Background and objectives
According to the negativity hypothesis, depressed individuals are over-pessimistic due to negative self-concepts. In contrast, depressive realism suggests that depressed persons are realistic compared to their nondepressed controls. However, evidence supporting depressive realism predominantly comes from judgment comparisons between controls and nonclinical dysphoric samples when the controls showed overconfident bias. This study aimed to test the validity of the two accounts in clinical depression and dysphoria.
Methods
Sixty-eight participants, including healthy controls (n = 32), patients with DSM-IV major depression (n = 20), and dysphoric participants with CDC-defined chronic fatigue syndrome (n = 16) performed an adjective recognition task and reported their item-by-item confidence judgments and post-test performance estimate (PTPE).
Results
Compared to realistic PTPE made by the controls, patients with major depression showed significant underconfidence. The PTPE of the dysphoric participants was relatively accurate. Both the depressed and dysphoric participants displayed less item-by-item overconfidence as opposed to significant item-by-item overconfidence shown by the controls.
Limitations
The judgment-accuracy patterns of the three groups need to be replicated with larger samples using non-memory task domains.
Conclusion
The present study confirms depressive realism in dysphoric individuals. However, toward a more severe depressive emotional state, the findings did not support depressive realism but are in line with the prediction of the negativity hypothesis. It is not possible to determine the validity of the two hypotheses when the controls are overconfident. Dissociation between item-by-item and retrospective confidence judgments is discussed.
The cognitive model of depression forwarded by Beck (Beck, 1967 and Beck, 1987) proposes that depressed individuals view themselves as defective, inadequate, diseased and deprived. As a result of these negative self-concepts, depressed persons believe that they are undesirable and worthless and tend to underestimate or criticize themselves. The “negativity hypothesis” (Clark et al., 1999 and Gilboa-Schechtman et al., 2002) thus leads to the prediction that depressed persons will be overly pessimistic in their self-referent evaluations (Dunn et al., 2007, Stone et al., 2001 and Whitton et al., 2008). However, several studies (Alloy and Abramson, 1979, Alloy and Ahrens, 1987 and Crocker et al., 1988) appear to support the contrary view of “depressive realism”. According to depressive realism, depressed persons are neither over-optimistic nor over-pessimistic but rather realistic.
Nevertheless, most studies supporting depressive realism have included only dysphoric/mildly depressed individuals –– as defined by Beck Depression Inventory (BDI) (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) scores between 10 and 19 – and often have used decision tasks for which there is no objectively correct answer (e.g., the contingency judgment paradigm). The results obtained from a mildly depressed population may be different from those demonstrated by severely depressed individuals (Clark et al., 1999) and decision tasks without objectively correct answers cannot determine the extent to which an individual is over-optimistic, over-pessimistic or realistic (Haaga and Beck, 1995 and Stone et al., 2001).
A further shortcoming of many previous studies involves the inclusion of only one outcome measure or task, on which the healthy controls showed overconfidence. It is not possible to differentially test the validity of the two hypotheses if one only compares the judgment accuracy of depressed or dysphoric versus nondepressed individuals when healthy controls show positive bias in their confidence judgments because both hypotheses make the same prediction under this experimental condition (Dobson and Franche, 1989 and Fu et al., 2005). The two hypotheses make competing predictions only when healthy controls show either accurate judgments or underconfidence (Stone et al., 2001). Under these conditions, the negativity hypothesis predicts that depressed individuals will demonstrate a self-deprecating bias, whereas depressive realism predicts that depressed individuals still will be realistic.
The current investigation sought to address each of these methodological shortcomings. First, a clinically depressed group, a dysphoric group, and matched controls were tested. The dysphoric group was comprised of individuals with chronic fatigue syndrome (CFS). Previous research has shown that the BDI scores of these chronic fatigue patients often indicate dysphoria (Johnson et al., 1996 and Moss-Morris and Petrie, 2001). Inclusion of the dysphoric CFS group was to further test the validity of depressive realism in individuals with mild depressive symptoms in the absence of clinical depression.
Second, we used a recognition memory task for which there were objectively correct answers thereby facilitating the measuring of the degree of judgment accuracy. Thirdly, according to our pilot study, healthy controls (n = 45) demonstrated the required experimental condition to contrast the validity of the two hypotheses on this recognition task. Specifically, healthy controls showed overconfidence when judgments were made at an item-by-item level, but underconfidence on a retrospective judgment, that is, a post-test performance estimate (PTPE). Because the healthy controls showed differential judgment-accuracy patterns on the two confidence assessments, both types of judgments were included.
In summary, we used a recognition task in which the healthy controls showed distinct patterns of judgment accuracy to test the validity of the depressive realism versus the negativity hypotheses in two patient groups: individuals experiencing a current episode of major depression, and dysphoric individuals with CFS. Under the situations where the healthy controls showed realistic or underconfident judgment, we hypothesized that (1) the confidence assessment of the individuals with depression and dysphoria should be realistic if the depressive realism account is correct; On the other hand, (2) if the negativity hypothesis is valid, then the depressed patients should demonstrate greater underconfidence compared to the healthy controls.
