ABSTRACT
Repetitive negative thinking (RNT) describes a recursive, unproductive pattern of thought that is commonly observed in individuals who experience anxiety and depression. Past research on RNT has primarily relied on self-report, which fails to capture the potential mechanisms that underlie the persistence of maladaptive thought. We investigated whether RNT may be maintained by a negatively biased semantic network. The present study used a modified free association task to assess state RNT. Following the presentation of a valenced (positive, neutral, negative) cue word, participants generated a series of free associates, which allowed for the dynamic progression of responses. State RNT was conceptualised as the length of consecutive, negatively valenced free associates (i.e. chains). Participants also completed two self-report measures that assessed trait RNT and trait negative affect. Within a structural equation model, negative (but not positive or neutral) response chain length positively predicted trait RNT and negative affect, and this was only the case for positive (but not negative or neutral) cue words. These results suggest that RNT tendencies may be reflected in semantic retrieval and can be assessed without self-report.
Acknowledgements
The authors would like to thank Evelyn Behar, PhD for helpful discussions and valuable comments.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Although much research and theory on state RNT has focused on a distinction between rumination and worry, recent work suggests that these behaviours reflect the same underlying cognitive process, with differences relating to disorder-specific content and temporal orientation (Wahl et al., Citation2019).
2 We opted to not introduce response time constraint to the task in order to minimise task-induced stress (e.g., participants who are less adept at speed typing might feel rushed).
3 Although it is possible that this score reflects the participant’s true perseverative thinking tendencies, we were concerned that the score may instead reflect a responding strategy that indicates inattention. Since the PTQ does not include reverse-scored items, a score of 0 could be the result of adopting a straight-lining approach when the participant responded to the PTQ. To err on the side of caution, we opted to remove the participant’s data from analyses.
4 Since PTQ and STAI-T are highly correlated in our sample (r = .72), we opted for path analysis as our primary analytic approach, as it allowed us to model that intercorrelation and to apply a rigorous test of whether the variables from the FAST would predict unique variance in trait anxiety and repetitive negative thinking, with both scores included together in all study models. Another approach is to conduct a linear regression model with a composite internalising variable (by averaging the z-scores of PTQ and STAI-T). We tested this model with the new composite variable as the outcome variable, and the same three predictor variables (negative chain lengths for positive, negative, and neutral seed). The findings from this model are consistent with the results of the reported path analysis, where only the positive seed/negative chain significantly predicted the composite internalising score (β = .258, p = .012; other p > .56).