Individual Alpha Peak Frequency Predicts 10 Hz Flicker Effects on Selective Attention

Abstract

Rhythmic visual stimulation ("flicker") is primarily used to "tag" processing of low-level visual and high-level cognitive phenomena. However, preliminary evidence suggests that flicker may also entrain endogenous brain oscillations, thereby modulating cognitive processes supported by those brain rhythms. Here we tested the interaction between 10 Hz flicker and endogenous alpha-band (∼10 Hz) oscillations during a selective visuospatial attention task. We recorded EEG from human participants (both genders) while they performed a modified Eriksen flanker task in which distractors and targets flickered within (10 Hz) or outside (7.5 or 15 Hz) the alpha band. By using a combination of EEG source separation, time-frequency, and single-trial linear mixed-effects modeling, we demonstrate that 10 Hz flicker interfered with stimulus processing more on incongruent than congruent trials (high vs low selective attention demands). Crucially, the effect of 10 Hz flicker on task performance was predicted by the distance between 10 Hz and individual alpha peak frequency (estimated during the task). Finally, the flicker effect on task performance was more strongly predicted by EEG flicker responses during stimulus processing than during preparation for the upcoming stimulus, suggesting that 10 Hz flicker interfered more with reactive than proactive selective attention. These findings are consistent with our hypothesis that visual flicker entrained endogenous alpha-band networks, which in turn impaired task performance. Our findings also provide novel evidence for frequency-dependent exogenous modulation of cognition that is determined by the correspondence between the exogenous flicker frequency and the endogenous brain rhythms.SIGNIFICANCE STATEMENT Here we provide novel evidence that the interaction between exogenous rhythmic visual stimulation and endogenous brain rhythms can have frequency-specific behavioral effects. We show that alpha-band (10 Hz) flicker impairs stimulus processing in a selective attention task when the stimulus flicker rate matches individual alpha peak frequency. The effect of sensory flicker on task performance was stronger when selective attention demands were high, and was stronger during stimulus processing and response selection compared with the prestimulus anticipatory period. These findings provide novel evidence that frequency-specific sensory flicker affects online attentional processing, and also demonstrate that the correspondence between exogenous and endogenous rhythms is an overlooked prerequisite when testing for frequency-specific cognitive effects of flicker.

Keywords: SSVEP; alpha oscillations; attention; entrainment; flicker; steady-state visual-evoked potentials.

Figure 1.

Stimuli and task. A, Trials began with the presentation of a mask consisting of hash marks for 2000 ms followed by imperative stimulus, presentation of which lasted until a button press or until the deadline of 1200 ms was exceeded, and was followed by an intertrial interval (ITI) of 1000 ms. B, Each stimulus consisted of a target letter flickering within (10 Hz) or outside (7.5 or 15 Hz) the alpha-band range, while flanker letters flickered outside (7.5 or 15 Hz) or within alpha-band range respectively. C, Frequency tagging was implemented by sine-wave modulation of stimulus luminance.

Figure 2.

Topographical maps of frequency- and stimulus-specific SSVEP spatial filters and frequency spectra of RESS component time series. A, RESS topographical maps for different tagging frequencies at target and flanker positions. Note that topographical distribution of SSVEPs elicited by the target is more centrally focused than that of the flankers. B, Frequency spectra of RESS component time series, expressed in SNR (signal-to-noise ratio) units, highlights frequency-specificity of RESS SSVEPs. Note that 10 Hz target and 10 Hz flanker spectra are based on averaging 10T/xF and xT/10F conditions (where x is 7.5 or 15 Hz).

Figure 3.

Occipital alpha sources. A, Locations of equivalent dipoles for occipital alpha ICs. Each blue dot represents a participant. The red dot represents the centroid of the cluster based on all participant dipoles, the scalp projection of which is depicted in the top right. B, Power spectra of alpha IC time series, normalized to the power of each participant's alpha peak (for comparability across participants). Each row represents a participant, color corresponds to normalized spectral power, and the white dot denotes alpha peak frequency.

Figure 4.

Behavioral performance. A, Mean RT and (B) error rates on congruent versus incongruent trials plotted as a function of condition. Gray barplots in each graph represent congruency effect (RT and error-rate difference between incongruent and congruent trials). Error bars denote SEM.

Figure 5.

Conflict-related theta-band power. A, Condition-average changes in power relative to the baseline period (−500 to −200 ms) at electrodes FCz and Cz (top), and theta component derived using source separation (bottom; see Materials and Methods for details). B, Theta-band (3–7 Hz) power distribution over the scalp in the time-frequency window indicated by the white square in A, and forward model of theta component. C, Condition-specific changes in theta-band power (data taken from the theta component).

Figure 6.

Attentional modulation of SSVEPs. Grand-average SSVEP amplitudes (determined from 500 to 2600 ms time window and expressed in SNR units) for each stimulus type (target and flankers) and each flicker frequency. The data plotted here is taken from the frequency spectra depicted in Figure 2, except that 10 Hz target flicker conditions are separated. Error bars denote SEM. *p < 0.05; n.s. indicates p > 0.05.

Figure 7.

Single-trial analysis results. A, Graphical representation of the best-fitting statistical model (fixed effects only), where: congruency_CONGRUENT indicates comparison between congruent and incongruent trials; congruency_CONGRUENT:alphaABS indicates differences in the effect of alphaABS on incongruent versus congruent trials; condition_10T/15F indicates comparison between conditions 10T/7.5F and 10T/15F; condition_7.5T/10F indicates comparison between conditions 10T/7.5F and 7.5T/10F; condition_15T/10F indicates comparison between conditions 10T/7.5F and 15T/10F. Error bars indicated 95% confidence intervals. Numbers denote fixed-effects coefficients. Statistical significance of fixed-effects coefficients is marked with asterisk symbols, where *p < 0.05, **p < 0.01, ***p < 0.001, and nonsignificant coefficients are marked with empty circles. B, Graphical summary of fixed effects marked with a green rectangle in A, which illustrates that closer match between IAF and 10 Hz flicker resulted in overall slower RTs, and that this effect was stronger for incongruent trials (steeper regression line slope). Shaded areas represent 95% confidence intervals around the slope of regression line. Note that for LME modeling log-transformed RTs were used; however, here mean RTs are left in original units for interpretation clarity.