The computational nature of memory modification

Human Pavlovian fear conditioning conforms to probabilistic learning

Learning to predict threat from environmental cues is a fundamental skill in changing environments. This aversive learning process is exemplified by Pavlovian threat conditioning. Despite a plethora of studies on the neural mechanisms supporting the formation of associations between neutral and aversive events, our computational understanding of this process is fragmented. Importantly, different computational models give rise to different and partly opposing predictions for the trial-by-trial dynamics of learning, for example expressed in the activity of the autonomic nervous system (ANS). Here, we investigate human ANS responses to conditioned stimuli during Pavlovian fear conditioning. To obtain precise, trial-by-trial, single-subject estimates of ANS responses, we build on a statistical framework for psychophysiological modelling. We then consider previously proposed non-probabilistic models, a simple probabilistic model, and non-learning models, as well as different observation functions to link learning models with ANS activity. Across three experiments, and both for skin conductance (SCR) and pupil size responses (PSR), a probabilistic learning model best explains ANS responses. Notably, SCR and PSR reflect different quantities of the same model: SCR track a mixture of expected outcome and uncertainty, while PSR track expected outcome alone. In summary, by combining psychophysiological modelling with computational learning theory, we provide systematic evidence that the formation and maintenance of Pavlovian threat predictions in humans may rely on probabilistic inference and includes estimation of uncertainty. This could inform theories of neural implementation of aversive learning.

Nitrous oxide may interfere with the reconsolidation of drinking memories in hazardous drinkers in a prediction-error-dependent manner

Weakening drinking-related reward memories by blocking their reconsolidation is a potential novel strategy for treating alcohol use disorders. However, few viable pharmacological options exist for reconsolidation interference in humans. We therefore examined whether the NMDA receptor antagonising gas, Nitrous Oxide (N₂O) could reduce drinking by preventing the post-retrieval restabilisation of alcohol memories in a group of hazardous drinkers. Critically, we focussed on whether prediction error (PE; a key determinant of reconsolidation) was experienced at retrieval. Sixty hazardous drinkers were randomised to one of three groups that retrieved alcohol memories either with negative PE (Retrieval + PE), no PE (Retrieval no PE) or non-alcohol memory retrieval with PE (No-retrieval +PE). All participants then inhaled 50% N₂O for 30 min. The primary outcome was change in beer consumption and alcohol cue-driven urge to drink from the week preceding manipulation (baseline) to the week following manipulation (test). The manipulation did not affect drinking following the intended retrieval+/- PE conditions However, a manipulation check, using a measure of subjective surprise, revealed that the group-level manipulation did not achieve the intended differences in PE at retrieval. Assessment of outcomes according to whether alcohol-relevant PE was actually experienced at retrieval, showed N₂O produced reductions in drinking in a retrieval and PE-dependent fashion. These preliminary findings highlight the importance of directly testing assumptions about memory reactivation procedures in reconsolidation research and suggest that N₂O should be further investigated as a potential reconsolidation-blocking agent.

Reminder duration determines threat memory modification in humans

Memory reminders can return a memory into an unstable state such that it will decay unless actively restabilized into long-term memory through reconsolidation. Exposure to a memory reminder, however, does not always lead to destabilization. The 'trace dominance' principle posits that the extent of exposure to memory reminders governs memory susceptibility to disruption. Here, we provide a first systematic investigation of reminder duration effects on threat memory modification in humans. Reminder duration was parametrically varied across 155 participants in a three-day protocol. We found that short reminders (1 s and 4 s) made the memory prone to interference from post-retrieval extinction, suggesting that the memory had been updated. In contrast, no reminder or long reminders (30 s and 3 min) made the memory resistant to such interference, and robustly return. Reminder duration therefore influences memory stability and may be a critical determinant of therapeutic efficacy.

Multiple memory systems, multiple time points: how science can inform treatment to control the expression of unwanted emotional memories

Memories that have strong emotions associated with them are particularly resilient to forgetting. This is not necessarily problematic, however some aspects of memory can be. In particular, the involuntary expression of those memories, e.g. intrusive memories after trauma, are core to certain psychological disorders. Since the beginning of this century, research using animal models shows that it is possible to change the underlying memory, for example by interfering with its consolidation or reconsolidation. While the idea of targeting maladaptive memories is promising for the treatment of stress and anxiety disorders, a direct application of the procedures used in non-human animals to humans in clinical settings is not straightforward. In translational research, more attention needs to be paid to specifying what aspect of memory (i) can be modified and (ii) should be modified. This requires a clear conceptualization of what aspect of memory is being targeted, and how different memory expressions may map onto clinical symptoms. Furthermore, memory processes are dynamic, so procedural details concerning timing are crucial when implementing a treatment and when assessing its effectiveness. To target emotional memory in its full complexity, including its malleability, science cannot rely on a single method, species or paradigm. Rather, a constructive dialogue is needed between multiple levels of research, all the way 'from mice to mental health'.This article is part of a discussion meeting issue 'Of mice and mental health: facilitating dialogue between basic and clinical neuroscientists'.

Event segmentation protects emotional memories from competing experiences encoded close in time

Fear memories are characterized by their permanence and a fierce resistance to unlearning by new experiences. We considered whether this durability involves a process of memory segmentation that separates competing experiences. To address this question, we used an emotional learning task designed to measure recognition memory for category exemplars encoded during competing experiences of fear-conditioning and extinction. Here we show that people recognized more fear-conditioned exemplars encoded during conditioning than conceptually related exemplars encoded immediately after a perceptual event boundary separating conditioning from extinction. Selective episodic memory depended on a period of consolidation, an explicit break between competing experiences, and was unrelated to within-session arousal or the explicit realization of a transition from conditioning to extinction. Collectively, these findings suggest that event boundaries guide selective consolidation to prioritize emotional information in memory—at the expense of related but conflicting information experienced shortly thereafter. We put forward a model whereby event boundaries bifurcate related memory traces for incompatible experiences. This stands in contrast to a mechanism that integrates related experiences for adaptive generalization¹²³, and reveals a potentially distinct organization by which competing memories are adaptively segmented to select and protect nascent fear memories from immediate sources of interference.

A preregistered, direct replication attempt of the retrieval-extinction effect in cued fear conditioning in rats

In 2009, Monfils and colleagues proposed a behavioral procedure that was said to result in a permanent attenuation of a previously established fear memory, thereby precluding a possible return of fear after extinction (Monfils, Cowansage, Klann, & LeDoux, 2009). By presenting a single retrieval trial one hour before standard extinction training, they found an enduring reduction of fear. The retrieval-extinction procedure holds great clinical potential, particularly for anxiety patients, but the findings are not undisputed, and several conceptual replications have failed to reproduce the effect. These failures have largely been attributed to small procedural differences. This preregistered study is the first endeavor to exactly replicate three key experiments of the original report by Monfils et al. (2009), thereby gauging the robustness of their seminal findings. Despite adhering to the original procedures as closely as possible, we did not find any evidence for reduced return of fear with the retrieval-extinction procedure relative to regular extinction training, as assessed through spontaneous recovery, reinstatement and renewal. Behavior of animals in the control condition (extinction only) was comparable to that in the original studies and provided an adequate baseline to reveal differences with the retrieval-extinction condition. Our null findings indicate that the effect sizes in the original paper may have been inflated and question the legitimacy of previously proposed moderators of the retrieval-extinction effect. We argue that direct experimental evaluation of purported moderators of the retrieval-extinction effect will be key to shed more light on its nature and prerequisites.

On the transition from reconsolidation to extinction of contextual fear memories

Retrieval of an associative memory can lead to different phenomena. Brief reexposure sessions tend to trigger reconsolidation, whereas more extended ones trigger extinction. In appetitive and fear cued Pavlovian memories, an intermediate "null point" period has been observed where neither process seems to be engaged. Here we investigated whether this phenomenon extends to contextual fear memory. Adult rats were subjected to a contextual fear conditioning paradigm, reexposed to the context 2 d later for 3, 5, 10, 20, or 30 min, with immediate injections of MK-801 or saline following reexposure, and tested on the following day. We observed a significant effect of MK-801 with the 3- and 30-min sessions, impairing reconsolidation and extinction, respectively. However, it did not have significant effects with 5-, 10-, or 20-min sessions, even though freezing decreased from reexposure to test. Further analyses indicated that this is not likely to be due to a variable transition point at the population level. In conclusion, the results show that in contextual fear memories there is a genuine "null point" between the parameters that induce reconsolidation and extinction, as defined by the effects of MK-801, although NMDA receptor-independent decreases in freezing can still occur in these conditions.