Bumblebees learn to forage like Bayesians.

Area of Science:

• Behavioral Ecology
• Animal Cognition

Background:

• Bayesian foraging models predict animals use prior information about resource distribution.
• This prior information can be innate or learned, influencing patch-selection strategies.
• Understanding learned priors is crucial for explaining foraging plasticity.

Purpose of the Study:

• To investigate if bumblebee foragers can rapidly learn prior information about resource distribution.
• To determine if learned foraging strategies adapt to different environmental variances (high vs. uniform).
• To confirm the role of learned prior expectations in optimizing foraging behavior.

Main Methods:

• Experimental manipulation of patch quality in high-variance and uniform environments.
• Observation of bumblebee foraging behavior and reward intake rates.
• Application of Cox regression models to analyze patch-staying tendencies.
• Testing for carry-over effects of learned behavior in a common environment.

Main Results:

• Bumblebees demonstrated rapid learning of prior information within hours.
• Bees adjusted their patch-staying behavior (incremental vs. decremental responses) according to environmental predictions.
• Learned foraging strategies improved reward intake rates to an average of 80% of the predicted maximum.
• Adaptive behaviors persisted after training, indicating memorized prior information.

Conclusions:

• Bumblebee foragers exhibit remarkable adaptive plasticity through learned prior expectations.
• This study provides the first clear evidence of isolating the adaptive use of learned priors in Bayesian foraging.
• Learned prior information significantly enhances foraging efficiency in complex, patchy environments.