point as the primary difference.

Main Methods:

• Utilizing instance-based learning theory (IBLT) to model behavior across both paradigms.
• Developing a single cognitive model incorporating a stopping point rule for the sampling paradigm.
• Employing quantitative model comparison to evaluate IBLT against existing specialized models.

Main Results:

• Behavior in both sampling and repeated-choice paradigms is explained by common cognitive processes within IBLT.
• A single IBLT-based model accurately captures human choices and predicts choice sequences across paradigms.
• IBLT significantly outperforms the best models developed for each paradigm individually.

Conclusions:

• The instance-based learning theory (IBLT) provides a unified framework for understanding decisions from experience.
• The stopping point is the critical factor differentiating sampling and repeated-choice experimental paradigms.
• This research integrates previously disparate findings, advancing the psychology of decision making.