Categorization of complex visual images by rhesus monkeys. Part 2: single-cell study

Area of Science:

• Neuroscience
• Cognitive Science
• Visual Perception

Background:

• Understanding the neural basis of visual categorization is crucial for deciphering complex cognitive functions.
• The anterior temporal cortex is implicated in processing complex visual information and object recognition.

Purpose of the Study:

• To investigate the neural coding mechanisms underlying ordinate-level visual categorization.
• To determine how neurons in the anterior temporal cortex represent categories like 'trees' versus 'non-trees'.

Main Methods:

• Single-cell recordings were performed in the anterior temporal cortex of rhesus monkeys.
• Monkeys were trained to categorize color images of trees versus other objects.
• Neural responses were analyzed for selectivity and invariance to stimulus transformations.

Main Results:

• Neurons exhibited significant selectivity for complex color images.
• Approximately 25% of neurons showed category-specific responses (trees vs. non-trees).
• About 10% of neurons responded almost exclusively to trained category exemplars.
• Neural responses were largely invariant to changes in stimulus position and size.
• Response invariance was insufficient to cover the full variability within a category, challenging prototype theories.
• Within-category selectivity was strong, suggesting limitations of single-neuron representation.

Conclusions:

• Ordinate-level visual categorization relies on a population of neurons, not single prototype-representing neurons.
• Each neuron likely represents a limited set of exemplars within a category.
• This population coding model better explains the observed neural selectivity and invariance.