Abstract
It is well known that spatial attention facilitates information processing at selected locations. For example, human observers are typically better at discriminating stimuli at attended relative to unattended locations. Several theories for this basic finding have been proposed. One possibility is that selective attention enhances the gain of neural responses in early visual areas (response enhancement). A second possibility is that attention reduces the variability of individual neurons and decorrelates noise within neural populations (noise reduction). A third possibility is that attention improves the efficiency with which sensory responses are "read out" by later decision-making and sensory motor responses (efficient selection). Here, we used a combination of psychophysics, electroencephalography (EEG), and computational modeling to evaluate these alternatives. Participants performed a two-interval-forced-choice contrast discrimination task while attending one or two locations (focused and divided attention conditions, respectively). Contrast-response functions (CRFs) were generated by plotting contrast-dependent changes in the mean amplitudes of early (the P1 ERP component) and late (late positive deflection or LPD) positive-going potentials that peaked over contralateral posterior-occipital (~80-130ms) and central posterior electrodes (~230-330ms), respectively. Consistent with the 'response enhancement' model, we found that focused attention had a multiplicative effect on the P1 and LPD CRFs. Moreover, a computational model that incorporated only changes in multiplicative response gain vastly outperformed models that assume noise reduction or efficient selection. These findings suggest that spatial attention facilitates behavioral performance primarily by enhancing the multiplicative response gain of sensory responses that cascade across processing stages.
Meeting abstract presented at VSS 2014