Cognitive Therapy and Research

, Volume 41, Issue 4, pp 519–529 | Cite as

Change in Explanatory Flexibility and Explanatory Style in Cognitive Therapy and its Components

  • Michael T. Moore
  • David M. Fresco
  • Jeremiah A. Schumm
  • Keith S. Dobson
Original Article


The current study represents a secondary analysis of the dismantling study of cognitive therapy of depression originally conducted by Jacobson et al. (J Consult Clin Psychol 64:295–304, 1996). New analyses examined the role of explanatory flexibility and explanatory style in the recovery from depression. Results indicated that BA treatment responders, but not AT or CT participants evidenced significant improvement in explanatory flexibility, whereas patients from all three study arms, irrespective of responder status demonstrated improvements in explanatory style. Improvement in explanatory flexibility was associated with decreases in symptoms of depression for CT, but not BA or AT, participants. Further, the combination of high explanatory flexibility and low explanatory style conferred maximal protection over relapse. These results suggest that explanatory flexibility is a viable candidate as a process associated with treatment gains in CT. In addition, the results suggest that important cognitive change is possible without an explicit, deliberate focus on the part of the therapist.


Cognitive therapy Depression Explanatory flexibility Explanatory style 


Cognitive diathesis-stress theories of depression (Abramson et al. 1978; Beck 1967, 1976) have advanced our understanding of the etiology, maintenance, and treatment of depression in a number of ways. These theories posit that vulnerability to depression arises through early life experiences that lead one to develop a depressogenic view of the world. Specifically, the Reformulated Learned Helplessness Theory (Abramson et al. 1978) and Hopelessness Theory (Abramson et al. 1989) both conceptualize vulnerability to depression in terms of a depressogenic or pessimistic explanatory style (the tendency to view negative events as arising from stable, global, and internal causes). Similarly, Beck’s (1967, 1976) theory of depression posits that vulnerability to depression is associated with dysfunctional attitudes and negative schema regarding the self, world and future.

Cognitive diathesis-stress theories of depression have informed and influenced efforts to develop psychotherapies that include techniques to teach individuals how to identify and challenge pessimistic causal attributions for actual events (Seligman 1980) or dysfunctional thoughts (Beck et al. 1979). Empirical findings consistently support the efficacy of cognitive therapy of depression (DeRubeis et al. 1990; Elkin et al. 1989; Evans et al. 1992; Segal et al. 1999), which in turn, are consistent with the cognitive change hypothesis.

The specifics and mechanisms of cognitive change in cognitive therapy remain a topic of great interest in the field and a detailed review is beyond the scope of the current study. Still, several points raised during the discussion of cognitive change are important to the current study. Specifically, what facets of cognition are the appropriate targets for change in cognitive therapy? Hollon et al. differentiate between two main kinds of cognitions: cognitive structures and cognitive products (Ingram and Hollon 1986; Hollon and Garber 1988; Hollon and Kriss 1984; Kendall and Ingram 1989).1 Cognitive structures represent “the way or manner in which information is represented in memory” (Ingram and Hollon 1986, p. 263). Cognitive structures play an active role in the processing of information. Schemata represent a form of cognitive structure important to the cognitive theories and therapies for depression. By contrast, cognitive products represent directly accessible, conscious thoughts such as self-statements, automatic thoughts, and causal attributions. Such products result from the processing of sensory input information through cognitive structures.

The distinction between cognitive structures and cognitive products is important with respect to cognitive therapy of depression. For example, theorists caution that targeting cognitive products will yield limited clinical utility, as such interventions amount to symptomatic treatments (Hollon and Kriss 1984; Safran et al. 1986). Rather, Beck et al. (1979) state explicitly that changes in cognitive structures or core schema represent critical change mechanisms in cognitive therapy. Safran et al. (1986) assert that efforts at cognitive change should focus on core processes. Finally, Beck (1984) cautioned that depressed individuals would remain vulnerable for relapse when underlying cognitive structures were not targeted and changed.

Although the focus of the Beck et al. (1979) approach is on the assessment and treatment of dysfunctional cognitions, some of the treatment methods do not explicitly target cognition. For example, the first component of Beck et al. manual devotes considerable time to what has come to be termed behavioral activation (BA). This component consists of a suite of behaviorally oriented, non-cognitive skills (e.g., monitoring of daily activities, assessment of pleasure and mastery, role-plays and cognitive rehearsal of activities, etc.; see Jacobson et al. 1996 for a more complete description) designed to engage people in their natural environment. Interestingly, patients treated with cognitive therapy often report improvements as well as cognitive change in the early sessions—before the introduction of explicit cognitive strategies (Rush et al. 1977).

Component Analysis of Cognitive Therapy of Depression

Jacobson et al. (1996) conducted a component analysis experiment designed to evaluate the degree to which the major components of cognitive therapy of depression (CT; Beck et al. 1979) were associated with recovery from depression as well as cognitive change. In that study, 150 patients with current major depression were randomly assigned to one of three treatments: a treatment focused exclusively on the behavioral activation (BA) component of CT, a treatment that included both BA and the teaching of skills to modify automatic thoughts (AT), but excluding the components of CT focused on core schema, or the full CT treatment. Of relevance to the current study, the AT manual specifically identified negative causal attributions as a potential target for change, and encouraged clinicians to identify and modify them. The acute treatment phase lasted 20 weeks and all treatments received behavioral activation albeit in different amounts. Similarly, the techniques included in the AT condition were consistent with Beck et al. (1979), but were delivered more frequently relative to the CT condition. Using a definition of recovery as no longer meeting diagnostic criteria for major depression and being asymptomatic on a measure of clinician-assessed depression, all three treatments demonstrated equivalent success by the end of the acute treatment phase. There was also equivalent success in preventing relapse across the three conditions in the 2-years follow-up period (Gortner et al. 1998). Jacobson et al. (1996) also examined how change in how causal attributions was associated with symptom change. They found that these variables were related for participants in BA, but not CT. Many other studies of CT have examined cognitive products, such as causal attributions, as mediators or moderators of treatment outcome (e.g., Barber and DeRubeis 1989; DeRubeis et al. 1990; Seligman et al. 1988). Recently, evidence is accruing related to a variable thought to be a cognitive process counterpart to attributional or explanatory style: explanatory flexibility.

Explanatory Flexibility

A finding that has emerged across an array of the biomedical and psychological literatures is the role of rigidity (e.g., physiological, behavioral, cognitive) in the pathogenesis of physical and psychological disorders (Brosschot and Thayer 2004; Thayer and Lane 2002; Wilson and Murrell 2004). The list of candidate processes associated with a rigid approach range from decreased heart rate variability, to biases in the allocation of attention, to perseverative thinking (e.g., worry, rumination), to behavioral inactivity or inflexible patterns of behavior (Brosschot and Thayer 2004; Thayer and Lane 2002; Wilson and Murrell 2004). Similarly, rigidity has demonstrated an association to heart disease and hypertension, as well as anxiety and mood disorders (Thayer and Lane 2002).

One way to account for the emergence of rigidity as a response to threatening or stressful circumstances is in its survival value from an evolutionary perspective (Wilson and Murrell 2004). In case of threat, the organism capable of narrowing its behavioral repertoire, allocating attention resources to the danger, and exhibiting high autonomic arousal will have a greater likelihood of fighting, escaping from, or remaining undetected when confronted with a predator or other danger. Similarly, in the case of depression or reinforcement deprivation, an organism which finds itself in an environment devoid of positive reinforcement (actual or perceived) will likely make a frantic effort to escape and, failing to do so, may become behaviorally inactive -possibly in an effort to preserve energy, until the circumstances have changed (Wilson and Murrell 2004). Thus, although rigidity can represent functional responses to idiosyncratic environmental circumstances, rigid response styles are often emotionally maladaptive, in that behavioral inactivity does nothing to actively change the stimuli presented by threatening environments or environments lacking positive reinforcers. This conceptualization of the etiology and treatment of emotional problems is more consistent with traditional (Ferster 1973; Lewinsohn 1974) and contemporary (Hayes et al. 1999; Jacobson et al. 2001; Linehan 1993) behavioral approaches to psychopathology than cognitive models, in that they emphasize a functional analysis to promote less reliance on escape or avoidance behaviors, as well as to promote behaviors associated with positive reinforcement. In essence, the goal of behavioral treatment approaches is to help individuals cultivate a flexible and versatile array of behaviors so that they can maximize access to positive reinforcement.

Fresco et al. (2006, 2007b) have proposed a cognitive process construct called explanatory flexibility. Fresco et al. (2007b) defined explanatory flexibility as the degree to which individuals balance their interpretation of events with historical and current contextual factors and make effective use of all available information. Explanatory flexibility has been operationalized as the variability of an individual’s responses to the stable and global items for negative events from the Attributional Style Questionnaire (ASQ; Peterson et al. 1982). Fresco et al. (2007b) based this measure on the stable and global dimensions of the ASQ to make explanatory flexibility comparable to the hopelessness theory (Abramson et al. 1989), which places greater emphasis on the stability and globality dimensions and refers to this index as generality. A small standard deviation is conceptualized as a reflection of less flexibility or more rigidity, whereas a large standard deviation is interpreted as an indication of flexibility. In a sample of 78 college students, Fresco et al. (2007b) found explanatory flexibility to be only modestly correlated (r = −0.25) with generality. Importantly, explanatory flexibility, but not generality, moderated the relationship of negative life events to levels of depression measured 8 weeks later. The effect size (f2 = 0.09) for the interaction term approached Cohen’s (1988) convention for a medium effect. The pattern of this interaction was such that there was a strong association between the number of intervening life events and subsequent depression for participants with low explanatory flexibility (i.e., rigidity), whereas no association between negative life events and subsequent depression was observed with high explanatory flexibility. The interaction of explanatory flexibility and negative life events remained even when generality was controlled for in the model, as a main effect and in interaction with negative life events.

Fresco et al. (2006) examined the effects of a mood priming challenge on explanatory flexibility. Dysphoric participants with a history of depression endorsed a more stable and global explanatory style for negative events (generality) before the mood priming challenge as compared to euthymic participants with a history of depression and never depressed participants. The latter two groups did not differ from one another on baseline generality. Further, the three groups did not differ from one another on baseline explanatory flexibility. Following the mood priming challenge, euthymic participants with a history of depression experienced drops in explanatory flexibility (i.e., became more rigid) whereas dysphoric participants with a history of major depression and never depressed participants did not. The magnitude of the change in explanatory flexibility corresponded to a large effect size (Cohen’s f = 0.34). Conversely, dysphoric participants with a history of major depression experienced increases in generality (f = 0.24) whereas euthymic participants with a history of major depression and never depressed participants did not. These results suggest that the interaction of cognitive reactivity and explanatory flexibility might represent a trait-like risk factor for the recurrence of depression in a manner that differentiates explanatory flexibility from explanatory style.

Subsequent research has expanded upon what is known about explanatory flexibility. Moore and Fresco (2007) replicated prior work demonstrating that explanatory flexibility and style, while related, were empirically distinct (r = −0.27). Fresco et al. (2014) that university students with self-reported generalized anxiety disorder (GAD), became more rigid in their causal attributions in response to a negative mood prime. Healthy control participants, however, did not demonstrate this cognitive reactivity to the mood prime. Finally, Lackner et al. (2015), in the first study of explanatory flexibility in a clinical sample, found that individuals who met DSM-IV criteria for GAD or major depressive disorder (MDD) demonstrated lower explanatory flexibility than individuals who met criteria for other Axis I disorders. Despite the increasing amount of research on explanatory flexibility, an important question that remains to be answered is the role of flexibility in the context of treatment.

Current Study

The current study is a secondary analysis of Jacobson et al. (1996) clinical trial. The original series of two studies demonstrated that the components of Beck et al. (1979) cognitive therapy of depression produced comparable rates of recovery and relapse prevention. Further, the treatments did not result in differential changes in hypothesized mechanisms, including explanatory style. The current study does not restate these findings, but extends these findings by focusing on the relationship of treatment modality and treatment outcome to explanatory style and explanatory flexibility. Specifically, although participants in all three conditions were evaluated, the current study made predictions regarding changes in cognitive style following acute treatment in a protocol that offers a high concentration of non-cognitive interventions (BA) versus a treatment that provides a high concentration of interventions to identify and dispute distorted cognitive products (AT). Specifically, given that the AT condition offered a concentration of cognitive interventions designed to target and change cognitive products, it was hypothesized that response to treatment in the AT condition would be associated with decreased pessimistic explanatory style. Conversely, consistent with the traditional behavioral emphasis on increased flexibility and behavioral approach, it was hypothesized that response to treatment in the BA condition would be associated with increased explanatory flexibility. The current study also sought to examine the relationship of post-treatment levels of explanatory style and explanatory flexibility to depression relapse.



Details about participant selection, exclusionary criteria, assignment procedures, therapy conditions, inter-rater reliabilities, and assessment of outcome have been described previously (see Gortner et al. 1998; Jacobson et al. 1996). The sample consisted of 150 participants with MDD according to criteria of the Diagnostic and Statistical Manual of Mental Disorders (3rd edition, revised [DSM-III-R]; American Psychiatric Association 1987) assessed with the Structured Clinical Interview for the DSM-III-R (Spitzer et al. 1987), a score of 12 or higher on the Hamilton Rating Scale for Depression (HRSD; Hamilton 1967), and a score of 20 or higher on the Beck Depression Inventory (BDI; Beck et al. 1988). Eligible participants were also matched on a number of variables (e.g., gender, number of previous depressive episodes). Participants were then randomly assigned to one of three treatment conditions: BA, AT, or CT. Participants met individually with one of four experienced therapists who were trained to provide treatment in all of the three conditions. Treatment interventions entailed up to 20 sessions over a 16-weeks period. Symptoms of depression, diagnostic status, and cognitive measures were assessed at pre- and post-treatment. In addition, diagnostic status was assessed at 2-years follow-up.


All three treatment conditions are based on techniques and procedures described in detail elsewhere (see Beck et al. 1979; Jacobson et al. 1996). The BA condition focused on helping participants identify specific life problems and engage in semi-structured activities that contained naturally occurring reinforcement agents. The AT condition incorporated all the components of the BA condition along with the identification and modification of dysfunctional patterns of thinking that were related to specific situations. Techniques used in the AT condition included the identification and monitoring of dysfunctional thinking, challenging the validity of such thoughts, empirically testing beliefs, and practicing more functional responses related to thought patterns. The CT condition included all components of the BA and AT conditions, and added techniques related to the identification and modification of generalized core beliefs that are presumed to be causally related to dysfunctional beliefs and depressive reactions (Beck et al. 1979).

Cognitive Measures

The Expanded Attributional Style Questionnaire (EASQ; Peterson and Villanova 1988) is a self-report measure that assesses causal attributions for 24 hypothetical negative events along the dimensions of internality, stability, and globality. The first six hypothetical events are retained from the original Attributional Style Questionnaire (Peterson et al. 1982) with the addition of 18 new hypothetical negative events. Consistent with hopelessness theory (Abramson et al. 1989), generality scores were computed by averaging the values of stable and global items on the EASQ to provide a composite measure of depressogenic explanatory style with higher scores representing greater pessimism.

Explanatory flexibility was computed as the standard deviation of stable and global items. In the current study, the internal consistency based on the stable and global items for all 24 negative hypothetical events was very good (pre-treatment: α = 0.90; post-treatment: α = 0.93). However, to be directly comparable to previous studies (Fresco et al. 2006, 2007a, b, 2009), which used the original ASQ, composite scores for generality and explanatory flexibility were computed from the first six stable and global items on the EASQ. Thus, generality was computed as the average of the 12 stable and global items; explanatory flexibility was computed as the standard deviation of those same 12 items. Internal consistencies for the abbreviated scales were acceptable (pre-treatment: α = 0.76; post-treatment: α = 0.81) and they were highly correlated with their longer counterparts (pre-treatment generality, r = 0.86; pre-treatment explanatory flexibility, r = 0.76; post-treatment generality, r = 0.88; post-treatment explanatory flexibility, r = 0.77). Similar to the results reported by Fresco et al. (2007b), generality and explanatory flexibility were modestly correlated with one another at pre- (r = −0.38) and post-treatment (r = −0.34).

Outcome Measures

The Longitudinal Interview Follow-up Evaluation (LIFE-II; Keller et al. 1987) is a semi-structured interview used to measure participants’ recovery from or occurrence of new depressive episodes and the occurrence of relapse in depression.

The Hamilton Rating Scale for Depression (HRSD; Hamilton 1967) is a clinician-administered measure of depression symptom severity that correlates highly with the Beck Depression Inventory (BDI; Beck et al. 1979). The present study used the 17-item version of the HRSD.

Data Analyses

Analyses were conducted only with participants who had completed at least 12 of their 20 therapy sessions and who had complete posttest and follow-up data. As previously reported, 12-months attrition rates using these criteria were only 8% and were comparable across treatment conditions (Gortner et al. 1998). For the purposes of the current study, treatment responder status was defined as post-treatment HRSD score of <8 and minimal or no depressive symptoms on the LIFE-II interview.2 These criteria correspond to the definition of depression recovery developed by Frank et al. (1991).


Exploratory Analyses of Missing EASQ Data

Although the earlier published reports from this study described outcomes for a sample of 150 participants, there were considerable missing data (n = 36) for the EASQ at the post-acute treatment assessment session. Thus, participants were compared to test differences among patients with missing and completed post-treatment EASQ data. Analyses revealed that individuals with missing EASQ data did not differ statistically from participants with completed EASQ data on pretreatment HRSD scores [M = 19.56, SD = 5.01, M = 18.18, SD = 3.83 respectively, t(149) = 1.75, p = 0.08, Cohen’s d = 0.29], post-treatment responder status χ2(1, N = 147) = 1.67, p = 0.19, as well as demographic characteristics such as number of children [M = 0.77, SD = 0.94, M = 0.92, SD = 1.10 respectively, t(147) = −0.72, p = 0.46, d = 0.12], marital status χ2(4, N = 140) = 6.90, p = 0.14, education χ2(6, N = 147) = 4.50, p = 0.61, or ethnicity χ2(4, N = 143) = 3.44, p = 0.49. Further, treatment groups did not significantly differ on the number of participants with missing post-treatment EASQ data. There was no missing pre-treatment, and little missing post-treatment, HRSD data (n = 10; 7%).

Pre-Treatment Group Differences

A one-way MANOVA was conducted to test pretreatment group differences on cognitive measures (i.e., generality and explanatory flexibility scores). The omnibus test revealed no significant group differences for both treatment group, F(4, 264) = 0.93, p = 0.45, f = 0.12, as well as responder status, F(2, 131) = 0.59, p = 0.56, f = 0.09.

Treatment Condition and Responder Effects on Change in Cognitive Measures3

To test the effects of treatment condition (3) and responder status (2) on pre- to post-treatment change in cognitive measures, one mixed-design MANOVA was conducted using generality and explanatory flexibility as the dependent measures. The findings of these analyses were evaluated using customary standards of significance (e.g., p < 0.05) as well as Cohen’s (1988) conventions for small (f = 0.10, d = 0.20), medium, (f = 0.25, d = 0.50), and large (f = 0.40, d = 0.80) effects. The univariate tests (reported below) were only interpreted if a parent omnibus test was statistically significant. The Shapiro–Wilk W statistic was used to evaluate if pre- and post-treatment measures of explanatory flexibility and style were normally distributed. The results of these tests indicated that this assumption of the ANOVA was met. In addition, Mauchly’s tests of sphericity indicated that this assumption was met for the within-subjects effects for explanatory flexibility and style. Results for explanatory flexibility revealed significant time by condition, F(2, 132) = 3.59, p < 0.03, f = 0.23, and time by recovery status interactions, F(1, 132) = 5.56, p < 0.02, f = 0.20 (see Table 1 for descriptive statistics). Main effects for time, F(1, 132) = 0.002, p = 0.97, f < 0.001, condition, F(2, 132) = 0.48, p = 0.62, f = 0.08, recovery status, F(1, 132) = 0.98, p = 0.33, f = 0.08, the interactions of condition and recovery status, F(2, 132) = 0.35, p = 0.70, f = 0.07, and the three-way interaction of all three variables, F(2, 132) = 1.31, p = 0.27, f = 0.14, were all nonsignificant. Follow-up analyses, decomposing the significant two-way interactions, revealed that patients receiving BA experienced increases in explanatory flexibility that, while not statistically significant, t(52) = 1.67, p = 0.10, were associated with a close to medium effect, d = 0.46. However, participants receiving AT, t(39) = 0.66, p = 0.52, d = 0.21, and CT, t(45) = −1.21, p = 0.23, d = 0.36, did not experience significant changes in explanatory flexibility during active treatment. With regard to responder status, responders experienced slight increases in explanatory flexibility, t(76) = 1.58, p = 0.12, d = 0.36, while nonresponders experienced slight decreases in flexibility, t(60) = −1.52, p = 0.13, d = 0.39, with both of these findings not meeting criteria for statistical significance.

Table 1

Means and (standard deviations) of acute treatment changes in generality and explanatory flexibility by treatment condition and clinician-assessed responder status



Post-acute Tx





4.82 (0.68)

4.49 (0.82)

4.19 (0.64)


4.76 (0.68)

5.05 (0.74)

4.03 (0.82)


4.86 (0.71)

4.53 (0.66)

4.24 (0.76)



1.42 (0.41)

1.48 (0.42)

1.56 (0.49)


1.52 (0.40)

1.62 (0.38)

1.51 (0.40)


1.56 (0.40)

1.47 (0.50)

1.51 (0.31)

*p < 0.05; BA Behavioral Activation, AT Behavioral Activation plus Automatic Thoughts Challenging, CT Full suite of Cognitive Therapy, GEN Generality, FLEX Explanatory flexibility

Results for generality revealed significant main effects of time, F(1, 132) = 44.00, p < 0.001, f = 0.58, recovery status, F(1, 132) = 7.23, p = 0.008, f = 0.23, and significant interactions of time and recovery status, F(1, 132) = 16.28, p < 0.001, f = 0.35, and condition and recovery status, F(2, 132) = 4.12, p < 0.02, f = 0.25. The main effect of condition, F(2, 132) = 0.50, p = 0.61, f = 0.09, interactions of time and condition, F(2, 132) = 0.57, p = 0.57, f = 0.10, and the three-way interaction of all three variables, F(2, 132) = 0.09, p = 0.92, f = 0.03, were all nonsignificant. Generality decreased significantly in all participants, regardless of condition or responder status, t(138) = −7.10, p < 0.001, d = 1.21. Responders demonstrated significant decreases in generality, t(76) = −7.39, p < 0.001, d = 1.69, while nonresponders evidenced smaller decreases, t(60) = −2.39, p = 0.02, d = 62.

Association between Changes in Explanatory Flexibility, Generality, and Symptom Change During Acute Treatment

Similar to the analyses conducted by Jacobson et al. (1996), we sought to examine the association between changes in explanatory flexibility and symptom change during acute treatment in the three treatment arms. Given that the current study represents the first examination of explanatory flexibility in the context of treatment for depression, these analyses should be viewed as exploratory and designed to inform future research. Jacobson et al. found that change in the stable and global items from the EASQ correlated significantly with HRSD symptom change for BA (stable: r = 0.45, p < 0.01; global: r = 0.38, p < 0.05), but not CT (stable: r = 0.03, ns; global: r = 0.22, ns). Inconsistent with the previously reported findings, pre- to post-treatment residual change in flexibility was not correlated with residual change in HRSD scores in the BA, r(52) = 0.03, p = 0.90, or AT conditions, r(39) = −0.14, p = 0.44, but approached statistical significance in the CT condition, r(46) = −0.25, p = 0.09. Decomposing the results for CT further, CT responders were compared to nonresponders with regard to the correlation between change in flexibility and symptom change. Although an association between these two variables was found for CT responders that was statistically significant at the trend level, r(29) = −0.33, p = 0.08, no such relationship was found for CT nonresponders, r(17) = 0.02, p = 0.92. Change in generality scores from pre- to post-treatment was only significantly associated with change in HRSD scores for participants receiving AT, r(39) = 0.39, p < 0.02. For participants receiving BA, the association between change in generality and depression was significant at a trend level, r(52) = 0.26, p = 0.06, and was not statistically significant for those receiving CT, r(46) = 0.18, p = 0.23. Averaging across all treatment conditions, a statistically significant correlation was found between pre- to post-treatment change in explanatory flexibility and style, r(139) = −0.35, p < 0.001. This relationship was strongest for participants in the BA, r(53) = −0.42, p = 0.002, and CT conditions, r(46) = −0.43, p = 0.003, and not statistically significant for patients receiving AT, r(40) = −0.19, p = 0.31.

Effects of Post-Treatment Cognitive Measures on Durability of Treatment Gains

The effects of post-treatment cognitive measures in predicting the durability of treatment gains through 2-years follow-up was examined for treatment responders. Well weeks were defined as weeks in which participants demonstrated no or minimal depressive symptomatology on the LIFE-II. Relapse was defined as meeting DSM-III-R criteria for depression for at least two consecutive weeks. The number of well weeks ranged from 0 to 104 weeks.

Given the relative novelty of the explanatory flexibility construct, we opted to examine the relationship of both cognitive measures in association with one another with respect to the durability of treatment gains. A Cox proportional-hazards survival analysis was computed to estimate the impact of post-treatment cognitive style on time until relapse. Post-treatment explanatory flexibility and generality and their cross-product were entered into the model as covariates. To examine the assumption of proportionality of hazards, the interactions between the natural logarithm of time until relapse and all covariates were examined (Tabachnik and Fidell 2013). After adjustment for multiple comparisons, none of these interactions terms were statistically significant, indicating that this assumption was met.

A survival function was derived first without any predictors (−2 Log Likelihood = 273.15). Entering the covariates into the model resulted in a drop in the survival function value (−2 Log Likelihood = 266.59). The difference was tested using a Chi square statistic with k degrees of freedom (Allison 1984). The covariates resulted in a near significant improvement in the prediction of survival, χ2(3) = 6.57, p = 0.087. In looking at the individual predictors, post-treatment explanatory flexibility, B = −6.15, SE = 2.68; Wald(1) = 5.27, p = 0.02, and explanatory style, B = −1.77, SE = 0.89; Wald(1) = 3.97, p = 0.05, significantly added to the prediction of survival. These main effects were qualified by a significant interaction, B = 1.44, SE = 0.63; Wald(1) = 5.23, p = 0.02. However, as Allison (1984) explains, statistical power is affected by both sample size and the number of cases that are censored or in this case, who survived the entire 2-years follow-up period (n = 46; 31%). Thus, this interaction was interpreted and the Cox Regression model was solved at either + 1 or −1 SD of the sample mean on post-treatment generality, and either + 1 or −1 SD of the sample mean on post-treatment explanatory flexibility. Solving the regression equation in this manner estimated the proportional hazards as a function of low generality/low flexibility (Rigid Optimists), low generality/high flexibility (Flexible Optimists), high generality/high flexibility (Flexible Pessimists), and high generality/low flexibility (Rigid Pessimists). Findings revealed that lower pessimism combined with higher flexibility was associated with the most durable treatment gains. Conversely, a combination of high pessimism and high explanatory flexibility was associated with the highest rates of relapse (see Fig. 1).

Fig. 1

Deconstruction of the interaction of Time 2 generality and Time 2 explanatory flexibility from Cox proportional-hazards survival analysis predicting time to relapse


Given the emphasis on increasing flexibility consistent with a behavioral approach, BA alone was predicted to be associated with the largest increases in explanatory flexibility and that this change would be associated with decreases in depression. Given the focus on changing cognitive products, AT was predicted to be associated with the largest changes in explanatory style (generality) which would, in turn, be associated with symptom improvement. These hypotheses were partially supported by the data. No statistically significant pre-treatment differences were found, on generality or explanatory flexibility, between conditions. While BA was not associated with statistically significant increases in explanatory flexibility, this result corresponded to a medium effect. AT and CT, however, were neither associated with statistically significant, nor clinically meaningful, change in explanatory flexibility. Although not statistically significant, treatment responders, regardless of condition, experienced improvements in explanatory flexibility with an effect size in the medium range. Counter to predictions, however, increases in explanatory flexibility were associated (albiet at trend level) with decreases in depression for participants receiving CT (with this relationship being stronger for CT responders than nonresponders). Also counter to predictions, all treatment conditions experienced roughly equivalent decreases in generality (with treatment responders experiencing a larger decrease than nonresponders). Consistent with predictions, drops in generality were most strongly related to drops in depression for AT participants. With regard to relapse, both explanatory flexibility and style were better predictors than either variable alone; highly flexible and optimistic individuals were the least likely to relapse. Finally, results were consistent with past research (e.g., Moore and Fresco 2007) demonstrating that explanatory flexibility and style are empirically related, but distinct, constructs.

The current study represents one of a number of first steps in the exploration of the interaction of cognitive content and process variables in CT. Whereas all of the components of CT seemed equally capable of producing changes in cognitive content, BA seemed particularly well-suited for producing change in our variable of cognitive process, explanatory flexibility. Traditional behavioral approaches view depression as a combination of maladaptive escape or avoidance behaviors, as well as a lack of behavioral responses capable of producing positive reinforcement (Ferster 1973; Lewinsohn 1974). Thus, a functional analytic approach to depression focuses on decreasing the reliance on escape or avoidance behaviors, while increasing an individual’s behavioral repertoire to increase the availability of positive reinforcement. The BA condition used in the current study is a precursor to the more elaborated model and treatment manual of Martell et al. (2001). However, the work of Martell et al. (2001) may provide a framework for understanding how the BA condition in the current study may be particularly associated with increases in explanatory flexibility. Building on the work of Ferster (1973), the Martell et al. (2001) BA intervention conceptualizes depressed individuals as having developed a narrow repertoire of behavior that predominantly features escape or avoidance of aversive stimuli and consequences. Thus, a goal of treatment with the BA approach is to promote a broader repertoire of behaviors and to reduce the reliance on behaviors of escape or avoidance. This objective is accomplished, in part, by teaching clients to self-monitor their activities. Through this action, clients gain a greater awareness of the antecedents and consequences of their actions and, in doing so, recognize the function of certain behaviors. Although not as deliberately articulated as in Martell et al. (2001), the Jacobson et al. (1996) BA condition contained many of the interventions to promote a broader repertoire of behaviors and, relative to the AT and CT conditions, was provided in greater concentrations. Thus, one way to account for the increases in explanatory flexibility in BA is that 20 sessions of therapy designed to increase approach behaviors may have led to behavioral versatility, which is being reflected by increases in explanatory flexibility.

The lack of statistically significant change in explanatory flexibility for participants in CT, coupled with the association between increases in flexibility and decreases in symptoms of depression for this group, is puzzling and will require future research to adequately address. These findings are particularly intriguing given that links between decreases in generality and symptom change were found for participants receiving BA and AT, but not CT. The association between change in flexibility and symptoms of depression in CT suggests that something that differentiates this group from the other two, such as work on core beliefs, is responsible for the specificity of the association. The lack of significant increases in flexibility may indicate that CT is better able to translate cognitive change to symptom change than BA or AT. Other possibilities are equally plausible. For example, an unmeasured variable(s) (i.e., homework compliance, therapeutic alliance, decentering) could account for treatment response and/or the pattern of findings described above.

The results of the current study add to a growing body of literature demonstrating relationships between explanatory flexibility and psychological well-being. These results are also relevant to the more general claim that flexibility in cognition, emotion, behavior and physiology is adaptive. Kashdan and Rottenberg (2010), in their review, claim that individuals able to vary their emotional and behavioral responses to suit their environments are better able to respond to these incredibly complex, shifting, and variable contexts. They present evidence that both negative and positive emotions can be adaptive, that coping variability better predicts positive outcome than consistent use of any one coping strategy, and relationships between various forms of psychological inflexibility and psychopathology. While links between explanatory and coping flexibility have already been found (Fresco et al. 2006), how explanatory flexibility relates to other indices of psychological flexibility remains unanswered.

The current study examined links between explanatory flexibility and generality, a more traditional potential mediator of efficacy in CT. However, explanatory flexibility can also be conceptualized in the context of other potential mechanisms of action in CT, such as decentering (e.g., Ingram and Hollon 1986; Safran and Segal 1990). Decentering has been defined as the ability to observe one’s thoughts and feelings as temporary, objective events in the mind, as opposed to reflections of the self that are necessarily true. It is thought that the variability in causal attributions demonstrated in flexible individuals is a reflection of incorporating varying contextual information. It is also possible that it is a reflection of various cognitive and emotional components of decentering. Bernstein et al. (2015) describe a theoretical model of decentering, the Meta-Cognitive Process Model. In this model, decentering is posited to be composed of three components: meta-awareness, disidentification from internal experience, and reduced reactivity to thought content. Of these three, we would expect meta-awareness to be most strongly related to explanatory flexibility. Awareness of thoughts, feelings, and contexts, and the relationships between them should result in more variable causal attributions, in addition to the other two components of the model, but future research will be needed to test this hypothesis.


Although the findings from the present study are encouraging, they must be interpreted in light of the study’s limitations. The most notable limitation is that the study emerged as a secondary analysis of an existing dataset. The hypotheses of the current study were certainly not in the minds of the investigators of the original research. Thus, in some cases, statistical power and the manner in which the original data were collected do not permit more sophisticated modeling of the role of explanatory style and explanatory flexibility in the recovery process from depression. Still, the current findings warrant other studies that deliberately attempt to demonstrate how explanatory style and explanatory flexibility play a role in the treatment of depression. In addition, given the low attrition of individuals in the current study, results cannot be generalized to treatment non-completers. Finally, the results related to the association of cognitive and symptom changes were not guided by strong, a priori hypotheses. While this was due to the fact that research in explanatory flexibility is in its relative infancy, replication will be necessary to establish that the results of the current study represent broadly generalizable phenomena.

Future Research Directions

Findings from the current study add to a growing body of research suggesting that explanatory flexibility represents a distinct cognitive factor related to depression (Fresco et al. 2006, 2007b) and other emotional problems (Fresco et al. 2009). However, explanatory flexibility is a relatively recent variable in the literature on risk factors in depression, and requires further investigation before it can complement the extant literature of behavioral and cognitive approaches to the etiology and treatment of emotional disorders. Although the current study provides evidence that BA might provide a specific avenue for improving explanatory flexibility, studies using larger treatment samples are required to examine whether explanatory flexibility represents a unique mechanism of change in BA when compared to other treatment modalities in preventing depression relapse. It is possible that explanatory flexibility represents a transdiagnostic index of psychological well-being. Initial research has found a role for explanatory flexibility in disorders other than unipolar mood disorders (Lackner et al. 2015). However, future research will be needed to both highlight the precise role that explanatory flexibility plays and determine the utility of explanatory flexibility as an indicator of psychotherapy process in a variety of treatments that highlight behavioral and cognitive flexibility. Likewise, future studies are needed to clarify whether improvements in pessimistic attributional style are unique mechanisms of change in cognitive treatments when compared to other treatment modalities that protect against depression relapse.


  1. 1.

    See the primary source citations for a more elaborated and nuanced discussion of cognitive structures and cognitive products

  2. 2.

    Affirmative treatment response was also defined as post-treatment BDI scores of <9 and minimal or no depressive symptoms on the LIFE-II interview. Patterns of results were comparable in both classification schemes. For brevity of the current paper, these findings are omitted, but are available from the corresponding author

  3. 3.

    At the recommendation of an anonymous review, we also conducted a logistic regression with treatment response as the criterion and residual change in generality, explanatory flexibility, condition, and their interactions as covariates. The results of this analysis were that only residual change in generality served as a significant predictor, b = −1.07, SE = 0.63, Wald (df = 1) = 2.89, p < 0.001, while flexibility, condition, and the interactions between residual change in generality, flexibility, and condition, did not


Compliance with Ethical Standards

Conflict of Interest

Michael T. Moore, David M. Fresco, Jeremiah A. Schumm, Keith S. Dobson declare that they have no conflict of interest.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. For this type of study formal consent is not required.

Informed Consent

Informed consent was obtained from all individual participants included in the study.

Animal Rights

This article does not contain any studies with animals performed by any of the authors.


  1. Abramson, L., Metalsky, G. I., & Alloy, L. B. (1989). Hopelessness depression: A theory-based subtype of depression. Psychological Review, 96, 358–372.CrossRefGoogle Scholar
  2. Abramson, L., Seligman, M., & Teasdale, J. (1978). Learned helplessness in humans: Critique and reformulation. Journal of Abnormal Psychology, 87, 102–109.CrossRefPubMedGoogle Scholar
  3. Allison, P. D. (1984). Event History Analysis: Regression for Longitudinal Event Data. Newbury Park, CA: Sage Publications.CrossRefGoogle Scholar
  4. American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (3rd ed., rev.). Washington, DC: American Psychiatric Association.Google Scholar
  5. Barber, J. P., & DeRubeis, R. J. (1989). On second thought: Where the action is in cognitive therapy for depression. Cognitive Therapy and Research, 13, 441–457.CrossRefGoogle Scholar
  6. Beck, A. T. (1967). Depression: Causes and treatment. Philadelphia: University of Pennsylvania Press.Google Scholar
  7. Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International Universities Press.Google Scholar
  8. Beck, A. T. (1984). Cognition and therapy. Archives of General Psychiatry, 41, 1112–1114.CrossRefPubMedGoogle Scholar
  9. Beck, A. T., Rush, A. J., Shaw, B. F., & Emery, G. (1979). Cognitive therapy of depression. New York: Guilford.Google Scholar
  10. Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, 77–100.CrossRefGoogle Scholar
  11. Bernstein, A., Hadash, Y., Lichtash, Y., Tanay, G., Shepherd, K., & Fresco, D. M. (2015). Decentering and Related Constructs: A Critical Review and Metacognitive Processes Model. Perspectives on Psychological Science, 10(5), 599–617.
  12. Brosschot, J. F., & Thayer, J. F. (2004). Worry, preservative thinking, and health. In I. Nyklicek, L. Temoshok & A. Vingerhoets (Eds.), Emotional expression and health: Advances in theory, assessment, and clinical application (pp. 99–114). New York: Brunner and Rutledge.Google Scholar
  13. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd edn.). San Diego: Academic Press.Google Scholar
  14. DeRubeis, R. J., Evans, M. D., Hollon, S. D., Garvey, M. D., Grove, W. M., & Tuason, V. B. (1990). How does cognitive therapy work? Cognitive change and symptom change in cognitive therapy and pharmacotherapy for depression. Journal of Consulting and Clinical Psychology, 58, 862–869.CrossRefPubMedGoogle Scholar
  15. Elkin, I., Shea, M. T., Watkins, J. T., Imber, S. D., Sotsky, S. M., Collin, J. F., Glass, D. R., Pilkonis, P. A., Leber, W. R., Fiester, S. J., Docherty, J., & Parloff, M. B. (1989). National Institute of Mental Health Treatment of Depression Collaborative Research Program: General effectiveness of treatments. Archives of General Psychiatry, 46, 971–982.CrossRefPubMedGoogle Scholar
  16. Evans, M. D., Hollon, S. D., DeRubeis, R. J., Piasecki, J., Grove, W. M., Garvey, M. J., & Tuason, V. B. (1992). Differential relapse following cognitive therapy and pharmacotherapy for depression. Archives of General Psychiatry, 49, 802–808.CrossRefPubMedGoogle Scholar
  17. Ferster, C. B. (1973). A functional analysis of depression. American Psychologist, 28, 857–870.CrossRefPubMedGoogle Scholar
  18. Frank, E., Prien, R. F., Jarrett, R. B., Keller, M. B., Kupfer, D. J., Lavori, P. W., Rush, A. J., & Weissman, M. M (1991). Conceptualization and rationale for consensus definitions of terms in major depressive disorder: Remission, recovery, relapse, and recurrence. Archives of General Psychiatry, 48, 851–855.CrossRefPubMedGoogle Scholar
  19. Fresco, D. M., Heimberg, R. G., Abramowitz, A., & Bertram, T. L. (2006). The effect of a negative mood priming challenge on dysfunctional attitudes, explanatory style, and explanatory flexibility. British Journal of Clinical Psychology, 45, 167–183.CrossRefPubMedGoogle Scholar
  20. Fresco, D. M., Mennin, D.S., Hambrick, J. & Heimberg, R.G. (2009). The effect of a negative mood priming challenge on explanatory flexibility and state emotional intelligence in individuals with and without generalized anxiety disorder. Manuscript under review.Google Scholar
  21. Fresco, D. M., Moore, M. T., van Dulmen, M., Segal, Z. V., Teasdale, J. D., Ma, H., & Williams, J. M. G. (2007a). Initial psychometric properties of the Wider Experiences Questionnaire: A self-report survey of decentering. Behavior Therapy, 38, 234–246.CrossRefGoogle Scholar
  22. Fresco, D. M., Rytwinski, N. K., & Craighead, L. W. (2007b). Explanatory flexibility and negative life events interact to predict depression symptoms. Journal of Social and Clinical Psychology, 26, 595–608.Google Scholar
  23. Fresco, D. M., Williams, N. L., & Nugent, N. R. (2006). Association of explanatory flexibility and coping flexibility to each other and to depression and anxiety. Cognitive Therapy and Research, 30, 201–210.CrossRefGoogle Scholar
  24. Fresco, D. M., Moore, M. T., Mennin, D. S., Heimberg, R. G., & Hambrick, J. (2014). Changes in explanatory flexibility among individuals with generalized anxiety disorder in an emotion evocation challenge. Cognitive Therapy and Research, 38, 416–427.CrossRefGoogle Scholar
  25. Gortner, E. T., Gollan, J. K., Dobson, K. S., & Jacobson, N. S. (1998). Cognitive-behavioral treatment for depression: Relapse prevention. Journal of Consulting and Clinical Psychology, 66, 377–384.CrossRefPubMedGoogle Scholar
  26. Hamilton, M. (1967). Development of a rating scale for primary depressive illness. British Journal of Social and Clinical Psychology, 6, 276–296.CrossRefGoogle Scholar
  27. Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (1999). Acceptance and commitment therapy: An experiential approach to behavior change. New York: Guilford.Google Scholar
  28. Hollon, S. D., & Garber, J. (1988). Cognitive therapy. In L. Y. Abramson (Ed.), Social cognition and clinical psychology (pp. 204–253). New York: Guilford.Google Scholar
  29. Hollon, S. D., Kendall, P. C., & Lumry, A. (1986). Specificity of depressotypic cognitions in clinical depression. Journal of Abnormal Psychology, 95, 52–59.CrossRefPubMedGoogle Scholar
  30. Hollon, S. D., & Kriss, M. R. (1984). Cognitive factors in clinical research and practice. Clinical Psychology Review, 4, 35–76.CrossRefGoogle Scholar
  31. Ingram, R. E., & Hollon, S. D. (1986). Cognitive therapy for depression from an information processing perspective. In R. E. Ingram (Ed.), Information processing approaches to clinical psychology (pp. 259–281). San Diego: Academic Press.Google Scholar
  32. Jacobson, N. S., Dobson, K. S., Truax, P. A., Addis, M. E., Koerner, K., Gollan, J. K., Gortner, E., & Prince, S. E. (1996). A component analysis of cognitive-behavioral treatment for depression. Journal of Consulting and Clinical Psychology, 64, 295–304.CrossRefPubMedGoogle Scholar
  33. Jacobson, N. S., Martell, C.R., & Dimidjian, S. (2001). Behavioral activation treatment for depression: Returning to contextual roots. Clinical Psychology, 8, 255–270.Google Scholar
  34. Kashdan, T. B., & Rottenberg, J. (2010). Psychological flexibility as a fundamental aspect of health. Clinical Psychology Review, 30, 865–878.CrossRefPubMedPubMedCentralGoogle Scholar
  35. Keller, M. B., Lavori, P. W., Friedman, B., Nielsen, E., Endicott, J., McDonald-Scott, P., & Andreason, N. C. (1987). The longitudinal interval follow-up evaluation: A comprehensive method for assessing outcome in prospective longitudinal studies. Archives of General Psychiatry, 44, 540–548.CrossRefPubMedGoogle Scholar
  36. Kendall, P. C., & Ingram, R. E. (1989). Cognitive behavioral perspectives: Theory and research on depression and anxiety. In P. C. Kendall & D. Watson (Eds.), Anxiety and depression: Distinctive and overlapping features (pp. 27–53). San Diego: Academic Press.Google Scholar
  37. Lackner, R. J., Moore, M. T., Minerovic, J., & Fresco, D. M. (2015). Explanatory flexibility and explanatory style in treatment-seeking clients with Axis I psychopathology. Cognitive Therapy and Research, 39, 736–743.CrossRefPubMedPubMedCentralGoogle Scholar
  38. Lewinsohn, P. M. (1974). A behavioral approach to depression. In R. J. Friedman & M. M. Katz (Eds.), The psychology of depression: Contemporary theory and research. Oxford: John Wiley & Sons.Google Scholar
  39. Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York: Guilford Press.Google Scholar
  40. Martell, C. R., Addis, M. E., & Jacobson, N. S. (2001). Depression in context: Strategies for guided action. New York: Norton & Co.Google Scholar
  41. Moore, M. T., & Fresco, D. M. (2007). Depressive realism and attributional style: Implications for individuals at risk for depression. Behavior Therapy, 38, 144–154.CrossRefPubMedGoogle Scholar
  42. Peterson, C., Semmel, A., von Baeyer, C., Abramson, L. Y., Metalsky, G. I., & Seligman, M. E. P. (1982). The Attributional Style Questionnaire. Cognitive Therapy and Research, 6, 287–300.CrossRefGoogle Scholar
  43. Peterson, C., & Villanova, P. (1988). An expanded attributional style questionnaire. Journal of Abnormal Psychology, 97, 87–89.CrossRefPubMedGoogle Scholar
  44. Rush, A., Beck, A., Kovacs, M., & Hollon, S. (1977). Comparative efficacy of cognitive therapy and pharmacotherapy in the treatment of depressed outpatients. Cognitive Therapy and Research, 1, 17–37.CrossRefGoogle Scholar
  45. Safran, J. D., & Segal, Z. V. (1990). Interpersonal process in cognitive therapy. New York: Basic Books.Google Scholar
  46. Safran, J. D., Vallis, T. M., Segal, Z. V., & Shaw, B. F. (1986). Assessment of core cognitive processes in cognitive therapy. Cognitive Therapy and Research, 10, 509–526.CrossRefGoogle Scholar
  47. Segal, Z. V., Gemar, M., & Williams, S. (1999). Differential cognitive response to a mood challenge following successful cognitive therapy or pharmacotherapy for unipolar depression. Journal of Abnormal Psychology, 108, 3–10.CrossRefPubMedGoogle Scholar
  48. Seligman, M.E.P. (1980). A learned helplessness point of view. In L. Rehm (Ed.), Behavior therapy for depression (pp. 123–142). New York: Academic Press.Google Scholar
  49. Seligman, M. E., Castellon, C., Cacciola, J., Schulman, P., Luborsky, L., Ollove, M., & Downing, R. (1988). Explanatory style change during cognitive therapy for unipolar depression. Journal of Abnormal Psychology, 97(1), 13.CrossRefPubMedGoogle Scholar
  50. Spitzer, R. L., Williams, J. B. & Gibbon, M. (1987). Instruction manual for the Structured Clinical Interview for DSM—III—R. (Available from the Biometrics Research Department, New York, NY: New York State Psychiatric Institute.Google Scholar
  51. Tabachnik, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th edn.). Boston: Pearson.Google Scholar
  52. Thayer, J. F., & Lane, R. D. (2002). Perseverative thinking and health: neurovisceral concomitants. Psychology and Health, 17, 685–695.CrossRefGoogle Scholar
  53. Wilson, K. G., & Murrell, A. R. (2004). Values work in acceptance and commitment therapy: Setting a course for behavioral treatment. In S. C. Hayes, V. M. Follette & M. M. Linehan (Eds.), Mindfulness and acceptance: Expanding the cognitive-behavioral tradition (pp. 120–151). New York: Guilford.Google Scholar

Copyright information

© Springer Science+Business Media New York 2017

Authors and Affiliations

  1. 1.Gordon F. Derner Institute of Advanced Psychological StudiesAdelphi UniversityGarden CityUSA
  2. 2.Department of PsychologyKent State UniversityKentUSA
  3. 3.School of Professional PsychologyWright State UniversityDaytonUSA
  4. 4.Department of PsychologyUniversity of CalgaryCalgaryCanada

Personalised recommendations