Background Adult patients with anxiety disorders (ADs) as well as healthy children and adolescents generalize learned fear responses to a larger degree than healthy adults. Although such overgeneralization of fear is central to theories of AD, little research has examined the respective mechanisms in pediatric populations. Therefore, we investigated behavioral and neural processes associated with fear generalization in adolescent patients with AD. Methods Magnetoencephalography (MEG) was recorded while 30 adolescent patients with AD and matched healthy controls (HCs) performed a prevalidated fear conditioning and generalization paradigm. Differently tilted Gabor gratings served as conditioned and generalization stimuli (CS+, CS−, GS) and a screaming female face as the unconditioned stimulus (UCS). Participants rated CS and GS regarding fear and UCS expectancy. The final analyses included participants showing good data quality and contingency awareness (15 AD, 16 HC; mean age: 15.65 ±1.02, 14–17 years). Results On the behavioral level, evidence for qualitatively different generalization patterns between groups was weak. Instead, fear and UCS‐expectancy ratings were overall higher in the AD group. On the neural level, qualitative differences emerged as a function of time. AD patients showed lower neural responses to the CS− and CS−‐like GS in frontoparietal regions (330–370 ms), while HCs showed the reverse pattern. Additionally, AD patients showed lower neural responses to the CS+ and CS+‐like GS in a sensory occipitotemporal region (>230 ms) that later showed higher responses to the CS+ (a positive gradient) in both groups (>360 ms). Conclusions Results argue against strong qualitative differences in fear generalization in adolescent AD on a behavioral level. The observed time‐dependent qualitative differences in magnetoencephalographic responses occurred in brain regions associated with inhibitory processes (frontoparietal regions) and motivated attention (occipitotemporal regions). Further research applying temporally highly resolving electrophysiological neuroimaging appears promising to investigate the interplay of developmental and pathological generalization and its neurocognitive basis.
Roesmann et al. (Thu,) studied this question.