Related Experiment Videos
Feature-based attention increases the selectivity of population responses in primate visual cortex
Julio C Martinez-Trujillo1, Stefan Treue
1Centre for Vision Research, 4700 Keele St., York University Toronto, Ontario M3J 1P3, Canada.
Current Biology : CB
|May 4, 2004
Summary
Feature-based attention modulates neuronal responses multiplicatively, enhancing selectivity for attended features. This differs from spatial attention, which only changes response gain without altering selectivity.
Area of Science:
- Neuroscience
- Visual Cortex Research
- Attention Mechanisms
Background:
- Spatial attention modulates neuronal gain in the visual cortex, strengthening attended locations without altering selectivity.
- The effect of feature-based attention on neuronal selectivity remains unknown.
Purpose of the Study:
- To investigate whether feature-based attention modulates neuronal responses similarly to spatial attention.
- To determine if feature-based attention alters neuronal selectivity in the visual cortex.
Main Methods:
- Recorded responses of 135 direction-selective neurons in the middle temporal area (MT) of macaques.
- Presented an unattended distractor stimulus within a neuron's receptive field while animals attended to a stimulus in the opposite hemifield.
- Manipulated the direction of the attended stimulus relative to the neuron's preferred direction.
Main Results:
- Feature-based attention modulated neuronal responses by a factor of approximately 12% when the attended direction was constant.
- Changing the attended direction systematically altered modulation from enhancement to suppression, increasing directionality by about 20%.
- Neuronal responses were multiplicatively modulated, with modulation strength depending on feature similarity.
Conclusions:
- Feature-based attention exerts a multiplicative modulation on neuronal responses, consistent with the feature-similarity gain model.
- Feature-based attention increases feature selectivity at the population level by enhancing responses to preferred features and suppressing responses to non-preferred features.