Predictive processing in volatile environments.

Area of Science:

• Decision-making and computational neuroscience
• Animal behavior and ecology

Background:

• Adaptive behavior involves balancing exploitation and exploration under uncertainty.
• Diverse fields like ecology and neuroscience study this using various frameworks (e.g., reinforcement learning, Bayesian inference).

Purpose of the Study:

• To synthesize diverse research on decision-making under uncertainty within a predictive processing framework.
• To explore how animals infer environmental structure and adapt behavior to variability.

Main Methods:

• Review and synthesis of experimental and theoretical studies from ecology and neuroscience.
• Analysis of laboratory paradigms (probabilistic reversal learning, foraging tasks) and their relevance to decision variability.
• Examination of neural mechanisms (cortical/subcortical circuits, neuromodulators) underlying adaptive decisions.

Main Results:

• Decision variability often reflects adaptive sampling, not suboptimal noise.
• Animals regulate learning rates, persistence, and exploration based on inferred sources of outcome variability.
• Neural circuits and neuromodulatory systems encode predictions and modulate evidence integration.

Conclusions:

• Adaptive decision-making is a hierarchical uncertainty resolution process across timescales and contexts.
• This framework links ecological decision rules, laboratory models, and neural mechanisms.
• Understanding adaptive behavior requires integrating insights from multiple scientific disciplines.