seeing red: affect modulation and chromatic color

Seeing red:Affect modulation and chromaticcolor responses on the Rorschach

Johanna C. Malone, PhDMichelle B. Stein, PhDJenelle Slavin-Mulford, PhDIruma Bello, PhDS. Justin Sinclair, PhDMark A. Biais, PsyD

Psychoanalytic theories suggest that color perception on the Ror-schach relates to affective modulation. However, this idea has mini-mal empirical support. Using a clinical sample, the authors exploredthe cognitive and clinical correlates of Rorschach color determinantsand differences among four affective modulation subtypes: Con-trolled, Balanced, Under-Controlled, and Flooded. Subtypes weredifferentiated by measures of affective regulation, reality testing/con-fusion, and personality traits. Initial support for the relationship ofchromatic color response styles and affective modulation was found.(Bulletin of the Menninger Clinic, 77[1], 70-93)

It has long been realized that there must exist a very close relati-onship between color and affectivity. The gloomy person is oneto whom everything looks "black, " while the cheerful person issaid to see everything through rose-colored glasses. (Rorschach,1921/1942, p. 99)

Colors are not merely recognized, they are felt to be exciting orsoothing, dissonant or harmonious, clamorous and shrill or tran-

Dr. Malone, Dr. Stein, Dr. Sinclair, and Dr. Biais are at Massachusetts CeneralHospital and Harvard Medical School, Boston, Massachusetts. Dr. Slavin-Mulford isat Augusta State University, Department of Psychology, Augusta, Georgia. Dr. Bello isat New York University Medical Center and Bellevue Hospital, New York, New York.Correspondence may be sent to Dr. Malone at Massachusetts General Hospital andHarvard Medical School, Psychological Evaluation and Research Laboratory (PEaRL),One Bowdoin Square, Suite 701, Boston, MA 02114; e-mail: [email protected]. (Copyright © 2013 The Menninger Foundation)

70 Vol. 77, No. 1 (Winter 2013)

Rorschach chromatic color and affect modulation

quil, warm or cool, cheerful or drab, disturbing and distracting orconducive to tranquility and concentration. (Schachtel, 1966, p.161)

The relationship between color perception on the Rorschach andaffect is controversial, with a long history of theoretical support butminimal empirical evidence. Despite this, contemporary Rorschachscoring systems hold that color responses (perception) providean indication of the subject's capacity to modulate affect. Exner(2003) describes affective processes as both a "valuable asset" anda "serious Uability." Ideally, individuals harness their emotionsto guide thoughts, relationships, and behaviors. However, affectregulation may range from extremes of constricted expression toflooded dysregulation. On the Rorschach, individuals are thoughtto experientially demonstrate their affective style as they are movedby and in turn integrate color into their responses. As opposed toother measures of self-reported emotional dysregulation, the pro-cess of affect modulation observed on the Rorschach may have thepotential to be more nuanced by elucidating an implicit processbeyond what a patient can articulate through immediate self-re-flection. The goal of the present study is first to present the theoryand rationale behind the linking of color-form perception to affectmodulation; second, to summarize the extant empirical literatureon this subject; and third, to empirically investigate the relation-ship between chromatic color determinants and clinical correlates,including developmental history, neuropsychological functioning,and self-reported psychological functioning.

The most recent scoring systems for examining Rorschachchromatic color responses utilize the following determinants:Eorm Color (FC), Color Eorm (CE), and Color (C)' (see Exner,2003; Meyer, Viglione, Mihura, Erard, & Erdberg, 2011). Affec-tive modulation is thought to be most controlled (EC response)when the form in the responses takes a primary role with color(affect) used to accurately support the description (e.g., "7i lookslike a frog—it's a green frog with legs, eyes, body, and mouth.").In contrast, a CE response is dominated by color but still draws

1. The present article does not discuss Color Naming, (Cn) or Color Projection (CP),both of which utilize chromatic color.

Vol. 77, No. 1 (Winter 2013) 71

Malone et al.

upon form (e.g., "a beautiful rosebush—the colors of the red andpink petals with the green stem and leaves underneath'^).Whenform is absent (e.g., pure C), it is believed that emotional respon-sivity due to the color stimulus has left the person without theability to structure and organize his or her thinking (e.g., "blood,it's all red and bloody").

Theory

The theoretical link of color perception to affect has been de-veloped across psychoanalytic theoretical schools with founda-tions in a phenomenological approach (Schachtel, 1966) and egopsychology and drive theory (Rapaport, Gill, & Schäfer, 1968).Schachtel (1966) outlined three core features that link colorperception and affective experience. First, he described passivi-ty, stating, "Color seizes the eye, but the eye grasps for form"(Schachtel, 1966, p. 160). He paralleled color perception to thatof emotional experience, which can sweep over a person with-out the individual's active pursuit. Next, Schachtel emphasizedthe common features of color perception and affect through im-mediacy, giving the example of the neon street sign that drawsthe individual's attention almost by force. He stated, "the eye isassailed by color whether it wants to or not" (Schachtel, 1966,p. 160). He identified feeling tone or mood quality as a final el-ement shared by color perception and affectivity. For this lastpoint, Schachtel identified cultural meanings linked to certain col-ors or anecdotally referred to the ways in which interior decora-tors speak to how colors may be linked to the feeling of a room.Rorschach (1921/1942) himself also held that feeling tone was acentral link between affect and color. Overall, Schachtel proposeda theoretical vantage point that emphasized how the experienceof color perception and affect have commonalities related not tospecific feelings but more to how the person is vulnerable to thepositive and negative effects of the impact of each domain.

Using a drive theory perspective, Rapaport et al. (1968) ex-panded on this work not only by commenting on the process ofaffective experience but also by providing an elaborate theory for

72 Bulletin of the Menninger Clinic


bow tbe Rorscbacb assessed affect modulation tbrougb the waysin wbicb form quality was integrated into tbe color response.As tbe responses became less and less form dominated, tbe in-dividual was seen as baving less and less regulatory control overtbe experience of affect. Rapaport and colleagues (1968) positedtbat color triggers impulsive responsiveness as part of a need forinstinctual release of tension or affective discbarge. In explainingtbe process of affect modulation, they suggested tbat tbe way anindividual incorporated form into tbe color response representedthe ability to "delay" and tberefore adapt to bave a more flexiblecontrol tbat integrated more tbougbt and structure. Tbis wouldtranslate into greater affect modulation, being cbaracterized byprofiles witb more ECs tban CEs and Cs.

Utilizing tbese tbeories, conceptualizations were made regard-ing bow profiles cbaracterized by each of tbe tbree cbromaticcolor determinants would translate into tbe presentations of pa-tients in clinical settings. EC determinants were expected to bebigbest in "neurotics" and reflect capacity for "affective rapport"and "emotional adaptation" (Rapaport et al., 1968, p. 380). Ror-scbacb described tbese responses as tbe "successful integrationof color and form," bigbligbting tbe value of being able to con-trol and modulate emotional experiences. In contrast, tbe CE wastbougbt to be more common in individuals witb "vivid unfetteredaffectivity" in wbicb tbere was less restraint and capability formanaging affect (Exner, 2003; Rapaport et al., 1968). Rapaportet al. suggested tbat tbese responses would be most common in"unclassified scbizopbrenics" and "bysterics." Overton (2000)articulated tbe difference between CE and EC by suggesting tbattbe EC is like being in love as compared to tbe CE, wbicb is likefalling in love. Finally, tbe pure C determinant, response witb noform wbatsoever, was described by Rapaport and colleagues astbe "abandonment of control over affects and actions" (p. 386)tbat would be found at tbe bigbest levels in scbizopbrenia. Draw-ing on tbe work of Scbacbtel, tbe pure C response bigbligbts tbepassivity of tbe experience because tbere may be an inappropri-ateness of tbe flattening of affect.

Vol. 77, No. 1 (Winter 2013) 73

Malone et al.

Research

As presented above, there is a well-developed psychoanalytic the-oretical basis linking the relationship of chromatic color determi-nants to affective adaptation. However, to date there is minimalempirical evidence to support this claim. For example, using asample of undergraduate students, Stevens and colleagues (1993)compared subgroups of high-affect modulators (high FC and lowCF-HC; n = 6) and low-affect modulators (low FC and high CF+C;n = 10), with "high" and "low" defined by median values of thedeterminants in the sample. Although a small sample was used,this research found no support for the affect modulation hypoth-esis using discrepancy scores between a person's perception of hisor her affect and a coder's perception of the person's affect in avideotaped discussion. Frank (1976, 1993) reviewed the litera-ture and asserted that there was no empirical foundation linkingcolor to affect and speculated that the relationship has been false-ly perpetuated by cultural lore. He (1993) goes as far as to state,"clinicians should not interpret persons' responses to color asrepresentative of the vicissitudes of their emotional life; the datado not justify such an interpretation" (p. 13). However, despitethis assertion, more recently several studies have linked increasedcolor responses to personality disorders characterized by affectivedysregulation. Specifically, Biais, Hilsenroth, and Fowler (1998)found that the sum of color determinants was significantly posi-tively correlated with not only specific symptoms of histrionicpersonality disorder but also the number of symptoms of this dis-order. In another study, Zodan, Charnas, and Hilsenroth (2009)found that the weighted sum of color responses could differenti-ate between those with and without a borderline personality dis-order diagnosis. Couples engaged in custody disputes have alsobeen found to have higher rates of domestic violence when theyhave greater levels of C and less FC, suggesting less control overstrong affects (Miller, 1999; cf. Exner, 2003). When studying thecolor determinants in relationship to psychotherapy, Exner andcolleagues (Exner, 1978; Exner &c Murillo, 1975; Exner, Murillo,& Cannavo, 1973) found that "successful" posttreatment out-comes were associated with color responses shifting to be moredominant by form than by color (a shift in the FC:CF-HC ratio). In

74 Bultetin of the Menninger Ctinic


an older study, Weiner (1964) found tbat pure C responses couldbelp in differentiating tbose witb and witbout scbizopbrenia.

Tbis bandful of studies suggests tbat tbere may be value infurtber testing tbe relationsbip between chromatic color and af-fective modulation. Wbile emotional regulation can be assessedtbrougb self-report and observations in clinical interview, tbevalue of using a performance-based measure is to assess aspectsof a patient's emotional world tbat may not be readily accessiblefor tbe patient to articulate. Tberefore, drawing upon tbe existingtbeoretical foundation and long bistory of application of colordeterminants to understand affect modulation in tbe context ofindividual assessment protocols, tbis study sougbt to empiricallyassess tbe relationsbip of cbromatic color determinants to mea-sures of cognition, personality, and psycbopatbology. Tbe firstpart of tbe study used exploratory analyses examining bivariatecorrelations of color determinants treated dimensionally witbbistorical, cognitive, psycbological, and personality variables.On tbe basis of existing tbeory, we expected tbat tbe FC vari-able would be associated witb variables linked to variables acrossdomains tbat are tied to better abilities to modulate, wbile tbeCF and C would be tied to bigber levels of psycbopatbology andmaladaptive personality. In tbe second part of tbe study, a prioribypotbeses were tested regarding tbe differences between Exner'scategorical FC:CF+C affective subgroups (see Table 6) in domainsof personality and psycbopatbology.

Method

ParticipantsParticipants were a clinical sample of 88 individuals (67% male)referred for psycbological testing at a large urban bospital. Aspart of tbis assessment, tbe Rorscbacb was administered in tbecontext of a larger battery and tben scored witb tbe Exner Com-prehensive System (CS). Tbe large majority of tbe participantswere Caucasian (90%), and tbeir ages ranged from 16 to 69 (M= 37.6; SD = 13.8). Most of tbe participants bad never been mar-ried (68.2%). Approximately balf (48%) of tbe participants badgraduated from college, and 56% were unemployed at tbe time

Vol. 77, No. 1 (Winter 2013) 75

Malone et al.

of the assessment. The preassessment diagnoses received from thereferring clinician included 44.3% depressive disorder, 17.0% bi-polar disorder, 13.6% anxiety disorder, and 4.5% psychotic dis-order. Our assessment clinic maintains an Institutional ReviewBoard (IRB)-approved deidentified data repository containing allassessment data and extensive demographic information for allevaluations conducted. The assessment and demographic data re-ported in this study were obtained from this IRB-approved datarepository.

MeasuresPEaRL Semi-structured Clinical Interview (PEaRL-SSCI). The

PEaRL-SSCI (Biais, 2011) aims to systematically capture clini-cally relevant information regarding a patient's current function-ing along with his or her psychiatric, developmental, educational,relational, occupational, medical, and legal history. Data fromthis measure have demonstrated interrater reliability. In a previ-ous study (Slavin-Mulford et al., 2012), the interrater reUabilityof the developmental history variables has been strong, rangingfrom .61 (past psychiatric hospitalization) to .85 (for both suicideattempt and education [education was coded categorically: <12years, 12 years, >12 years, college graduate, and >16 years]). Allparticipants in the present study completed this interview as partof the standard assessment protocol.

Rorschach Ink Blot Test. The Rorschach (Rorschach,1921/1942) is a performance-based measure used to assess per-sonality, psychopathology, and psychological functioning. In thisstudy, all participants completed the Rorschach, which was ad-ministered and scored using the Exner Comprehensive System(2003). The Rorschach protocols were scored and administeredby six different individuals, and two licensed clinical supervisorsreviewed the scoring of predoctoral interns and postdoctoral fel-lows. The use of multiple examiners and supervisors reduces therisk that a single individual could have systematically influencedthe Rorschach scores. Rorschach protocols used in this study hadmore than 14 responses and no Card rejections. Only the threechromatic color determinants (FC, CF, C) are used in the analysesin the present article.



Personality Assessment Inventory (PAI). The PAI (Morey,1991, 2007) is a self-report multiscale personality measure usedto assess psychopathology, substance abuse, treatment-related is-sues, and interpersonal style. It contains 344 items comprising 22nonoverlapping scales: 4 validity scales, 11 clinical scales, 5 treat-ment scales, and 2 interpersonal scales. Norms for the PAI weredeveloped from three separate samples: a census-matched sample(N = 1,000), a college sample (N = 1,051), and a clinical sample(N = 1,246). In this study, we examined the following 13 PAIfull scales: Somatic Complaints (SOM), Anxiety (ANX), Anxi-ety Related Disorders (ARD), Depression (DEP), Mania (MAN),Paranoid (PAR), Schizophrenia (SCZ), Borderline (BOR), Anti-social (ANT), Alcohol Problems (ALC), Drug Problems (DRG),Suicidal Ideation (SUI), and Aggression (AGG). Ten of these PAIscales also contain subscales (ALC, DRG, and SUI do not havesubscales) that were also explored in the study. Of the total sam-ple in this study, 85 participants completed the PAI.

NEO Five Factor Inventory (NEO-FFI). The NEO-EFI (Costa&C McCrae, 1992; McCrae & Costa, 2004) is a 60-item person-ality inventory that yields five factors: Extroversion, Agreeable-ness, Openness to Experience, Neuroticism, and Conscientious-ness. Participants rated how true a statement was for them ona five-point Likert-type scale ranging from "strongly agree" to"strongly disagree." Of the total sample in this study, 77 partici-pants completed the PAI.

Performance-based neurocognitive abilitiesThe Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler,1999) was used to assess Full Scale, Verbal, and Performance IQ.Attention, Concentration, and processing speed were assessedwith Trail Making Test Part A (Monsch et al., 1992; Reitan,1958; Reitan & Wolfson, 1985) and the Digit-Coding Scale ofthe Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV;Wechsler, 2008). The Wechsler Memory Scale-Fourth Edition(WMS-IV; Wechsler, 2009) and the Repeatable Battery for theAssessment of Neuropsychological Status (RBANS; Randolph,1998) were used to assess logical memory (i.e., informationplaced in the context of a story both in terms of immediate recall

Vol. 77, No. 1 (Winter 2013) 77

Malone et al.

and recall of information following a delay). Measures of execu-tive functioning included Trail Making Test Part B and Stroop(Trenerry, Crosson, Deboe, &c Leber, 1989). Tbese tests botbmeasure the ability to inbibit responses.

ProceduresAll patients were evaluated under standard clinical procedures,and tbe Rorscbacb was included as part of tbe assessment proto-col. Participants were referred for psychological or neuropsycbo-logical assessments by a treating professional. Tbe assessmentswere conducted by licensed psycbologists or by postdoctoralfellows and predoctoral psycbology interns under supervision.All assessment evaluations included a semistructured clinical in-terview and tbe Rorscbacb. Additional measures were includedbased on tbe protocol from tbe specific assessment. After eacbassessment, approximately 45 nontest variables (obtained fromtbe semistructured interview or medical record) and all psycbo-logical and neuropsycbological test scores were entered into tbedata repository.

Data analysesIn tbe first set of data analyses, we treated tbe cbromatic colordeterminants dimensionally. We used bivariate correlations inexploratory analyses to identify tbe empirical correlates of tbetbree cbromatic color determinants in tbe domains of life bistory,neuropsycbological functioning, and self-reported psycbologicalfunctioning.

In tbe second set of analyses, we created categorical subgroupsusing different ratios of tbe EC:CE+C determinants outlined byExner (2003; see Table 6). On tbe basis of Exner's (2003) descrip-tions, researcb, and tbeoretical work (Overton, 2000; Rapaportet al., 1968; Scbacbtel, 1966), we developed a priori bypotbesesregarding tbe relationsbip of tbese variables to domains of af-fect regulation (PAI BOR subscales and SUI), reality testing (PAISCZ subscales and PAR subscales), and NEO personality traits.In tbese analyses, scores on domains on tbese subscales werecontinuous variables, and affect modulation classification was acategorical variable witb four levels (controlled, balanced, under-



Table 1. Means, standard deviations, skewness, and kurtosis for color determinants(N=88)

Color determinant M SD Range Skewness Kurtosis

Current sample

FC

CF

C

1.49

1.36

0.27

1.47

1.40

0.74

0-6

0-5

0-4

1.08

0.89

3.54

0.82

-0.04

14.11

controlled, and flooded). Using multinomial logistic regression,we examined whether self-reported variables in these domainsincreased the likelihood of belonging to a subgroup with moreproblematic affect modulation. The logistic model interprets re-gression coefficients by creating an odds ratio, a ratio of the prob-ability of whether a person is a case to the probability that aperson is a noncase (Cohen, Cohen, West, &c Aiken, 2003).

Results

Dimensional correlationsDescriptive statistics for chromatic color determinants in the cur-rent sample are presented in Table 1. These results can be com-pared to previous clinical and nonclinical samples in Table 2. AsTable 1 shows, both FC and CF were normally distributed but Chad a nonparametric distribution. The nonnormal distributionof C is consistent with findings from both clinical (Exner, 2001)and nonclinical samples (Exner & Erdberg, 2005). We furtherevaluated our data using a box-and-whiskers plot as suggested byTukey (1977). The results of this analysis indicated that all scoresgreater than zero, that is, any case with a single C response, wereoutliers (falling above the 1.5 interquartile range) as such anytransformation of the distribution of C risks profoundly alteringthe meaning of the variable. Therefore, given that our findingsregarding the nature of C's distribution is consistent with past re-search and that our planned comparisons involve forming groups(categories) based upon configurations of the Color variables, wechose not to normalize/transform the distribution of C.

Vol. 77, No. 1 (Winter 2013) 79

Malone et al.

Table 2. Descriptive statistics (means and standard deviations) for color variables forpatient and nonpatient samples

FC CF C N

Inpatient schizophrenia

Inpatient depression

Character disorder

Nonpatients

1.54/1.60

1.58/1.95

0.98/1.29

2.97/1.78

1.24/1.38

1.58/1.38

0.86/1.08

2.80/1.64

0.42/0.72

0.72/0.99

0.47/0.86

0.17/0.45

320»

315»

180»

450''

Note. 'From Exner (2001). 'From Exner and Erdberg (2005).

As part of exploratory analyses, tbe bivariate correlations wereexamined witb criteria from clinical interviews, neuropsycbologi-cal tests, PAI full scale and subscale scores, and NEO-FFI traits(Tables 3-7). Correlations are only reported wben coefficientsreacbed statistical significance (p < .05) or tbe level of a trend{p < .10). Wben examining tbe cbromatic color determinants inrelation to data gatbered from tbe clinical interview (Table 3),we found tbat FC was negatively correlated witb markers of ag-gression, psycbosis, and bead injuries, wbile CF was positivelycorrelated witb bistories of depressed affect and suicidal ideation.

An examination of tbe neuropsycbological variables listed inTable 4 revealed no significant correlations between tbe cbromat-ic color determinants and bigber order cognitive abilities (FullScale, Verbal, and Performance IQ) or measures of executivefunctioning. However, tbe FC determinant was correlated witbbetter scores on a range of attention, concentration, and mem-ory tests. Tbese are intriguing findings because attentional andmemory processes are considered to be important components ofemotion regulation (Ocbsner & Gross, 2005).

In contrast to tbe results from neuropsycbological data, FCwas unrelated to full scales on tbe PAI (Table 5). Instead, CFwas correlated witb a broad range of psycbological difficulties,including suicidal ideation. In contrast to CF, pure C was relatedless to anxiety and stress and more to social isolation, but stillbad elevated depression and scbizopbrenia subscales. Data fromtbe PAI subscales (Table 6) revealed a more nuanced and complexpattern of correlations. Pure C was associated witb conversionsymptoms. It is notable tbat bigber pure C was also related tobigber endorsement of pain in tbe clinical interview. Pure C was

80 Buttetin of the Menninger Ctinic


Table 3. Correlations of chromatic color determinants with datafrom clinical interview (N = 88)

FC CF C

History of suicide ideation

History of homicide attempts

History of auditory hallucinations

History of depression

Patient rating current depression

Patient rating current confusion

Patient rating current pain

History of mania

Number of previous psychiatric hospitalizations

History of arrests

Head injury

-.22»

-.26»

-.19+

-.29»»

-.21»

.30»»

.22»

.36»

.59»»

.21+

.34+

Note, -fp < .10, 'p < .05; "p < .01.

also related to aspects of paranoia (hypervigilance and resent-ment) and depressive experience. Higher C was also tied to great-er social isolation. In examining the PAI subscales, we found thatCE continued to be tied to a broad range of psychopathology, in-cluding anxiety, depression, confusions, endorsement of psychot-ic experience, and dysregulated affect. Both the Suicide PotentialIndex and the Violence Potential Index (composite variables ofrisk factors) also showed noteworthy correlations to more color-dominated responses.

The associations between NEO traits and the color chromaticdeterminants (Table 7) revealed that, consistent with the findingson the PAI, CE was positively correlated with neuroticism. PureC was associated with being more closed to experience and intro-verted, while the FC determinant was positively correlated withopenness to new experience.

Categorical subgroupsHaving identified patterns of correlations using the exploratoryanalyses, we next sought to examine the chromatic color determi-nants using the subtypes defined by the ratio of FC:CF+C. Exner(2003) and Overton (2000) outline the ways in which it is not the

Vol. 77, No. 1 (Winter 2013) 81

jlmih

Highlight

jlmih

Highlight

jlmih

Highlight

Malone et al.

Table 4. Neuropsychologica

Gognitive-WASI

Full-scale IQ

Verbal IQ

Performance IQ

Attention

Digit symbol coding

Trails A

Memory

Logical memory immediate

Logical memory delayed

Executive functioning

Stroop

Trails B

1 testing means, standard deviations,chromatic color determinants

Mean (SD)

106.2 (16.0)

107.3 (15.2)

102.8 (16.0)

91.13 (17.2)

94.8 (14.5)

88.1 (19.8)

91.6 (20.2)

98.6 (16.4)

87.7 (19.9)

N

87

85

85

75

56

75

75

86

55

and correlations with

Correlations withcolor determinants

FG GF G

.28»

.23+

.32»

.36»

Noie. +p < .10, °p < .05.

determinant alone but the ratio of the determinants that deter-mines the style of affective modulation. A color ratio/subgroupapproach to studying the chromatic color determinants has alsobeen utilized in research studying cognitive processes of objectrecognition (Kron, Cohen, Benzimana, &c Ben-Shakhara, 2009).Exner (2003) proposes 10 subtypes (see Table 8) based on theratio of color determinants. On the basis of the distribution ofpeople across the 10 subtypes, we further collapsed the subtypesinto five groups (see Table 9): Emotional Control, Balanced, Ve-neered Under-Control, Under Control, and Flooded. Descrip-tions of these groups and their incorporation of the FC:CF+Cratio are included in Table 9.

Exner's subtypes: Multinomial logistic regressionThree stepwise multinomial logistic regression models were con-ducted to explore whether there were differences in the affec-tive modulation groups, based on (1) difficulties with emotionalregulation (PAI BOR subscales and SUI scale); (2) reality testing

82 Bultetin of the Menninger Clinic.


Table 5. PAI full-scale means, standard deviations, and correlations with chromaticcolor determinants (N= 85)

Mean {SD) Correlations with color determinants

Anxiety (ANX)

Anxiety-related Dx (ARD)

Depression (DEP)

Mania (MAN)

Schizophrenia (SCZ)

Borderline (BOR)

Suicidal ideation (SUI)

Stress (STR)

Nonsupport (NON)

T-score

65.1 (13.0)

60.6 (14.4)

69.8 (15.4)

51.4(10.5)

61.9 (13.0)

63.2(11.5)

65.9 (19.4)

61.0(12.9)

58.4(14.3)

FC CF

.25»

.26»

.24»

.20+

.29»*

.19+

.35»»

.25*

.18+

C

.25*

.26*

.23*

.27»

Note. +p < .10, 'p < .05; " p < .01.

and confusion scores (PAI SCZ and PAR subscales); and (3) nor-mative personality traits (NEO) (see Table 10). Tbese domainswere cbosen based on tbe extant tbeory, researcb, and resultsfrom exploratory analyses. In tbese analyses, we did not utilizetbe veneered under-control group because tbere was only one in-dividual in tbis group. Tbe emotional control group was used asa comparison group for tbe tbree otber affect modulation groups(i.e., balanced, under-control, and flooded).

In tbe first model, tbe four PAI Borderline Subscales (i.e., nega-tive relationsbips, identity problems, self-barm, and affective dys-regulation) and tbe Suicidal Ideation (SUI) scale were entered intbe model. Tbe goodness of fit cbi-square statistic indicated themodel was well-fitting (% (̂240) = 264.90, ns; Nagelkerke pseudoR-square = .11). Tbe results indicated tbat if an individual's scoreon tbe suicide scale was bigber, tbe person bad a greater likeli-bood of being categorized as under-controlled and flooded as op-posed to emotionally controlled.

In tbe second model, tbe PAI Scbizopbrenia subscales (i.e., psy-cbotic experience, social detachment, tbougbt disorder) and PAIParanoia subscales (i.e., bypervigilance, persecution, resentment)were entered in tbe model. Tbe goodness of fit cbi-square statis-tic indicated tbe model was well-fitting (x^(225) = 225.07, ns;Nagelkerke pseudo R-square = .30). Here results again indicated

Vol. 77, No. 1 (Winter 2013) 83

jlmih

Highlight

jlmih

Highlight

Malone et al.

Table 6. PAI subscale means, standard deviations, and correlations with chromaticcolor determinants (N= 85)

Mean (SD) Correlations with color determinants

T-score FC CF C

SOM-C Conversion

SOM-H Health Concerns

Anx-Cognitive

Anx-Affective

ARD-Obsessive-Compulsive

ARD-Trauma

DEP-Cognitive

DEP-Affective

DEP-Physical

PAR-Hypervigilance

PAR-Resentment

SCZ-Psychotic Experience

SCZ-Social Detachment

SCZ-Thought Disorder

BOR-Affective Instability

Suicide Potential Index

Violence Potential Index

Treatment Process Index

Mean Clinical Index

56.6 (14.3)

58.2(13.1)

64.6 (13.7)

64.9 (13.2)

52.2(12.1)

63.2(15.1)

69.7 (15.7)

70.8 (15.7)

60.4 (13.1)

56.7(11.5)

57.8 (12.9)

50.3 (14.1)

60.7(12.3)

64.2 (14.7)

63.3 (11.7)

65.7(13.9)

56.5 (12.8)

59.1 (15.3)

59.2 (8.0)

-.24»

-.22»

.22»

-.26»

.20+

.32»»»

.31»»»

.28»

.25»

.33»»»

.24»

.27»

.22».43»»»

.25*

.27*

.25*

.30»»»

.19+

.18+

.22»

.35»»»

.26»

.27»

Noie. +p < .10, *p < .05, **p < .01, "*p< .001.

that if an individual endorsed more psychotic experience, he orshe had a higher likelihood of being classified as both under-con-trolled and flooded as opposed to controlled.

In the final stepwise multinomial logistic regression, the fiveNEO traits were entered into the model. The goodness of fit chi-square statistic indicated the model was well-fitting (^^(207) =218.56, ns; Nagelkerke pseudo R-square = .28). The results re-vealed that if an individual was higher in neuroticism, there wasa greater likelihood of being under-controlled (only a trend forflooded) as compared to emotionally controlled. In contrast, in-dividuals higher in openness were less likely to be flooded thanemotionally controlled.

84 Buttetin of the Menninger Ctinic


Table 7. NEO-FFI means, standard deviations, and correlations with chromatic colordeterminants {N = 77)

Mean {SD) Correlations with color determinants

T-score FC CF

Neuroticism 63.8 (13.3) .35*

Extraversion 40.4 (13.4) -.28*

Openness 54.9 (11.2) .24* -.25*

Agreeableness 50.2 (13.5)

Conscientiousness 39.2 (14.9)

Noie, "p < .05.

Discussion

Tbe link between cbromatic color and affect on tbe Rorscbacb isconsidered by some to be based on nostalgia for past tbeoreticalmodels and cultural lore ratber tban empirical data. If tbis is true,it is of great concern because tbe ratio between cbromatic colordeterminants on tbe Rorscbacb is routinely used in tbe context ofa full psychological assessment battery to inform tbe clinician'sunderstanding of bis or ber patient.

In tbe present study, we sougbt to consider tbis topic again witbfresb eyes, wbile remaining informed by past tbeoretical modelsand tbe limited body of empirical researcb. Overall, our resultsprovide moderate support for the relationsbip between processesof affect modulation and the chromatic color determinants in tbecontext of a clinical setting of patients receiving psycbologicaland neuropsychological assessment.

Wben treated dimensionally, tbe cbromatic color determinantsdifferentially correlated to measures of cognitive, psycbological,and personality functioning in ways tbat are generally consistentwitb existing tbeory. For example, tbe FC determinant was relatedto better control over one's experiences and affective modulationas sbown by stronger neuropsycbological scores on concentrationand memory tests, tbe absence of association to psycbopatbol-ogy, and bigber openness to engage witb experience in tbe world.Interestingly, tbere appears some evidence for wby Rapaport andcolleagues linked tbe CF determinant to "vivid unfettered affec-tivity" because it correlated witb broad-ranging psycbopatbology.

Vol. 77, No. 1 (Winter 2013) 85

Malone et al.

Exner possiblefinding

Pfl

Pf2

Pf3

Pf4

Pf5

Pf6

Pf7

Pf8

Pf9

PflO

Table 8. Exner's FC:CF+C ratio subgroups

N in CurrentSample

• 7

0

23

2

0

1

36

10

7

2

FC:CF+C ratio

C=0 &c FC> CF+C+1& FC<2(CF+C)

C=0 & FC>2(CF+C)&c FC< (CF+C)

C=0; &c FC>3(CF+C)

C=l; & FC> CF+C+1& FC<2(CF+C)

C>1 ÔC FC>2(CF+C)

C>1 & FC>CF+C+1& FC<2(CF+C)

C=0; & FC=CF+C &CF+C>FC+2

C>1 & FC=CF+C &cCF+C>FC+2

C=0 & FC+3<CF+C

C>1 & FC+3<CF+C

Exner description of affectmodulation

Modulates affect well

More stringent controlover emotional dis-charge

Much more overcontrol-ling than most; con-stricted

Modulates well but somelapses

In most cases modulatedaffect well; control isvulnerable to failure;in conflict regardingemotions

Strives to modulate emo-tions effectively butserious lapses inmodulation

Less stringent control,more obvious andintense emotions; notnecessarily bad

Serious modulation prob-lems

Modulates emotions lesswell than others

Significant lapses in modu-lating affect

suicidality, dysregulation, self-harm, aggression, and neuroticism.Einally, pure C's positive correlation with negative affect, para-noia, social isolation, lower extraversion, and openness is con-sistent with Schachtel's description of the flattening of affect andpassivity that occurs when someone is completely overwhelmedand even paralyzed by emotional experience.

In our second set of analyses, we created categorical subgroupsof individuals based on their affective modulation style. Theseresults generally support the traditional Rorschach interpretationthat a Color Ratio emphasizing form-dominated color responses



Table 9. Collapsed affect modulation subgroups

Name P£ group « FG:CF+C ratio Description

Emotionalcontrol

Balanced

Veneered under-controlled

Under-controlled

Flooded

Pf3

P f l , 4

Pf6

Pf7

23

36

Pf 8, 9,10 19

G=0; & FG>3(GF+G) Stringent control;constricted;defensive

FG>GF+G+1&FG<2(GF+G)Balanced ratio;G<1

Balanced ratio; G>1

C=0; &c FC=CF+C&C GF+G>FG+2

Modulates affectwell or modu-lates well butsome lapses

Strives to modulateemotions effec-tively but seri-ous lapsesin modulation

Less stringent con-trol, more obvi-ous and intenseemotions; notnecessarily bad

GF+G levels become Significant lapsesgreater than PF7; in modulatingG>0 affect

(FC) reflects affect modulation or control. Furthermore, theseanalyses revealed that the ratio of chromatic color determinantsmay be seen as a possible risk factor for suicidal ideation andpsychotic experiences. Theoretically this could be understoodas relating to the lack of ability to delay impulsive responses tostrong emotions that become enacted and experienced throughpsychotic-like symptoms or even wanting to die. It highlights theways in which treatments aimed at difficulties modulating affectmay not only reduce suicidality (e.g., dialectical behavior therapy,mentalization-based treatment) but also may have the potentialto diminish psychotic disturbance.

Results examining normative personality traits were also re-vealing, as it appears that individuals higher in neuroticism weremore likely to be in the under-controlled group than in the emo-tionally controlled group. This is consistent with the hypothesesof Rapaport and colleagues, who thought that these individualswould be most likely to be "hysterics." It also suggests that thisgroup experiences affect in ways that likely leave them dysregu-lated and emotionally labile. In contrast, the flooded group was

Vol. 77, No. 1 (Winter 2013) 87

Malone et al.

cohaWo.

Eo(Jo.

3Oh.rol gcou

naloloui;ivith E'ses 1cco

reg

u•a

gis

_o(A1S

trolits.Resu10.Table

o_olie[OJJcoler-cUnd

•a«jc

Exp(b)

,.^xp(l

Mte}

—O.

Ui

Ul

toiable

ISo

IVO00omo00o

+otvOo

I

nrt S

•O

L/5

g Ihot

oy

c

2 O

88B

ulletin of the Menninger

Clinic


differentiated by lower levels of openness and extraversion ascompared to the emotionally controlled group. Clinically, thissuggests a person who, rather than acting out when emotion-ally overwhelmed, instead shuts down and closes off experience.As the ratio of FC:CF+C becomes more heavily weighted on theCF+C side, treatment interventions may greatly differ based onknowledge of how the person may be coping with the affectiveintrusions in his or her life.

An interesting finding that emerged in these analyses was thelack of evidence to suggest that the emotionally controlled groupwas in any way overly inhibited and constricted emotionally.While the FC determinant individually was correlated with higherlevels of obsessive control, when the categorical affective modula-tion subgroups were used, no such finding emerged (e.g., consci-entiousness could have been highly elevated in this group). Thisresult may be related to the specific nature of our clinical sample,in which greater adaptive control is likely to be a strong protec-tive factor in relation to whatever issue the participants were be-ing assessed for. However, it also may indicate that the FC:CF-i-Cratio is most informative when looking at the increasing levels ofCF and C, which appear to move from more traditional affectivedysregulation to more extreme affective flooding that has resultedin a modulation "shutdown."

Strengths and limitationsThis study has a number of strengths. It is perhaps the most com-prehensive set of analyses to date to examine the chromatic colordeterminants in relation to a full psychological assessment. Thisis particularly timely given the development of new Rorschachcoding systems, which thankfully are more stringently empha-sizing the empirical foundations of the scores. Next, the clinicalsample provides an image of real patients who are being referredfor evaluations, thus having naturalistic validity. There is cur-rently minimal research studying Rorschach profiles as used inclinical settings (rather than in research studies). Third, this studyrelied on multimethod and multitrait measures. Sources of dataincluded the clinical interview, self-report measures, and neuro-psychological and psychological performance-based measures.

Vol. 77, No. 1 (Winter 2013) 89

Malone et al.

Despite tbese strengtbs, tbere are also a number of limitations.First, the subjects were not part of a random sample and samplesize was relatively small. However, given tbat our limited samplesize reduced statistical power, tbis in fact means tbat our signifi-cant results are conservative. In addition, our sample, althoughsomewbat representative of the bospital population, was ulti-mately lacking in diversity and tberefore generalization. Finally,tbe cbromatic color determinants from tbe Rorscbacb were notdouble coded by independent raters.

Future directions of tbis work include replicating tbese find-ings in a larger diverse normative sample. Tbere is a need to con-tinue empirically testing tbe assumptions tied to variables on theRorscbacb. As mentioned, it would be interesting to see wbetberbigber levels of FC in tbe FC:CF+C ratio migbt be associatedwitb negative consequences of emotional constriction in non-clinical samples. Next, given tbat tbe meaning of a bigber CF+Cappeared to cbange as tbat side of tbe ratio grew, it would bevaluable to furtber parse tbis finding. For example, are pure Csand CFs making different contributions to disturbances in affectmodulation capacities? Finally, tbere is a need to better defineand differentiate affect modulation as assessed by tbe Rorscbacband emotional regulation. It appears possible tbat affective mod-ulation is a different process tbat is less tied to managing moodlability and more representative of having the means to engageand shape affective responses to furtber expression as opposed tobeing dominated by tbem eitber tbrougb a bursting of frenzy orsimply being flattened alive.

In conclusion, tbe Rorscbacb's cbromatic color determinantsmay bave tbe potential to measure aspects of a patient's affectivemodulation. Utilizing tbe Rorscbacb in conjunction witb otberself-report and clinician-report measures allows for an empiri-cal view of implicit process tbat may be outside of the patient'sawareness yet still affecting tbe person in his or ber daily life.

- '" Buttetin of the Menninger Ctinic


References

Biais, M. A. (2011). PEaRL Semi-structured Clinical Interview. Boston, MA:Massachusetts General Hospital.

Biais, M. A.. Hilsenroth, M. J., & Fowler, J. C. (1998). Rorschach correlatesof the DSM-IV histrionic personality disorder, journal of Personali-ty Assessment, 70, 355-364.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multipleregression/correlation analysis for the behavioral sciences (Vol. 3).Mahwah, NJ: Lawrence Erlbaum Associates.

Costa, P. T., Jr., & McCrae, R. R. (1992). NEO PI-R professional manual:Revised NEO Personality Inventory (NEO OI-R) and NEO Five-Factor Inventory (NEO-FFI). Lutz, FL: Psychological AssessmentResources.

Exner, J. E. (1978). The Rorschach: A comprehensive system. Volume 2:Current research and advanced interpretation (1st ed.) New York:John Wiley.

Exner, J. E. (2001). A Rorschach workbook for the comprehensive system.Asheville, NC: Rorschach Workshops, Inc.

Exner, J. E. (2003). The Rorschach: A comprehensive system. Volume 1:Basic foundations and principles for interpretation (4th ed.). NewYork: John Wiley.

Exner, J. E., & Erdberg, P. (2005). The Rorschach: A comprehensive system.Volume 2: Advanced interpretation. Hoboken, NJ: Wiley.

Exner, J. E., & Murillo, L. G. (1975). Early prediction of posthospitalizationrelapse, journal of Psychiatric Research, 12, 231-237.

Exner, J. E., Murillo, L. G., &C Cannavo, E. (1973). Disagreement betweenpatient and relative behavioral reports as related to relapse in non-schizophrenic patients (Vol. 34). Washington, DC: Eastern Psycho-logical Association.

Frank, G. (1976). On the validity of hypotheses derived from the Rorschach:I. The relationship between color and affect. Perceptual and MotorSkills, 43, 411-427.

Frank, G. (1993). On the validity of hypotheses derived from the Rorschach:The relationship between color and affect, update 1992. Psycholo-gical Reports, 73, 12-14.

Kron, A., Cohen, A., Benzimana, H., &c Ben-Shakhara, G. (2009). What is itthat color determinants determine? The relation between the Ror-schach Inkblot Method and cognitive object-recognition processes.journal of Personality Assessment, 91, 137-142.

McCrae, R., & Costa, R (2004). A contemplated revision of the NEOFive-Factor Inventory. Personality and Individual Differences, 36,587-596.

Meyer, G. J., Viglione, D. V., Mihura, J. L., Erard, R. E., & Erdberg, R(2011). Rorschach performance assessment system: Administration,

Vol. 77, No. 1 (Winter 2013) 91

Malone et al.

coding and technical manual. Toledo, OH: Rorschach PerformanceAssessment System.

Miller, T. A. (1999). Rorschach assessment of object relations and affectcontrol in domestic violent and non-violent couples. Dissertati-on Abstracts International: The Humanities and Social Sciences,59(11-A), 406^09.

Monsch, A. U., Bondi, M. W., Butters, N., Salmon, D., Katzman, R., ôcThal, L. J. (1992). Comparisons of verbal fluency tasks in the detec-tion of dementia of the Alzheimer's type. Archives of Neurology,49, 1253-1258.

Morey, L. C. (1991). Personality Assessment Inventory: Professional manu-al. Odessa, FL: Psychological Assessment Resources, Inc.

Morey, L. C. (2007). Personality Assessment Inventory: Professional manual(2nd ed.). Odessa, FL: Psychological Assessment Resources.

Ochsner, K. N., &: Gross J. J., (2005). The cognitive control of emotion.Trends in Cognitive Sciences, 9, 242-249.

Overton, C. G. (2000). A relational interpretation of the Rorschach colordeterminants. Journal of Personality Assessment, 75, 426-448.

Randolph, C. (1998). RBANS manual: Repeatable Battery for the Assess-ment of Neuropsychological Status. San Antonio, TX: Psychologi-cal Corp (Harcourt).

Rapaport, D., Gill, M. M., ôc Schafer, R. (1968). Diagnostic psychologicaltesting. New York: International Universities Press.

Reitan, R. M. (1958). Validity of the Trail Making Test as an indicator oforganic brain damage. Perceptual and Motor Skills, 8, 271-276.

Reitan, R. M., & Wolfson, D. (1985). The Halstead-Reitan neuropsycholo-gical test battery: Theory and clinical interpretation. Tucson, AZ:Neuropsychology Press.

Rorschach, H. (1942). Psychodiagnostics. Berne, Switzerland: Hans. (Origi-nal work published 1921)

Schachtel, E. G. (1966). Experiential foundations of Rorschach's test. NewYork: Basic Books.

Slavin-Mulford, J., Sinclair, S. J., Stein, M., Malone, J. C , Bello, I., & Biais,M. A. (2012). External validity of the Personality AssessmentInventory (PAI) in a clinical sample. Journal of Personality Assess-ment, 94, 593-600.

Stevens, D. T., Edwards, K. J., Hunter, W. E, &c Bridgman, L. (1993). An in-vestigation of the color-affect hypothesis in Exner's ComprehensiveSystem. Perceptual and Motor Skills, 77, 1347-1360.

Trenerry, M. R., Crosson, B., Deboe, J., & Leber, W. (1989). Stroop Neu-ropsychological Test (SNST). Odessa, FL: Psychological ResourcesInc.

Tukey, J. W. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley.

92, Bulletin of the Menninger Clinic


Wechsler, D. (1999). Wechsler Abbreviated Scale of Intelligence (WASI). SanAntonio, TX: Harcourt Assessment.

Wechsler, D. (2008). Wechsler Adult Intelligence Scale-Fourth Edition. SanAntonio, TX: Pearson.

Wechsler, D. (2009). Wechsler Memory Scale-Fourth Edition. San Antonio,TX: Pearson.

Weiner, I. B. (1964). Pure C and color stress as Rorschach indicators ofschizophrenia. Perceptual and Motor Skills, 18, 484.

Zodan, J., Chamas, J., & Hilsenroth, M. J. (2009). Rorschach assessment ofreality testing, affect and object representation of borderline pathology:A comparison of clinical samples. Bulletin of the Menninger Clinic, 73,121-142.

Vol. 77, No. 1 (Winter 2013) 93

Copyright of Bulletin of the Menninger Clinic is the property of Guilford Publications Inc. and its content may

not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written

permission. However, users may print, download, or email articles for individual use.

seeing red: affect modulation and chromatic color

Documents