Biological Psychology 70 (2005) 141–151 www.elsevier.com/locate/biopsycho

Cognitive restructuring and EEG in major depression Patricia J. Deldin a,*, Pearl Chiu b a

University of Michigan, Department of Psychology, 2252 East Hall, Ann Arbor, MI 48109-1109, USA b Harvard University, Department of Psychology, William James Hall, Cambridge, MA 02138, USA Received 30 October 2002; accepted 5 January 2005 Available online 7 March 2005

Abstract Techniques based on cognitive therapy and electroencephalography (EEG) were used to investigate the predictive utility of EEG alpha power with regard to mood improvement. Controls and individuals with major depression participated in four EEG recording blocks. Blocks 1 and 4 were resting baselines. During Block 2, Ss were asked to think about their ‘‘most troubling life difficulty.’’ Next, Ss were introduced to cognitive views of depression and techniques used in cognitive therapy. For Block 3, Ss were asked to use these methods to think again about their life difficulty. Ss who reported greater post- than pre-intervention happiness (i.e., ‘‘Responders’’) exhibited greater overall cortical activity than Nonresponders. Depressed Responders further exhibited a cortical asymmetry of greater right relative to left activity in frontal areas. The predictive utility of EEG is discussed with regard to identifying individuals who show mood improvement following cognitive restructuring. # 2005 Elsevier B.V. All rights reserved. Keywords: Electroencephalography; Major depression; Cognitive therapy; Treatment predictors

1. Cognitive restructuring and EEG in major depression Cognitive theories have been shown to be appropriate for describing major depression (e.g., Haaga et al., 1991; Williams et al., 1997), and have led to an efficacious treatment, cognitive psychotherapy, for the disorder (Elkin et al., 1989; DeRubeis and Crits-Christoph, 1998). Indeed, cognitive therapy is considered to be a highly effective treatment – as effective as pharmacotherapy in many cases – and potentially more effective when used in conjunction with medications (Dobson, 1989; Otto et al., 1996). Moreover, research indicates that depressed individuals who receive cognitive therapy either alone or with pharmacotherapy may be less likely to experience a relapse than patients who receive other forms of treatment (Evans et al., 1992; Thase and Simons, 1992). However, despite evidence for the efficacy of cognitive therapy, relatively little is known about the patient characteristics of those who exhibit a therapeutic response. * Corresponding author. Tel.: +1 734 647 9863; fax: +1 734 615 0573. E-mail address: [email protected] (P.J. Deldin). 0301-0511/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.biopsycho.2005.01.003

The human and fiscal cost of failed treatment motivates the attempt to refine measures that may discern those who may benefit from psychotherapy from those who will not. Potential predictors, mediators, and concomitants of positive response to cognitive therapy have traditionally been examined in studies that dismantle components of psychotherapy (e.g., Jacobson et al., 1996; Jarrett and Nelson, 1987), that compare pharmacotherapy outcome with psychotherapy outcome (e.g., Peselow et al., 1990; Fava et al., 1994), or that attempt to identify predictors of treatment response using a range of self-report measures or therapist variables (e.g., Castonguay et al., 1996; Dohr et al., 1989). Although such studies have provided some insight into the efficacy and changes associated with various aspects of cognitive therapy, the results have largely been inconsistent in yielding a comprehensive system of variables that may be of clinical utility with regard to therapeutic response. Investigations of psychosocial factors have also been inconclusive regarding the utility of any particular variable, or group of variables, as an outcome predictor of positive response to cognitive therapy (for reviews, see Scott, 1996; Whisman, 1993). Indeed, although traditional

142

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

approaches to identify these predictors have implicated such assorted variables as therapeutic empathy, perceived helpfulness of the therapy, homework compliance, degree of life stress, and symptom severity, as important with regard to the therapeutic process, none has emerged as a sufficient predictor of response to cognitive therapy (for review, see Salkovskis, 1996). Clearly, the patient characteristics that may be associated with therapeutic response require further investigation. There is a growing literature suggesting that examining brain physiology in addition to psychosocial variables may be of utility in exploring mechanisms and identifying predictors of treatment response (Miller and Keller, 2000). For example, pre-treatment hemispheric advantage on a dichotic listening task has been found to predict response to both cognitive–behavioral therapy and fluoxetine therapy (Bruder et al., 1996, 1997). In addition, abnormal sleep EEG profiles (based on REM density, REM latency, and sleep efficiency) have been associated with non-response to both cognitive–behavioral therapy and interpersonal psychotherapy in major depression (Thase et al., 1996, 1997). Recent biofeedback studies also suggest that EEG techniques may be of utility in exploring mechanisms of therapeutic change (Baehr et al., 1997; Rosenfeld et al., 1995; Allen et al., 2001). As major depression is likely to be an interaction between psychosocial and biological variables, it seems reasonable to suggest that therapeutic response may be associated with cortical physiology. Elucidating this relationship clearly has significant implications for not only the design and implementation of effective therapeutic practices but also for the understanding of the maintenance of major depression. Thus, the present research investigates the role and predictive utility of electroencephalographic (EEG) patterns and changes with regard to positive therapeutic response. 1.1. EEG alpha The phenomena of EEG alpha (8–12 Hz) synchronization during mental relaxation and desynchronization during mental activity has long been observed (for review, see Klimesch, 1999), and has led researchers to employ EEG alpha power as an economical, yet precise, inverse index of cortical activity (Gevins, 1998). More generally, increased alpha power during cognitive tasks is thought to reflect a relative decrease in the proportion of local cortical neurons that are engaged in a particular task performance (Gevins and Smith, 2000; Klimesch, 1999). Research has consistently found a decrease of alpha power during task solution (e.g., Rugg and Dickens, 1982; Ramos et al., 1993; McEvoy et al., 2000), as well as taskspecific changes in regional alpha power (e.g., Earle, 1988; Molle et al., 1999; Bell and Fox, 2003). EEG alpha power has also received increasing attention as an index of particular performance and task variables. Specifically,

evidence from studies with tasks varying attentional requirements suggests that alpha is inversely related to attentional level and demand during cognitive processes (e.g., Ray and Cole, 1985; Dujardin et al., 1993). Moreover, increases in task difficulty have been found to be associated with a decrease of power in the lower alpha band (8–10 Hz; Gevins and Smith, 2000). Gevins and Smith (2000) further noted that a high-ability group of participants (as measured by scores on the WAIS-R) displayed relatively larger practice-related increase in alpha power, suggesting that this group adapted more quickly to task demands, or that they were able to adopt task performance strategies that took advantage of processing resources in other cortical regions. EEG alpha asymmetry has also been used as a measure of cortical activity in an extensive range of studies investigating the localization of emotional processing to general cortical regions. Specifically, it has been suggested that the left and right frontal regions are specialized, respectively, for the experiential aspects of positive (i.e., approach-related) and negative (i.e., withdrawal-related) emotion (e.g., Davidson et al., 1979; Sackheim et al., 1982; for review, see Davidson, 1998). Evidence further suggests that a frontal asymmetry (of greater left relative to right pre-frontal alpha power) may be characteristic of depression and may predispose an individual to respond with predominantly negative affect given a negatively valenced stimulus (e.g., Davidson, 1995, 1998; Harmon-Jones, 2003). Thus, existing research on EEG emphasizes that both asymmetry and overall levels of cortical activity are likely to influence an individual’s response, or capacity to respond, to a particular cognitive or emotional demand. Indeed, an extensive literature has validated EEG alpha as a tool for delineating specific cognitive states and processes; similarly, much literature exists regarding the cortical asymmetry models of depression and affect. Hence, if cognitive therapy for depression is considered a task with high cognitive demand, an individual’s levels and patterns of cortical activity are likely to moderate the therapeutic response. 1.2. Cognitive restructuring and EEG The present study combines the techniques of electroencephalography and cognitive restructuring to examine whether baseline patterns and levels of cortical activity may be useful indices of mood response to a cognitive therapy analog. We anticipate that pre-existing levels or patterns of cortical activity may have predictive utility regarding whether an individual will exhibit mood improvement following a brief cognitive intervention. Specifically, we hypothesize that (1) those who exhibit greater baseline cortical activity will be at an advantage to report a therapeutic response to the mood intervention, and/or (2) that a baseline cortical asymmetry will predispose individuals to respond more (i.e., those who exhibit greater relative left frontal activity) or less (i.e., those with relatively

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

greater right frontal activity) favorably to our cognitive mood intervention. We will also investigate patterns of cortical activation concomitant with changes in mood and cognitive style. It should be emphasized that though our cognitive restructuring task is based upon the techniques of cognitive therapy for depression (Beck et al., 1979; Burns, 1999), it is a brief, one-time intervention in a laboratory setting, and any generalization of results to applications of cognitive therapy, or of our brief intervention to a vehicle of therapeutic response must be cautious. Nevertheless, our task was based upon the cognitive restructuring strategies outlined in one of the most accessible and highly recommended resources for individuals suffering from depression (Burns, 1999; Norcross et al., 2003). Thus, the potential of the task to be an efficient analog to cognitive therapy and to provide a focused examination of particular components of the therapeutic process in the laboratory may be considered.

143

education level of 14.6 years (S.D. = 2.5). The mean duration of the current depressive episode was 14 months, and 87% (n = 13) of the depressed individuals had experienced at least one previous episode of depression. The groups did not differ in age (F = .801, p > .5), sex (F = .61, p > .8), or education (F = .707, p > .4). One control participant and five depressed participants reported having been in a psychotherapy in which difficulties were discussed in a manner similar to that employed in our paradigm. Three individuals were currently taking antidepressant medication. Preliminary analyses indicate no difference in EEG reactivity in medicated and unmedicated participants; thus, all subjects were included in subsequent analyses. In accord with Harvard University Institutional Review Board guidelines, informed consent was obtained and it was emphasized that participants could withdraw from the study at any time with no adverse consequences. Participants were compensated US$ 10 for each hour spent in the laboratory. 2.2. Materials

2. Method 2.1. Participants Thirty-three participants, aged 18–65, were recruited from the Boston area through advertisements in local newspapers advertising an ‘‘information processing’’ or ‘‘depression’’ study. Exclusionary criteria for the depression group included: major medical illnesses or conditions, cognitive impairments, head injuries resulting in unconsciousness for over 10 min, and any current or past Axis I disorder other than major depression, except anxiety disorders. Exclusionary criteria for the control group further comprised all Axis I disorders. A brief phone interview served as the preliminary screening for study participation. The Structured Clinical Interview for DSM-IV, Patient Edition (SCID-I/P; First et al., 1995) was subsequently administered by a Ph.D. level clinical psychologist (PJD) or by advanced graduate students trained in SCID administration. In order to confirm accuracy of the diagnoses, 25% of the interview tapes were reviewed by a graduate student trained in SCID administration. Participants who met criteria for current major depressive episode (MDE) or normal control, as determined by the SCID, were invited to take part in the physiology studies. Because evidence suggests that patterns of cortical activity in cognitive and affective processes may differ in left-handed individuals (e.g., Savage, 1993), only right-handed individuals as assessed with the Edinburgh Handedness Inventory were included in the present study (Oldfield, 1971). The control group included 13 women and 5 men, ranging in age from 21 to 61 (M = 38.4, S.D. = 13.2), with a mean education level of 16.1 years (S.D. = 1.9). The depressed group included 12 women and 3 men, ranging in age from 18 to 65 (M = 40.6, S.D. = 13.0), with a mean

Prior to beginning the physiology recording, individuals were asked to complete a battery of self-report questionnaires. The Beck Depression Inventory (BDI) was used to assess participants’ severity of depression (Beck et al., 1961). The Beck Hopelessness Scale (BHS) was used to measure participants’ expectations regarding future events (Beck and Weissman, 1974). The Spielberger State/Trait Anxiety Inventory (STAI) was employed as an index of participants’ state and trait anxiety (Spielberger et al., 1970). The Dysfunctional Attitudes Scale (DAS) provided a measure of cognitive distortions (Weissman, 1980). Participants also marked visual analog scales (21 cm) of Happiness (HAP), ranging from ‘‘the happiest I have ever been’’ to ‘‘not happy at all,’’ and Sadness (SAD), ranging from ‘‘the saddest I have ever been’’ to ‘‘not sad at all’’. As expected, diagnostic groups differed significantly on all self-report measures (see Table 1). 2.3. Procedure As outlined below, each individual participated in four 6 min, eyes-closed EEG recording blocks, and engaged in an introduction to both cognitive views of depression and Table 1 Participant self-report scores by diagnostic group, mean (S.D.) Measure

Control (n = 18)

Major depression (n = 15)

BDI ** BHS** STAI-S** STAI-T** DAS** HAP (cm)** SAD (cm)**

2.2 (.06) 2.1 (2.6) 27.5 (4.5) 29.3 (5.7) 218 (23.9) 13.3 (2.8) 1.5 (2.7)

23.8 (8.3) 13.2 (4.2) 47.5 (12.7) 55.9 (13.7) 164 (39.6) 6.6 (4.2) 11.6 (4.9)

**

p < .01.

144

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

methods used in cognitive psychotherapy to restructure negative thoughts. Baseline 1: Subjects were instructed to relax while a baseline EEG was recorded. Think 1: For the second recording block, subjects were asked to think about the ‘‘difficulty in your life that is troubling you the most’’. Cognitive restructuring: Prior to the third recording block, each participant engaged in a two-part introduction to cognitive restructuring. The information provided was based upon cognitive theories and therapy of depression (Beck et al., 1979; Burns, 1999). The first 10–15 min of the session involved introducing participants to cognitive views of depression and to types of negative or unrealistic thinking. Specifically, participants were introduced to the idea that distorted cognitions can contribute to negative moods, and that three types of distorted cognition are ‘‘all-or-nothing thinking,’’ ‘‘magnification,’’ and ‘‘should statements’’. The latter half of the session included a description of several methods of identifying and changing negative thinking. Participants were given strategies to identify unrealistic or negative thinking and were informed that three ways to restructure such thinking include: (a) assigning a number between 0 and 100 to the severity of the difficulty, (b) examining the facts by making a list of the ways negative thoughts could be false, and (c) substituting neutral phrases for emotionally loaded words. The entire session lasted approximately 30 min, during which the experimenter was seated next to the participant with a small table between. To ensure full understanding of the information, several reality-based examples were provided with each concept, subjects were given a review sheet, and prior to the next EEG recording session, participants were asked to review the material verbally with the experimenter. If necessary, additional examples were provided, and if they chose, participants were given the opportunity to practice identifying and restructuring negative thoughts. For full text of the cognitive restructuring protocol, please contact the authors. Think 2: During the third recording session, participants were asked to think again about the same life difficulty they had identified previously in Think 1. They were further asked to try to use the information provided in the cognitive restructuring session to identify and restructure specific negative thoughts about the difficulty. Baseline 2: Participants were instructed to relax while a post-task baseline EEG was recorded. Finally, participants again marked visual scales of Happiness and Sadness, and also completed a questionnaire to determine adherence to the task. Participants were then thanked and fully debriefed. 2.4. Physiological recording EEG was recorded from F3, F4, P3, and P4 using a lycra stretchable cap (manufactured by Electro-Cap, International

Inc., Eaton, OH) with electrodes positioned according to the international 10–20 system. During recording, all sites were referenced to vertex1 (Cz). Physiological signals were amplified and filtered through an S.A. Instruments Custom 37/64 Bioamp polygraph. In order to facilitate artifact scoring, electrooculogram (EOG) data were recorded using tin electrodes placed lateral to the outer canthi, and above and below the left supraorbital and suborbital positions. High- and low-pass analog filter settings were set at 0.01 Hz and 30 Hz, respectively. Digital sampling occurred at 512 Hz for 6 min during each of the four recording blocks of the study. All electrode impedances were kept below 10 kV. Participants were instructed to keep their eyes closed during each recording block. 2.5. Data reduction and analysis Combined EEG and EOG data were visually inspected to identify ocular and muscular artifact. When artifact occurred in a given channel, data from all channels were rejected (Barlow, 1986). In order to compute power density of the alpha (8–12 Hz) band, Fast Fourier Transform (FFT) using James Long Company Software was applied to all windows of artifact-free data that were 1 s in duration, with FFT windows overlapping by 50%. The FFT was performed for each 6 min recording block. Statistical analyses were conducted on log (ln) transformed alpha power density for each 6 min recording block. Asymmetry scores were calculated for frontal and parietal regions, for each recording block, as the difference between the log alpha power density in the right hemisphere lead and log alpha power density in the homologous left hemisphere lead (e.g., ln F4 alpha power ln F3 alpha power). Higher asymmetry scores thus indicate greater relative left hemisphere activity. 2.6. Statistical analyses In order to examine whether particular patterns of cortical activity were associated with improved mood in response to our intervention, scores of pre-intervention happiness were subtracted from post-intervention happiness, and the difference was used as a measure of mood response to the cognitive intervention. This difference score will subsequently be referred to as the ‘‘happiness response’’ 1 The vertex reference was chosen based on its extensive use in studies of EEG and emotion at the time the study was conducted. Since the study was implemented, some papers have noted the potentially problematic use of a sole Cz reference and cited the preferential use of other reference montages (e.g., averaged mastoids) or concurrent reporting of analysis from several different montages (e.g., Davidson, 1998; Hagemann et al., 1998, 2001; Reid et al., 1998). However, our recording montage at the time does not allow for re-referencing to another montage (i.e., averaged mastoids) for comparison. Although no reference montage has been definitively found to be ideal (Coan and Allen, 2003; Hagemann et al., 2001), the potential impact of the vertex reference is further noted in Section 4.

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

with greater positive values indicating greater self-reported happiness after the intervention than before. In order to assess the predictive utility of EEG with regard to mood improvement, those participants who reported greater postthan pre-intervention happiness were classified as ‘‘Responders.’’ Similarly, those who showed negative values of happiness response, or who showed no change in happiness response, were classified as ‘‘Non-responders.’’ An inspection of the range of scores reported on the Sadness scale revealed that a majority of control participants reported both pre- and post-task scores of zero on the Sadness scale, indicating that no change in sadness was measurable. Thus, analyses differentiating Responders and Non-responders using the Sadness scale were conducted only with the depressed group. Separate 2  4  2  2 repeated measures ANOVAs were conducted to investigate Response status (Responders, Non-responders)  Block (Baseline 1, Think 1, Think 2, Baseline 2)  Region (frontal, parietal)  Laterality (left, right) as well as Diagnosis (control, major depression)  Block  Region  Laterality. In addition, 2  2 repeated measures ANOVAs for Response status  Region and Diagnosis  Region were conducted for cortical asymmetry scores (ln right alpha ln left alpha). Pearson correlations were computed between degree of happiness response and total alpha power at each site2 (F3, F4, P3, and P4) during each recording block (Baseline 1, Think 1, Think 2, Baseline 2). All correlations were computed both within diagnostic groups and with diagnostic groups combined. Logistic regression analyses were conducted to assess the predictive utility of baseline alpha power with regard to identifying those individuals who are likely to exhibit increased happiness after the cognitive intervention task (i.e., ‘‘Responders’’). In order to account for multicollinearity between activity within cortical regions, new variables ‘‘frontal’’ (ln F3 alpha + ln F4 alpha) and ‘‘parietal’’ (ln P3 alpha + ln P4 alpha) were entered as predictor variables for the categorical dependent variable of happiness Response status (i.e., whether or not an increased happiness score was observed after our intervention). Finally, a mean split by alpha power was conducted in order to determine the specificity (1 false positive rate, i.e., rate at which alpha level misidentifies those who exhibit mood improvement) and sensitivity (true positive rate, i.e., rate at which alpha level accurately identifies mood response) of using alpha power to differentiate Responders to our intervention. A discriminant function analysis was performed to confirm and augment the above analyses to assess whether EEG alpha variables would correctly discriminate those who exhibit mood improvement from those who do not. 2

Given the number of sites recorded, we were unable to compute power at individual sites residualized for variance associated with whole-head power (Pivik et al., 1993; Wheeler et al., 1993).

145

3. Results As the happiness response (i.e., change in positive affect; (post-restructuring happiness minus pre-restructuring happiness)) reported by controls and depressives was not significantly different, F(1, 32) = 1.3, p > .1, participants for analyses incorporating mood Response status were grouped according to response on the Happiness scale. Mood ‘‘Responders’’ are those who exhibited positive values of happiness response; individuals who reported no-change or negative happiness response values are classified as ‘‘Nonresponders’’. Thirty-nine percent of the control participants (7 of 18) and 60% of the depressed participants (9 of 15) were thus classified as Responders. As expected, happiness response scores differed significantly between Responders and Non-responders, F(1, 32) = 23.9, p < .0001. See Table 2 for self-report scores by Response status and diagnostic group. 3.1. EEG topography In order to confirm that our physiological data were consistent with that reported in the literature, we examined the scalp distribution of our EEG recordings. An ANOVA investigating Diagnosis  Region  Laterality revealed a Table 2 Participant self-report scores by diagnostic group and Response status, mean (S.D.) Measure

Non-responder

Responder

All participants BDI+ BHS+ STAI-S* STAI-T DAS HAP (cm)* SAD (cm)+

8.3 (9.4) 5.3 (4.9) 31.9 (11.6) 37.3 (14.0) 192 (35.8) 12.4 (4.6) 3.9 (5.5)

15.6 9.4 42.9 47.1 188.6 8.4 8.3

n

17

Control BDI BHS STAI-S+ STAI-T DAS * HAP (cm)+ SAD (cm) n

n +

16 (2.7) (1.3) (4.3) (5.2) (23.9) (2.6) (2.6)

11

Depressed BDI+ BHS* STAI-S STAI-T+ DAS+ HAP (cm)+ SAD (cm)

*

2.4 2.3 26.0 31.3 206.2 14.4 1.3

18.8 10.6 42.6 48.2 161.0 8.6 8.8 6

p  .05. p  .1.

(14.8) (7.5) (15.2) (20.2) (47.9) (4.2) (6.5)

1.5 1.8 30.3 27.3 230.5 12.1 2.1

(2.8) (1.1) (3.8) (7.0) (17.4) (2.7) (3.0)

7 (7.3) (4.6) (13.3) (18.9) (41.4) (5.3) (6.4)

27.7 16.0 53.7 64.1 152.7 5.2 13.6 9

(8.1) (2.3) (12.5) (6.7) (32.9) (2.6) (2.4)

146

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

between happiness response score and alpha power at all sites and blocks, indicating that an increased level of overall cortical activity is associated with greater happiness response (see Table 3). The depressed group, as reported in Table 3, demonstrated significant or marginally significant negative correlations between happiness response and alpha power at frontal sites during most recording blocks. Although the correlations between happiness response and cortical activity at parietal sites did not achieve traditional levels of significance, the magnitude of the Pearson correlation coefficients (ranging from .27 to .49) indicate a medium to large effect size. Thus with a larger sample size, significance is likely to be achieved. No correlations were observed between cortical activity and mood improvement on the Sadness scale within the depressed group. Within the control group, as shown in Table 3, significant negative correlations were also observed between happiness response and alpha power at all sites and during each recording block except Baseline 2. No significant correlations were observed between happiness response scores and alpha asymmetry values. Fig. 1. Log alpha power (uV2) by site and recording block for each diagnostic group. Note: less alpha indicates greater cortical activity.

3.3. Absolute alpha power main effect of Region, F(1, 32) = 125, p < .0001, indicating that, as expected, EEG alpha was greater in parietal compared with frontal regions. No other effects or interactions emerged. Mean EEG alpha power within the control and depression groups by task block is provided in Fig. 1. Groups did not differ according to the number of FFT windows retained per condition (mean (S.D.), controls: Baseline 1 = 652.17 (70.64); Think 1 = 623.00 (61.81); Think 2 = 610.06 (98.83); Baseline 2 = 651.18 (59.95). Major depression: Baseline 1 = 612.87 (106.54); Think 1 = 633.53 (85.36); Think 2 = 621.00 (119.54); Baseline 2 = 629.33 (84.55)).

The 2  4  2  2 repeated measures ANOVAs investigating Response status  Block  Region  Laterality revealed a main effect of Response status, indicating that Responders exhibit decreased alpha power (i.e., increased cortical activity) at all cortical sites, Response status: F(1, 32) = 5.3, p < .03. Responders’ and Non-responders’ mean alpha power (with standard deviations in parentheses) were 2.28 (0.2) and 3.14 (0.2), respectively. Although analyses including the Diagnosis  Response status interaction term did not reach statistical significance ( p > .2), post hoc analyses of Responders and Non-responders within each diagnostic group suggest that the decreased alpha power of Responders was carried by the depressed Responders (control Response status: F(1, 17) = 1.2, ns; depressed Response status: F(1, 14) = 3.6, p < .09). No significant effects emerged in the ANOVA investigating Diagnosis  Block  Region  Laterality.

3.2. Correlational analyses between happiness response and alpha power When the control and depressed participants were examined together, significant negative correlations emerged

Table 3 Pearson correlations between ln alpha power and happiness response at each recording site and block Site

All participants Baseline 1

F3 F4 P3 P4 * ** +

**

.48 .51 ** .45 * .43 *

Depressed participants

Think 1 **

.47 .49 ** .40 * .39 *

p  .05. p  .01 (two-tailed). p  .1.

Think 2 **

.53 .56** .49** .47**

Baseline 2 **

.46 .49 ** .39 * .35+

Baseline 1 *

.57 .58 * .49 .40

Think 1 +

.51 .55+ .37 .31

Control participants Think 2 *

.58 .64 * .49 .42

Baseline 2

Baseline 1

+

*

.53 .58 * .42 .27

.52 .54* .51* .58*

Think 1 +

.48 .45+ .47+ .55 *

Think 2 *

.55 .51 * .54 * .61 **

Baseline 2 .31 .35 .26 .35

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

3.4. Asymmetry differences In order to investigate baseline differences in patterns of alpha power between Responders and Non-responders, an ANOVA investigating Response status  Region was performed for asymmetry scores, F(1, 32) = 4.3, p < .05. Further breakdown indicated that Responders and Nonresponders exhibited opposite patterns of frontal asymmetry with the Responders exhibiting negative asymmetry scores (i.e., ln F4 < ln F3 alpha power) and Non-responders showing positive asymmetry scores (i.e., ln F4 > ln F3), F(1, 32) = 5.47, p < .03. Responders’ and Non-responders’ mean alpha asymmetry differences (ln F4 ln F3; standard deviations in parentheses) were .04 (0.1) and .07 (0.1), respectively. Although analyses including the Diagnosis  Response status interaction term did not reach statistical significance ( p > .3), post hoc examination of control and depressed Responders and Non-responders suggest that the asymmetry effect in Responders was carried primarily by the asymmetry differences between depressed Responders and Non-responders (control: F(1, 17) = .55, ns; depressed: F(1, 14) = 6.2, p < .03). No further differences emerged in ANOVAs investigating either Response status  Region or Diagnosis  Region for baseline cortical asymmetry scores. 3.5. Predictive utility of alpha power The predictive utility of baseline alpha power for determining happiness response was first examined through logistic regression analyses. Analyses indicate that both frontal and parietal activity are of predictive utility with regard to identifying those who are likely to respond to the mood intervention3 (frontal entered first: Step 1 x2 = 9.1, p < .003; Step 2 x2 = .01, ns; parietal entered first: Step 1 x2 = 7.2, p < .007; Step 2 x2 = 2.0, ns). Specifically, the odds ratios indicate that for every unit increase in level of cortical activity, an individual has between 41% (1 .59; parietal) and 47% (1 .53; frontal) chance of exhibiting mood improvement in response to the cognitive intervention. The potential utility of alpha power for identifying those likely to exhibit a positive happiness response was further evaluated using the mean of baseline alpha power as a cutoff score. This split was successful in differentiating Responders to our intervention. Though the specificity (1 falsepositive rate, i.e., percentage of ‘‘Responders’’ identified by alpha power who actually were ‘‘Responders’’) was just above chance (56%), the sensitivity (true positive rate, i.e., percentage of all ‘‘Responders’’ who were accurately identified by alpha power) was moderately high (81%), further indicating that examining pre-treatment cortical activity may be of significant clinical utility. 3

It should be noted that some shrinkage of predictive accuracy is expected when implementing logistic regression with small sample sizes (Cohen et al., 2003; Copas, 1997).

147

As the above analyses do not identify a particular cortical site to be of predictive utility (rather, baseline EEG across all sites is implicated), baseline alpha at sites F3, F4, P3, and P4 were included in the discriminant function analyses. Seventy-five percent of the Non-responders, and 87.5% of the Responders were correctly identified. Overall classification was 81.3% (Wilks l = 0.586, x2 = 14.962, p < .01) indicating that the EEG predictors differentiated between the two response groups. All four of the variables contributed significantly to the classification.

4. Discussion Most striking was the presence and predictive utility of overall greater cortical activity for identifying those who reported greater post- than pre-intervention happiness compared with those who did not report such an improvement. Several lines of research may provide insight into this finding. First, a decrease of alpha rhythm (increase in cortical activity) has consistently been observed during task solution (i.e., Dolce and Waldeier, 1974; Rugg and Dickens, 1982; Gutierrez and Corsi-Cabrera, 1988; Ramos et al., 1993), and it may be argued that those who reported a mood improvement were merely more engaged in the restructuring task. However, the Responders showed decreased alpha power even at pre-task baseline, with no additional decrease during either unguided or guided thinking. This specificity suggests that overall cortical activity is a pre-existing, task-independent factor that may be associated with, and of predictive utility with regard to, mood reactivity. Second, if unilateral relative activity of the right or left hemisphere is consistent with the preferential experience and expression of negative and positive emotions (e.g., Wheeler et al., 1993; Tomarken et al., 1992), perhaps bilateral activity, as observed in the present study, is accompanied by an advantage for the experience and expression of emotion more generally. If so, such an advantage may, in part, moderate the mood improvement reported by this group. Indeed, the anhedonia and psychomotor retardation characteristic of depressed individuals may be attributed to a deficit in the ‘‘approach’’ subset of positive emotion (Davidson et al., 1990). Thus, it may be that those who exhibited overall greater activity were also relatively less impaired in processes that lead to the experience of emotion more generally. Although the small sample size and preliminary nature of this study hinder the meaningful analysis of moderator effects of baseline EEG upon therapeutic response (Baron and Kenny, 1986; McClelland and Judd, 1993), our data provide intriguing support for the inclusion of EEG or other biological measures in addition to the psychosocial client characteristics that have traditionally been examined as potential predictors of treatment outcome in cognitive therapy for depression (Whisman, 1993).

148

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

It is also of interest that the depressed Responders exhibited a baseline frontal asymmetry of greater right than left activity compared with depressed Non-responders. As the Responders reported greater scores on a range of scales assessing depressogenic symptoms, this pattern of cortical activity is consistent with evidence that an asymmetry toward greater right relative to left activity in frontal areas is both associated with greater negative affect and also observed in individuals in remission from depression (for review, see Davidson, 1995, 1998; Henriques and Davidson, 1990). However, the observed mood reactivity in our sample also seems inconsistent with the literature suggesting that individuals with such patterns of frontal asymmetry are predisposed to respond to negative stimuli with greater negative affect (Davidson, 1995, 1998). Notably, these studies largely employ passive viewing tasks that use asymmetry scores to predict the degree of affect elicited by valenced stimuli. As the current study assesses mood reactivity in response to a cognitive restructuring task, the substantial difference between our task and traditional tasks renders it difficult to draw a direct comparison; indeed, no investigations to our knowledge have examined the mechanisms of therapeutic mood response compared with those of affect elicited by valenced stimuli. Of importance, in an explicit mood induction procedure, Gotlib et al. (1998) found that, contrary to expectations, greater pre-induction right frontal cortical activity did not predict the degree of induced sadness. They posit that brain systems subsuming the production of ‘internally generated’ emotion may differ from those involved with emotion elicited in response to external stimuli. The asymmetry patterns observed herein may also be compared with the findings of Bruder et al. (1996, 1997) that on a dichotic listening task, Responders to both cognitive therapy and fluoxetine exhibited a right ear (left hemisphere) advantage and left ear (right hemisphere) disadvantage, respectively. Bruder et al. (1997) further posited that dichotic listening paradigms are likely to involve processes in more posterior temporoparietal regions which would be consistent with the pattern of parietal activity in depression proposed by Heller et al. (1995). In the current study, parietal asymmetry differences between Responders and Non-responders were not observed. Together, the present findings suggest that absolute cortical activity may be integral to mood improvement, and provide further evidence that frontal patterns of cortical asymmetry may be an index of depressive state or increased risk for depression (e.g., Davidson, 1995; Henriques and Davidson, 1990), as evidenced by the greater right frontal activity exhibited by the participants reporting more severe depression symptoms. Finally, no changes in patterns of cortical asymmetry were observed between any of the four recording blocks. Several plausible interpretations may be provided for this observation. First, the task may have not affected preexisting patterns of cortical activity. Indeed, the evidence that EEG patterns shift with mood reactivity is inconsistent

(for review, see Hagemann et al., 1998). Second, the cognitive restructuring task may have required activation of other cortical or sub-cortical regions than we measured, or may have not elicited changes in alpha power. Third, mood responsivity could be a trait characteristic, one of whose identifying factors is a relatively greater overall cortical activity that predisposes an individual to mood reactivity. It should also be noted that our data indicate no between-group differences in bilateral baseline cortical activity in controls and individuals with depression, a finding that at first seems inconsistent with the bulk of the literature pointing to frontal hypoactivity in depression (for review, see Pollock and Schneider, 1990). However, given that our data were recorded as participants were meeting a new experimenter and anticipating the beginning of a new task, that depressed individuals may have been more alert to the testing conditions, and that anticipatory motivation may decrease alpha suppression (increase cortical activity; e.g., Zinser et al., 1999), a direct comparison of our data to this literature must be tentative. Although our findings clearly highlight the multi-factorial nature of depression, the limitations in our protocol provide motivation for further investigation. First, the use of the active vertex reference may distort EEG asymmetry calculations (for discussion, see Hagemann et al., 2001); nevertheless, our data are consistent with findings of relatively greater right frontal activity in individuals with greater depression severity. Unfortunately, our recording montage at the time does not allow for off-line calculation of an alternate reference (e.g., averaged ears) for comparison. Also, the lack of a placebo group and the use of a single experimenter to conduct the cognitive restructuring task render it difficult to assess whether the observed mood response was a consequence of our cognitive restructuring task, a spontaneous mood improvement, a result of a pleasant interaction, or an artifact of the experimental situation. Refining and expanding the current post-task questionnaires to address more specifically the nature of mood improvement and of each participant’s thought process during both the unguided and structured thinking tasks would facilitate the interpretation of our data. Further, although our task was based upon the techniques of cognitive therapy for depression (Beck et al., 1979; Burns, 1999), and a significant increase in happiness was observed in a majority of the depressed participants subsequent to our cognitive intervention, it should be emphasized that the task is a brief, one-time intervention in a laboratory setting. Indeed, given the importance of interpersonal factors not only in the maintenance of depressed mood (e.g., Segrin and Abramson, 1994; for review, see Ingram et al., 1996), but also in positive therapeutic outcome (e.g., Burns and Auerbach, 1996; Arnow et al., 2003; Klein et al., 2003), greater rapport between the experimenter and some participants may have, in part, contributed to positive mood response in these participants. Thus, any generalization of our brief intervention to cognitive therapy, or of our results to the therapeutic process, must be cautious.

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

4.1. In conclusion Pre-treatment therapeutic empathy, REM latency, dichotic listening asymmetry, perceived helpfulness of the therapy, and homework compliance are among the characteristics that have been implicated as predictors of response to cognitive therapy for depression (for reviews, see Scott, 1996; Salkovskis, 1996; Whisman, 1993). The diversity of findings regarding characteristics that may predict treatment outcome emphasizes the many interactions that contribute to major depression and highlights how little is known about variables that may account for individual differences in the response to treatment. We suggest that EEG activity may be of clinical utility with regard to establish a patient profile for assigning treatment protocols. We further propose that as an index of general affective reactivity, global cortical activity has largely, and surprisingly, been overlooked in favor of investigating of asymmetry metrics that reflect individual differences in emotional response tendencies. In sum, it is increasingly evident that a systems approach toward establishing a patient profile of positive therapeutic response is necessary. Indeed, it is highly unlikely that symptom alleviation at a cognitive, emotional, biological, or other level occurs exclusive of changes in the remaining systems. Clarifying the interaction among these systems is critical to the understanding of the maintenance of depression, and to the refining and development of more effective and efficient treatments. Our findings add to the growing literature and provide evidence that baseline levels and patterns of cortical activity may be of predictive utility with regard to identifying a subset of depressed individuals who have an advantage for mood improvement.

Acknowledgements This research was made possible by NIMH B/START Grant #1R03M H57694-01 to Patricia J. Deldin, as well as grants from the Harvard College Research Program and Harvard Committee on Mind, Brain, and Behavior to Pearl Chiu. We gratefully acknowledge the technical assistance of James Long.

References Allen, J., Harmon-Jones, E., Cavender, J., 2001. Manipulation of frontal EEG asymmetry through biofeedback alters self-reported emotional responses and facial EMG. Psychophysiology 38, 685–693. Arnow, B., Manber, R., Blasey, C., Klein, D., Blalock, J., Markowitz, J., et al., 2003. Therapeutic reactance as a predictor of outcome in the treatment of chronic depression. Journal of Consulting and Clinical Psychology 71, 1025–1035. Baehr, E., Rosenfeld, P., Baehr, R., 1997. The clinical use of an alpha asymmetry protocol in the neurofeedback treatment of depression: two case studies. Journal of Neurotherapy 2, 10–23.

149

Barlow, J., 1986. Artifact processing (rejection and minimization) in EEG data processing. In: Lopes da Silva, F.H., Storm van Leeuwen, W., Remond, A. (Eds.), Handbook of Electroencephalography and Clinical Neurophysiology, vol. 2. Elsevier, Amsterdam, revised series ed., pp. 15–62. Baron, R., Kenny, D., 1986. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51, 1173–1182. Beck, A., Rush, A., Shaw, B., Emery, G., 1979. Cognitive Therapy of Depression. Guilford Press, New York. Beck, A., Ward, C., Mendelson, M., Mock, J., Erbaugh, J., 1961. An inventory for measuring depression. Archives of General Psychiatry 4, 561–571. Beck, A., Weissman, A., 1974. The assessment of pessimism: the hopelessness scale. Journal of Consulting and Clinical Psychology 42, 861– 865. Bell, M., Fox, N., 2003. Cognition and affective style: individual differences in brain electrical activity during spatial and verbal tasks. Brain and Cognition 53, 441–451. Bruder, G., Otto, M., McGrath, P., Stewart, J., 1996. Dichotic listening before and after fluoxetine treatment for major depression: relations of laterality to therapeutic response. Neuropsychopharmacology 15, 171– 179. Bruder, G., Stewart, J., Mercier, M., Agosti, V., Leite, P., Donovan, S., Quitkin, F., 1997. Outcome of cognitive–behavioral therapy for depression: relation to hemispheric dominance for verbal processing, Journal of. Abnormal Psychology 106, 138–144. Burns, D., 1999. The Feeling Good Handbook. William Morrow Company, New York. Burns, D., Auerbach, A., 1996. Therapeutic empathy in cognitive–behavioral therapy: does it really make a difference? In: Salkovskis, P. (Ed.), Frontiers of Cognitive Therapy. Guilford Press, New York, pp. 35–164. Castonguay, L., Goldfried, S., Raue, P., Hayes, A., 1996. Predicting the effect of cognitive therapy for depression: a study of unique and common factors. Journal of Consulting and Clinical Psychology 64, 497–504. Coan, J., Allen, J., 2003. Frontal EEG asymmetry and the behavioral activation and inhibition systems. Psychophysiology 40, 106–114. Cohen, J., Cohen, P., West, S., Aiken, L., 2003. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Erlbaum Associates, Mahwah, New Jersey. Copas, J., 1997. Using regression models for prediction: shrinkage and regression to the mean. Statistical Methods in Medical Research 6, 167– 183. Davidson, R., 1995. Cerebral asymmetry, emotion, and affective style. In: Davidson, R., Hughdahl, K. (Eds.), Brain Asymmetry. MIT Press, Cambridge, pp. 361–388. Davidson, R., 1998. Affective style and affective disorders: perspectives from affective neuroscience. Cognition and Emotion 12, 307–330. Davidson, R., Ekman, P., Saron, C., Senulis, J., Friesen, W., 1990. Approachwithdrawal and cerebral asymmetry: emotional expression and brain physiology I. Journal of Personality and Social Psychology 58, 330–341. Davidson, R., Schwartz, G., Saron, C., Bennett, J., Goleman, D., 1979. Frontal vs. parietal EEG asymmetry during positive and negative affect. Psychophysiology 16, 202–203. DeRubeis, R., Crits-Christoph, P., 1998. Empirically supported individual and group psychological treatments for adult mental disorders. Journal of Consulting and Clinical Psychology 66, 37–52. Dobson, K., 1989. A meta-analysis of the efficacy of cognitive therapy for depression. Journal of Consulting and Clinical Psychology 57, 414–419. Dohr, K., Rush, J., Bernstein, I., 1989. Cognitive biases and depression. Journal of Abnormal Psychology 98, 263–267. Dolce, G., Waldeier, H., 1974. Spectral and multivariate analysis of EEG changes during mental activity in man. Electroencephalography and Clinical Neurophysiology 36, 577–584.

150

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151

Dujardin, K., Derambure, P., Defebvre, L., Bourriez, J., Jacqesson, J., Guieu, J., 1993. Evaluation of event-related desynchronization during a recognition task: effect of attention. Electroencephalography and Clinical Neurophysiology 86, 353–356. Earle, J., 1988. Task difficulty and EEG alpha asymmetry: an amplitude and frequency analysis. Neuropsychobiology 20, 96–112. Elkin, I., Shea, T., Watkins, J., 1989. National Institute of Mental Health Treatment of Depression Collaboration Treatment Program. Archives of General Psychiatry 49, 971–982. Evans, M., Hollon, S., DeRubeis, R., Piaseck, J., Grove, W., Garvey, M., Tuason, V., 1992. Differential relapse following cognitive therapy and pharmacotherapy for depression. Archives of General Psychiatry 49, 802–808. Fava, M., Bless, E., Otto, M., Pava, J., Rosenbaum, J., 1994. Dysfunctional attitudes in major depression: changes with pharmacotherapy. Journal of Nervous and Mental Disease 182, 45–49. First, M., Spitzer, R., Gibbon, M., Williams, J., 1995. Structured Clinical Interview for DSM-IV Axis I Disorders-Patient Edition (SCIP-I/P, Version 2.0) American Psychiatric Press, Washington, DC. Gevins, A., 1998. The future of electroencephalography in assessing neurocognitive functioning. Electroencephalography and Clinical Neurophysiology 106, 165–172. Gevins, A., Smith, M., 2000. Neurophysiological measures of working memory and individual differences in cognitive ability and cognitive style. Cerebral Cortex 10, 829–839. Gotlib, I., Ranganath, C., Rosenfeld, J.P., 1998. Frontal EEG alpha asymmetry, depression, and cognitive functioning. Cogn. Emotion 12, 449– 478. Gutierrez, S., Corsi-Cabrera, M., 1988. EEG activity during performance of cognitive tasks demanding verbal and/or spatial processing. International Journal of Neuroscience 42, 149–155. Haaga, D., Dyck, M., Ernst, D., 1991. Empirical status of cognitive theory of depression. Psychological Bulletin 110, 215–236. Hagemann, D., Naumann, E., Becker, G., Maier, S., Bartussek, D., 1998. Frontal brain asymmetry and affective style: a conceptual replication. Psychophysiology 35, 372–388. Hagemann, D., Naumann, E., Thayer, J., 2001. The quest for the EEG reference revisited: a glance from brain asymmetry research. Psychophysiology 38, 847–857. Harmon-Jones, E., 2003. Clarifying the emotive functions of asymmetrical frontal cortical activity. Psychophysiology 40, 838–848. Heller, W., Etienne, M., Miller, G., 1995. Patterns of perceptual asymmetry in depression and anxiety: implications for neuropsychological models of emotion and psychopathology. Journal of Abnormal Psychology 104, 327–333. Henriques, J., Davidson, R., 1990. Regional brain electrical asymmetries discriminate between previously depressed and healthy control subjects. Journal of Abnormal Psychology 99, 22–31. Ingram, R., Scott, W., Siegle, G., 1996. Depression: social and cognitive aspects. In: Millon, T., Blaney, P., Davis, R. (Eds.), Oxford Textbook of Psychopathology. Oxford University Press, New York, pp. 203–226. Jacobson, N., Dobson, K., Truax, P., Addis, M., Koerner, K., Gollan, J., Gortner, E., Prince, S., 1996. A component analysis of cognitive– behavioral treatment for depression. Journal of Consulting and Clinical Psychology 64, 295–304. Jarrett, R., Nelson, R., 1987. Mechanisms of change in cognitive therapy of depression. Behavior Therapy 18, 227–241. Klein, D., Schwartz, J., Santiago, N., Vivian, D., Vocisano, C., Castonguay, L., et al., 2003. Therapeutic alliance in depression treatment: controlling for prior change and patient characteristics. Journal of Consulting and Clinical Psychology 71, 997–1006. Klimesch, W., 1999. EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Research Reviews 29, 169–195. McClelland, G., Judd, C., 1993. Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin 114, 376– 390.

McEvoy, L., Smith, M., Gevins, A., 2000. Test–retest reliability of cognitive EEG. Clinical Neurophysiology 111, 457–463. Miller, G., Keller, J., 2000. Psychology and neuroscience: making peace. Current Directions in Psychological Science 9, 212–215. Molle, M., Marshall, L., Wolf, B., Fehm, H., Born, J., 1999. EEG complexity and performance measures of creative thinking. Psychophysiology 36, 95–104. Norcross, J., Santrock, J., Campbell, L., Smith, T., Sommer, R., Zuckerman, E., 2003. Depression. In: Authoritative Guide to Self-Help Resources in Mental Health, Guilford Press, New York, pp. 187–201. Oldfield, R., 1971. The assessment and analysis of handedness: the Edinburgh Inventory. Neuropsychologia 9, 91–113. Otto, M., Pava, J., Sprich-Buckminster, S., 1996. Treatment of major depression: applications and efficacy of cognitive–behavioral therapy. In: Pollack, M., Otto, M., Rosenbaum, J. (Eds.), Challenges in Clinical Practice: Pharmacologic and Psychosocial Strategies. Guilford Press, New York, pp. 31–52. Peselow, E., Robins, C., Block, P., Barouche, F., Fieve, R., 1990. Dysfunctional attitudes in depressed patients before and after clinical treatment and in normal control subjects. American Journal of Psychiatry 147, 439–444. Pivik, T., Broughton, R., Coppola, R., Davidson, R., Fox, N., Nuwer, R., 1993. Guidelines for quantitative electroencephalography in research contexts. Psychophysiology 30, 547–558. Pollock, V., Schneider, L., 1990. Quantitative, waking EEG research in depression. Biological Psychiatry 27, 757–780. Ramos, J., Corsi-Cabrera, M., Guevara, M., Arce, C., 1993. EEG activity during cognitive performance in women. International Journal of Neuroscience 69, 185–195. Ray, W., Cole, H., 1985. EEG alpha reflects attentional demands, and beta activity reflects emotional and cognitive processes. Science 228, 750– 752. Reid, S., Duke, L., Allen, J., 1998. Resting frontal electroencephalographic asymmetry in depression: inconsistencies suggest the need to identify mediating factors. Psychophysiology 35, 389–404. Rosenfeld, J., Cha, G., Blair, T., Gotlib, I., 1995. Operant (biofeedback) control of left–right frontal alpha power differences: potential neurotherapy for affective disorders. Biofeedback and Self-Regulation 20, 241–248. Rugg, M., Dickens, A., 1982. Dissociation of alpha and theta activity as a function of verbal and visuo-spatial tasks. Electroencephalography and Clinical Neurophysiology 53, 201–207. Sackheim, H., Greenberg, M., Weissman, A., Gur, R., Hungerbuler, J., Geschwind, N., 1982. Hemispheric asymmetry in the expression of positive and negative emotions: neurologic evidence. Archives of Neurology 39, 210–218. Salkovskis, P. (Ed.), 1996. Frontiers of Cognitive Therapy. Guilford Press, New York. Savage, C., 1993. Information processing and interhemispheric transfer in left- and right-handed adults. International Journal of Neuroscience 71, 201–219. Scott, J., 1996. Cognitive therapy of affective disorders: a review. Journal of Affective Disorders 37, 1–11. Segrin, C., Abramson, L., 1994. Negative reactions to depressive behaviors: a communication theories analysis. Journal of Abnormal Psychology 11, 43–70. Spielberger, C., Gorsuch, R., Lushene, R., 1970. Manual for the State Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto. Thase, M., Kupfer, D., Fasiczka, A., Buysse, D., 1997. Identifying an abnormal electroencephalographic sleep profile to characterize major depressive disorder. Biological Psychiatry 41, 964–973. Thase, M., Simons, A., 1992. Cognitive behavior therapy and relapse of nonbipolar depression: parallels with pharmacotherapy. Psychopharmacology 28, 117–122. Thase, M., Simons, A., Reynolds, C., 1996. Abnormal electroencephalographic sleep profiles in major depression: association with response to cognitive behavior therapy. Archives of General Psychiatry 53, 99–108.

P.J. Deldin, P. Chiu / Biological Psychology 70 (2005) 141–151 Tomarken, A., Davidson, R., Wheeler, R., Doss, R., 1992. Individual differences in anterior brain asymmetry and fundamental dimensions of emotion. Journal of Personality and Social Psychology 62, 676–687. Weissman, A., 1980. Dysfunctional attitudes scale. In: Fischer, J., Corcoran, K. (Eds.), Measures for Clinical Practice: A Sourcebook, 1994. Free Press, New York, pp. 187–190. Wheeler, R., Davidson, R., Tomarken, A., 1993. Frontal brain asymmetry and emotional reactivity: a biological substrate of affective style. Psychophysiology 30, 82–89.

151

Whisman, M., 1993. Mediators and moderators of change in cognitive therapy of depression. Psychological Bulletin 114, 248–265. Williams, J., Watts, F., MacLeod, C., Mathews, A., 1997. Cognitive Psychology and Emotional Disorders. Wiley, New York. Zinser, M., Fiore, M., Davidson, R., Baker, T., 1999. Manipulating smoking motivation: impact on an electrophysiological index of approach motivation. J. Abnorm. Psychol. 108, 240–254.