Figure 1812

Interpretive bias in high-anxious and low-anxious participants as a function of stress condition (high or low) and context-target delay. Adapted from Calvo and Castillo (1997).

normal controls to interpret such sentences in a threatening way, and there were no group differences in response bias.

More detailed information about interpretive bias was reported by Calvo and Castillo (1997). They presented ambiguous sentences concerned with ego threat, physical threat, or neutral events under low or high stress conditions. Each sentence was followed by a disambiguating sentence containing a target word confirming or disconfirming the consequence implied by the ambiguous sentence. The target word was presented either 500 (short delay) or 1200 milliseconds (long delay) after the preceding context, and the task was to name the target word as rapidly as possible.

Interpretive bias was shown when the target word was named faster when it confirmed a threatening interpretation than when it disconfirmed such an interpretation. The high-anxious participants showed strong interpretive bias at the long delay under high stress conditions, but not at the short delay or with low stress (see Figure 18.12). What do these findings mean? Interpretive bias depends on state anxiety or anxiety as a mood stateas well as on anxiety as a personality dimension.In addition, interpretive bias does not occur rapidlyand automatically, but rather involves subsequentstrategic processes.


The effects of depression on emotion-congruent processing have been studied in various attentional and perceptual tasks. Normal and clinical depression have been considered. Depression in normals has often been assessed by questionnaires such as the Beck Depression Inventory.

Attentional bias

There is little convincing evidence for the existence of attentional bias among depressed individuals. In a study using the dot-probe task, MacLeod et al. (1986) found that anxious patients showed an attentional bias when a threatening and a neutral word were presented together. However, depressed patients showed no attentional bias.

Gotlib, McLachlan, and Katz (1988) used a modified version of the task employed by MacLeod et al. (1986), but presented emotionally positive words as well as emotionally negative and neutral words. There was an effect of depression on attention, in that the non-depressed participants attended selectively to positive stimuli. However, Gotlib et al. (1988) did not establish whether the attentional bias was due to anxiety rather than to depression. Mogg et al. (1991) carried out a replication of the Gotlib et al. (1988) study, and found that the attentional bias was due to state anxiety rather than to depression.

Most studies on attentional bias in depression have probably measured fairly automatic attentional processes. An exception is a study by Matthews and Antes (1992) on controlled attentional processes. They measured the eye movements of depressed and non-depressed individuals who were presented with slides containing "sad" and "happy" regions. The depressed participants focused relatively more on the sad regions and less on the happy regions.

Interpretive bias

The effects of depression on interpretation of ambiguity have been assessed in several studies. The evidence consistently indicates that there is an interpretive bias in depressed individuals. Various studies (discussed by Rusting, 1998) have made use of the Cognitive Bias Questionnaire. Events are described briefly, with participants having to select one out of four possible interpretations of each event. Depressed patients consistently select more negative interpretations than controls.

Pyszczynski, Holt, and Greenberg (1987) carried out a study in which depressed and non-depressed students rated various possible future events. The depressed participants rated negative future events as more likely to happen than did the non-depressed participants, whereas the opposite was the case for positive future events.


Rusting (1998) reviewed the relevant research literature more thoroughly than has been possible here. As a result, we will start with one of her main conclusions (1998, pp. 189-190):

Most of the traditional emotion-congruency literature has examined the effects of moods and traits on the processing of emotional stimuli separately. However, this literature has yielded mixed findings across perception, attention, interpretation/judgement, recall and recognition, and autobiographical memory tasks. Although many studies do obtain evidence for mood-congruency and trait-congruency, some studies have found mood-incongruency effects, and others have found no effects of mood or personality at all.

There are several possible reasons why the findings have been somewhat inconsistent. First, participants who are in a negative mood state may use various strategies (e.g., thinking about positive events) to improve their mood. If these strategies succeed, they will weaken any findings. Second, some studies of mood congruity may have produced non-significant findings because of discrepancies between the emotional state of the participants and the specific emotional content of the learning material. For example, participants in a depressed mood may be asked to learn anxiety-related words. Relevant evidence was reported by Ingram et al. (1987). Depressed but non-anxious participants had enhanced recall of depressionrelevant words, whereas anxious but non-depressed participants had greater recall of anxiety-relevant words. Third, there have generally been more significant findings when the participants had to process the learning material with reference to themselves. Why is this? According to Rusting (1998, p. 183), "studies that incorporate self-referent processing tasks may actually be tapping into stable structures in memory, rather than producing effects that are purely dependent on temporary mood states."

There is only limited support for the traditional approach, in which it was assumed that personality and mood have independent effects on emotional processing. What about the mediator and moderator approaches? There have been practically no attempts to assess the mediator approach. The evidence is most consistent with the moderator approach. According to Rusting (1998, p. 188), the evidence for this approach, "is fairly consistent across attention, interpretation/judgement, recall and recognition, and autobiographical memory tasks. In each of these areas there is some evidence indicating that certain personality traits moderate whether mood-congruency or mood-incongruency effects are obtained."

It makes sense that emotional processing should depend interactively on traits and mood states. Any given personality trait is associated with a richly interconnected network of relevant emotional information and knowledge. However, there is a huge amount of information in long-term memory, and so this trait-relevant information is only likely to influence emotional processing when it is activated by the appropriate mood state. Thus, maximal effects on emotional processing will be obtained when individuals possess the relevant knowledge structures (personality traits), and when these knowledge structures are fully activated (mood states).

Theories of emotional processing

The research findings provide more support for the theoretical approach of Williams et al. (1988, 1997) than for Bower's network theory or Beck's schema theory. For example, there is strong evidence that anxiety is associated with an attentional bias, but the evidence is much weaker so far as depression is concerned. According to network and schema theories, individuals in a depressed mood should have facilitated processing of (and attention to) mood-congruent stimuli, and should thus show attentional bias. In contrast, Williams et al. (1997) argued that depressed individuals do not engage in excessive perceptual processing of threat-related stimuli, and so they should not have an attentional bias.

The theory of Williams et al. is also better equipped to handle the findings on explicit and implicit memory biases. Network and schema theories predict that anxious and depressed individuals should show both kinds of memory bias. However, it follows from the Williams et al. theory that anxiety should tend to produce an implicit memory but not an explicit memory bias, whereas the opposite should be the case with depression. There is some support for these predictions, but anxiety is often associated with an explicit memory bias.

Network and schema theories also have difficulties with the findings from studies on perceptual word recognition of emotional and neutral words. In most of these studies, words were presented very briefly but for progressively longer until they were identified correctly. As Niedenthal, Setterlund, and Jones (1994, p. 93) concluded, "Research designed to explore emotional influences in perception has revealed little systematic evidence for changes in the efficiency of word recognition as a function of emotional state." For example, Gerrig and Bower (1982) put hypnotised participants into a happy or angry mood, and then presented positive, negative, and neutral words. There was no evidence of any emotion-congruent effects in two experiments, although this is what is predicted by network theory.

Why have most studies failed to obtain evidence for facilitated processing of emotion-congruent information in perceptual word recognition? Selective attention was not involved, in that the participants were presented with only one stimulus at a time. If mood states influence selective attentional processes, as is proposed by Williams et al. (1988, 1997), this would explain why there is evidence for attentional bias but not for emotion-congruent effects in word recognition.

What of the future? Theoretically, there is a need to develop the approach of Williams et al. They have made good use of the distinction between perceptual and conceptual processes to shed light on differences between anxious and depressed individuals in the processing of threat. However, most tasks involve a mixture of perceptual and conceptual processes, and it is often hard to identify their respective contributions. Human information processing is very complicated, and so much more complex theories will be required, which may attach less importance to the imprecise distinction between perceptual and conceptual processes.

Experimentally, there has been too much emphasis on the processing of threat-related environmental stimuli (e.g., words). Anxious individuals often exhibit cognitive biases for internal stimuli. For example, patients with panic disorder catastrophically misinterpret their own physiological activity (Clark, 1986), and patients with social phobia have an interpretive bias for the adequacy of their own social behaviour (Stopa & Clark, 1993), assuming it to be much less adequate than is actually the case.

Finally, it is important to emphasise the potential relevance of research on emotional processing for an understanding of the anxiety disorders and clinical depression. For example, attentional, interpretive, and memory biases can all produce increased levels of anxiety or depression in individuals who are already anxious or depressed. Eysenck (1997, p. 100) discussed some of the implications so far as anxiety is concerned:

Cognitive biases applied to the processing of threat-related information increase the level of state anxiety, and elevated state anxiety exaggerates the cognitive biases. This can create a positive feedback loop which eventually creates extremely high levels of uncontrollable experienced anxiety. Anxious patients often have more pronounced cognitive biases than normals high in trait anxiety, and so may be especially likely to become trapped in such positive feedback loops.

Happiness Today

Happiness Today

Unhappy With Your Current Situation? Follow The Principles Within This New audio series and Guide and Create True Happiness Everyday Of Your Life. Download To Discover How To Live A More Fulfilling And Happy Life Starting Today.

Get My Free Ebook

Post a comment