Big Thanks to my collaborators and colleagues at the Department of Neurology @MPI-CBS.
Concept
Imagine an individual visiting a neighbour who hears a kennel gate open and sees a dog. Later, while inside the house, the dog becomes aggressive and injures the person. Cognitive psychology teaches us that the place where the aversive incident happened is likely to be well remembered in our long-term episodic memory, because it can warn us about the future face-to-face encounter with the dangerous dog. Likewise, seeing the dog again mobilises our defensive system to cope with the anticipated harm, which is elegantly described by first-hand Pavlovian threat conditioning — a fundamental mechanisms of associative learning for recognising threats. But what about the kennel gate outside the house? For example, the sound it makes when opening?
This project has shown that cues, such as the specific sound of the gate in our hypothetical scenario, may become also memorable. These inconspicuous cues can be prioritised in long-term episodic memory when they are retrospectively inferred to be of significance in predicting future threats. To support such inference, individuals can employ their pre-existing knowledge of environmental structure (e.g., a kennel gate is typically associated with a dog). In other words, recognizing stable environmental patterns may help retrospectively connect events encoded in separate learning episodes and contexts to aid inference about danger. Importantly, this mechanism allows us to identify signals that indicate the risk of harm before the threat is actually present.
Abstract
Linking memories of temporally separated neutral and threatening experiences can support later recognition of early warning signals. Yet it remains unclear what cognitive processes might allow the seemingly inconsequential neutral events – encoded in safe contexts and thus prone to forgetting – to nevertheless be preserved in memory well enough to inform future threat predictions. Here, we tested whether Pavlovian threat conditioning retroactively strengthens long-term episodic memory for previously encoded neutral events by leveraging prior knowledge of their predictive relations. Young adults first learned to anticipate the order of semantic categories in two deterministic sequences (A->B->C and D->E->F) through repeated presentation of images depicting category exemplars. They then incidentally encoded trial-unique exemplars from each category during sequence viewing. In a next phase, exemplars from the third-position categories were repeated such that one (C) predicted mild electric shocks, whereas the other (F) was never reinforced, creating a category-level threat association through conditioning. After 24 hours, recognition memory for exemplars of categories encoded before conditioning depended on sequence and position: when a sequence ended with a shock-predictive category, memory for its starting category (A) was prioritized over the following category (B) – an effect absent in the control categories (D and E). These results suggest that aversive learning can retroactively reshape episodic memories of neutral events when they are scaffolded by prior sequence knowledge, bridging temporal gaps and thereby potentially promoting early threat anticipation
Acknowledgment
This work was supported by the Max Planck Society. BMB was supported by the Max Planck Society, the International Max Planck Research School “NeuroCom”, and the University of Leipzig.
License
This work is licensed under CC BY 4.0.