What transcranial direct current stimulation intensity is best for cognitive enhancement?

Abstract

In a recent study published in The Journal of Neurophysiology, Ehrardt et al. (Ehrhardt SE, Filmer HL, Wards Y, Mattingley JB, Dux PE. J Neurophysiol 125: 385–397, 2021) report that moderate intensity (1 mA/25 cm²) transcranial direct current stimulation (tDCS) is optimal for improving performance on a stimulus-response matching task, as opposed to a lower 0.7 mA/25 cm² or higher 2 mA/25 cm² dose. This result suggests that behavioral effects of tDCS do not follow a linear dose-response curve. Potential neurobiological and neurocognitive implications of these findings, as well as suggested directions for future research, are discussed.

Research into potential noninvasive forms of cognitive enhancement for psychiatric disorders, neurodegenerative diseases, and healthy populations is of great interest.One such form, transcranial direct current stimulation (tDCS), has become widely utilized in both basic and clinical studies due to its low cost, ease of administration, safety (with only minor adverse effects commonly reported), and demonstrated neurocognitive and associated neurophysiological effects (1). Nonetheless, well-powered studies that have compared the effects of a range of tDCS “doses,” e.g., current amplitudes or stimulus durations, are much less common than studies that have only used one dose. Rarer still are studies that have examined if the effects of tDCS stimulation on performance in one task are transferrable to other tasks. Such research is necessary to maximize tDCS efficacy and “real-world” applicability to ensure that tDCS effects are not specific to a particular set of laboratory conditions.

In a recently accepted paper in The Journal of Neurophysiology, Ehrardt et al. (2) compared the effect of three left prefrontal anodal tDCS stimulation intensities [20 min of 0.7 mA/25 cm², 1 mA/25 cm², and 2 mA/25 cm², or sham stimulation (with 30-s ramp-up) administered during task training on 4 consecutive days] on reaction time and accuracy on the trained task as well as three untrained tasks. A between-subjects design was used in which participants [n = 123, age (SD) = 21.95 (2.97) yr, 79 females] were randomly assigned to one of the four stimulation intensities [0.7 mA/25 cm² (n = 31), 1 mA/25 cm² (n = 32), and 2 mA/25 cm² (n = 31), or sham (n = 29)]. Participants were also blinded to stimulus condition. Blinding was effective, as participants performed near chance level (22.1%; chance = 25%) when asked what stimulation group they were in. Stimulation was administered using 5- × 5-cm saline-soaked electrodes. The trained task was a stimulus-response mapping task in which participants pressed different buttons depending on the auditory or visual stimulus presented. A more difficult dual-stimulus condition (simultaneous auditory and visual presentation) was also included. The three untrained tasks were 1) a task identical to the trained task, except using different stimulus-response maps; 2) a visual search task in which individuals searched for a 90° rotated letter “T” among distractor “L” stimuli; and 3) a Stroop task in which individuals reported the color of a word under a congruent condition (letter color matching word, e.g., “blue” in blue font), incongruent condition (letter color not matching word, e.g., “blue” in green font), or neutral condition (e.g., “pan” in blue font). Participants were asked to respond as quickly and accurately as possible. Behavioral measures of interest were accuracy and reaction time (RT).

tDCS results were dose dependent for the trained task. Specifically, Ehrhardt and colleagues found that the 1 mA/25 cm² dose had the biggest effect on RT, significantly [t (P) = 2.52 (0.007)] decreasing RTs versus sham with an effect size (Cohen’s d) of 0.65 during the single-stimulus condition. In contrast, the 0.7 mA/25 cm² and 2 mA/25 cm² doses decreased RTs versus sham with d = 0.60 [t (P) = 2.31 (0.012)] and d = 0.49 [t (P) = 1.93 (0.029)] (respectively) for the single-stimulus condition. A similar dose-dependent pattern was observed for the dual-stimulus condition and for reaction time decreases for the single-stimulus condition in the stimulus-response transfer task. Effects on accuracy followed the same dose-dependent pattern for the trained task during the dual-stimulus condition [d = 0.52, t (P) = 1.94 (0.028); d = 0.58, t (P) = 2.24 (0.014); d = 0.47, t (P) = 1.83 (0.037)] and for 0.7 mA/25 cm², 1 mA/25 cm², and 2 mA/25 cm² stimulations versus sham, respectively. No effects of stimulation were observed for the other two transfer tasks.

The results of this study suggest that tDCS effects on neurocognition may follow an inverted U-shaped dose-response curve, in which a moderate stimulus intensity has the strongest effects on performance. These results further suggest that tDCS may enhance the effects of cognitive training as well as transfer to novel stimulus-response pairings in the same task. Transferability may be limited, however, as tDCS did not enhance performance on the visual search or Stroop tasks. These findings are also in alignment with previous studies showing nonlinear relationships between stimulus intensity and neuronal excitation in humans (3). Ehrhardt et al. (2) postulate that the curve may be due to tDCS inducing an optimum level of excitatory/inhibitory (E/I) balance to maximally enhance cognitive ability. tDCS modulation of E/I balance may occur in various forms, including at the synaptic level (ratio of excitatory/inhibitory inputs), circuit level (ratio of pyramidal/interneuronal cell activation), and neurotransmitter level (ratio of synaptic glutamate/GABA) (4).

As noted by the authors, a limitation of the study was that tDCS did not transfer beyond the stimulus-response task, possibly due to limitations in dynamic range as a result of ceiling effects. Accordingly, larger differences may have been observed using more difficult tasks. Due to the limited transferability, it is also unclear if the enhancements observed in the study would carry over to the plethora of task demands associated with everyday cognitive function.

The finding that tDCS effects transferred to a new stimulus-response map but not the other two tasks is somewhat surprising. It is possible that the observed task-specific transfer was due to differences in the neuronal systems engaged in each task. The stimulus-response task heavily involves activation within and connectivity between prefrontal areas and brain regions associated with sensorimotor control (e.g., motor cortex, cerebellum, and dorsal striatum) (5). Visual search tasks heavily recruit visual areas such as the frontal eye fields and visual cortex, in addition to frontal regions (e.g., 6). Stroop tasks, on the other hand, are primarily associated with frontal executive and salience networks (e.g., the dorsolateral prefrontal cortex, inferior frontal gyrus, and anterior cingulate) (7). Although the prefrontal cortex was likely recruited during all three tasks, it is possible that tDCS-induced plasticity associated with performance enhancement occurred within the particular network activated during stimulation and was therefore specific to the neuronal processes engaged by the task being performed while stimulation took place. Alternatively, it is possible that the optimum levels of prefrontal activation differed between the tasks; additional conditions in which the visual search task or Stroop tasks are used as the training task would be needed to test this hypothesis.

What neuronal mechanisms may underlie the inverted U-shaped dose-response curve observed in the Ehrhardt et al. (2) study? When administered at doses typical for human studies (0.5 mA–2 mA with 25 or 35 cm² sponges), tDCS is unlikely to have measurable effects on neuronal spike firing rates (8). Instead, it may modulate excitability by altering cortical electric fields and membrane potentials, thereby changing spike timing and subsequently inducing frequency-dependent variations in oscillatory dynamics (i.e., power and synchrony) or synaptic plasticity (9). Indeed, tDCS effects on plasticity may help explain the relatively low task transferability observed in the Ehrhardt et al. study, as it may have affected encoding specifically for that particular task. Related to the observed inverted U-shaped response, one prominent theory of healthy brain function is that the brain continuously maintains homeostatic control of E/I balance (e.g., ratio of excitatory vs. inhibitory neuronal activation) to maximize its dynamic range for plasticity and neuronal control over information transfer (10). In this framework, overexcitation may induce a ceiling effect, providing limited maneuverability for plasticity to occur. A certain level of inhibition is also required to evoke synchronous oscillations, as neurons must collectively be below firing threshold before they are able to discharge in unison. As such synchrony may be required for learning and long-range neuronal information flow (11), it is possible to speculate that overexcitation may disturb this synchrony and actually have deleterious effects on cognition.

Related to this point, it is important to consider that the optimum intensity for tDCS stimulation may vary between individuals. For this reason, a strength of the Ehrhardt et al. (2) report is their observation that the four stimulation groups did not differ in their baseline performance on any task. One factor that may account for differences in optimum intensity is anatomical variation, as bone shape, structure, and composition, as well as cortical (gray and white matter) architecture, can all influence the amount of current that is received by the brain during tDCS (12, 13). A second factor is that different people may have different baseline E/I setpoints and therefore require varying levels of stimulation to show enhancement. In this vein, if tDCS effects differ based on baseline E/I balance, it may also suggest that the ideal stimulus intensity may vary for certain demographic or disease groups. For example, women may show differences in brain excitability and GABA/glutamate ratio depending on menstrual cycle phase (14) as well as differences in cranial anatomy versus men (e.g., skull thickness, composition), and a few studies have accordingly found significant sex differences in tDCS effectiveness (15–17). With regard to disease, magnetic resonance spectroscopy studies of schizophrenia have found that the illness may be associated with reduced cortical glutamate/GABA ratios relative to healthy individuals (18). If true, this suggests that ameliorating cognitive deficits in schizophrenia with tDCS would require stronger stimulation than in healthy participants. Indeed, this was the case in a study by (19), who compared the effects of sham, 1 mA/35 cm², and 2 mA/35 cm² left dorsolateral prefrontal cortex tDCS stimulation on working memory in individuals with schizophrenia. Hoy et al. (19) found that only the 2 mA/35 cm² dose significantly improved performance, in contrast to the result in healthy individuals presented by Ehrhardt and colleagues.

A limitation of the Ehrhardt et al. (2) study is that only behavioral measures were examined. Future studies may employ functional neuroimaging or neurophysiology [e.g., electroencephalography (EEG)] to determine if tDCS is having its intended targeted neuronal effects as well as determine the biological mechanisms by which stimulation modulates performance on the stimulus-response and other tasks. For example, previous work indicates that evoked (i.e., stimulus-locked) activation in the β frequency range (13–30 Hz) in frontal and occipital electrodes is associated with faster reaction times in a similar stimulus-response task (20). One may hypothesize, therefore, that tDCS may preferentially affect activity in this frequency range during this task, and that the strongest effects would be observed at 1 mA/25 cm². EEG or magnetic resonance spectroscopy may also be used in future studies to determine baseline levels of activation and/or E/I balance (i.e., prefrontal glutamate/GABA ratio), with the hypothesis that people with lower levels of activation would require stronger stimulation to show optimal results. People with higher levels of baseline activation, conversely, would thus require weaker stimulation to show maximal improvement.

Overall, the results of Ehrhardt et al. (2) suggest that across a group of healthy, relatively young (age ∼22 yr) individuals, an intermediate tDCS dose (1 mA/25 cm²) is more effective than 0.7/25 cm² or 2 mA/25 cm² doses at inducing transferable procognitive effects. As many factors may contribute to individual variability in tDCS response, future studies may examine sources of variation in the observed pattern of results to determine the magnitude for which each factor contributes to tDCS response. Indeed, it should be noted that because tDCS results often do not replicate and interindividual variability in response to tDCS may be high (up to ∼50/50% responders/nonresponders) (21, 22), the results presented in the study should be verified in an independent sample.

GRANTS

Dr. J. Smucny is funded in part by a grant from the National Institute of Mental Health (F32 MH114325).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.S. drafted manuscript; J.S. edited and revised manuscript; J.S. approved final version of manuscript.