INTRODUCTION

Recent active neurophysiological studies of creativity, are focused on elucidating the patterns of cognitive processes that underlie the mechanisms for finding original solution to a problem, including the role of intelligence and execution control in solving a task or the contribution of insight strategy [1–5]. There is still no consensus in understanding the relationships between creativity and intelligence, apparently due to the fact that these psychometric constructs represent a complex set of mental operations, including attention, memory, imagination, association formation, which are used to varying degrees in solving experimental problems. Recent studies of the interaction between creativity and intelligence, mainly using tomographic methods, indicate their common structural elements in the neural architecture of the brain [5], and psychometric data point to functional similarity of information selection processes [4]. However, it remains unclear whether this similarity is ensured by the generalized or specific neural systems required to perform experimental tasks of verbal or figural nature. For these reasons, the issues whether creativity is a component of intelligence, or on the contrary, is intelligence included in the structure of creativity, or are creativity and intelligence connected through common neural mechanisms of information selection and the use of memory resources, remain relevant.

In addition to the data on wide representation of brain structures in association with both creativity [6–11] and intelligence [12–16], there is evidence that the association of these cognitive constructs relies mainly on the interaction of frontal cortical areas involved in executive functions and the posterior cortical regions or frontoparietal network (FPN) and default mode network (DMN) systems [5, 6, 16]. Moreover, the executive control of attention is considered as a factor that regulates the contribution of fluid intelligence and “mind wandering” to creative thinking, which at its different stages have both positive and negative value [5].

At present, much attention is paid to the mechanisms of FPN and DMN cooperation, since their pre-tuning at rest and the dynamics of functional interaction is associated with individual variety of task performance strategies in both verbal and figural creativity tests [3, 17]. For example, it was demonstrated that the FPN was a mediator of the association between verbal creativity and activity of the anterior DMN, and between figurative creativity and the DMN posterior part [17]. The increased contribution of the FPN reflects the dominance of executive control in the search for new solutions to the problem, while the DMN contribution reflects spontaneous generation of ideas [6, 7, 18]. According to different authors, creative thinking is accompanied by dynamic reorganization of these systems with regional expansion of functional neural networks, including the left middle temporal gyrus and auditory system for verbal creativity [19] or the temporoparietal and prefrontal areas for figurative creativity [20]. Moreover, in the latter case, the specificity of the involvement of individual areas of the indicated cortical regions was determined by factors formed from different creativity indices. In particular, the factor related to fluency and originality was found to be positively associated with brain volume in structures close to the DMN, and the factor that combined indices of elaboration of the drawing and fixation resistance was associated with the frontotemporal regions.

In the process of data accumulation, the initial idea on the right hemisphere dominance in solving experimental creative tasks [21] was changed, since the involvement of the left dorsolateral prefrontal cortex [22] or the left anterior cingulate cortex [20] was demonstrated, for example, in non-verbal creativity testing. Moreover, lateral changes of functional connectivity measured in the dorsolateral prefrontal cortex associated with nonverbal creativity were found to be dependent on FPN and DMN activity [23]. At the same time, new evidence of the crucial importance of right frontal cortex functions for the success of divergent thinking has appeared, which were obtained on the basis of the analysis of different intelligence components, namely fluid, crystallized (which was tested using verbal tasks), and visual-spatial [5]. Therefore, to understand the formation patterns of different functional neural systems that provide different forms of creativity, and the features of their reorganization depending on the structure of intelligence, further studies are required.

To study the functional importance of the resting brain, not only its tomographic, but also encephalographic characteristics are used [1, 23–25]. Among different EEG frequency ranges, synchronization/desynchronization of α-oscillations is considered more often than others, which is due to their information content in relation to the specificity of inhibition/activation processes in neural networks reflecting both creativity and intelligence (IQ) [1, 26–28]. It was demonstrated that the fluid IQ variability was associated with the updating of working memory, while the creativity predictors were represented not only by this component of the executive system, but also by inhibitory functions [27]. The balance of resting state activity in the frontal and posterior cortex may reflect the individual style of solving problems, including the preference for insightful or analytical strategy [2, 27, 29, 30]. Moreover, not only α, but also low-frequency ∆-, θ-, and high-frequency β-oscillations are considered as indicators of such a balance [18, 24, 31–33]. The ∆-rhythm is of interest for the analysis of creativity, since an increase in its power reflects the suppression of learned dominant behavior and indicates the effectiveness of new learning, while θ- and β-oscillations are considered as encephalographic correlates of information processing due to DMN and FPN functions [31, 34].

Previously, we revealed an increase in the interaction of neural ensembles of the anterior cortex with the posterior parts of the left hemisphere in individuals characterized by relatively high levels of intelligence and creativity compared to those who had lower levels [1]. Later, it was found that frontal ∆- and β-rhythms could serve as predictors of both non-verbal creativity and verbal intelligence [32]. The objective of this study was to elucidate the regional features of the ∆- and β‑rhythm patterns, as pre-tuning for the realization of verbal or non-verbal creative activity and its potential connection with the values of the verbal and visual-spatial components of intelligence.

MATERIALS AND METHODS

The study involved 37 individuals (university students, 18 ± 1.1 years old; 27 females and 10 males).

To determine verbal and figurative (visual-spatial) components of intelligence, the Amthauer intelligence structure test was used, i.e., mean values of the two verbal (subtests 2 and 3) (IQv) and two non-verbal visual-spatial tasks (subtests 7 and 8) (IQs) performance. Non-verbal figurative creativity was assessed using the Torrance “Circles” and “Incomplete figures” subtests, while the verbal one was assessed using the Guilford’s Alternative Uses Test and creation a meaningful sentence with the inclusion of three stimulus words (nouns from distant semantic categories). The originality of response indices upon the performance of the first three methods were calculated on the basis of the corresponding database as reciprocal of a number of the same ideas [35]. The sentence originality was assessed by three trained experts with experience in working with this technique (Cronbach’s α for the estimates was 0.82).

EEG recording in the state of quiet wakefulness with eyes closed was performed using Mitsar-201 equipment and software (Russia) through 19 leads (Fp1, Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, T6, O1, O2) arranged according to the 10/20 system with an integrated ear reference electrode and ground electrode in lead Fpz. The electrode resistance was less than 5 kOhm. The bandpass filter, 50/60 Hz; high- (HP) and low-pass (LP) filter parameters, 0.53 and 50 Hz, respectively. For the analysis of brain activity, 2-s artifact-free EEG segments with an overlap of 50% with a total duration of 60 s were selected. The sampling rate was 250 Hz. For each electrode site, the EEG spectral density was calculated using the fast Fourier transform method in six frequency ranges: ∆ (1–4 Hz), θ (4–7 Hz), α1 (7–10 Hz), α2 (10–13 Hz), β1 (13–20 Hz), and β2 (20–30 Hz). For statistical analysis, the natural logarithm of EEG power values in the ∆- and β2-ranges was used according to the goal of the study.

Statistical data treatment was performed using the Statistica 13.3 (SN: JPZ912J057923CNET2ACD-K) licensed software package.

RESULTS

Table 1 shows the results of testing the intelligence and creativity indices.

Table 1.   Descriptive statistics for the intelligence and creativity indices
Full size table

Cluster organization of the creativity and intelligence indices. At the first step of data analysis, organization of the creativity and intelligence indices was examined by means of agglomerative hierarchical clustering of pre-normalized data (normalization was performed by dividing by the mean value). Evaluation of the different methods showed that the best clustering was provided by Ward’s method with the Euclidean metric (Fig. 1). Similar structure of clusters that combine the intelligence and creativity indices is provided by complete-linkage clustering, which is based on determining the maximum pairwise distances between objects in different clusters. Since Ward’s method is based on determining minimum variance in hypothetical clusters, the formed hierarchical structure of intelligence and creativity indices can be considered optimally stable for interpreting the result.

Fig. 1.
figure 1

Dendrogram of proximity measures for the indices of verbal and non-verbal components of intelligence and creativity according to Ward’s clustering. IQv, verbal component of intelligence; IQs, visual-spatial component of intelligence; Оc, originality index upon the performance of the “Circles” subtest; Оuu, originality index upon the performance of the “Unusual use” subtest; Оif, originality index upon the performance of the “Incomplete figures” subtest; Оsm, originality index upon the performance of the “Sentence making” subtest.

Full size image

The cluster structure shown in Fig. 1 indicates the closest relationship between IQs and figural originality upon the performance of the “Circles” subtest, and grouping this cluster together with IQv and the verbal originality measure for the “Alternative Uses Test” into a common cluster (cluster 1). The originality of response indices upon the performance of the other two subtests (“Incomplete figures” and “Sentence making”) form a cluster separate from the first group of variables, which points to the different strategy for finding a solution to the problem (cluster 2).

For further analysis of the EEG data, samples were formed by clustering of the intelligence and creativity indices with the help of the K-Means algorithm, which made it possible to separate two groups that differed in the values of the considered indices. Composition of these groups was adjusted based on the results of hierarchical clustering. In particular, GRCIQ consisted of individuals who showed relatively high scores in the “Circles” and “Alternative Uses Test” (included in cluster 1), and GRC consisted of individuals with relatively high scores in the “Incomplete figures” and “Sentence making” subtests (included in cluster 2).

Analysis of variance of all creativity indices showed the effect of GROUP factor (F5,166 = 6.85; P < 0.0001; η2 = 0.17). A two-way ANOVA with independent variables, SEX (2) and GROUP (2), performed for each measure of intelligence and creativity revealed the effect of only GROUP factor (2.69 < F1,33 < 12.5, 0.001 < P < 0.1) with higher IQ values, Oc and Ouu in GRCIQ, versus relatively low IQ values, but higher Oif and Osm values in GRC (Table 2).

Table 2.   Quantitative composition and normalized intelligence and creativity indices in two groups formed on the basis of the results of cluster analysis using the K-Means method
Full size table

Thus, the objective of further study was to elucidate the EEG correlates of creativity, which differed in groups formed according to the identified different strategies for finding a solution to the problem. The analysis limited us to only ∆- and β-rhythms, since their informative value as predictors of non-verbal creativity and verbal intelligence was previously demonstrated [32].

Cluster organization of ∆- and β2-rhythms in groups differing in the structure of creativity and intelligence. To elucidate the patterns of frequency-spatial organization of low- and high-frequency oscillations in the selected groups, GRCIQ and GRC, Ward’s hierarchical agglomerative clustering with a Euclidean metric was also used. Figure 2 shows the dendrograms of formed clusters obtained for these groups for the ∆-rhythm, and Fig. 3, for the β2-rhythm.

Fig. 2.
figure 2

Dendrograms of ∆-rhythm clusters formed using Ward’s method and representing groups that differ in the creativity organization: correlated with intelligence (a) and not correlated with intelligence (b).

Full size image
Fig. 3.
figure 3

Dendrograms of β2-rhythm clusters formed using Ward’s method and representing groups that differ in the creativity organization: correlated with intelligence (a) and not correlated with intelligence (b).

Full size image

According to the ∆-rhythm cluster organization, two clusters were distinguished in both groups. However, GRCIQ was characterized by greater distance of the cluster representing the anterior cortical areas with the leading role of oscillations in F8. At the same time, in GRC the resulting clusters were similar in terms of connectivity, and in the combination of the anterior cortex areas the leading role belonged to F7 and F8 (Fig. 2) (the specificity of the inclusion of these sites in the clusters identified for GRCIQ and GRC was proved by an additional cluster analysis of the ∆-rhythm only for the anterior part of the cortex).

Regional organization of the β2-rhythm is represented by two clusters in GRCIQ (Fp1 and Fp2 form one cluster, the remaining electrode sites form the second cluster) and four clusters in GRC (sites Т3, Т6, and Т4 reflect three separate clusters, and all other sites are included into another, common cluster) (Fig. 3). Moreover, the leading role in grouping of two β‑rhythm clusters in GRCIQ is played by the region represented by Т4, and in GRC, by F7, F8 and Fp1.

To determine the degree of interaction between high-frequency β2 and low-frequency ∆ biopotentials in GRCIQ and GRC, a correlation analysis was performed, the results of which are shown in Fig. 4 (0.50 < r < 0.63 at 0.008< P < 0.05; the sites for the ∆‑rhythm are shown vertically, and the sites for β2 are shown horizontally).

Fig. 4.
figure 4

Correlation maps between ∆- (vertical leads) and β2 (horizontal leads) rhythms for groups that differ in the organization of creativity: correlated with intelligence (GRCIQ, a) and not correlated with intelligence (GRC, b).

Full size image

The data indicate that the differences between groups are represented by more pronounced associations between the ∆-rhythm in the frontal cortex and β2-oscillations in the posterior cortex for GRCIQ (Fig. 4a), while GRC is characterized by more widely distributed correlations of β2- and ∆-rhythms with the concentration in central-parietal cortical areas (Fig. 4b).

Thus, it can be concluded that the two selected groups of study participants, differing in the level of intelligence and originality of responses while performing the tasks which require the rejection of stereotypical ideas (GRCIQ) or solve a problem under conditions of a given variety of stimuli (GRC), are characterized by different forms of regional organization of ∆‑ and β2-rhythms and their relationships. In GRCIQ, two clusters were identified in each frequency range, representing different forms of association of the anterior and posterior cortical regions, reflected in the correlation of low-frequency biopotentials of the frontal cortex and the generalized high-frequency activity of β‑activity. GRC is distinguished by regionally more differentiated clustering of β2-activity at its diffusely distributed correlation with the ∆-rhythm while the exclusion of anterior frontal areas.

DISCUSSION

The results indicate that the stimulus material itself, despite the general instruction “to be original,” determines the strategy for finding a solution to the problem, even to a greater extent than its verbal or figural nature. Repetitive stimuli (circles or an ordinary object) contribute, first of all, to the generation of stereotyped responses. Therefore, in order to reject them and continue the search with critical assessment of the ideas, arising during the search, such pre-tuning of resting-state brain activity is required, in which the prefrontal cortex is involved (Figs. 2a–4a). This situation is characterized by a combination of originality and IQ indices (Fig. 1). Under the conditions of testing creativity using a variety of stimuli belonging to different semantic categories, the response strategy in a widely represented network of associations is preferable for successful completion of a task, which is reflected by the dominance of diffusely associated areas of the posterior cortex (Figs. 3b, 4b). It can be suggested that such functional association of different cortical regions is provided by the synchronization of neural networks at the ∆- and β-range frequencies. In particular, with the dominance of prefrontal regions in the case of a tendency to the strategy of critically conditioned rejection of the stereotype, but temporo-parietal-occipital regions, in the cases of the given stimuli causing actualization of distant associations. It is noteworthy that in the latter case, there is a relatively greater effect of synchronization of the ∆- and β2‑rhythm amplitudes in the sites of the right hemisphere (Т4, О2), while in the first case, of the left hemisphere (Fp1, F7). That is, the observed effect of a change in hemispheric dominance associated with creativity [20–22] can be explained by individual preferences in using the strategies of “intellectual” or “spontaneous” realization of divergent thinking, accompanied by a shift in activity in the FPN and DMN upon problem solving [23]. The data testify to the importance of the resting-state FPN and DMN and are in good agreement with data on different regional associations of creativity components, i.e., fixation resistance, with frontotemporal areas, and the originality, with DMN [20].

The “intellectual” strategy is understood as an internally directed search for a response, the originality of which is determined by intellectual abilities (resources of knowledge and logical thinking), while the “spontaneous” strategy is defined as the insurance of divergent thinking due to a given variety of stimuli. The success of divergent thinking is mediated by the involvement of different components of intelligence, the combination of which makes it possible to predict about 46% of neural networks involved in creativity [5], and the reason for its failure may be the lack of desire to search for information [36].

According to the literature data, the biopotentials of the ∆-range are considered as an indicator of the motivational component of activity that modulates functional activity of neural networks [37, 38], while β-activity reflects the information load and factors of cognitive control reorganization through the FPN functions [29, 39, 40]. In this case, the revealed regional specificity of the correlation patterns of these rhythms in GrCIQ and GRC can be interpreted as motivational pre-tuning of flexible reorganization of the selective processes in order to search for an original idea, based on executive control or a system of distant semantic associations, respectively.

Thus, analysis of the resting state EEG makes it possible to assess the individual resources of the reorganization of the brain structures depending on the conditions for creativity testing, which first of all require the flexibility of thinking or fluency in the search for ideas. It seems likely that due to the summation of such different strategies for solving a problem, in psychometric creativity testing, different forms of relationships between originality, fluency, and flexibility are observed [41, 42]. This fact of different ways of achieving a result is emphasized in the dual pathway to creativity model, i.e., as a function of flexible thinking and perseverance [43], or the association of originality and flexibility indices not only with different brain structures, but also with their multidirectional associations [44, 45]. Thus, as one of the factors of the observed diversity in the interaction of brain structures in the analysis of creativity [4, 8–11], different testing conditions should be considered, since the tasks most often used in both EEG and fMRI studies are “Alternative Uses Test” or “Incomplete figures” are different in nature (verbal and non-verbal, respectively) and, as was demonstrated in the present study, require different strategies for finding an answer.

CONCLUSIONS

The revealed frequency-spatial patterns of ∆- and β2-rhythms and their relationships reflect pre-tuning of functional cortex activity to use different strategies for searching for an original idea in creativity testing. The combination of low-frequency biopotentials of the frontal cortex and generalized high-frequency β‑activity can be associated with the strategy of an “intelligent” search for an original answer under the pressure of stereotypical decision. Differentially presented clustering of β2-activity with its more diffusely distributed correlation with the ∆-rhythm with the exclusion of the anterior frontal areas reflects the search for a solution based on spontaneous, less controlled associations.