Guiding the Emotion in Emotional Memories: The Role of the Dorsomedial Prefrontal Cortex

The dmPFC Serves Control Functions

Circumscribed lesions to the dmPFC are rare (as noted by Bzdok et al., 2013), and so neuroimaging research has played a key role in clarifying the contributions of the dmPFC in cognitive and socio-affective domains. On cognitive tasks, the dmPFC plays a key role in performance monitoring, evaluating decisions and their outcomes (e.g., Alexander & Brown, 2014). Regarding socio-affective tasks, the dmPFC is particularly implicated in social cognition (Lieberman et al., 2019) and in the updating of social impressions, especially when information must be integrated (Ferrari et al., 2016). Although the dmPFC may not strongly represent the affective value of information (Lieberman et al., 2019), it is associated with the regulation of emotion (Silvers et al., 2015).

What these varied abilities may share in common is their reliance on control processes, including the selection, prioritization, and updating of information. Activation in the dmPFC consistently has been linked to the flexible control of behavior (Venkatraman & Huettel, 2012) across motor, memory, and decision-making tasks. The dmPFC may be particularly implicated when these control demands span cognitive and socio-affective domains. For instance, the dmPFC is important for the control of motor outputs when they occur within emotional contexts (Coombes et al., 2012) and for guiding decision making under uncertainty (Venkatraman & Huettel, 2012).

In this article, we describe evidence that the dmPFC plays an important role in guiding memory, particularly when the memories contain emotional content. Although its role in emotional memory appears to span explicit and implicit domains, we focus on the role of the dmPFC in episodic memory (Fig. 2). The prefrontal cortex (PFC) as a whole provides top-down constraints on memory processes and narrows what information is encoded, stored, and retrieved. Extensive research has revealed that the lateral PFC can provide top-down signals to filter out task-irrelevant information and to control the content held in memory by selecting stimuli on the basis of their sensory features or category membership (see review by D’Esposito & Postle, 2015). The dmPFC also appears to provide top-down signals to control the contents of memory, but we suggest that its functionality and connectivity make it ideally suited to guide the narrative framing—and the affect—of episodic memories.

OPEN IN VIEWER

Fig. 2.

Evidence for the role of the dorsomedial prefrontal cortex (dmPFC) in emotional episodic memory. During the encoding phase, the dmPFC is engaged in the regulation of emotion by reappraisal (activation shown in the three-dimensional brain; Morris et al., 2014) and by guided attention (activation shown in the insert; Allard & Kensinger, 2014). In the moments after encoding (storage phase), participants with stronger connectivity between a seed region in the amygdala (in green) and the dmPFC (in purple; connectivity denoted by arrow) show a greater positive memory bias compared with those with weaker amygdala-dmPFC connectivity (Kark & Kensinger, 2019). During the retrieval phase, the dmPFC guides retrieval of specific event detail and can reduce the emotional intensity of memories: The top left image shows activation during retrieval of event details (Ford et al., 2011). The top right image shows activation associated with reduced emotional intensity of memories in young adults (Holland & Kensinger, 2013), and the lower left image shows activation corresponding with reduced vividness of older adults’ negative (not positive) memories (Ford & Kensinger, 2017). The lower right image shows a dmPFC region whose activation, across two different data sets, was linked to reduced activity in the hippocampus for negative (not positive) memories (Ford & Kensinger, 2018).

The dmPFC Guides Episodic Memory

The role of the dmPFC in guiding memory begins during the encoding of an emotional experience. Specifically, the dmPFC plays a key role in constructing the meaning of an event and modulating the emotional intensity of the experience. The dmPFC is commonly recruited in the service of emotion regulation during event experiences (Silvers et al., 2015), serving to either upregulate or downregulate emotional responses (Frank et al., 2014). More generally, the dmPFC is thought to play a role in the appraisal of experiences (Ma et al., 2017). In other words, the dmPFC gives affective meaning to an event and can adjust that meaning on demand.

Although this role of the dmPFC is often considered within the specific context of affective neuroscience, the dmPFC may play a more general role in the creation of narrative meaning during events. In their STRing (structural and temporal representation binding) theory, Krueger et al. (2009) proposed that the medial PFC (mPFC) supports knowledge about social events by holding event simulators that represent information stored throughout the neocortex. They also proposed that more dorsal mPFC regions represent information tied to the goal states and actions of other people. But the dmPFC may play a role in event representation even for content that is not overtly social. For instance, Baetens and colleagues (2014) demonstrated that the dmPFC is important for the representation of conceptual knowledge—what they referred to as “high construal”—of nonsocial as well as social objects. Chen and colleagues (2017) provided further evidence that the dmPFC abstracts representations from complex events. They asked participants to view a movie and to verbally recall the movie while undergoing functional MRI. Their results demonstrated that the dmPFC was among the regions in which created representations were shared across people. In fact, within the dmPFC, the representational overlap across people was greater when they were recalling the movie than when they were originally viewing it, which underscores the dmPFC’s role in abstracting memory-driven narratives of experienced events.

The connectivity of the dmPFC makes it well suited to the creation of these abstracted representations. The anterior portion of the dmPFC anchors a subsystem of the default-mode network (Andrews-Hanna et al., 2010) with strong functional connectivity with the temporoparietal junction, a supramodal association area that is part of a ventral attention network, and with the lateral temporal cortex and temporal pole, regions that together are important for the representation of semantic knowledge. The more posterior portions participate in networks linked to attention and salience (Eickhoff et al., 2016). This placement of the dmPFC at the nexus of multiple different resting-state networks may make it well suited for executing the control and abstraction of memory representations. Indeed, the dmPFC is well connected with multiple association areas across the lateral frontal, temporal, and parietal lobes (Bzdok et al., 2013), which may enable it to effectively guide memory representations. Recent research has suggested that the dmPFC, like other PFC regions, shows white-matter connectivity with the amygdala (Goetschius et al., 2019), and thus is well suited to register and modulate the affect generated by semantic content during memory encoding (Kaneda et al., 2017). Although the dmPFC itself is not typically considered to be part of a reward network, interactions between the dmPFC and ventromedial PFC may provide a way for the ongoing processing in the dmPFC to be influenced by valuation signals (Kuzmanovic et al., 2018). The dmPFC can, in turn, modulate other cortical regions: For instance, transcranial magnetic stimulation of the dmPFC was shown to modulate the excitability of other prefrontal regions, as well as regions within the occipital lobe, during the experience of fear (Gonzalez-Escamilla et al., 2018). Thus, the dmPFC can tip the scales to influence the content that is encoded into memory.

The role of the dmPFC in the representation of events in memory continues as memories are stored. The dmPFC aids in the stabilization of visual percepts in memory over short periods of time (Schwiedrzik et al., 2018), and it continues to play a role in their stabilization over longer-term periods. For instance, when memories are cued over periods of slow-wave sleep—a phase of sleep linked to the consolidation of memories—this leads to strengthened connectivity between the dmPFC and occipital regions (Berkers et al., 2018).

Through its connectivity with the amygdala, the dmPFC also appears to modulate the emotional content of events that are stabilized in memory. For instance, Kark and Kensinger (2019) demonstrated that increases in the resting-state connectivity (i.e., correlations in activity in the absence of an external task to perform) between the dmPFC and the amygdala from before to after encoding correlated with participants’ ability to retain positive (not negative) content over a subsequent 24-hr delay. These results suggest that the way dmPFC interconnects with limbic regions in the moments during and after encoding can affect the likelihood that emotional aspects of an encoding episode are remembered, which explains some of the individual differences in whether people are more likely to focus on the positive or the negative. Together, the research suggests that the dmPFC can play a key role in modulating the way that events are encoded and stored in memory, influencing both direct and indirect routes that support emotional memory encoding and consolidation (Fig. 3).

OPEN IN VIEWER

Fig. 3.

Integrating the dorsomedial prefrontal cortex (dmPFC) into models of encoding and consolidation of emotional episodic memories. Existing models of encoding and consolidation of emotional memories (a) focus primarily on how emotion leads to direct modulation of memory circuitry via the amygdala (modulation model, shown in red; e.g., McGaugh, 2000) and on the additional role of indirect routes of memory modulation via effects of emotion on attention and elaboration processes implemented by lateral prefrontal regions, including lateral prefrontal cortex (lateral PFC; mediation model, shown in green; e.g., Talmi, 2013). These models focus on explaining emotional enhancements of memory, or why emotional experiences often are more likely to be remembered than neutral experiences. During the experience of an emotional event (b), the dmPFC receives inputs from the amygdala and from other prefrontal regions, including the ventromedial prefrontal cortex (vmPFC). These inputs enable the dmPFC to respond to goal states and to represent appraisals of emotional events. The dmPFC, in turn, can provide top-down signals that modulate the function of other regions, including those that influence emotional memory via direct (modulation model) or indirect (mediation model) routes (c). By doing so, the dmPFC is positioned to influence the event properties that are attended and elaborated (Path 1), the affective intensity experienced and the affective content encoded and stored (Path 2), and the way that other event details are integrated into a memory representation (Path 3). By extending encoding and consolidation of emotional memories to include these influences of the dmPFC, models may better account for not only whether emotional events will be remembered, but also the framings and affective tones with which those events will be remembered. MTL = medial temporal lobe.

At retrieval, the dmPFC participates in the integration of affective and cognitive components of memories, setting up networks and narratives that either emphasize or de-emphasize particular mnemonic content. We have already discussed how the dmPFC may be responsible for the abstraction of meaning from remembered information (Chen et al., 2017), and there is further evidence suggesting that it is involved in the retrieval of shared societal knowledge of past events (Gagnepain et al., 2020). In other words, the dmPFC may direct retrieval of a particular event on the basis of the broader societal context or pertinent information from related events.

One way the dmPFC may shape memories at the time of retrieval is through accessing and emphasizing event context. When young adults are asked to retrieve personal memories associated with familiar musical clips (e.g., Ford et al., 2011), the dmPFC is recruited to a greater extent when retrieved memories are specific to a particular place and time, rather than semantic (factual) or generalized (from repeated or extended events). In other words, the dmPFC is associated with the extent to which people retrieve the full event context. The role of the dmPFC in supporting specific memories is consistent with neuropsychological findings from Chapados and Petrides (2015), who revealed that patients with left dmPFC damage showed impairments in the retrieval of the relations between items and their contexts.

The dmPFC may also play an important role in controlling the emotionality of a memory in accordance with an individual’s goals at the time of retrieval. Recruitment of the dmPFC is greater when younger adults are asked to decrease the negative emotionality of negative events through reappraisal than when they are asked to maintain or increase the negative emotionality (Holland & Kensinger, 2013). This is consistent with the role of the dmPFC in emotion-regulation tasks (Frank et al., 2014) and with demonstrations that the dmPFC is engaged when participants intentionally attempt to control their memory at retrieval (e.g., Anderson et al., 2016).

There is also evidence that the dmPFC can respond to less explicit motivational shifts during memory retrieval. For instance, it has been proposed that older adults are more motivated to maintain positive affect during cognitive tasks (Mather & Carstensen, 2005), such that their retrieval-related goals are shifted away from retrieval of details and toward retrieval of positive elements. Two recent studies revealed age differences in dmPFC recruitment during retrieval that are consistent with this motivational shift. The first study revealed an age-by-valence interaction in the relation between dmPFC recruitment and subjective vividness ratings, driven by an age-related reversal in the relation between recruitment and vividness of negative images (Ford & Kensinger, 2017). Specifically, whereas recruitment of the dmPFC was associated with increased vividness ratings for negative events in young adults, recruitment of the same dmPFC region was associated with decreased vividness ratings for negative events in older adults. In contrast, age had no effect on the relation between dmPFC recruitment and vividness ratings for positive events. In the second study, a similar interaction during retrieval of memories associated with a highly emotional public event was found (Ford & Kensinger, 2019). In this study, age was associated with a reversal in the relation between dmPFC recruitment and ratings of memory negativity: In young adults, recruitment of the dmPFC was associated with higher negativity ratings, whereas in older adults, it was associated with lower negativity ratings. Age was not associated with the relation between dmPFC recruitment and positivity ratings.

In both studies, age was also associated with a shift in how the dmPFC worked together with the hippocampus to support retrieval of negative events (Ford & Kensinger, 2018). Whereas connectivity between these two regions was positive in young adults, which suggests that the regions may work together to support memory for negative events, connectivity was negative in older adults. In other words, on trials when older adults recruited the dmPFC more, they recruited the hippocampus to a lesser extent. There was no effect of age on connectivity during retrieval of positive events, which suggests a valence-specific effect rather than a compensatory role of the dmPFC following a reduced hippocampal response. The findings suggest that the dmPFC may be recruited to enhance or diminish particular emotional details, depending on the goals at the time of retrieval (Fig. 4).

OPEN IN VIEWER

Fig. 4.

Visualizing the role of the dorsomedial prefrontal cortex (dmPFC) in emotional memory retrieval. During retrieval of emotional memories (a), the primary role of the dmPFC may be to coordinate current retrieval context and goals to emphasize or de-emphasize particular elements of the memory trace. The illustrations in (b) provide an example of the same memory trace—in this case, a child’s bicycle accident—being retrieved under two different retrieval contexts. Because the memory is being retrieved for a different goal in each context, the framing of the memory differs, and different types of details are disproportionately remembered. In this figure, ability to enhance or diminish retrieval of particular content is represented by the gauges being dialed up to higher levels or down to lower levels.

Toward an Expanded Model of Emotional Memory

The dmPFC plays a key role in flexibly guiding the way that information is remembered. It plays this role from the moment that an event is experienced until the time it is brought to mind and elaborated upon at retrieval. Throughout the phases of episodic memory, the connections of the dmPFC appear to enable it to control the narrative—and the affective tone—of the memory. Activation of the dmPFC is not always revealed in analyses of successful emotional memory (e.g., Dahlgren et al., 2020), so it may not modulate whether emotional events are remembered. Yet the body of research reviewed here suggests that it has a role in controlling how emotional events are remembered, that is, in controlling the content and affective framing of memories. In some cases, an understanding of how events are remembered may be at least as important as an understanding of whether they are remembered. For instance, the way people frame an event in memory and the features they focus on may help to distinguish those who can adaptively learn from an event from those who ruminate counterproductively. We propose that by incorporating the dmPFC, models of emotional memory can capture these differences in how emotional events are remembered. Doing so may bring new insights into affective disorders and a new clarity in understanding individual and situational variability in the way that emotional experiences are remembered.

Recommended Reading

Dahlgren, K., Ferris, C., & Hamann, S. (2020). (See References). A review of the literature on the neuroscience of emotional memory, including a meta-analytic denotation of the regions that relate to successful emotional memory and a description of regions implicated at encoding and retrieval of episodic memory.

Gagnepain, P., Vallée, T., Heiden, S., Decorde, M., Gauvain, J. L., Laurent, A., Klein-Peschanski, C., Viader, F., Peschanski, D., & Eustache, F. (2020). (See References). A report on a functional MRI study that examined neural activity as participants recalled information related to World War II that was presented during a museum tour; results revealed the sensitivity of the dorsomedial prefrontal cortex to the influence of collective memory schemas.

Kensinger, E. A., & Ford, J. H. (2020). (See References). A review of how emotional experiences are remembered at the time of retrieval and how that process of remembering can influence the way an event is represented in mind and subsequently retrieved.

Lieberman, M. D., Straccia, M. A., Meyer, M. L., Du, M., & Tan, K. M. (2019). (See References). A review of methods and meta-analytic approaches used to clarify the roles of different subregions of the medial prefrontal cortex (focusing on Brodmann’s areas 9, 10, and 11), incorporating evidence from lesion studies, transcranial magnetic stimulation, and functional MRI research.

Talmi, D. (2013). (See References). Proposes a mediation model of emotional memory that complements a directmodulation model by explaining how emotional memory can be enhanced immediately after an event’s occurrence by cognitive factors such as attention and organization.

Supplemental Material