Introduction
Creativity and innovative thinking in the arts, science, stage performance, the commercial enterprise, and business innovation is a multidimensional construct (Kaufman
2007) and based on diverse psychological and cognitive processes (Csikszentmihalyi
1999; Kaufman and Beghetto
2009; Gaut
2010; Sawyer
2011; Perlovsky and Levine
2012; Khalil et al.
2019). Explicitly, creative cognition is at the root of extraordinary performance in arts and sciences (Baas et al.
2015). There are certain types of creative processes, such as divergent thinking (DT) and convergent thinking (CT), that are selectively affected by inhibitory control (IC) (Radel et al.
2015; Cassotti et al.
2016; Khalil et al.
2019).
IC is a central component of executive function (EF) and allows the suppression of automatic, prepotent, or inappropriate actions and ideas (Aron et al.
2014). This inhibition of automatic and prepotent thoughts, actions, and responses is crucial for creativity (Benedek et al.
2014; Radel et al.
2015). An association between IC and creative performance has been established (for review, cf. Khalil et al. (
2019)). For example, Radel et al. (
2015) exposed their participants to an Eriksen Flanker (Eriksen and Eriksen
1974) or Simon (Simon
1990) task before performing creativity tests. In these cognitive tasks, perception of (Eriksen Flanker) or response to (Simon) stimuli distracting from the target stimuli need to be inhibited. Exhausting the participants' inhibitory control resources by these tasks led to enhanced fluency (i.e., number of ideas) and originality (i.e., generation of uncommon ideas) in the Alternative Uses Task (AUT; Carroll and Guilford (
1968)) but not the Remote Associate Task (RAT; Mednick (
1962)). AUT and RAT are well-established measures for DT and CT, respectively. Thus, a lack of resources for inhibition might lead to facilitation of the novelty (i.e., originality) of thoughts (i.e., ideas). Accordingly, one could hypothesize that particular idea generation processes profit from a depletion of resources for inhibition.
The prefrontal cortex (PFC) is known to be part of a deliberate inhibition control network and considered to be a central node for problem solving and idea generation from adolescence to adulthood (Cassotti et al.
2016). Goghari and MacDonald (
2009) postulated that a shared prefrontal network (including left and right IFG) mainly serves for processes of response selection and inhibition. However, inhibition is a diverse cognitive dimension, and different sub-dimensions have been attributed to various, distinct frontal cortical sites and hemispheres. Also, the degree of the recruitment of such networks dramatically depends on specific task demands (Aron et al.
2014).
For instance, the left IFG is commonly involved when controlled responses are required, while the right IFG is activated more when task demands are more robust for response inhibition (Goghari and MacDonald
2009). Also, Swick et al. (
2008) indicated a role of the left IFG for the successful implementation of IC over motor responses. Whereas the right IFG might also involve response inhibition but is triggered through automatic, bottom-up processing. For example, engagement of right IFG was reported for stop-signal tasks (SST) and interpreted as reprogramming action plans (Lenartowicz et al.
2011). Moreover, Aron et al. (
2014) affirmed inhibition as a central component of executive control that depends upon the right IFG and associated networks.
Using transcranial direct current stimulation (tDCS), Cunillera et al. (
2014) revealed the involvement of the right IFG in two kinds of inhibition processes (i.e., reactive and proactive inhibition). tDCS is a non-invasive tool that modulates brain function through hyper- or hypopolarization of neurons (Stagg and Nitsche
2011; Medeiros et al.
2012). Stramaccia and coauthors reported evidence about the role of the right IFG in inhibition accuracy in SST (Stramaccia et al.
2015) and, in interference control during memory retrieval processes (Stramaccia et al.
2017).
Even training and developmental studies have hinted to a pivotal role of the IFG in IC. Practicing inhibition tasks reduced the neural activity within the prefrontal inhibitory networks to inhibition trials reaffirming the role of the prefrontal cortex, especially the IFG, in IC (Manuel et al.
2013; Berkman et al.
2014; Chavan et al.
2015). Also, Hartmann et al. (
2016) reported enhanced activity of the right frontal cortex, including right IFG, in association with top-down IC after Go NoGo task (GNGT) training. GNGT is a mutual task used to assess IC in humans and animals where participants have to respond quickly to frequently occurring ‘Go’ stimuli and to inhibit responses to infrequent ‘NoGo’ stimuli (Tamm et al.
2002; Swick et al.
2008; Luijten et al.
2011; Vara et al.
2014; Hartmann et al.
2016; Wilson et al.
2016). Using GNGT, Tamm et al. (
2002) postulated an association between increasing left IFG activation and response inhibition abilities during development. Conversely, Vara et al. (
2014) revealed bilateral inferior frontal activation in adolescents, but more right lateralized inferior frontal activity in adults; see Table
3 of Vara et al. (
2014).
Taken together, the IFG is a crucial brain region associated with various dimensions of IC. The left IFG seems to be more related to rapid response execution (Tamm et al.
2002), the successful implementation of IC over motor responses (Swick et al.
2008) and DT (Ivancovsky et al.
2019), while the right IFG is mainly associated with unconscious, automatic, tonic inhibition and IC (Lenartowicz et al.
2011; Aron et al.
2014; Cunillera et al.
2016; Campanella et al.
2017).
Besides IFG, the right dorsolateral prefrontal cortex (rDLPFC) may be engaged in active inhibitory processing of both motor and higher level memory representations (Penolazzi et al.
2014; Friehs and Frings
2018,
2019; Sandrini et al.
2020). Using tDCS, Penolazzi et al. (
2014) reported that cathodal stimulation over rDLPFC leads to decreased inhibition during the standard retrieval-practice paradigm (RPP). Also, Friehs and Frings (
2018) examined the inhibitory role of rDLPFC on Stop-Signal Reaction Time (SSRT) using tDCS. They reported a reduction in SSRT and the number of omission errors after anodal tDCS (a-tDCS). The involvement of rDLPFC in monitoring the need to stop and stepping into action when top-down IC is required (Fuster
2015) is in line with the previous findings. However, a later study by Friehs and Frings (
2019) did not find a modulation of error rates in any form but only a significant increase in SSRT after cathodal tDCS (c-tDCS). Sandrini et al. (
2020) further verified that both the rDLPFC and right inferior parietal cortex (rIPC) represent an essential part of the fronto-basal-ganglia network, which is critical for rapid response inhibition. Accordingly, Aron et al. (
2014) proposed that DLPFC implements task rules rather than inhibition. Finally, Zmigrod et al. (
2015) provided direct evidence for the role of the left DLPFC (lDLPFC) in CT and DT but a mediating role of the PFC in problem solving behavior, presumably through attentional processes.
Support for the idea that IC and creativity are closely related and that both functions depend on shared neural substrates, particularly in the IFG, comes from lesion studies and studies experimentally manipulating brain regions underlying IC. For instance, lesions leading to an attenuation of cognitive inhibition allow patients to be more creative (Kapur
1996; Miller et al.
2000; Miller and Hou
2004; Seeley et al.
2008; Shamay-Tsoory
2011). Seeley et al. (
2008) reported an enhancement of right hemisphere activation in a patient after progressive degeneration of the left frontal hemisphere (left IFG included). Along the same lines, Miller and colleagues reported on patients with left hemispheric degeneration (Miller et al.
1996) who developed creative abilities such as musical or artistic skills (Miller et al.
2000; Miller and Hou
2004). It might be speculated that reduced left IFG activation might have resulted in decreased IC, which in turn was associated with enhanced creativity. Accordingly, Kapur (
1996) postulated a
”paradoxical functional facilitation
” theory, where he explained the increment of creativity as the result of brain damage affecting areas involved in attenuation (i.e., left temporoparietal and inferior frontal regions).
Moreover, Shamay-Tsoory (
2011) revealed a positive correlation between lesions in left parietal areas and increased levels of creativity. Combined, one can expect that in the process of being creative, IC might be a cognitive control mechanism important for developing original ideas or giving non-conventional answers. Therefore, reduced activation of left frontal regions (i.e., left IFG) and increased activation of right frontal regions (i.e., right IFG) should influence the creative performance. However, up to now, a causal relationship between creativity and IC has not been revealed. With this current study, we intended to establish such a causal relationship between creativity and IC, as one dimension of creative cognition (Benedek et al.
2012; Mok
2012; Cassotti et al.
2016; Khalil et al.
2019) through experimentally manipulating IC using tDCS.
Several studies used brain stimulation (i.e., tDCS) to modulate and to explore components of IC and its association with creativity (Mayseless and Shamay-Tsoory
2015; Zmigrod et al.
2015; Lucchiari et al.
2018). Findings from tDCS studies by Mayseless and Shamay-Tsoory (
2015) supported a Balance Hypothesis, according to which creativity demands a balance of activation between both hemispheres of the frontal lobes (and more specifically, between the right and the left IFG). These authors applied a bilateral tDCS stimulation with the cathode over the right IFG, and the anode over the left IFG, and compared this condition with the contrary one. Results revealed increased DT scorings with left cathodal and right anodal stimulation but no effect on creativity in the reverse condition. Unimodal stimulation with either the anode or the cathode over the left or right IFG alone, however, was not sufficient to alter the creative process (Mayseless and Shamay-Tsoory
2015).
Recently, Lucchiari et al. (
2018) presented a critical review of original research articles investigating the various influences of tDCS on creativity and its underlying mechanisms (cf Table 1 of Lucchiari et al. (
2018)). They concluded that tDCS effects are considerably unspecific, modulating only the likelihood of more creative thinking. They further expressed the necessity for a more comprehensive framework related to creativity research and brain stimulation (Lucchiari et al.
2018).
Table 1
Effects of tDCS on creativity dimensions as revealed by ANCOVA analysis (cf. Fig.
3)
Fluency | Intercept | 391.97 | 1 | 95.44 | 1.555e−11 | 0.731 |
Baseline values* | 38.25 | 1 | 9.31 | 0.004** | 0.210 |
tDCS condition | 0.51 | 1 | 0.12 | 0.727 | 0.003 |
Baseline values:tDCS condition | 0.22 | 1 | 0.05 | 0.818 | 0.001 |
Residuals | 143.74 | 35 | |
Originality | Intercept | 240.56 | 1 | 39.60 | 3.18e−07 | 0.531 |
Baseline values | 8.75 | 1 | 1.44 | 0.238 | 0.039 |
tDCS condition* | 33.25 | 1 | 5.47 | 0.025* | 0.135 |
Baseline values:tDCS condition | 0.41 | 1 | 0.07 | 0.797 | 0.002 |
Residuals | 212.60 | 35 | |
Flexibility | Intercept | 110.58 | 1 | 135.40 | 1.394e-−13 | 0.794 |
Baseline values | 0.13 | 1 | 0.16 | 0.688 | 0.005 |
tDCS condition | 0.04 | 1 | 0.05 | 0.831 | 0.001 |
Baseline values:tDCS condition | 0.98 | 1 | 1.20 | 0.282 | 0.033 |
Residuals | 28.58 | 35 | |
To provide additional evidence for formulating such a framework, we intended to address the question of whether changes in IFG brain activity and, thus, IC would mediate changes in creativity. For that purpose, we used AUT to measure creativity in terms of DT and applied GNGT to examine the IC before and after tDCS. Tests of DT are probably one of the most commonly used assessments of creativity and do provide valuable information about creative potential. In the AUT task, participants are asked to name different, alternative uses for everyday objects. AUT measures three dimensions: fluency (i.e., number of ideas), originality (i.e., generation of uncommon ideas), and flexibility (i.e., the ability to change strategy) (Horne
1988; Chávez-Eakle et al.
2007; Scibinetti et al.
2011). From the AUT, scores for ideational fluency, ideational originality, and ideational flexibility are calculated. Ideational fluency represents the number of ideas an individual gives, while ideational originality expresses the statistical infrequency or uniqueness of ideas, and ideational flexibility depicts the number of different conceptual categories used by the individual (Runco et al.
1987; Runco and Jaeger
2012; Beketayev and Runco
2016). Thus, ideational flexibility is considered to be extremely important as it allows an individual to avoid ruts and routines when solving problems. In turn, it not only contributes to creative problem solving but is also related to adaptability and the ability to shift perspectives while solving a problem.
Based on the previous evidence, we hypothesized that, on one hand, hyperpolarization of left IFG through cathodal stimulation coupled with anodal stimulation of the right IFG (i.e., L −R +) should reveal a facilitative effect on creativity. On the other hand, creativity should be decreased in another treatment group, in which hyperpolarization of the right IFG was coupled with depolarization of the left IFG (i.e., L + R −). More precisely, we assumed that L + R − stimulation would lower AUT scores through enhancing IC as measured with the GNGT. On the contrary, the group treated with the reversed stimulation arrangement, i.e., L −R + should express higher AUT scores based on decreased IC. In particular, we expected that changes in IC induced by c-tDCS targeting the left IFG coupled with a-tDCS targeting the right IFG should result in altered originality and flexibility, but not necessarily fluency in the AUT.
Discussion
Our study had two primary goals: first, to investigate the effect of tDCS on divergent thinking (DT; through AUT) and inhibitory control (IC; through GNGT), and second, to explore the relationship between AUT, GNGT, and the activity of the left and right IFG.
Concerning the first goal, in line with other studies (Mayseless and Shamay-Tsoory
2015; Lucchiari et al.
2018; Ivancovsky et al.
2019) and, in an agreement with the Balance Hypothesis (Mayseless and Shamay-Tsoory
2015), we found an enhancement of ideational originality after L −R + (facilitation of right IFG and inhibition of left IFG) but not after the opposite stimulation regime. However, the study by Mayseless and Shamay-Tsoory (
2015) showed only a trend in increasing performance in all three dimensions of AUT, i.e., ideational fluency, originality, and flexibility, while we did not find any direct effect of tDCS on flexibility and fluency.
Regarding IC, no direct effect of tDCS could be revealed. This result might be due to the simplicity of GNGT, i.e., the ease to discriminate between Go and NoGo stimuli (Sallard et al.
2018). Moreover, the lack of offline effects of tDCS on GNGT performance reported by Sallard et al. (
2018) corroborates findings by previous tDCS/GNGT studies stimulating the right IFG (Cunillera et al.
2016; Campanella et al.
2017). Research has demonstrated that offline stimulation improved IC during a SST (Cai et al.
2016) but not GNGT (Campanella et al.
2017). Similarly, using SST, other studies found an enhancement of inhibition accuracy after tDCS over the right IFG (Jacobson et al.
2011; Ditye et al.
2012; Stramaccia et al.
2015; Cai et al.
2016). While Cunillera et al. (
2014) reported a simultaneous modulation of two kinds of inhibition processes (reactive and proactive inhibition processes), by a-tDCS on the right IFG, using SST and GNGT, respectively. Additionally, Stramaccia et al. (
2017) revealed that tDCS over the right IFG disrupts control over interference using memory inhibition tasks. Therefore, one should take into consideration the type of task used to measure IC.
Other factors that might explain our results regarding the effect of tDCS on GNGT could be baseline activity levels. A study by Sallard et al. (
2018) suggested that the baseline level of engagement of the brain areas of interest might be a critical factor in determining the functional effect of tDCS, which was confirmed by changes in the BOLD signal after a-tDCS manifesting only under conditions of low task-related activity.
With regards to the second goal, regarding the relationship between AUT, GNGT, and the activity of the left and right IFG, as we did not find a direct effect of tDCS on GNGT, no mediation effect of IC can be assumed. However, ANCOVA analysis showed a potential moderation effect (as indicated by a significant interaction effect of tDCS condition and change in
d’): for both, originality and flexibility, but not fluency, post-stimulation scores were higher after L −R + as compared to L + R − conditions, only when associated with increased IC as revealed by higher
d’ (90/10 condition). It is not surprising that this moderation effect was found only for the 90/10 but not the 50/50 condition of the GNGT, as the demand on IC resources is higher in the 90/10 condition (Swick et al.
2008). No difference between stimulation conditions was revealed when there was a decline in GNGT (negative
d’delta90 values; cf. Fig.
5).
A potential explanation for the described moderation effect, i.e., that tDCS is only effective in enhancing originality and flexibility through left cathodal and right anodal tDCS when
d’ in the GNGT is on an enhanced level. This enhanced
d’ level might reflect a specific mind state or another latent factor that facilitates or attenuates the effects of tDCS on creativity (i.e., the moderation effect of
d’). Dependencies of the tDCS effect on the subjects' neurocognitive states have been suggested previously (Learmonth et al.
2015; Hsu et al.
2016). We admit that these arguments are speculative and need to be further investigated. Consequently, better control of all the previous aspects could thus help to improve the reliability of the effects of tDCS on brain activity, and by extension, on its behavioral consequences.
In conclusion, this study emphasizes the effectiveness of shifting activity from left to right IFG through tDCS for creative performance in a DT task. Of interest, tDCS stimulation did not significantly modulate performance in all three AUT dimensions. c-tDCS targeting the left IFG associated with a-tDCS over the right IFG resulted in increasing originality and flexibility, but not fluency in the AUT.
Open Question, Limitations and Future Directions
Our current study highlighted the possibility of a latent factor (LF) that determines the moderation effect of d′ induced by tDCS on creativity (i.e., either facilitation or attenuation). Therefore, it would be of great interest to explore such factor (s) in future studies. Such latent factors could relate to the biophysical properties of the tissue (and thus the efficiency of the tDCS) or individual differences in cognitive status or mindset.
Regarding the tDCS protocol, we relied on using offline tDCS. Thus, it would be interesting to examine whether offline vs. online tDCS would result in a similar effect on AUT and GNGT and their associated functional activity. Also, due to this specific electrode montage (i.e., bilateral bipolar-balanced montage) that we applied in our study, and the non-focal nature of tDCS, we cannot rule out that observed results might reflect a combined effect of stimulation of the IFG and other regions of the frontal cortex, such as the fronto-polar region. Thus, it would be fascinating to elaborate on the tDCS effects on brain activity and network connectivity underlying these performances in the AUT and the GNGT. Lastly, we could not ignore the impact of potential individual differences and variations in the mind states on manipulating creativity through tDCS. We used the sample size of
N = 40, which is relatively similar to what had been used in several previous publications related to original creativity research conducted using tDCS techniques (see Table
1 of Lucchiari et al.
2018). However, this sample size might be at the lower limit of our current observation.