Methods 42 (2007) 68–76

www.elsevier.com/locate/ymeth

Creativity meets neuroscience: Experimental tasks for the neuroscientiWc study of creative thinking

Andreas Fink ¤, Mathias Benedek, Roland H. Grabner, Beate Staudt, Aljoscha C. Neubauer

Institute of Psychology, University of Graz, Universitaetsplatz 2/III, A-8010 Graz, Austria

Accepted 2 December 2006

Abstract

The psychometric assessment of diVerent facets of creative abilities as well as the availability of experimental tasks for the neuroscien-tiWc study of creative thinking has replaced the view of creativity as an unsearchable trait. In this article we provide a brief overview ofcontemporary methodologies used for the operationalization of creative thinking in a neuroscientiWc context. Empirical studies arereported which measured brain activity (by means of EEG, fMRI, NIRS or PET) during the performance of diVerent experimental tasks.These tasks, along with creative idea generation tasks used in our laboratory, constitute useful tools in uncovering possible brain corre-lates of creative thinking. Nevertheless, much more work is needed in order to establish reliable and valid measures of creative thinking, inparticular measures of novelty or originality of creative insights.© 2006 Elsevier Inc. All rights reserved.

Keywords: Human cognition; Creativity; Creative thinking; Divergent thinking; Neuroscience

1. Introduction

Creativity is deWnitely a complex Weld of research. Onthe one hand, it pervades almost all areas of our everydaylife: It is important in the pedagogical, cultural, and in thescientiWc domain. Likewise, creativity is advantageous ineconomy and in the job. On the other hand, no conclusivedeWnition of this mental ability construct has been achievedyet. Most researchers agree that creativity is the ability toproduce work that is novel (original, unique), useful andgenerative [1]. Accordingly, creativity is regarded as a per-formance or ability trait, preferably manifested in original,valuable, and socially accepted ideas, products, or works ofart. This view is also reXected in the presumption that thecreativity level of an individual can be assessed by means ofperformance measures derived from creative thinking tasksor psychometric tests. But what kind of measures are these?

As originally suggested by Guilford, creative talent orcreative ability can be assessed by a number of variables

* Corresponding author. Fax: +43 316 380 9811.E-mail address: andreas.Wnk@uni-graz.at (A. Fink).

1046-2023/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ymeth.2006.12.001

such as ideational Xuency (i.e., number of ideas), the degreeof novelty (or uniqueness/originality) of ideas, or the Xexibil-ity of the mind (i.e., the ability to produce diVerent types ofideas, as opposed to rigidity) [2]. InXuenced by Guilford’ssuggestions many creativity measures have been developed,among the most inXuential are the Torrance Tests of Crea-tive Thinking (TTCT; [3]), Mednick’s Remote AssociatesTest [4], or Guilford’s divergent production tests [5].

The availability of creativity measures as well as theavailability of experimental tasks for the study of creativethinking has also motivated other scientiWc disciplines toenter into this complex mental ability domain. Recentresearch eVorts in the Weld of cognitive sciences and partic-ularly in the Weld of neurosciences have expanded ourknowledge about creativity to a considerable extent. DiVer-ent frameworks and theories about possible mechanismsunderlying creative thinking have been proposed [6,7]. Basi-cally, theoretical and empirical advances in these disciplineshave—along with psychometric approaches—displaced theviewpoint of creativity as an unsearchable phenomenon.This is nicely illustrated in Ward et al.’s concept of creativecognition which is considered as an extension of recent

A. Fink et al. / Methods 42 (2007) 68–76 69

work in cognitive psychology or cognitive science to thedomain of creative thinking [8]. Concepts, theories, andmethods that have already been employed in many noncre-ative research Welds of cognitive psychology are adapted forthe study of creative thought. Among the most prominentexamples of the creative cognition approach are, as out-lined in Ward et al., the study of insightful problem solving,creative imagery, extending concepts (or conceptual expan-sion, respectively), or the study whether creative productsare the result of recently activated knowledge (e.g., previ-ously seen examples). Likewise, Weisberg’s knowledge the-ory of creativity has also contributed to a betterunderstanding of this complex mental ability domain [9].He emphasizes the role of domain-speciWc knowledge as animportant prerequisite for creative functioning. Along thesame lines, other researchers pay attention to intellectualability as a key variable in creative thinking [10].

The viewpoint of creativity as performance or abilityoriented trait has been further underpinned by Dietrichwho provides a comprehensive review of contemporaryresearch in the Weld of cognitive science and neuroscience[6]. Dietrich argues that creativity requires a variety of clas-sic (frontal lobe demanding) cognitive abilities such asworking memory, sustained attention, or cognitive Xexibil-ity. Creative thinking involves, among others, the ability tobreak conventional rules of thinking or to develop newstrategies. Moreover, producing novel ideas by combiningalready stored knowledge elements [6] presumably alsoinvolves working memory, which is conceptualized as theability to temporarily maintain information in mind uponwhich concurrent information processing takes place.

Recent advances in the development of brain imagingtechniques like the quantiWcation of task- or event-related(de)synchronization of brain activity in the electroencepha-logram (EEG), the measurement of regional cerebral bloodXow (rCBF) via positron emission tomography (PET), orfunctional magnetic resonance imaging (fMRI) techniquesallow us to look at the brain when engaged in creativethinking [see also 11]. However, presumably due to diYcul-ties in operationalizing creative performance during EEG,PET, or fMRI measurements, neuroscientiWc studies thataim at identifying possible brain mechanisms related to cre-ative thinking are comparatively rare at the present time. Inthe following we provide a brief overview of existing meth-odologies used for the operationalization of creative think-ing in a neuroscientiWc context.

1.1. Operationalization of creativity in neuroscientiWc research

Table 1 summarizes empirical studies which measuredbrain activity during the performance of diVerent experi-mental tasks. The employed tasks cover diVerent aspects ofcreative thinking ranging from creative story generation,over mental imagery, to mental composition of music. Pet-sche, for instance, used verbal, visual, and musical tasks[12]. Bhattacharya and Petsche asked their participants to

mentally compose a drawing [13]. In other studies partici-pants were requested to solve match problems [14] or togenerate a story with given stimulus words [15,16]. Most ofthe tasks presented in this table were adapted from, or atleast inXuenced by Torrance’, Mednick’s or Guilford’s testsof creative thinking [3–5]. For example, Jung-Beeman et al.[17; see also 18] studied neural activity during the perfor-mance of compound remote associate problems which wereadapted from Mednick’s Remote Associates Test [4]. In thistask, Jung-Beeman et al. present three stimulus words (e.g.,pine, crab, sauce) and instruct their participants to producea single solution word (apple) that represents a compoundwith each of the three stimulus words (pineapple, crab apple,applesauce). Carlsson et al., Folley and Park, and Mölleet al. employed modiWed versions of the well-knownunusual uses task (see TTCT), which requires participantsto name as many alternative or unusual uses of a commonobject as possible [19–21].

An issue directly related to the experimental tasksreviewed here is the experimental design that should allowconclusions on the neuronal bases of creative thinking.Near infrared spectroscopy (NIRS), fMRI, and PET mea-sure brain activity indirectly by hemodynamic (in PET alsometabolic) parameters. The observed changes in brainactivity (e.g., from a baseline condition to an activationinterval) occur rather slowly (e.g., the BOLD signal infMRI reaches its maximum at 4–6 s and needs about 15 s todecline) which considerably limits the investigation of thetime-course of creative cognition. With respect to fMRI,perhaps the most important question is how to designdiVerent task conditions that can be used to isolate thebrain areas involved in creative cognition by means of thesubtraction method [22,23]. The mere comparison of a crea-tive thinking task with a resting period seems to be unsatis-factory as it is not known which cognitive processes takeplace in the resting condition. Binder et al., for instance,found that language regions are active during a restingperiod which was attributed to “mental soliloquizing” ofthe participants [24]. A more appropriate approach is thecomparison of tasks or conditions requiring creative think-ing to a diVerent extent. In this vein, Jung-Beeman et al.contrasted brain activity during problem solving with vs.without insight as indicated by the participants [17]. How-ard-Jones et al., in contrast, varied the extent of creativeengagement via the instruction to generate either creativeor uncreative stories [15].

While fMRI enables insights into the neuroanatomicalbases of creative cognition with high spatial accuracy, theprimary advantage of EEG lies in its high temporal resolu-tion (in the range of milliseconds) and the availability ofdiVerent parameters. All EEG studies presented in Table 1analyzed oscillatory EEG activity which is associated withfunctional network formation and dynamic interactionswithin and between brain structures during cognitive infor-mation processing [25–27]. In light of the complexity of cre-ative thinking, presumably requiring a highly coordinatedinterplay of diVerent neural networks, the analysis of

70 A. Fink et al. / Methods 42 (2007) 68–76

oscillatory activity seems to be the choice method in EEGresearch on creative cognition. In this context, at least twotypes of EEG parameters with diVerential functional sig-niWcance have to be distinguished: (a) analyses of (task-related changes in) the amplitude of EEG backgroundactivity in diVerent frequency bands [17,28] and (b) analysesof functional couplings (in amplitude and/or phase)between diVerent electrode sites [12,13]. In studies from ourresearch group, we primarily focused on task-related EEGpower changes in the alpha frequency band as this measurehas proved to be a useful and appropriate technique tomeasure the level and also the topographical distribution ofcortical activity during cognitive task performance [29].

The studies presented in Table 1 are organized by theneuroscientiWc measurement method. As can be seen in thistable, most of these studies measured brain activity bymeans of EEG, some of them used PET or fMRI, and in

one study NIRS was employed. In this context it is impor-tant to note that the applied measurement method placesseveral limitations on the experimental tasks and designsthat can be realized. Most of the creative thinking tasksreported in literature are paper-and-pencil tasks andrequire the participant to write down or even to draw ideasto a given stimulus. This response mode is very diYcult torealize in neurophysiological settings, as (a) time segmentsof writing or drawing may cause artifacts (e.g., muscle arti-facts in EEG or activation artifacts due to a task-relatedmotor activity in fMRI) and consequently reduce the num-ber of reliable (artifact-free) time segments that can be ana-lyzed, and (b) the test environment does not, or only to acertain extent, allow free-hand writing or drawing (e.g., withthe participants lying supine inside the fMRI scanner). Thisimportant restriction may also be a reason why most of thestudies employed verbal tasks. However, one way to

Table 1Overview of studies employing neuroscientiWc methods to the study of creative thinking

NeuroscientiWc measurement method

Task Description/example Scoring

[13] EEG Creative imagery Mentally compose a drawing on one’s own choice

[30–32,37,56] EEG Creative idea generation tasks Insight task, utopian situations, alternative uses, word ends

Ideational Xuency, originality of ideas

[28,34] EEG Closed vs. open problems Open problem task: Plan a day’s activities by considering certain task constraints (e.g., time needed to accomplish all tasks)

Solving eYciency (time, correctness of each solution etc.)

[16,35] EEG Creative story generation, drawing

Fantasy speech task: “A man meets a woman and asks her out on a date. Make up a story about who the people are, how they met and what will happen. Use your imagination.”

External ratings of creative quality [35]

[33] EEG Remote Associates Test; Alternate Uses Test

Name a fourth word that is related to a given stimulus triplet; Alternate uses for a brick, a shoe and a newspaper

[21] EEG Convergent vs. divergent thinking

Consequences: “Imagine you are able to Xy”; Alternate uses, similarities and meanings in pictures

Productivity (number of diVerent responses)

[12] EEG Verbal, visual and musical tasks

Creative story generation; Mentally create a picture; Mentally compose a short piece of music

[36,57] EEG Convergent vs. divergent thinking (hypothesis generation)

“There were hundreds of poisonous snakes in the zoo. How will it be possible to measure the lengths of each snake?”

Productivity (Xuency)

[17] EEG, fMRI Remote associates problems Form a compound word/phrase to, e.g., pine, crab, sauce

Subjective ratings (experience of AHA!)

[14] fMRI Match problems An arrangement of matches must be reorganized to make other patterns by removing matches

Reaction time and accuracy

[15] fMRI Creative story generation Generate a story using three presented words

External ratings: creative quality

[20] NIRS Divergent thinking tasks Generate uses for real objects Number of singular and combinatory uses (i.e., use for at least two objects within the stimulus set), response time

[19] rCBF via 133xenon inhalation method

Verbal Xuency task; uses of bricks

Verbal Xuency task: Generate as many words as possible beginning with a speciWed letter; Brick uses: Name as many diVerent uses of bricks

Fluency, number of categories

[58] rCBF via PET Verbal creativity tasks Creative story generation using words from diVerent semantic areas

A. Fink et al. / Methods 42 (2007) 68–76 71

circumvent this limitation is to isolate the time interval ofcreative thinking from the response interval. Bhattacharyaand Petsche, for example, asked the participants to men-tally compose a drawing while looking at a white wall [13].After EEG recording, they had to sketch the visualized pic-ture on paper. A similar procedure was administered in thefMRI study by Howard-Jones et al. who required partici-pants to (silently) generate creative vs. uncreative storiesfrom sets of three words visually presented inside the scan-ner [15]. Immediately after leaving the scanner the partici-pants had to identify the word sets from a list and to recalltheir stories from a selection of word sets. In verbal creativethinking tasks, an alternative approach is to require oralresponses which can be recorded by the experimenter. Thisprocedure which has been repeatedly employed in the EEGstudies from our research group shall be introduced later inthis article [30–32].

DiVerent methodologies and experimental designs havebeen realized in the reviewed literature depending on therespective research question. Brain activation patterns werecompared between (a) low vs. high creative individuals[19,33], (b) between tasks involving low vs. high creativity[21], and (c) between low vs. high original ideas within onetask [17,30]. Particularly in the latter approach measures ofthe creative quality of the produced ideas are needed. Thispoint is not satisfactorily addressed in the literature inas-much as most researchers merely focus on quantitativemeasures such as ideational Xuency (i.e., number of ideas).Some studies focused on parameters of “solving eYciency”such as reaction time and accuracy measures [14,34]. Quali-tative measures such as measures of novelty, uniqueness, ororiginality are almost completely lacking (cf. right columnin Table 1). Only few studies used qualitative measures ofcreative task performance. Howard-Jones et al., forinstance, applied an external rating procedure in order toobtain a measure of creative quality of the generated stories[15; see also 35]. In the study by Jung-Beeman et al. self-rat-ings of the given responses were obtained [17]. In theremaining studies summarized in Table 1 no qualitativemeasures of creative task performance are reported.

What do the studies summarized in Table 1 tell us aboutpossible brain correlates of creative thinking? How canthese studies contribute to a better understanding of thiscomplex mental ability domain? On the whole, they appearto indicate that diVerent modes of thinking are accompa-nied by diVerent activity patterns of the brain. For example,when participants’ task is to name as many unusual uses ofa common, everyday object or to think of as many conse-quences as possible of a given hypothetical situation, theydisplay a diVerent pattern of electrophysiological brainactivity than during the performance of intelligence relatedtasks (e.g., mental arithmetic, or continuation of a row ofletters; cf. [21]). Mölle et al. reported higher EEG complex-ity during the performance of more “free-associative” typesof tasks which could be the result of a larger number ofindependently oscillating neural assemblies during this typeof thinking [21]. Similarly, Jausovec and Jausovec as well as

Razoumnikova reported Wndings which also indicate thatthese diVerent modes of thinking are accompanied bydiVerent activity patterns in the EEG [34,36]. Hence, brainactivation patterns during free-associative thinking appearto stand in contrast to activation patterns that are usuallyobserved during the performance of conventional cognitivetasks, e.g., intelligence test tasks. The studies summarized inTable 1 also suggest that during creative thinking, the brainactivation patterns of highly creative individuals are signiW-cantly diVerent from those observed in lower creative indi-viduals [13,33]. Finally, neuroscientiWc studies also revealthat brain states that come along with highly original ideasdiVer from those observed during the production of lessoriginal, conventional ideas [17,30,32].

2. Combining novel tasks of creative thinking with EEG

Our on-going research deals with the exploration ofdiVerent aspects of creative thinking. In pursuing a neuro-scientiWc approach, we measured brain activity (by meansof EEG) while participants were engaged in the perfor-mance of creative idea generation tasks [30–32,37]. In thefollowing we provide a detailed description of our method-ologies with special focus on combining novel tasks of crea-tive thinking with EEG. In this context it is important tonote that the choice and modiWcation of tasks was guidedby the goal to realize them appropriately within an EEGrecording session. Some limiting factors that have an eVecton the realization of experimental tasks in neuroscientiWcstudies have already been outlined in previous sections ofthis article and we will refer to these when we discuss therationale of our task procedures in more detail.

Perhaps the most important restriction in the choice andmodiWcation of experimental tasks was that the task con-straints should be low to enable the participant to producea broad range of ideas (i.e., many, diVerent, and originalideas). This was not only done to achieve a large number oftrials (i.e., allowing reliable EEG measurements) but alsomotivated by our research aim to contrast brain states thatcome along with qualitatively diVerent responses or ideas(e.g., more vs. less original). In order to select the most suit-able tasks out of literature, several pilot studies were runprior to the EEG sessions in which several creative thinkingtasks adapted from the Torrance Tests of Creative Think-ing [3] and from well-established German creativity tests,viz., the “Verbaler Kreativitäts-Test” [39] and the “BerlinerIntelligenzstruktur-Test” [40] were analyzed [38].

2.1. Description of experimental tasks

In Fig. 1 an overview of the selected tasks is given. Fourtasks with two diVerent items each were used. In the insighttask (IS) participants are confronted with unusual, hypo-thetical situations that are in need of explanation. They arerequired to think of as many and diVerent causes or circum-stances as possible. Similarly, in the utopian situation task(US) participants are instructed to put themselves in the

72 A. Fink et al. / Methods 42 (2007) 68–76

Fig. 1. Overview of creative idea generation tasks and schematic time-course of experimental procedure. Brain activity during creative idea generation wasquantiWed by means of task- or event-related power changes in the EEG alpha band. A 10 s time interval during the presentation of the Wxation cross (pre-stimulus reference interval R) as well as a 1 s time interval 250 ms before pressing the IDEA-button (activation interval A) were used for EEG analyses.For quantifying task-related power changes (TRP) in the alpha band [59] the (log-transformed) power during the reference interval was subtracted fromthe (log-transformed) power during the activation interval for each electrode i according to the formula: TRPi D log [Powi activation]¡ log [Powi refer-ence]. Decreases in alpha band power from the reference to the activation interval are reXected in negative TRP values, whereas task-related increases areexpressed in positive values [59]. As indicated in the lower part of this Wgure, somewhat higher TRP values (i.e., increases in alpha activity from the pre-stimulus reference to the activation interval; as symbolized by blue colors in the brain maps) were found in the AU, IS and US tasks as compared to theWE task, most prominent in posterior regions of the cortex (cortical areas: AF, anteriofrontal; F, frontal; FC, frontocentral; CT, centrotemporal; CP,centroparietal; PT, parietotemporal; PO, parietooccipital). (For interpretation of the references to color in this Wgure legend, the reader is referred to theweb version of this paper.)

A. Fink et al. / Methods 42 (2007) 68–76 73

given utopian situations that will actually never happenand to produce as many and original ideas as possible. Inthe alternative uses test (AU) participants’ task is to nameas many and unusual/original uses of a conventional, every-day object. And Wnally, in the word ends task (WE), Ger-man suYxes are presented that have to be completedoriginally by the participants.

Prior to the presentation of the experimental tasks, par-ticipants were familiarized with the response mode by pre-senting them a simple association task. To this end, theywere instructed to produce associations to the concept“holiday” and to press an IDEA-button whenever an asso-ciation or idea related to this stimulus word came into theirmind. Immediately after pressing the IDEA-button a mes-sage appeared on screen asking the participants to vocalizethe idea (which was recorded by the experimenter) andWnally to conWrm it by pressing this button again where-upon the stimulus reappeared on screen. Subsequently, theexperimental session started. All tasks began with the pre-sentation of a warming up trial (e.g., IS: “Person A wearsonly designer clothes”; or US: “What would happen, if sud-denly an ice-age would break in”). Possible solutions/ideas tothe warming up items were given in order to clarify therange of thinkable responses (IS: “is rich”, “attaches impor-tance to good quality”, “is sponsored by the designer”; US:“stronger hairiness in animals and men”, “increasing heatingcosts”, “insolvency of open air baths”). The item presenta-tion time (or the response interval, respectively) was 3 minfor each test item (see Fig. 1). The choice of item presenta-tion time (or the response interval, respectively) was guidedby the following considerations: Relevant literature sug-gests that the idea rate usually decreases over time, espe-cially within the Wrst 6 min [41]. On the other hand, there isa general increase in idea originality over time [42–44], withmore common ideas being more likely produced at thebeginning of the task, followed by more remote ones. Atask presentation time (or a response interval, respectively)of 3 min has proven to be long enough to cover more thaninitial common ideas and short enough to keep mean idearate up to about four ideas per minute.

2.2. Scoring of task performance

In addition to the assessment of ideational Xuency (i.e.,counting the number of given responses) we also measuredthe originality of ideas. This was realized twofold: First, inorder to obtain a subjective measure of originality, weprompted our participants to judge the originality of theirown ideas given during the performance of the experimen-tal tasks. To this end, the given responses were presented tothe participants subsequent to the EEG recording sessionand participants were required to rate each single idea on aWve-point rating scale ranging from 1 (highly original) to 5(not original at al ). Second, the originality of ideas was alsoassessed by means of an external rating procedure similarto the Consensual Assessment Technique (CAT) proposedby Amabile [45]; see also [46,47]. The CAT technique is

based on a consensual deWnition of creativity. Accordingly,“ƒcreativity can be regarded as the quality of products orresponses judged to be creative by appropriate observersƒ” [45], p. 1001. In our studies, the ideas of each partici-pant were rated by three female and three male raters(advanced diploma psychology students) with regard totheir uniqueness or originality. SpeciWcally, the raters wereinstructed to evaluate the ideas separately for each task (i.e.,separately for the IS, US, AU, and WE task, respectively)and each participant. They were prompted to read allresponses a participant gave in one task to gain a Wrst over-all impression of its originality range and—based on this—to rate the originality of the given responses on a Wve-pointrating scale ranging from 1 (highly original) to 5 (not origi-nal at all). The raters were asked to utilize the completescale range as far as possible. In order to obtain a measureof internal consistency of the ratings we computed Cron-bach alpha coeYcients, separately for each of the fourexperimental tasks (in considering the six available ratingsas “items”). We found satisfactory internal consistencybetween the ratings (Cronbach alphas of .93, .92, .91 and .87for the IS, US, AU, and WE task, respectively). Subse-quently, the ratings were averaged over all six raters (foreach of the four tasks separately) so that one originalitymeasure was available for each task and participant.

2.3. Rationale of task procedure

As already mentioned above, the choice as well as somemodiWcations of the tasks (e.g., task timing, IDEA buttonprocedure) was guided by the goal to realize them appro-priately within an EEG recording session. In the following,the rationale of the task procedure is outlined more thor-oughly.

As illustrated in Fig. 1, each task started with the presen-tation of a Wxation cross for a time period of 15 s, whichserved for the assessment of baseline EEG activity. Brainactivity during creative idea generation was measured bymeans of an event- or task-related approach. Concretely,EEG activity during the performance of the experimentaltasks (i.e., the generation of ideas, cf. abbreviation “A” inFig. 1) was contrasted with EEG activity during the pre-stimulus reference period (abbreviation “R” in the Figure).The term “event-“ or “task-related” refers to the principlethat brain activity during the performance of a given cogni-tive task is related to a pre-stimulus reference interval dur-ing which no task is performed [29]. Provided that diVerentexperimental groups or tasks do not diVer in pre-stimulusreference brain activity, any diVerences in event- or task-related brain activity are due to group or task diVerences inthe process of task performance (in this context creativeidea generation) and not to diVerences in baseline brainactivity. In this regard it should be noted that the basicprinciple and rationale of this baseline interval is notdirectly comparable to a resting period in fMRI studies.While the latter should reXect a task-related control condi-tion used to “subtract” all cognitive processes from the

74 A. Fink et al. / Methods 42 (2007) 68–76

experimental condition that are of no interest for theresearch question, the function of the reference interval inthis EEG approach is to adapt for individual EEG powerdiVerences in diVerent frequency bands and to quantifychanges in the oscillatory dynamics that are induced by anevent or task [25]. These changes can then be interpretedwith respect to the functional signiWcance of the respectivefrequency band (see below).

As already outlined above, we requested the participantsto press a so-called IDEA-button whenever a creative ideacomes into mind. In applying this procedure, the experi-menter can obtain information about an individual’s brainactivity at any point of time, for instance, immediately priorto the production of creative ideas. In this particular con-text, the EEG method provides a special advantage overother brain imaging techniques (such as fMRI) as it enablesthe analysis of brain activity with a comparatively hightemporal resolution. Given that the peak of the BOLD sig-nal (in response to a stimulus) is time-delayed and the gen-eral sluggishness of this hemodynamic response, fMRIstudies do not allow for such a Wne-grained temporal analy-sis of brain activity. After pressing the IDEA-button, par-ticipants were requested to vocalize the idea and to WnallyconWrm it with a button-press. With this procedure we werenot only able to analyze brain activity immediately prior tothe production of creative ideas but also to precisely local-ize (and exclude) EEG segments that were contaminatedwith the production of speech, body movements, eye blinks,etc. This experimental design would also be very diYcult torealize in an fMRI study. If the participant’s responses(ideas) occur at a comparably high rate with short intervalsin between, the long-lasting BOLD signal does not allowdiVerentiating brain activity related to creative thinking(before the response) and the oral response itself. As a con-sequence, any activation of language-related brain areasmight not solely be due to the semantic processing of thepresented verbal stimulus (e.g., in a coarse semantic net-work presumably required for the production of remote ormore original ideas, respectively; [48]) but also to the speechproduction itself.

The task procedure used in our previous studies alsoenables—provided that a suYcient number of ideas isgiven—the analysis of brain activity that is associated withthe production of highly original ideas, as compared tobrain responses that come along with the production ofconventional, customary ideas. This is very similar to theapproach adopted by Jung-Beeman et al. who analyzedbrain activity during the subjective experience of “AHA!”[17]. In that study, participants were asked to indicatewhether they had solved the problem with or withoutinsight, herewith providing trials that were either solvedwith or without the unique experience of “AHA!”. Like-wise, the responses a participant gave in the experimentaltasks described above can also quite easily be classiWed intolower original and higher original ones (e.g., by means of amedian split of the ratings within each person). This classiW-cation can be based on both, external and subjective ratings

of the given responses which can elucidate the neurophysio-logical bases of creative cognition from diVerent perspec-tives [30,32].

In contrast to most creative thinking tasks which arecommonly used as paper-and-pencil tasks, participantswere requested to respond orally to the stimuli presented onthe screen. The choice of the oral response mode wasguided by the following reasons: First, the production ofcreative ideas—in particular ideational Xuency (i.e., thenumber of ideas)—is not contaminated by writing or typingspeed, respectively. In addition, oral responding to the stim-uli is far less time consuming than typing into the com-puter, particularly in those participants who are notcapable of typewriting or are not familiar with computers.Also, given that the participants are requested to respondorally, they can exclusively focus on the production ofnovel and original ideas, being less likely distracted from“technical surroundings”. The application of EEG (butalso NIRS and PET) is also advantageous over fMRI withrespect to this response mode. In fact, the loud noise of theMR scanner due to switching of the gradient coils makes itvery diYcult to record oral responses with the standardmicrophones installed in the scanner. Therefore, either spe-cial microphones have to be used or the image acquisitionhas to be suspended for a pre-deWned response period.

2.4. Analyses of experimental tasks

There were several behavioral and neurophysiologicalanalyses run in order to assess the psychometric quality(i.e., reliability and validity) of the presented creative ideageneration tasks. The main results of these analyses shouldbe brieXy discussed, particularly with respect to aspects ofconvergent vs. discriminant validity of the tasks. Generally,the behavioral analyses reveal that the tasks may draw ondiVerent cognitive functions, ranging from more intelli-gence-related (e.g., the WE Taskß as evident by a signiWcantcorrelation with verbal IQ) to more free-associative tasks(i.e., IS, US, AU), which are not correlated with verbal intel-lectual ability at all [31,38]. These latter tasks are, unlike theWE task, considerably associated with creativity-relatedpersonality variables such as “openness to new experi-ences” which is frequently found to be related to creativity[49]. It appears to be noteworthy that task diVerences wereapparent on the neurophysiological level as well. As illus-trated in Fig. 1, the IS, US, and AU task elicited strongerincreases in EEG alpha activity (i.e., synchronization ofalpha activity; symbolized by blue colors in the maps) thanthe more intelligence-related WE task which displayed thelowest increase in alpha activity (symbolized by yellow col-ors in the maps). In addition to this, stronger increases inEEG alpha activity (relative to rest) were also found inmore original ideas as compared to less original ones[30,32].

But what is the functional meaning of synchronization ofEEG alpha activity? What does this tell us about possiblebrain correlates of creative thinking? The traditional view-

A. Fink et al. / Methods 42 (2007) 68–76 75

point considers task- or event-related alpha synchronizationas a cortical idling phenomenon, presumably reXecting areduced state of active information processing in the underly-ing neuronal networks [50]. Hence, synchronization of alphaactivity during creative idea generation could indicate that areduced or a lower activity level of the brain is needed to pro-duce novel, original ideas (cf. Martindale’s low arousal the-ory [51]). However, besides the view of alpha synchronizationas a cortical “idling” phenomenon, there is also recent evi-dence suggesting that synchronization of alpha activity couldreXect an active cognitive process, viz., an active inhibition oftask-irrelevant brain regions [26,52,53]. Thus, alpha synchro-nization observed during creative thinking may also reXectan inhibitory mechanism that prevents internal informationprocessing to be disturbed by external input or conXictingoperations [54]. However, further studies which combinediVerent neuroscientiWc measurement methods (such asfMRI and EEG) within one and the same experimentaldesign are needed to clarify the functional meaning of EEGalpha synchronization during creative thinking.

3. Concluding remarks

In this article we have focused on contemporary method-ologies used for the operationalization of creative thinking inneurophysiological research settings. One important factorthat should be kept in mind in evaluating neuroscientiWc cre-ativity literature is that well-designed, widely acknowledgedexperimental tasks that have been successfully employed inbehavioral studies of creative thinking can often not beadopted one-to-one to neuroscientiWc studies. In many cases,comparatively “simple” experimental tasks were used duringthe performance of neurophysiological measurements (suchas EEG, NIRS, PET, or fMRI). Nevertheless, these studieshave yielded useful information about possible brain corre-lates of creative thinking, herewith extending our knowledgeabout this complex mental ability domain.

The tasks employed in our laboratory may also be usefulin the neuroscientiWc study of creative thinking. Partici-pants are presented verbal problems (e.g., they have tothink of original causes or consequences, or to think of asmany, as diVerent, and as original uses of conventionaleveryday objects) for a certain period of time. Wheneverthey have an idea, they are requested to press the IDEAbutton and to vocalize it. Thereby, the time-point of ideaproduction is self-directed by the participants, each singleidea can be recorded, and the IDEA button responses pro-vide triggers for analyzing critical EEG time intervals (e.g.,immediately before the production of creative or uncreativeideas). In this experimental design, the high temporal reso-lution of the EEG is deWnitely advantageous over otherneurophysiological measurement methods such as fMRI.The presented tasks also allow for the generation of a suY-cient number of ideas (more than 10 ideas within 3 min)which, in turn, results in an adequate number of trials nec-essary to obtain reliable neurophysiological measures ofcreative thinking.

There are, nevertheless, also some important shortcom-ings of the presented methodologies that need to beaddressed in future studies. First, the creative idea genera-tion tasks presented in this article are exclusively verbaltasks (i.e., employing verbal stimulus material). In this par-ticular context it is also important to note that the pre-sented tasks appear to draw on diVerent verbal demands.While in the IS or US task participants were required torespond to verbally presented scenarios or hypothetical sit-uations, in the WE task presented suYxes had to com-pleted. Obviously, the tasks considerably diVer with respectto the involvement of language or speech [38]. This was alsoreXected in the Wnding that performance in the WE taskwas more strongly related to verbal intelligence than per-formance in the remaining tasks. Future research shouldalso invest into the construction of experimental tasks suit-able for the neuroscientiWc study of non-verbal creativitysuch as creative imagery or visual art [13].

Moreover, comprehensive measures of creative task per-formance are needed. When brain activity patterns duringcreative thinking are analyzed we need not only to examinewhether or to which extent the participants are actuallyengaged in creative thinking, we also need to know howsuccessfully they are in the production of novel, originalideas. This holds particularly true when brain states duringthe production of highly creative ideas are contrasted tothose observed during the production of customary, con-ventional ideas. As evident in Table 1, qualitative measuresof creative task performance are rare. Due to diYculties indeWning and measuring the uniqueness, novelty, or origi-nality of ideas, creative task performance has mostly beenassessed by means of the Xuency of ideas. Operationalizingnovelty or originality solely in terms of statistical infre-quency may be not satisfying given that crazy, useless, orsilly answers are normally also statistically infrequent.Besides subjective ratings of the given responses [17,32] anexternal rating procedure with trained raters [cf. also 45]could be a possible alternative. Along this way, it is notonly possible to obtain a quality measure of creativityreXecting novelty and uncommonness but, depending onthe respective rating instruction, also a measure that takesusefulness and appropriateness into account.

Another unresolved issue is the question as to how theWndings obtained with creative idea generation tasks relateto real-life creativity? Can brain activity patterns that havebeen observed during the performance of these relativelybasic tasks tell us anything about creativity? We think so.In the presented idea generation tasks participants areexplicitly instructed to produce ideas that cover a broadrange of diVerent ideas (i.e., diVerent types of ideas) and toproduce unusual, original ideas no one else would think of.Hence, the presented tasks appear to represent more thanmerely verbal Xuency tasks, they can be viewed as a usefulestimate of the potential for creative thought [55]. Also, at aneurophysiological level, brain activity patterns consider-ably diVer depending on whether an individual is creativelyresponding to hypothetical situations (i.e., performing the

76 A. Fink et al. / Methods 42 (2007) 68–76

IS task) or engaged in the completion of suYxes (i.e., per-formance of the WE task). This is in line with neuroscien-tiWc research suggesting that diVerent modes of thinking areassociated with functionally diVerent activity patterns of thebrain (cf. Table 1). In applying diVerent measurement meth-ods (such as fMRI and EEG) on well-established experi-mental tasks of creative thinking we will get along in theunderstanding of possible brain mechanisms related to cre-ative thinking.

The work presented in this article should motivateresearchers from diVerent disciplines to enter into this men-tal ability domain. Creativity is often viewed as a diYcultWeld of research. However, recent advances in cognitive,psychometric and neuroscientiWc research have qualiWedthis viewpoint to a considerable extent. We hope that wecould demonstrate that valuable approaches to the neuro-scientiWc study of creative thinking are available now.

Acknowledgements

The research presented in this article was supported by agrant from the Austrian Science Foundation (P16393). Theauthors wish to thank Silvana Weiss and Anna Kanape fortheir help in the work presented in article.

References

[1] R.J. Sternberg, T.I. Lubart, Am. Psychol. 7 (1996) 677–688.[2] J.P. Guilford, Am. Psychol. 5 (1950) 444–454.[3] E.P. Torrance, Torrance Tests of Creative Thinking, Scholastic Test-

ing Service, Bensenville IL, 1966.[4] S.A. Mednick, Psychol. Rev. 69 (1962) 220–232.[5] J.P. Guilford, The Nature of Human Intelligence, McGraw-Hill, New

York, 1967.[6] A. Dietrich, Psychon. B Rev. 11 (2004) 1011–1026.[7] D.K. Simonton, Am. Psychol. 55 (2000) 151–158.[8] T.B. Ward, S.M. Smith, R.A. Finke, in: R.J. Sternberg (Ed.), Hand-

book of Creativity, Cambridge University Press, Cambridge, 1999, pp.189–212.

[9] R.W. Weisberg, in: R.J. Sternberg (Ed.), Handbook of Creativity,Cambridge University Press, Cambridge, 1999, pp. 26–250.

[10] F. Barron, D.M. Harrington, Annu. Rev. Psychol. 32 (1981) 439–476.[11] K.M. Heilman, S.E. Nadeau, D.O. Beversdorf, Neurocase 9 (2003)

369–379.[12] H. Petsche, Int. J. Psychophysiol. 24 (1996) 145–159.[13] J. Bhattacharya, H. Petsche, Hum. Brain Mapp. 26 (2005) 1–14.[14] V. Goel, O. Vartanian, Cereb. Cortex 15 (2005) 1170–1177.[15] P.A. Howard-Jones, S.-J. Blakemore, E.A. Samuel, I.R. Summers, G.

Claxton, Cogn. Brain Res. 25 (2005) 240–250.[16] C. Martindale, D. Hines, L. Mitchell, E. Covello, Pers. Indiv. DiVer. 5

(1984) 77–86.[17] M. Jung-Beeman, E.M. Bowden, J. Haberman, J.L. Frymiare, S.

Arambel-Liu, R. Greenblatt, P.J. Reber, J. Kounios, PLOS Biol. 2(2004) 500–510.

[18] E.M. Bowden, M. Jung-Beeman, J. Fleck, J. Kounios, Trends Cogn.Sci. 9 (2005) 322–328.

[19] I. Carlsson, P.E. Wendt, J. Risberg, Neuropsychologia 38 (2000) 873–885.[20] B.S. Folley, S. Park, Schizophr. Res. 80 (2005) 271–282.

[21] M. Mölle, L. Marshall, B. Wolf, H.L. Fehm, J. Born, Psychophysiol-ogy 36 (1999) 95–104.

[22] M.E. Raichle, J. Neurosci. 23 (2003) 3959–3962.[23] R.A. Poldrack, NeuroImage 12 (2000) 1–13.[24] J.R. Binder, J.A. Frost, T.A. Hammeke, P.S.F. Bellgowan, S.M. Rao,

R.W. Cox, J. Cogn. Neurosci. 11 (1999) 80–93.[25] M. Bastiaansen, P. Hagoort, Cortex 29 (2003) 967–992.[26] W. Klimesch, M. Doppelmayr, J. Schwaiger, P. Auinger, T. Winkler,

Cogn. Brain Res. 7 (1999) 493–501.[27] G. Pfurtscheller, F.H. Lopes da Silva, in: E. Niedermeyer, F.H. Lopes

da Silva (Eds.), Electroencephalography: Basic Principles, ClinicalApplications, and Related Fields, 5th ed., Lippincott Williams & Wil-kins, Philadelphia, 2005, pp. 1003–1016.

[28] N. Jausovec, Intelligence 28 (2000) 213–237.[29] C. Neuper, W. Klimesch, Event-related Dynamics of Brain Oscilla-

tions, Series: Progress in Brain Research, Elsevier, Amsterdam, 2006.[30] A. Fink, A.C. Neubauer, Int. J. Psychophysiol. 62 (2006) 46–53.[31] A. Fink, R.H. Grabner, M. Benedek, A.C. Neubauer, Eur. J. Neurosci.

23 (2006) 2241–2246.[32] R.H. Grabner, A. Fink, A.C. Neubauer, Behav. Neurosci., in press.[33] C. Martindale, D. Hines, Biol. Psychol. 3 (1975) 71–80.[34] N. Jausovec, K. Jausovec, Int. J. Psychophysiol. 36 (2000) 73–88.[35] C. Martindale, N. Hasenfus, Biol. Psychol. 6 (1978) 157–167.[36] O.M. Razoumnikova, Cogn. Brain Res. 10 (2000) 11–18.[37] A. Fink, A.C. Neubauer, Eysenck meets Martindale: The relationship

between extraversion and originality from the neuroscientiWc perspec-tive, submitted for publication.

[38] M. Benedek, A. Fink, A.C. Neubauer, Creativity Res. J. 18 (2006) 317–328.[39] K. Schoppe, Verbaler Kreativitätstest (V-K-T) [Verbal creativity test],

Hogrefe, Göttingen, 1975.[40] A.O. Jäger, H.-M. Süß, A. Beauducel, Berliner Intelligenzstruktur-

Test [Berlin Intelligence Structure Test], Hogrefe, Göttingen, 1997.[41] P.A. Howard-Jones, S. Murray, Creativity Res. J. 15 (2003) 153–166.[42] J.D. Moran, R.M. Milgram, J.K. Sawyers, V.R. Fu, Child Dev. 54

(1983) 921–926.[43] P.V. Olczak, M.F. Kaplan, Am. J. Psychol. 82 (1969) 157–167.[44] W.C. Ward, Child Dev. 40 (1969) 869–878.[45] T. Amabile, J. Pers. Soc. Psychol. 43 (1982) 997–1013.[46] D. Hocevar, J. Pers. Assess. 45 (1981) 450–464.[47] J.A. Plucker, J.S. Renzulli, in: R.J. Sternberg (Ed.), Handbook of Cre-

ativity, Cambridge University Press, Cambridge, 1999, pp. 35–61.[48] M. Jung-Beeman, Trends Cogn. Sci. 9 (2005) 512–518.[49] G.J. Feist, Pers. Soc. Psychol. Rev. 2 (1998) 290–309.[50] G. Pfurtscheller, A. Stancak Jr., C. Neuper, Int. J. Psychophysiol. 24

(1996) 39–46.[51] C. Martindale, in: R.J. Sternberg (Ed.), Handbook of Creativity, Cam-

bridge University Press, Cambridge, 1999, pp. 137–152.[52] O. Jensen, J. Gelfand, J. Kounios, J.E. Lisman, Cereb. Cortex 12

(2002) 877–882.[53] P. Sauseng, W. Klimesch, M. Doppelmayr, T. Pecherstorfer, R. Freun-

berger, S. Hanslmayr, Hum. Brain Mapp. 26 (2005) 148–155.[54] N.R. Cooper, R.J. Croft, S.J.J. Dominey, A.P. Burgess, J.H. Gruzelier,

Int. J. Psychophysiol. 47 (2003) 65–74.[55] M.A. Runco, in: M.A. Runco, S.R. Pritzker (Eds.), Encyclopedia of

Creativity, Academic Press, San Diego, CA, 1999, pp. 577–582.[56] B. Staudt, A.C. Neubauer, High Abil. Stud. 17 (2006) 3–16.[57] O.M. Razoumnikova, Neurosi. Lett. 362 (2004) 193–195.[58] N.P. Bechtereva, A.D. Korotkov, S.V. Pakhomov, M.S. Roudas, M.G.

Starchenko, S.V. Medvedev, Int. J. Psychophysiol. 53 (2004) 11–20.[59] G. Pfurtscheller, in: 4th ed., G. Pfurtscheller, F.H. Lopes da Silva

(Eds.), Event-Related Desynchronization, Handbook of Electroen-cephalography and Clinical Neurophysiology, vol. 6, Elsevier,Amsterdam, 1999, pp. 89–105.