ارزشیابی از موقعیت غیردرون فردی عاطفی بیماران مبتلا به اختلال شخصیت مرزی

Abstract According to cognitive models of borderline personality disorder (BPD), an important cause for the instability of patients with BPD is dichotomous thinking (DT). Object-relation theories assume that the similar phenomenon of splitting is central in BPD. Previous studies focusing on interpersonal situations found support for DT being prominent in BPD. The aim of this study was to investigate whether patients with BPD also make use of dichotomous and schema-specific evaluations in noninterpersonal situations. An experiment was designed in which a frustrating and rewarding situation was induced by computer games that subjects had to play. Participants evaluated both themselves and the games. Patients with BPD (n=24)(n=24) were characterized by somewhat more extreme game evaluations in the emotionally negative situations than normal controls (n=25)(n=25), participants with a cluster C (n=10)(n=10) or an anti-social personality disorder (ASPD) (n=16)(n=16). Patients with BPD appeared to be characterized best by a general negative evaluative style, more than by DT or splitting. ASPD participants showed a positivity bias in both conditions.

. Introduction A characteristic feature of patients with borderline personality disorder (BPD) is their instability of affects, self-image and relationships causing problems and conflicts in these domains (American Psychological Association (APA), 1994). The Beckian cognitive model of BPD described by Pretzer (1990) and Arntz (2004) attributes this instability to (a) dichotomous (black-or-white) thinking (DT), i.e., evaluating situations in mutually excluding extremes, and (b) incompatible basic beliefs. It is assumed that borderline patients make extreme, dichotomous evaluations of the world, instead of seeing the world in shades of grey (Arntz, 2004 and Pretzer, 1990). Lacking intermediate evaluative categories these patients show extreme emotions and behavior, as well as abrupt shifts from one extreme to the other. Although DT is seen as a common cognitive distortion, which may also be displayed by other patients, this thinking style is thought to be highly prominent in borderline patients. Besides DT, the basic beliefs or schemas of borderline patients (Arntz, 2004 and Pretzer, 1990) are also assumed to increase instability in BPD. According to the model, borderline patients see (i) themselves as powerless and vulnerable, and (ii) as unacceptable, and (iii) the world or others as dangerous and malevolent. The first schema turns them towards others, seeking help and protection, whereas the other schemas drive them in the very opposite direction, that is away from others, preventing punishment and abuse. The combination of dichotomous thinking and the incompatible schemas is very potent and might explain the unstable patterns of BPD. Most studies on evaluations by patients with BPD (Arntz and Veen, 2001, De Bonis et al., 1998, Leichsenring et al., 1992, Sheffield et al., 1999 and Veen and Arntz, 2000) support the hypothesis that DT is prominent in BPD. Overall, it was found that borderline patients evaluated others more extremely than clinical and nonclinical controls. Veen and Arntz (2000) found that this was only the case when borderline patients were asked to evaluate persons in a BPD-specific role, such as the role of perpetrator or victim of abuse. Evaluations of persons in an emotional role that was not BPD-specific or a neutral role, were not more extreme than those of others. Possibly, activating BPD schemas is a prerequisite for patients with BPD to think extremely. The primary aim of this study was to investigate whether the range of situations in which patients with BPD make use of dichotomous thinking also extends to noninterpersonal situations. The studies described above focused on interpersonal evaluations. This focus was probably chosen because having unstable interpersonal relationships is a salient feature of borderline patients. Another reason is that object-relation and attachment theories conceptualize extreme evaluations of borderline patients as direct or indirect reiterations of adverse interpersonal childhood experiences (e.g., Fonagy et al., 2000, Kernberg, 1996 and Sable, 1997). It is however not clear whether DT in BPD is a general characteristic which is applied in frustrating or rewarding interpersonal as well as noninterpersonal situations. Besides having interpersonal problems, patients with BPD also struggle with performing noninterpersonal tasks such as studies, jobs, and household tasks. This kind of problems might stem from extreme noninterpersonal evaluations. A secondary aim of this study was to investigate the way DT in BPD manifests itself in noninterpersonal situations, and how to exactly conceptualize noninterpersonal DT: as (schema related or unrelated) cognitive distortions, as the defense mechanism ‘splitting’, or as negative thinking. Instead of a cognitive distortion, object-relation theories (e.g., Kernberg, 1966 and Kernberg, 1996) conceive extreme evaluations of borderline patients as expressions of ‘splitting’, a defense mechanism thought to separate good and bad aspects of self- and other-representations. It is assumed that split evaluations (1) are unidimensional or univalent, i.e., all good or all bad, and (2) can be extremely positive (idealization) as well as extremely negative (devaluation). Studies focusing on interpersonal evaluations found evidence for both unidimensional DT ( Arntz & Veen, 2001), as well as multidimensional or multivalent (for example, very good and very bad) DT ( De Bonis et al., 1998, Sheffield et al., 1999 and Veen and Arntz, 2000). According to Arntz and Veen (2001), response format might explain these ambiguous findings. A structured response format might protect patients with BPD from primitive, unidimensional thinking levels and help them to perform on higher levels than usually in normal life. Extreme positive thinking was not found in these studies, but this might be caused by having used only neutral and negative stimuli. Extreme evaluations in borderline patients might also be expressions of a negative thinking style or negativity bias. Several studies, mostly using projective stimuli, found that a lower affect-tone of the object world was the most specific feature of borderline patients spontaneous narratives or evaluations (Arntz and Veen, 2001, Kurtz and Morey, 1998, Nigg et al., 1992, Segal et al., 1992, Spear and Sugarman, 1984, Stuart et al., 1990 and Westen et al., 1990; Westen, Ludolph, Lerner, Ruffins, & Wiss, 1990). As mentioned by Westen and co-workers (e.g., Stuart et al., 1990), this thinking style might be explained by the projection of aggressive impulses, higher levels of constitutional aggression and/or by expectations or schemas of the object world as being malevolent and untrustworthy possibly resulting from childhood traumas (see for the latter relation, Arntz, 1994, Arntz, 2004 and Ornduff, 2000) (see for a review of studies on childhood trauma in BPD, Sabo, 1997). We investigated these issues by asking borderline patients, personality disorder (PD) controls (i.e., patients with a cluster C PD (CPD) and participants with an anti-social PD (ASPD)), and normal controls, to play two computer games and to evaluate these games as well as themselves. In this way we experimentally manipulated a noninterpersonal situation. By varying the degree of difficulty of the games, we created a frustrating, respectively rewarding situation which, we hypothesized, would activate schemas on powerlessness (of the self) and malevolence (of the noninterpersonal world), and would offer opportunity for extreme evaluations, possibly also extremely positive ones. In summary, the main question of this study was whether BPD is characterized by dichotomous and schema-specific noninterpersonal evaluations. The hypotheses were as follows: (1) extreme noninterpersonal evaluations are a specific feature of BPD; (2) patients with BPD evaluate their nonpersonal environment more extremely than control subjects on dimensions related to the schema ‘the world is dangerous and malevolent’; (3) patients with BPD evaluate themselves more extremely than control subjects on dimensions related to the schemas ‘I am powerless and vulnerable’ and ‘I am unacceptable’; (4) the extreme evaluations in BPD are unidimensional; (5) patients with BPD show extremely negative evaluations as well as extremely positive evaluations.

3. Results 3.1. Manipulation check In order to check whether the easy and difficult games were experienced as intended, we analysed the game scores (objective measure), game evaluations on the scales difficult–easy, frustrating–rewarding, not doable–doable, and changes in emotional state from baseline and to post-game measurement (subjective measures). 3.1.1. Game scores Overall, the mean game score on the difficult games (M=-357M=-357, Sd=493Sd=493) was significantly lower than the mean score on the easy games (M=6.80M=6.80, Sd=201Sd=201), Z=-5.38Z=-5.38, p<.0005p<.0005 (one-tailed test). The difference between scores on the difficult and the easy games was similar over the groups, χ2(3,N=75)=.58χ2(3,N=75)=.58, ns. 3.1.2. Game evaluations The evaluations were analysed separately for each scale. Overall, subjects evaluated the difficult games as less easy (M=-1.90M=-1.90, Sd=33.3Sd=33.3), Z=-2.16Z=-2.16, p<.05p<.05, less rewarding (M=-18.2M=-18.2, Sd=24.4Sd=24.4), Z=-3.86Z=-3.86, p<.0005p<.0005, and less doable (M=7.00M=7.00, Sd=32.8Sd=32.8), Z=-6.06Z=-6.06, p<.0005p<.0005, than the easy games (easy: M=9.20M=9.20, Sd=29.3Sd=29.3; rewarding: M=-6.00M=-6.00, Sd=24.5Sd=24.5; doable: M=23.7M=23.7, Sd=25.8Sd=25.8) (one-tailed tests). The differences between evaluations of easiness of the difficult and easy games differed over the groups, χ2χ2(3,N=75)=9.16(3,N=75)=9.16, p<.05p<.05. The groups evaluations of rewardfulness and doableness differed similarly from easy to difficult games over the groups, χ2χ2's (3,N=75)<3.90(3,N=75)<3.90, ns. For easiness, the difference between the games was smaller for the BPD group than for the ASPD group, t(38)=2.29t(38)=2.29, p<.05p<.05, and in further analyses it appeared that of the four groups only the ASPD group evaluated the easy game as significantly easier (M=9.56M=9.56, Sd=30.3Sd=30.3) than the difficult game (M=-22.6M=-22.6, Sd=20.8Sd=20.8), Z=-2.64Z=-2.64, p<.005p<.005 (one-tailed test). 3.1.3. Emotional state Only emotions with clear valences were analysed, i.e., happiness, sadness, fear, and anger. After the difficult game, the subjects reported less happiness (M=39.5M=39.5, Sd=23.3Sd=23.3), Z=-4.35Z=-4.35, p<.0005p<.0005, and more anger (M=20.3M=20.3, Sd=24.7Sd=24.7), Z=-3.18Z=-3.18, p<.0005p<.0005, than at baseline (happiness: M=55.4M=55.4, Sd=25.3Sd=25.3; anger: M=12.0M=12.0, Sd=17.3Sd=17.3) (one-tailed tests). Sadness (baseline: M=24.9M=24.9, Sd=26.0Sd=26.0; difficult game: M=23.5M=23.5, Sd=26.1Sd=26.1) and fear (baseline: M=24.0M=24.0, Sd=28.6Sd=28.6; difficult game: M=21.1M=21.1, Sd=26.5Sd=26.5) did not significantly increase from baseline to difficult game, |Z||Z|'s <.88<.88, ns (one-tailed tests). After the easy game, the subjects reported less sadness (baseline: M=24.9M=24.9, Sd=26.0Sd=26.0; easy game: M=19.7M=19.7, Sd=23.4Sd=23.4), Z=-2.24Z=-2.24, p<.05p<.05, but not more happiness, Z=-2.78Z=-2.78, ns, nor less anger or fear, |Z||Z|'s <.62<.62, ns (one-tailed tests). In fact, there was a decrease of reported happiness after the easy game, (easy game: M=46.0M=46.0, Sd=23.1Sd=23.1), Z=-2.78Z=-2.78, p<.005p<.005 (two-tailed test). Most emotion changes from baseline to post-game measurement were similar over the groups, χ2χ2's (3,N=75)<7.07(3,N=75)<7.07, ns. Only the decrease in fear from baseline to easy game differed significantly over the groups, χ2χ2(3,N=75)=10.4(3,N=75)=10.4, p<.05p<.05. The BPD group showed a significant decrease of fear after the easy game (baseline: M=36.7M=36.7, Sd=29.8Sd=29.8; easy game: M=30.5M=30.5, Sd=27.6Sd=27.6), Z=-1.77Z=-1.77, p<.05p<.05, whereas the others did not (one-tailed tests). 3.1.4. Manipulation check summarized Playing the difficult game resulted in low game scores, was found to be not very rewarding and doable, decreased happiness, and increased anger. Compared to the difficult game, the easy game resulted in high game scores and was evaluated as rewarding and doable. Although playing the easy game reduced happiness, it also reduced sadness in all groups and fear in the target BPD group. We conclude that we successfully manipulated noninterpersonal frustration and reward. 3.2. Game evaluations 3.2.1. DT scores DT scores are presented in Fig. 1. Contrasts on the nonneutral scales were in the expected direction with exception of the contrast BPD vs. ASPD for DT scores of the easy game on the specific scale. The contrasts however were mostly not significant, as is shown in Table 2. For the difficult game, only the contrast BPD vs. CPD for evaluations on the nonspecific scales reached significance. For evaluations on the specific scales this contrast (BPD vs. CPD) was a trend. For the easy game, all contrast were nonsignificant. DT scores (with standard errors) per subject group, game, and scale type. Fig. 1. DT scores (with standard errors) per subject group, game, and scale type. Figure options Table 2. Test resultsa of the group contrasts for the DT, splitting, and NT scores of the game evaluations Contrast Game evaluation scores DT Splitting NT Difficult game on specific scales BPD vs. CPD −1.37†† −1.43†† 1.00 BPD vs. ASPD −0.32 −0.45 3.61*** BPD vs. NP −1.46†† 210* 1.70* Difficult game on nonspecific scales BPD vs. CPD −1.77* 86.0†† 1.48†† BPD vs. ASPD −0.49 −0.99 2.48** BPD vs. NP −1.09 223†† 1.70* Difficult game on neutral scales BPD vs. CPD 0.18 BPD vs. ASPD 0.53 BPD vs. NP −0.58 Easy game on specific scales BPD vs. CPD −0.28 107 −0.08 BPD vs. ASPD 176 1.29 1.83* BPD vs. NP 240 284 271 Easy game on nonspecific scales BPD vs. CPD −0.53 119 0.55 BPD vs. ASPD −0.78 159 1.27†† BPD vs. NP −1.48†† 291 0.66 Easy game on neutral scales BPD vs. CPD −0.08 BPD vs. ASPD 0.18 BPD vs. NP 0.84 ††.05<p<.10.05<p<.10. * p<.05p<.05. ** p<.01p<.01. *** p<.001p<.001. a Values <5<5 are tt-values of the tt-test, values >100>100 are UU-values of the Mann–Whitney test. Table options 3.2.2. Splitting scores Fig. 2 illustrates the splitting scores. Contrasts on the nonneutral scales were again in the expected direction with exception of the contrast BPD vs. ASPD for evaluations of the easy game on the specific scales. Just as the DT scores, however, the splitting scores also did not show large group effects (see Table 2). For the difficult game, the contrast BPD vs. NP for evaluations on the specific scales was the only significant contrast. Splitting scores (with standard errors) per subject group, game, and scale type. Fig. 2. Splitting scores (with standard errors) per subject group, game, and scale type. Figure options 3.2.3. NT scores Mean NT scores of the game evaluations on the nonneutral scales are presented in Fig. 3. Patients with BPD evaluated the difficult game more negatively than most other groups. Unlike the DT and splitting scores, contrasts with BPD for the NT scores were, at least for the difficult game, mostly significant. Only the contrasts BPD vs. CPD failed to reach significance, although we did find a trend for this contrast for evaluations on the nonspecific scales. NT scores (with standard errors) per subject group, game, and scale type. Fig. 3. NT scores (with standard errors) per subject group, game, and scale type. Figure options Because the ASPD group evaluated the games mostly more positively instead of more negatively, we also looked at the contrasts ASPD vs. CPD, and ASPD vs. NP. For evaluations of the difficult game, the contrasts ASPD vs. CPD for the specific scale, t(32)=-1.96t(32)=-1.96, p<.05p<.05, and ASPD vs. NP for the specific scale, t(47)=-2.13t(47)=-2.13, p<.05p<.05, were significant. For the easy game, these contrasts were nonsignificant. DT, splitting, and NT scores were not corrected for group differences in age, sex, and computer use because regression coefficients for these scores were nonsignificant in all game by scale conditions. 3.3. Self-evaluations 3.3.1. Baseline At baseline, patients with BPD showed lower DT scores for self-evaluations than the ASPD group on the nonspecific scales (BPD: M=23.8M=23.8, Sd=9.26Sd=9.26; ASPD: M=28.3M=28.3, Sd=10.8Sd=10.8), t(38)=2.29t(38)=2.29, p<.05p<.05, and than the NP group on the specific scales (BPD: M=26.4M=26.4, Sd=9.90Sd=9.90; NP: M=34.7M=34.7, Sd=11.2Sd=11.2), t(47)=2.81t(47)=2.81, p<.01p<.01. DT scores were not corrected for group differences in sex, age, and computer use because regression coefficients were nonsignificant for all the scales. Patients with BPD showed also low splitting scores for self-evaluations at baseline. Two-tailed tested contrasts that reached significance were the contrast with the ASPD group for the nonspecific scale (BPD: M=13.4M=13.4, Sd=10.4Sd=10.4; ASPD: M=27.0M=27.0, Sd=10.0Sd=10.0), U=68.0U=68.0, p<.0005p<.0005, and with the NP group for both scales: specific (BPD: M=13.0M=13.0, Sd=9.13Sd=9.13; NP: M=19.4M=19.4, Sd=9.19Sd=9.19), U=191U=191, p<.05p<.05, and nonspecific scales (NP: M=24.4M=24.4, Sd=13.1Sd=13.1), t(47)=3.47t(47)=3.47, p<0.005p<0.005 (two-tailed tests). The regression coefficient of the factor computer use was significant for splitting scores on the specific scales (β=.28β=.28), t=2.45t=2.45, p<.05p<.05. All other regression coefficients were nonsignificant. The analyses were corrected by adding computer use as covariate to the ANOVA. For the Mann–Whitney test, we recalculated the scores for each subject for one common level of computer use, using the regression coefficient. The corrections did not change the results. NT scores of the self-evaluations at baseline were also low for patients with BPD. The BPD group evaluated themselves less positively than the ASPD group on the specific scales (BPD: M=-6.07M=-6.07, Sd=14.9Sd=14.9; ASPD: M=15.9M=15.9, Sd=9.78Sd=9.78), t(38)=5.80t(38)=5.80, p<.0005p<.0005, and the nonspecific scales (BPD: M=5.07M=5.07, Sd=16.4Sd=16.4; ASPD: M=27.0M=27.0, Sd=10.0Sd=10.0), U=48.0U=48.0, p<.0005p<.0005 (two-tailed tests). The BPD group evaluated themselves also less positively than the NP group, again on both scales, viz., the specific scales (NP: M=19.4M=19.4, Sd=9.19Sd=9.19), U=46.5U=46.5, p<.0005p<.0005, and the nonspecific scales (NP: M=24.3M=24.3, Sd=13.1Sd=13.1), t(47)=4.87t(47)=4.87, p<.0005p<.0005 (two-tailed tests). Mean evaluation scores were not corrected for group differences in sex, age, and computer use because regression coefficients were nonsignificant for all the scales. 3.3.2. Effect of games on self-evaluations The effects of the games on the self-evaluations were analysed with ANCOVAs with scores at baseline as covariates and the DT, splitting or NT scores as dependent variables, or with Mann–Whitney tests of the change in scores (post-game score–score at baseline). The BPD group showed in almost all conditions lower instead of higher DT and splitting scores than the other groups. No contrast reached significance. Regression coefficients of sex, age, and computer use were nonsignificant for the DT and splitting scores in all game by scale conditions. For the NT self-evaluation scores after the difficult game, we found that patients with BPD evaluated themselves significantly more negatively than those with ASPD on the specific scales (BPD: M=6.63M=6.63, SE=2.04SE=2.04; ASPD: M=11.8M=11.8, SE=2.33SE=2.33), t(38)=1.89t(38)=1.89, p<.05p<.05. No significant contrasts were found for the easy game. Again, we also investigated indications for a positivity bias in ASPD by looking at the contrasts with this group. The ASPD group showed significantly more positive self-evaluations than the NP group on the specific scales after both the difficult game (changes from baseline to difficult game: ASPD: M=1.15M=1.15, Sd=9.08Sd=9.08; NP: M=-1.99M=-1.99, Sd=7.00Sd=7.00), U=119U=119, p<.05p<.05, and the easy game (changes from baseline to easy game: ASPD: M=2.76M=2.76, Sd=4.51Sd=4.51; NP: M=-.72M=-.72, Sd=5.59Sd=5.59), U=110U=110, p<.05p<.05. All other contrasts were nonsignificant. Regression analyses yielded significant coefficients for the factor sex on the nontransformed self-evaluation score on the specific scales after the difficult game (β=-.29β=-.29, i.e., women show lower scores), t=-2.53t=-2.53, p<.05p<.05, and for the factor computer use on the same scales after the easy game (β=.25β=.25, i.e., subjects high on computer use show higher scores), t=-2.16t=-2.16, p<.05p<.05. All other regression coefficients were nonsignificant. We corrected for possible biases of group differences in sex on the analyses for the specific scales after the difficult game by redoing these analyses with the female subjects only. For the female subjects, the contrasts BPD vs. ASPD patients and ASPD vs. NP group for nontransformed self-evaluations on the specific scales of the difficult game disappeared. Biases of group differences in computer use on the analyses for the specific scales after the easy game were corrected for by adding computer use as an additional covariate, or (for nonparametric analyses) by recalculating the scores for each subject for one common level of computer use, using the regression coefficient. After correction, a borderline significant contrast for this condition appeared between the BPD and ASPD group (BPD: M=10.4M=10.4, SE=2.50SE=2.50; ASPD: M=11.2M=11.2, SE=3.56SE=3.56), t(38)=1.78t(38)=1.78, p=.05p=.05. 3.4. Multidimensional evaluations versus one-dimensional evaluations In order to investigate whether the evaluations were multidimensional or unidimensional, we compared the DT scores and splitting scores for the games and self-evaluations. The BPD group showed significantly higher DT scores than splitting scores for both game and self-evaluations in all conditions, .0005<p.0005<p's <.03<.03, indicating that DT of these patients was multidimensional instead of unidimensional.