Abstract
Although fMRI has played a central role in the study of attention, growing evidence suggests that the BOLD signal may be insensitive to attentional modulation of stimulus-driven responses. While electrophysiological studies find that attention multiplicatively scales contrast-response functions (CRFs), fMRI studies find a different pattern: attention produces an additive baseline shift in the BOLD signal that is independent of stimulus contrast. This finding suggests that attentional effects measured with fMRI reflect top-down inputs to visual cortex, rather than the modulation of stimulus-driven activity. If this is the case, fMRI may provide little insight about how attention improves sensory coding. Here, we re-examined whether fMRI is sensitive to multiplicative effects of attention, using a feature-based attention (FBA) paradigm designed to preclude additive effects. We manipulated FBA in one hemifield, and measured responses to a probe stimulus in the opposite hemifield, exploiting the well-established spatial spread of FBA’s effects. Specifically, we cued human observers to attend to one of two gratings (one vertical, one horizontal) in the relevant hemifield, and we measured responses evoked by a horizontal probe in the other hemifield. Thus, the probe could match or mismatch the attended orientation. To measure CRFs in visual areas V1-V3, we parametrically varied the contrast of the probe grating (0-96% contrast). We found no additive effects with attention. Instead, FBA increased contrast gain, shifting CRFs leftwards, improving contrast sensitivity. This effect was larger in V2 and V3 than in V1. For a subset of eccentricities, we also found an increase in response gain in V2 and V3, a vertical scaling of the contrast-response function. Simulations revealed that these multiplicative effects of FBA can be explained with a normalization model of attention. Importantly, these non-additive effects of FBA provide positive evidence that fMRI is sensitive to attentional modulation of stimulus-driven activity.