Abstract

Cognitive bias (Stroop interference and implicit memory bias) for masked and unmasked threat words (illness words and negative emotion words) was investigated for its associations with emotion (anxiety, anger/aggression, and positive affect) and somatic complaints in a randomly selected community sample of 138 individuals. Because measures of cognitive bias are inherently bipolar, the data were tested for both linear and curvilinear trends. Cross-sectional analysis showed that the associations between somatic anxiety and Stroop interference for unmasked threat words were better described in terms of curvilinear than linear models, and that degree of somatic complaints was associated with Stroop facilitation for masked threat words. Longitudinal analysis showed that Stroop interference for masked threat words predicted lack of positive affect during an ensuing eight weeks daily recording period. Implicit memory bias was found to be negatively associated with measures of anger/aggression,both in the cross-sectionaland longitudinal analyses. We suggest that: (a) cognitive bias should be treated as a bipolar construct in future research; (b) the phenomenon of Stroop facilitation for threat words deserves further exploration; and (c) more research should address the question about the relationship between cognitive bias and other emotions than anxiety (e.g., positive affect and anger).