Orienting auditory attention in time: Lateralized alpha power reflects spatio-temporal filtering

The deployment of neural alpha (8-12 Hz) lateralization in service of spatial attention is well-established: Alpha power increases in the cortical hemisphere ipsilateral to the attended hemifield, and decreases in the contralateral hemisphere, respectively. Much less is known about humans’ ability to deploy such alpha lateralization in time, and to thus exploit alpha power as a spatio-temporal filter. Here we show that spatially lateralized alpha power does signify - beyond the direction of spatial attention - the distribution of attention in time and thereby qualifies as a spatio-temporal attentional filter. Participants (N = 20) selectively listened to spoken numbers presented on one side (left vs right), while competing numbers were presented on the other side. Key to our hypothesis, temporal foreknowledge was manipulated via a visual cue, which was either instructive and indicated the to-be-probed number position (70% valid) or neutral. Temporal foreknowledge did guide participants’ attention, as they recognized numbers from the to-be-attended side more accurately following valid cues. In the magnetoencephalogram (MEG), spatial attention to the left versus right side induced lateralization of alpha power in all temporal cueing conditions. Modulation of alpha lateralization at the 0.8-Hz presentation rate of spoken numbers was stronger following instructive compared to neutral temporal cues. Critically, we found stronger modulation of lateralized alpha power specifically at the onsets of temporally cued numbers. These results suggest that the precisely timed hemispheric lateralization of alpha power qualifies as a spatio-temporal attentional filter mechanism susceptible to top-down behavioural goals.


39
Selective attention refers to the prioritization of task-relevant sensory stimuli at the expense of 40 distraction (Desimone & Duncan, 1995). Time

93
Neuroscience work has shown that brain mechanisms of spatial and temporal attention interact.

94
Studies that provided participants with spatial and temporal foreknowledge about visual target 95 stimuli found overlapping parietal cortex activations in functional magnetic resonance imaging and 96 positron emission tomography (Coull & Nobre, 1998), as well as interactive effects of both stimulus 97 dimensions on early components in the event-related potential in the EEG (Doherty et al., 2005). In

106
In the present study, we employ magnetoencephalography (MEG) to record neural responses in 107 human participants. We ask whether and how the human brain implements an attentional filter 108 mechanism that is both spatially and temporally specific. To this end, we augmented an established 109 auditory spatial attention paradigm (Tune et al., 2018;Wöstmann et al., 2016;Wöstmann et al., 2018) 110 5 and employed an additional cue to provide temporal foreknowledge. While pre-stimulus alpha 111 lateralization was unaffected by temporal foreknowledge, we here show that rhythmic modulation 112 of alpha lateralization at the speech stimulus rate increases following a cue that allows temporal 113 foreknowledge. This modulation of alpha lateralization was temporally specific in the sense that 114 alpha modulation was more pronounced at time points of temporally cued versus non-cued speech 115 items.

117
The design of the present study follows closely the procedure of previous auditory spatial attention 118 paradigms used in our lab (Tune et al., 2018;Wöstmann et al., 2016). The major difference is that in 119 addition to a spatial cue (to indicate whether to focus attention to the left versus right side), we here 120 also employed a temporal cue to guide a listener's attention in time to one out of five stimulus  background (see Fig. 1A). In one half of trials, the temporal cue was instructive, meaning that one 135 bar was larger in height than the other bars and indicated that the respective number position was 136 likely probed and should thus be attended. In the other half of trials, the temporal cue was neutral, 137 meaning that all bars were of the same height and no number position was cued. For each trial with 138 an instructive temporal cue, the number to be probed in the end of a trial was drawn from a 139 distribution that contained the cued number positions with 70% probability and the remaining 140 number positions with 30% probability. Instructive cue trials can be considered valid in case the cued 141 number position was probed and invalid otherwise. The expected value of cue validity was 70%, 142 which effectively ranged between 64% and 81% across participants in the present study.

143
Second, a spatial cue was provided to indicate whether spoken numbers on the left or right side 144 had to be attended. The spatial cue was a monaural 1000-Hz pure tone of 0.5-s duration, with a 0.05-145 s linear onset ramp. The spatial cue was presented on the to-be-attended side and was valid in 100%

154
For each experimental trial, ten different numbers were randomly selected. Numbers were 155 grouped into two streams of five numbers, each. Sound presentation was dichotic, meaning that 156 one stream was presented to the left ear and the other to the right ear. For concurrent numbers, 157 perceptual centers were temporally aligned, and digits were distinct in their ten and one positions 158 (e.g., co-occurrences of "35" and "37" or "81" and "21" were avoided). The onset-to-onset interval of 159 every two subsequent numbers was 1.25 s, resulting in a number presentation rate of 0.8 Hz. Finally, 160 broadband background noise (at a signal-to-noise ratio, SNR, of +10 dB) was added for the entire 161 trial duration (spatial cue onset until final number offset).

163
Design and procedure. Prior to the presentation of auditory materials, the temporal cue was 164 presented for the duration of 2.5 s and was afterwards replaced by a fixation cross. Next, after a time 165 interval of ~1 s (jittered randomly between 0.8 and 1.2 s), the auditory spatial cue was presented 166 either on the left or right ear to indicate that spatial attention had to be directed to the left or right 167 side, respectively. 1.5 s after spatial cue onset, the presentation of competing numbers on the left 168 and right side started. Approximately 1 s after the offset of the last pair of competing numbers 169 (jittered randomly between 0.8 and 1.2 s), a response screen was presented containing two numbers.

170
One of the probe numbers -the target -was a number from the to-be-attended side. The other 171 number -the lure -was a number not presented at all during the respective trial. In case of a valid 172 temporal cue trial (~70 % of instructive cue trials), the target number was the number presented on 173 the to-be-attended side at the temporally cued position. In case of an invalid temporal cue trial (~30 174 % of instructive cue trials), the target number was a number presented on the to-be-attended side

186
The experiment was divided in 5 blocks; participants took brief breaks in-between the blocks. The         3 Hz with ±2 Hz spectral smoothing (time period: 0-7.5 s relative to spatial cue onset). This filter was 296 applied to single-trial Fourier Transforms (1-5 Hz, in steps of ∼0.133 Hz). ITPC at each grid point was 297 calculated and averaged across frequencies.

298
To localize the 0.8-Hz modulation of ITPC, the spatially adaptive filter was applied to time-

303
To localize alpha power lateralization prior to the onset of lateralized speech items, we first 304 calculated a common filter for each participant, based on the leadfield and the cross-spectral density 305 of Fast Fourier Transforms centered at 10 Hz with ±2 Hz spectral smoothing (time period: 0-1.5 s 306 relative to spatial cue onset; calculated on all trials). Next, the common filter was used to localize 307 alpha power separately for attend-left versus attend-right trials, followed by calculation of the alpha 308 lateralization index for each grid point (according to Equation 1).

309
To localize the 0.8-Hz modulation of alpha power lateralization, a common filter was calculated 310 for every participant with the same parameters stated above, but for the time period 1.5-7.5 s 311 relative to spatial cue onset. The common filter was used to localize time-resolved alpha power (0.5-

350
For circular statistics, we used the Rayleigh test for uniformity of circular data and the parametric

351
Hotelling paired sample test for equal angular means (Zar, 1999), implemented in the circular 352 statistics toolbox for Matlab (Berens, 2009

451
Auditory encoding, quantified as low-frequency inter-trial phase coherence (ITPC), exhibited 452 pronounced rhythmic activity at the number presentation rate of 0.8 Hz (Fig. 3A-D). Across         Figure S4 shows that the lateralization of 535 oscillatory power around number onsets was specific to the alpha frequency band). It might at first 536 glance seem surprising that we did not observe more positive, but instead more negative alpha 537 lateralization for cued versus uncued numbers (Fig. 5C) and for cued versus neutral numbers (Fig.   538   5D) in a ~0.5-s long time window including number onset. However, as we know from the whole-539 trial analysis (Fig. 3), alpha lateralization temporally lags the presentation and sensory encoding of 540 numbers by ~0.46 s. Thus, there is a more negative state of alpha lateralization at number onset, 541 followed by a state of alpha lateralization close to zero ~0.46 s thereafter. For cued compared with 542 uncued and neutral numbers, the difference between these two states is more pronounced, which 543 effectively results in a stronger modulation of lateralized alpha power.      (Fig. 5). Presumably, the temporally lagging alpha lateralization is a neural signature of a 642 reactive filter mechanism, which serves to select the previously presented number on the to-be-