To successfully predict important events, the representations in memory on which we rely need to be constantly updated and transformed to best reflect a complex and dynamic world. Here we employed a novel paradigm to investigate how memories of threat learning affect the flexible recombination across distinct but overlapping experiences, an ability referred to as relational memory. Participants (n=35) visited the lab to first encode neutral associations (A - B), which were reactivated and predictively associated with a new aversive or neutral element (B - C) on the following day, whilst pupil dilation was measured as an index of arousal. Then, again one day later, the accuracy of relational memory judgements (A - C?) was tested. Novel association to threat was found to impair relational memory. Unexpectedly, this effect was not moderated by arousal. We propose that compartmentalization of threat learning events could be a function of a healthy memory, preventing maladaptive ‘episodic overgeneralization’ of threat to previously encoded episodes.

The powerful ways future behavior and cognition can be affected by emotional events are typically studied either by means of Pavlovian conditioning or episodic memory paradigms. However, due to their incompatible methods, little is known about how Pavlovian conditioning and episodic memory relate to each other, or work in concert to affect behavior and other cognitive processes. Here, we present a novel stimulus set consisting of 60 negative, neutral, and positive image/sound combinations (Affective Multi-modal Stimuli; AMS) that elicited strong subjective and psychophysiological emotional responses, the latter of which transferred to arbitrary predictors already after a single paired presentation in a new paradigm (Episodic Conditioning). Real-life emotional events are likely to involve both Pavlovian conditioning and episodic memory processes. The episodic conditioning paradigm enables their simultaneous assessment, thereby providing an opportunity to gain more holistic insight into how these process interact in sickness and in health.

Whereas many studies have investigated generalization of learned threat across semantic dimensions, little attention has been given to the possibility that Pavlovian threat responses may spread beyond what is directly learned through episodic associations. Here, we increased the demand on episodic memory in a sensory preconditioning task to assess whether and how episodic memory contributes to threat generalization. Participants encoded forty unique episodes, each consisting of two neutral stimuli, after which one of each pair was newly associated with either an aversive or a neutral stimulus on the following day. Another day later, both stimuli of the original memories were found to trigger enhanced pupil dilation if one was indirectly linked with an aversive stimulus. This effect was independent of whether the declarative associations encoded on day one were accurately retained on the day of testing, and confined to trials on which the indirectly associated stimulus was consciously brought to mind, suggesting an important role for relational memory processes. There was no moderating effect of trait-anxiety. The present study demonstrates that the human defensive system is remarkably adept at quickly anticipating threat based on information acquired over separate events, and gives a first glimpse into how this ability is facilitated by episodic memory.

Contextual overgeneralization of emotional memory is believed to be a core aspect of affective disorders. Identifying methods to restrict emotional memory activation to its original encoding context is therefore of significant clinical interest. Preliminary evidence from rodent research points to a promising approach: reexposure to the context in which fear conditioning took place seems to reduce fear generalization to other contexts. However, it remains unknown which mechanisms underlie these effects, even though such fundamental knowledge is crucial for successful translation to interventions. Importantly, exposure to a context that resembles – but is not identical to – the encoding context may lead to diminished contextual dependency of memory by integration of additional contextual cues. Here, we therefore assessed in a large-scale study (N=180) whether reexposure to the encoding context enhances contextual dependency of episodic memory and whether exposure to a similar context impairs it. We also tested whether such effects are predicted by the strength of memory retrieval during context (re)exposure. The results showed that correct recognition depends on context and that such contextual dependency is lower for emotional than neutral memories, which replicates prior research. However, exposure to the encoding context or a similar context did not affect contextual dependency of emotional or neutral memory. Also, retrieval strength did not predict any effects. Thorough insight into factors underlying the effects of context (re)exposure on contextual dependency seems key to eventually attain a therapeutic memory recontextualization intervention