Introduction

Individuals are often highly motivated to improve their cognitive abilities by means of learning, technologies and interventions, while also trying to reduce effort and achieve high performance at the lowest possible cost. In recent years, transcranial direct current stimulation (tDCS) has frequently been presented as an easy-to-use technology for simple upregulations of brain activity and cognitive performance. On the one hand, this simple conception is ignorant of the complexity of the human brain, its involvement in solving cognitive tasks with success, and of individual differences on several neurophysiological and psychological layers. On the other hand, an enhancement of neuronal activity is often assumed to automatically result in better cognitive performance. Empirically, this direct link between the quantity of neural activity and the quality of performance is not always applicable, and cognitive improvements may also result from inhibitory administrations of tDCS. The aim of this paper is to sketch distinct mechanisms that allow for such paradoxical beneficial effects of inhibitory, cathodal tDCS on cognitive performance.

When directed through the scalp, weak electrical currents can hyperpolarize or depolarize neuron populations by shifting their resting membrane potential, mostly inhibiting or facilitating the release of action potentials. In the electrical fields underneath externally placed electrodes, a relatively uniformly situated majority of the neuronal population directs the hyper- or depolarizing effect of stimulation. In sum, the dominant responses can produce behaviourally relevant inhibitory or excitatory effects, thus decreased or increased brain activity is often assumed. This physiological basis of modern tDCS methodology was investigated in motor cortex regions by means of muscle activity triggered by (standardized) single magnetic pulses (Nitsche and Paulus 2000) and supported by the modulation of visual evoked potentials in the occipital lobe (Accornero et al. 2007). Moreover, excitability changes are consistently documented to outlast the stimulation duration and produce after-effects in the range of hours, thus indicative of neuroplastic processes. However, a broad range of potential tDCS applications in psychiatric interventions and modulations of cognitive performance include other brain regions with different architectures, such as the prefrontal and parietal cortices. Here, a direct physiological evaluation of immediate stimulation effects is nearly impossible. Moreover, inhibitory stimulations with cathodal tDCS produce ambivalent outcomes (Jacobson et al. 2012), raising the question of whether this method is efficient for modulating higher order cortical functions in healthy participants (Horvath et al. 2015; see also Dedoncker et al. 2016).

Further complicating the inhibitory notion of cathodal tDCS is the fact that higher stimulation intensities (namely, 2 mA instead of 1 mA) over motor regions did not reduce but rather increase the cortical excitability (Batsikadze et al. 2013; Jamil et al. 2016). This finding has been suggested to reflect ceiling effects based on calcium influx, with larger postsynaptic calcium concentrations due to stronger stimulation intensities causing long-term potentiation rather than long-term depression (Lugon et al. 2015).

Potential Mechanisms for Paradoxical Beneficial Effects of Cathodal tDCS

Reduction of Hyperactivity in Patient Groups

Whereas there indeed exist studies that confirm the negative impact of 1-mA cathodal tDCS on critical cognitive functions such as cognitive control (e.g., Wolkenstein et al. 2014), response inhibition (Nieratschker et al. 2015), or working memory (Zaehle et al. 2011), there is also evidence for beneficial effects on cognitive processes. From early on, it was proposed that pathological elevations of cortical excitability in patient groups could be attenuated by cathodal inhibition and subsequently induced neuroplasticity, such as synaptic mediated after-effects (Nitsche et al. 2003b). Using a bicephalic 2-mA tDCS montage to reduce activity in the left temporo-parietal junction (cathode) and concurrently increase activity in the left prefrontal cortex (anode), auditory verbal hallucinations were markedly reduced in schizophrenic patients (Brunelin et al. 2012). This is likely due to a dual-mechanistic effect of reduced internal speech generation (temporo-parietal inhibition) and enhanced cognitive control (left prefrontal cortical excitation). In another patient group with treatment-resistant obsessive compulsive disorder, a condition often linked to hyperactivity in the left orbitofrontal cortex, 10 sessions of 2-mA cathodal tDCS in this region led to a significant reduction of symptoms (Bation et al. 2016). Both studies used stimulations at rest and showcase the beneficial effect of reducing hyperactivity in patient groups.

Reduction of Distractive Network Activity in Healthy Participants

Similarly, certain tasks may profit from attenuation of disturbing neurocognitive processes. For example, diminished internal regulation of bottom-up sensory information from reduced prefrontal activity by 1.5-mA cathodal tDCS was suggested as the mechanism that improved performance in the semantic generation of uncommon tool uses (Chrysikou et al. 2013). In our current research, we consistently observe that 1-mA cathodal tDCS can counteract implicit (potentially dysfunctional) associations in internally generated but not perceptually available stimulus–response compatibility effects (Schroeder et al. 2016).

Theoretically and empirically, system networks activated by a task can be more sensitive to the modulating effects of electrical fields introduced by tDCS and could provide a more fruitful reasoning for its modulations than the topographical perspective (see Fertonani and Miniussi 2016, for elaborated discussion). For instance, in an episodic memory task, effects of stimulation were only observed with task instructions eliciting cognitive activity (Zwissler et al. 2014).

Improvement of Signal-to-Noise Ratio

Another interesting conception of the working mechanisms underlying beneficial cathodal stimulation is that of a noise filter: By attenuating spontaneous neuron firing, goal-relevant firing may benefit from an improved signal-to-noise ratio. This focusing effect was documented in different paradigms (visuomotor tracking and visual search) with stimulations over occipital and parietal areas with 1 and 2 mA, respectively (Antal et al. 2004b; Moos et al. 2012). In a comparison of different placements with 1.5 mA, the correct target responses in a working memory task were outlined to depend on modulations below the cathode (Heinen et al. 2016). Prefrontal cathodal tDCS during episodic memory encoding also improved memory accuracy, whereas anodal tDCS had a detrimental effect (Zwissler et al. 2014). Furthermore, the noise-filter hypothesis is consistent with the improvement of target-flanker processing during 1.5-mA prefrontal cathodal tDCS under high attentional load (O’Neil and Adamson 2012).

Homeostatic Plasticity and Priming

In a comparable variant of the stimulus-stimulus conflict flanker task and with a more ventrolateral cathode placement over F7, Nozari et al. (2014) experimentally manipulated the timings of task and 1.5-mA stimulation. A task-specific facilitatory tDCS effect was observed only after 20 min of stimulation. Interestingly, concurrent task completion in the cathodal group produced generally increased response times only, compared to sham, and this effect was reproduced if participants engaged in another difficult task concurrent to stimulation and before they performed the flanker task. When cathodal tDCS was administered prior to another anodal stimulation during motor training, motor cortex excitability, performance and learning were further enhanced (Christova et al. 2015). Regulative adaptation processes were also observed from occipital cathodal tDCS in a visual orientation discrimination task and here, performance improved the most when the stimulation was applied before the task with 1.5 mA for 22 min (Pirulli et al. 2014). Thus, homeostatic plasticity may restore brain activity following cortical depression by cathodal tDCS and eventually result in excitability enhancements.

Distant Disinhibition

Lastly, effects of tDCS are not limited to the brain region under the electrode but can affect a widespread cortical and subcortical network (Bestmann et al. 2015). Modulation of distant functionally connected networks was observed in neuroimaging (Keeser et al. 2011), while cathodal tDCS to the cerebellum could improve arithmetic performance by route of disinhibition of frontal areas (Pope and Miall 2012). Thus, performance improvements might also be caused by stimulation effects other than those on the targeted brain region.

Table 1 Cognitive enhancement by cathodal tDCS
Full size table

Need for Polarity-Specific Models?

Although we often observed opposing detrimental effects of anodal tDCS in tasks with beneficial cathodal tDCS (Zwissler et al. 2014; Schroeder et al. 2015; Plewnia et al. 2013; Nieratschker et al. 2015), polarity-specific cognitive effects are not necessarily granted and might also tap different mechanisms. Although desirable for the correct identification of theoretical and computational models, further empirical validations of the presented mechanisms (Table. 1) are challenged by the involvement of task- and state-dependent brain activities that may interact differentially with direct currents dependent on their direction. The apparent differentiation of neuromodulation underneath cathodes and anodes is further subject to partially different physiological mechanisms. At the minimum, AeCi polarity effects of tDCS are asymmetrical and appear to be more effective underneath anode or cathode, dependent on the domain and paradigm under study (Antal et al. 2004a; Jacobson et al. 2012). In motor cortex stimulations, tDCS often demonstrates sustained shifts of excitability and inhibition exceeding the exact stimulation duration. The neuroplastic after-effects of cathodal tDCS, especially, were found to depend on modulation of glutamatergic inter-neurons, whereas anodal tDCS seems to rely on additional GABAergic synapses (Stagg and Nitsche 2011). A blockage of voltage-dependent sodium channels can only nullify immediate anodal but not cathodal tDCS effects on motor cortex excitability, whereas NMDA receptors are generally involved in after-effects of both anodal and cathodal tDCS (Nitsche et al. 2003a).

Individual Differences

Distinct neurochemical dependencies can predict stimulation effects in groups with different neurotransmitter distributions, e.g., due to a specific genetic profile. For instance, in a group of genotyped individuals, an impairment of response inhibition by cathodal tDCS was entirely driven by Val/Val homozygous COMT allele carriers, which demonstrates a dependency of optimal performance on prefrontal dopamine signalling (Nieratschker et al. 2015). Thus, such non-linear interactions may account for the observation that further enhancements by tDCS can deteriorate performances (Plewnia et al. 2013; Schroeder et al. 2015). In addition to the influence of the genetic makeup on stimulation effects, a dimensional conception of interacting factors has to consider cortical and skull structure, personality and baseline performance. Electrical field modelling of individual MRI data confirmed that the current strength at a targeted region was related to tDCS behavioural effects but also showcased notable variability in stimulation response for different head and cortical structures (Kim et al. 2014). Furthermore, current density and direction are potentially susceptible to practically relevant differences in hair thickness and saline saturation, sweat, pressure by rubber bands and other (unexplored) montage variations (Horvath et al. 2014).

Beyond these relevant individual characteristics, there are also indirect modulations of stimulation effects. Baseline performance may predict stimulation successes (Heinen et al. 2016), while unrelated cognitions and emotion can alter the direction of neuromodulation. For example, high- and low-math anxiety groups were stimulated with identical prefrontal anodal tDCS configuration when solving simple arithmetic decisions, and their performance was improved or impaired, respectively (Sarkar et al. 2014). A recent study on psychopathic personality traits demonstrated this principle with prefrontal cathodal tDCS and showed a correlation of stimulation effect (response inhibition improvement) with participants’ cold-heartedness personality score (Weidacker et al. 2016).

Current Directions

What effect could different cortical structures have on tDCS effectivity? This question should be elaborated on in much greater detail, but we will formulate a few speculations. One instructive idea could be drawn from the observation that prefrontal TMS with the identical stimulus intensity led to smaller EEG responses than motor TMS (Kähkönen et al. 2004), which has not been tested with tDCS. Cognitive tasks differ critically from TMS-induced motor evoked potentials in the involvement of distributed networks with potential compensation processes. Correspondingly, latency and accuracy as behavioural measures are subject to further external noise (Jacobson et al. 2012). Finally, cytoarchitecture and receptor density can vary notably between functionally different cortical areas, e.g., motor and multimodal association cortices (Zilles et al. 2002).

Further reducing the apparent AeCi dichotomy, development of more focal stimulation electrodes (e.g., concentric electrodes or HD-tDCS) or multicortical placements (bilateral, bicortical) could profit from refinement of current flow neural polarization effects. Recent studies have already justified the concurrent up- and downregulation of respective brain areas with standard configurations (e.g., Bation et al. 2016) or with concentric montages, which appear feasible and effective in modulating response inhibition (Hogeveen et al. 2016). By turning necessary return electrodes into pragmatic co-generators of a more focal current field, concentric montages may also reduce the current lack of standardization and motivate theory-guided stimulation approaches.

Conclusion

We described beneficial effects of cathodal tDCS on cognitive functions and their assumed rationales (cf. Table 1). Performance improvements were observed in reduction of hyperactivity in patients, suppression of dysfunctional networks in healthy participants, noise reduction, homeostatic plasticity and distant disinhibition. Future research should explore interactions between the described mechanisms that may have additive or neutralizing relevance for different behaviours and could result in unwanted cognitive side effects. The acknowledgement of cathode placements as an active contribution to successful enhancements with tDCS, paired with (individualized) network-function definitions and specification of electrical parameters, will help us avoid diffuse stimulation effects and provide new options for an effective, targeted and knowledge-driven modulation of cognitive and behavioural functions.