Reinforcing saccadic amplitude variability in a visual search task.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Neuroscience

Background:

• Human visual search often involves rigid scanning patterns, potentially hindering performance.
• Saccadic amplitude variability plays a role in efficient visual exploration.

Purpose of the Study:

• To investigate if reinforcement learning can induce specific saccadic amplitude variability during visual search.
• To determine if learned saccadic strategies transfer to standard visual search tasks.

Main Methods:

• A novel gaze-contingent visual foraging task was developed.
• Reinforcement learning principles were applied by making target discovery contingent on specific saccadic amplitudes.
• A control group received noncontingent reinforcement.

Main Results:

• Reinforcing rare saccade amplitudes increased uncertainty (U values) by 54.89% on average.
• Reinforcing frequent saccade amplitudes decreased uncertainty by 41.21%.
• A control group showed no consistent change in variability.

Conclusions:

• Reinforcement learning is a viable mechanism for modulating saccadic amplitude variability in visual search.
• Learned saccadic strategies demonstrated transferability to conventional visual search paradigms.
• Findings highlight the adaptive nature of oculomotor control through reinforcement learning.