A key functional property of slow oscillations is the rhythmic shifting of excitability in local neuronal ensembles (Lakatos et al., 2005). It has been shown in the monkey visual cortex that delta oscillations can entrain differentially to 2 interleaved streams of auditory and visual stimuli depending on the modality to which attention is paid, resulting in increased response to visually-attended stimuli (Lakatos et al., 2008). The goal of the present study was to extend these results to other parts of the cortex and to the human .

The experiment was run on 7 epileptic patients undergoing surgical evaluation and implanted with subdural electrodes over various parts of the cortex. The task before the patients was to detect either auditory or visual targets in a stream of alternating auditory and visual standard stimuli presented just above threshold. The stimulation rate in a given modality was1.54 Hz (delta rhythm), with a normally distributed random jitter.

Out of a total of 584 electrodes showing no or little pathological activity, we first selected 197 electrodes showing some entrainment to at least one of the stimulus streams (i.e. significantly non-uniform delta phase at the time of stimulation across trials, rayleigh test, p<.05 non-corrected). We then compared the phase of delta oscillations between auditory-attended and visual-attended trials. Out of 197 electrodes, 72 showed a significant phase difference (permutation test p<.05, corrected for multiple tests per patient), showing that attention can modulate the way delta rhythm is entrained to the stimulus stream in various parts of the brain.

Significant attentional phase shifts ranged from very small values (less than 15º) to complete phase reversals as observed in monkey V1. The value of the phase shift was significantly correlated and inversely proportional to the amplitude of post-stimulus evoked activity, as indexed by a post-stimulus increase in phase-locking values (PLV) between 3 and 30 Hz in a 50-350ms post-stimulus time-window.
In electrodes over the parietal and frontal cortex, the attentional phase shift was generally larger and not associated to an increase in PLVs, suggesting that, at least in these areas, entrainment could not be explained by evoked activity at the rate of stimulation.
In contrast, electrodes over the superior temporal (auditory) cortex, the occipital and subtemporal (visual) cortex and the primary motor/somatosentory areas showed higher PLVs, corresponding to evoked activity that might have hidden effects on ongoing delta oscillations. Interestingly. auditory and visual sensory areas showed classical attention effects, the PLVs increasing over the sensory cortex of a given modality when patients paid attention to this modality.
Importantly, electrodes showing significant effects barely overlapped with those showing frequent or occasional interictal activity as identified by epileptologists.

These results demonstrate that in the human brain, ongoing delta oscillations, reflecting variations in membrane excitability, can entrain to sensory streams in extended regions of the cortex and that this entrainment can be modulated to serve intermodal attentional selection, possibly leading to the classically observed increase in sensory responses to the stimuli.