Attentional load modulates responses of human primary visual cortex to invisible stimuli

Abstract

Visual neuroscience has long sought to determine the extent to which stimulus-evoked activity in visual cortex depends on attention and awareness. Some influential theories of consciousness maintain that the allocation of attention is restricted to conscious representations [1, 2]. However, in the load theory of attention [3], competition between task-relevant and task-irrelevant stimuli for limited-capacity attention does not depend on conscious perception of the irrelevant stimuli. The critical test is whether the level of attentional load in a relevant task would determine unconscious neural processing of invisible stimuli. Human participants were scanned with high-field fMRI while they performed a foveal task of low or high attentional load. Irrelevant, invisible monocular stimuli were simultaneously presented peripherally and were continuously suppressed by a flashing mask in the other eye [4]. Attentional load in the foveal task strongly modulated retinotopic activity evoked in primary visual cortex (V1) by the invisible stimuli. Contrary to traditional views [1, 2, 5, 6], we found that availability of attentional capacity determines neural representations related to unconscious processing of continuously suppressed stimuli in human primary visual cortex. Spillover of attention to cortical representations of invisible stimuli (under low load) cannot be a sufficient condition for their awareness.

Figure 1

Stimulus Configuration Used in the Experiment (A) A continuous central RSVP task was combined with presentation of invisible, task-irrelevant stimuli in the periphery. While a subject wore red-blue anaglyph glasses, the nondominant eye was presented with low-contrast red line drawings of two objects in two of the four visual-field quadrants along the 45° or 135° diagonal. The dominant eye was presented with four highly salient, high-contrast, and rapidly changing blue masks, one in each of the four quadrants. (B) Rapid serial visual presentation (RSVP) task at the fovea. Depending on task instruction, the same pseudorandom stream of colored letters served as both the low-load (detection of “pop out” target letter ‘T’ irrespective of its color) and high-load (detection of specific “conjunctions” of letter form and color, i.e., white N and blue Z) tasks.

Figure 2

Experimental Results (A) In both load conditions, localization of the tool images occurred at the level of chance (50%; one-sample t test; p = 0.919 and 0.737 for low and high loads, respectively). Individual analyses also confirmed that none of the participants were better than chance. (B) Reaction time for RSVP target detection as a function of load. Observers were significantly slower under high-load conditions (paired t test; t(6) = 5.792; p = 0.001). (C) Differential V1 BOLD response to invisible images under high and low loads. The y axis shows the percent signal change, averaged over the selected V1 voxels, for suppressed-stimulus presence minus absence. Data are plotted for each participant (lines) as well as the group means (bars). All error bars indicate 1 standard error of the mean.