Effects of tDCS during inhibitory control training on performance and PTSD, aggression and anxiety symptoms: a randomized-controlled trial in a military sample

Background Post-traumatic stress disorder (PTSD), anxiety, and impulsive aggression are linked to transdiagnostic neurocognitive deficits. This includes impaired inhibitory control over inappropriate responses. Prior studies showed that inhibitory control can be improved by modulating the right inferior frontal gyrus (IFG) with transcranial direct current stimulation (tDCS) in combination with inhibitory control training. However, its clinical potential remains unclear. We therefore aimed to replicate a tDCS-enhanced inhibitory control training in a clinical sample and test whether this reduces stress-related mental health symptoms. Methods In a preregistered double-blind randomized-controlled trial, 100 active-duty military personnel and post-active veterans with PTSD, anxiety, or impulsive aggression symptoms underwent a 5-session intervention where a stop-signal response inhibition training was combined with anodal tDCS over the right IFG for 20 min at 1.25 mA. Inhibitory control was evaluated with the emotional go/no-go task and implicit association test. Stress-related symptoms were assessed by self-report at baseline, post-intervention, and after 3-months and 1-year follow-ups. Results Active relative to sham tDCS neither influenced performance during inhibitory control training nor on assessment tasks, and did also not significantly influence self-reported symptoms of PTSD, anxiety, impulsive aggression, or depression at post-assessment or follow-up. Conclusions Our results do not support the idea that anodal tDCS over the right IFG at 1.25 mA enhances response inhibition training in a clinical sample, or that this tDCS-training combination can reduce stress-related symptoms. Applying different tDCS parameters or combining tDCS with more challenging tasks might provide better conditions to modulate cognitive functioning and stress-related symptoms.


Influence of baseline inhibitory control on tDCS effects in SST training
The SSRT change score (ΔSSRT = SSRT at session 5 -SSRT at session 1) was regressed on preassessment Go/no-go scores (RT and accuracy, separately) together with the predictor Stimulation group, and their interaction. The results are presented in Table S.1. Go-RT and No-go accuracy did not interact significantly with Stimulation group in predicting SSRT change, suggesting that baseline go/no-go performance did not influence tDCS effects on SSRT enhancement. No-go accuracy did have a main effect on the SSRT change score. Correlation analysis showed that lower no-go accuracy at baseline was associated with stronger SSRT improvements during training (r = .24, p = .036), implying that worse inhibitory control performance at baseline may leave more room for performance improvement during inhibitory control training.

IAT -Quad model details and outcomes
The Quad model (Conrey et al., 2005) includes the following components that drive response behavior in the IAT: association activation (AC: "the likelihood that automatic bias is activated by a stimulus"), discriminability (D: "the likelihood that a correct response can be determined"), overcoming bias (OB: "the likelihood that automatic bias is overcome"), and guessing (G: "the likelihood that, in the absence of other information, a guessing bias drives responses). The D and G parameters were defined for target words and attribute words separately (i.e., for target words: Dtarget and Gtarget, and for attribute words: Dattribute and Gattribute). A single parameter was defined for AC and OB, as bidirectional associations were assumed (Greenwald et al., 2002). The model was fitted on the number of correct and incorrect responses per trial category and task phase from all participants, separately for the preand post-assessment and for the active tDCS and sham groups.
To test group differences in the overcoming bias (OB) parameter at post-assessment, we tested the free model (for parameter estimations, see Table S

Dot-probe task
The dot-probe task measures attentional biases for threat. In this task, a pair of face cues (one angry face and one neutral face) were presented on a computer screen, divided over the upper and lower half of the screen. After a variable cue-stimulus interval (CSI; 200, 400, 600, 900 or 1200 ms), the face cues were replaced by a probe stimulus ('>>>' or '<<<') and a distractor stimulus ('\/\/' or '/\/\').
Participants were instructed to identify the direction of the probe stimulus (left or right) as fast as possible by pressing the correct button on the keyboard: 'F' or 'J'. The probe stimulus randomly appeared in the angry face cue location or in the neutral face cue location. An attentional bias toward the threat (here: angry face) location is induced by the fast attention capture of threat cues, especially at short CSIs (Cisler & Koster, 2010). A threat avoidance bias is also found in PTSD and anxiety patients, especially at longer CSIs. The attentional bias is computed as the RT difference between stimuli in the threat vs. neutral location (RT threat -RT neutral). The dot-probe task was only assessed post-intervention.
Results. Data were not available for 8 participants who did not complete the dot-probe task, leaving a sample of 88 for attention bias analysis (42 active tDCS, 46 sham). The attentional bias scores across CSI durations showed a very low split-half reliability of r = -0.10. This is not surprising in light of recent insights: dot-probe performance does often not reliably measure attentional bias (McNally, 2019), although this task has also shown reliable results (see e.g. [Gladwin & Vink, 2020]).
Considering that the tDCS intervention could have modulated aspects of dot-probe task performance, we carried out the preregistered analysis in spite of the low reliability.
Overall, both active tDCS and sham groups showed very small attentional bias scores that did not significantly differ from zero (attentional bias score in ms -active tDCS: M = 5.3 ±27.7; sham: -