If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
High-definition transcranial direct current stimulation (HD-tDCS) for the enhancement of working memory – A systematic review and meta-analysis of healthy adults
Department of Psychiatry, Psychotherapy and Psychosomatics, Faculty of Medicine, RWTH Aachen, Pauwelsstraße 30, Aachen, 52074, North Rhine-Westphalia, Germany
Department of Psychiatry, Psychotherapy and Psychosomatics, Faculty of Medicine, RWTH Aachen, Pauwelsstraße 30, Aachen, 52074, North Rhine-Westphalia, GermanyInstitute of Neuroscience and Medicine, JARA-Institute Brain Structure Function Relationship (INM 10), Research Center Jülich, Wilhelm-Johnen-Straße, 52438, Jülich, Germany
Department of Psychiatry, Perelman School of Medicine at the University of Pennsylvania, 3535 Market Street, Suite 3080, Philadelphia, PA, 19104-3309, USA
Department of Psychiatry, Psychotherapy and Psychosomatics, Faculty of Medicine, RWTH Aachen, Pauwelsstraße 30, Aachen, 52074, North Rhine-Westphalia, Germany
Overall, no significant effect of left prefrontal HD-tDCS on WM performance emerged.
•
HD-tDCS effects on WM performance are heterogeneous.
•
Combination of HD-tDCS with multiple training sessions improves effectiveness.
•
Within-subject designs increase the homogeneity of results across studies.
Abstract
Background
High-definition transcranial direct current stimulation (HD-tDCS) administers weak electric current through multiple electrodes, enabling focal brain stimulation. An increasing number of studies investigate the effects of anodal HD-tDCS on the enhancement of working memory (WM). The effectiveness of the technique is, however, still unclear.
Objective/hypothesis
This systematic review analyzed the current literature on anodal HD-tDCS for WM enhancement, investigating its effectiveness and the influence of different moderators to allow for comparison with conventional tDCS.
Methods
Following the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines, a comprehensive literature review was conducted using PubMed, Web of Science, and Scopus. Sixteen single- or double-blind, sham-controlled studies were included in the review. Eleven studies were included in the meta-analysis, focusing solely on stimulation of the left prefrontal cortex (PFC).
Results
No significant effect of anodal HD-tDCS on the left PFC for WM accuracy (g = 0.23, p = 0.08), and reaction time (g = 0.03, p = 0.75 after trim-and-fill) was found. Further analysis revealed heterogeneity in the accuracy results. Here, moderator analysis indicated a significant difference between studies that repeatedly used HD-tDCS enhanced WM training and studies with one-time use of HD-tDCS (p < 0.001), the latter having a smaller effect size. Another moderator was the research design, with differences between within-subjects-, and between-subjects designs (p < 0.05). Within-subject studies showed lower effect sizes and substantially lower heterogeneity. Qualitative analysis reinforced this finding and indicated that the motivation of the participant to engage in the task also moderates the effectiveness of HD-tDCS.
Conclusion
This review highlights the importance of inter-individual differences and the setup for the effectiveness of anodal, HD-tDCS augmented WM training. Limited evidence for increased sensitivity of HD-tDCS to these factors as compared to conventional tDCS is provided.
]. Here, WM is divided into the phonological loop (auditory processing), the visuospatial sketchpad (visual processing), and the central executive which coordinates attentional control [
]. WM is essential for complex cognitive functions such as language acquisition, mathematic calculations, or problem-solving, and disruptions of WM translate to deficits in various daily activities [
]. Transcranial direct current stimulation (tDCS) presents a promising NIBS technique to modulate brain activity. Conventional tDCS is administered via two electrodes that are externally applied to the head and administer positive or negative currents by an anode or cathode, respectively. In its conventional protocol, and when targeting the primary motor cortex, anodal tDCS enhances neuronal excitability while cathodal tDCS is shown to decrease it [
Feredoes E, Delgado Garcia JM, Ae FF, et al. Anodal transcranial direct current stimulation of prefrontal cortex enhances working memory. doi:10.1007/s00221-005-2334-6.
]. While the excitatory effects of anodal stimulation on cognitive functions are quite robust, the inhibitory effects of cathodal stimulation are not [
]. An explanation might be found in the distributed nature of networks constituting cognitive functions, potentially providing compensatory mechanisms to counter inhibition, or even benefitting from it through decreased neuronal competition [
]. Consequently, an increasing amount of research is investigating the modulation of these functions using tDCS, mostly focusing on anodal stimulation [
]. Findings also demonstrate the potential of tDCS to modulate human cognition but heterogeneity in the results of its application presents an obstacle to its use as a standardized treatment. Presumably, the reason can be found in the influence of individual differences on tDCS effects [
]. Despite the heterogeneity that is introduced by these factors, a meta-analysis that investigated the effect of anodal tDCS on WM improvement reported a small significant effect [
]. While a general effect of tDCS on WM performance did not withstand correction for publication bias, an effect of tDCS in combination with WM training remained significant [
]. TDCS effects emerged exclusively following anodal stimulation of the left dorsolateral prefrontal cortex (DLPFC), but not after stimulation of the right DLPFC and the parietal cortex.
Current fMRI literature shows that the left DLPFC is robustly activated during WM tasks [
]. Presumably, different parts of this network specialize for different sub-functions of WM, with the PFC constituting a region of particular relevance [
]. Support comes from brain lesion studies, indicating that damage to the left DLPFC results in deficits in the manipulation of information held in WM, an ability that is similar to the proposed function of the central executive [
]. The central executive represents higher-level WM operations independent of task modality, allocating attention to, and integrating information from other WM sub-systems.
With recent advancements in NIBS technology, it is possible to target brain areas more focally using High-Definition tDCS (HD-tDCS). HD-tDCS is using a setup of multiple smaller electrodes instead of the two larger pad-electrodes of conventional tDCS. The most common electrode placement is a 4 × 1 ring configuration with a central electrode surrounded by four return electrodes. The four return electrodes help to isolate the stimulated region and enable more focal brain stimulation with longer-lasting effects compared to conventional tDCS [
]. While application of HD-tDCS to improve WM is becoming more prevalent, evaluation of its effectiveness is aggravated by heterogeneity in the results [
Focalised stimulation using high definition transcranial direct current stimulation (HD-tDCS) to investigate declarative verbal learning and memory functioning.
Up to date, no meta-analysis or systematic review about HD-tDCS effects on WM exists. The current meta-analysis aims to fill this gap and provide an overview of the increasing number of HD-tDCS studies intended to enhance WM. Due to the excitatory effects of anodal tDCS and evidence from conventional tDCS research, the meta-analysis will focus on anodal HD-tDCS only. To enable comparison with a previous meta-analysis of conventional tDCS [
], this meta-analysis partially adapts its methods and focuses on anodal HD-tDCS targeting the left DLPFC. This meta-analysis is embedded in a systematic review aimed to explore the effects of HD-tDCS on WM and the influence of potential moderators such as training, task performance before and during the task, and WM task.
2. Methods
2.1 Search strategy
The review followed the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines (Fig. 1) [
Page MJ, Mckenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Research Methods and Reporting. Published online 2021. doi:10.1136/bmj.n71.
]. The online databases PubMed, Web of Science, and Scopus were searched between the 9th and the August 13, 2021. The search was complemented by a hand search on google scholar which yielded no additional results. Due to the novelty of HD-tDCS, it was not necessary to restrict the time of publication.
Fig. 1PRISMA flow diagram of the literature search.
The reference sections of articles were screened for additional literature. The languages of the articles were restricted to German and English. The primary outcome measure was accuracy, the secondary reaction time (RT). Using the same search method, the literature research was updated on the 15th of October 2021 and on the 10th of May 2022, restricting publication dates of articles to 2021 and onwards. Sixteen studies were included in the final analysis, eleven in the meta-analysis.
2.3 Eligibility criteria
Participants had to be healthy adults (age ≥18 years). Studies that were limited to patient cohorts were excluded. Experiments comparing HD-tDCS effects in clinical samples and healthy controls were included, using only the data of healthy controls.
Included studies had to use double-blind or single-blind, sham-controlled research designs. Both English and German papers were eligible. Due to a lack of German studies, only English studies were included. Both within- and between-subject designs were included. Studies using only conventional tDCS or cathodal stimulation were excluded. Case studies were also excluded.
2.4 Outcome variables
Included studies used at least one WM task, either as a primary or a secondary outcome. These included n-back tasks, Sternberg tasks, Operation Span tasks, and a delayed-response WM task. If possible, and as a substitute for accuracy, metrics from signal detection theory such as d’ and k’ were used due to their advantage to protect against bias. Additionally, RT measures were used for the secondary analysis. If data was not available, the corresponding author was contacted via email to request additional information. Eventually, data was obtained from published figures with the ImageJ software [
The meta-analysis included 10 studies that performed HD-tDCS of the left DLPFC, and one study on the left ventrolateral prefrontal cortex (VLPFC). The latter was included due to an apparent overlap between electric field (EF) simulations of this setup, and the ones used to stimulate the DLPFC. The analysis was performed using R (2020). The methods were based on a recent publication providing a comprehensive explanation, codes, and packages for conducting a meta-analysis in R35. Based on the “metagen” package [
], a random-effects model was chosen to pool effect sizes, which were weighted using the inverse variance method. This decision was based on the heterogeneity of study parameters included in the analysis (e.g.: different WM measurements, different current strengths, different stimulation durations). The whole code can be found in Supplement 1. The primary meta-analysis focused on accuracy with a secondary analysis conducted on RT.
2.6 Effect size
For each study, the standardized mean difference and the pooled standard deviation were calculated using the difference between verum and sham conditions. If data on the differences between pre-and post-tests was available, this was chosen instead. For the effect size, Hedges' g was used [
]. Hedges' g is interpreted similar to Cohen's d, with effect sizes 0.2, 0.5, and 0.8 representing small, medium, and large effects, respectively. In comparison to Cohen's d, Hedges' g can adjust for the sample size of the studies, countering the potential bias introduced by studies with small sample sizes [
], the standard formula to compute the effect sizes was adapted to include the observed standard deviation of the sham group for both verum and sham stimulation. This approach analyzes the effect size of a study relative to normal variability (i.e. the variability in the sham group). The formula is depicted below.
Using the inverse-variance method, the effect sizes were pooled. The between-study variance was estimated with the Q-profile method [
], quantifying the percentage to which the observed Q value exceeds the expected Q value under the assumption of homogeneity, and the statistic, which quantifies the variance of the data's true effect size. All steps are included in the metagen package [
Most studies in the meta-analysis included multiple effect sizes. Here, different effect sizes were calculated and then averaged to provide one effect size for each study. This approach was chosen to satisfy the independence assumption. While this approach prevents underestimation of heterogeneity, collapsing different effect sizes into one can also eliminate potentially relevant information on factors influencing the effectiveness of HD-tDCS. An overview of the data can be found in Supplement 2.
2.9 Moderators
To identify potential moderators, prior meta-analyses on conventional tDCS were consulted and adapted, leading to the following moderators [
Effects of anodal transcranial direct current stimulation on working memory: a systematic review and meta-analysis of findings from healthy and neuropsychiatric populations.
Inspection of the data also revealed a sixth potential moderator, the study design. Hence, it was included post-hoc in the moderator analysis.
6.
Study design (within-vs between-subjects design)
Using a mixed effect model, a moderator analysis was conducted. Here, each study was allowed to contribute multiple effect sizes, making use of all available data.
2.10 Qualitative analysis
Using an explorative approach, all 16 studies were included in the analysis. Articles were analyzed for potential moderators which then informed the creation of a coding scheme. The results of the meta-analysis were also incorporated into the coding scheme. In an iterative process, this scheme was then used to code the articles. The following codes were chosen: Current density, stimulation duration, age and sex of participants, blinding, baseline differences, outcome variable, task modality, anode placement, congruency of training task and outcome task, task difficulty, multiple HD-tDCS training sessions.
3. Results
The meta-analysis investigated the effects of anodal HD-tDCS of the left DLPFC on WM performance. The second part of the analysis includes a qualitative assessment of factors influencing the effectiveness of HD-tDCS on WM. An overview of the included studies can be found in Table 1.
Table 1Summary of studies included in the systematic review.
Studies employed an online WM task during each training session.
No increase in performance one day after the training program. Participants receiving active stimulation performed better in the shape 6-back (transfer task) one day after the end of training compared to the sham condition but not in the verbal 6-back.
Significant increase in WM performance for verbal and shape n-back. Higher learning rates during HD-tDCS augmented WM training.
Naka et al., 2018
10|10
F3/Fz,Fp1,C3,F7
0.375 mA/cm2
16
Online & Offline
Between- Participants, Single-blind
3-back
Single Session
Significant increase in hit rate for verbal n-back, but not for the auditory n-back task both online and offline.
Nikolin et al., 2015
26
F3/AF3,F5,FC3,F1
0.32 mA/cm2
20
Offline
Within- Participants, Single-blind
3-back
Single Session
No significant effect.
Hill et al., 2017
20
F3/Fz,Fp1,C3,F7
0.318 mA/cm2
20
Offline
Within- Participants, Single-blind
2-back
Single Session
No significant effect.
Lu et al., 2021
22|21
F3/AF3,F5,FC3,F1
N.A.
20
Offline
Between- Participants, Single-blind
2-back
9 Sessions
No significant effect.
Maldonado & Bernard, 2021
26|26
F3,F5/FP1,F4,P3, C2, CP2, TP7, F9
N.A.
20
Offline
Between- Participants, Single-blind
Sternberg verbal
Single Session
Interaction of anodal stimulation and WM-load on RT with higher RT in the high-load condition. Close-to-significant interaction of anodal stimulation and WM-load on the accuracy, with higher accuracy under medium load.
Splittgerber et al., 2020
24
AF7,AF3,F3/Fp2, T7
0.637 mA/cm2
20
Online
Within- Participants, Single-blind
2-back
Single Session
Interaction of stimulation and initial task performance, with low-performers increasing in performance and high performers decreasing in performance.
Weintraub-Brevda & Chua, (2019)
20|20
F7/F9,F5,FT7,FC5
N.A.
20
Online
Between- Participants, Single-blind
Delayed response task.
Single Session
Significantly increased accuracy compared to sham stimulation.
Gözenman & Berryhill, (2016)
24
P4/Pz,C4,P8,O2 P3/Pz,C3,P7,O1
1.184 mA/cm2
20
Online
Within- Participants, Single-blind
Retro-Cue task.
Single Session
Interaction effect between WM capacity and stimulation, with individuals having low WM capacities benefitting from HD-tDCS.
A table with the risks of biases for each domain and for each study can be seen in Fig. 2. Overall, there were some concerns for a risk of bias. This risk was mainly apparent for the randomization process as well as the selection of the reported results. The main problem for the randomization was the lack of specification about (1) how random allocation to groups was achieved in parallel designs or (2), how sessions were randomized in crossover designs. In the absence of any specific information, some concerns were apparent. For the potential bias due to selectively reported results, the main concern was the lack of availability of a pre-defined analysis plan.
Fig. 2Risk of Bias. D1 refers to bias due to the randomization. D2 refers to bias due to deviations from the intended interventions. DS refers to bias arising from period and carryover effects (only applicable for crossover designs). D3 refers to bias due to missing outcome data. D4 refers to bias due to measurement of the outcome. D5 refers to bias due to the selection of the reported results.
11 studies were included in the meta-analysis (Supplement 3). The results indicated a non-significant trend towards improvement of WM performance in the group that received anodal HD-tDCS over the left PFC as compared to sham: Hedges'g = 0.283, p = 0.082, 95%CI [-0.0431; 0.6084]. An overview can be seen in Fig. 3. The data indicated heterogeneity: (10) = 18.85, p = 0.049, = 45.5%. The funnel plot showed a slight skew of the data (Fig. 4) and analysis with Egger's test indicated asymmetry (t = 2.871,p = 0.019). Subsequent use of the trim-and-fill procedure filled in one study, decreasing the overall effect size to: Hedges'g = 0.228, p = 0.189, 95%CI [-0.1311; 0.5876]. Heterogeneity was significant: (11) = 23.58, p = 0.015, = 53.3%. Due to the sensitivity of the trim-and-fill procedure to high-heterogeneity [
Fig. 4Funnel Plot of publication bias for High-Definition transcranial direct current stimulation of the left dorsolateral prefrontal cortex for accuracy.
10 studies were included for the meta-analysis of the RT (Supplement 4). The results did not show a significant effect of anodal HD-tDCS on WM improvement: Hedges'g = 0.112, p = 0.157, 95%CI [-0.0518; 0.2751] and showed no signs of heterogeneity: (9) = 4.28, p = 0.89, = 0.0% (Fig. 5). Egger's test showed asymmetry (t = 3.215,p = 0.003) and the trim-and-fill procedure filled in three studies, decreasing the effect size to: Hedges'g = 0.028, p = 0.746, 95%CI [-0.1529; 0.2079] (Fig. 6).
Fig. 5Forest plot of anodal left dorsolateral prefrontal cortex stimulation for reaction time.
Fig. 6Funnel Plot of publication bias for High-Definition transcranial direct current stimulation of the left dorsolateral prefrontal cortex for reaction time. White datapoints indicate studies that were imputed by the trim-and-fill procedure.
Effects of single versus dual-site High-Definition transcranial direct current stimulation (HD-tDCS) on cortical reactivity and working memory performance in healthy subjects.
]. For accuracy, a significant difference between studies applying HD-tDCS during multiple online WM training sessions and studies with one-time use of HD-tDCS emerged: Q(1) = 16.34, p < 0.0001. The effect of the two studies that augmented WM training with HD-tDCS was significant: Hedges'g = 0.859, 95%CI = [0.6569; 1.0608], p<0.001, with close to no heterogeneity: Q(1) = 0.27, I2 = 0.0%. The same moderator was also found for RT; Q(1) = 16.02, p < 0.0001. Studies that augmented HD-tDCS with multiple online WM training sessions significantly decreased RT: Hedges'g = 0.570, 95%CI = [0.2328; 0.9073], p < 0.05.
A significant difference between studies using within-, and between-subject designs was also found: Q(1) = 5.17, p = 0.023. Heterogeneity for within-subject designs was substantially smaller compared to between-subject designs: Q(5) = 1.18, I2 = 0.0% and Q(16) = 45.71, I2 = 67.2%, respectively. Between-subject studies showed a significant effect (multiple effect sizes): Hedges'g = 0.492, 95%CI = [0.0700; 0.9147], p = 0.025 while within-subject designs showed no effect: Hedges'g = 0.001, 95%CI = [-0,204; 0.223], p = 0.911. The other moderators were non-significant.
3.5 Qualitative analysis
The meta-analysis revealed some heterogeneity. A coding scheme was created to investigate systematic differences that may exist between groups with large and small effect sizes. Studies that were not included in the meta-analysis were added to maximize the available information. An overview of the moderators is presented in Table 2.
Table 2Differences between studies with-, and studies without a significant HD-tDCS effect on WM performance.
Identical training- and outcome task (offline-designs)
9/9
3x congruent task
4x no task, 1x congruent task, 1x incongruent task
+
Multiple Sessions
16/16
2x multiple sessions
1x multiple sessions
+
Note. The table shows the effects of the investigated moderators on the study outcomes. Data availability shows the number of applicable studies that reported the data. Effect on outcome indicates either no observable influence of the moderator on HD-tDCS effect on WM performance (−), inconclusive evidence for the assessment of an effect (∼), or evidence for an influence on the effect (+).
a One study was excluded due to the reported data being three standard deviations below or above the average.
b For between-participant designs, the reported results are collapsed over sham and active conditions.
Due to the limited number of studies that included WM tasks based on auditory stimuli, interpretation of the effect of task modality on the outcome is inconclusive. For the group of tasks that used a visual WM paradigm, whether the task employed verbal or non-verbal stimuli (e.g.: letters vs shapes), did not affect the results.
11 of the included studies employed at least one stimulation setup that targeted the left DLPFC. Nine of them positioned the anode at F3, and two used multiple anodes, with at least one of them positioned over F3. While minor differences in the placement of the return electrodes were apparent, no effect on the outcome was observed. The number of setups targeting other brain areas is too small to delineate a potential effect of anode placement.
The difficulty of the task might play a role. Two studies found an interaction effect between WM capacity [
], and HD-tDCS effects. Another study found a significant interaction between task load, stimulation, and RT on a WM task. Due to a lack of objective criteria, assessing systematic differences in task difficulty between studies with significant HD-tDCS effects and those without is difficult. Studies that found an effect, however, appear to be more challenging, focusing on adaptive n-back tasks, 3-back tasks, and a delayed response WM task with distractors. One of the studies used both a single and a dual n-back task and indicated numerically higher effect sizes in the more challenging dual n-back task [
Effects of high-definition transcranial direct current stimulation (HD-tDCS) of the intraparietal sulcus and dorsolateral prefrontal cortex on working memory and divided attention.
Nine of the studies employed an offline design, i.e., measurement of WM performance after application of HD-tDCS. In the three studies that showed a significant effect of HD-tDCS on WM performance, participants performed a WM task during stimulation that was identical to the one used to measure WM performance. Studies in which participants did not perform any task during stimulation did not find a significant effect.
Another potential driver for the HD-tDCS effects is the combination of HD-tDCS with multiple training sessions. Five studies used multiple HD-tDCS sessions, four of them with concurrent administration of a WM task. Of these four studies, two found HD-tDCS effects on accuracy measurements [
High-definition transcranial direct current stimulation (HD-tDCS) enhances working memory training.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.2019; 1: 329-332https://doi.org/10.1109/EMBC.2019.8856976
Effects of HD-tDCS combined with working memory training on event-related potentials.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.2020; 2020-July: 3553-3556https://doi.org/10.1109/EMBC44109.2020.9176063
]. The two studies that found a significant result were the same studies that employed an adaptive WM task, pointing towards a potential interaction of training and task difficulty. The study that reported reduced variance in the learning rate of the HD-tDCS group employed a distinct type of n-back task and stimulation of either the right DLPFC or the left primary motor area [
]. Interestingly, the study that reported multiple HD-tDCS sessions without simultaneous administration of a WM task did not find a significant effect [
Nine of the studies provided additional information on the tolerability of HD-tDCS (Supplement 5). Overall, three out of 329 participants had to be excluded due to HD-tDCS related side effects, reporting a sharp stinging and being in the active condition. One study reported significantly increased perception of pain in the HD-tDCS group [
Effects of high-definition transcranial direct current stimulation (HD-tDCS) of the intraparietal sulcus and dorsolateral prefrontal cortex on working memory and divided attention.
]. Seven studies provided information on blinding; all of them were successful.
4. Discussion
The present systematic review and meta-analysis focused on the effectiveness of anodal HD-tDCS for the enhancement of WM. Despite four of the studies showing a large effect size (hedges’ g > 0.8), the overall effect of the meta-analysis only revealed a non-significant trend of beneficial anodal HD-tDCS effects on WM accuracy, and none on RT. Further analysis revealed heterogeneity in study results and indicated a moderating effect of study design and use of HD-tDCS in combination with multiple WM training sessions. The moderating effect of training was also apparent in the RT analysis. The systematic review reinforced these results and additionally highlighted the importance of task difficulty, demanding a level that keeps the participant engaged. While other factors, such as duration of stimulation, blinding success, age, current density, or baseline WM performance, are likely to be important for HD-tDCS guided interventions on WM, the influence in the reviewed sample was low. The trim-and-fill procedure, while being susceptible to heterogeneity in the data, indicated the sensitivity of the results to the outcome of single studies [
]. A reason is the small number of studies, especially for the meta-analysis, which should be taken into account for further discussion and warrants caution for the interpretation of the results.
A comparison between HD-tDCS and conventional tDCS showed only small differences in their effect sizes. For conventional tDCS of the left DLPFC, effect size was estimated at 0.17 for single stimulation, and 0.29 for stimulation in combination with multiple WM training sessions [
]. Due to the limited number of studies on HD-tDCS, the present meta-analysis collapsed effect sizes for both training- and single-session designs, resulting in an effect size of 0.28. The analysis also revealed higher heterogeneity when compared to conventional tDCS.
4.1 HD-tDCS and conventional tDCS
The heterogeneity of this meta-analysis deviates from meta-analyses on WM enhancement using conventional tDCS [
Effects of anodal transcranial direct current stimulation on working memory: a systematic review and meta-analysis of findings from healthy and neuropsychiatric populations.
]. The biggest difference between the techniques lies in the increased focality of HD-tDCS. EF simulations comparing HD-tDCS and conventional tDCS of the motor cortex showed that an increase in stimulation focality leads to increased inter-individual variability in the generated EFs [
]. The distance between the electrodes also becomes increasingly important for HD-tDCS. A larger distance results in an increased peak of the EF and deeper penetration [
]. The focal approach of HD-tDCS might target sub-processes of WM more specifically. The importance of a specific sub-process for the respective WM task, and whether this sub-process creates a performance bottleneck for an individual, might affect HD-tDCS results and increase heterogeneity.
Recent studies on conventional tDCS also highlight the influence of cathode placement for the EF as well as behavioral outcomes and functional connectivity [
Soleimani G, Saviz M, Bikson M, et al. Group and individual level variations between symmetric and asymmetric DLPFC montages for tDCS over large scale brain network nodes. Sci Rep |. 123AD;11:1271. doi:10.1038/s41598-020-80279-0.
]. Conventional tDCS protocols commonly place the anodal and cathodal (or reference) electrodes on separate hemispheres, leading to a broader influence on the networks constituting a specific task. The HD-tDCS protocols included in this meta-analysis employed set-ups targeting the left hemisphere only. This demonstrates an important difference between both stimulation set-ups, and between conventional tDCS and HD-tDCS in general. Specifically with regard to other areas involved in WM processes, such as the right DLPFC, the need for HD-tDCS to target other areas becomes apparent. One of the studies in the qualitative analysis indeed indicated that stimulation the right DLPFC can also increase WM [
]. Potentially, HD-tDCS setups might also be adapted to target whole brain-networks more specifically, utilizing both anodal, and cathodal effects and deviating from the 4 × 1 ring setup that is often used. There is, however, only limited information on the influence of cathode placement for HD-tDCS. In the current analysis, only one study investigated effects of cathodal HD-tDCS, indicating decreased performance for high WM load when applied over the left PFC. Further research is needed to explore the effects of cathodal stimulation as well as different cathode placements in HD-tDCS studies.
4.2 Potential moderators of HD-tDCS effects
The meta-analysis and the systematic review indicated that HD-tDCS in combination with online working memory training in multiple sessions may facilitate WM improvement. Evidence demonstrates that training during conventional tDCS leads to increased skill acquisition when compared to offline training [
], where online and offline studies were analyzed separately and tDCS in combination with online training sessions produced larger effect sizes.
The observed benefit of WM training and concurrent anodal HD-tDCS application may result from a synergy of tDCS-induced increases in cortical excitability [
], and a practice-induced increase of activation in the networks required for task performance. This synergy might benefit long term-potentiation (LTP) according to the principles of Hebbian plasticity [
The neurobiology of prefrontal transcranial direct current stimulation (tDCS) in promoting brain plasticity: a systematic review and meta-analyses of human and rodent studies.
]. Evidence for the benefit of multiple tDCS sessions suggests a cumulative effect on neuronal excitability when tDCS is administered over a 1-week-period [
Daily transcranial direct current stimulation (tDCS) leads to greater increases in cortical excitability than second daily transcranial direct current stimulation.
]. Other studies show that combining tDCS with multiple training sessions enhances the consolidation of skill acquisition between training sessions when compared with sham stimulation [
]. These findings support the idea that HD-tDCS effects on WM are maximized when paired with multiple training sessions but should be interpreted with caution based on the limited sample size.
Qualitative analysis indicated that initial task performance [
] might also influence the effectiveness of HD-tDCS on WM enhancement. Both relate to the perceived difficulty of the task, suggesting another potential moderator. Conventional tDCS studies indicate that the ‘optimal’ difficulty, which is challenging but not overwhelming, is beneficial for tDCS effects [
Modulating the interference effect on spatial working memory by applying transcranial direct current stimulation over the right dorsolateral prefrontal cortex.
]. The underlying reasons are speculative. Some studies show that incentive-driven motivation acts as a facilitating factor for the effectiveness of tDCS on WM capacity [
Reward motivation and neurostimulation interact to improve working memory performance in healthy older adults: a simultaneous tDCS-fNIRS study HHS Public Access.
]. Interestingly, three of the studies used an n-back task that adapted to the performance of the participants in multiple training sessions. Two studies reported beneficial HD-tDCS effects with high effect sizes [
High-definition transcranial direct current stimulation (HD-tDCS) enhances working memory training.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.2019; 1: 329-332https://doi.org/10.1109/EMBC.2019.8856976
] pointing towards a potential synergy of optimal task difficulty and multiple online training sessions.
A final moderator is the study design, namely a within-subject versus a between-subject design. This highlights the already mentioned importance of individual differences for the efficacy of tDCS [
] as well as intra-individual differences. Recently observed interactions of circadian preferred time of the day with cortical excitability and WM performance act as an illustrative example of intra-individual factors and their potential influence on tDCS effectivity [
], individual differences might exert an even more pronounced influence on the outcome compared to conventional tDCS. Future studies should take these factors into consideration and control for differences and their potential effect on the outcome. This would help to better understand the influence of these factors and enable more personalized treatment protocols, eventually increasing the benefit of HD-tDCS. As the results show, the use of within-subject designs can also help containing their influence. It might also be advisable to create individualized head models and map task-related activity to implement the stimulation parameters individually.
While these moderators can inform the setup of future studies, the number of studies they are based on is limited. It might therefore be better to use them for the creation of new hypothesis warranting validation.
4.3 Limitations
The number of participants in the meta-analysis was rather low, with a mean of 21.4 participants per group/study, and most of the studies employed between-subject designs. This might underpower these studies for the detection of small effects.
For the meta-analysis, the low number of studies should be taken into account, being sensitive to single studies as shown by the trim-and-fill procedure. It was tried to cope with the problem by combining both quantitative and qualitative analysis and comparing the results of the meta-analysis with HD-tDCS studies targeting brain areas other than the left PFC. Still, the low number of studies dictates caution for the interpretation of potential moderators.
For calculation of the pooled effect sizes, different task modalities (e.g.: verbal and spatial), different timepoints (e.g.: 5 and 30 min after stimulation) and different difficulties (e.g.: 2-back and 3-back) were combined, potentially masking the effects of specific HD-tDCS parameters. We tried to account for this by including multiple effect sizes per study in the moderator analysis.
Due to the low number of studies, potential interactions between moderators were not investigated in the analysis. It is likely that some combinations, such as the duration of the stimulation, its intensity, and the region that is stimulated, differently affect HD-tDCS.
In the qualitative review, moderator analysis followed a post-hoc, exploratory approach. The results found must therefore be interpreted in the context of hypothesis-generating work, and should be integrated with available evidence to be tested in future research.
5. Conclusion
The meta-analysis indicated a trend of prefrontal anodal HD-tDCS effects for the enhancement of WM. Further examination revealed that HD-tDCS showed substantially higher heterogeneity compared to conventional tDCS, and subsequent analysis indicated the presence of moderators. Similar to conventional tDCS, the combination of HD-tDCS with multiple online training sessions increased effectiveness as indicated by the meta-analysis and the qualitative review. Results further revealed that using a within-participant design substantially reduces heterogeneity. In previous meta-analyses on conventional tDCS, the study design was not reported as a moderator. The qualitative review indicated that task difficulty might also moderate the effects of HD-tDCS on WM, with more demanding tasks showing higher effectiveness. In combination with multiple online training sessions, effects were further increased. However, due to the small number of studies, the precise extent to which these variables modulate the effects of HD-tDCS can only be estimated.
Overall, the results show that HD-tDCS might have the potential to increase WM performance, but appears to be more susceptible to moderating effects and other variables such as inter-individual differences. Due to the increased focality of HD-tDCS, individual differences in brain structure and structure-function mapping might be of special importance. One way to mitigate such influences could be the use of structural, as well as task-related MRI scans to inform the creation of individualized HD-tDCS setups. Future studies should consider factors reported in our analyses. Nevertheless, further research is needed to shed light onto the mechanisms underlying stimulation success to fully exploit the potential of HD-tDCS.
Author contributions
Dario Müller: Conceptualization, formal analysis, methodology, investigation, data curation, writing – original draft, writing – review & editing, visualization. Ute Habel: Investigation, resources, supervision, funding acquisition, writing - review & editing. Edward S. Brodkin: Investigation, supervision, writing - review & editing. Carmen Weidler: Investigation, data curation, conceptualization, funding acquisition, supervision, writing - review & editing, validation.
Funding
This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation – 269953372/GRK2150) and the START-Program of the Faculty of Medicine of the RWTH Aachen University.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Feredoes E, Delgado Garcia JM, Ae FF, et al. Anodal transcranial direct current stimulation of prefrontal cortex enhances working memory. doi:10.1007/s00221-005-2334-6.
Focalised stimulation using high definition transcranial direct current stimulation (HD-tDCS) to investigate declarative verbal learning and memory functioning.
Page MJ, Mckenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Research Methods and Reporting. Published online 2021. doi:10.1136/bmj.n71.
Effects of anodal transcranial direct current stimulation on working memory: a systematic review and meta-analysis of findings from healthy and neuropsychiatric populations.
Effects of single versus dual-site High-Definition transcranial direct current stimulation (HD-tDCS) on cortical reactivity and working memory performance in healthy subjects.
Effects of high-definition transcranial direct current stimulation (HD-tDCS) of the intraparietal sulcus and dorsolateral prefrontal cortex on working memory and divided attention.
High-definition transcranial direct current stimulation (HD-tDCS) enhances working memory training.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.2019; 1: 329-332https://doi.org/10.1109/EMBC.2019.8856976
Effects of HD-tDCS combined with working memory training on event-related potentials.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS.2020; 2020-July: 3553-3556https://doi.org/10.1109/EMBC44109.2020.9176063
Effects of high-definition transcranial direct current stimulation (HD-tDCS) of the intraparietal sulcus and dorsolateral prefrontal cortex on working memory and divided attention.
Soleimani G, Saviz M, Bikson M, et al. Group and individual level variations between symmetric and asymmetric DLPFC montages for tDCS over large scale brain network nodes. Sci Rep |. 123AD;11:1271. doi:10.1038/s41598-020-80279-0.
The neurobiology of prefrontal transcranial direct current stimulation (tDCS) in promoting brain plasticity: a systematic review and meta-analyses of human and rodent studies.
Daily transcranial direct current stimulation (tDCS) leads to greater increases in cortical excitability than second daily transcranial direct current stimulation.
Modulating the interference effect on spatial working memory by applying transcranial direct current stimulation over the right dorsolateral prefrontal cortex.
Reward motivation and neurostimulation interact to improve working memory performance in healthy older adults: a simultaneous tDCS-fNIRS study HHS Public Access.
00236-4&id=gr1.jpg){kind=link}
00236-4&id=gr2.jpg){kind=link}
00236-4&id=gr3.jpg){kind=link}
00236-4&id=gr4.jpg){kind=link}
00236-4&id=gr5.jpg){kind=link}
00236-4&id=gr6.jpg){kind=link}