Modeling the integration of expectations in visual search with centre-surround neural fields

Area of Science:

• Computational Neuroscience
• Cognitive Psychology
• Vision Science

Background:

• Eye movements, or saccades, are crucial for visual search and information gathering.
• Understanding the neural mechanisms underlying saccade generation is a key challenge in neuroscience.
• Previous models have explored information integration for saccade targeting, but a comprehensive dynamic model is needed.

Purpose of the Study:

• To investigate the Centre-Surround Neural Field (CSNF) model as a mechanism for integrating information into target likelihood maps for saccade guidance.
• To compare the predictive accuracy and saccadic latencies of the CSNF model against human behavioral data in visual search tasks.
• To assess the model's ability to capture eye movement patterns in multi-target search scenarios.

Main Methods:

• The Centre-Surround Neural Field (CSNF) model, a dynamic neural network, was employed.
• The model simulates excitation of nearby regions and inhibition of distant locations to mimic saccadic competition.
• Simulations were conducted under conditions analogous to naturalistic search tasks with targets present (expected/unexpected) or absent, and in multi-target scenarios.

Main Results:

• The CSNF model successfully predicted accuracy patterns similar to human participants in visual search tasks.
• The model showed discrepancies in predicting saccadic latencies, differing from human behavioral findings in some conditions.
• The model qualitatively captured eye movement behavior in multi-target search conditions.

Conclusions:

• CSNF models offer a plausible mechanism for information integration in saccade targeting, particularly regarding search accuracy.
• Discrepancies in saccadic latency predictions highlight areas for refinement in current neural field models.
• The CSNF model demonstrates potential for explaining complex eye movement behaviors in visual search.