Further evidence for biased semantic networks in obsessive-compulsive disorder (OCD): When knives are no longer associated with buttering bread but only with stabbing people

Background and objectives Semantic network models suggest that individuals with obsessive-compulsive disorder (OCD) process words with multiple meanings (e.g., "knife") more likely in an OC-related (i.e., "weapon") than in a neutral way (i.e., "cutlery"). Initial evidence was found in an online study. The aim of the current study was to investigate semantic networks in a clinical OCD sample and particularly to identify whether changes in semantic networks following the add-on intervention association splitting (AS) exceeded changes expected through cognitive behavioral therapy (CBT) alone. Methods An association task was presented to 36 healthy controls and 70 OCD patients over a period of eight weeks with OCD patients receiving CBT and an add-on intervention (randomized allocation to either AS or a computerized cognitive training). Participants were asked to generate up to five associations to standardized (OC-relevant, negative, neutral) and individual cue words. Associations were rated with regard to OC-relevance and valence. Results Analyses revealed that OCD participants produced a) significantly more OC-relevant associations and b) more negative associations than controls for cue words. In the OCD sample, the OC-relevance and valence of associations changed after therapy for personal cue words. This effect was associated with AS at statistical trend level. Limitations No clinical control group was recruited; no inter-rater reliability was assessed for the association task. Conclusions Further evidence for biased associative networks in OCD was found. Associations of individually chosen cue words proved to be modifiable by therapy.