Is creative insight task-specific? A coordinate-based meta-analysis of neuroimaging studies on insightful problem solving

    Is creative insight task-specific? A coordinate-based meta-analysis of neuroimaging studies on insightful problem solving Wangbing Shen, Yuan Yuan, Chang Liu, Xiaojiang Zhang, Jing Luo, Zhe Gong PII: DOI: Reference:

S0167-8760(16)30716-4 doi:10.1016/j.ijpsycho.2016.10.001 INTPSY 11178

To appear in:

International Journal of Psychophysiology

Received date: Revised date: Accepted date:

14 May 2016 24 August 2016 2 October 2016

Please cite this article as: Shen, Wangbing, Yuan, Yuan, Liu, Chang, Zhang, Xiaojiang, Luo, Jing, Gong, Zhe, Is creative insight task-specific? A coordinate-based meta-analysis of neuroimaging studies on insightful problem solving, International Journal of Psychophysiology (2016), doi:10.1016/j.ijpsycho.2016.10.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT RUNNING TITLE: TASK-DEPENDENCE OF BRAIN-BASED INSIGHT

T

Is creative insight task-specific? A coordinate-based meta-analysis of

IP

neuroimaging studies on insightful problem solving Wangbing Shen1, 2, Yuan Yuan2, *, Chang Liu2,*, Xiaojiang Zhang2, Jing Luo3, 4, *, Zhe Gong 2 2

School of Public Administration and Institute of Applied Psychology, Hohai University, China

SC R

1

School of Psychology and Laboratory of Cognitive Neuroscience, Nanjing Normal University, China 3

Beijing Key Laboratory of Learning and Cognition, Capital Normal University, China

Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences, China

NU

4

Acknowledgments—This work was supported by the National Natural Science Foundation of

MA

China (31500870), the National Social Science Foundation of China (15BZX129), the Fundamental Research Funds for the Central Universities (2014B15314), and Beijing Municipal Commission of Education Key Program of Science and Technology (KZ201410028034).

TE

D

* Corresponding authors.

Date of Submission: August 23, 2016

AC

CE P

Address: Chang Liu School of Psychology Nanjing Normal University Nanjing 210097, No. 122, Ninghai Road, China Tel: +86-25-86212351 Email: [email protected]; [email protected] And Jing Luo Beijing Key Laboratory of Learning and Cognition, Department of Psychology, Capital Normal University, Beijing 100191, China Email: [email protected]

1

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Is creative insight task-specific? A coordinate-based meta-analysis of neuroimaging studies on insightful problem solving

T

Abstract: The question of whether creative insight varies across problem types has recently come

IP

to the forefront of studies of creative cognition. In the present study, to address the nature of

SC R

creative insight, the coordinate-based activation likelihood estimation (ALE) technique was utilized to individually conduct three quantitative meta-analyses of neuroimaging experiments that

NU

used the compound remote associate (CRA) task, the prototype heuristic (PH) task and the Chinese character chunk decomposition (CCD) task. These tasks were chosen because they are

MA

frequently used to uncover the neurocognitive correlates of insight. Our results demonstrated that

D

creative insight reliably activates largely non-overlapping brain regions across task types, with the

TE

exception of some shared regions: the CRA task mainly relied on the right parahippocampal gyrus, the superior frontal gyrus and the inferior frontal gyrus; the PH task primarily depended on the

CE P

right middle occipital gyrus (MOG), the bilateral superior parietal lobule/precuneus, the left inferior parietal lobule, the left lingual gyrus and the left middle frontal gyrus; and the CCD task

AC

activated a broad cerebral network consisting of most dorsolateral and medial prefrontal regions, frontoparietal regions and the right MOG. These results provide the first neural evidence of the task dependence of creative insight. The implications of these findings for resolving conflict surrounding the different theories of creative cognition and for defining insight as a set of heterogeneous processes are discussed. Key words: insight; neuroimage; prototype-heuristic; chunk decomposition; remote association

2

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT 1 Introduction Creative insight, a type of creative cognition, has an important role in advancing social

IP

T

development and in enabling adaptation to an increasingly challenging world. Abundant evidence

SC R

from historical documents and anecdotal recordings has demonstrated the importance of creative insight in scientific discoveries, technological innovations, and artistic creations as well as in an individual‘s success (see Dietrich, & Kanso, 2010; Radel, Davranche, Fournier, & Dietrich, 2015).

NU

As a complex and multifaceted process, creative insight has been scientifically examined for only

MA

a century since Kohler‘s pioneering research. With the development of neuroscientific techniques, especially the coupling of functional resonance imaging (fMRI) and traditional cognitive measures,

D

a new approach to creative insight, namely, brain-based insight, is being characterized. This

TE

approach has greatly expanded understanding of the essence of creative insight by contributing

CE P

new knowledge of the neural and brain correlates that underlie the creative process and by offering strategies to facilitate insight via evidence-based brain stimulation. However, some key theoretical

AC

problems remain. One important problem currently under debate is whether creative insight is task-specific or task-independent. This question is rooted in heated controversy concerning the domain-general and domain-specific theories of creative cognition (Chen, Himsel, Kasof, Greenberger, & Dmitrieva, 2006; Dow & Mayer, 2004; Reiter-Palmon, Illies, Cross, Buboltz, & Nimps, 2009). 1.1 Domain-general and domain-specific hypotheses of insight Insight is ambiguous due to its suddenness, directness and continuousness (e.g., Epstein, Kirshnit, Lanza, & Rubin, 1984). The psychological nature of creative insight is both of great interest and highly debated. Sternberg and Davidson published a landmark book, The Nature of Insight, in which creative

3

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT insight was examined using different influential approaches that were divided into two categories: domain-general and domain-specific. According to domain-general theory, ―insight problems are

IP

T

thought of as a single class of problems that all require the same general problem-solving strategy‖

SC R

(Dow & Mayer, 2004). As a result, creative insights underlying the successful solution of these problems are very similar and do not have distinct cognitive mechanisms (recruiting different cognitive skills or sub-processes) or neural bases (do not activate the same brain regions and connections). In

NU

contrast, domain-specific theory argues that ―insight problems can be broken down into coherent

MA

subcategories such as verbal, mathematical, and spatial insight problems, each requiring a different type of problem-solving strategy‖ (Dow & Mayer, 2004), which suggests that the cognitive or neural

D

underpinnings of creative insight are modulated by domain and even cognitive tasks. In other words,

TE

insight might vary depending on the different processes needed to solve a problem.

CE P

Considerable evidence concerning differences in creative thinking (Kaufman & Baer, 2005; Plucker & Zabelina, 2009) and insight (Dow & Mayer, 2004; Weisberg, 1995; Cunningham,

AC

MacGregor, Gibb, & Haar, 2009), particularly regarding task performance in different domains and the origins and causes of such differences, has been presented. To address the issue of domain specificity, especially the task-independence of creativity, Reiter-Palmon et al. (2009) required participants to solve one of three realistic creative problems that differed in terms of their complexity, involvement, and problem-based efficacy. They observed that the participants‘ creativity was influenced by the type of problem solved (problem type accounted for approximately 4-12% of the variance after the potential impact of individual ability was excluded) as well as by the measure of creativity used to assess the solution, which suggests that not all real-world creative problems are equivalent and emphasizes the importance of how problem solvers respond to different creative thinking tasks. Similarly, Dow and

4

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Mayer (2004) directly examined the training effect of task type on insight problem solving and observed that the successful solution of different insight problems depended on distinct cognitive

IP

T

strategies or abilities. Specifically, they found that solutions to verbal insight problems mainly relied on

SC R

the definition and analysis of the terms included in a problem; solutions to mathematical insight problems primarily depended on a novel approach to numbers; and solutions to spatial insight problems principally depended on the removal of self-imposed constraints. In the study, 158 participants were

NU

required to complete a series of measures that tested motivational and personality traits as well as

MA

intellectual abilities in addition to three creative thinking tasks, which involved artistic, verbal and mathematical domains, with different instructions. The results showed that with the exception of one

D

intellectual factor that was linked with the domain-specific component of mathematical creativity under

TE

the explicit ―be creative‖ instruction condition, all of the other measures and cognitive tasks were

CE P

associated with or relied on domain-general components of creativity, providing strong evidence of domain generality and little evidence of domain specificity (Chen et al., 2006). Therefore, whether

AC

creative insight is a domain-general process that transcends specific domains or tasks (i.e., task-general or task-independent) or encompasses a range of domain-specific processes that vary from domain to domain or from task to task (i.e., task-specific or task-dependent) remains an open question that is worthy of further research. 1.2 Three task types commonly used to explore insight Most brain imaging studies concerning creative insight have found that brain-based insight can be investigated and examined through many different tasks. In addition to the widely used compound remote associate (CRA) problems, the prototype-heuristic (PH) task, the riddle-guessing task (e.g., Mai, Luo, Wu, & Luo, 2004), the degraded picture recognition task (e.g., Ludmer, Dudai, & Rubin, 2011),

5

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT the number reduction task (NRT; e.g., Darsaud et al., 2011), the classical insight problems solving task (e.g., the Matchstick problem; Goel & Vartanian, 2005), the chunk decomposition (CD) tasks ①, and the

IP

T

magic demystifying task (Danek, Fraps, von Müller, Grothe, & Öllinger, 2013) have also been used to

SC R

investigate cognition, especially the neural mechanisms underlying creative insight. Due to their domain characteristics and extensive use, this work focused on three types of task②: the CRA task, the PH task and the CD tasks. The following section will provide a brief introduction to these three tasks.

NU

1.2.1 The CRA task

MA

The CRA task, which was created by Bowden and Jung-Beeman (1998, 2003), is a variant of the remote associates task (RAT) developed by Mednick (1962). The broad utility of the CRA task in

D

neuroscience studies of creative insight was highlighted by Jung-Beeman et al. (2004), although this

TE

task was already being used to determine the behavioral and psychological correlates of creative insight

CE P

prior to this publication. This task involves a set of three words, for which participants are required to identify a solution word that can individually form a compound word or phrase with each of the three

AC

given words. For example, if the three problem words ―tree,‖ ―sauce,‖ and ― pine‖ are presented, the participants must identify a common word, such as ―apple,‖ that can be matched with each of them, forming ―apple tree,‖ ―applesauce,‖ and ―pineapple‖. In general, the CRA task has at least two main advantages. First, because the problems associated with the task can be solved in a relatively short period of time, many trials can be attempted in a single experimental session. Second, this task is easier

①

There may be some differences between, for example, matchstick algebra (with mathematical processes) and

logographic character decomposition (without any mathematical process). However, they involve a common process (CD), and it may be more reasonable to call them CD tasks (plural) due to the above differences. ②

The three insight tasks were chosen as the targets of this work for the following reasons: First, these tasks are

the most commonly used in neuroimaging studies of spontaneous insight; therefore, a sufficient number of studies was available to generate reliable results for the present meta-analyses. Second, the three tasks are the most commonly used insight tasks and involve more than one well-defined domain. 6

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT to administer than other classical insight problems, which allows the exclusion of as many extraneous variables as possible (Bowden & Jung-Beeman, 2003). These advantages are responsible for the

IP

T

extensive utilization of the task in studies of insight problem solving, creative thinking (e.g., Lee &

SC R

Therriault, 2013), and associative thought (e.g., Razumnikova, 2007). The CRA task is considered a suitable tool for measuring convergent thought (Chermahini, Hickendorff, & Hommel, 2012; Cerruti & Schlaug, 2009) and creative insight (Razumnikova, 2007) and is not only implemented in English

NU

speaking countries, having also been translated into Russian (Razumnikova, 2007), Dutch (Chermahini,

MA

et al., 2012), German, Hebrew, Japanese, and Jamaican versions (Bowden & Jung-Beeman, 2003; Shen, Yuan, Liu, Yi, & Dou, 2016). Thus, this task can be applied for behavioral and neuroscientific studies

TE

1.2.2 The CD tasks

D

across a range of contexts.

CE P

Every stimulus can be considered a ―chunk‖ or a portion of a chunk. When an individual encounters an unfamiliar task, he or she tries their best to familiarize themself by automatically

AC

recognizing instances of chunks in the environment; however, in most cases, the problem solver does not know which previously acquired chunks are relevant for the solution. The term ―chunk decomposition‖ comes from representation restructuring theory (RCT; Ohlsson, 1992; Knoblich, Ohlsson, Haider, & Rhenius, 1999). RCT proposes that successful insight can be achieved through constraint relaxations or CD, which destructs familiar patterns into elements (e.g., strokes or radicals; see Luo & Knoblich, 2007, for details) that can be regrouped in another meaningful way. In simple terms, CD is the opposite of chunking. As a specific form of insightful thinking, CD has been shown to be linked with special perceptual, linguistic, and executive processes distinct from ordinary modes of thinking (Huang, Fan, & Luo, 2015).

7

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Knoblich and colleagues (1999) first introduced the term CD in empirical studies of insight and defined CD sub-processes underlying matchstick arithmetic problems involving Roman numerals. As a

IP

T

logographic language system, Chinese characters are ideal cases of perceptual chunks. Specifically,

SC R

Chinese characters are composed of (nonsensical) radicals, which in turn are composed of strokes that carry meaning. In general, one Chinese character consists of several strokes, which are considered sub-chunks of the character (Luo & Knoblich. 2007). Due to the highly structured property of Chinese

NU

characters, Luo, Niki and Knoblich (2006) developed Chinese character chunk decomposition (CCD),

MA

which has been widely adapted and used to reveal the neurocognitive mechanisms underlying CD and creative insight (e.g., Luo et al., 2006; Wu, Knoblich, Wei, & Luo, 2009; Wu, Knoblich, & Luo, 2013;

D

Tang, Pang, Nie, Conci, Luo, Luo, 2015; Huang et al., 2015). In the Chinese CCD task, two Chinese

TE

characters are placed on the left and right ends of a line. Participants are required to remove a portion

CE P

of the right character and add it to the left one to form two new valid characters (see Luo et al., 2006). Given the adequacy and superiority of Chinese characters for this purpose, nearly all neuroscientific

AC

studies of the neural bases of CD have adapted Chinese characters as materials.

1.2.2 The PH task

Heuristics are very important shortcuts for the insight that results in effective presentations and solutions. The PH task was designed according to Prototype Heuristic Theory (PHT), which was originally proposed by Zhang and colleagues (Luo et al., 2011; Qiu et al., 2010; Luo, Li, Qiu, Wei, Liu, & Zhang, 2013); according to PHT, insight can be reached through two stages: the ―representations connection‖ stage and the ―relationship mapping‖ stage. In the first stage, problem solvers identify similarities between the representations of the required function in a problem and the feature function in a prototype based on established semantic representations of the problems and prototypes, forming a

8

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT connection between the representations of the technical problem and the prototype. In the second stage, ―relationship mapping‖ is further established; namely, the relationship between a certain construction

IP

T

and the feature function of the prototype is mapped onto the representation of the technical problem

SC R

(Ming, Tong, Yang, Qiu, & Zhang, 2014). Briefly, PHT proposes that insight occurs if the critical heuristic information embedded in the prototype is suddenly obtained, providing the orientation needed to identify the correct solution to the problem from among the possible solution space (Ming et al.,

NU

2014). In the PH task, participants are asked to learn a well-defined prototype containing heuristic

MA

information (e.g., having the same mapping rule) and to then work out on their own the subsequently provided insight problems (e.g., the solution to the riddle ―有口难言‖, which means ―being unable to

D

speak even with a mouth‖, is ―哑‖, which literally means ―mute‖) that are superficially semantically

TE

irrelevant to the base or prototype problems (the solution to the problem ―有眼难见‖, which means

CE P

―being unable to see even with eyes‖, is ―盲‖, which literally means ―blind‖). To date, this task has been extensively utilized to uncover the neural correlates of creative insight in knowledge-sparse and

AC

knowledge-rich fields through the use of two types of well-validated stimuli: Chinese logogriphs (e.g., Qiu et al., 2010) and scientific innovation problems (e.g., Luo et al., 2013; Tong et al., 2015).

1.3 The goal of the present study The present work aimed to determine whether domain-generality theory or domain-specificity theory more accurately describes insight. One key problem is whether insight is task-independent or task-dependent. According to domain-generality theory, insight problems are considered a single class of tasks that all require the same general problem-solving strategy. In contrast, domain-specific theory argues that insight problems can be divided into coherent subcategories, each of which can be solved through the same strategy (Dow & Mayer, 2004). From the perspective of task characteristics, the key

9

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT to the problem of whether creative insight is domain-specific or domain-general is the dependence of the neural or mental processes that underlie creative insight on the specific task used to measure it. In

IP

T

other words, the question of whether creative insight is task-specific reveals the nature of creative

SC R

insight, particularly in relation to individual insight in problem solving.

As mentioned earlier, many studies have noted the potential influence of task type on creativity and have proposed the necessity of conducting a detailed comparison of multiple tasks. The majority of

NU

previous studies used only one creative insight task, with the exception of a few investigations that

MA

adapted several creativity tasks (Abraham, Beudt, Ott, & von Cramon, 2012; Fink, Grabner, Gebauer, Reishofer, Koschutnig, & Ebner, 2010; Reiter-Palmon et al., 2009; Kaufman & Baer, 2005). Moreover,

D

several studies (e.g., Chen et al., 2006; Kaufman & Baer, 2005) have noted that existing arguments in

TE

favor of domain-specificity theory and domain-generality theory have various weaknesses. For

CE P

example, the most fundamental shortcoming of domain-specificity theory is that it relies ―almost exclusively on the absence of good evidence favoring domain-generality rather than on direct evidence

AC

either for domain-specificity or against domain-generality‖ (Chen et al., 2006). Therefore, it remains unclear whether some brain regions are involved in only certain insight tasks while others are specifically engaged in other insight tasks. Given that very few empirical studies (i.e., Abraham et al. 2012) have examined potential differences in the neurocognitive mechanisms underlying the various tasks (e.g., the CRA task and the candle problem, which measures classical insight) used to investigate creative insight, in the present work, we performed a meta-analysis of brain activation patterns elicited by three types of novel insight task that have been frequently used in previous neuroimaging studies. To achieve this goal, the activation likelihood estimation (ALE) method was employed to conduct a quantitative meta-analysis of fMRI experiments of creative insight. The application of a meta-analysis

10

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT has significant advantages over empirical investigations in differentiating the potential effect of task type among these three insight tasks because neuroimaging (e.g., fMRI or PET) studies are intrinsically

IP

T

underpowered. In general, such studies have only approximately 20 participants, and they largely rely

SC R

on indirect measures of neuronal activity that are susceptible to several technical and biological confounders. As a result, ―divergence in results can arise from subtle differences in design,‖ which in turn suggests that ―it is difficult to evaluate what is specific to a particular report and what would

NU

generalize across a number of contexts, tasks, stimuli, and subjects‖ (Grosbras, Beaton, & Eickhoff,

MA

2012). Use of the ALE method, which has been widely validated and is shown to be an effective and important method for synthesizing neuroimaging results, enabled us to overcome these shortcomings,

D

integrate neuroimaging results across studies, and obtain reliable brain activation data associated with

TE

different types of insight tasks. For these reasons, we adopted the ALE meta-analysis method to assess

2 Methods

CE P

potential task modulation effects on creative insight.

AC

2.1 Study selection: inclusion and exclusion criteria The PRISMA guidelines (http://www.prisma-statement.org/) were followed while conducting this meta-analysis (Moran, Schroder, Keip, & Moser, 2016; Moser, Moran, Kneip, Schroder, & Jarson, 2016). Studies were selected by searching the PNAS, PLoS, MIT Press, Oxford Press, Elsevier Science, Wiley-Blackwell, Sage, and Springer databases using the keywords creative insight, insight problem, prototype heuristic, remote associate, and chunk decomposition. Furthermore, the selected articles were required to include one of the following keywords: functional resonance imaging/fMRI, neuroimaging, or brain imaging. Additional articles were identified by further searching the reference sections of all obtained articles. The following inclusion criteria were utilized to screen studies (see also Figure 1):

11

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT (i) Studies reporting on the association between brain activation and one of the three insight tasks focused on in the current work.

IP

T

(ii) Peer-reviewed studies (written in English or Chinese) that used fMRI to evaluate healthy

SC R

participants and reported coordinates using standard Montreal Neurological Institute or Talairach space. (iii) Studies presenting clear comparisons of brain activation for an insight condition and that for a non-insight or other baseline condition.

NU

(iiii) Studies focused on the process of insight rather than on insight preparation or the

MA

post-solution stage of verification.

[INSERT FIGURE 1 HERE]

D

It can be argued that focusing on the moment of insight might increase the likelihood of

TE

identifying task-specific activity because the final solution might require the integration or

CE P

activation of very task-specific information. However, if insight is task-independent or shows a general, across-task pattern, any task-invariant brain activation patterns related to the insight

AC

process should also be present at the moment of insight. Additionally, creative insight is a multistage and dynamic process. Focusing on more than one stage and on different types of creative insight would make it more difficult to isolate the specific neural correlates of each stage of this insight and to explain the cognitive functions of the different activated brain regions. Furthermore, previous findings have demonstrated that four is the reasonable minimum number of studies to include in an ALE meta-analysis (e.g., Mincic, 2015; Simmonds, Pekar, & Mostofsky, 2008). The number of available studies that examined other stages of insight was not sufficient to conduct a reliable ALE meta-analysis; a larger number of studies focused on insight-based solutions due to special interest in the concept of a ―flash of insight‖. A total of 14 published fMRI

12

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT studies (including 15 experiments) on creative insight met the inclusion criteria and were included in the meta-analysis (see Figure 1). As listed in Table 1, these studies yielded 135, 30 and 29 foci for the

IP

T

meta-analyses of the CCD, CRA, and PH tasks, respectively; the contrasts yielding these foci were

SC R

plotted in Talairach space using BrainMap and Sleuth V2.4 (http://www.brainmap.org/sleuth/).

2.2 ALE technique and activation display

ALE is a common statistical modeling technique designed to address invariance among

NU

neuroimaging (e.g., fMRI) studies and can therefore be used to merge and integrate neuroimaging

MA

findings across studies (Turkeltaub, Eden, Jones, & Zeffiro, 2002; Laird et al., 2005). In contrast to arbitrary methods, such as narrative and table-based literature reviews, foci-based ALE is a

D

novel, effective, and quantitative meta-analysis technique that can be conducted independently of

TE

the traditional tabular method of establishing agreement across studies and any self-supplied

CE P

anatomical labels, which are generally obtained by treating markedly activated foci as the peaks of a 3D Gaussian probability distribution rather than as single points. Technically, the ALE technique

AC

pools peak activation coordinates across studies to identify a potential convergent effect of interest by comparing activation likelihoods calculated for the observed activation foci with a null random distribution (null distribution of random spatial associations between foci) and mapping all of the coordinates onto a single template, such as Montreal Neurological Institute or Talairach space (Laird et al., 2005; Fox, Spreng, Ellamil, Andrews-Hanna, & Christoff, 2015). Currently, most meta-analyses of neuroimaging studies utilize the software program Ginger ALE 2.3.5 debug version (San Antonio, TX: UT Health Science Center Research Imaging Institute), which has an embedded ALE algorithm (Laird et al., 2005; Turkeltaub et al., 2002, 2012) that clearly reveals the spatial uncertainties of activation foci identified in various neuroimaging results and enables 3D

13

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Gaussian distributions to be summed to create a voxel-based map (foci whose activation is greater than expected by chance) with improved goodness of fit (Laird et al., 2005).

IP

T

[INSERT FIGURE 2 HERE]

SC R

To achieve the goal of determining whether creative insight is domain-specific or domain-general (i.e., task-independent or task-dependent), brain activation regions drawn from neuroimaging studies using the CRA task, the PH task and the CCD task were grouped into three

NU

meta-analyses. According to prior studies (e.g., Simmonds et al., 2008), the ideal way to examine

MA

the task-dependence of insight is to conduct individual meta-analyses focused on the three types of insight tasks rather than a global meta-analysis of all three types of insight task. Due to

D

inconsistencies in the coordinate space of the reported brain activation patterns across the

TE

investigations included in the present meta-analysis, all results reported in Montreal Neurological

CE P

Institute coordinates were converted into Talairach space using non-linear transformations via the

Brett transformation implemented in the Ginger ALE 2.3.5 software package. To ensure that the

AC

current meta-analysis reflected cross-study convergence, a relatively liberal statistical threshold (uncorrected p=0.01; Turkeltaub et al., 2002)③ and a minimum volume threshold (a cluster threshold of k=1000 mm3; Fox et al., 2015) were used to optimize result visualization. The calculated clusters were overlaid on a standard brain in Talairach space provided by brainmap.org ③

This liberal statistical threshold was chosen for two reasons: First, the number of available studies in each

category is very limited and far below the requirement for a more conservative false discovery rate (FDR) threshold (the minimum number of studies is more than 10 when using the FDR method). Second, we used the same method for generating uncorrected p results that has been used in previous studies (e.g., Turkeltaub et al., 2002), and an algorithm is available that can transform these results into FDR-corrected results (see Laird et al., 2005). Importantly, if the present meta-analytical results were generated using 3dClustSim with a voxelwise threshold of 0.01, it would result in a minimum cluster volume >882 mm3 based on the nearest-neighbor chain algorithm, for a corrected p value <0.05 at the cluster level (see Bonino, Ricciardi, Bernardi, Sani, Gentili, Vecchi, & Pietrini, 2015). This suggests that the statistical threshold in the present work is stricter than using a corrected p significance level of 0.05 at the cluster level. 14

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT (Colin_tlrc_2x2x2.nii) and visualized using Multi-Image Analysis GUI image-viewing software (http://ric.uthscsa.edu/mango/mango.html).

IP

T

[INSERT TABLE 1 HERE]

SC R

3 Results

Figure 2 shows that the brain activation foci reported by previous fMRI studies of the three types of creative insight task examined here exhibited different distributions across the cerebral

NU

hemispheres. To explore the task independence of creative insight, fifteen experiments (listed in

MA

Table 1) from fourteen studies were included in our three meta-analyses. The ALE-based meta-analysis of the five studies that used the Chinese CCD task, which

D

compared tight CD④ with an appropriate solution with loose CD without an appropriate result,

TE

uncovered greater activity in a network of bilateral prefrontal regions, namely, the bilateral

CE P

precentral gyri (PreG), the right extra-nuclear gyrus, and the right cingulate gyrus. In contrast, stronger brain activation in the right parahippocampal gyrus (PHG), the right superior frontal

AC

gyrus (SFG), and the right inferior frontal gyrus (IFG) was found in the CRA insight solution compared to the CRA non-insight solution. Additionally, our meta-analysis of the PH studies revealed increased activity in the right middle occipital gyrus (MOG), the bilateral superior parietal lobule/precuneus (SPL/PreC), the left lingual gyrus, the left middle frontal gyrus (MFG), and the left inferior parietal lobule (IPL) in the insight condition compared to the non-insight condition (Figure 3). Because several of the included studies of heuristic insight (26 foci from three experiments in

④

In the Chinese CCD task, novel results were obtained by decomposing unfamiliar tight (stroke-level) chunks of

Chinese characters or Roman numerals. The appropriateness of chunk decomposition mainly relies on whether the decomposition generates real Chinese characters. 15

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT two studies; Qiu et al., 2010; Luo et al., 2013) and CD insight (34 foci from two experiments reported by Luo et al. (2006) and Tang et al. (2015)) reported activation foci in reverse

IP

T

comparisons, two additional meta-analyses were conducted for each of these tasks, with one being

SC R

applied per comparison. The results showed enhanced activity in the bilateral medial frontal gyrus (BA 9/10), the left superior temporal gyrus (BA 22; STG) and the right postcentral gyrus (BA 40) in the heuristic non-insight condition, whereas suprathreshold ALE activation foci were observed

NU

in the left cuneus (BA 18) and clustered in the left precentral gyrus (BA 6/43), the insula (BA 13),

MA

the IPL (BA 40), and the STG (BA 22) in the reverse comparison for the CCD task.

[INSERT TABLE 2 HERE]

D

4 Discussion

TE

The present work sought to identify common brain activation patterns across neuroimaging

CE P

studies to establish associations between creative insight and brain activity. Toward this end, three quantitative meta-analyses of neuroimaging investigations of three different insight tasks were

AC

conducted. For the CRA task, our results showed that the brain activation foci identified across the included fMRI studies mainly converged in the right PHG, the SFG, and the IFG. In contrast, heuristic insight induced by the PH task primarily activated the bilateral SPL/PreC, the right MOG, the left lingual gyrus, the left IPL, and the left MFG. However, numerous brain activation foci have been reported by previous fMRI studies investigating the neural underpinnings of the CCD process, resulting in the creation of a brain-based CD network that encompasses most bilateral prefrontal regions, the bilateral PreG, the right extra-nuclear gyrus, and the right cingulate gyrus. It can be seen from this network that, for CD insight, only the right IFG cluster partially overlapped with the brain activation foci for associative insight. Also, the CD insight shared some (i.e., the

16

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT left MFG) but not all of the brain activation foci triggered by the heuristic insight of the PH task (see Figure 4). Additionally, the creative insight induced by the CCD task included only some of

IP

T

the brain regions associated with heuristic insight. These highly task-specific results suggest that

SC R

creative insight may not be a homogeneous process across tasks and instead may reflect a heterogeneous set of task-dependent mental events, supporting the task dependence (and even domain specificity to some degree) of insight. Below, we discuss the roles of the main brain foci

NU

identified and the potential implications of these findings.

MA

[INSERT FIGURE 3 HERE] 4.1 Convergent brain activation in the CRA insight task

D

The PHG was obviously activated during insight-based problem solving in the CRA task;

TE

this brain structure has been increasingly reported to be activated during engagement in different

CE P

types of creativity tasks (e.g., Gao & Zhang, 2014; Ellamil, Dobson, Beeman, & Christoff, 2012; Jung-Beeman et al., 2004) and is thought to be responsible for constructing episodic simulations

AC

(Schacter & Addis, 2009) or establishing new semantic associations (Ellamil et al., 2012) by accessing old associations stored in the semantic and episodic memory systems. Relative to the other two types of insight task, the CRA task seems to result in greater formation of novel associations, although the breaking of impasse-related constraints (see Smith, Huber, & Vul, 2013) has also been reported. In this context, the observed PHG may reflect the formation of novel or weak semantic associations between distant and seemingly irrelevant information, consistent with previous findings that the parahippocampal region is more strongly recruited for the formation of contextual associations compared to non-contextual associations (Aminoff, Gronau, & Bar, 2007). Consistent with previous findings (e.g., Subramaniam, Kounios, Parrish, & Jung-Beeman,

17

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT 2008; Ellamil et al., 2012), the identification of the right SFG in our meta-analysis of the CRA task suggests that this region may, together with the right IFG (e.g., Goel & Vartanian, 2005, p.

IP

T

1176), be responsible for the formation of ―a plan for top-down activation of semantic fields

SC R

relevant to the problem at hand‖ (Binder & Desai, 2011), whose damage may cause ―no loss of stored knowledge per se [but rather] impairs the ability to access this knowledge for creative problem solving‖ (Binder & Desai, 2011, p. 532). Unexpectedly, activation of the right anterior

NU

STG (aSTG), which was reported to show the greatest difference in activation for insight and

MA

analytic solutions to CRA problems (Jung-Beeman et al., 2004), was not observed here. We believe that the lack of identification of the right aSTG may be partially due to its adjacency to the

D

PHG and partially skewed by the meta-analytic algorithms employed here. In ALE analyses, the

TE

only information taken into account is the anatomical position of the peak activation voxel of each

CE P

suprathreshold cluster, which is treated as the center of a 3D Gaussian distribution; thus, neither the specific magnitude nor the extent of activation of each of the clusters captured in the included

AC

fMRI studies was taken into account (Simmonds et al., 2008, p. 230). In this regard, the absence of the identification of the right aSTG may result from the meta-analytic algorithms employed to analyze the ALE results (i.e., the results may have been skewed). Conversely, the left aSTG, which was identified in the reverse comparisons of both CD and heuristic insight, concordat with Dietrich and Kanso‘s results (2010, p. 839), may play a key role in ―passive insight‖ processes as well as in language comprehension involving syntactic violations (Friederici, Rüschemeyer, Hahne, & Fiebach, 2003). Our results indicate that, unlike CD insight (mainly relying on frontal regions) and heuristic insight (requiring the collaboration of bi-hemispheric and anterior-posterior regions), associative insight, particularly the moment that the CRA task is solved via insight, relies

18

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT on the right brain, corroborating previous findings of the superiority of the right hemisphere in insight (Jung-Beeman et al., 2004; Shen et al., 2013). Additionally, the above patterns of

IP

T

hemispheric collaboration revealed by our meta-analyses of the three insight tasks support the

SC R

task-dependence of creative insight.

[INSERT FIGURE 4 HERE] 4.2 Convergent brain activation in the PH insight task

NU

Heuristic insight involves ―representation connection‖ and ―relationship mapping‖ stages,

MA

although the former is considered to be the core process (Ming et al., 2014). Once critical heuristic information embedded in a prototype is abruptly recognized, the solver immediately achieves

D

insight or finds the correct problem search space with the help of the heuristic prototype. As a

TE

brain region associated with visual processing, the right MOG may participate in creating visual

CE P

imagery. Indeed, this region has a well-established role in visual attention research (e.g., Macaluso, Frith, & Driver, 2000) and serves as the basis for the formation and application of functional

AC

associations between the heuristic prototype and a target problem. In accordance with this idea, previous studies have shown that this region could provide critical information for representational changes and the reorganization of visual imagery during insightful problem solving (Hao et al., 2013; Luo et al., 2006; Qiu et al., 2010). In contrast, the lingual gyrus is an anatomical structure located near the ventral visual pathway that is believed to play a vital role in vision and dreaming (Dresler et al., 2015; Luo et al., 2013). It may be activated in response to binding or the establishment of original (i.e., inexperienced) connections between the different representations of the heuristic prototype and a target problem by stimulating and particularly re-orienting (e.g., guiding the search of) problem space. As cited in Luo et al. (2013), ―activity in the lingual gyri

19

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT would be highlighted when novel stimuli are presented in the spatially unattended visual field.‖ Thus, the lingual gyrus becomes activated due to an initially established association (through the

IP

T

participants‘ own initiatives) between a prototype problem and a subsequently presented target

SC R

problem. In accordance with prior heuristic prototype studies (e.g., Tong et al., 2013), the activation of BA 39, which is part of the left IPL and also the location of the angular gyrus according to fMRI studies, may participate in conveying retrieved heuristic knowledge to aid in

NU

the formation of novel associations.

MA

As a hub of the creativity-relied default mode network (DMN; see Huang et al., 2015), the bilateral SPL/PreC might be involved in the automatic retrieval of useful information, such as

D

prototype-related heuristics or the so-called prototype activation referred to in the PHT. According

TE

to the PHT (Ming et al., 2014), the process of relationship mapping is actually spontaneous and

CE P

insensitive to mental resources, mirroring the marked activation that has been observed in the left MFG. In accordance with our findings, the left MFG has often been reported to play a similar role

AC

in the transformation and mapping of various rules, including the conversion of orthography to phonology and orthography-to-semantics mapping (Liu et al., 2006; Luo et al., 2006). The left MFG is activated automatically, which is consistent with the PHT, indicating that relationship mapping is automatically and immediately achieved after solvers establish a context-appropriate connection between the required-function representations of a target problem and its prototype (Ming, et al., 2014; Hao et al., 2013). This automaticity also fits with the role of the left MFG as an essential component of the task-negative DMN (Luo et al., 2013). 4.3 Convergent brain activation in the CCD insight task The function of MOG activation during the CCD task may be similar to its function in the

20

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT solving of the PH task, i.e., the generation of visual imagery, including mental simulation (e.g., decomposing, removing or re-chunking strokes or radicals) of a decomposed character for CD,

IP

T

which is in line with well-established findings that early perceptual processing is especially

SC R

critical for CD (Wu et al., 2009; Wu et al., 2013; Luo et al., 2006). Moreover, the widespread activation observed in the prefrontal regions of the cerebral hemispheres, including nearly all of the prefrontal subdivisions, such as the bilateral MFG and the left SFG, and almost all of the brain

NU

areas distributed in the prefrontal cortex, such as BA 6 and BA 47, converged into significant

MA

activation patterns during the CCD task for insight-based CD. Excluding a few brain regions (e.g., BA 47, including the right IFG, the extra-nuclear gyrus, and the MdFG) in the ventral prefrontal

D

cortex (VPFC), most of these frontal regions are considered part of the well-known dorsolateral

TE

prefrontal cortex (DLPFC), which consists of BA 8, BA 9, BA 11, and BA 46 (Ciesli, 2013) and is

CE P

widely considered the executive system responsible for working memory, attention, and other processes required to solve creative problems (Luo et al., 2006; Goel & Vartanian, 2005).

AC

The results of our analysis of the Chinese CCD task are supported by abundant evidence from both homogeneous task applications (e.g., Wu et al., 2013; Luo et al., 2006; Goel & Vartanian, 2005; Knoblich et al., 1999) and heterogeneous task applications (e.g., Aziz-Zadeh et al., 2009; Qiu et al., 2010; Chein & Weisberg, 2014), suggesting that the DLPFC is primarily involved in sustaining internally focused but wandering attention states (left SFG; Manna et al., 2010). The DLPFC is organized through loosely related elements into novel chunks (i.e., the left MFG), candidate regions for retrieving and selecting competitive associations (e.g., the left MFG/IFG), and strongly inhibitory but inappropriate associations (i.e., the right MFG). In contrast, the VPFC is

primarily involved in maintaining the motivation to solve problems by strengthening progress

21

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT monitoring or emotional regulation (i.e., the right MdFG) in addition to forming novel chunks (i.e., BA 47 and the right IFG together with the left IFG; see Goel & Vartanian, 2005 for details). In

IP

T

addition to the involvement of the bilateral precentral gyrus in movement simulation, the role of

SC R

the right extra-nuclear gyrus (BA 47), which is part of the VPFC, might covary with that of the right IFG; this region also likely participates in maintaining motivational drive, whereas the right cingulate gyrus may be activated in response to conflict between old (familiar but inappropriate

NU

chunk) and new (novel and appropriate chunk) cognitive models. The activation of this cingulated

MA

region has been particularly highlighted as a source of difficulty (e.g., Wu et al., 2013) involved in steering individuals away from a novel method of problem solving (Huang et al., 2015).

D

4.4 Conclusion and limitations

TE

Taken together, our results demonstrated that creative insight induced by different tasks is

CE P

underpinned by the activation of distinct brain regions. The insights individually triggered by the CRA, CCD, and PH tasks shared few overlapping brain areas, suggesting that the neural correlates

AC

of insight are highly task-specific or task-dependent and providing important taxonomic support for insight (e.g., associative insight vs. heuristic insight). However, given the complexity of phase-sensitive insight (Sandkühler & Bhattacharya, 2008; Zhao et al., 2013) and the poor temporal resolution of fMRI in isolating the precise moment of insight solution, it is very difficult to exclude some confounding effects from temporal nuances when assessing the insight stages of the three types of insight task⑤. Moreover, previous findings indicate that insight likely includes ⑤

Subramaniam et al. (2009) examined brain activations associated with mental preparation preceding associative

insight, but they also reported on brain activation patterns at the insight solution stage that the present work assessed. It seems impractical to limit all comparisons to a single time frame because the tasks themselves are very different and there are limited target studies using fMRI, many of which have poor temporal resolution. As such, there are not enough available studies to generate reliable meta-analytical results or even to perform a convergent meta-analysis. Indeed, most journals require that studies be novel or present new findings in order to achieve 22

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT somewhat task-general components. First, substantial evidence indicates that creative insight mainly consists of or relies on common or routine process of cognition, such as working memory

IP

T

or attention, which are part of other high-level cognitive or psychological processes, suggesting

SC R

that insight does not involve specialized components or distinct cognitive processes (see Plucker & Beghetto, 2004). All types of insight may require such task-general cognitive processes or cross-domain fundamental abilities (devoted to creativity and insight problem solving), such as

NU

reasoning or mental representation (Plucker & Beghetto, 2004). Second, researchers have not yet

MA

identified a unique region of brain activation that is specific to insight alone or that is only activated by a certain insight task. In this work, two brain activation regions (i.e., the MOG and

D

the MFG) were shared between heuristic and CD insight. Finally, the three types of insight task

TE

examined here may share some processes (e.g., attention) with their corresponding baselines.

CE P

Accordingly, creative insight is a task-dependent process with certain cross-task components. This study had several limitations that should be considered. A common flaw of

AC

meta-analytical studies, particularly meta-analyses of neuroimaging studies, is the failure to include consistent baselines (for further discussion on this topic, see Luo & Knoblich, 2007). However, this is not a fatal shortcoming because all the selected studies were designed strictly based on the principle of Donders‘ cognitive subtraction technique. The included brain activation foci represent the supra/subthreshold amplitudes that the baselines were subtracted from based on well-matched experimental conditions rather than the amplitude of brain activation based on only one experimental condition (see Luo & Knoblich, 2007; Weisberg, 2013). The relatively small publication, which means that identifying studies that have examined the same topic while focused on the same time points may not always be possible. One hypothesis that was assumed in our meta-analysis is that all of the original included studies tried their best to exclude the confounding factors that may influence conclusions drawn from the cognitive subtraction of brain activation patterns elicited by experimental trials from the baseline patterns. 23

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT number of studies included in the current meta-analyses should be noted, although the sample size reached the minimum sample size criterion needed to generate reliable meta-analytical results.

IP

T

Although the neural basis of creative insight has recently attracted considerable attention, only a

SC R

limited number of studies met our inclusion criteria due to the complexity and diversity of the experimental tasks that have been used to study insight. The studies included in our meta-analyses were thus limited. Another disadvantage of the present work is that it cannot offer more empirical

NU

information regarding the task independence of insight or details regarding the extent to which

MA

creative insight is task-specific. Many additional neuroimaging studies will be needed to sufficiently address this topic. Nevertheless, the current study provides extremely valuable

D

information. To the best of our knowledge, this work provides the first neural evidence that

TE

improves our understanding of the task generality and task specificity of insight. Furthermore, this

CE P

study broadens our knowledge of the nature of insight problem solving and helps conceptualize (define) creative insight, facilitating taxonomic study of insight (see Dow & Mayer, 2004).

AC

Moreover, the results from the present study also contain significant implications for methods of facilitating insight and for the design of ideal insight tasks (Weisberg, 2013) or corresponding baseline counterparts (Luo & Knoblich, 2007). In summary, the present work performed three separate meta-analyses to explore the task dependence of insight and to provide a quantitative overview of the brain regions that are consistently activated during engagement in different types of creative insight. The results from our meta-analyses revealed many cerebral regions that are recruited during the three types of insight processes studied here, providing support for the task dependence of creative insight, at least at the neural level. There are of course domain- or process-specific components when solving

24

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT any problem; however, it seems there are also some general features/processes associated with insight (e.g., the ability to switch between ideas or to pull weakly formed ideas into

IP

T

consciousness), and the activation patterns of the brain regions corresponding to such are likely

SC R

offset by cognitive subtraction. Taken together, these findings suggest that creative insight occurs through a hybrid process that requires both task-dependent and task-independent sub-processes and that experimental tasks (see Weisberg, 2015 for details) can modulate the neural

MA

NU

underpinnings of insight.

References

* indicated the paper was included in one of the meta-analyses.

D

Abraham, A., Beudt, S., Ott, D. V., & von Cramon, D. Y. (2012). Creative cognition and the brain: Dissociations

TE

between frontal, parietal–temporal and basal ganglia groups. Brain Research, 1482, 55-70. Anderson, J. R., Anderson, J. F., Ferris, J. L., Fincham, J. M., & Jung, K. J. (2009). Lateral inferior prefrontal cortex and anterior cingulate cortex are engaged at different stages in the solution of insight

CE P

problems. Proceedings of the National Academy of Sciences, 106(26), 10799-10804. * Aminoff, E., Gronau, N., & Bar, M. (2007). The parahippocampal cortex mediates spatial and nonspatial associations. Cerebral Cortex, 17(7), 1493-1503. Aziz-Zadeh, L., Kaplan, J. T., & Iacoboni, M. (2009). ―Aha!‖: The neural correlates of verbal insight

AC

solutions. Human Brain Mapping, 30(3), 908-916. Binder, J. R., & Desai, R. H. (2011). The neurobiology of semantic memory. Trends in cognitive sciences, 15(11), 527-536.

Bonino, D., Ricciardi, E., Bernardi, G., Sani, L., Gentili, C., Vecchi, T., & Pietrini, P. (2015). Spatial imagery relies on a sensory independent, though sensory sensitive, functional organization within the parietal cortex: A fMRI study of angle discrimination in sighted and congenitally blind individuals. Neuropsychologia, 68, 59-70. Bowden, E. M., & Jung-Beeman, M. (1998). Getting the right idea: Semantic activation in the right hemisphere may help solve insight problems. Psychological Science, 9(6), 435-440. Bowden, E. M., & Jung-Beeman, M. (2003). Normative data for 144 compound remote associate problems. Behavior Research Methods, Instruments, & Computers, 35(4), 634-639. Cerruti, C., & Schlaug, G. (2009). Anodal transcranial direct current stimulation of the prefrontal cortex enhances complex verbal associative thought. Journal of Cognitive Neuroscience, 21(10), 1980-1987. Chen, C., Himsel, A., Kasof, J., Greenberger, E., & Dmitrieva, J. (2006). Correlates of domain-general and domain-specific components of creativity. Journal of Creative Behavior, 40(3), 179-199. Chein, J. M., & Weisberg, R. W. (2014). Working memory and insight in verbal problems: Analysis of compound remote associates. Memory & Cognition, 42(1), 67-83. Chermahini, S. A., Hickendorff, M., & Hommel, B. (2012). Development and validity of a Dutch version of the

25

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Remote Associates Task: An item-response theory approach. Thinking Skills and Creativity, 7(3), 177-186. Cieslik, E. (2013). Is there "one" DLPFC in cognitive action control? Evidence for heterogeneity from co-activation-based parcellation. Cerebral Cortex, 23 (11), 2677–2689. Cranford, E. A., & Moss, J. (2011). An fMRI study of insight using compound remote associate problems. In

T

Proceedings of the 33rd annual conference of the cognitive science society (pp. 3558–3563). Austin, TX: Cognitive Science Society.

IP

Cunningham, J. B., MacGregor, J. N., Gibb, J., & Haar, J. (2009). Categories of insight and their correlates: An

analogies. The Journal of Creative Behavior, 43(4), 262-280.

SC R

exploration of relationships among classic‐type insight problems, rebus puzzles, remote associates and esoteric

Danek, A. H., Fraps, T., von Müller, A., Grothe, B., & Öllinger, M. (2013). Aha! experiences leave a mark: facilitated recall of insight solutions. Psychological Research, 77(5), 659-669.

Darsaud, A., Wagner, U., Balteau, E., Desseilles, M., Sterpenich, V., Vandewalle, G., ... & Maquet, P. (2011).

NU

Neural precursors of delayed insight. Journal of Cognitive Neuroscience, 23(8), 1900-1910. Dietrich, A., & Kanso, R. (2010). A review of EEG, ERP, and neuroimaging studies of creativity and insight. Psychological Bulletin, 136(5), 822-848.

MA

Ding, X., Tang, Y. Y., Cao, C., Deng, Y., Wang, Y., Xin, X., & Posner, M. I. (2015). Short-term meditation modulates brain activity of insight evoked with solution cue. Social Cognitive and Affective Neuroscience, 10(1), 43-49.

Dresler, M., Wehrle, R., Spoormaker, V. I., Steiger, A., Holsboer, F., Czisch, M., & Hobson, J. A. (2015). Neural

D

correlates of insight in dreaming and psychosis. Sleep Medicine Reviews, 20, 92-99.

TE

Dow, G. T., & Mayer, R. E. (2004). Teaching students to solve insight problems: Evidence for domain specificity in creativity training. Creativity Research Journal, 16(4), 389-398. Ellamil, M., Dobson, C., Beeman, M., & Christoff, K. (2012). Evaluative and generative modes of thought during

CE P

the creative process. Neuroimage, 59(2), 1783-1794. Epstein, R., Kirshnit, C. E., Lanza, R. P., & Rubin, L. C. (1984). ‗Insight‘ in the pigeon: antecedents and determinants of an intelligent performance. Nature 308, 61-62. Fink, A., Grabner, R. H., Gebauer, D., Reishofer, G., Koschutnig, K., & Ebner, F. (2010). Enhancing creativity by

AC

means of cognitive stimulation: Evidence from an fMRI study. Neuroimage, 52(4), 1687-1695. Fox, K. C., Spreng, R. N., Ellamil, M., Andrews-Hanna, J. R., & Christoff, K. (2015). The wandering brain: Meta-analysis of functional neuroimaging studies of mind-wandering and related spontaneous thought processes. NeuroImage, 111, 611-621. Friederici, A. D., Rüschemeyer, S. A., Hahne, A., & Fiebach, C. J. (2003). The role of left inferior frontal and superior temporal cortex in sentence comprehension: localizing syntactic and semantic processes. Cerebral Cortex, 13(2), 170-177. Gao, Y., & Zhang, H. (2014). Unconscious processing modulates creative problem solving: Evidence from an electrophysiological study. Consciousness and cognition, 26, 64-73. Goel, V., & Vartanian, O. (2005). Dissociating the roles of right ventral lateral and dorsal lateral prefrontal cortex in generation and maintenance of hypotheses in set-shift problems. Cerebral Cortex, 15(8), 1170-1177. Grosbras, M. H., Beaton, S., & Eickhoff, S. B. (2012). Brain regions involved in human movement perception: A quantitative voxel‐based meta‐analysis. Human Brain Mapping, 33(2), 431-454. Hao, X., Cui, S., Li, W., Yang, W., Qiu, J., & Zhang, Q. (2013). Enhancing insight in scientific problem solving by highlighting the functional features of prototypes: An fMRI study. Brain Research, 1534, 46-54. * Huang, F., Fan, J., & Luo, J. (2015). The neural basis of novelty and appropriateness in processing of creative chunk decomposition. NeuroImage,113, 122-132. *

26

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Jung-Beeman, M., Bowden, E. M., Haberman, J., Frymiare, J. L., Arambel-Liu, S., Greenblatt, R., ... & Kounios, J. (2004). Neural activity when people solve verbal problems with insight. PLoS Biology, 2(4), 500-510. * Kaufman, J. C., & Baer, J. (Eds.). (2005). Creativity across domains: Faces of the muse. Psychology Press. Knoblich, G., Ohlsson, S., Haider, H., & Rhenius, D. (1999). Constraint relaxation and chunk decomposition in

T

insight problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(6), 1534-1555.

IP

Kounios, J., Frymiare, J. L., Bowden, E. M., Fleck, J. I., Subramaniam, K., Parrish, T. B., & Jung-Beeman, M. (2006). The prepared mind neural activity prior to problem presentation predicts subsequent solution by sudden

SC R

insight. Psychological Science, 17(10), 882-890.

Laird, A. R., Fox, P. M., Price, C. J., Glahn, D. C., Uecker, A. M., Lancaster, J. L., ... & Fox, P. T. (2005). ALE meta-analysis: Controlling the false discovery rate and performing statistical contrasts. Human Brain Mapping, 25(1), 155-164.

NU

Liu, C. L., Hue, C. W., Chen, C. C., Chuang, K. H., Liang, K. C., Wang, Y. H., ... & Chen, J. H. (2006). Dissociated roles of the middle frontal gyri in the processing of Chinese characters. NeuroReport, 17(13), 1397-1401.

MA

Ludmer, R., Dudai, Y., & Rubin, N. (2011). Uncovering camouflage: Amygdala activation predicts long-term memory of induced perceptual insight. Neuron, 69(5), 1002-1014. Luo, J., & Knoblich, G. (2007). Studying insight problem solving with neuroscientific methods. Methods, 42(1), 77-86.

D

Luo, J., Niki, K., & Knoblich, G. (2006). Perceptual contributions to problem solving: Chunk decomposition of

TE

Chinese characters. Brain Research Bulletin, 70(4), 430-443. * Luo, J., Li, W., Fink, A., Jia, L., Xiao, X., Qiu, J., & Zhang, Q. (2011). The time course of breaking mental sets and forming novel associations in insight-like problem solving: An ERP investigation. Experimental Brain

CE P

Research, 212(4), 583-591.

Luo, J., Li, W., Qiu, J., Wei, D., Liu, Y., Zhang, Q., & Kilner, J. (2013). Neural basis of scientific innovation induced by heuristic prototype. PloS One, 8(1), e49231. * Macaluso, E., Frith, C. D., & Driver, J. (2000). Modulation of human visual cortex by crossmodal spatial

AC

attention. Science, 289(5482), 1206-1208. Mai, X. Q., Luo, J., Wu, J. H., & Luo, Y. J. (2004). ―Aha!‖ effects in a guessing riddle task: An event‐related potential study. Human Brain Mapping, 22(4), 261-270. Manna, A., Raffone, A., Perrucci, M. G., Nardo, D., Ferretti, A., Tartaro, A., ... & Romani, G. L. (2010). Neural correlates of focused attention and cognitive monitoring in meditation. Brain Research Bulletin, 82(1), 46-56. Mednick, S. (1962). The associative basis of the creative process. Psychological Review, 69(3), 220-232. Mincic, A. M. (2015). Neuroanatomical correlates of negative emotionality-related traits: A systematic review and meta-analysis. Neuropsychologia, 77, 97-118. Ming, D., Tong, D., Yang, W., Qiu, J., & Zhang, Q. (2014). How can we gain insight in scientific innovation? Prototype heuristic is one key. Thinking Skills and Creativity, 14, 98-106. Moser, J. S., Moran, T. P., Kneip, C., Schroder, H. S., & Larson, M. J. (2016). Sex moderates the association between symptoms of anxiety, but not obsessive compulsive disorder, and error-monitoring brain activity: A meta-analytic review. Psychophysiology, 53, 21-29. Moran, T. P., Schroder, H. S., Kneip, C., & Moser, J. S. (2016). Meta-analysis and psychophysiology: A tutorial using depression and action-monitoring event-related potentials. International Journal of Psychophysiology. doi: http://dx.doi.org/10.1016/j.ijpsycho.2016.07.001 Ohlsson S. (1992). Information processing explanations of insight and related phenomenon (pp.1 -44). In M.

27

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Keane, K. & Gilhooly (Eds.), Advances in the psychology of thinking. London: Harvester-Wheatsheaf. Pang, J., Tang, X., Niki, K., & Luo, J. (2009). Brain activities related to the Chinese character chunking tasks: An fMRI study. In Natural Computation, 2009. ICNC'09. Fifth International Conference on (Vol. 5, pp. 32-37). IEEE. * Beghetto, R. A. (2004). Why creativity is domain general, why it looks domain specific, and

T

Plucker, J. A. &

why the distinction does not matter (pp. 153-167). In: R J. Sternberg & E. L., Grigorenko, & J. L., Singer. (Eds).

IP

Creativity: From potential to realization. Washington, DC, US: American Psychological Association.

Mathematics Education , 41(1-2), 5-11.

SC R

Plucker, J., & Zabelina, D. (2009). Creativity and interdisciplinarity: one creativity or many creativities?. ZDM Qiu, J., Li, H., Jou, J., Liu, J., Luo, Y., Feng, T., ... & Zhang, Q. (2010). Neural correlates of the ―Aha‖ experiences: evidence from an fMRI study of insight problem solving. Cortex, 46(3), 397-403. * Radel, R., Davranche, K., Fournier, M., & Dietrich, A. (2015). The role of (dis) inhibition in creativity: Decreased

NU

inhibition improves idea generation. Cognition, 134, 110-120.

Razumnikova, O. M. (2007). Creativity related cortex activity in the remote associates task. Brain Research Bulletin, 73(1), 96-102.

MA

Reiter-Palmon, R., Illies, M. Y., Cross, L. K., Buboltz, C., & Nimps, T. (2009). Creativity and domain specificity: The effect of task type on multiple indexes of creative problem-solving. Psychology of Aesthetics, Creativity, and the Arts, 3(2), 73-80.

solving. PLoS One, 3(1), e1459.

D

Sandkühler, S., & Bhattacharya, J. (2008). Deconstructing insight: EEG correlates of insightful problem

TE

Schacter, D. L., & Addis, D. R. (2009). On the nature of medial temporal lobe contributions to the constructive simulation of future events. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1521), 1245-1253.

CE P

Shen, W. (2014). Associative insight: A joint study on its body-mind-brain mechanism. Doctoral dissertation, Nanjing Normal University. *

Shen, W., Liu, C., Zhang, X., Zhao, X., Zhang, J., Yuan, Y., & Chen, Y. (2013). Right hemispheric dominance of creative insight: An Event-related potential study. Creativity Research Journal, 25(1), 48-58.

AC

Shen, W. Yuan, Y., Liu, C., Yi, B., & Dou, K. (2016). The development and validity of a Chinese version of the Compound Remote Associate Test. American Journal of Psychology, 129(3), 245-258. Simmonds, D. J., Pekar, J. J., & Mostofsky, S. H. (2008). Meta-analysis of Go/No-go tasks demonstrating that fMRI activation associated with response inhibition is task-dependent. Neuropsychologia, 46(1), 224-232. Smith, K. A., Huber, D. E., & Vul, E. (2013). Multiply-constrained semantic search in the Remote Associates Test. Cognition, 128(1), 64-75. Subramaniam, K., Kounios, J., Parrish, T. B., & Jung-Beeman, M. (2009). A brain mechanism for facilitation of insight by positive affect. Journal of Cognitive Neuroscience, 21(3), 415-432. * Tang, X., Pang, J., Nie, Q. Y., Conci, M., Luo, J., & Luo, J. (2015). Probing the cognitive mechanism of mental representational change during chunk decomposition: A parametric fMRI study. Cerebral Cortex, doi: 10.1093/cercor/bhv113. * Terai, H., Miwa, K., & Asami, K. fMRI Study in Insight Problem Solving Using Japanese Remote Associates Test Based on Semantic Chunk Decomposition (pp. 3516-3521). In Proceedings of the 33rd Annual Conference of the Cognitive Science Society. Tong, D., Li, W., Tang, C., Yang, W., Tian, Y., Zhang, L., ... & Zhang, Q. (2015). An illustrated heuristic prototype facilitates scientific inventive problem solving: A functional magnetic resonance imaging study. Consciousness and Cognition, 34, 43-51. *

28

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Tong, D., Zhu H. X., Li W. F., Yang W. J., Qiu J., & Zhang Q. (2013). Brain activity in using heuristic prototype to solve insightful problems. Behavioural Brain Research, 253, 139-144. * Turkeltaub, P. E., Eickhoff, S. B., Laird, A. R., Fox, M., Wiener, M., & Fox, P. (2012). Minimizing within experiment and within - group effects in activation likelihood estimation meta-analyses. Human Brain Mapping,

T

33(1), 1-13. Turkeltaub, P. E., Eden, G. F., Jones, K. M., & Zeffiro, T. A. (2002). Meta-analysis of the functional neuroanatomy

IP

of single-word reading: method and validation. NeuroImage, 16(3), 765-780.

Weisberg, R. W. (1995). Prolegomena to theories of insight in problem solving: A taxonomy of problems (pp.

SC R

157-196). In R. J. Sternberg. & J. E. Davidson. (Ed), (1995). The nature of insight. Cambridge, MA, US: The MIT Press.

Weisberg, R. W. (2013). On the ―demystification‖ of insight: A critique of neuroimaging studies of insight. Creativity Research Journal, 25(1), 1-14.

NU

Weisberg, R. W. (2015). Toward an integrated theory of insight in problem solving. Thinking & Reasoning, 21(1), 5-39.

Wu, L., Knoblich, G., Wei, G., & Luo, J. (2009). How perceptual processes help to generate new meaning: An EEG

MA

study of chunk decomposition in Chinese characters. Brain Research, 1296, 104-112. Wu, L., Knoblich, G., & Luo, J. (2013). The role of chunk tightness and chunk familiarity in problem solving: Evidence from ERPs and FMRI. Human Brain Mapping, 34(5), 1173-1186. * Zhao, Q., Zhou, Z., Xu, H., Chen, S., Xu, F., Fan, W., & Han, L. (2013). Dynamic neural network of insight: a

AC

CE P

TE

D

functional magnetic resonance imaging study on solving Chinese ‗chengyu‘ riddles. PLoS One, 8(3), e59351.

29

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT

57 potentially relevant studies 39 studies were eliminated:

IP

T

7 review studies on neural correlates of insight elicited by the task of CRA or CCD; 19 studies used other brain stimulation techniques like transcranial direct current stimulation (tDCS) or behavioral measures; 13 neuroimaging studies used other insight tasks;

SC R

18 studies met criteria

4 studies were excluded:

NU

1 study did not provide specific standard coordinates (Cranford & Moss, 2013); 1 study on the effect of meditation on the CRA solving (Ding et al., 2014); 1 study involved two different ways achieving insight (Terai, Miwa, & Asami, 2013); 1 study only focused on mental preparation of insight (Kounios et al., 2006);

15 experiments drawn form 14 studies; 1 study reported 2 experiments:

MA

5 studies used the CCD task with a subtotal sample of 78 4 studies used the CRA task with a subtotal sample of 73 5 studies (6 experiments) used the PH task with a subtotal sample of 117

AC

CE P

TE

D

Figure 1 flowchart depicting the selection of studies in this meta-analysis

30

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Table 1 Details of studies included in the quantitative meta-analysis

insight task

contrasts

scanner

n

foci

Tang et al. (2015)

CCD task

tight chunk vs. loose chunk vs. baseline

3T

22

23

Huang et al., (2015)

CCD task

novel-appropriate vs. familiar-inappropriate

3T

15

13

Wu et al. (2013)

CCD task

(un)familiar-tight vs. (un)familiar-loose

3T

14

58

Pang et al. (2009)

CCD task

tight chunk vs. loose chunk

3T

13

3

Luo et al. (2006)

CCD task

tight chunk vs. loose chunk

3T

14

38

Shen (2014)

CRA task

insight solution vs. noninsight solution

3T

13

10

Anderson et al. (2009), Exp.1

CRA task

insight solution vs. baseline solution

3T

20

6

Subramaniam et al. (2008)

CRA task

insight solution vs. noninsight solution

3T

27

7

Jung-Beeman et al. (2004), Exp. 1

CRA task

insight solution vs. noninsight solution

1.5 T

13

7

Tong et al. (2015)

PH task

illustrated depictions vs. non-illustrated depictions

3T

32

3

Tong et al. (2013)

PH task

successfully solved trials vs. unsolved trials

3T

16

2

Hao et al. (2013)

PH task

highlighted heuristics vs. no-highlighted heuristics

3T

17

2

Luo J L et al. (2013), Exp.1

PH task

novel problems vs. old problems

3T

19

1

Luo J L et al. (2013), Exp.2

PH task

novel problems vs. old problems

3T

17

2

Qiu et al. (2010)

PH task

aha trails vs. non-aha trails

3T

16

19

AC

CE P

TE

D

MA

NU

SC R

IP

T

study

31

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Table 2 List of brain structures activated in the ALE meta-analysis (insight > noninsight) BA

Peak ALE value (×10-3)

x

y

z

1520

34

8.97

16

-11

-14

Right superior frontal gyrus

1400

9

13.78

7

51

27

Right inferior frontal gyrus

1376

46

7.45

47

33

6

Left middle frontal gyrus

2272

9

9.86

-47

10

34

Left superior parietal lobule/precuneus

2240

SC R

PH

7/19

8.07

-33

-64

37

39

7.81

-38

-64

38

7/19

7.65

31

-66

42

19

15.81

48

-57

-8

18

11.31

0

-74

-2

9/46

14.59

-50

20

30

6

13.24

-48

2

26

8

12.32

4

24

41

8

9.74

-4

18

50

32

7.74

2

20

40

2408

46

23.01

45

34

18

2352

47

15.50

33

26

-8

47/11

9.74

28

24

-4

9

12.69

46

9

27

6

7.74

36

2

36

Left inferior parietal lobule superior

parietal

1808

lobule/precuneus

1528

Right middle occipital gyrus

1376

Left lingual gyrus 5904

MA

Left middle/inferior frontal gyrus Left precentral gyrus

2840

Right medial frontal gyrus

Right cingulate gyrus

TE

Right middle frontal gyrus

D

Left superior frontal gyrus

Right extra-nuclear

NU

Right

Right inferior/middle frontal gyrus

1680

CE P

Right inferior frontal gyrus Right precentral gyrus

1656

6

13.18

-25

2

49

Left middle/superior frontal gyrus

1088

6

10.67

26

1

49

Right middle frontal gyrus

AC

CCD

Talairach Center

Right parahippocampal gyrus

T

CRA

Volume (mm3)

Brain region

IP

tasks

32

ACCEPTED MANUSCRIPT

SC R

IP

T

TASK-DEPENDENCE OF BRAIN-BASED INSIGHT

Figure 2 Schematic illustration of various brain activation foci reported in the literature included in the

NU

meta-analysis. In the figures, the black squares represent the foci reported for the Chinese CCD task, the red triangles represent the foci for the CRA task, and the red crosses represent the foci for the PH task. rPHG = right parahippocampal gyrus; aSTG = anterior superior temporal gyrus; CCD= character chunk decomposition; CRA =

AC

CE P

TE

D

MA

compound remote associates; PH = prototype heuristic.

33

ACCEPTED MANUSCRIPT

NU

SC R

IP

T

TASK-DEPENDENCE OF BRAIN-BASED INSIGHT

Figure 3 Summary of ALE meta-analysis activation clusters for insight solutions compared to non-insight solutions

MA

across the three tasks. CRA = compound remote associates; PH = prototype heuristic; CCD= character chunk

AC

CE P

TE

D

decomposition.

34

ACCEPTED MANUSCRIPT

TE

D

MA

NU

SC R

IP

T

TASK-DEPENDENCE OF BRAIN-BASED INSIGHT

CE P

Figure 4 The pattern of overlap in the meta-analytical ALE activation clusters between the CRA, PH, and CCD tasks.

AC

CRA = compound remote associates; PH = prototype heuristic; CCD= character chunk decomposition.

35

ACCEPTED MANUSCRIPT TASK-DEPENDENCE OF BRAIN-BASED INSIGHT Highlights

T

1 The issue of whether insightful process is task-specific or task-independent was examined.

SC R

3 Different brain networks engaged in distinct insight tasks.

IP

2 The fMRI studies used three types of cognitive insight tasks were meta-analyzed.

AC

CE P

TE

D

MA

NU

4 The findings are the first neural evidence for the task-dependence of brain-based insight.

36