No Evidence for Phase-Specific Effects of 40 Hz HD-tACS on Multiple Object Tracking

Abstract

Phase synchronization drives connectivity between neural oscillators, providing a flexible mechanism through which information can be effectively and selectively routed between task-relevant cortical areas. The ability to keep track of objects moving between the left and right visual hemifields, for example, requires the integration of information between the two cerebral hemispheres. Both animal and human studies have suggested that coherent (or phase-locked) gamma oscillations (30-80 Hz) might underlie this ability. While most human evidence has been strictly correlational, high-density transcranial alternating current stimulation (HD-tACS) has been used to manipulate ongoing interhemispheric gamma phase relationships. Previous research showed that 40 Hz tACS delivered bilaterally over human motion complex could bias the perception of a bistable ambiguous motion stimulus (Helfrich et al., 2014). Specifically, this work showed that in-phase (0° offset) stimulation boosted endogenous interhemispheric gamma coherence and biased perception toward the horizontal (whereby visual tokens moved between visual hemifields-requiring interhemispheric integration). By contrast, anti-phase (180° offset) stimulation decreased interhemispheric gamma coherence and biased perception toward the vertical (whereby tokens moved within separate visual hemifields). Here we devised a multiple object tracking arena comprised of four quadrants whereby discrete objects moved either entirely within the left and right visual hemifields, or could cross freely between visual hemifields, thus requiring interhemispheric integration. Using the same HD-tACS montages as Helfrich et al. (2014), we found no phase-specific effect of 40 Hz stimulation on overall tracking performance. While tracking performance was generally lower during between-hemifield trials (presumably reflecting a cost of integration), this difference was unchanged by in- vs. anti-phase stimulation. Our null results could be due to a failure to reliably modulate coherence in our study, or that our task does not rely as heavily on this network of coherent gamma oscillations as other visual integration paradigms.

Keywords: coherence; gamma; interhemispheric integration; multiple object tracking; phase-locking; transcranial alternating current stimulation.

Figure 1

Manipulating ongoing phase relationships with multifocal stimulation. The tACS output is split into multiple sites of stimulation, changing the phase relationship between cerebral hemispheres. (A) Bilateral 4 × 1 ring electrodes allowed for the application of perfectly in-phase (0° offset) tACS over the target area (human motion complex, V5; centroid electrodes positioned at P7/PO7 and P8/PO8), where the centroid electrodes continuously share current of the same polarity. (B) Anti-phase tACS applied at the same scalp locations, where centroid electrodes continuously share current of the opposite polarity (180° offset). (C) Realistic simulations of current flow for the two montages (Soterix HD-Explore software).

Figure 2

Manipulating object boundaries to change interhemispheric transfer demands. (A) With objects deflecting off the horizontal bar but passing over the vertical bar (darkened for illustration only), objects freely move between visual hemifields—bound only to the two uppermost or lowermost quadrants. (B) With objects deflecting off the vertical bar but passing over the horizontal bar, objects are bound within separate visual hemifields (illustrated by red and blue shaded areas over the left and right quadrants).

Figure 3

Trial sequence and experimental design. (A) Participants were asked to fixate centrally while four targets were cued (one in each quadrant). The cue disappeared during the pretrial period before all identical objects (comprised of four targets and four non-targets) began to move, deflecting linearly off the horizontal bar (between-hemifield trials) or the vertical bar (within-hemifield trials). Participants then chose the four objects they believed were the cued targets before receiving feedback (green, correct; red, incorrect). (B) Counterbalanced across participants, the two experimental sessions were each comprised of a sham block (always first; gray) and active stimulation block (either in-phase, green; or anti-phase, yellow). The preceding sham block set the baseline for each session, and captured any training effect across sessions. The two sessions were conducted 1 week apart. All blocks were 15 min in length, with 5-min breaks between blocks. There was an opportunity for participants to have a small break (1 min or less) within each block, but stimulation continued throughout this break.

Figure 4

No phase-specific effect of stimulation on overall tracking performance. The cost of integration (within-hemifield performance minus between-hemifield performance) was not significant for the pooled sham conditions. However, this became highly significant during both in-phase and anti-phase tACS (^***ps < 0.001). This may reflect a general effect of time (since sham always preceded an active block), but only within-hemifield trials improved during stimulation (p < 0.001), with between-hemifield performance generally decreasing (p = 0.369). This larger cost of integration may therefore reflect a detrimental effect of tACS (both in-phase and anti-phase) on between-hemifield tracking. Error bars represent within-participant 95% confidence intervals.

Figure 5

Mixed evidence for an emergent phase-specific effect of stimulation. To capture an emergent effect of stimulation, rank correlations were computed between trial number and performance (where positive correlations suggest a general increase in performance over time). For between hemifield trials, performance generally improved more over time during in-phase vs. anti-phase tACS (^**p < 0.01), with no phase-specific effect of tACS on the rank correlations observed during within-hemifield trials. However, the interaction was not significant. Error bars represent within-participant 95% confidence intervals. Fisher's z-transformation (ρ to ρ') was performed on the rank correlations, though the outcomes were unchanged by this transformation.