Abstract: The operational similarities between the retrieval extinction paradigm (reconsolidation-based) and the traditional extinction paradigm (extinction-based)—involving subtle procedural settings during the retrieval phase—may prevent memories from entering the reconsolidation stage, thus avoiding the long-lasting inhibitory effect on fear relapse. Studies have shown that the duration of retrieval exposure regulates the phase into which memories enter; this mechanism is related to the number of prediction errors elicited during retrieval. This study is intended to investigate the regulatory role of retrieval exposure duration from memory reconsolidation to extinction and to clarify its mechanism by quantifying prediction errors.
In this study, skin conductance responses and subjective expectancy values were collected and combined with a simplified Rescorla-Wagner learning model to fit the PE quantification curve. Following the experimental design idea of transitioning from the traditional extinction paradigm to the retrieval extinction paradigm, extinction training was divided into retrieval and extinction phases based on different numbers of retrieval trials to form four retrieval exposure durations: traditional extinction, single-trial retrieval extinction, double-trial retrieval extinction, and quadruple-trial retrieval extinction. The effects of conditioned fear memory elimination under these four conditions were compared.
The results revealed that in the spontaneous recovery test, both the single-trial and quadruple-trial retrieval extinction groups showed better inhibitory effects on spontaneous fear recovery, whereas the traditional and double-trial retrieval extinction groups exhibited evident spontaneous recovery effects. In the fear reinstatement test, only the single-trial retrieval extinction group showed a better inhibitory effect on fear reinstatement, while the other three groups all demonstrated fear reinstatement effects. Furthermore, the subjective expectancy results for the CS+ trials on the second day among the four groups indicated that there were no intergroup differences in the participants’ subjective expectations of whether CS+ was accompanied by a shock. However, in terms of quantified prediction errors, the single-trial retrieval extinction group showed the most significant change in prediction errors on the second day of the CS+ trials, with the overall lowest prediction error values.
This study determined that the number of retrieval trials played a crucial role in the successful transition from the traditional extinction paradigm to the retrieval extinction paradigm in human fear conditioning. This study emphasizes the importance of prediction errors during the retrieval phase as a potential determinant of memories undergoing reconsolidation or extinction. It is recommended that studies delve deeper into identifying specific neural markers for memory reconsolidation and extinction in humans, with the aim of designing more reliable clinical treatments for maladaptive emotional memories.

Key words: exposure duration, traditional extinction paradigm, retrieval extinction paradigm, prediction errors