Choice of saccade endpoint under risk

Area of Science:

• Visual perception
• Decision-making under uncertainty
• Oculomotor control

Background:

• Eye movements, specifically saccades, are crucial for visual information acquisition by directing gaze to high-resolution retinal areas.
• Prior research indicates observers optimize fixation points for information gain and reduced uncertainty.
• The role of explicit rewards in guiding saccade endpoint selection remains an area for investigation.

Purpose of the Study:

• To investigate whether human observers select saccade endpoints that maximize reward gain in a detection task with explicit, location-dependent rewards.
• To compare human saccade endpoint selection strategies against an ideal observer model maximizing expected gain.

Main Methods:

• An 8-alternative forced-choice detection task was employed using a contrast-defined target in noise.
• Participants performed a single saccade to indicate target location, with varying reward magnitudes associated with each potential location.
• Human saccade endpoints were compared to those of an ideal observer using a visibility map (d') derived from detection performance across retinal locations.

Main Results:

• The spatial distribution of human saccade endpoints was significantly influenced by the location of the highest reward.
• Both human and ideal observers demonstrated a tendency to saccade towards high-reward and actual target locations.
• Human observers exhibited a greater number of saccades to locations distant from the highest reward and target, diverging from the ideal observer's strategy.

Conclusions:

• Explicit rewards significantly modulate saccade endpoint selection in visual search tasks.
• Human saccade strategies can be suboptimal compared to an ideal observer, with a tendency to deviate towards less optimal locations.
• Suboptimal saccade endpoint choice and information integration may impair target detection and reward maximization.