Lapses in perceptual decisions reflect exploration

Area of Science:

• Neuroscience
• Decision-Making
• Behavioral Economics

Background:

• Perceptual decision-making often includes errors (lapses) assumed to be independent of evidence strength and caused by external noise.
• Previous models treated these lapses as a nuisance, unrelated to the core decision process.

Purpose of the Study:

• To investigate the stimulus dependence of lapses in a multisensory decision task in rats.
• To propose and test a novel model where lapses represent a strategic trade-off between exploitation and exploration in decision-making.
• To explore the neural basis of lapses by linking them to specific brain regions.

Main Methods:

• Utilized a multisensory decision task in rats.
• Manipulated reward magnitude and probability of specific actions to test model predictions.
• Employed disruption experiments targeting the posterior striatum and secondary motor cortex to assign neural computations.

Main Results:

• Demonstrated that lapses are stimulus-dependent, contradicting the noise hypothesis.
• Confirmed model predictions: manipulations of action-specific rewards selectively altered lapses associated with those actions.
• Identified posterior striatum and secondary motor cortex as crucial for decision-related computations reflected in lapses.

Conclusions:

• Lapses are not mere noise but an integral component of strategic decision-making, reflecting a balance between exploiting known rewards and exploring uncertain options.
• Lapses provide valuable information about action values, even under disrupted brain states.
• This work redefines the role of lapses, offering a new framework for understanding decision computations and their neural underpinnings.