Inconsistent Effects of Parietal α-tACS on Pseudoneglect across Two Experiments: A Failed Internal Replication

Abstract

Transcranial electrical stimulation (tES) is being investigated as an experimental and clinical interventional technique in human participants. While promising, important limitations have been identified, including weak effect sizes and high inter- and intra-individual variability of outcomes. Here, we compared two "inhibitory" tES-techniques with supposedly different mechanisms of action as to their effects on performance in a visuospatial attention task, and report on a direct replication attempt. In two experiments, 2 × 20 healthy participants underwent tES in three separate sessions testing different protocols (10 min stimulation each) with a montage targeting right parietal cortex (right parietal-left frontal, electrode-sizes: 3cm × 3cm-7 cm × 5 cm), while performing a perceptual line bisection (landmark) task. The tES-protocols were compared as to their ability to modulate pseudoneglect (thought to be under right hemispheric control). In experiment 1, sham-tES was compared to transcranial alternating current stimulation at alpha frequency (10 Hz; α-tACS) (expected to entrain "inhibitory" alpha oscillations) and to cathodal transcranial direct current stimulation (c-tDCS) (shown to suppress neuronal spiking activity). In experiment 2, we attempted to replicate the findings of experiment 1, and establish frequency-specificity by adding a 45 Hz-tACS condition to α-tACS and sham. In experiment 1, right parietal α-tACS led to the expected changes in spatial attention bias, namely a rightward shift in subjective midpoint estimation (relative to sham). However, this was not confirmed in experiment 2 and in the complete sample. Right parietal c-tDCS and 45 Hz-tACS had no effect. These results highlight the importance of replication studies, adequate statistical power and optimizing tES-interventions for establishing the robustness and reliability of electrical stimulation effects, and best practice.

Keywords: landmark task; replication; tACS; tDCS; tES reliability.

FIGURE 1

(A) Schematic representation of a single trial structure. A fixation cross was presented for 1000 ms, followed by a transected line flashed for 150 ms. The fixation cross reappeared for the response time period. (B) Schematic representation of the experimental procedure of both experiments. On each day, after a practice session (P) and transcranial electrical stimulation (tES) set up, participants were asked to complete six blocks of the landmark task. During block 3 and block 4 participants received active tES or sham in a counter-balanced order across days. (C) Results of electric field distribution modeling are shown together with the electrode position (F3 in red, P6 in gray). The electric field strength is scaled from 0 (blue) to maximum (red). The simulation indicates diffuse activation within the right parietal cortex together with a more restricted activation of a left frontal region.

FIGURE 2

Test re-test reliability of spatial bias in perceptual line bisection (point of subjective equality, PSE) for experiment 1 (A) and experiment 2 (B). Test–retest reliability is significant across three measurement points collected before tES (baseline) at three different days, as indicated by both Spearman’s rho and Shepherd’s Pi values (see lower right corner of each graph). x- and y-values represent the distance from the veridical center in pixels (dashed lines represent 95% confidence intervals).

FIGURE 3

Spatial bias in perceptual line bisection (PSE during landmark task performance) over time (six blocks) in three experimental sessions varying in terms of the tES protocol used (experiment 1). tES was applied with a right parietal – left frontal montage for 10 min using either α-tACS, c-tDCS or sham tES during blocks 3–4 (B3–4, shaded area). (A) Overall group mean PSE values (19 participants, within-subject design) as a function of blocks (B1–6) and tES (line plots). During the second block of tES (block 4), α-tACS significantly differed from sham and c-tDCS. The asterisk indicates a significant difference between Sham and α-tACS (p < 0.05). (B) Performance split according to “rightward shifters” (upper main graph, 14/19 participants) and “leftward shifters” (lower inset, 5/19 participants) during α-tACS (as compared to sham in block 4). The pattern of results is consistent with α-tACS having shifted spatial bias rightward (reduced pseudoneglect) in the majority of participants. (C) Performance split according to “rightward shifters” (upper main graph, 12/19 participants) and “leftward shifters” (lower inset, 7/19 participants) during c-tDCS (as compared to sham in block 4). No effects of active stimulation (c-tDCS vs. sham) are discernible in this condition and population. The error bars represent 95% confidence interval corrected for a within subjects design (Cousineau, 2005).

FIGURE 4

(A) Spatial bias in perceptual line bisection (PSE during landmark task performance) over time (six blocks) as a function of tES condition in experiment 2. The montage, intensity and total stimulation duration were identical to experiment 1. tES conditions consisted of α-tACS, 45 Hz (γ-tACS) or sham tES applied during blocks 3–4 (B3-4, shaded area). PSE values (20 participants, within-subject design) showed no significant difference across conditions in this population. Error bars represent 95% confidence interval corrected for a within subjects design. (B) Performance split according to “rightward shifters” (right panel, 9/20 participants) and “leftward shifters” (left panel, 11/20 participants) during α-tACS (as compared to sham in block 4). Unlike experiment 1, α-tACS reduced pseudoneglect in 45% of the population (inconsistent, random distribution).

FIGURE 5

Spatial bias in perceptual line bisection (PSE) over time (six blocks) as a function of tES condition when data from experiment 1 and 2 were collapsed over common conditions (Sham, α-tACS). PSE values showed no difference across conditions in the overall population (39 participants). Error bars represent 95% confidence interval corrected for a within subjects design.