Predicting rhesus monkey eye movements during natural-image search

Area of Science:

• Neuroscience
• Cognitive Science
• Vision Science

Background:

• Predicting human visual search in natural scenes involves salience, relevance, and context.
• Macaque monkeys are a key model for studying the neural control of eye movements.

Purpose of the Study:

• To evaluate the predictive power of salience, relevance, and context models on macaque visual search behavior.
• To determine if macaque visual search performance is comparable to humans in natural scenes.
• To investigate the role of scene context in macaque visual search, particularly with inverted images.

Main Methods:

• Monkeys were trained on a pedestrian search task previously used for human studies.
• Salience, relevance, and context models were applied to predict monkey eye fixations.
• Monkey search performance was tested using both upright and inverted natural scene images.

Main Results:

• Salience, relevance, and context models successfully predicted monkey eye fixations.
• The combined predictive accuracy for monkey fixations was similar to that observed in humans.
• Monkeys maintained high pedestrian detection rates even with inverted images, with model predictions for inverted images best explained by rotated original image predictions.

Conclusions:

• The macaque monkey serves as a robust model for understanding human visual search in natural environments.
• The findings support the use of macaque models to elucidate the neural basis of eye movement control during natural scene search.
• Environmental context significantly influences visual search, and macaque models can effectively replicate these effects.