Neurophysiology of intermodal selective attention in humans

Abstract

To function normally, primates must selectively attend a subset of the profusion of sensory information that constantly bombards the sense organs. We investigated the neural mechanisms involved in selectively attending to one of two simultaneously presented sensory modalities; namely audition and vision. The primary focus was on the mechanisms by which attention was deployed and sustained in an expectancy period preceding arrival of bi-sensory stimulation. This was achieved by using a cueing paradigm, whereby visually presented cue-words instructed subjects which of the two modalities was to be attended on a given trial. By examining neural activity in the intervening period between cue and imperative stimulus (approximately 1 second), we assessed the processes of attentional deployment. We employed the method of high-density mapping of human event-related potentials (ERP - 128 channels). Processing of cue-words was enhanced in early visual sensory cortices relative to passive viewing. A robust early activation was seen over left lateral frontal cortex (peaking at 180ms), likely representing encoding of cue-words. At 230ms, differential activations were observed over mid-frontal and mid posterior parietal cortices dependent upon which modality was to be attended, and likely representing differential deployment of anticipatory attention by the fronto-parietal attention system. In the later period (after 300ms), we found a pair of bilateral fronto-polar positivities that were of greater amplitude during attend-auditory trials and these were accompanied by enhancement of a fronto-central negativity. We interpreted this pattern as anticipatory biasing of auditory association cortices, driven by top-down inputs from frontal cortex. Midline frontal activation was enhanced during attend-visual trials, accompanied by enhancement of activity over parieto-occipital cortex. We interpreted this pattern as frontally driven preparatory attention in parieto-occipital visual cortices. Attention could also be biased by selective inhibition of incoming stimuli of the distracter modality. We hypothesized that oscillatory alpha-band activity might constitute such inhibitory mechanisms. Cueing attention to audition resulted in higher parieto-occipital alpha in the period preceding the onset of the imperative stimulus than when attention was cued to vision. We propose that this alpha activity reflects a disengaged visual attentional system in preparation for anticipated auditory input that is attentionally more relevant.