A computational account of transsaccadic attentional allocation based on visual gain fields

Area of Science:

• Neuroscience
• Cognitive Science
• Visual Perception

Background:

• Goal-directed behavior relies on tracking object locations.
• Early visual areas map objects by retinal position, not world coordinates.
• Computing world-referenced spatial information across eye movements is a key debate.

Purpose of the Study:

• To investigate if covert attention shifts accurately track real-world object locations during eye movements.
• To explore the precision of attentional selectivity across saccades.
• To develop a model explaining attention allocation in world-referenced coordinates.

Main Methods:

• Human observers performed tasks involving covert attention shifts during saccades.
• Attentional selectivity was measured for its precision and temporal accuracy.
• A computational observer model was developed based on neurophysiological data.

Main Results:

• Attentional selectivity demonstrated remarkable precision, minimally affected by saccades.
• The observer model successfully recapitulated human data.
• The model explained attention allocation to task-irrelevant locations across eye movements.

Conclusions:

• Visual attention operates within a real-world coordinate system.
• This world-referenced spatial computation occurs rapidly in early cortical processing.
• Findings challenge previous models and offer a parsimonious explanation for attentional phenomena across eye movements.