Given that the cognitive theories of depression are vulnerability-stress models in which depressogenic cognitive styles are proposed to confer vulnerability to depression when individuals confront negative life events, it is important to evaluate the interaction between cognitive style and the occurrence of negative life events in predicting onset/recurrence of depression. As indicated earlier, several previous studies have found support for the vulnerability-stress hypothesis (Alloy & Clements, 1998; Alloy et al., 1997; Anderson, 1990; Metalsky et al., 1987,1993; Metalsky & Joiner, 1992; Nolen-Hoeksema et al., 1986, 1992; Panak & Garber, 1992;
FIG. 2.1. Prospective incidence rate of DSM-III-R major depressive disorder (left) and hopelessness depression (right) as a function of cognitive risk group status and Time 1 stress-reactive rumination. LR = Low Risk; HR = High Risk; Lo-SRR = Low Stress-Reactive Rumination; Hi-SRR = High Stress-Reactive Rumination.
DSM-/W-R M DD HOPELESSNESS DEPRESSION
FIG. 2.1. Prospective incidence rate of DSM-III-R major depressive disorder (left) and hopelessness depression (right) as a function of cognitive risk group status and Time 1 stress-reactive rumination. LR = Low Risk; HR = High Risk; Lo-SRR = Low Stress-Reactive Rumination; Hi-SRR = High Stress-Reactive Rumination.
Robinson, Garber, & Hilsman, 1995) in predicting depressive symptoms. However, there have been a handful of published studies that have found no support for this hypothesis (Cole & Turner, 1993; Joiner & Wagner, 1995; Tiggemann, Winefield, Winefield, & Goldney, 1991).
To date, little evaluation of the vulnerability-stress hypothesis has been conducted with the CVD project data. Preliminary investigation of this hypothesis using the Temple site data only, in which negative life events were coded as loss and/or danger events using the criterial definitions of Finlay-Jones and Brown (1981), revealed that negative cognitive style, loss events, and danger events all significantly predicted the occurrence of depressive episodes (Safford & Alloy, 1999). However, there were no vulnerability-stress interactions found for episodes of either MD or HD. This particular study evaluated the effects of cognitive style and life events on the first depressive episode experienced during the course of the CVD project, controlling for past history of depression. Further analyses are needed to evaluate if this lack of significant findings is true when both sites' data are used and for all negative life events or just for these two subgroups of life events. In a second preliminary evaluation of the occurrence of depressive episodes across the first 2% years of follow-up in the CVD project (Robinson, 1997), also using Temple site data only, marginally significant to significant relations were found between the cognitive style x number of negative life events interaction and number of episodes of any depressive disorder, including both minor and major episodes. A third preliminary study of the occurrence of MD and HD episodes across the first 9 months of follow-up in the CVD project using only Temple site data (Panzarella, Alloy, & Whitehouse, in press) found that the interaction between cognitive risk and the number of negative life events predicted onsets of both MD and HD. Moreover, the cognitive vulnerability-stress interaction was further moderated by social support. HR participants with high stress and poor social support had a higher likelihood of developing a MD or HD episode than participants with 0, 1, or 2 of the 3 risk factors. As such, more definitive work remains to be done to evaluate the validity of the vulnerability-stress hypothesis in predicting depressive episodes in the CVD project data from both sites.
Characteristics of Cognitively Vulnerable Individuals
Self-Referent Information Processing. The cognitive theories of depression hypothesize that individuals who are cognitively vulnerable to depression tend to process information about themselves in a negatively biased fashion. For example, Beck (1967, 1987) hypothesized that certain individuals possess negative self-schemata that negatively color their perception, interpretation, and memory of personally relevant experiences.
Similarly, the hopelessness theory (Abramson et al., 1989) proposes that individuals with a negative cognitive style tend to infer negative self-characteristics following the occurrence of negative life events.
Data from the Self-Referent Information Processing (SRIP) Task Battery, administered at the Time 1 assessment of the CVD project, were used to test these hypotheses. Given Beck's hypothesis that individuals with a negative self-schema should demonstrate biased information processing only for depression-relevant stimuli (i.e., stimuli related to themes of incompetence, worthlessness, and low motivation), Alloy et al. (1997) predicted that HR participants would demonstrate information-processing biases for depression-relevant, but not depression-irrelevant, self-referent adjectives. Partial support was obtained for this hypothesis. Specifically, as predicted, compared to LR participants, HR participants showed preferential self-referent processing compared to LR participants for negative depression-relevant material (e.g., words like "failure," "passive," and "useless") as evidenced by relatively greater endorsement, faster processing, greater accessibility, better recall, and higher predictive certainty of this material. In addition, HR participants were less likely to process positive depression-relevant stimuli (e.g., words such as "resourceful," "energetic," and "important") than were LR participants. Finally, although contrary to prediction, there were risk group differences for the depressionirrelevant material on two of the tasks. Specifically, LR participants were more likely than HR participants to judge positive depression-irrelevant stimuli as self-descriptive (e.g., words like "thoughtful") and believed they were more likely to engage in future positive depression-irrelevant behaviors (e.g., giving up a seat on a bus for an elderly lady). However, the group differences were larger for depression-relevant than for depression-irrelevant stimuli even on these tasks. Importantly, all of the risk group differences were maintained even after participants' levels of depressive symptoms were statistically controlled. These findings are unique in demonstrating that the information-processing biases previously demonstrated in depressed individuals (see Ingram, Miranda, & Segal, 1998; Segal, 1988) also extend to nondepressed individuals at high cognitive risk for depression.
We also examined whether the negative self-referent processing exhibited by HR individuals mediated or moderated the predictive association between cognitive risk and prospective onsets of depressive episodes (Steinberg, Oelrich, Alloy, & Abramson, 2004). A composite of the five dependent measures from the SRIP Task Battery partially mediated the cognitive risk effects for prediction of HD episodes, but not MD or MiD episodes. This finding is interesting because HD is hypothesized specifically to be a cognitively mediated subtype of depression. In addition, the negative SRIP composite interacted with cognitive risk to predict first onsets, but not recurrences, of MD and HD. Individuals who both exhibited negative cognitive styles and negative information processing about themselves were at increased risk for a first onset of depression compared to individuals with either of these risk factors alone.
Cognitive Vulnerability and Personality Characteristics. In addition to evaluating negative information processing about the self in individuals prone to depression, it is also important to evaluate the relation between cognitive vulnerability to depression and other personality characteristics and disorders. For example, it has been proposed that individuals with negative cognitive styles might be at increased risk for Axis II personality dysfunction (Smith et al., 2004). In support of this hypothesis, previous studies have indicated that comorbidity between depression and personality disorders is high, ranging from 30% to 70% (see Farmer & Nelson-Gray, 1990). In addition, depressed inpatients with comorbid personality disorders, especially borderline personality disorder, have been found to be more likely to exhibit negative cognitive styles than are depressed inpatients without comorbid personality disorders (Rose, Abram-son, Hodulik, Halberstadt, & Leff, 1994). Finally, many personality disorders are partially defined by cognitive patterns that are consistent with definitions of depressogenic cognitive style (Beck et al., 1990). For example, the Cluster C personality disorders (Avoidant, Dependent, and Obsessive-Compulsive) were found to be associated with feelings of incompetence, helplessness, and weakness.
Although the relative specificity of depressogenic cognitive styles has been demonstrated for Axis I psychopathology (e.g., Alloy et al., 2000; Alloy, Abramson, Whitehouse, et al., in press), it has been proposed that such cognitive specificity would not be likely to occur for Axis II personality disorders (Smith et al., 2004). This hypothesized nonspecificity is due to the fact that personality disorders are frequently comorbid with each other and all three of the personality clusters have been associated with depression (Farmer & Nelson-Gray, 1990). In addition, all of the personality clusters are associated with cognitive, behavioral, and interpersonal characteristics consistent with those likely to be found in individuals who are cognitively vulnerable to depression (Smith et al., 2004). In contrast to this proposed nonspecificity between negative cognitive styles and the various personality disorders, there is some evidence that negative cognitive style and HD may be more specifically related to borderline and dependent personality functioning than to other personality disorders (Akhavan, 2000).
To further examine the relation between cognitive style and personality dysfunction, the Personality Disorder Examination (PDE; Loranger, 1988) was administered to all participants at the beginning and end of the CVD
project. The PDE interview provides DSM-III-R categorical personality disorder diagnoses as well as dimensional scores for each disorder. In analyses of personality dysfunction, as assessed by the PDE at Time 1 of the CVD project, the cognitively high risk group had significantly more diagnosable personality disorders than the low risk group (5.4 vs. 1.7%; Smith et al., 2004). Although these percentages are low for both groups, it must be kept in mind that participants in the CVD project were quite young (mean age = 19 years) and, therefore, relatively unlikely to have full-blown personality disorder. For example, Loranger (1988) indicated that it is difficult to accurately diagnose personality disorder in individuals younger than age 25. Therefore, the fact that HR individuals were more than twice as likely to have a diagnosable personality disorder at such a young age is a significant finding. In addition, although there were not enough participants with diagnosable personality disorders to examine the rates of each personality disorder category separately, all three personality clusters were represented in those diagnosed.
When evaluating personality dysfunction using the dimensional scores, the HR group was rated higher than the LR group for Cluster A paranoid and schizotypal dimensions, Cluster B borderline, histrionic, and narcissistic dimensions, and Cluster C avoidant, dependent, obsessive-compulsive, and passive-aggressive dimensions. Schizoid, antisocial, and sadistic personality dysfunction were the only personality dimensions on which no significant risk group differences were found. The strength of these findings is bolstered by the fact that, except for the narcissistic, passive-aggressive, and self-defeating dimensions, these HR-LR differences remained after statistically controlling for the participants' depressive symptom levels, based on their BDI scores. In addition, the risk group differences in the lifetime prevalence of episodic unipolar depressive disorders (major, minor, and HD) reported by Alloy et al. (2000) remained after statistically controlling for the effects of personality dysfunction. Therefore, although an association exists between negative cognitive style and personality disorder, personality dysfunction does not appear to be a sole mediator of the relation between cognitive vulnerability and depression.
Aside from DSM personality dysfunction, other personality characteristics have been linked to depression and deserve attention in regard to cognitive vulnerability for depression as well. For example, sociotropy and autonomy (Beck, 1983) represent two personality subtypes believed to confer vulnerability to depression when an individual experiences negative life events that are congruent with these personality traits. Sociotropy is believed to be a personality style characterized by concern about interpersonal relatedness and fear of rejection or abandonment. Individuals exhibiting this personality style are hypothesized to be prone to depression when they experience interpersonal stresses and losses, such as the breakup of a relationship or a fight with a friend. On the other hand, autonomy is believed to be a personality style characterized by concern for achievement, independence, and self-definition. Autonomous individuals are hypothesized to be at risk for depression when they experience failure to achieve goals they set for themselves, such as receiving a poor grade in school, or failing to get a promotion at work.
To date, there has been little published research on the relation between depressogenic cognitive styles and these two personality subtypes. In some analyses of the CVD project data, this relation was examined (Abramson, Alloy, & Hogan, 1997). HR participants showed greater socio-tropy than did LR participants, even after controlling for their current level of depression. However, the opposite was found to be true with regard to autonomy. Specifically, HR individuals showed a trend toward less autonomy than LR individuals both before and after controlling for current depression levels. This suggests that HR individuals should be more prone to sociotropic, but not autonomous, depressions.
The aforementioned relations between cognitive vulnerability for depression and various personality dysfunction, disorders, and subtypes represent an important contribution to the continued evaluation and expansion of cognitive theories of depression. By incorporating an examination of personality and interpersonal functioning, these findings extend the growing body of research investigating the cognitive-behavioral-interpersonal configurations that confer risk for depression (e.g., Alloy, Fedderly, Kennedy-Moore, & Cohan, 1998; Gotlib & Hammen, 1992; Joiner, Alfano, & Metalsky, 1992; Panzarella et al., in press; Segrin & Abramson, 1994).
Developmental Antecedents of Cognitive Vulnerability to Depression
Evidence shows that negative cognitive styles do indeed confer vulnerability to future episodes of both depression and suicidality, so it is important to understand factors that may contribute to the development of such styles. Data from the CVD project allow an initial examination of several factors that may contribute to the development of these cognitive styles (see Alloy et al., 2004, for an in-depth review of the developmental findings from the CVD project). As part of the CVD project, we assessed the cognitive styles and lifetime history of psychopathology of 335 of our participants' parents (217 mothers, 118 fathers). In addition, participants and their parents were asked to report the parents' inferential feedback styles and parenting styles. Finally, participants' reports of childhood maltreatment were assessed.
Parental Psychopathology. Given previous research suggesting that children of depressed parents are at increased risk for the development of negative attributional styles (e.g., Garber & Flynn, 2001; Hammen, 1992) and episodes of depression (e.g., Downey & Coyne, 1990), data from the CVD project were used to examine the relation between participants' cognitive risk status and their parents' history of depression. Parental history of depression was assessed using the reports of our participants (i.e., family history RDC method; Andreason, Endicott, Spitzer, & Winokur, 1977), and from direct interviews of the parents themselves, using the expanded SADS-L.
Preliminary data from the CVD project suggest that there was a relation between participants' cognitive risk group status and their parents' histories of depression (Abramson et al., 2004). This relation, however, appears to be stronger for mothers' than for fathers' histories of depression. Specifically, HR participants, compared to LR participants, reported that their mothers were significantly more likely, and their fathers were marginally more likely, to have a past history of depression. In the direct interviews of the parents, mothers of HR participants were more likely to have had a past history of depression than were mothers of LR participants. There were no group differences, however, for fathers' histories of depression. These findings are unique in demonstrating a relation between parents' histories of depression and the cognitive styles of nondepressed individuals, and provide support for explorations of possible mediators of the association between parental depression and offspring's cognitive vulnerability to depression.
Modeling and Parental Inferential Feedback. Parents may influence the cognitive styles of their children through modeling their own negative cognitive style or by providing negative inferential feedback regarding the causes and consequences of negative events in their children's lives. However, studies have provided only limited support for a direct relation between parents' and their children's negative cognitive styles. For example, in the CVD project, the mothers of HR participants had more dysfunctional attitudes than did mothers of LR participants, even after controlling for the mothers' levels of depressive symptoms (Alloy et al., 2001). In contrast, there were no risk group differences in mothers' or fathers' inferential styles or in fathers' dysfunctional attitudes. In another study, however, third, fourth, and fifth graders' attributional styles were significantly related to those of their mothers, but not their fathers (Seligman et al., 1984). Finally, in a third study, no relation was found between sixth graders' attributional styles and those of their mothers (Garber & Flynn, 2001). Thus, although there is some evidence that children may model the cognitive styles of their parents, especially their mothers, future studies are needed to further examine this relation. Given the mixed results obtained thus far, future studies should examine potential moderating factors that may either strengthen or weaken the relation (e.g., amount of time spent with parent).
Studies have provided more consistent support for the hypothesis that negative parental inferential feedback may contribute to the development of a negative cognitive style in their children. For example, according to both participants' and their parents' reports and controlling for respondents' levels of depressive symptoms, both mothers and fathers of HR participants in the CVD project provided more stable, global attributional feedback than did mothers and fathers of LR participants (Alloy et al., 2001). Similarly, controlling for respondents' levels of depressive symptoms, mothers of HR participants also provided more negative consequence feedback for negative events in their child's life than did mothers of LR participants, according to both respondents' reports, as did fathers of HR participants according to the participants' reports. In addition, negative attributional and consequence feedback from mothers interacted with a history of high levels of childhood stressful life events to predict HR status (Crossfield, Alloy, Abramson, & Gibb, 2002). Moreover, the negative inferential feedback from parents predicted prospective onsets of depressive episodes in their children over the 2%-year follow-up period, mediated, in part or totally, by the children's cognitive risk status (Alloy et al., 2001).
These results have been supported in other studies. For example, sixth graders' attributional styles for positive and negative life events were correlated with their mothers' attributional styles for the same child-relevant events (Garber & Flynn, 2001). In addition, adolescents' attributional styles were significantly related to their fathers', but not mothers', attributional styles for the same child-relevant events (Turk & Bry, 1992). Thus, there is some evidence that parents may contribute to the development of negative cognitive styles in their children, not by the children modeling the attributions their parents make for negative events in the parents' lives, but by the attributional and consequence feedback the children receive from their parents for negative events in the children's own lives.
Parenting Styles. Studies have suggested that certain parenting styles may also contribute to the development of a negative cognitive style in children. For example, both HR participants in the CVD project and their fathers reported that the fathers exhibited less warmth and acceptance (and more rejection) than did fathers of LR participants (Alloy et al., 2001). There were no group differences, however, for fathers' levels of either psychological autonomy versus control or firm versus lax control (discipline), nor were there any group differences in the parenting styles of mothers. Fathers' acceptance scores also predicted prospective onsets of MD, MiD, and HD episodes in their children, but only the prediction of HD episodes was mediated by the children's cognitive risk status (Alloy et al., 2001). In a longitudinal study of sixth graders and their mothers, higher levels of maternal psychological control were associated with increasing negativity of their children's attributional styles over a 1-year follow-up, even after controlling for the mothers' histories of mood disorders (Garber & Flynn, 2001). In this study, neither parental acceptance versus rejection nor firm versus lax control were related to changes in the children's attributional styles. Finally, undergraduates with a negative cognitive style reported less maternal care when growing up than did undergraduates with a positive cognitive style (Whisman & Kwon, 1992). In this study, undergraduates' cognitive styles were not related to the degree of maternal overprotection reported during childhood.
Thus, although these studies suggest a relation between certain parenting styles and children's cognitive styles, they do not agree as to which parenting styles are the most detrimental. In addition, only one study (Alloy et al., 2001) from the CVD project has examined the parenting practices of fathers. Future studies, therefore, should seek to clarify the relation between parenting practices and children's negative cognitive styles and should seek to include fathers in this evaluation.
Childhood Maltreatment. In extending the etiological chain of the hopelessness theory, Rose and Abramson (1992) proposed a developmental pathway by which negative life events, especially childhood maltreatment, may lead to the development of a negative cognitive style. Specifically, they suggested that when maltreatment occurs, individuals attempt to understand the cause, consequences, and meanings of the abuse so that future negative events may be avoided and hopefulness may be maintained. Thus, after the occurrence of maltreatment, children may initially make hopefulness-inducing attributions about its occurrence. For example, children may initially explain being beaten or verbally abused by their father by saying, "He was just in a bad mood today," which is an external, unstable, and specific explanation. With the repeated occurrence of maltreatment, however, these hopefulness-inducing attributions may be disconfirmed, leading children to begin making hopelessness-inducing attributions about its occurrence. For example, children may explain the maltreatment by thinking, "I'm a terrible person who deserves all the bad things that happen to me," which is an internal, stable, and global explanation that entails negative consequences and negative self-characteristics. Over time, these attributions may generalize to initially unrelated negative events. In this way, a relatively stable and global negative cognitive style may develop.
Researchers have only recently begun to evaluate the relation between childhood maltreatment and cognitive styles. These initial evaluations, however, have supported Rose and Abramson's hypotheses. For example, controlling for their levels of depressive symptoms, HR participants in the CVD project reported significantly higher levels of childhood emotional, but not physical or sexual, maltreatment than did LR participants (Gibb, Alloy, Abramson, Rose, Whitehouse, Donovan, et al., 2001). In addition, participants' cognitive risk status fully mediated the relation between reported levels of childhood emotional maltreatment and the occurrence of DSM—III—R and RDC nonendogenous MD during the first 2% years of follow-up. Further, participants' cognitive risk status fully mediated, and their average levels of hopelessness partially mediated, the relation between reported levels of childhood emotional maltreatment and the occurrence of HD during the first 2% years of follow-up. To address the possibility that the association of childhood emotional maltreatment with negative cognitive styles is actually due to genetic influences or a negative family environment in general, Gibb, Abramson, and Alloy (2004) also examined the relation between emotional maltreatment by nonrelatives (i.e., peer victimization) during development and negative cognitive styles. Gibb et al. (2004) found that even when parental maltreatment and parental history of psychopathology were controlled, peer victimization still was significantly associated with cognitive HR status. These findings are not easily explained by third variable accounts such as genetic influence or a general negative family context.
Similarly, examining the CVD project participants' average levels of suicidality (both questionnaire- and interview-assessed) across the first 2% years of follow-up, participants' cognitive risk status and average levels of hopelessness partially mediated the relation between reported levels of childhood emotional maltreatment and average levels of suicidality (Gibb, Alloy, Abramson, Rose, Whitehouse, & Hogan, 2001). The results of a recent cross-sectional study were also supportive of Rose and Abramson's developmental model. Specifically, the results were consistent with the hypothesis that high levels of childhood emotional maltreatment lead to more negative inferences about that maltreatment, which then lead to the development of a negative inferential style, and this inferential style then leaves one vulnerable to hopelessness and the symptoms of hopelessness depression (Gibb, Alloy, Abramson, & Marx, 2003).
In addition to supporting Rose and Abramson's (1992) developmental model, these results also support their hypothesis that childhood emotional maltreatment may be more likely than either childhood physical or sexual maltreatment to contribute to the development of a negative cognitive style. Specifically, Rose and Abramson hypothesized that, with childhood emotional maltreatment, the depressogenic cognitions are directly supplied to the child by the abuser. With childhood physical and sexual maltreatment, however, children must supply their own negative cognitions. In this way, childhood physical and sexual maltreatment may allow greater opportunity for the child to make less depressogenic attributions and inferences for the occurrence of maltreatment. Although these studies provide preliminary evidence for a relation between childhood emotional maltreatment and negative cognitive styles, future longitudinal research is needed that assesses the degree to which emotional maltreatment contributes to increased negativity in cognitive styles over time.
Many important theoretical issues remain to be addressed with the CVD project data. Although analyses to date have indicated that individuals with negative cognitive styles are at higher risk for experiencing clinically significant depression, future analyses will need to be done to evaluate whether nondepressed individuals at high cognitive risk are more likely than low risk individuals to develop depression only when they experience stressful life events, or whether cognitive risk may confer vulnerability to depression even in the absence of negative life events. In the CVD project, negative life events were assessed repeatedly (every 6 weeks for the first 2^> years of follow-up) and dated to the day they occurred, which makes prospective evaluation of this cognitive vulnerability-stress hypothesis possible. Although preliminary investigations of the vulnerability-stress hypothesis have been conducted, a more thorough evaluation is necessary. In addition, it will be important to test if any predictive effect of the cognitive risk x stress interaction for future depressive episodes is mediated by the occurrence of hopelessness, as predicted by the hopelessness theory, and if it is specific to HD as opposed to other possible subtypes of depression. Other environmental and individual difference variables that may serve as protective factors against the development of hopelessness and depression will also need to be explored. For example, there is substantial evidence that social support can help buffer against the occurrence of depression when people experience stressful events (e.g., Cohen & Wills, 1985; Panzarella et al., in press). Future analyses of the CVD project data will allow an investigation of these potential protective factors.
There is much room for future research on cognitive vulnerabilities to depression outside of the CVD project as well. Most importantly, the CVD project combined both inferential styles and dysfunctional attitudes in defining negative cognitive style. Although this method of selecting participants provides the strongest possible test of the cognitive theories of depression, it does not allow examination of the unique contribution of inferential styles and dysfunctional attitudes in the prediction of depression. As such, the CVD project represents an important step in research examining cognitive vulnerability to depression. Future studies are needed, however, to more specifically test the predictions of Beck's theory and the hopelessness theory, separately (e.g., Haeffel et al., 2003). The role played by positive events should also be addressed in future studies. That is, do positive events provide a buffering effect, protecting against the occurrence of depression? Given the CVD project findings thus far, which have indicated a significant prospective relation between cognitive vulnerability and future depression, studies should continue to examine the therapeutic impact of modifying individuals' cognitive styles. For example, one study has suggested that cognitive behavioral therapists may reduce clients' depressive symptoms by reducing the negativity of clients' attributional styles (DeRubeis & Hollon, 1995). In addition, there is some evidence that training children to make less negative attributions about the negative events in their lives can help protect against future levels of depression (Gillham, Reivich, Jaycox, & Seligman, 1995; Jaycox, Reivich, Gillham, & Seligman, 1994). Given that negative cognitive styles may be especially likely to contribute vulnerability to depression when exacerbated by rumination, depressogenic cognitive styles may also be altered indirectly by training individuals in more effective methods of coping with stressful events, rather than directly trying to alter their cognitive style. Alternatively, it might be necessary to help cognitively vulnerable individuals decrease the stress in their environments.
Further, building positive cognitive styles in children by educating parents to model and provide feedback about more benign inferences for stressful events, as well as direct training in generating positive interpretations of stressful events in schools might help reduce the occurrence of negative cognitive style, and therefore, depression. Finally, parenting classes that teach parents less abusive ways of raising their children may also aid in the prevention of cognitive vulnerability to depression. All of these treatment and prevention models, based on the theorized existence of depressogenic cognitive styles, require further investigation if we are to more fully understand and utilize what we have learned from research on the cognitive theories of depression.
Was this article helpful?