Differences in learning across the lifespan emerge via resource-rational computations.

Area of Science:

• Cognitive Psychology
• Developmental Psychology
• Computational Neuroscience

Background:

• Accurate belief formation is vital for adaptive behavior but computationally intensive.
• Learning involves predictive inference, with potential age-related differences linked to cognitive resource allocation.

Purpose of the Study:

• To investigate if age differences in learning arise from resource-rational computations.
• To model belief updating as a sampling process balancing accuracy and computational cost.

Main Methods:

• Developed a resource-rational model where belief updating uses a sampling process with a stopping criterion.
• Collected data from lifespan studies (N=129, N=90) and a younger adult study (N=94).
• Analyzed participant biases in belief updating, specifically perseveration and anchoring.

Main Results:

• Children and older adults exhibited biases consistent with a frugal sampling policy.
• Observed increased perseveration when updating prior beliefs.
• Found a stronger anchoring bias when updating from external values.

Conclusions:

• The findings support the resource-rational model, suggesting biases originate from sampling strategies.
• Perseverative and anchoring biases can emerge from efficient belief-updating computations.
• Resource-rational adjustments in sampling computations explain age-related learning changes.