Electrical stimulation

During four out of the six blocks in each experimental session, participants received different tACS conditions at α- and γ-frequency delivered to the left and right occipital/parietal cortex. Stimulation was administered in a single-blind design. As stimulation was remote controlled via the MATLAB code that controlled the entire experiment, we decided against a double blinding procedure to allow the experimenter to monitor and ensure the correct and safe tACS application of the system. In each condition, 1 mA (2 mA peak-to-peak) tACS was applied either in the α- (individual α-frequency) or γ- (47 Hz) range and administered to either the left or right occipital cortex. The γ-tACS frequency was chosen to target the center of the lower γ-band (35–65 Hz). During the two remaining blocks no stimulation was applied. All possible stimulation conditions were applied in each experimental session. Contrasting the aforementioned stimulation frequencies and target regions allows to assess the spatial and frequency specificity of tACS effects, as each of these conditions is expected to result in different modulations of participants’ RTs (Fig. 1a). Further this design allows to largely rule out alternative explanations for stimulation effects such as peripheral mechanisms that could give rise to sensory entrainment (e.g. stimulation of the retina or cutaneous nerves)^33–35. Due to the large distance of the stimulation electrodes to the retina, effects of retinal stimulation (phosphenes) should be spatially unspecific, whereas effects caused by stimulation of cutaneous nerves should cause effects opposite to those expected from a stimulation of the brain, as somatosensory input is processed contralaterally to the stimulated hemisphere. Alternatively, these visual or somatosensory sensations could simply distract participants (without entrainment of oscillations in the brain via sensory pathways). Indeed, after the experimental sessions, 13 participants reported sensations merely at the onset of stimulation, which, however, disappeared after few seconds. Assuming that sensations in the two stimulation conditions cause similar distractions, this would likely give rise to changes in RTs into the same direction for both stimulation frequencies and across cue types. In contrast, for a neurophysiological effect of stimulation we expect a differential effect of the two stimulation frequencies that reverses depending on the validity of the cue. Stimulation conditions were performed in randomized order with the constraint that stimulation-free blocks were balanced across the two sessions such that each of the stimulation-free blocks follows one out of the four stimulation conditions (Fig. 1b). Balancing was carried out to avoid systematic influences of carry-over effects from previous tACS conditions^25,26. A post-hoc inspection indicated a relatively equal distribution of stimulation conditions over time/blocks (Supplementary Fig. ^S1), ensuring that outlasting effects commonly reported after tACS administration level out across participants. However, as an exception, α-tACS over the right hemisphere was disproportionately often delivered in the first block. TACS was delivered using a hybrid electrical stimulation/EEG system (Starstim 8, Neuroelectrics, Barcelona, Spain), remote controlled using MATLAB 2016a and the MatNic (Neuroelectrics, Barcelona, Spain) interface. Two pairs of circular, sintered Ag/AgCl electrodes (PiStim, Neuroelectrics, Barcelona, Spain) with a radius of 1 cm were placed over locations O1-P3 and O2-P4 of the international 10–20 system, targeting left or right occipital cortex (Fig. 1c). Electrodes were filled with an electrically conductive gel (Signa Gel, Parker Laboratories Inc., Fairfield, NJ, USA). Impedances were kept below 10 kΩ. Stimulation was started with a 1-s fade-in at the beginning of each block of the attention task and maintained until the end of the respective block (~ 8-min). The first trial of each block started during the ramp-in period. Stimulation was faded-out over a 1-s period. The total stimulation duration across blocks was kept below 40-min per day.

Prior to the experiment, participants’ IAF was determined from a 3-min EEG recording during rest with eyes-closed. Signals were recorded at a rate of 250 Hz from electrode Pz (reference: right earlobe, ground: AFz). The EEG signal was filtered between 1 and 40 Hz and segmented into 1-s epochs. Fast Fourier Transforms (2-s zero-padding, Hanning window) were computed on each of the segments and the resulting power spectra were averaged. The power peak in the 8–12 Hz range was used as stimulation frequency for the α-tACS conditions. Due to the much larger frequency band and the lower signal-to-noise ratio, which hinders time efficient estimation of an individual γ-frequency, γ-tACS was applied at a fixed frequency in the center of the lower gamma band (47 Hz). The hybrid electrical stimulation/EEG device allows for time-efficient application of multiple stimulation and EEG electrodes and flexible remote control over stimulation signals. However, due to insufficient dynamic range of the EEG system, the large stimulation artifact resulted in saturation of the EEG signals precluding any analysis of EEG signals during stimulation²². We further refrained from analyzing task related EEG due to the low spatial sampling available from the system (4-channels).