Learning to minimize efforts versus maximizing rewards: computational principles and neural correlates

Area of Science:

• Neuroscience
• Computational Neuroscience
• Decision Science

Background:

• Reward maximization mechanisms are well-studied computationally and neurally.
• Learning to minimize action cost remains less understood.
• A unified learning mechanism for choice dimensions is hypothesized.

Purpose of the Study:

• To investigate if a single computational rule governs both reward maximization and effort minimization.
• To identify the neural correlates of learning to maximize rewards versus minimize effort.

Main Methods:

• Healthy volunteers performed a probabilistic instrumental learning task.
• Task parameters included varying physical effort and monetary outcomes.
• Brain activity was monitored using functional magnetic resonance imaging (fMRI).

Main Results:

• Behavioral data supported a unified computational rule using prediction errors for both reward and effort learning.
• The ventromedial prefrontal cortex encoded reward information.
• The anterior insula, dorsal anterior cingulate, and posterior parietal cortex encoded effort information.

Conclusions:

• The brain employs a consistent computational learning rule for both benefits (rewards) and costs (effort).
• Distinct neural networks differentially represent reward and effort dimensions.
• This suggests specialized neural systems for processing different aspects of choice options.