The Interaction Between Intelligence and Creativity at the Neural Level

Contents

Abstract

Introduction

Evolution of Thinking

Reasoning
Cognitive Models of Reasoning
Neuroscience of Reasoning

Intelligence
Neuroscience of Intelligence

Creativity
Neuroscience of Creativity

Instruments
ASK
DWI

White Matter Anatomy
Superior Longitudinal Fasciculus
Inferior Longitudinal Fasciculus
Cingulum
Corpus Callosum
Internal Capsule

Study

Materials and Methods
Participants
Image Acquisition
Data Analysis
Statistical Analysis
Anatomical Labeling

Results

Discussion

Bibliography

Acknowledgment

Many thanks to Dr. Nicola de Pisapia for his supervision and special thanks to Dr. Angelo Bifone for the supports and facilities he provided to make this study possible. We are highly thankful to Dr. Alessia Monti for her supports in cognitive assessments and relative data analysis. Thanks to Stefano Tambalo for his help in learning the basics of imaging data processing. In the end, special thanks to Dr. Jorge Jovicich for the opportunity he provided to enter the world of neuroimaging.

Abstract

Intelligence and creativity are two complementary thinking abilities necessary for the development of different aspects of human life. There are various approaches which try to explain the nature of these two cognitive abilities and many neuroscience studies have been conducted to explore their neural bases. Different models have been proposed to specify the role of various brain networks supporting these two products of cognition. What lacks in the neuroscience literature are the studies which examine the correlation of intelligence and creativity at the neural level. The present research, applying the diffusion-weighted imaging method, is focused on the interaction of these two thinking sections at the connectivity level. Thirty-nine adult subjects participated, inferential and creative thinking modules of ASK test were administered to assess intelligence and creativity. The FA values computed on DTI data which can be an index of connectivity were analyzed through TBSS protocol. The statistical results support the idea that developing both thinking abilities is associated with less connectivity of some brain areas which up to a threshold, their connections support the development of each skill.

In this work section “EVOLUTION OF THINKING” is inspired by Heyes (2012) 1, section “REASONING” by Barbey & Barsalou (2010) 2, section “INTELLIGENCE” by Haier (2017) 3, section “CREATIVITY” by Jung (2014) 4 and Haier (2017) 3, section “ASK” by Faraci et al. (2016) 5, section “DWI” by Jones et al. (2013) 6, section “WHITE MATTER ANATOMY” by Aralasmak et al. (2006) 7.

Introduction

Evolution of Thinking

The human evolution has given rise to new styles of living, social concepts, technologies and other developments which originate from the new forms of thinking. Abilities like causal reasoning, imitation, language, metacognition, and theory of mind can be considered as various evolved faculties of thought which is not exclusive to human beings. Comparing human mind to those of other animals alive today, the origins of thinking can be traced back to the common ancestors of extant eutherian mammals (125 Mya 8), primates (85 Mya9) and great apes (15 Mya 10).

Two main structural components of the human brain which are necessary for the emergence of thinking are neocortex and cerebellum. Conventionally, the neocortex has been considered to relate to higher cognition like planning and executive control while the cerebellum pertains to sensorimotor processing necessary for visually guided reaching and grasping. These two structures have been particularly tightly evolved together not only in primates but, in mammals more generally. Through millions of years, the co-evolution of higher and sensorimotor intelligence, thinking and acting, has added new processes that supervise and control more primitive ways of thinking and has given rise to robust and coordinated ‘embodied' modes of thought.

With this point of view, we can assert that the human thinking is a product of a gradual and incremental ‘co-evolution’ according to peculiar cognitive- developmental mechanisms in a domain-general manner and provide a standard set of computations to process information in a vast range of variety from technical and social domains. Sociocultural experiences, on the other hand, have brought about a set of domain-specific mechanisms which gave rise to abilities like tool-making, planning and imitating. The computational processes applied by these mechanisms are also present in other animals but, their application in human, in range, capacity and flexibility, made a substantial difference 1.

Reasoning

We can describe the reasoning as the process of transition from what is known or hypothesized to what is unknown or implicit in one’s thinking. One form of reasoning is deductive inference where the completeness of the evidence supports the validity of the conclusion. On the other hand, when the evidence is not complete and drawing a conclusion depends on some conditions of uncertainty, the reasoning takes the form of inductive inference. For example, when it is raining there is complete evidence for a deductive inference that a storm has emerged, whereas emerging a storm is an incomplete evidence which provides uncertain support for the inductive conclusion that it will rain.

Cognitive Models of Reasoning

To understand human reasoning ability, cognitive models and their neural correspondences are essential. Barbey and Barsalou 2 summarized cognitive models of reasoning and some evidence from neuroscience studies that are pertinent to these models.

Cognitive models of reasoning mainly follow two distinct theories. One that postulates specialized reasoning modules for the mind 11,12 and predicts that the neural systems for reasoning should be relatively localized implementing modules that are cognitively impenetrable and informationally encapsulated. The other theory named as dual-process theory proposes general-purpose reasoning systems for the mind 13,14,15,16, 17,18,19.

Dual-process in reasoning consists of two general-purpose systems: an associative system and a rule-based system. The associative system is responsible for generating quick and unconscious hypothesis and acts through basic cognitive operations such as association, similarity, and memory retrieval. The rule-based system accounts for conscious and intentional mechanisms, which are evolutionarily more advanced. According to this classification, inductive reasoning applies the associative system as it necessitates retrieval and evaluation of world knowledge, whereas deductive reasoning depends on rule-based, formal procedures. The dual-process model predicts that reasoning recruits different neural systems depending on cognitive demand 2,13. For simple reasoning tasks, the associative system suffices, and the recruited regions are primarily left inferior frontal gyrus, especially Broca’s area, while the temporal lobes and posterior parietal association cortex get involved. Difficult tasks instead, recruit rule-based system to correct reasoning which employs the prefrontal cortex especially the ventrolateral sub-region implicated in rule maintenance.

Barsalou proposes a second theory 20 that provides a general-purpose account for reasoning. According to this theory, reasoning is grounded in a somewhat different pair of systems: the linguistic system and the conceptual system. Like his view, the brain’s language system initially produces relatively superficial information about a reasoning problem such as word associates, syntactic structures and so on. When the linguistic forms are generated, the conceptual system gets increasingly involved in maintaining their meaning representations. An underlying assumption of this theory is that superficial processing based on linguistic representations may often suffice an appropriate reasoning performance. When this type of processing is not sufficient, the generation of conceptual images becomes necessary for more sophisticated reasoning. Consistent with the dual-process model, this approach to reasoning also predicts that different reasoning tasks (i.e., superficial vs. deep processing) may recruit different brain areas according to their cognitive demands.

Barbey and Barsalou assert that both dual coding frameworks predict that reasoning will recruit neural systems that support two forms of coding. One significant set of systems serves language processing employing the left frontotemporal language system. The second one maintains conceptual processing, mental simulation, and imagery which recruits bilateral sensorimotor areas.

Mental-models and Mental-logic theories have applied another account of general-purpose systems to explain the reasoning. According to Mental-model theory 21, in front of a reasoning task, a person generates possible circumstances which are alternative models of its premises and render them valid. If all the models validate the truth of conclusion then, the inference is deductive; otherwise, if only a limited number of models support the truth of the conclusion, then the inference has inductive or probabilistic form. Consecutively, this theory suggests that the underlying neural systems for deductive and inductive reasoning are shared and primarily reside in right-hemisphere regions. As long as mental models have a visuospatial organization, this theory considers a principal role in the reasoning process for parietal and occipital areas implicated in visuospatial processing. In contrast, mental logic theory 22,23,24 considers deductive reasoning as the application of formal syntactic operations in the form of deductive rules thus, according to this view, reasoning recruits left prefrontal and superior temporal regions involved in formal, rule-based operations.

Neuroscience of Reasoning

Deductive Reasoning

As mentioned before, the neural system engaged in a reasoning task depends on the characteristics of the task. One difference is due to the familiarity of semantic content 25. For example, a deductive reasoning task like [All dogs are pets / All poodles are dogs / Therefore, all poodles are pets] engages left frontotemporal system including left inferior frontal cortex (BA 47), left middle/superior temporal cortex (BA 21/22), and left temporal pole (BA 21/38). This system is active also in memory and language tasks that employ familiar semantic content. Linguistic processing plays a central role in such tasks.

On the other hand, considering a task which involves unfamiliar semantic content like [All P are B / All C are P / Therefore all C are B], reasoning employs a bilateral frontoparietal system including bilateral dorsal (BA 6) and inferior (BA 44) frontal lobes, bilateral superior and inferior parietal lobes (BA 7), and bilateral occipital lobes (BA 19). Activation of left prefrontal cortex in both reasoning tasks implies the crucial role of this region in deductive reasoning. Right prefrontal cortex shows activeness only when the reasoning task consists of unfamiliar semantic content or in conditions where a conclusion conflicts with prior beliefs (e.g., No harmful substances are natural / All poisons are natural / Therefore, no poisons are harmful). Barbey and Barsalou 2 conclude that while the language may often be the main component in both of deductive reasoning tasks, spatial/visual processing may often be central in unfamiliar reasoning.

Another kind of task dependency of deductive reasoning regards to whether a reasoning problem produces expected versus unexpected conclusions 26. We can study this kind of reasoning through inhibitory belief problems that to fulfill the task; subjects must inhibit a highly accessible belief that could interfere with the correct conclusion (e.g., No addictive things are inexpensive / Some cigarettes are expensive / Therefore, some cigarettes are not addictive). Here, the reasoning process could have three phases: 1. detect the conflict between the prior beliefs and the logical conclusion, 2. inhibit the response associated with the belief bias, and 3. execute the appropriate reasoning mechanisms. Failing to detect the conflict, and/or failing to impede the answer related to a belief bias would lead to an incorrect conclusion. Reasoning process to a correct conclusion in inhibitory belief problems engages right inferior prefrontal cortex which can be a cue to the role of this region in conflict detection and/or resolution between prior belief and logic. Besides, the ventromedial prefrontal cortex is active during the reasoning which leads to an incorrect conclusion that can reflect its involvement in non-logical or affective processing.

Evaluating the role of different cognitive theories in the explanation of these findings, Barbey and Barsalou summarize that the neural systems which underlie deduction vary considerably depending on task factors and cognitive demand. Consistent with the task specificity hypothesis, the areas that support deduction vary with the familiarity of the materials and according to whether belief violations occur and are detected. These findings show that difficult unfamiliar problems employ more neural regions than for easier familiar ones which are consistent with cognitive demand hypothesis. There is also evidence in favor of modularity view as left prefrontal cortex generally shows activeness during different reasoning tasks suggesting its essential role in deductive inference.

Deductive vs. Inductive Reasoning

Osherson et al. 27, using a categorical syllogism task, compared neural systems engaged in deductive vs. inductive reasoning. The experiment consisted of three conditions – deduction, induction, and baseline. Deduction condition was an inference about a valid or invalid categorical syllogism (e.g., None of the bakers play chess / Some of the chess players listen to opera / Therefore, some of the opera listeners are not bakers). Individuals were to decide whether the conclusion was valid or not. In the inductive reasoning condition, only invalid categorical syllogisms were presented (e.g., Some of the computer programmers play the piano / No one who plays the piano watches soccer matches / Therefore, some computer programmers watch soccer matches). The task was to choose between probable or improbable conclusion. A baseline condition consisted of a categorical syllogism with anomalous semantic content (e.g., All the engineers own a computer / None of the engineers has been to school / Therefore, all the people who own computers are married). Counterbalancing of categorical syllogisms was done to the same materials occur in every condition. Deductive reasoning (deduction minus baseline) showed theactivationintheleft dorsolateral frontal cortex (BA 6). As reported earlier left frontal system showed activation during deductive reasoning tasks which involve familiar semantic content. Analysis of direct comparison between deduction and induction conditions revealed a different pattern of brain activation in bilateral posterior regions, with a right-hemisphere prevalence, including associative visual cortex (e.g., middle and superior occipital gyri, cuneus, precuneus), along with thalamus and right superior parietal lobe (BA 7). These areas have been reported to be involved in visuospatial tasks that require form discrimination and imaginative operations 28,29,30,31. In a study, the researchers observed the activation of these areas in a deductive reasoning task which employed visuospatial materials 32. Activation of visuospatial processing system during such tasks can support the idea that people apply visuospatial representations such as Venn diagrams or Euler circles to reasoning on categorical syllogism 33. Besides, activation was also found in the right anterior cingulate (BA 24/32), implicating attention 28,34, and executive control 35 in deductive reasoning. Evaluation of inductive reasoning (induction minus deduction) revealed an activation pattern in the left dorsolateral frontal (BA 8 and 10) and right insular cortices. Other studies found the involvement of these regions in probabilistic reasoning tasks that require the estimation of relative frequencies 36 and different quantities 37 (e.g., How fast do racehorses gallop?).

A subsequent neuroimaging research 38 studied reasoning through a task that employed conditional statements instead of categorical syllogisms (e.g. 38, If he is an electrician, then he spent two years in night school / He is an electrician and owns a computer / Therefore, he spent two years in high school). Here, visuospatial processing does not seem to be relevant in conditional syllogisms and as a prediction, the new task and materials recruited brain regions which are not active during visuospatial processing. Right inferior frontal cortex (BA 44), right anterior cingulate (BA 24) and right middle temporal cortex (BA 21) were the regions with the most activity revealed in deductive reasoning (deduction minus induction). The researchers concluded that this frontotemporal system forms a logic-specific network in the right hemisphere parallel to the language-specific network in the left hemisphere. We can propose that this right-hemisphere system executes the calculation of mental transformations necessary for a formal deduction. Evaluation of inductive reasoning (induction minus deduction) instead, revealed a massive activation of the brain regions like the left inferior frontal (BA 47) and left insular cortices along with left posterior cingulate (BA 31), parahippocampal (BA 36), left medial temporal (BA 35), and superior and medial prefrontal cortex (BA 9). These areas are compatible with the left frontal (BA 8 and 10) and the insular regions found in the previous study known to be involved in the recall and evaluation of familiar world knowledge for which standard theories of inductive reasoning consider a central role during induction.

Another neuroimaging study 39 developed a category learning task to assess the component processes of inductive reasoning. The study aimed to find neural systems engaged in rule application versus rule inference during category learning. Individuals were presented with novel animals, and their task was to judge whether all the animals in a set were from the same category. To analyze the "rule application" condition; some criteria for category membership were provided. In the rule inference condition, there was no instruction, and the individuals had to infer the rule of membership. According to the computational demands of the task, the study divided the conditions into easy and difficult for further analysis. As the results revealed, rule inference (rule inference minus rule application) majorly employs bilateral hippocampus. Other studies have found that the activation of this area can be modulated by stimulus novelty 40,41,42. On the other hand, rule application analysis (rule application minus rule inference) revealed activation in the pre-supplementary motor area (BA 8). This area is involved in anticipation of motor activity and seems to reflect an anticipatory response to category exemplars while it is absent when the categorization rule is unknown. Beside episodic encoding of novel stimuli, generation and testing of hypotheses are the necessary components of inductive reasoning. For example, to infer the criteria of membership, it is inevitable to generate and test possible rules such as ‘‘has spots on the abdomen’’ or ‘‘has only two appendages’’ for a category of fictional animals. To examine the neural systems which are engaged in hypothesis selection, the study analyzed the task by difficulty interaction: {hard rule induction minus hard rule application} minus {easy rule induction minus easy rule application}. This comparison helps to evaluate the effects of difficulty due to the variations in the features of animals that correspond to or violate the category membership rule. The results revealed activation in the right lateral orbital prefrontal cortex (BA 47 and BA 11) which has shown its involvement in complex reasoning tasks such as analogical and metaphorical transfer 43,44,45. Challenging reasoning problems of different types of reasoning often activate this area.

Conclusion

Findings from various neuroimaging studies of either deduction or induction support the task sensitivity of reasoning. Compatible with the task specificity hypothesis, the neural engagement in a given type of reasoning may depend more on the study-specific tasks and materials than on the kind of reasoning per se.

Furthermore, the latter study revealed a different pattern of activation from the findings of the other study which has found a significant role of the left frontotemporal system in deductive inference in a categorical syllogism task. This difference on the same fundamental reasoning process of deduction could regard to methodological differences between the recruited tasks and materials across studies. Another consideration concerns to the left prefrontal cortex which is often active on deduction tasks while some of the studies discussed in this section have not included it in the pattern of activation. A consistent finding regards to the reasoning about familiar materials which tends to employ left-hemisphere language and knowledge networks while using less frequent problems tends to active bilateral systems that include the right-hemisphere network. This pattern is compatible with theories that propose two different reasoning systems, one for language processing, and one for processing of spatial/visual information.

The neuroscience evidence reviewed help evaluate current psychological theories. It is confirmed that reasoning typically recruits distributed neural systems and the provided indications show inconsistency with the predictions of modularity theory. On the other hand, the results demonstrate that reasoning is not exclusive to the left hemisphere regions for language and rule-based operations instead, reasoning often engages bilateral and posterior areas which are not considered by mental logic theories of reasoning. The other point is the emphasized role of right-hemisphere system predicted by the mental-models theory which seems inconsistent with several left-hemisphere and bilateral activation patterns observed across studies. We can assert that reviewed findings are in line with dual-process theory as long as it predicts the presence of broadly distributed neural systems beside multiple reasoning systems (e.g., associative vs. rule, language vs. knowledge). Effects of task specificity and cognitive demand which have been revealed by reviewed studies can be another confirmation for this theory.

DTI Studies of Reasoning

There are other studies which employed diffusion tensor imaging method to study the white matter integrity associated with reasoning ability. The results of a research conducted by Zachary et al. 46 revealed reasoning associations with the integrity of the corpus callosum likely interconnecting anterior PFC. Anterior PFC is thought to be involved in the processing of relationally complex, abstract information 47 such as those presented in reasoning tasks. They also found the cingulum bundle to be essential for reasoning. These tracts seem to be necessary for the integration of frontal, parietal, temporal, and thalamic regions 48,49. They also presented evidence of the engagement of cortico-thalamic connections in reasoning as revealed by the association they observed between the integrity of the anterior corona radiata and reasoning performance. Previous DTI studies conducted in older adults 50 have confirmed the role of these connections. Penke et al. 51, assessing reasoning as a part of a composite measure of fluid intelligence, found an important role of white matter integrity in subcortico-cortical fibers. These findings confirmed the role of subcortical areas in the regulation of cortical neurons in response magnitude, firing mode, and synchrony which serve as a basis for cognition 52.

Intelligence

Richard Haier 3 believes that reasoning literature should include "intelligence" as a critical word for indexing, and the relationship of reasoning tests to intelligence should be acknowledged in the discussion of results that show brain/reasoning relationships. A few cognitive neuroscience studies of reasoning use the word intelligence while tests of reasoning correlate highly to the g -factor 53. Generally, neuroimaging studies of reasoning revealed neural activation patterns that are consistent with intelligence.

Gottfredson 54 proposes a definition of intelligence which researchers seem to accept it :

[Intelligence is] "a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience … It is not merely book learning, a narrow academic skill, or test-taking smarts. Instead, it reflects a broader and more profound capability for comprehending our surroundings – 'catching on,' 'making sense' of things, or 'figuring out' what to do." 54

According to the "positive manifold" theory 55, all tests of mental abilities positively correlate with each other. That means doing well on one kind of mental ability test can be predictive of doing well on other tests (Figure 1 56).

Figure 1 shows a row of 15 different tests of specific abilities at the lower level. At a higher level, there is a group of similar skills which represent more particular factors: reasoning, spatial ability, memory, speed of information processing, and vocabulary. At the top level, there is g-factor. Scientific evidence confirms that all the factors derived from individual tests have something in common, and this common factor is called the general factor of intelligence, or g. The g -factor reflects the definitions of intelligence as a general mental ability and can be a function of specific skills that individual tests estimate.

Another model of intelligence 57,58 postulates two core factors of general mental ability: crystallized intelligence and fluid intelligence. Crystallized intelligence describes the ability to learn facts based on knowledge and

Abbildung in dieser Leseprobe nicht enthalten

Figure 1. The structure of mental abilities. The g -factor is common to all mental tests. Numbers are correlations that show the strength of relationship between tests, factors, and g. Note all correlations are positive 56.

experience. Fluid intelligence refers to inductive and deductive reasoning for novel problem-solving. Measures of fluid intelligence have shown a high correlation to measures of g, and these two are often considered equivalent. Crystallized intelligence shows relative stability over the life span while fluid intelligence decreases slowly with age.

Neuroscience of Intelligence

The neuroscience studies of intelligence use various measures with high g -loadings and try to find the basic neural system of general intelligence. Haier and Jung 59,60 reviewed 37 neuroimaging studies which employed structural MRI, PET, and fMRI techniques. In their review, they focused on common results among studies irrespective of imaging and assessment methods. Almost 50% of brain regions were common among the studies. The result of their review shows that the salient brain areas were distributed throughout the brain but focalized in parietal and frontal regions. Consequently, Haier and Jung proposed a model and called that the Parieto-frontal Integration Theory (PFIT) of Intelligence. The term of “Integration” in their theory emphasizes that communication among the salient neural systems is a crucial point to the model. They assert that identifying specific brain areas engaged in intelligence does not suffice and understanding the temporal and sequential interactions through networks is the primary goal.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2. The Parieto-frontal Integration Theory (PFIT) showing brain areas associated with intelligence 61.

Figure 2 illustrates the model proposed by Haier and Jung 61. The brain areas of PFIT constitute a global network and subnetworks that support intelligence. The areas are more concentrated in frontal and parietal lobes with left (blue circles) and right (red circles) hemisphere distribution. As shown in the figure with a yellow arrow, a major white matter tract of fibers connects the frontal and parietal lobes. This tract is the arcuate fasciculus and as authors proposed, serves as an essential tract for intelligence.

Functional Analysis of PFIT

PFIT proposes four stages of information processing during problem- solving and reasoning. In stage 1, information enters the posterior regions of the brain through sensory perception pathways. In stage 2, the data subsequently proceed to the associative areas of the brain where relevant memory is integrated. Stage 3 consists of the integrated information flow to the frontal lobes where weighing options and decision occur, and in the last stage 4 motor or speech areas get involved if required. In a complex situation, the procedure is unlikely to be strictly one-way and sequential but rather, multiple and parallel sequences in a back and forth manner through the networks are required.

According to the 4-stage PFIT model, an intelligent brain integrates sensory information in posterior areas and then, the data are further integrated to higher-level processing as they flow to anterior regions. Some PFIT areas have shown to be involved in memory, attention, and language and integrating these basic cognitive processes is fundamental to intelligence. Haier and Jung believe that individual differences in intelligence, whether the g -factor or other specific factors are rooted both in the structural characteristics of the particular PFIT areas and in the way information flows around these areas. Having more gray matter in crucial areas and/or more white matter fibers which connect them or even more efficient information flow across PFIT networks could be the source of inter-personal variability of intelligence.

DTI Studies of Intelligence

There are studies which investigate the correlation between white matter integrity and intelligence. Yu et al. 62, comparing between fifteen mental mentally disabled patients and 79 healthy controls, have reported positive associations between FA and intelligence (measured with Wechsler Adult Intelligence Scale) in the right uncinate fasciculus and the anterior part of the corpus callosum. Schmithorst et al. 63 found positive correlations of IQ scores with FA values in frontal and occipito-parietal white matter areas in a healthy pediatric population. Skranes et al. 64 showed that low IQ adolescents born with very low birth weight had low FA values in the external capsule and inferior and middle superior fasciculus. Barnea-Goraly et al. 65 found that females with fragile X exhibited lower FA values in white matter in frontostriatal pathways and parietal sensory- motor tracts. Peng et al. 66 revealed that the FA of the parieto-occipital central white matter positively correlates with verbal IQ in patients with malignant phenylketonuria.

Creativity

While we can consider intelligence as a stable evolutionary mechanism over the last 1.6 million years, creativity in human seems to be emerged at least, over the previous ∼30,000 years 67. Respect to intelligence, creativity is novel 4.

Creativity is a concept of individual differences which is intended to explain why some people have a higher potential to provide new solutions to old problems than others 68. Creative thinking necessitates changing the point of view about the world and can be a driving force toward civilization 69. How can we characterize creativity and what is a creative process?

Campbell proposes Blind Variation and Selective Retention (BVSR) theory of creative cognition to answer these questions incorporating evolutionary principles 70. According to this theory, human creativity “represent(s) cumulated inductive achievements, stage by stage expansions of knowledge beyond what could have been deductively derived from what had been previously known.” Campbell postulates three necessary conditions for a creative process: a mechanism for introducing [blind] variations, a consistent selection process, and a mechanism for preserving and reproducing the selected variations.

Subsequently, Simonton assessed and extended BVSR theory 71. He criticizes the imprecise definition of "blind variation” and asserts that creativity and discovery could not be blind unless to the extent that the utilities are initially unknown. In contrast to blind variation, he presents "sighted variation" or sighted ideas which are guided by prior applicable ideas (a.k.a. acquired expertise). According to Simonton, the “blind variation” component of the theory of Campbell does not mean the randomly generated ideas and states: "as long as the probabilities of any generated responses are decoupled from their utilities, the responses are blind without the necessity of being random."

According to Campbell’s theory, a creative process starts by simulation or “substitution” of mental representations of the environment while subsequently, the “solution” selection takes place in the mind among a pool of thinking experiments, “according to a criterion which is in itself substituting for an external state of affairs.” Campbell in his work reports multiple examples of thinkers and how they describe their creative thinking and maybe the most famous one is Poincaré who describes four stages of creative thought including preparation, incubation, illumination, and verification 72.

Psychologists have also proposed stages in a creative process 73. In Amabile’s model of creativity 74 five stages have been considered for creative thinking including problem/task presentation, preparation, response generation, response validation, and outcome. Another approach 75 considers problem construction, information encoding, category search, specification, combination and reorganization of best-fitting categories, idea evaluation, implementation, and monitoring the core stages of a creative process. Schuler and Goerlich 76 proposes an 8-stage-process: 1. problem finding (discovering, identifying and defining relevant problems); 2. information search (knowledge and retrieval of pertinent information); 3. concept combination (reorganization of existing categories, finding linksandanalogies); 4. ideageneration (ideation, characterized by originality, fluency, flexibility); 5. solution development (translation of the original idea in a functional solution); 6. idea evaluation (comparison of different solutions, finding pros and cons); 7. adaptation/customization (redesign, fitting of the original idea) and 8. implementation (communication, persuasion, integration). Here, stages can be considered interdependent to a certain degree, and it is often necessary to return to a stage to achieve a satisfactory solution. They believe that stages one to four can be summarized as "creativity," whereas stages five to eight rather represent "innovativeness."

The psychometric assessment of creative ability mainly relies on divergent thinking tasks 77, which involve the generation of creative ideas to open problems (e.g., find creative alternative uses for a brick 78). Scores in such tasks play the role of the measure of creativity or other characteristics like fluency, flexibility, and originality. Divergent thinking ability is a popular indicator of the potential for creative thought 79. Studies have found evidence of this indicator validity corresponding to real-life creativity 80,81.

Neuroscience of Creativity

In the literature there can be found some case studies where people have manifested extraordinary acquired creative ability, often artistic, after developing frontotemporal dementia (FTD) 82, 83, 84. Not all FTD patients demonstrate this ability. These observations can raise the possibility that creativity might be developed in more people if only certain brain conditions change. One probable explanation is that dis-inhibition of neural circuits caused by the disease could be a key element as long as dis-inhibition makes possible the associations among brain areas which do not communicate typically 3. Dis-inhibition of the particular neural networks related to creativity may also be possible without interfering the regular activity of other brain networks responsible for balance, coordination, memory, and judgment. Understanding neural correlates of creativity can help to have a clearer idea.

One early fMRI study scanned the brains of six male professional jazz pianists while performing two tasks of improvisation or over-learned musical sequences 85. The results revealed that compared to the over-learned sequence, improvisation was associated with a bilateral deactivation in some areas especially of the prefrontal cortex (BAs 8, 9, and 46) along with bilateral activation in some distributed regions especially in the frontal lobe (BA 10). Liu et al. 86 conducted a similar fMRI study with 12 male freestyle rap musicians comparing improvisation of rap lyrics to the repetition of previously memorized sequences. Musical background was the same for both conditions. This study also found the pattern of deactivations and activations, consistent with the results of the Limb and Braun study of jazz pianists.

A review by Arden et al. 87 of 45 functional and structural neuroimaging studies of creativity which also included other measures of creative ability other than musical, a similar neural pattern was found along with other brain areas engaged in creative thinking. It was not far from the expectation that the results showed little overlap among the studies.

Jung and Haier 61, focusing on consistencies among imaging and lesion studies, integrated neuroimaging findings from creativity studies. In their work,

Abbildung in dieser Leseprobe nicht enthalten

Figure 3. Different creativity findings from seven MRI studies. Each colored symbol shows activated brain areas related to creativity from a different study. There is little overlap of areas across studies 87.

they avoid problems of task-specific results which they consider a primary source of inconsistency among functional imaging studies. Based on a combination of these studies, they proposed the Frontal Dis-inhibition Model (F-DIM) of creativity. Figure 4 61 shows their model which can also serve for a comparison to the intelligence PFIT model.

A meta-analysis 88 of 34 functional neuroimaging studies which included 622 healthy adults tried to answer whether there is a consistent neural activation pattern across creative ability tasks despite the task diversity. The results reported some consistency among the studies with this limitation that those studies did not consider the deactivation patterns (Figure 5 88).

These results are consistent with F-DIM and other studies, and the central areas show a distribution across frontal and parieto-temporal regions, especially the lateral prefrontal cortex. The tasks in the recent meta-analysis majorly include generation of ideas or a combination of elements. Anterior areas are suggested to be involved in combining ideas creatively while more posterior

Abbildung in dieser Leseprobe nicht enthalten

Figure 4. Frontal Dis-inhibition Model (F-DIM) of creativity. Numbers indicate Brodmann areas associated with increased (up arrows) or decreased (down arrows) brain activity based on a review of studies. Blue is left lateralized; green is medial; purple is bilateral; yellow arrow is anterior thalamic radiation white matter tract 61.

regions may be engaged in freely generating novel ideas 3. We can conclude that regions in both the right and left hemispheres are associated with creativity which confirms that creativity is not an exclusive function of the right-sided brain.

DTI Studies of Creativity

Results of a DTI study 89 suggest negative correlations between creative thinking and white matter integrity within the inferior frontal lobes. They show that lower values of FA principally within left inferior frontal white matter (i.e., regions overlapping the uncinate fasciculus and anterior thalamic radiation) associated with higher scores in divergent thinking tasks.

Takeuchi et al. 90 conducted another DTI study, and positive correlations

[...]