cognitive vulnerability to psychological disorders

Journal of Social and Clinical Psychology, Vol. 25, No. 7, 2006, pp. 705-725 RISKINDDesign, Theory, AND ALLOY and Methods

COGNITIVE VULNERABILITY TO PSYCHOLOGICAL DISORDERS: OVERVIEW OF THEORY, DESIGN, AND METHODS JOHN H. RISKIND George Mason University LAUREN B. ALLOY Temple University

Cognitive vulnerability models offer popular ways of understanding the origins and causal factors that contribute to psychological problems. Cognitive vulnerabilities are typically purported to create liabilities to psychological disorders after individuals encounter stressful events, in a vulnerability–stress interaction, and to maintain the problems after their onset. We present a conceptual framework that describes common features shared by most cognitive models, including schematic biases in information processing, developmental factors, reciprocal feedback loops, and specific vulnerabilities for specific problems or disorders. We also examine issues related to the role of theory in determining the appropriate research design, and discuss actual design options and methods used in cognitive vulnerability research, and their strengths and limitations.

Over the past 30 years, cognitive perspectives, including a focus on information–processing, have been extended to depression, anxiety, eating disorders, substance abuse, marital problems, and a variety of psychological problems (e.g., Alloy & Riskind, 2006; Mathews & MacLeod, 1994). The present article reviews some of the basic issues relating to the theory, design, and methods in research on cognitive vulnerability to psychological problems. In the sections of this article, we first discuss basic tenets of cognitive models of psychological problems, including the concept of cognitive mediation, and vulnerability–stress Address correspondence to John H. Riskind, Department of Psychology MSN 3F5, George Mason University, Fairfax, VA 22030-4444; E-mail: [email protected]

705

706

RISKIND AND ALLOY

interaction. We next examine issues related to the role of theory in determining the appropriate research design, and discuss actual design options and methods used in cognitive vulnerability research, and their strengths and limitations. We conclude the article with a brief summary.

BASIC TENETS OF COGNITIVE MODELS OF PSYCHOLOGICAL PROBLEMS Cognitive clinical models assume that cognitive phenomena mediate the relationship between events that a person experiences and subsequent emotional responses. That is, between the situation and the individual’s responses comes the all–important step of information processing and cognitive appraisal (e.g., Fridja, 1987; Lazarus, 1991; Ortony, Clore, & Collins, 1988; Roseman, Spindel, & Jose, 1990). For example, individuals who attribute failures to stable personal defects are likely to become more depressed and exhibit more helplessness. This same guiding principle applies to internal stimuli (e.g., physical sensations, unwanted thoughts or emotions) as to stimuli in the external environment (e.g., criticism from others). The principle also applies to both “normal” and “abnormal” responses of individuals. That is, such models are also based upon the precept of a continuity of normal and abnormal cognitive processes. As Beck (1991) stated: “The [cognitive] model of psychopathology proposes that the excessive dysfunctional behavior and distressing emotions or inappropriate affect found in various psychiatric disorders are exaggerations of normal adaptive processes” (p. 370). At the same time, people can exhibit relatively habitual or stable patterns in the ways in which they appraise emotion–provoking stimuli (e.g., Abramson, Seligman, & Teasdale, 1978; Riskind et al., 2000; Weiner, 1985). For example, some individuals habitually attribute failures to personal defects. As a result of these differences, people can differ in their future risk for developing particular kinds of emotional disorders or psychological problems. Importantly, cognitive models of vulnerability distinguish distal phenomena, which were present before the problem developed, from proximal phenomena that occur close in time to onset of its symptoms (Abramson, Metalsky, & Alloy, 1989). Distal cognitive factors are typically cognitive predispositions (e.g., dysfunctional attitudes or depressive inferential styles) to respond to stressful situations in maladaptive ways. They tend to be more trait–like aspects of personality that are more distal to future episodes than proximal cognitions, which are situationally specific thoughts or mental processes that occur very close to, or even during, the future episode of the psychological problem. Proximal cognitions are typically produced when individuals process a

THEORY, DESIGN, AND METHODS

707

stressful experience in any situation through the underlying cognitive vulnerability. The cognitive vulnerability is developed from earlier social learning history or developmental events. COGNITIVE VULNERABILITY–STRESS PARADIGM Most theorists adopt a vulnerability–stress paradigm in which it is recognized that psychological disorders are caused by a combination of predisposing (constitutional or learned) and precipitating (environmental) factors. That is, precipitating events (e.g., stressful life events, early childhood traumas, a marital disagreement) can trigger the development of psychological problems or emotional disorders for certain individuals (e.g., see Alloy, Abramson, Raniere, & Dyller, 1999), but the degree and even direction of response can differ enormously from one person to another. Thus, precipitating environmental events are particularly likely to produce psychological problems among individuals who have a preexisting cognitive vulnerability to the disorders. Indeed, most individuals who are exposed to precipitating stressful events do not develop disorders. And, when disorders or problems emerge, their specific nature differs for different individuals and is not determined just by the precipitating stress alone. For example, stressful events are elevated in depression (Brown & Harris, 1978; Paykel, 1982), bipolar disorder and mania (Johnson & Roberts, 1995), anxiety disorders (Last, Barlow, & O’Brien, 1984; Roy–Byrne, Geraci, & Uhde, 1986), and even schizophrenia (Zuckerman, 1999). Thus, cognitive vulnerability–stress models are offered to help account for not only who is vulnerable to developing emotional disorder (e.g., individuals with a particular cognitive style), and when (e.g., after a stress), but to which disorders (e.g., depression, eating disorder, etc.) they are vulnerable (Riskind & Alloy, 2006). Researchers working in the cognitive paradigm often prefer the term “vulnerability” rather than “diathesis,” because this former term embraces the idea of learned and modifiable predispositions, instead of immutable genetic or biological traits (e.g. Just, Abramson, & Alloy, 2001), which is implied by the latter term. For example, Beck (1967, 1976), the earliest expounder of a cognitive vulnerability–stress interaction, proposed that an individual’s cognitive vulnerability to later difficulties is brought about by maladaptive knowledge structures or schemata formed of core beliefs, attitudes, and concepts that individuals develop from earlier experiences. The individual’s schemata influence the idiosyncratic subjective meaning that the person attaches to events and thus the emotional impact of stressful events. As such, cognitive schemata

708

RISKIND AND ALLOY

can promote psychological and/or interpersonal maladjustment when they are poorly grounded in social reality or otherwise dysfunctional. According to Beck’s formulation, a change in the nature of information is fundamental to the onset of psychological problems. For example, once the underlying mechanism (e.g., schemata) for a relevant psychological problem (e.g., depression, anxiety, anger) is set off, the nature of self–referent cognition undergoes a profound change. For example, individuals may be biased to judge themselves as failures, which can increase vulnerability to problems of mood such as depression or anxiety. Moreover, people are thought to think differently about themselves, the world, the future, or the activities in their lives, and to remember, screen, select, and interpret information in different ways during states of depression, anxiety, etc., than during normal states.

COGNITIVE VULNERABILITY MODELS: A FRAMEWORK OF COMMON FEATURES While cognitive vulnerability models differ in their particulars and their subject matter, most share important common features (see Riskind & Alloy, 2006). As depicted by Figure 1, most tend to assume a series of causal chains by which cognitive vulnerabilities develop from earlier life experiences (such as faulty attachment relationships or modeling). Once individuals have developed cognitive vulnerabilities, these are likely to alter their responses, and with later stressor/vulnerability combinations are converted into the emotional or psychological problems. The cognitive vulnerabilities are assumed to have schematic function (e.g., Clark, Beck, & Alford, 1999). As such, the vulnerabilities shape and schematically bias the person’s selective processing, attention, and memory, and promote changes in the concomitant contents of the individual’s thinking (i.e., the ideation, imagery, or “automatic” thoughts). The schematic biases associated with their cognitive vulnerabilities can guide the whole range of activities involved in information processing (e.g., selective attention, encoding, elaboration and retrieval in memory, interpretation). An important characteristic of cognitive vulnerabilities is that they are associated with “disorder–specific” information processing biases (e.g., attention vs. memory) and in specific typical proximal cognitions (Beck & Clark, 1997). For example, the disorder–specific bias in anxiety is widely assumed to involve information relevant to threat (e.g., public humiliation), and accompanied by proximal thoughts such as “I’ll look like a fool.” In depression, the specific bias is for information relevant to past loss, irreversible defeat, and hopelessness. Such disorder–specific information–processing biases are presumably instigated when cogni-


709

A. Learning History Parenting, Attachment, & Developmental Experiences

B.

Relatively Trait-like, and Distal Cognitive Vulnerabilities Maladaptive Maladaptive Cognitive Cognitive Styles, Styles, Attitudes Attitudes

E. Self-Protective Responses and BiDirectional Feedback --e.g., Negative Effects of Reassurance Seeking, or Avoidance

C.

Combination(s) of Vulnerability-Stress and of lack of Suitable Protective Factors (e.g., Social Support, Coping Skills)

D.

Proximal Cognitive Processes Co-occurrent with Specific Episodes of Problems Biased Biased Information Information Biased Appraisals Processing Processing and and of Intensifying Inferences Inferences Danger Automatic Thoughts, Images Problematic Problematic Emotional Emotional States States and and Behavior Behavior

FIGURE 1. Theoretical Framework for Cognitive Vulnerability

tive vulnerabilities (distal factors) are engaged, but the stages of processing and subject matter that is implicated can differ with different models and psychological problems or disorders (e.g., Williams, Watts, MacLeod, & Mathews, 1988). In this regard, some kinds of problems may involve biased information processing at multiple levels of the cognitive continuum (e.g., attention, encoding, retrieval), whereas other problems may involve processing biasing concentrated at particular points on the continuum. Examples of early stage biases would be the hyper–vigilance for threat

710

RISKIND AND ALLOY

stimuli that occurs in people who are anxious, or an attentional bias for positive information that is likely to occur in people who are highly optimistic. Examples of later stage biases would be slanted tendencies to interpret and elaborate the meaning of stimuli in schema–congruent manners. The notion that cognitive content and biases differ in different psychological problems has been the impetus for considerable research. A variety of cognitive vulnerability factors have been suggested (Alloy & Riskind, 2006). Beck’s cognitive model emphasized the role of faulty beliefs and dysfunctional attitudes in creating vulnerabilities to depression, anxiety, or other psychological problems (e.g., Beck, 1976; Beck & Clark, 1997). Other current models of depression focus on dysfunctional inferential (attributional) styles that create a cognitive vulnerability (Abramson, Metalsky, & Alloy, 1989; Alloy, Abramson, Safford, & Gibb, 2001). A recent model of anxiety (Riskind, Williams, Gessner, Chrosniak, & Cortina, 2000; see Riskind, Williams, & Joiner, this issue) emphasizes the roles of maladaptive mental simulations of danger involving the generation of dynamic mental scenarios of rapidly rising risk (Riskind, Williams, Gessner, Chrosniak, & Cortina, 2000). In combination with cognitive vulnerability factors, the risk of developing psychological problems and their potential severity after exposure to stress is also influenced by protective and exacerbating factors. Even after exposure to stress, the presence of certain protective factors (e.g., social support, an intimate relation with a spouse or lover, effective coping mechanisms) may work against the development of psychological problems, or reduce the likelihood the problems will develop (e.g., Brown & Harris, 1978). Exacerbating factors are additional stresses or factors (e.g., further medical illnesses, psychological problems, or negative affect expressed by others) that impinge on the individual and act to worsen a psychological problem or disorder after its onset. As a result, a set of individuals could at any point begin with precisely the same combinations of vulnerability and stress, but still markedly differ in their future trajectories of problems because they differ in protective and exacerbating factors. In this context, many additional factors potentially related to vulnerability to emotional disorders such as personality factors, demographic factors, developmental experiences, or interpersonal patterns can be sensibly understood with the cognitive paradigm. For example, personality characteristics such as negative affectivity or neuroticism can be seen as the consequences of the activation of underlying negative schemata (e.g., Jolly & Kramer, 1994). Factors such as the availability of social support can be understood as serving to buffer stress by helping to impart more adaptive inferences about the causes and consequences of negative life events (Dobkin, Panzarella, Fernandez, Al-


711

loy, & Cascardi, 2004; Panzarella, Alloy, & Whitehouse, in press). Similarly, factors such as gender can be sensibly understood within the cognitive paradigm in terms of their relationship to dysfunctional rumination patterns that affect a person’s focus of attention on negative events (e.g., Nolen–Hoeksema, Morrow, & Frederickson, 1993). When possible, it can often be of value to conceptualize cognitive mechanisms by which “noncognitive” variables can interact with the cognitive system (Riskind & Alloy, 2006). Beyond these factors, vicious cycles involving bi–directional causal links (and feedback loops) can also contribute to the onset, maintenance, or recurrence of psychological problems (see Figure 1). For example, under the duress of the stress and intense symptoms, cognitively vulnerable individuals tend to engage in various maladaptive self–protective or compensatory behaviors. Depressed individuals often have a heightened inclination to engage in reassurance seeking from others (Joiner, Alfano, & Metalsky, 1992), which can alienate others or drive them away. Individuals who are depressed also contribute to the occurrence of self–generated life events (e.g., creating interpersonal conflicts or excessive demands) that can maintain and exaggerate their own depression (Hammen, 1991). Recent research suggests that individuals who are cognitively vulnerable to anxiety may also create their own stressful life events (Shahar, Black, & Riskind, 2004). Some individuals who are anxious engage in avoidance behaviors, some relatively subtle—such as avoiding eye contact to avoid seeing imagined rejection—that can self–generate stress and prove maladaptive. Psychological problems are believed to persist as long as the cognitive components of the problems are present, and improve when they are altered. Further, although temporary relief is produced by changes in proximal cognitive components of the problems, durable improvement requires changes of the underlying cognitive vulnerability factors.

ADDITIONAL ISSUES FOR COGNITIVE VULNERABILITY MODELS Several additional issues confront cognitive vulnerability models of psychological disorders (for a longer discussion of such issues, see Riskind & Alloy, 2006). These issues include the (1) specificity of cognitive vulnerability factors, (2) the specificity and nature of developmental antecedents, and (3) the role of cognitive vulnerability in comorbidity, SPECIFICITY OF COGNITIVE VULNERABILITY FACTORS. The first issue for cognitive models concerns the distinction between

712

RISKIND AND ALLOY

specific and nonspecific causal factors in psychological problems/emotional disorders. Specific causal factors are relatively unique or focal factors in that they influence and predict the development of a particular problem, but do not equally apply to all psychological problems or disorders in general. For example, some vulnerability factors might apply to just social anxiety, but not other forms of anxiety. In contrast, other cognitive vulnerability factors could be specific to the whole spectrum of different forms of anxiety, but not apply to depression (e.g., Riskind & Williams, 2006). Alternatively, nonspecific (or common) causal factors are characterized by their relatively low discriminatory power (Ingram, 1990; see also, Clark, 1997) and in this way, potentially cut across a range of different psychological problems (e.g., depression, anxiety, substance abuse disorders, marital problems). A likely instance of a nonspecific factor is prior exposure to uncontrollability (Chorpita & Barlow, 1998), which seems to play an important, but nonspecific, role in many disorders. It is, of course, important to understand nonspecific or common causal factors that apply to multiple psychological problems. At the same time, however, any adequate cognitive model of such problems also needs to identify causal factors that have discriminatory power and applies to specific problems and not others. By this view, it is necessary to identify possible factors that may play roles as specific cognitive vulnerabilities for specific problems. COMMON AND SPECIFIC DEVELOPMENTAL FACTORS As can be seen from the framework depicted in Figure 1, it is often assumed that people’s antecedent childhood experiences can help to mold the nature of the cognitive vulnerabilities they later develop. Recent research does, indeed, suggest that childhood experiences such as peer rejection or faulty attachment relationships may lead to cognitive vulnerabilities (e.g., Gibb, Alloy, Abramson, & Marx, 2003; Ingram, 2003; Ingram, Bailey, & Siegle, 2004; Riskind, Williams, Altman, Black, Balaban, & Gessner, 2004; Rogers, Reinecke, & Setzer, 2004; Safford, Alloy, Crossfield, Morocco, & Wang, 2004; Williams & Riskind, 2004). However, the specificity of such developmental factors to particular psychological problems is still relatively unknown. If such developmental variables were, indeed, to be found to constitute such common factors, they would be expected to play a general, but nonspecific, role in the origins of many psychological problems. In some cases, it is perhaps not just childhood experiences alone, but factors such as variability in the particular factual details of childhood experiences may also help account for differences in the vulnerabilities that are developed. For example, children could be more prone to de-


713

velop depression–related cognitive patterns if they have been subjected to unremitting or relatively constant parental abuse or criticism. In contrast, they might be more prone to develop anxiety–related patterns if they have been subjected to more variable parental negative events, and/or if some positive protective factors are present. Much more work on these developmental questions is clearly warranted. COGNITIVE VULNERABILITY AND COMORBIDITY Another issue is how cognitive models account for the comorbidity of psychological disorders (Alloy, Kelly, Mineka, & Clements, 1990). For example, it is well documented that anxiety and depression appear in a comorbid form (e.g., Zinbarg, & Barlow, 1996). Besides this, it is known that symptoms of anxiety and depression are often more severe when they co–occur than when they occur separately—such as when case controls only have one of the disorders (see Riskind, Moore, Harman, Hohman, Beck, & Stewart, 1991). Once we recognize that cognitive vulnerability factors can vary independently of each other, we might easily imagine that individuals with compound vulnerabilities (i.e., they combine cognitive vulnerabilities to different disorders) could be expected to be far likelier to develop comorbid emotional disorders. Preliminary support for just this notion is offered by a recent study that assessed both depressive explanatory style and looming maladaptive style (Riskind et al., 2000) and examined their main effects and interactions (Riskind & Williams, 2000). As expected, individuals who had the compound vulnerability, in that they combined both depressive inferential patterns and the looming cognitive style for anxiety (see Riskind, Williams, & Joiner, this issue), exhibited more severe symptoms of anxiety and depression than would be expected from the simple summation of utterly separate effects. Thus, the study of cognitive vulnerability factors has promise in helping to illuminate the nature of psychological antecedents of comorbid emotional disorders.

THE ROLE OF THEORY IN DESIGN SELECTION: THE INTERFACE OF THEORY AND RESEARCH METHODS Having examined the central theoretical facets of models of cognitive vulnerability, we consider the role that such theory plays in design selection, and how it is necessary to tailor the research design selected to test our research questions or the specific logic of the cognitive vulnerability models of interest. We will begin by examining several questions related to the kinds of causal relations specified, the nature of the vulnerability–stress combination, and the role of cognitive priming; then we focus

714

RISKIND AND ALLOY

on specific logical criteria needed to support a hypothesized vulnerability factor. HYPOTHESIZED CAUSAL RELATIONS IN COGNITIVE VULNERABILITY–STRESS MODELS One set of issues to consider in selecting an appropriate research design concerns the type of causal relations (e.g., distal or proximal; necessary, sufficient, contributory) posited by a given cognitive model. For example, early negative life events (e.g., attachment relationships in childhood) can be viewed as a distal cause of psychological disorder (by way of a cognitive vulnerability). In contrast, later negative life events that are proximal to the onset of the psychological problem or disorder can be viewed as the precipitating event for an episode of the disorder. The appropriate design of a research study is also influenced by whether the hypothesized causal role is necessary, sufficient, or contributory. In these terms, a necessary cause is an etiological factor that is an essential condition (either in the present or the past) for the disorder to occur. In the absence of the etiological factor, the disorder cannot occur, although the factor by itself does not require the disorder to occur (i.e., the factor is necessary but not sufficient). A sufficient cause is an etiological factor that guarantees the occurrence of the disorder, although the factor may not be necessary for the disorder (e.g., the factor is sufficient but not necessary). A contributory cause increases the statistical likelihood that the disorder will occur by playing a supporting causal role, but is neither necessary nor sufficient for the occurrence of the disorder. In all, the hypothesized causal relations in the model should guide the nature of the research design as well as the choice of statistical analyses. In some cognitive models, it may be hypothesized that the cognitive vulnerability and stress combine in an additive fashion as a straightforward summation. In other models, in contrast, there is true statistical interaction (e.g., Abramson, Alloy, & Hogan, 1997; Alloy et al., 1999; Monroe & Simons, 1991) that predicts outcomes synergistically beyond the separate additive effects of the vulnerability and stress. For instance, the combination of a high level of cognitive vulnerability and high stress will be far likelier to lead to an episode of the psychological disorder in question than either factor (or their additive combination) alone. The specific manner in which the cognitive vulnerability and stress are postulated to combine as causal factors is critical in determining the appropriate statistical analysis as well as levels of the cognitive vulnerability–stress combination that must be sampled (or experimentally manipulated) to test the model (see Alloy et al., 1999).


715

Another key aspect of the hypothesized causal relations lies in the distinction between moderating and mediating factors (Baron & Kenny, 1986; Holmbeck, 1997). A moderator is a third variable that codetermines the outcomes by affecting the relationship between the independent variable (e.g., the cognitive vulnerability or stress or both) and the dependent variable (disorder). In essence, a moderator statistically interacts with the vulnerability or stress (or both) and affects the direction or strength of the relationship between the vulnerability–stress combination and disorder. For example, let’s suppose that gender is posited as a moderator of the relationship between a cognitive vulnerability–stress combination and depression (e.g., women may be more susceptible to such effects). What this means is that the direction or strength of the cognitive vulnerability–stress combination would depend on whether subjects were men or women. In contrast, a mediator is a third variable that is assumed to account for the relation between an independent variable (e.g., the distal cognitive vulnerability and proximal stress, or their combination) and the dependent variable (e.g., the disorder). The mediator can be seen as the intermediary process or mechanism by which the cognitive vulnerability–stress combination becomes converted into an episode of disorder. For example, as we saw before (Figure 1), biases in mental processes of memory or attention could be seen as a third variable that mediates the relation between a cognitive vulnerability and the onset of disorder. In sum, moderators specify the conditions under which a vulnerability–stress combination will lead to a disorder. In contrast, mediators specify how or why the vulnerability–stress combination leads to disorder. Baron and Kenny (1986) discuss appropriate differences in the statistical tests required to test moderation as opposed to mediation. THE PRIMING OR ACTIVATION OF COGNITIVE VULNERABILITIES During periods when such vulnerabilities are in an inactive or latent state, a cognitive priming task such as a relevant mood–induction or activating provocation task (Ingram, Miranda, & Segal, 1998; Riskind & Rholes, 1984) could be needed. One can make an interesting analogy between a cardiac stress test (which is used to reveal a hidden coronary dysfunction) and a cognitive priming task (which is used to reveal a hidden cognitive vulnerability). It follows that it may sometimes be necessary to bring out latent vulnerability with an appropriate priming test (for evidence, see Ingram, Miranda, & Segal, 2006). But not all cognitive vulnerabilities may require a priming procedure (Just et al., 2001). The extent to which cognitive activation and assessment of vulnerabilities requires priming in different models remains an open question.

716

RISKIND AND ALLOY

LOGICAL CRITERIA NECESSARY TO SUPPORT A PUTATIVE COGNITIVE VULNERABILITY There are four key logical criteria that must be satisfied to strongly empirically support a hypothesized cognitive vulnerability. First, the temporal precedence and stability of the vulnerability independent of the symptoms of the disorder must be established (e.g., Alloy, Abramson, Raniere, & Dyller, 1999; Ingram, Miranda, & Segal, 1998). That is, the putative vulnerability must temporally precede the initial onset of the psychological disorder, or, in the case of a vulnerability factor for the course of a disorder, it must precede episodes or symptom exacerbations of the disorder (i.e., it has predictive validity). Second, it must exhibit some degree of stability independent of the symptoms of the disorder. That is, the vulnerability must be shown to be more than just a transient state–manifestation or consequence of the changing symptoms of the disorder. Third, and of equal importance, alternative explanations of results must be eliminated as plausible options. This is achieved in part by establishing that the predicted relationships are not due to potential third variables (or confounds; Cohen & Cohen, 1983). Confidence in the validity of the putative vulnerability is also increased by providing supplementary evidence that the vulnerability factor plays a causal role in the development of symptoms or onset of disorder (i.e., obtaining evidence for its construct validity). This being so, important evidence on these issues is supplied by a network of findings showing predicted differences in personality characteristics, information processing, coping patterns, etc., of individuals who are high or low in cognitive vulnerability. For example, results that can be interpreted as demonstrating the developmental antecedents or mediating processes (e.g., information processing deficits) specifically predicted by the cognitive theory (but not by a “rival” third factor), offers strong support to the system of hypotheses and the construct validity of the proposed cognitive vulnerability. A fourth, and final, criterion for causal status is also worth highlighting. If a theory of interest claims that the vulnerability factor is restricted or specific to a certain disorder, it must further be shown that the factor is largely applicable to the disorder of interest and not to other disorders (i.e., it has discriminant validity). Evidence that a putative vulnerability has stability or temporal precedence is insufficient for establishing that it has specificity to given psychological problems (Riskind & Alloy, 2006). Thus, there is a need to directly test the specificity of the predicted outcomes.


717

ISSUES IN RESEARCH DESIGN Based on the preceding logical criteria, we now consider several design options and their relative tradeoffs in terms of their strengths and limitations. DESIGN IN A BEST–CASE SCENARIO The unique importance of experimental designs in all research is that independent variables are directly manipulated and can normally provide the strongest evidence for cause–effect relationships. In true experimental designs, group differences in psychological problems can be inferred to be due to the experimental treatments (e.g., manipulations of cognitive vulnerability and stress), because experimental control over sources of error permits a relatively strong basis for testing cause–effect relationships, as the experimental groups are most probably equivalent an all other variables (e.g., other individual differences in variables that predated the experiment). As it turns out, experimental manipulation studies, in which participants are randomly assigned to vulnerability conditions, are normally almost impossible to implement when studying cognitive vulnerability to psychological problems. One can hardly imagine, for both ethical and pragmatic reasons, that researchers would try to randomly assign young children to conditions that manipulate their attachment relationships (e.g., “good” versus “bad”) to test their vulnerability to psychological disorders. For this reason, cognitive vulnerability researchers almost inevitably rely on other research designs, including analogue, quasi–experimental, and correlational research designs (Alloy et al., 1999). ANALOGUE AND QUASI–EXPERIMENTAL STUDIES As true experimental designs cannot usually be implemented in studying human cognitive vulnerability, some cognitive vulnerability research uses analogue and quasi–experimental designs. Analogue studies (which can use laboratory animals or nonclinical human participants as proxies for actual clinical patients) can sometimes have value for testing parts of cognitive vulnerability theories. For example, experimental manipulations in animal analogue studies have been used to test potential causal variables featured in the learned helplessness model of depression in humans (e.g., see Seligman, 1975). Likewise, experimental manipulations in analogue studies with humans have tested cognitive models by randomly assigning normal (i.e., nondisordered) partici-

718

RISKIND AND ALLOY

pants to different mood–induction conditions (e.g., depression vs. elation vs. neutral) and assessing mood changes or changes in cognitive biases (Riskind, 1989). If such analogue studies are to have validity as analogues for human psychological problems, they must meet certain criteria (Abramson & Seligman,1977). In particular, they must establish: (1) close similarities between the analogue model (e.g., learned helplessness) and the clinical disorder (e.g., human depression), and (2) close similarities between essential features of the disorder (e.g., the motivational and cognitive deficits in depression) and the features of the analogue model (e.g., behavioral and learning deficits in helpless animals). When these criteria are met, analogue studies can provide a useful type of convergent evidence for the “construct validity” of the causal mechanisms hypothesized by a cognitive vulnerability model. A quasi–experimental design offers another alternative for testing cognitive vulnerability models. Despite containing some elements of experimental control (i.e., there is at least one experimental manipulation), such designs are not true experiments because they do not assign participants on a random basis to one of the key independent variables (i.e., the “quasi–experimental variable”). For example, responses of high–risk (cognitively vulnerable) and low–risk (nonvulnerable) individuals might be compared on an experimental laboratory task of memory that manipulates a “within–subjects” factor (such as the positive or negative valence of information). Such quasi–experimental designs are wholly adequate for some purposes, such as testing causal inferences within the high cognitive vulnerability group, as half of the participants are randomly assigned to serve as a control for the other half of the participants who receive a different manipulation or intervention. But any statistical main effects or interaction effects pertaining to the quasi–experimental variable of interest (e.g., the cognitive vulnerability), are confounded by the nature of the “self–selection” rather than random assignment of individuals of the quasi–experimental groups. Thus, the design does not permit an unambiguous test of causal hypotheses about the quasi–experimental factors. For example, the cognitive vulnerability (or stress) groups can differ on more than the selected characteristic, and when selecting people for a given cognitive vulnerability, one may also be inadvertently selecting individuals for neuroticism, gender, other psychopathology, or any other number of third variables that are correlated with the quasi–experimental variable. Two main strategies to minimize this difficulty of quasi–experimental designs use a “participant matching” approach, so as to equate the different study groups on potential third variables (or confounds). In one variant, the researcher ensures that when the groups as a whole are compared, they are shown not to differ (“samplewide matching”) on the


719

third variables. In the second variant, each group participant is paired (matched) on a case–by-case basis with a participant in another group who has similar characteristics (“case-by-case matching”). For example, each participant in a cognitive vulnerability group can be paired with another participant who is matched on potentially confounding third variables such as gender, educational level, or other demographic variables. Alternatively, statistical methods such as hierarchical regression analysis or analysis of covariance can be used to remove the effects of potential confounding variables, but none of these strategies fully replaces experimental control of the variables. CORRELATIONAL DESIGNS: CROSS–SECTIONAL, LOOK–BACK, AND PROSPECTIVE “Cross–sectional” (case control) designs compare a group with a psychological disorder to a normal control group (or perhaps, to groups with other disorders) on characteristics such as their respective scores on cognitive vulnerability measures. Such studies can be seen as preliminary tests or sources of hypotheses, but are inadequate for establishing the temporal precedence or stability of a vulnerability independent of the symptoms of the disorder. Their chief weakness is that they cannot rule out the chance that scores on vulnerability measures are simply correlates or consequences or “scars of the disorder, rather than risk factors or antecedent causes of the disorder (Just et al., 2001; Lewinsohn, Steinmetz, Larson, & Franklin, 1981). Similar limits apply to the causal inferences or interpretations that can be drawn from “remitted disorder” designs, in which previously symptomatic individuals are examined in a remitted state to see whether a hypothesized cognitive vulnerability is present. Such remitted designs cannot be used to determine whether the cognitive vulnerability factors of interest were actually present before the episode, or whether they are really an outcome of the disorder (they can be useful for circumscribed purposes, see Just et al., 2001; Riskind & Alloy, 2006). Retrospective designs are longitudinal studies in which the thrust is to “look backward” (instead of forward) over time—such as when participants are asked to recall information about their cognitive vulnerabilities (or past stresses) before their first episodes. The main limitation of these designs is that recall of participants can be influenced by forgetting, cognitive biases, or even the disorder itself. For example, depressed individuals could have biased recall of stressful events or past dysfunctional attitudes as a consequence of their current moods. In “follow–back” studies, which are more unbiased in these respects, objective records (e.g., medical records, personal diaries) of participants are lo-

720

RISKIND AND ALLOY

cated that existed before the onset of the first episode of a psychological disorder and compared for group differences. The major reason to prefer follow–back studies over retrospective designs is that the “study” groups examined are compared on objective data, or at least data that are coded independent of the experimenters’ knowledge of the diagnostic status of the groups. A potential problem such studies can encounter is that there have been changes in the methods or procedures by which the original objective data were recorded. A further major limitation of using all “look backward” designs as well as remitted disorder and cross–sectional (case control) designs in testing cognitive models is that individuals in the disorder groups developed clinical disorders for highly heterogeneous etiological reasons. For example, the origins of depressive disorders for some individuals might mainly involve a biological diathesis or dysfunctional interpersonal patterns, instead of the hypothesized cognitive vulnerability of interest. Thus, if researchers were simply to study groups of individuals who exhibit emotional disorders that are superficially similar in their manifestations, they may be examining disorders that are generated by means of quite different causal processes. That is, the individuals’ disorders have seemingly similar phenotypes, but different underlying causes or genotypes (see Alloy, Abramson, Safford, & Gibb, 2001, on hopelessness depression). If no differences in the hypothesized cognitive vulnerability are obtained, the results provide a particularly ambiguous basis for inference. It could prove important to sort out many other emotional disorders or their subtypes by their putative causal processes or underlying genotypes. Because researchers are ordinarily unable to conduct true experimental studies of the development of psychological disorder, the preferred design is a prospective design. In such a design, the potential cognitive vulnerability is assessed in participants prior to the onset of an episode (or symptoms) of their disorder. Prospective designs are valuable as they can help to establish both the vulnerability factor’s temporal precedence and independence from symptoms or previous diagnosis and psychological problems (Alloy et al., 1999). That is, “high risk” participants have not yet ever experienced the clinical disorder, which has the additional benefit of removing the potentially confounding effects of the previous presence of the disorders (e.g., of medication, hospitalization). Possible experimenter bias is also eliminated because the researcher does not know who will eventually develop the disorder. Prospective studies can also be used to establish whether the hypothesized cognitive vulnerability applies with specificity to the clinical disorder of interest and not other disorders (i.e., discriminant validity). In behavioral high–risk designs (e.g., Alloy & Abramson, 1999), a preferred form of prospective designs, researchers select participants who


721

are presently nondisordered because they possess behavioral (or cognitive) characteristics hypothesized to make them vulnerable to developing a particular disorder in the future. The researchers then follow these “high risk” participants prospectively, along with a comparison group of individuals who score low on the hypothesized risk factor (Alloy, Abramson, Whitehouse, Hogan, Panzarella, & Rose, 2006). The behavioral high–risk design allows researchers to establish the precedence and stability of the hypothesized cognitive vulnerability factor in individuals who do not presently possess the disorder of interest. The researchers can also examine the role of other factors (e.g., stress, protective factors) in influencing which high–risk participants later develop the emotional disorder. In all, evidence from multiple designs may provide the most compelling convergent validity for the effects of cognitive vulnerabilities. True experimental studies cannot usually be implemented for both ethical and practical reasons. Behavioral high–risk studies provide the best compromise that is presently available. However, other research designs can provide useful supplemental support for the construct validity of the hypothesized cognitive vulnerability (e.g., for proposed information–processing biases). Although we do not have space to delve into these here, there are also several difficult conceptual and methodological issues in cognitive vulnerability research. These include difficulties in the sometimes unwarranted assumption of the independence of measurements of cognitive vulnerability, stress, and psychological disorder (for more such issues, see Alloy et al., 1999). For example, sometimes, there can be direct effects of psychological disorders on behavior, effects of cognitive vulnerability to disorders on stress, and so on.

SUMMARY AND CONCLUSIONS The cognitive vulnerability paradigm provides a way of understanding the factors that contribute to the onset and recurrence of psychological problems and disorders, as well as the factors that contribute to their maintenance. Cognitive vulnerability models distinguish distal from proximal causes, developmental events from precipitating events, specific from nonspecific causes, and recognize that a variety of protective and exacerbating factors can affect the likelihood that cognitively vulnerable individuals will develop psychological disorders. Some cognitive vulnerability models may focus on common, or nonspecific factors that lack discriminating power, whereas other models provide ways of understanding factors that create a specific liability to particular disorders. Cognitive models also may provide ways to understand some rea-

722

RISKIND AND ALLOY

sons for the comorbidity of disorders. As has been seen, there are important logical criteria that must be satisfied to support the claim of a putative cognitive vulnerability and must guide the selection of research designs for testing such claims. The four remaining articles in this special issue illustrate the extension of cognitive vulnerability research to a variety of psychological problems, including anxiety, depression and bipolar disorders, and psychopathy.

REFERENCES Abramson, L.Y., Alloy, L.B., & Hogan, M.E. (1997). Cognitive/personality subtypes of depression: Theories in search of disorders. Cognitive Therapy and Research, 21, 247–265. Abramson, L.Y., Metalsky, G.I., Alloy, L.B. (1989). Hopelessness depression: A theory–based subtype of depression. Psychological Review, 96, 358–372. Abramson, L.Y., & Seligman, M.E.P. (1977). Modeling psychopathology in the laboratory: History and rationale. In J. Maser and M.E.P. Seligman (Eds.), Psychopathology: Experimental models. San Francisco, CA: Freeman. (pp. 1-26). Abramson, L.Y., Seligman, M.E.P. & Teasdale, J.D. (1978). Learned helplessness in humans: A critique and reformulation. Journal of Abnormal Psychology, 87, 49-74. Alloy, L.B., & Abramson, L.Y. (1999). The Temple–Wisconsin cognitive vulnerability to depression (CVD) project: Conceptual background, design and methods. Journal of Cognitive Psychotherapy, 13, 227–262. Alloy, L.B., Abramson, L.Y., Hogan, M.E., Whitehouse, W.G., Rose, D.T., Robinson, M.S., Kim, R.S., & Lapkin, J.B. (2000). The Temple-Wisconsin Cognitive Vulnerability to Depression project: Lifetime history of Axis I psychopathology in individuals at high and low cognitive risk for depression. Conceptual background, design, and methods. Journal of Abnormal Psychology, 109. Alloy, L.B., Abramson, L.Y., Raniere, D., & Dyller, I. (1999). Research methods in adult psychopathology. In P.C. Kendall, J.N. Butcher, & G.N. Holmbeck (Eds.), Handbook of research methods in clinical psychology (2nd ed.). New York: Wiley. Alloy, L.B., Abramson, L.Y., Safford, S.M., & Gibb, B.E. (2001). The cognitive vulnerability to depression (CVD) project: Current findings and future directions. In LB. Alloy & J.H. Riskind, Cognitive vulnerability to emotional disorders. New Jersey: Erlbaum. Alloy, L.B., Abramson, L.Y., & Viscusi, D. (1981). Induced mood and the illusion of control. Journal of Personality and Social Psychology, 41, 1129–1140. Alloy, L.B., Abramson, L.Y., Whitehouse, W.G., Hogan, M.E., Panzarella, C., & Rose, D.T. (2006). Prospective incidence of first onsets and recurrences of depression in individuals at high and low cognitive risk for depression. Journal of Abnormal Psychology, 115, 145–156. Alloy, L.B., Kelly, K.A., Mineka, S., & Clements, C.M. (1990). Comorbidity of anxiety and depression disorders: A helplessness–hopelessness perspective. In J.D. Maser & C.M. Cloninger (Eds.), Comorbidity of mood and anxiety disorders. Washington, DC: American Psychiatric Press. Alloy, L.B., & Riskind, J. H. (Eds). (2006) Cognitive Vulnerability to Emotional Disorders. Hillsdale, NJ: Erlbaum. Barron, R.M., & Kenny, D.A. (1986). The moderator–mediator variable distinction in social


723

psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Beck, A.T. (1967). Depression: Clinical, experimental, and theoretical aspects. New York: Harper & Row. Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International Universities Press. Beck, A. T. (1991). Cognitive therapy: A 30-year retrospective. American Psychologist, 46, 368–375. Beck, A. T., & Clark, D. A. (1997). An information processing model of anxiety: Automatic and strategic processes. Behaviour Research and Therapy, 35, 49–58. Brown, G.W., & Harris, T.O. (1978). Social origins of depression. London: Tavistock. Chorpita, B.F., & Barlow, D.H. (1998). The development of anxiety: The role of control in the early environment. Psychological Bulletin, 124, 3–21. Clark, D.A. (1997). Twenty years of cognitive assessment: Current status and future directions. Journal of Consulting & Clinical Psychology, 65, 996–1000. Clark, D.A., Beck, A.T., & Alford, B. A. (1999). Scientific Foundations of Cognitive Theory and Therapy of Depression. New York: John Wiley & Sons. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Earlbaum. Dobkin, R.D., Panzarella, C., Fernandez, J., Alloy, L.B., & Cascardi, M. (2004). Adaptive inferential feedback, depressogenic influences and depressed mood: A laboratory study of the expanded hopelessness theory of depression. Cognitive Therapy & Research, 28, 487–509. Fridja, N. (1987). The Emotions. New York: Cambridge University Press. Gibb, B.E., Alloy, L.B., Abramson, L.Y., & Marx, B.P. (2003).Childhood maltreatment and maltreatment-specific inferences: A test of Rose and Abramson’s (1992) extension of the hopelessness theory. Cognition & Emotion, 17, 917–931. Hammen, C. (1991). The generation of stress in the course of unipolar depression. Journal of Abnormal Psychology, 100, 555–561. Holmbeck, G.N. (1997). Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: Examples from the child–clinical and pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65, 599–610. Freud, S. (1919/1959). Lines of advance in psychoanalytic therapy. Translated by J. Riviere and J. Strachey. In J. Strachey (Ed.), The Standard Edition of the Complete Psychological Works of Sigmund Freud. London: Hogarth Press. Ingram, R.E. (1990). Self–focused attention in clinical disorders: Review and a conceptual model. Psychological Bulletin, 107, 156–176. Ingram, R.E. (2003). Origins of cognitive vulnerability to depression. Cognitive Therapy & Research, 27, 77–88. Ingram, R.E., Bailey, K., & Siegle. (2004).Emotional information processing and disrupted parental bonding: Cognitive specificity and avoidance. Journal of Cognitive Psychotherapy: An International Quarterly, 53–64. Ingram, R.E., Miranda, J., & Segal, Z.V. (1998). Cognitive vulnerability to depression, New York: Guilford Press. Ingram, R.E., Miranda, J., & Segal, Z. (2006). Cognitive vulnerability to depression: Theory, research, and future directions. In LB. Alloy & J.H. Riskind, Cognitive vulnerability to emotional disorders. New Jersey: Erlbaum. Johnson, S.L., & Roberts, J.E. (1995). Life events and bipolar disorder: Implications from biological theories. Psychological Bulletin, 117, 434–449. Joiner, T.E., Alfano, M.S., & Metalsky, G.I. (1992). When depression breeds contempt: Reas-

724

RISKIND AND ALLOY

surance–seeking, self–esteem, and rejection of depressed college students by their roommates. Journal of Abnormal Psychology, 101, 165–173. Jolly, J.B., & Kramer, T.A. (1994). The hierarchical arrangement of internalizing cognitions. Cognitive Therapy and Research, 18, 1–14. Just, N., Abramson, L.Y., & Alloy, L.B. (2001). Remitted depression studies as tests of the cognitive vulnerability hypotheses of depression onset: A critique and conceptual analysis. Clinical Psychology Review, 21, 63–83. Last, C.G., Barlow, D. H., & O’Brien, G,.T. (1984). Precipitants of agoraphobia: Role of stressful life events. Psychological Reports, 54, 567–570. Lazarus, R. (1991). Emotion and Adaptation. New York: Oxford University Press. Lewinsohn, P.M., Steinmetz, J.L., Larson, D.W., & Franklin, J. (1981). Depression-related cognitions: Antecedent or consequence? Journal of Abnormal Psychology, 90, 213-219. Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and emotional disorders. Annual Review of Psychology, 45, 25–50. Monroe, S.M., & Simons, A.D. (1991). Diathesis–stress theories in the context of life stress research: Implications for the depressive disorders. Psychological Bulletin, 110, 406–425. Nolen–Hoeksema, S., Morrow, J., & Frederickson, B.L. (1993). Response styles and the duration of episodes of depressed mood. Journal of Abnormal Psychology, 10, 20–28. Ortony, A., Clore, G.L., & Collins, A. (1988). The cognitive structure of emotions. New York: Cambridge University Press. Panzarella, C., Alloy, L.B., & Whitehouse, W.G. (in press). Expanded hopelessness theory of depression: On the mechanisms by which social support protects against depression. Cognitive Therapy and Research, 30. Paykel, E.S. (Ed.). (1982). Handbook of affective disorders. New York: Guilford. Riskind, J.H. (1989). The mediating mechanisms in mood and memory: A cognitive–priming formulation. Journal of Social Behavior and Personality, 4, 173–184. Riskind, J.H., & Alloy, L.B. (2006). Cognitive vulnerability to emotional disorders: Theory, design, and methods. In L.B. Alloy & J.H. Riskind (Eds.), Cognitive vulnerability to emotional disorders (pp. 1–29). Hillsdale, NJ: Erlbaum. Riskind, J.H., Moore, R., Harman, B., Hohman, A.A., Beck A. T., & Stewart, B. (1991). The relation of generalized anxiety disorder in general and dysthymic disorder in particular. In R.L. Rapee & D.H. Barlow (Eds.), Chronic anxiety: Generalized anxiety disorder and mixed–anxiety depression. New York: Guilford Press. Riskind, J. H., & Rholes, W. S. (1984). Cognitive accessibility and the capacity of cognitions to predict future depression: A theoretical note. Cognitive Therapy and Research, 8, 1–12. Riskind, J.H., & Williams, N.L. (2006). A unique vulnerability common to all anxiety disorders: The looming maladaptive style. In LB. Alloy & J.H. Riskind, Cognitive vulnerability to emotional disorders. New Jersey: Erlbaum. Riskind, J.H., Williams, N.L., Altman, M.D., Black, D.O., Balaban, M.S., & Gessner, T.L. (2004). Developmental antecedents of the looming maladaptive style: Parental bonding and parental attachment insecurity. Journal of Cognitive Psychotherapy: An International Quarterly, 18, 43–52. Riskind, J. H., Williams, N. L., Gessner, T., Chrosniak, L, & Cortina, J. (2000). A Pattern of Mental Organization and Danger Schema Related to Anxiety: The Looming Maladaptive Style. Journal of Personality and Social Psychology, 79, 837–852. Rogers, G.M., Reinecke, M.A., & Setzer, N. J. (2004). Childhood attachment experience and adulthood cognitive vulnerability: Testing state dependence and social desirability hypotheses. Journal of Cognitive Psychotherapy: An International Quarterly, 793–90. Roseman, I.J., Spindel, M.S., & Jose, P.E. (1990). Appraisals of emotion-soliciting events:


725

Testing a theory of discrete emotions. Journal of Personality and Social Psychology, 59, 899–915. Roy–Byrne, P. B., Geraci, M., & Uhde, T. W. (1986). Life events and the onset of panic disorder. American Journal of Psychiatry, 143, 1424–1427. Safford, S.M., Alloy, L.B., Crossfield, A.G., Marocco, A.M., & Wang, J.C. (2004). The relationship of cognitive style and attachment to depression and anxiety in young adults. Journal of Cognitive Psychotherapy: An International Quarterly, 25–42. Seligman, M.E.P. (1975). On depression, development, and death. San Francisco, CA: Freeman. Shahar, G., Black, D., & Riskind, J.H. (2004). Looming vulnerability and a stress generation model. Manuscript in preparation. Williams, N.L., & Riskind, J.H. (2004). Adult romantic attachment and cognitive vulnerabilities to anxiety and depression: Examining the interpersonal basis of vulnerability models. Journal of Cognitive Psychotherapy: An International Quarterly, 18, 2–24. Williams, J.M.G., Watts, F.N., MacLeod, C., & Mathews, A. (1988). Cognitive psychology and emotional disorders. Chichester, England: Wiley. Zinbarg, R. E., & Barlow, D. H. (1996). Structure of anxiety and the anxiety disorders: A hierarchical model. Journal of Abnormal Psychology, 105, 181–193. Zuckerman, M. (1999). Vulnerability to psychopathology: A biosocial model. Washington, DC: American Psychological Association.