The interplay between attention, interpretation and memory in dysphoria: an investigation of the combined cognitive bias hypothesis

Introduction. Cognitive accounts assert that emotionally biased information-processing mechanisms play a central role in the onset and maintenance of depression. Although research has yielded substantial empirical data demonstrating cognitive biases in depression, scientific understandings of interrelations among these biased cognitive processes remains limited in depressed samples. This study examined the interplay between attention, interpretation, and memory biases in dysphoria. It was hypothesized that different cognitive bias parameters would be correlated. To test whether cognitive biases would operate in isolation or in concert, path models including and excluding mutual relations among biases were evaluated in a theory-driven manner. Method. The sample consisted of 19 dysphoric and 37 non-dysphoric undergraduate students. Participants completed a computerized version of the scrambled sentences test while their eye movements were recorded. A subsequent incidental free recall task assessed memory for the unscrambled sentences. Results. Attention biases were indexed by the total fixation durations on negative vs. positive words within a scrambled sentence. Interpretation biases were assessed by the percentage of emotional sentences unscrambled negatively. Memory biases were measured by the percentage of emotional sentences recalled negatively. Dysphoric individuals demonstrated negative biases in attention (p=.07), interpretation (p<.01) and memory processes (p<.05) compared with non-dysphoric individuals. Significant positive correlations between all bias indices emerged (all p’s <.05). Path analyses revealed an acceptable fit only for a model including interrelations among cognitive biases (χ²(3)=4.16, p=.24, CFI=0.98, TLI=0.96, RMSEA=0.08). Parameter estimates indicated that dysphoria group status predicted attention (p=.07) and interpretation biases (p<.001). Attention biases predicted interpretation biases (p=.05) which in turn predicted memory biases (p<.001). Discussion. The observed pattern of correlations among bias indices provides support for the broader construct of information-processing bias. Consistent with predictions by cognitive accounts, results suggest that biases rather operate in concert than in isolation and provide preliminary support for a cascading effect of attention bias on interpretation and memory biases. Implications for future research will be discussed