Re-starting a neural race: anti-saccade correction

Area of Science:

• Neuroscience
• Cognitive Psychology
• Computational Modeling

Background:

• The anti-saccade task involves conflicting pro-saccade and anti-saccade responses.
• Previous models explained reaction times for correct and error responses but not corrections.
• The occurrence and timing of error corrections remained unstudied.

Purpose of the Study:

• To investigate the mechanism underlying error corrections in the anti-saccade task.
• To test a novel hypothesis that decision races restart after errors.
• To model the timing of corrective saccades using a computational approach.

Main Methods:

• Human volunteers performed an anti-saccade task, with all responses recorded.
• A computational model based on racing Linear Approach to Threshold at Ergodic Rate (LATER) units was employed.
• The model was extended with an assumption that incorrect pro-saccades initiate corrective anti-saccades.

Main Results:

• The model accurately predicted the timing of corrective saccades without additional parameters.
• The occurrence and timing of corrections were explained by the decision race restarting mechanism.
• The model successfully accounted for complex behavioral aspects of error correction.

Conclusions:

• Error correction in the anti-saccade task involves restarting a neural decision race.
• This mechanism explains the timing of corrective saccades and adds weight to the model.
• The concept of restarting decision races may be crucial for understanding sequential decision-making.