The network theories of Bower (see Chapter 3) and the schema theory of Beck (see above) predicted that a wide range of cognitive biases should be found in emotional disorders such as anxiety and depression. The failure to find such global biases prompted Williams et al. (1988, 1997) to propose an empirically based model in which cognitive biases were specific to specific emotional disorders.
Williams et al. (1988, 1997) took as their theoretical starting point the distinction made by Graf and Mandler (1984) between "priming" and "elaboration" of stimuli. That is, priming is an automatic stage of processing in which the stimulus may be linked with its representation in long-term memory, whereas elaboration refers to subsequent strategic or resource-demanding processes. We should note, however, that this particular distinction between "priming" and "elaboration" is not an ideal one, because of the more general meaning of "priming" in the cognitive literature which can refer to both automatic and strategic or controlled processes (e.g., Neely, 1977). This caveat should be borne in mind therefore during the following discussion of the theory.
A summary of the Williams et al. (1988, 1997) model is presented in Figure 4.4. In the case of anxiety disorders Williams et al. propose that automatic priming processes are biased towards the detection of anxiety-relevant stimuli or situations. For example, the individual with a dog phobia may automatically process a stimulus out of focal awareness as if it were a dog, whereas the normal individual would be
Figure 4.4 Summary of the Williams et al. (1997) model of anxiety and depression.
more likely to perceive the object as a non-threatening four-legged table. Thus the preattentive processes become "tuned" to detect personally significant stimuli that range from the innocuous, such as one's name, to objects with which the individual may have had unpleasant experiences, be it dogs or particular individuals. However, Williams et al. (1988, 1997) further propose that although initial priming or automatic processes are biased towards the detection of threat in anxiety, subsequent elaborative processes are biased away from the processing of threat. This proposal is based primarily on the general failure to find mnemonic biases in anxious individuals, despite the fact that they demonstrate preattentive biases (see Harvey et al., 2004, and Mathews & MacLeod, 2005, for recent summaries). Williams et al. state that this shift of strategic resources away from threatening stimuli is evident, for example, in the Watts Trezise, and Sharrock (1986; also Watts, 1986) study of spider-phobic people, in which they found poorer free recall of spider-related information in spider-phobic people despite the same individuals showing biases on a Stroop task, in which spider-related words were found to interfere with the naming of the ink colours in which the words were presented (see Chapter 6).
In contrast to anxiety, Williams et al. (1988, 1997) propose that the main cognitive biases evident in depression are resource-demanding elaborative ones that are most apparent in mnemonic tasks. They interpret studies of attentional biases in depression either as direct evidence that there are no such biases (e.g., MacLeod, Mathews, & Tata, 1986) or as evidence that anxiety levels were not accounted for in studies where attentional biases have been reported for depression (e.g., Gotlib & McCann, 1984). However, there is overwhelming evidence that depressed individuals show biases in the recall of emotionally valent material in a range of free recall, cued recall, negative mood induction, and autobiographical memory tasks (e.g., Blaney, 1986; Bower & Forgas, 2000). There remains some question, though, as to whether the mnemonic biases obtained with depressed individuals are mainly due to a "loss of the positive" in that positive self-related material is retrieved more slowly and less readily (cf. Figure 3.8 and the discussion in Chapter 3 of Bower's studies of positive and negative mood induction), as much as they may be due to a "gain of the negative", a question that we will consider in more detail in Chapter 7. A further issue of course is that depression and anxiety are highly correlated. Given that they exert their effects supposedly independently with the Williams et al. model, it remains perplexing why depressed individuals, who are almost always also anxious, do not also show preattentive anxiety-related bias, nor why anxious individuals, who are often also depressed, do not show depressogenic elaborative biases.
The great strength of the framework presented by Williams et al. (1988, 1997) is that it provided a focus for the dissatisfaction felt by many researchers with the predictions for global cognitive biases that emerged from the theories of Beck and Bower. The failures to find biases across a range of tasks and across a range of time periods were drawn together in Williams et al.'s empirically driven proposals. However, the Williams et al. approach provides a starting point rather than an aetiological theory; it draws attention to the possibility that certain cognitive biases are more likely to be characteristic of certain emotional disorders than of others. Nonetheless the evidence, as Williams and his colleagues themselves acknowledge, is by no means as clear-cut as their framework suggests; for example, biased recall for negative material has been found for agoraphobic individuals (Nunn, Stevenson, & Whalan, 1984) and for those with generalised anxiety disorder (Friedman, Thayer, & Borkovec, 2000), and priming effects have been obtained for depressed patients as well as for anxious ones (Dalgleish, Cameron, Power, & Bond, 1994; Power et al., 1996). One of the weaknesses of studies of attentional biases in depression may be that the material used has not been of a sufficiently personally relevant form; thus, studies of specific phobias may typically use stimuli relevant to the phobic object in addition to more general emotionally valent material, whereas studies of depression may need to use more personally relevant material than that employed to date, for example material related to the individual's most valued role or goal, before such attentional and pre-attentional biases can be excluded. Equally, failures to find elaborative biases in anxiety may also be a consequence of asking anxious individuals the wrong questions. That is, although the important question in relation to depression may be the processing of material in relation to the self, in anxiety it may be the estimated likelihood of threatening outcomes from which the biases stem; thus, the central role of 'worry' in anxiety disorders (e.g., MacLeod, Rose, & Williams, 1993) surely represents the elaborative processing of turning molehills into mountains or, more appropriately, turning moles into cancerous growths. If worry is based on elaborative processes but is not associated with mnemonic biases for worry-related material, then it will be necessary to consider inhibitory effects in retrieval as an alternative to the non-elaboration model of Williams et al. We await the proper test of these ideas, although we will consider more of the relevant empirical data in Chapters 6 and 7 when we discuss anxiety and depression, respectively.
Was this article helpful?