Individual differences in alpha frequency drive crossmodal illusory perception

Abstract

Perception routinely integrates inputs from different senses. Stimulus temporal proximity critically determines whether or not these inputs are bound together. Despite the temporal window of integration being a widely accepted notion, its neurophysiological substrate remains unclear. Many types of common audio-visual interactions occur within a time window of ∼100 ms. For example, in the sound-induced double-flash illusion, when two beeps are presented within ∼100 ms together with one flash, a second illusory flash is often perceived. Due to their intrinsic rhythmic nature, brain oscillations are one candidate mechanism for gating the temporal window of integration. Interestingly, occipital alpha band oscillations cycle on average every ∼100 ms, with peak frequencies ranging between 8 and 14 Hz (i.e., 120-60 ms cycle). Moreover, presenting a brief tone can phase-reset such oscillations in visual cortex. Based on these observations, we hypothesized that the duration of each alpha cycle might provide the temporal unit to bind audio-visual events. Here, we first recorded EEG while participants performed the sound-induced double-flash illusion task and found positive correlation between individual alpha frequency (IAF) peak and the size of the temporal window of the illusion. Participants then performed the same task while receiving occipital transcranial alternating current stimulation (tACS), to modulate oscillatory activity either at their IAF or at off-peak alpha frequencies (IAF±2 Hz). Compared to IAF tACS, IAF-2 Hz and IAF+2 Hz tACS, respectively, enlarged and shrunk the temporal window of illusion, suggesting that alpha oscillations might represent the temporal unit of visual processing that cyclically gates perception and the neurophysiological substrate promoting audio-visual interactions.

Figure 1

Individual Alpha Frequency Correlates with the Temporal Profile of the Double-Flash Illusion (A) Across-participants average probability of perceiving the illusion plotted as a function of interbeep delay. The red curve represents the sigmoid fit determining the amplitude of the window of illusion, corresponding to the inflection point of the sigmoid. (B) Across-participants average brain topography of oscillatory alpha activity during task performance and corresponding average FFT showing the peak frequency in the alpha band (light blue rectangle). (C) Scalp topography of the correlation index (Pearson’s R) between individual inflection points and alpha peak frequency (IAF) at each electrode, showing maximal correlation (r = 0.697; p < 0.001) around occipital electrodes (O1, O2, Oz). Scatterplot of the significant correlation between each individual’s inflection points (y axis) and the duration of one occipital alpha cycle (i.e., IAF; x axis).

Figure 2

tACS at Different Frequencies Modulates the Size of the Temporal Window of Illusion The main plot shows the sigmoid fit (with aligned inflection points) of the average perceived illusion across participants (y axis) at different interbeep delays (x axis) in the three tACS conditions (Cz-Oz montage): tACS at IAF (black dots/curve), IAF+2 Hz (green dots/curve), and IAF−2 Hz (red dots/curve). Note that all the inflection points fall within the range of alpha frequency band, represented by the light-blue rectangle. Right inset shows the significant shifts of the average inflection points calculated for each participant sigmoid fit as a function of tACS condition. Error bars represent SEM. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001.

Figure 3

tACS at IAF±2 Selectively Modulates the Probability of Double-Flash Illusion around the Inflection Point, without Affecting the Overall Probability To assess whether tACS at different frequencies (IAF+2 versus IAF−2) induces a selective shift of inflection points versus an overall change in proneness to experience the double-flash illusion, we compared via t test the probability of illusion at each interbeep interval in the IAF±2 Hz tACS conditions. The graph shows that t values (y axis) were significantly different between the two tACS conditions only at 108 ms interbeep interval (t(11) = 4.4, p = 0.015), with a trend toward significance at 96 ms interbeep interval (t(11) = 2.98, p = 0.09). The probability of illusion at all the other interbeep intervals did not change between IAF±2 Hz conditions (all t values <2.4, all p values >0.23), demonstrating that tACS selectively shifts the inflection points but not the overall probability of experiencing the illusion.

Figure 4

Alpha Power Is Inversely Correlated with Proneness to Experience the Sound-Induced Double-Flash Illusion Scatterplot of the correlation between average probability of perceiving the illusion across all SOAs (x axis) and individual alpha power at occipital electrodes (O1, O2, Oz; y axis).