creativity and depression

1677

The Dark Side of Creativity: BiologicalVulnerability and Negative Emotions Lead toGreater Artistic Creativity

Modupe AkinolaWendy Berry MendesHarvard University

intense negative emotions can create powerful self-reflective thought and perseverance, leading toincreased creativity (De Dreu, Baas, & Nijstad, 2008;Kaufman & Baer, 2002; Verhaeghen, Joormann, &Khan, 2005; cf. Isen, 2000). In this article, we exploreindividual differences and situational factors related tocreativity. We show that when individuals are biologi-cally vulnerable to experiencing negative affect and areexposed to a situation that brings about intense negativeemotion, they show the most artistic creativity.

Individual Differences in CreativityDecades of empirical research on personality traits of

highly creative individuals have identified a relativelyconsistent set of core characteristics of creative individ-uals. These traits include, for example, introversion,emotional sensitivity, openness to experience, andimpulsivity (see Feist, 1998, for a review). At theextreme of affective personality factors linked to cre-ativity are mood disorders. Historical figures in a vari-ety of creative domains, ranging from Emily Dickinsonto Robert Schumann, have been found to have suffered

Authors Note: We thank Teresa Amabile for her assistance on the cre-ativity task and for her intellectual guidance. We also thank the manydedicated lab assistants at the Health and Psychophysiology Lab atHarvard, especially Marina Nasman and Kathy Lee for their assistancein conducting this experiment. Wendy Berry Mendes was supported bya National Institute of Heart, Lung, and Blood grant (RO1 HL079383)and by a Cooke research grant. Please address correspondence toWendy Berry Mendes, Harvard University, 33 Kirkland, WJH 1420,Cambridge, MA 02138; e-mail: [email protected].

PSPB, Vol. 34 No. 12, December 2008 1677-1686DOI: 10.1177/0146167208323933 2008 by the Society for Personality and Social Psychology, Inc.

Historical and empirical data have linked artistic cre-ativity to depression and other affective disorders. Thisstudy examined how vulnerability to experiencing neg-ative affect, measured with biological products, andintense negative emotions influenced artistic creativity.The authors assessed participants baseline levels of anadrenal steroid (dehydroepiandrosterone-sulfate, orDHEAS), previously linked to depression, as a measureof affective vulnerability. They then manipulated emo-tional responses by randomly assigning participants toreceive social rejection or social approval or to a nonso-cial situation. Participants then completed artistic col-lages, which were later evaluated by artists. Resultsconfirmed a person-by-situation interaction. Socialrejection was associated with greater artistic creativity;however, the interaction between affective vulnerability(lower baseline DHEAS) and condition was significant,suggesting that situational triggers of negative affectwere especially influential among those lower inDHEAS, which resulted in the most creative products.These data provide evidence of possible biological andsocial pathways to artistic creativity.

Keywords: creativity; social rejection; neuroendocrine;DHEAS; affective vulnerability

What makes someone creative? Certainly, someindividuals are more creative than others. Wemerely need to compare da Vinci and Monet master-pieces to our own prosaic attempts at drawing a bowl offruit to conclude that artistic creativity is something thatis individualized and immutable. However, there is sub-stantial research that shows evidence for strong situa-tional factors influencing creativity. In some cases,

at NATIONAL UNIV LIBRARY on May 8, 2015psp.sagepub.comDownloaded from

from depression and other mood disorders (Goodwin &Jamison, 1990; Ramey & Weisberg, 2004; Weisberg,1994). Clinical, empirical, and biographical studies ofcreative individuals have shown that those in the cre-ative arts suffer from significantly higher rates of mooddisorders compared to matched controls (Andreasen,1987; Ludwig, 1995), and mood disorders are 8 to 10times more prevalent in writers and artists than in thegeneral population (Jamison, 1993).

One way to index vulnerability to experiencingdepression or negative mood is with biological products.Major depression is often associated with dysregulationof the hypothalamic-pituitary-adrenocorticol axis,which controls the release of adrenal steroids. Oneadrenal steroid that is commonly implicated in depres-sion is dehydroepiandrosterone (DHEA) and its activemetabolite DHEA sulfate (DHEAS). DHEAS levels havebeen found to be lower in those being treated for depres-sion (Wolkowitz et al., 1997), and adrenal stimulationresulting in increased DHEA levelshas been associatedwith lower symptom severity and improved mood inpatient samples (Rasmusson et al., 2004). The anxiolyticeffects of DHEA are believed to stem from its ability tocounter-regulate the deleterious effects of high cortisoland as a precursor to sex hormones (see Epel, Burke, &Wolkowitz, 2007, and Southwick, Vythilingam, &Charney, 2005, for reviews). In this study, we examineDHEAS level and its link to artistic creativity.

Emotional Influences on CreativityAlthough dispositional traits have been reliably linked

to creativity, person-level factors are only half of thestory; situational factors have been related to creativityas well. Literature on mood and creativity has demon-strated that both positive and negative affect can influencecreative performance. Although some evidence suggeststhat positive mood can enhance creativity (see Isen, 2000,for a review), many other studies have demonstrated thatnegative affect can have a facilitative effect on creativity(see Kaufmann, 2003, for a review).

The contradictory findings regarding the effects ofemotion on creativity stem from a variety of reasons,but the critical factors seem to be the type of task thatis being used as the outcome of creativity and the natureof the emotional experience. Regarding the type of cre-ativity, in some cases optimal creative performance mayrequire elaboration and analytic processing with highdetail orientation (Mackie & Worth, 1991; Schwarz,1990). For example, it has been found that negativemood can result in enhanced solution frequency on cre-ative tasks, particularly during tasks that require con-centration, precise execution, divergent thinking, andanalogical problem solving (Abele, 1992; Jausovec,

1989; Kaufmann & Vosburg, 1997). In other cases,optimal creative performance may require increasedreliance on rapid, less effortful judgment heuristicstrategies that show little systematic and analytic pro-cessing (Fiedler, 2000; Isen, Daubman, & Nowicki,1987).

The type of emotional state appears to have an influ-ence on creativity as well. In a recent meta-analysis,Baas, De Dreu, and Nijstad (in press) propose that inaddition to positive and negative mood, level of activa-tion should be examined in an effort to better under-stand the relationship between mood and creativity.Specifically, they offer evidence that activating moodstates (i.e., anger, fear, happiness) versus deactivatingmood states (i.e., calm, relaxed, sad, depressed) can dif-ferentially affect creative performance. In their dualpathway model of creativity (De Dreu et al., 2008), acti-vating moods that are negative in hedonic tone arebelieved to enhance creative fluency and originalitythrough enhanced perseverance. In contrast, activatingmoods that are positive in hedonic tone enhance cre-ative fluency and originality through enhanced cogni-tive flexibility. In this research, we focus on a creativetask that required concentration, high detail orienta-tion, and originality (Amabile, 1996), and we manipu-lated high-arousal, activating emotional states.

Person Situation Model of Artistic CreativityResearch suggests that creativity can be explained

partly by personality characteristics but also by situa-tional variables related to changing or enhancing affec-tive states. A third option considers the interaction ofperson and situational variables, as evidenced through theclassic interactionism approach (Endler & Magnusson,1976). This approach suggests that situational factorscan moderate the effect of person factors and has beenused in a wide variety of socialpsychological phenom-ena (see Mischel, 2004). In this study, we examined bio-logical products linked to depressionDHEASandmanipulated emotional states to test their combinedeffects on artistic creativity. We anticipated that engen-dering high-arousal negative emotions would bringabout increased artistic creativity and that this effectwould be exacerbated among those with lower levels ofDHEASan index of affective vulnerability.

Study OverviewThis experiment engendered high-arousal emotional

states by exposing participants to a laboratory task(Trier Social Stress Task; Kirschbaum, Pirke, &Hellhammer, 1993) designed to elicit strong and endur-ing positive or negative emotional responses. We chosesocial feedback as our experimental manipulation

1678 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN


because decades of socialpsychological research hasshown that social approval, versus rejection, differen-tially affects mood, self-esteem, behavior, and physiol-ogy (Crocker, Cornwell, & Major, 1993; Dickerson,Gruenewald, & Kemeny, 2004; Leary et al., 2003;Mendes, Major, McCoy, & Blascovich, 2008).

During the initial task, participants received eitherexplicit positive or negative feedback from a panelof interviewers during a mock job interview. Neuro-endocrine responses (DHEAS) were assessed at thebeginning of the experiment, and self-reported emotionswere assessed throughout the experiment. Artistic cre-ativity was assessed prior to and following the manipu-lation of emotional states.

There were three primary predictions that we tested.First, the affective vulnerability prediction tested the ideathat participants who were biologically vulnerableasindexed by lower DHEAS levelswould experiencegreater increases in negative affect after receiving reject-ing social feedback. Second, we predicted that partici-pants receiving social rejection would produce the mostcreative artistic products. Finally, we predicted a person situation interaction such that biological products(DHEAS) would interact with situational triggers, elic-iting high-arousal negative affect. Specifically, weexpected that participants who were lower in DHEASand who received rejecting social feedback would pro-duce the most creative artistic products.

METHOD

ParticipantsWe recruited 96 young adults (65 females) to take

part in a 2-hour study on physiological responses duringvarious laboratory tasks. Participants were recruitedvia newspaper advertisements in the Boston area speci-fying adults between the ages of 18 and 25. We pre-screened prospective participants for general healthconditions and provided study day instructions thatwere intended to reduce factors that would influenceneuroendocrine products (e.g., consuming dairy prod-ucts with live cultures, use of caffeine, recent exercise,etc.; Kirschbaum & Hellhammer, 1994).

ProcedureEach participant arrived at the lab during afternoon

hours and, after a 30-minute rest period, provided asaliva sample that we later assayed for DHEAS. Salivawas obtained in sterile tubes using the passive droolmethod, which requires participants to expectorate intoa cryovial via a plastic straw. Upon completion of thestudy, saliva samples were stored immediately at 80 C

until they were shipped overnight on dry ice to a labo-ratory in College Park, Pennsylvania. Saliva samples wereassayed for DHEAS using a highly sensitive enzymeimmunoassay (Salimetrics).1 The test used 50 ul ofsaliva per determination and has a lower limit of sensi-tivity of 10 pg/mL and a range of standard curve from10.2 to 1,000 pg/mL; its average intra- and interassaycoefficients of variation are 4.9% and 3.45%, respec-tively. Five samples could not be assayed because ofeither blood contamination or not enough saliva.

Baseline creativity. After the saliva sample wasobtained, participants completed a creativity task, theAbbreviated Torrance Test for Adults (ATTA; Goff &Torrance, 2002), which was intended to serve as a base-line level of creativity. Participants were given 3 minutesto complete the task, which consisted of nine triangleson a page, and they were instructed to use the trianglesto make some pictures that were unusual andinteresting and to give each picture a title along acommon theme.

Social evaluation task. Following the baseline cre-ativity task, participants were instructed that theywould be preparing and delivering an 8-minute speechfollowed by a 5-minute question-and-answer (Q&A)period in a mock job interview. Participants were ran-domly assigned to one of three evaluative conditions. Inthe two experimental conditions, the participants wereinstructed that they would be delivering the speech totwo evaluators; in the control condition, the partici-pants were instructed they would be delivering thespeech alone in the room. Once the participants con-sented to this part of the study, for those in the experi-mental conditions two experimenters (one male, onefemale) entered the room to reiterate the task instruc-tions. For those in the control condition, the experi-menter repeated the instructions. Following theinstructions, all participants were left alone for 2 min-utes to prepare for the speech.

For those in the experimental conditions, after thepreparation period the evaluators reentered the roomand participants began the speech. At this point, theexperimental conditions diverged into either the socialapproval or social rejection conditions. The role of theevaluators was carefully scripted, coordinated, andtimed so that all participants had a consistent experi-ence. Accordingly, the evaluators were trained to pre-sent themselves with confidence and assumed authority.The social approval condition consisted of the evalua-tors giving explicit positive feedback (e.g., You arevery clear and manage to put your personality across.You are very self-assured and authentic, really greatjob) as well as exhibiting positive nonverbal behavior

Akinola, Mendes / DARK SIDE OF CREATIVITY 1679


during the participants speech, for example, smiling,nodding, and leaning forward. In contrast, the socialrejection condition consisted of evaluators shakingtheir heads, frowning, and giving explicit negative feed-back during the speech (e.g., I felt that you could bemuch clearer and more articulate. Think about whatyou are saying before you say it).

Participants in the control condition were instructedby the experimenter to give a speech while alone in theroom and then to answer questions that appeared onindex cards. The control condition was designed torequire similar metabolic demands associated with speak-ing but not engender any specific strong emotional state.

Self-report measures. We assessed demand andresources appraisals, emotional states, and participantsperceptions of how the interviewers evaluated them.Immediately following the speech, participants werequeried regarding their resource appraisals of the situa-tion (e.g., I had the abilities to perform well on the task;Mendes, Gray, Mendoza-Denton, Major, & Epel,2007). Participants rated their agreement with each sen-tence on a 1 (strongly disagree) to 7 (strongly agree)scale. Self-reported emotions were assessed both priorto and after the speech task using the Positive andNegative Affect Schedule (Watson, Clark, & Tellegen,1988), which measures high-arousal, activated emo-tions. Participants rated their feelings on 20 emotionalstates (10 positive and 10 negative) using 5-point scalesranging from 1 (not at all) to 5 (a great deal). Positiveand negative emotion scales were calculated for eachtime point and had high reliability (alphas ranged from.85 to .91). After both the speech and the Q&A tasks,participants assigned to the social evaluation conditionsrated how well they believed each of the interviewersthought they performed (e.g., She thought I performedwell on the task). Responses for the male and femalejudges were highly correlated ( = .91), so we combinedthese responses into a single score.

Creativity task. After the speech task, participantswere given 10 minutes to complete a second artistic cre-ativity task.2 Following procedures outlined by Amabile(1982), participants were given a 10--15-inch pieceof cardboard, a bottle of glue, a bottle of glitter, and54 pieces of felt and paper in various shapes, sizes, andcolors and were told to make a collage on the card-board. Participants were given the following specificinstructions:

Feel free to use any of the materials that you like, but youdont have to use any that you dont want toit is entirelyup to you. You are not required to use anything in partic-ular; just make a collage that you find interesting.

This test of creativity has been demonstrated to yieldreliable assessments of artistic creativity (Amabile,1982). Following this task, participants were debriefedand thanked.

Creativity AssessmentWe recruited local artists to judge the creative prod-

ucts generated by the participants following the consen-sual assessment technique guidelines (Amabile, 1982).Eight artists (four professional artists and four graduatestudents in studio arts) with an average of 10 years ofstudio art experience were scheduled one at a time tojudge each collage, and a subset of these artists judgedthe baseline creativity task (the ATTA).3

Prior to evaluating each collage, the judges wereasked to create their own collages and were given thesame instructions as participants. They then proceededto judge participants collages. Each collage was ratedon 21 dimensions, which were assessed by having thejudges mark an X on an 18 mm line anchored from lowto high with a midpoint labeled medium. Consistentwith the consensual assessment technique, judges wereasked to use their own subjective definition in ratingeach of the dimensions.

We performed exploratory factor analysis with vari-max rotation on the 21 dimensions and found that 15 ofthe 21 dimensions loaded high on one factor (factor load-ings > .38).4 Interrater reliabilities were acceptable (alphasranged from .65 to .88). Therefore, we created a singleindex of collage creativity per participant, averagedacross judges and across the 15 dimensions, which formsour primary dependent variable of creativity ( = .96).

RESULTS

Manipulation ChecksWe first examined if the social feedback manipula-

tions were perceived as intended. There was no ambi-guity regarding the manipulated conditions of socialrejection and approval. Participants in the socialapproval condition perceived the evaluators as likingtheir interviews more than those in the social rejectioncondition, F(1, 44) = 26.33, p < .0001 (Table 1). Socialapproval also resulted in participants perceiving them-selves as having more resources than those in the rejec-tion condition, F(1, 88) = 4.20, p < .05. We created anindex of emotional responses (negative emotion posi-tive emotion) to provide a single index of negative affectwithout the buffering effects of positive emotion. Thenegative affect index differed by condition, with thoseassigned to social rejection reporting less positive and morenegative emotion (controlling for baseline emotional

1680 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN


responses) than those assigned to social approval, F(1,85) = 4.54, p < .04. Given these findings, we were con-fident that we successfully manipulated social rejectionand approval.

As an initial examination of our predictions, weexplored zero-order and first-order partial correlationsamong our primary dependent variables by condition(Table 2). The partial correlations include pretask nega-tive affect as a control variable to best assess changes inemotional responses as a result of the manipulatedsocial setting. Significant relationships were foundamong DHEAS, creativity, and negative emotionalresponses, most notably from the social rejection condi-tion. These relationships are explored below, organizedby the specific predictions.

Affective VulnerabilityOur first prediction was that lower DHEAS would

bring about greater increases in negative affect whenparticipants received rejecting social feedback.However, we anticipated that this vulnerability wouldnot be related to changes in negative affect followingsocial approval or the nonsocial condition. To establishthis initial effect, we conducted hierarchical regressionanalyses in which the first step included baselineDHEAS, condition effects (two effect-coded predictorvariables to represent the three levels of evaluation; wecoded social rejection as the reference condition), andpretask emotions as a covariate to predict the poststres-sor negative affect index. This first step produced anoverall significant model (R2 = .487, p < .0001). Thesecond step included all the initial predictors plus theinteraction terms (condition by DHEAS). As expected,the inclusion of the interaction terms significantlyincreased model fit (R2 = .041, p < .04). Supporting theaffective vulnerability prediction, following rejectingsocial feedback the lower the participants DHEAS, thegreater their negative emotional reactions (b = .07, p /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox false /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (U.S. Web Coated \050SWOP\051 v2) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown

/CreateJDFFile false /SyntheticBoldness 1.000000 /Description >>> setdistillerparams> setpagedevice

creativity and depression

Documents