Extinction-based exposure is effective but often incomplete, partly due to timing- and awareness-dependent factors that differentially recruit subcortical defensive and cortical cognitive systems. We report two parallel human fear-conditioning experiments that manipulated CS–US lead interval (short: 100–200 ms; long: 1000–4000 ms) and predictability (paired vs. unpaired), conducted in either a standard psychophysiology lab (Experiment 1; N=59 analysed) or a fully immersive VR context (Experiment 2; N=81 analysed). Eyeblink startle indexed acquisition and extinction; trial-wise expectancy was recorded in the lab, and interoceptive awareness (MAIA-2) was measured in VR. Across contexts, acquisition and extinction exhibited highly similar timing functions: CS+/CS– discrimination emerged robustly at longer intervals (≥1000 ms), whereas short intervals (≤200 ms) showed elevated startle when the CS–US relation was unpredictable and blunted startle when predictable. Extinction reduced startle at long but not short intervals in predictable conditions; unpredictability amplified early startle and attenuated later responses. Expectancy reports in the lab tracked this pattern closely. In VR, higher interoceptive awareness magnified startle at 100 ms under unpredictability and at long intervals under predictability. Pooled confirmatory mixed-effects analyses (Context × Phase × Lead Interval × Predictability × CS Type) revealed no evidence that the context (physical vs. VR) altered the shape of acquisition, extinction, or immediate retention curves. These results validate VR as a laboratory analogue of physical exposure for timing-dependent fear learning and suggest a functional boundary near ~1000 ms separating “imminent-threat” from “anticipated-threat” processing. Implications for optimizing exposure timing and transdiagnostic personalization via interoception are discussed.
Bergsnev et al. (Wed,) studied this question.