Memories for emotional experiences are typically well remembered. This is thought to be a result of immediate encoding effects, together with strengthening of subsequent consolidation processes. The selective prioritization of emotional experience can be adaptive insofar as it preserves memories that are personally significant. But persistent, intrusive emotional memories can be maladaptive. Hence, there is motivation to understand how emotional memories might be selectively attenuated so as to diminish the psychological impact of negative experiences. Here, we describe new research on the mechanisms involved in selecting and prioritizing long-term memories for emotional experiences in humans. Dr. Kroes will describe data showing that episodic memories can provide an organizational scaffolding to fear learning and expression. Dr. de Voogd will provide evidence that goal-directed attention with a high cognitive demand alters large-scale brain circuits involved in fear learning, thereby reducing recovery of emotional experiences. Dr. Talmi will provide data showing that the memory advantage of experiences that signal reward depends on competition that occur during recall. Dr. Dunsmoor will show that event boundaries that separate competing experiences of threat and safety adaptively segments emotional experiences into separate long-term memory traces. Collectively, the empirical data presented in this symposium make claims about the conditions by which long-term memories for emotional experiences are selectively strengthened as well as attenuated. They will be discussed critically by Dr. Smith, taking the adversarial philosophical perspective that emotions are not natural kinds, and therefore do not have a unique influence on memory.
A critical challenge for animals is to predict potential danger in novel situations. When an animal learns that a specific stimulus predicts an aversive outcome it can generalize fear responses to novel stimuli that resemble it. Such generalization of fear responses based on sensory feature overlap is advantageous but also limited and inflexible. A more flexible strategy would be if animals could also use prior knowledge about the structure of the world to infer potential danger in novel situations. Recent research from the field of episodic memory indicates that animals can infer relationships between stimuli across distinct episodic experiences to form a “cognitive map” of the structure of the world. One form of inferential reasoning is transitive inference whereby animals can deduce that if stimulus A is related to stimulus B and stimulus B to stimulus C then stimulus A must be related to stimulus D. It is still unclear, however, whether animals can use transitive inference based on prior episodic experiences to predict danger in novel situation. Here I will present a study showing that humans can learn relationships between stimuli across distinct episodic experiences using transitive inference and that they use these prior episodic relational memories to generalize threat responses to novel stimuli. Hence, the relational organization of prior episodic memories provide humans with a flexible strategy to infer danger in novel situations.
When people encounter items that they believe will help them gain reward, they later remember them better than those that do not. While it is adaptive to preferentially remember experiences that will be useful later, it is unknown how the competition for memory resources is implemented in time, through the processes of encoding, consolidation, and retrieval. In two experiments we promised participants £1 for remembering some pictures, but only 10 pence for remembering others. Their ability to describe the pictures was tested after one minute and after 24 hours. Memory at immediate test showed effects of list composition, suggesting local competition at encoding and/or retrieval. These results are consistent with our recently-proposed emotional Context Maintenance and Retrieval model [Talmi, Lohnas, & Daw, Psychological Review, in press], supporting it as a general account of motivated memory. In contrast, relative to this baseline, more valuable memories were not preferentially retained following delay, suggesting no detectable role of competition for consolidation.
Anxiety-related disorders are the most prevalent among all psychiatric disorders and have a high comorbidity with other disorders. Despite this high prevalence (lifetime prevalence is ~30%), treatment often remains unsuccessful. Here I will propose a neural framework for understanding the cognitive regulation of anxiety which can serve as a heuristic framework for guiding hypotheses to improve treatment of anxiety-related disorders. I will discuss evidence showing that cognitively demanding tasks induce a reorganization between large-scale networks. For example, goal-directed eye-movements deactivate the amygdala, a key structure of the salience network, and when embedded during safety learning they attenuate later fear recovery (de Voogd et al., 2018). This downregulation engages a ventromedial prefrontal pathway known to be involved in cognitive regulation of emotion suggesting that cognitively demanding tasks can be used to regulate anxiety. I will present new data, using a meta-analytic approach, showing that a cognitively demanding task outside the context of emotion engages a similar amygdala deactivation as cognitive reappraisal does, a widely-used emotion regulation technique. I also will present data showing this amygdala deactivation is linearly related to the load of the cognitive demand. We are currently investigating whether systematically increasing cognitive demand during safety learning indeed leads to a stronger reorganization between large-scale networks and a reduction in anxiety-related symptoms. The notion that increasing cognitive demand may benefit treatment of anxiety-related disorders provides a mechanistic account of already existing techniques, such as Tetris, Eye movement desensitization and reprocessing (EMDR), and cognitive reappraisal.
The inability to override fear associations is at the core of many anxiety and stress disorders. Popular treatments, such as exposure therapy, typically involve forming new memories meant to compete with and replace fear associations. This approach, which is built on the principles of Pavlovian fear extinction, can be effective. But symptoms often return, perhaps because extinction memories are context-specific and fear suppressed in the extinction context often re-emerges in a different context. This effect is known as contextual renewal. Here, we harness multivoxel pattern analysis during functional MRI to identify context reactivation in the human brain during the retrieval of extinction memories. This technique allows us to to discover whether contextual renewal (i.e., extinction-retrieval failure) occurs due to insufficient reactivation of the extinction context at a later time. We “tagged” the mental context during extinction learning by presenting task-irrelevant pictures (from separate categories) between trials, and then used fMRI pattern classifiers in occipitotemporal cortex to quantify the (category-specific) reinstatement of the extinction context the next day. We compared results in healthy adults to people with PTSD, a disorder characterized by deficits in fear inhibition. This population exhibited weaker neural reactivation of the extinction context during threat ambiguity, which may in part explain deficits in extinction retention in anxiety and stress disorders.