Learning expectations shape cognitive control allocation

Area of Science:

• Cognitive Psychology
• Decision Science
• Neuroscience

Background:

• Cognitive control allocation is often modeled as expected utility maximization, balancing benefits against costs.
• Recent theories suggest incorporating the value of learning into control allocation decisions is normative, even if it leads to longer deliberation.
• This challenges intuitive notions of immediate reward maximization.

Purpose of the Study:

• To empirically test whether individuals allocate greater cognitive control to tasks perceived as learnable.
• To investigate if perceived learnability influences willingness to expend effort, such as longer deliberation times.
• To examine compliance with formal models of control allocation that include learning value.

Main Methods:

• Participants performed a simple perceptual dot-motion task across blocks.
• Willingness to deliberate (expend effort) was measured.
• Participant proficiency and learning rate in initial blocks were assessed to predict subsequent control allocation.

Main Results:

• Task learnability significantly predicted participants' willingness to allocate greater cognitive control.
• Higher proficiency and learning rates in early task engagement correlated with increased deliberation in later stages.
• Behavior aligned with models predicting effort allocation based on future learning value.

Conclusions:

• Individuals do consider task learnability when deciding how much cognitive control to allocate.
• Findings support theoretical frameworks where agents weigh the future discounted value of learning when making effort-based decisions.
• This provides evidence for normative models of cognitive control that integrate learning into reward calculations.