Engagement and Disengagement: From the Basic Science of Emotion Regulation to an Anxiety Spectrum

ABSTRACT

Emotion regulation strategies vary in depth of processing. For instance, reappraisal requires greater engagement than distraction. This affects short‐term and long‐term response to stimuli. In this review, I describe how the “engagement‐disengagement dimension” improves understanding of emotion regulation in normative contexts and in internalizing psychopathology. Part 1 reviews work from my laboratory and others, suggesting that relatively disengaged emotion regulation strategies (e.g., distraction), may have short‐term benefits (e.g., faster implementation), but may come with long‐term costs (e.g., increased processing of stimuli at subsequent encounter). Therefore, depending on the desired outcome, the adaptive selection of an emotion regulation strategy will be determined by extent of emotional engagement–disengagement. In Part 2, I describe how individuals with more comorbid internalizing psychopathology (e.g., multiple anxiety and depressive diagnoses) are characterized by disengagement from negative stimuli as measured by the late positive potential (LPP). In addition, I introduce a brain profile I have termed, HARM‐A (heightened “alarm” and reduced motivated attention), which is characterized by a combination of heightened “alarm” (i.e., increased amygdala) and emotional disengagement (i.e., blunted LPPs) in response to negative stimuli. HARM‐A prospectively predicts worse outcomes over 2 years in a mixed internalizing sample. As such, chronic disengagement from negative stimuli appears to contribute to more comorbid and more severe internalizing psychopathology. Overall, emotional disengagement can be beneficial in the short term but may be poorly suited to emotional coping in the longer term.

Impact statement

Emotion regulation strategies vary in depth of processing/emotional engagement (e.g., reappraisal > distraction). This dimension provides a useful framework for explaining the relative advantages of various strategies in the short term versus long term. Furthermore, emotional disengagement, particularly when combined with heightened “alarm” (e.g., amygdala)—that is, the HARM‐A (heightened “alarm” and reduced motivated attention) brain profile—corresponds to worse outcomes in internalizing psychopathology.

Every day, we are confronted with situations in which we attempt to downregulate negative emotion. Sometimes, we do this by engaging directly with the situation. For example, we might question our assumptions about a situation, ultimately revising its meaning and therefore the emotions it elicits. At other times, we reduce emotional response by disengaging from the situation. That is, we may direct our attention elsewhere or busy ourselves with another task that consumes our mental resources, leaving little left over to process and respond to a negative event. Both engaged and disengaged strategies can be effective at reducing negative emotion, and most people use both at various times and in response to different types of situations (Aldao 2013; Sheppes et al. 2014). For instance, people tend to use engaged strategies for less intense emotional situations and disengaged strategies for more intense emotional situations (Shafir et al. 2016). Critically, however, engaged versus disengaged strategies may have radically different effects in the longer term (Kross and Ayduk 2008; Sheppes and Meiran 2007, 2008). In what follows, I will describe work suggesting that emotional engagement is necessary for lasting change in stimulus processing. Moreover, I will show how emotional disengagement may play a role in worsening anxiety and depression, potentially because it limits deeper processing necessary to adapt to emotional challenges (Wilson and Gilbert 2008) and learn from new information (Craske et al. 2014; Foa and Kozak 1986). To begin, I will briefly review prominent models of emotion regulation.

1. Models of Emotion Regulation

As the emotion regulation literature began to balloon in the late 1990s, James Gross introduced a seminal model that helped organize the field. Gross' (1998b) process model emphasized for the first time the importance of the time course of emotional response in the selection and deployment of emotion regulation strategies. The process model identified the various timepoints along the unfolding of emotional response at which emotion regulation could be accomplished (Gross 1998b). Strategies that could be employed prior to or during the emergence of an emotional response were referred to as “antecedent‐focused,” whereas strategies that were employed on fully developed emotional responses were “response‐focused.” The earliest stages of emotional response included “situation selection” and “attentional deployment,” and involved the use of disengaged emotion regulation strategies, such as looking away from emotionally arousing stimuli. A later stage was “cognitive change,” which involved more engaged emotion regulation strategies, such as reappraisal (i.e., meaning change). Even later was the “response modulation” stage, involving disengaged strategies such as suppression (i.e., inhibiting a fully developed emotional response).

Gross' (1998a) model was fundamental in highlighting the importance of when emotion regulation is accomplished, with implications for the types of strategies that could be used at each timepoint. Expanding on this model, subsequent work has suggested that disengaged and engaged emotion regulation strategies may not be bound to particular stages of emotional response (Sheppes and Meiran 2007). Furthermore, although antecedent‐focused strategies are generally thought of as superior to response‐focused strategies (e.g., Gross 1998a, 2002), there appear to be important distinctions between the various antecedent‐focused strategies in terms of how emotion regulation is accomplished.

One model of emotion regulation that has focused on how emotion regulation takes place is the automatic emotion regulation model (Gyurak, Gross, and Etkin 2011; Mauss, Bunge, and Gross 2007). This model introduced for the first time the notion that emotion regulation can occur implicitly. Implicit strategies modify emotional response without this being the explicit goal and include affect labeling (Lieberman et al. 2007) and emotional go/no‐go and Stroop tasks (Buhle, Wager, and Smith 2010), in which emotion is regulated as a byproduct of nonaffective task goals. Explicit strategies are those for which emotion regulation is the goal and include reappraisal (Gross 1998a), distraction (Sheppes and Meiran 2007), and other strategies (Mauss, Bunge, and Gross 2007). Though this initial model made a binary distinction between implicit and explicit strategies, a subsequent multidimensional model distinguished not only between emotion regulation goals (implicit vs. explicit) but also process (automatic vs. controlled; Braunstein, Gross, and Ochsner 2017).

Another model of how emotion regulation is accomplished is Sheppes' and colleagues' (2014) model of emotion regulation choice. This model introduced for the first time the distinction between engaged versus disengaged emotion regulation, with more engaged strategies involving greater depth of processing. Sheppes and colleagues (2014) used this framework to explain why people choose certain emotion regulation strategies over others. For instance, they suggested that people prefer to use disengaged strategies for highly intense emotional situations because these strategies can be applied early on, minimizing emotional arousal. On the other hand, people chose to use an engaged strategy (reappraisal) rather than a disengaged strategy (distraction) when they were given the goal to improve their longer‐term response to stimuli (Sheppes et al. 2014). Therefore, people prefer to avoid deeper processing when stimuli are highly intense, but have a sense that deeper processing will facilitate longer‐term change in emotional response (Sheppes et al. 2011).

In what follows, I will build on this work to argue that the engagement–disengagement dimension is a critical determinant of the effectiveness of different emotion regulation strategies. Part 1 describes basic science work from my laboratory and others that has examined the short‐ and long‐term effectiveness of engaged and disengaged emotion regulation strategies. Most empirical work on emotion regulation has focused on its immediate effects, though consideration of longer‐term effects is critical for the real world, where similar situations are often encountered more than once. In Part 2, I will describe how disengagement may play a role in internalizing psychopathology such as anxiety and depression. Though there is considerable evidence of heightened reactivity to negative stimuli in internalizing psychopathology, I suggest that patients with the most severe psychopathology and/or multiple psychiatric disorders (i.e., more comorbid) are often characterized by blunted electrocortical response to negative stimuli. Moreover, work from my laboratory suggests that emotional disengagement is especially pernicious when combined with heightened “alarm” (i.e., amygdala BOLD) to negative stimuli—i.e., a profile termed, HARM‐A (heightened “alarm” and reduced motivated attention). Though the mechanism behind blunted electrocortical responding in these individuals is still unknown, possibilities include willful disengagement (e.g., avoidance) or more habitual or automatic disengagement (including via an evolutionarily conserved response; Nesse 2000). A common thread across Part 1 and Part 2 is that emotional disengagement can be beneficial in the short term but may be poorly suited to emotional coping in the longer term.

The studies I describe have primarily used ERPs and fMRI BOLD response, elicited by pictures of emotional scenes from standardized databases like the International Affective Picture System (IAPS; Lang, Bradley, and Cuthbert 2008). Despite the benefits of neurobiological measures for assessing emotional response, (e.g., minimizing demand characteristics and the need for introspection), emotion encompasses physiological, subjective, and behavioral levels (Gross 1999). Therefore, the studies I describe below should be taken for what they are—as providing insight into a limited subset of emotional response that is not all‐inclusive but complementary to work performed at other levels.

2. Using the Late Positive Potential (LPP) to Measure Emotion Regulation

One ERP that has been used widely to measure the effects of emotion regulation, including in my own work, is the late positive potential (LPP). The LPP is centroparietally maximal, begins approximately 300 ms after stimulus onset, and is larger for emotional compared to neutral stimuli (Cuthbert et al. 2000; Hajcak, MacNamara, and Olvet 2010; MacNamara, Joyner, and Klawohn 2022). In addition to being sensitive to the overall salience of stimuli, the LPP is responsive to more fine‐grained distinctions in stimuli meaning. For example, the LPP is larger to stimuli that are personally relevant, such as pictures of loved ones (Grasso and Simons 2011; Vico et al. 2010) or one's own name (Tacikowski and Nowicka 2010). It can also be modulated by changes in the extrinsic value of stimuli, such as task demands. For instance, the LPP is larger for target compared to standard stimuli (Ferrari et al. 2008; Schupp et al. 2007; Weinberg et al. 2012). The LPP can also be modulated by participants' attempts to increase (Baur et al. 2015; Bernat et al. 2011; Wilson and MacNamara 2020) or decrease (Hajcak and Nieuwenhuis 2006; Moser et al. 2006; Parvaz et al. 2012; Thiruchselvam et al. 2011) emotional salience. Therefore, the LPP is sensitive to both intrinsic and extrinsic effects and provides a useful measure of the motivational salience of stimuli (Bradley 2009; Hajcak and Foti 2020; Lang, Bradley, and Cuthbert 1997). In what follows, I will describe work using the LPP to provide insight into negative emotion downregulation and internalizing psychopathology. In spite of my laboratory's interest in positive emotion upregulation (Barley et al. 2021; Cheng, Peters, and MacNamara 2023; Wilson and MacNamara 2020, 2024) and its potential relevance to anxiety and depression (Craske et al. 2019; Jackson, Wilson, and MacNamara 2024), there is still only a relatively small literature in this area. Therefore, I have not attempted to review it here.

3. Part 1—An Engagement–Disengagement Perspective on Emotion Regulation

3.1. The Need to Consider Short‐Term and Long‐Term Effectiveness of Emotion Regulation

Beginning in the 1990s, the first empirical work on emotion regulation aimed to show that people could regulate emotion in the laboratory and that effects were evident using a variety of neurobiological measures. As the field has matured, researchers have taken a more nuanced approach (MacNamara, Joyner, and Klawohn 2023). For example, researchers have begun to use more varied types of stimuli and have examined moderators of emotion regulation success. More broadly, there has also been a growing interest in bridging understanding of emotion regulation from the laboratory to the real world (MacNamara, Joyner, and Klawohn 2023). This includes understanding longer‐term effects.

Because engaged emotion regulation strategies require deeper processing of stimuli than disengaged strategies, they have different short‐term and long‐term effects. In what follows, “short‐term” refers to the immediate effects of emotion regulation as it is undertaken. I use “long‐term” to refer to paradigms that have gone beyond this timeframe in any sense. These long‐term studies have typically examined effects 30 min or so after emotion regulation has been initially implemented (though some of our recent work has examined effects up to 24 h later; MacNamara, Joyner, and Klawohn 2022). As such, I use “long‐term” in a relatively limited sense, and many questions remain concerning longer timescales (e.g., weeks, months, or years). In addition, because there have been many reviews that have summarized the benefits of the short‐term effects of emotion regulation (e.g., Dickey, Politte‐Corn, and Kujawa 2021; Hajcak, MacNamara, and Olvet 2010; MacNamara, Joyner, and Klawohn 2022; McRae and Gross 2020; Webb, Miles, and Sheeran 2012), I will highlight here only the main points from this body of work and dedicate the majority of Part 1 to longer‐term effects—work that has not previously been summarized succinctly.

The relative strengths of engaged versus disengaged emotion regulation strategies is presented in Figure 1, along with examples of these strategies. Some strategies that may readily be thought of as engaged or disengaged forms of emotion regulation are not included because they do not generally reduce internal experience of negative emotion. These include rumination (e.g., Kollárik et al. 2020) and expressive suppression, in which individuals inhibit outward display of emotion such as facial expressions (Gross 1998a, 2002). ¹ Moreover, acceptance, which is listed in Figure 1, has been poorly defined in the literature and has not been incorporated into the most influential emotion regulation models (Wojnarowska, Kobylinska, and Lewczuk 2020). Here, I use the term acceptance to refer to an engaged emotion regulation strategy that involves cultivating openness to emotional experience (quelling resistance), while simultaneously knowing that emotions do not need to control thoughts and behaviors (Hayes et al. 1999; Keogh et al. 2005). Unlike related strategies like mindfulness, which overlaps in terms of present‐moment focus but focuses more on monitoring (Lindsay and Creswell 2019), acceptance involves relatively deep processing of emotion.

FIGURE 1.

Relative advantages of engaged versus disengaged emotion regulation and example strategies.

3.2. Short‐Term Effectiveness of Engaged Versus Disengaged Emotion Regulation Strategies

The most widely studied emotion regulation strategy is cognitive reappraisal—an engaged strategy that involves changing the meaning of stimuli. Reappraisal was first shown to reduce the LPP and subjective ratings of negative pictures in the early 2000s (e.g., Hajcak and Nieuwenhuis 2006; Moser et al. 2006). However, because participants are usually asked to generate their own alternative interpretations for stimuli, it was unclear whether meaning change was sufficient to bring about change in picture processing. An alternative possibility was that the cognitive effort involved in generating alternative explanations for pictures might be responsible for the reduction of picture processing—for example, via distraction. Therefore, as a graduate student in Greg Hajcak's lab, I set out to verify meaning change as the mechanism underlying reappraisal, by (a) testing whether reappraisal could downregulation negative emotion even when cognitive load was lessened and (b) testing whether reappraisal could also upregulate negative emotion.

To this end, we created a series of negative and neutral descriptions of pictures, so that participants would not need to generate alternative meanings of picture content themselves. Then we asked participants to listen to a randomly selected description of each negative or neutral picture before viewing it. Pictures preceded by neutral descriptions elicited smaller LPPs and less arousing and unpleasant ratings than pictures that had been described negatively, whereas negatively described pictures elicited larger LPPs and higher ratings (Foti and Hajcak 2008; MacNamara, Foti, and Hajcak 2009). Therefore, meaning change alone, rather than the cognitive effort involved in generating alternative explanations, could modulate the LPP.

A less studied engaged emotion regulation strategy is acceptance, which involves a present‐moment, nonjudgmental focus. Integration of acceptance into the engagement–disengagement framework is complicated not only by varied operationalizations of acceptance, but also by the use of different types of stimuli, such as autobiographical memories rather than standardized pictures. Some evidence suggests that acceptance may reduce self‐reported negative emotion (Goldin, Moodie, and Gross 2019; Kross et al. 2009) and neurobiological correlates of negative emotion processing such as amygdala (Goldin, Moodie, and Gross 2019) and subgenual anterior cingulate (Kross et al. 2009) BOLD. On the other hand, acceptance may also increase picture processing, at least initially. That is, when participants were asked to pay attention to emotions arising from pictures “in an accepting manner” (Uusberg et al. 2016, 97), the LPP was increased compared to distraction or simply attending to picture details (Uusberg et al. 2016). Therefore, more work is needed to bolster evidence of acceptance's effectiveness, particularly using a consistent definition of acceptance that goes beyond a present‐moment focus to include other aspects of acceptance (Hayes et al. 2006).

While engaged emotion regulation strategies like reappraisal may be well‐suited to situations in which a person has the time and energy to dedicate to extended processing of an emotional situation, people also need to be able to disengage from emotionally arousing stimuli rapidly (e.g., refocusing attention away from a roadside ad for a horror movie when someone pulls out in front of us). In these types of situations, emotion regulation strategies that require less in‐depth processing and take place without the explicit decision to regulate emotion should be more effective.

One way of reducing attention to task‐irrelevant emotional stimuli without an explicit emotion regulation goal is via nonaffective, cognitively demanding tasks (Braunstein, Gross, and Ochsner 2017). In graduate school, I tested this using a working memory task. In this work, (MacNamara et al. 2012; MacNamara, Ferri, and Hajcak 2011) and studies that my laboratory has conducted since (Barley et al. 2021; Cheng, Jackson, and MacNamara 2022; MacNamara et al. 2019), participants are asked to memorize 6 or 2 letters presented at the beginning of a trial. Next, they view a task‐irrelevant stimulus, such as a negative or neutral picture or a colored shape that has been paired with shock, prior to being asked to type out the letters presented at the beginning of the trial. Results have consistently shown that working memory load reduces the picture‐elicited LPP. Therefore, demanding cognitive tasks that engage prefrontal brain regions (Smith et al. 1998) and consume attentional resources provide a disengaged yet effective means of emotional downregulation, at least in the short term. ²

Looking away from emotional stimuli provides another means of rapidly shifting attention away from emotional stimuli. For instance, when participants were instructed to look at nonemotional or emotional parts of negative pictures, such as a rock rather than the face of a deceased child (Hajcak et al. 2013), the LPP was reduced (Dunning and Hajcak 2009; Hajcak et al. 2013), as were steady‐state visual evoked potentials (ssVEPs; Hajcak et al. 2013). Moreover, reductions in picture processing were observed within a few hundred ms of the onset of cues directing participants to look away from arousing picture regions (Hajcak et al. 2013). As such, spatial attention appears to be a highly effective and rapid modulator of emotional salience. In a convergent line of work, we modulated spatial attention toward arousing pictures more subtly, by encouraging participants to disengage visual attention from briefly presented stimuli, without varying eye gaze. To do so, we presented pairs of pictures in different locations on a computer screen. When we made one pair of pictures task‐irrelevant by asking participants to make decisions about the other pair of pictures, there was no emotional modulation of the LPP by task‐irrelevant pictures (MacNamara and Hajcak 2009). Together, these studies indicate that both overt and covert manipulation of spatial attention can lead to rapid reductions in the processing of emotional stimuli.

Comparatively, engaged strategies such as reappraisal may take some time to implement. For instance, work by Langeslag and van Strien (2018) showed that reappraisal of snake and spider pictures reduced the LPP but not the early posterior negativity (EPN), which is evident ~200–300 ms poststimulus onset. Similarly, multiple studies that have directly compared distraction and reappraisal have shown that distraction reduces emotional processing faster than reappraisal (Adamczyk et al. 2023; Paul et al. 2013; Schönfelder et al. 2014; Thiruchselvam et al. 2011; Yuan et al. 2015).

Because reappraisal (but not distraction or other disengaged emotion regulation strategies) involves engaging with the arousing aspects of emotional stimuli at least early on (Strauss, Ossenfort, and Whearty 2016), it risks sustaining or even elevating emotional response if an alternative meaning for the emotional situation is not successfully generated or maintained (Bernat et al. 2011; Cao, Li, and Niznikiewicz 2020; Langeslag and Surti 2017). In line with this, a number of studies have failed to find evidence of reappraisal on the LPP (e.g., Gan et al. 2015; Imburgio and MacNamara 2019; Kudinova et al. 2016; Mallorquí‐Bagué et al. 2020; Pedersen and Larson 2016). Critically, failed reappraisals appear to be more likely for highly intense emotional stimuli (Shafir et al. 2015). For instance, reappraisal is ineffective at reducing the LPP for pictures of stigmatized (e.g., homeless people; drug addicts) but not nonstigmatized (e.g., man holding a gun, injured people) individuals (Krendl, Zucker, and Kensinger 2017).

Participants seem to know that reappraisal is less effective for highly intense stimuli. That is, when given the option to use distraction or reappraisal for highly arousing pictures, participants chose distraction more often (Dorman Ilan et al. 2020), but reappraisal is chosen more often for low‐intensity pictures (Sheppes et al. 2011). In line with this, larger LPPs prior to emotion regulation are predictive of a preference for distraction over reappraisal (Shafir et al. 2016; see also Adamczyk et al. 2024). Another disengaged strategy—affect labeling—is also effective at reducing emotion elicited by high‐intensity stimuli, though it may increase distress in low‐intensity situations (Levy‐Gigi and Shamay‐Tsoory 2022).

In sum, both engaged and disengaged emotion regulation strategies can be effective at reducing response to negative stimuli in the short term. However, disengaged strategies may be preferable if speed of implementation is important or if stimuli are highly intense. For less intense stimuli, however, more engaged emotion regulation strategies may be more effective and—as reviewed below—could have preferable long‐term outcomes.

3.3. Long‐Term Effectiveness of Engaged Versus Disengaged Emotion Regulation Strategies

Laboratory‐based research has typically examined the immediate, short‐term effects of emotion regulation. In the real world, however, long‐term reductions in negative emotion are often desirable, as are effects that generalize to different, but similar stimuli. For instance, if someone had been working to think differently about why no one talked to them at last week's party, it would be beneficial if some of their effort to reframe this situation carried over to affect similar situations in the future (e.g., next week's party). As such, understanding the longer‐term and not just the shorter‐term effects of various emotion regulation strategies is essential for their adaptive and flexible implementation. Below, I will describe how engaged and disengaged emotion regulation strategies appear to differ in their longer‐term effects, even though some of these strategies are comparably effective at reducing emotional response in the moment.

While in graduate school, I performed an initial investigation of how reappraisal—an engaged emotion regulation strategy—might affect subsequent encounters with the same stimuli. As in our earlier work (MacNamara, Foti, and Hajcak 2009), participants listened to a negative or neutral description before viewing each picture. Approximately 30 min later, participants viewed the same pictures again, this time without descriptions. Results showed that reappraisal's effects were persistent across time. That is, pictures that had been previously reappraised (from negative to neutral via picture descriptions) elicited smaller LPPs when encountered later during passive picture viewing (MacNamara, Ochsner, and Hajcak 2011). Other work has directly compared the durability of engaged and disengaged negative emotion downregulation strategies. That is, Thiruchselvam et al. (2011) asked participants to use reappraisal or distraction to reduce emotion elicited by negative pictures or to simply view these pictures (as well as neutral pictures as a control). Thirty minutes later, participants were exposed to the same pictures again. Although in this study, pictures that had previously been reappraised did not elicit smaller LPPs (vs. those that had been simply viewed), those with a distract history elicited larger LPPs than those with a reappraise history (Thiruchselvam et al. 2011; see also Jiang, Chen, and Guo 2020; Paul, Kathmann, and Riesel 2016). Together, these studies suggest that meaning change leads to lasting reductions in the processing of emotional content, whereas temporarily diverting attentional resources can lead to short‐term reductions in the emotional salience of stimuli, but may preserve this salience for the next encounter (see also Hermann, Kress, and Stark 2017).

Acceptance—another engaged emotion regulation strategy—also seems to exert lasting reductions in the emotionality of stimuli, even when it is not effective at initially reducing the LPP. That is, Usberg and colleagues found that viewing negative stimuli from an accepting perspective led to initial increases in the LPP. Nonetheless, acceptance removed affective amplification of the LPP after additional trials, whereas distraction and attentive viewing did not. When pictures were presented again without instructions to regulate, emotional modulation of the LPP was still only evident for distraction and the control condition, and not acceptance. Therefore, while engagement with emotional stimuli via acceptance appeared to amplify picture salience early on, it eventually led to reductions in the emotional salience of stimuli that persisted across time, when compared to a more disengaged emotion regulation strategy.

A related idea is whether emotion regulation not only persists across time to affect encounters with the same stimuli but might also generalize to similar but previously unseen stimuli. In our first attempt at answering this question, we examined how reappraisal might generalize to new negative pictures that were similar to those that had been reappraised (Bautista et al. 2022). Participants used reappraisal to downregulate the emotional salience of to either pictures of snakes or guns (between‐groups design). Following this, they performed a passive picture‐viewing task in which they viewed new pictures of guns and snakes, with instructions to simply view and rate these pictures. In the first task, reappraisal did not reduce the LPP (Figure 2A). In the second task, however, new pictures in the same category as those that had been previously reappraised (e.g., snakes) elicited smaller LPPs compared with new pictures in the category that had been previously viewed (e.g., guns)—Figure 2B. Therefore, at least at the level of the LPP (generalization was not observed for ratings), this work suggested that reappraisal might generalize to similar but previously unencountered negative stimuli. Moreover, as reappraisal did not modulate the LPP in the first task, its effects seemed to grow stronger over time. ³

FIGURE 2.

Reappraisal generalizes to previously unseen pictures when shown later during passive picture viewing. Grand‐averaged waveforms and headmaps depicting condition differences, shown separately for (A) a reappraisal task, in which participants were asked to reduce their response to negative pictures and (B) a generalization task, in which participants passively viewed similar pictures. The shaded area indicates the time window in which the LPP was scored. (Adapted from Bautista et al. 2022).

Based on the results described above, engaged emotion regulation may lead to intrinsic or bottom‐up changes in stimulus significance that persist until the next time stimuli are encountered. By contrast, reductions in the emotional salience of stimuli that result from temporary diversion of cognitive resources might be less effective at modulating subsequent encounters with stimuli (Kross and Ayduk 2008) and might even lead to a rebound effect/increased processing (Thiruchselvam et al. 2011). Increased processing of stimuli following disengaged emotion regulation might occur if these stimuli are perceived as novel because they were minimally processed the last time they were encountered. Along these lines, Sheppes and Meiran found that participants could recall fewer details of sad films following distraction, whereas reappraisal did not interfere with participants' ability to recall the films (Sheppes and Meiran 2007, 2008).

To test the idea that disengaged emotion regulation might lead to pictures being processed as if they had never been seen before, we measured ERPs to negative and neutral pictures that had been presented under high or low working memory load and were subsequently shown in a surprise old–new task (Jackson and MacNamara 2025). Because working memory load reduces the LPP by diverting processing resources away from emotional stimuli (e.g., MacNamara, Ferri, and Hajcak 2011), we reasoned that pictures that had been viewed under high compared to low working memory load might elicit larger LPPs at a subsequent encounter, similar to pictures that were being seen for the first time (new pictures). In the first task, we found that high versus low working memory load reduced the LPP elicited by pictures (Figure 3A), as in our previous work (Barley et al. 2021; Cheng, Jackson, and MacNamara 2022; MacNamara et al. 2012, 2019; MacNamara, Ferri, and Hajcak 2011). In the second task, we found that negative pictures that had been presented on high‐load trials elicited LPPs that did not differ in magnitude from that elicited by new negative pictures and were larger than pictures that had been presented on low‐load trials, which elicited smaller LPPs than new negative pictures (Figure 3B). In addition, negative pictures that had previously been presented on high‐load trials as well as new negative pictures were rated as more arousing than pictures presented on low‐load trials. Recognition accuracy was also worse for pictures that had been presented on high‐load compared to low‐load trials. Therefore, less engaged processing of negative pictures during the first task (i.e., on high‐load trials) appeared to preserve the emotional salience of these pictures for a subsequent encounter—that is, as if these pictures were never seen before (Jackson and MacNamara 2025). As such, the gains associated with disengaged emotion regulation strategies may be short‐lived, and might even backfire in the future.

FIGURE 3.

Working memory load reduces the LPP but leads to larger LPPs when negative pictures are encountered later. Grand‐averaged waveforms, shown separately for (A) neutral and negative pictures interspersed with a working memory task and (B) a recognition task, in which participants viewed the same pictures that had previously been presented under high or low working memory load, as well as new pictures. The shaded area indicates the time window in which negative pictures previously presented under high load (> low load) elicited larger LPPs in the recognition task. (Adapted from Jackson and MacNamara 2025).

In sum, engaged strategies such as reappraisal may be preferable when individuals aim to create lasting change in stimulus processing, or when it is beneficial for effects to generalize to similar but previously unencountered emotional situations. Moreover, disengaged strategies may lead to worse memory for stimuli when encountered later, as well as relatively enhanced responding to these stimuli that is comparable to the processing stimuli receive at first encounter.

3.4. Summary

The results of the work reviewed above (summarized in Figure 1) suggest that disengaged ways of reducing negative emotion, such as when participants temporarily allocate attention to a demanding cognitive task (MacNamara, Ferri, and Hajcak 2011) may lead to rapid reductions in emotional response (Adamczyk et al. 2023; Thiruchselvam et al. 2011) and may be more effective and preferred when stimuli are highly intense (Sheppes et al. 2011). Nonetheless, over the longer term, disengaged strategies appear to lead to a rebound in stimulus processing, as evidenced in larger LPPs and higher subjective ratings when pictures are encountered a second time (Thiruchselvam et al. 2011; Jackson and MacNamara 2025). Indeed, it would seem that as far as the LPP is concerned, our brains process pictures that are presented under high working memory load as if we had never seen them before (Jackson and MacNamara 2025). That is, using work or thinking about distracting or cognitively demanding topics to avoid or reduce emotion may interfere with acclimation to these stimuli. Instead, engaged strategies, such as reappraisal, appears to lead to lasting reductions in the picture‐elicited LPP (MacNamara, Ochsner, and Hajcak 2011) and might generalize to the processing of similar, but previously unseen pictures encountered later (Bautista et al. 2022). Considering this, selecting the best emotion regulation strategy might consist of matching level of emotional engagement with desired short‐term versus long‐term effects.

Chronic use of emotion regulation strategies that work well in the moment but preserve or even increase the emotional salience of stimuli over time may lead to shifts in emotion processing that could underlie psychopathology (Craske et al. 2014; Foa and Kozak 1986). Although a great deal of research has focused on heightened neurobiological response to negative stimuli in the internalizing disorders (Ball et al. 2012; Etkin and Wager 2007; Hattingh et al. 2013; Hayes, Hayes, and Mikedis 2012; Killgore et al. 2013; Kinney, Burkhouse, and Klumpp 2019; Kujawa et al. 2015; Sartory et al. 2013), emotional disengagement—that is, blunted electrocortical response to negative stimuli—has also been reported (Foti et al. 2010; MacNamara et al. 2013; MacNamara, Kotov, and Hajcak 2016; Weinberg and Hajcak 2011), most often among patients with more severe and/or more comorbid psychopathology. In what follows, I will describe my work using EEG and fMRI measures of emotional dis/engagement to prospectively predict increased symptoms in the internalizing disorders. This work builds on what we have learned from our basic science studies—that emotional disengagement interferes with acclimation to negative stimuli and may therefore be maladaptive over the longer term.

4. Part 2—The Role of Emotional Disengagement in Internalizing Psychopathology

4.1. The Clinical Significance of Comorbidity Load

A reliable predictor of clinical outcomes in the internalizing disorders is comorbidity load—that is, number of psychiatric disorders. Greater comorbidity load is predictive of: higher risk of suicide, longer duration of illness, increased physical illness, greater role impairment, lower remission rates, and higher rates of relapse (Boden, Fergusson, and Horwood 2007; Bruce et al. 2001, 2005; Geller et al. 2003; Hunt, Issakidis, and Andrews 2002; Klein Hofmeijer‐Sevink et al. 2012; McAleavey, Castonguay, and Goldfried 2014; Rudd, Dahm, and Rajab 1993; Souĕtre et al. 1994; Wittchen et al. 1994). Despite its clinical significance, the mechanisms underlying comorbidity load are unknown, which may limit improvements in clinical care.

4.2. An Anxiety Spectrum

An implication of the current psychiatric diagnostic system (American Psychological Association 2013) is that patients with higher comorbidity load have multiple, independent, disease processes. However, this might misrepresent the pathophysiology underlying comorbidity, preventing identification of clinically meaningful, shared pathophysiological processes. In line with this, epidemiological and genetic work has suggested that anxiety‐disordered patients might fall along a spectrum. ⁴ At one end of this spectrum are patients with relatively focal fear, low comorbidity, as well as low levels of negative affectivity and dysphoria. At the other end are patients who are characterized by more diffuse anxiety, high comorbidity load, high negative affectivity, and high levels of dysphoria (Kendler et al. 2003; Krueger 1999; Watson 2005). Identification of neural mechanisms that characterize this spectrum might explain disease burden, beyond categorical diagnoses, and that is unaccounted for by the current diagnostic system.

A number of studies have supported the notion of an anxiety spectrum using the LPP. Increased LPPs (negative > neutral) have been found in relatively focal, noncomorbid anxiety disorders (Kinney, Burkhouse, and Klumpp 2019; Kujawa et al. 2015). However, as psychopathology becomes more diffuse and is characterized by higher levels of negative affectivity, the LPP to negative stimuli appears to become blunted (e.g., Foti et al. 2010; MacNamara et al. 2013; MacNamara, Kotov, and Hajcak 2016; Weinberg and Hajcak 2011). For instance, in one study, we measured the LPP to negative and neutral pictures in women with “pure” generalized anxiety disorder (GAD); those with “pure” major depressive disorder (MDD); patients comorbid for both, and psychiatrically healthy controls. Of note, this study was conducted before the advent of the National Institute of Mental Health's Research Domain Criteria (RDoC; Insel et al. 2010), which has transformed mental health research with its call for a more a transdiagnostic and dimensional approach to psychopathology. Therefore, at the time, recruiting patients into single disorder groups/free from unspecified comorbidities was the norm (and very challenging, given high rates of comorbidity). Despite its focus on recruiting distinct diagnostic groups (rather than a continuum), results from this study revealed associations with the LPP that appear to fit with the notion of an anxiety spectrum, characterized by blunted responding at the far end. That is, when performing a picture‐viewing task, individuals with “pure” GAD (i.e., no comorbid disorders) showed heightened LPPs to negative stimuli, whereas individuals with “pure” depression, as well as those with comorbid depression and GAD had blunted LPPs (MacNamara, Kotov, and Hajcak 2016). Though GAD is typically thought of as a relatively diffuse form of anxiety (because patients worry about multiple domains of their lives), individuals with pure GAD are a somewhat unique group. These individuals are free from comorbidities and might therefore be thought of as falling somewhere in the low end or middle of the anxiety spectrum. Depression, on the other hand, is associated with high levels of negative affectivity and might therefore be situated farther along the spectrum, with comorbid GAD/MDD also appearing toward the farther end of the anxiety spectrum. As such, our results can be thought of as supporting the idea of an anxiety spectrum. More specifically, they suggest that as individuals become more comorbid/more impaired, they are characterized by blunted response to negative stimuli (we have also observed similar results using negative imagery, Bauer and MacNamara 2021; and in youth, Kujawa et al. 2015).

Independent of the ERP literature, fMRI work has suggested that heightened neurobiological response to negative might be associated with greater liability to internalizing psychopathology. For example, more negative temperament in monkeys and humans (which likely underlies several categorical disorders) is associated with increased response to negative stimuli in brain regions associated with threat‐processing, including the amygdala, anterior insula, anterior hippocampus, bed nucleus of the stria terminalis, midcingulate cortex, and periaqueductal gray (Avery, Clauss, and Blackford 2016; Cavanagh and Shackman 2015; Fox and Shackman 2019; Shackman et al. 2016). While all of these regions are important, the amygdala and the anterior insula, which are part of the “salience network” (Seeley et al. 2007), have been most widely studied in relation to negative emotionality and internalizing psychopathology (e.g., MacNamara, DiGangi, and Phan 2016; MacNamara and Phan 2016; Phan et al. 2004).

While I was a postdoc with Luan Phan, I showed that clinician‐rated anxiety symptoms in individuals with mixed internalizing psychopathology were associated with greater activation in the bilateral insulae for angry faces (MacNamara et al. 2017). This work suggested that insula activation to negative stimuli might vary transdiagnostically across an anxiety spectrum. By contrast, primary diagnosis did not predict activation in the insula. Along these lines, several other studies have also found evidence of heightened activation in the insula and amygdala among patients with anxiety and/or depression who are more severe and more comorbid—that is, those that are farther along the anxiety spectrum (Gaffrey, Barch, and Luby 2016; Heitmann et al. 2016; Phan et al. 2006). The amygdala, in particular, has emerged as one of the most well‐known findings in case–control studies, with patients with anxiety and depression showing increased amygdala activation in response to negative stimuli compared to individuals without diagnoses (Etkin and Wager 2007; Feldker et al. 2016; Groenewold et al. 2012; Hamilton et al. 2012; Liu et al. 2022; McTeague et al. 2020). Amygdala activity is also correlated with the extent of anxiety severity within and across diagnoses (Brühl et al. 2011; Lin, Miltner, and Straube 2021; Peluso et al. 2009; Phan et al. 2006; Shah et al. 2009; Stein et al. 2007). Therefore, as internalizing psychopathology worsens, amygdala response to negative stimuli appears to increase across the anxiety spectrum.

As such, both the ERP and fMRI literatures appear to support the notion of an anxiety spectrum. Nonetheless, at least at first glance, the direction of findings in these two bodies of work appears somewhat contradictory. That is, the far end of this spectrum (i.e., patients who are more severe, more comorbid) appears to be characterized—in separate literatures—by both a blunted LPP and increased engagement of the anterior insula and amygdala. To make sense of these findings, it is worth considering the neural generators underlying the LPP, and its functional significance. The LPP receives contributions from a variety of brain regions, including visual cortex, medial and lateral prefrontal cortex, as well as the amygdala and insula (Liu et al. 2012; MacNamara et al. 2018). That is, the LPP does not arise from a single brain region. In line with this, the LPP, which reflects the elaborated processing of stimulus salience, is determined by both “bottom‐up” and “top‐down” factors (Hajcak, MacNamara, and Olvet 2010; Hajcak and Foti 2020). Given these characteristics of the LPP, results from the ERP and fMRI literatures might not be contradictory, but instead might tell us something about the pathophysiology of the most severely affected patients that is not evident using one measure alone. Below, I propose a neurobiological profile that considers these two literatures as well as recent work from my laboratory to explain, in part, why some patients acquire additional diagnoses and worse functioning over time, while others do not.

4.3. Heightened “Alarm” and Reduced Motivated Attention (HARM‐A)

Compared with the amygdala, which is implicated in early and automatic response to threatening and negative stimuli (Liddell et al. 2005; Öhman 2005; Phillips et al. 2003), the LPP may measure more integrated, prefrontally mediated appraisals of stimulus salience (Hajcak, MacNamara, and Olvet 2010; Liu et al. 2012; MacNamara et al. 2018; MacNamara, Joyner, and Klawohn 2022). Therefore, in the context of heightened amygdala activation, reduced LPPs could signal attenuated higher‐order processing that might otherwise help modify threat appraisals. In line with this notion, prominent models of anxiety suggest that to reduce fear, sustained engagement with threatening stimuli is necessary (Foa and Kozak 1986). By continuing to sit with and process the meaning of threatening stimuli throughout the duration of stimulus presentation (and potentially beyond), it may be possible for individuals to assimilate new information that is at odds with the “fear structure” underlying an anxiety disorder (Craske et al. 2014; Foa and Kozak 1986). Without this—that is, for individuals with greater disengagement, as evident in blunted LPPs, the threat value of stimuli may be maintained, and might worsen over time. Emotional disengagement might be especially harmful for individuals who are also characterized by greater “bottom up” assessment of threat salience, as indicated by heightened “alarm” (e.g., amygdala). Hereafter, I refer to this neural profile as Heightened “Alarm” and Reduced Motivated Attention, HARM‐A.

Figure 4 presents a model that depicts how HARM‐A might serve as predictor of increased internalizing psychopathology over time. This conceptual model is bidirectional; that is, we propose that some anxiety‐disordered patients have higher levels of HARM‐A, precipitating symptom escalation, but that reciprocal associations also emerge, such that repeated bouts of psychopathology serve to increase HARM‐A (Post et al. 1982), which may in turn promote additional, distress‐based diagnoses and increased dysphoria over time (i.e., a “downward spiral”). In line with this hypothesis, anxiety‐disordered patients with higher comorbidity load are more vulnerable to resurgence of psychopathology, compared with less comorbid, remitted patients (Bruce et al. 2005; Geller et al. 2003). In addition, patients with remitted internalizing psychopathology are characterized by abnormal amygdala neurocircuitry, supporting the notion that prior episodes of psychopathology may leave a lasting effect on the brain for some patients (Bhaumik et al. 2017). Moreover, informed by these findings, and by animal work on sensitization (Radley et al. 2015; Sousa and Almeida 2012), our model posits that HARM‐A can increase with each bout of psychopathology and persists during remission, leaving these individuals more vulnerable to symptom resurgence. On the other hand, patients who show at least partial normalization of HARM‐A neurocircuit activation over time might be protected against future episodes of psychopathology.

FIGURE 4.

HARM‐A (heightened “alarm” and reduced motivated attention). Conceptual model showing a reciprocal relationship between HARM‐A and internalizing psychopathology over time. Though future work may reveal that some paths are stronger than others, this remains to be determined and as such is not depicted in the model.

To test the idea that HARM‐A might characterize more comorbid patients and predict symptom worsening over time, we recruited 110 individuals with internalizing symptoms. Participants were required to have at least a “focal fear” diagnosis (fear of public speaking or simple phobia), in order to ensure a similar baseline level of fear across the sample but were permitted to vary in extent of additional comorbid disorders (“Patients”). Patients were divided into a low comorbid group, comprised of individuals with the focal fear diagnosis and 0–1 comorbidities (n = 44) and a high comorbid group, with 2+ comorbidities (n = 34). We also recruited individuals who were free from current and past psychopathology (“Controls”; n = 32). Participants performed a passive picture viewing task during an EEG recording, and separately, in an MRI scanner. During the task, participants viewed negative and neutral pictures, before rating each picture's valence.

We first examined whether comorbidity load was associated with heightened amygdala response and smaller LPPs cross‐sectionally. Next, we sought to determine whether individuals with both of these neural signals would show greater increases in dysphoria at follow‐up, approximately 2 years later, as this would be in line with the notion that heightened HARM‐A puts individuals at risk of worsening trajectories over time. Remarkably, our results (Bauer et al. 2023) aligned with existing fMRI and LPP work that had been completed in separate samples. That is, patients in the high comorbid group were characterized by increased fMRI BOLD in the amygdala for negative > neutral pictures (Figure 5). Moreover, these same individuals showed blunted LPPs to negative pictures (Figure 6).

FIGURE 5.

Patients with multiple internalizing disorders have greater amygdala activation to negative stimuli. (A) Mean negative > neutral blood oxygen level‐dependent (BOLD) response (extracted β weights and arbitrary units) from the anatomically defined left amygdala, shown separately for control, low comorbid, and high comorbid groups. Error bars indicate the standard error of the mean. *p < 0.05. (B) One‐sample voxelwise t‐maps overlaid on a canonical brain rendering (Montreal Neurological Institute coronal) showing negative > neutral amygdala activation for control, low comorbid, and high comorbid groups, unmasked and depicted at p < 0.001. (Reprinted with permission from Bauer et al. 2023).

FIGURE 6.

Patients with multiple internalizing disorders have more blunted LPPs to negative stimuli. Grand‐averaged negative minus neutral difference waveforms at a centroparietal pooling, shown separately for control, low comorbid, and high comorbid groups, as well as headmaps depicting condition differences for each group. The shaded area indicates the time window in which significant group differences were observed. (Adapted from Bauer et al. 2023).

Next, we looked at the predictive utility of HARM‐A. As shown in Figure 7, individuals with a combination of heightened amygdala and a blunted LPP at baseline showed greater increases in dysphoria approximately 2 years later (controlling for baseline dysphoria; Bauer et al. 2023). Therefore, a combination of heightened “alarm” to negative stimuli (amygdala BOLD) as well as reduced motivated, elaborative processing of these same stimuli (LPP)—that is, HARM‐A—appeared to track a mechanism by which psychopathology is maintained and worsens over time. In addition, we found that a combination of heightened HARM‐A and greater past‐month stress (using the Perceived Stress Scale; Cohen, Kamarck, and Mermelstein 1983) was associated with the highest levels of internalizing symptoms. Therefore, one way in which HARM‐A might contribute to psychopathology is by strengthening the link between stressful events and negative affect.

FIGURE 7.

HARM‐A (heightened “alarm” and reduced motivated attention) predicts increased dysphoria 2 years later. Left: One‐sample voxelwise t‐maps overlaid on a canonical brain rendering (Montreal Neurological Institute coronal), showing baseline negative > neutral amygdala activity; middle: Baseline grand‐averaged waveforms at the centroparietal pooling where the LPP was scored, shown separately for negative and neutral pictures (shaded area indicates the time window contributing to HARM‐A); and right: Headmaps depicting the baseline voltage difference (negative > neutral) from 3500 to 7000 ms following picture onset—shown separately for participants (A) low in Year 2 minus baseline dysphoria and (B) high in Year 2 minus baseline dysphoria (based on a median split, for illustrative purposes only). (Reprinted with permission from Bauer et al. 2023).

In our ongoing work, we are continuing to test the utility of the HARM‐A brain profile in a new sample, while also using ecological momentary assessment (EMA) to obtain a more fine‐grained assessment of day‐to‐day stress. If HARM‐A holds up as a predictor of symptom worsening, this would help unite previously disparate fMRI and LPP literatures. Moreover, HARM‐A might provide a path forward for improved classification and more refined treatment targets for patients with the greatest clinical burden.

4.4. Summary

The work I have reviewed above describes how individuals with more severe and comorbid internalizing psychopathology are characterized by both heightened amygdala and insula responsivity to negative stimuli, as well as blunted LPPs to these same stimuli. Furthermore, patients with both of these neural signatures (i.e., heightened HARM‐A) appear to be at the greatest risk of increased dysphoria over time and may be more sensitive to stressful life events. Overall, this work unites prior literatures and suggests the need for further investigation of the HARM‐A brain profile as a transdiagnostic risk factor for worsening psychopathology in the internalizing disorders.

5. Overall Summary and Future Directions

5.1. An Engagement–Disengagement Perspective on Emotion Regulation

Temporary diversion of attentional resources away from emotional stimuli, while effective in reducing emotional response and stimulus processing in the moment, does not lead to lasting changes in stimulus processing. Moreover, disengaged emotion regulation strategies may even interfere with naturally occurring reductions in stimulus salience that would otherwise take place (Thiruchselvam et al. 2011; Jackson and MacNamara 2025). But on the other hand, disengaged strategies are faster and likely easier to implement than reappraisal, at least for certain kinds of stimuli (Adamczyk et al. 2023; Schönfelder et al. 2014; see also Langeslag and Surti 2017; Thiruchselvam et al. 2011). I have argued here that the engagement–disengagement dimension provides a useful means of characterizing emotion regulation strategies, particularly when it comes to weighing the trade‐offs between short‐term effort versus longer‐term gains.

Going forward, continued work is needed to increase understanding of the long‐term effects of emotion regulation. For instance, while several studies have compared the immediate effects of engaged and disengaged emotion regulation strategies (e.g., reappraisal vs. distraction), more research directly comparing the long‐term effects of such techniques is needed. In addition, examination of effects over longer time periods (e.g., several days or weeks) is virtually nonexistent and would help increase translation of laboratory findings to everyday life.

Another future line of work might involve testing the effects of engaged versus disengaged emotion regulation strategies using more varied stimulus modalities, such as films or auditory stimuli. In addition, imagined stimuli may be particularly relevant to increasing understanding of psychopathological processes, such as worry and rumination. Along these lines, we recently showed that the LPP can be elicited when participants are asked to imagine themselves in a negative or neutral scene described using standardized audio recordings, without any visual percept (MacNamara 2018). Moreover, we found that individuals higher in intolerance of uncertainty—a risk factor for anxiety and depression (e.g., Norr et al. 2013; Shihata, McEvoy, and Mullan 2017)—showed blunted LPPs to imagined negative scenes (MacNamara 2018), in line with our HARM‐A work. In another study, we found that people can upregulate their response to positive imagined scenes using savoring, but that this is compromised in depression (Jackson, Wilson, and MacNamara 2024). As such, emotion regulation using imagined stimuli might reveal effects of psychopathology more readily than emotional pictures (e.g., Kinney, Burkhouse, and Klumpp 2019; Kivity and Huppert 2018). Overall, using imagined stimuli and stimuli of different modalities should lead to a more complete and rigorous understanding of the relevance of the engagement–disengagement dimension to emotion regulation (see also Thiruchselvam, Hajcak, and Gross 2012).

5.2. Mechanisms Underlying Emotional Disengagement in Internalizing Psychopathology

In our work in clinical samples, described above, emotional disengagement as measured using the LPP was associated with worse outcomes over the longer term, particularly if combined with greater “alarm” (amygdala activity). Though an emerging body of work now suggests that more severe internalizing psychopathology is associated with smaller LPPs to negative stimuli (e.g., Bauer et al. 2023; Foti et al. 2010; MacNamara et al. 2013; MacNamara, Kotov, and Hajcak 2016; Weinberg and Hajcak 2011), the mechanism behind this effect is still largely unclear. Going forward, it will be important to understand this mechanism and to integrate findings with the broader literature on internalizing psychopathology. For instance, one of the most well‐established means of emotional disengagement in the anxiety disorders is avoidance (Craske, Hermans, and Vervliet 2018).

Emotional disengagement at the far end of the anxiety spectrum might be attributable to avoidance—either explicit, strategic avoidance, or that which has been practiced so extensively that it becomes implicit (Braunstein, Gross, and Ochsner 2017). On the other hand, more automatic processes could alternatively explain emotional disengagement among more severe and comorbid patients. For instance, emotional disengagement could reflect “bottom‐up,” neuroplastic changes that reduce motivated attention to salient stimuli in a nonwillful way. Evolutionary theory suggests that emotional disengagement may be an adaptive response to sustained adversity (Nesse 2000). That is, when an organism is under extreme duress, disengagement from the external environment may increase the odds of survival. Take, for example, a deer that has been unable to find food in the woods for several days because of heavy and unremitting snowfall. At some point, standing still would become the best way for the deer to survive (e.g., the snow might melt or food might fall near the deer). In this way, severe symptoms of depression or internalizing psychopathology and accompanying disengagement from stimuli that would normally elicit action tendencies (Frijda 1987) might emerge in response to chronic adversity because it is adaptive. These adaptations could involve alterations in emotion generation as well as emotion regulation—processes that are likely best thought of as parallel rather than sequential (Gross and Barrett 2011; see also Dixon et al. 2017; Yih et al. 2019).

With increased insight into the factors contributing to emotional disengagement in the internalizing disorders, it may be possible to identify new targets or adjunctive approaches that could help patients with internalizing psychopathology who fail to benefit from current treatments. Although exposure therapy is effective for many, a substantial proportion (e.g., ~50%; Hofmann et al. 2012) of patients are unhelped or have residual symptoms. Emotional disengagement might explain some of these treatment failures. For instance, we (Stange et al. 2017) and others (Barch et al. 2020; Bunford et al. 2017; Paul et al. 2022) have found that patients with smaller LPPs to negative pictures at baseline benefit less from psychotherapy for anxiety and depression (also see Kinney, Burkhouse, and Klumpp 2019). Therefore, patients characterized by emotional disengagement might benefit from alternative or adjunctive treatments. However, the form that such treatments take will likely depend on the mechanisms thought to underlie a patient's disengagement. For instance, high levels of implicit avoidance prior to exposure therapy might be addressed by having patients engage in attentional training to increase emotional engagement. On the other hand, neuroplastic changes leading to difficulty perceiving and identifying with emotional stimuli might be treated by prefacing exposure therapy with strategies designed to “jump‐start” emotional systems. This latter approach might involve behavioral activation, in which patients engage in activities that are pleasurable, in order to increase overall emotional engagement with the environment (e.g., Wagner et al. 2007). Alternatively, neurostimulation might be used to modulate brain regions involved in avoidance or emotional arousal.

Part of understanding the mechanisms underlying emotional disengagement in psychopathology will also involve knowing its temporal relation to symptoms—for example, causal or correlational. To parse these possibilities, we need longitudinal work. Moreover, the answers to these questions may be complex. For instance, relationships between emotional disengagement and psychopathology may be bidirectional. Repeated bouts of depression could lead to neuroplastic changes that alter emotional engagement, resulting in less acclimation to negative stimuli and increased likelihood of another depressive episode the next time a negative event occurs (e.g., Figure 4). Going forward, it will be necessary to parse these and other potential directions of associations, in order to advance the most parsimonious model.

5.3. Specificity of Emotional Disengagement in Psychopathology

Another potential avenue to consider going forward will be to understand the contribution of depression to emotional disengagement and to the HARM‐A brain profile. Emotional context insensitivity theory (ECI; Rottenberg, Gross, and Gotlib 2005) suggests that depression is associated with blunted response to both positive and negative stimuli. This hypothesis has been supported using a variety of psychophysiological measures, such as eyeblink startle and the LPP, as well as behavior (e.g., facial expression) and subjective report (Bylsma 2021; Bylsma, Morris, and Rottenberg 2007). Therefore, one possibility is that emotional disengagement across the anxiety spectrum is attributable primarily to depression. Teasing out the unique contribution of depression to emotional disengagement is not an easy task, given that depression is confounded with the presence of other distress disorders and with loss of function that is manifest at the higher end of the anxiety spectrum. In some studies, we have found that depression (vs. anxiety) uniquely accounts for blunted LPPs to negative stimuli (Bauer and MacNamara 2021). Nonetheless, other work in my laboratory, led by Kayla Wilson, suggests that broader impairments in overall function (not depression specifically) may account for varied findings. That is, in a mixed internalizing sample, Kayla found that internalizing symptoms (both anxiety and depression) were associated with increased LPPs to negative pictures, whereas only greater functional impairment (measured using the World Health Disability Schedule; Üstün et al. 2010) was associated with blunted LPPs to negative pictures (Wilson and MacNamara 2025). Therefore, it is not yet clear whether depression has a unique effect on negative emotion processing that is opposite to anxiety, or whether more general decrements in function might offer a more parsimonious explanation for emotional disengagement across the anxiety spectrum.

One factor complicating mechanistic explanation of emotional disengagement at the far end of the anxiety spectrum and reconciling it with ECI's predictions is that most prior work has failed to include positive stimuli. ⁵ Therefore, going forward, it will be important to clarify whether emotional disengagement in internalizing psychopathology is evident for both negative and positive stimuli (in line with ECI theory but not Kayla's work, which found no associations for positive pictures). An alternative possibility is that blunting to emotional stimuli might stem in part from heightened response to neutral stimuli. That is, some work has suggested that neutral stimuli are particularly salient for individuals with anxiety and depression because ambiguous stimuli can be perceived as threatening (Premo et al. 2021) or due to difficulties with attentional inhibition (e.g., Moran and Moser 2015). Moreover, other models, such as the contrast avoidance model (CAM; Newman and Llera 2011) suggest that excessive worry and generalized anxiety disorder (GAD) are characterized by a consistently elevated state of negative emotion in response to all stimuli. While on the surface CAM's predictions seem in contrast to the notion of emotional blunting/disengagement, heightened response to innocuous stimuli might in part explain reduced response to emotional stimuli, if operationalized as the relative response to negative versus neutral stimuli. Therefore, going forward, it will also be important to work towards more consistent operationalization of emotional blunting/disengagement, in order to clarify whether severe internalizing psychopathology is best characterized by reduced differential or absolute response to negative stimuli.

6. Conclusion

I have argued here that engagement–disengagement is a key factor influencing the short‐term and long‐term effectiveness of negative emotion downregulation. Specifically, emotional disengagement, while rapidly deployed and more effective for high‐intensity stimuli in the short term, does not lead to long‐term reductions in emotional salience. Emotional disengagement also has implications for psychopathological processes underlying internalizing psychopathology. I have described a brain profile, HARM‐A, that is characterized by heightened amygdala and blunted LPPs to negative stimuli and have shown that this brain profile is related to increased dysphoria 2 years later in individuals with internalizing psychopathology. Though much work has focused on the role of excessive emotional response in anxiety and depression, continued work on emotional disengagement could yield new insight into the classification and treatment of the most severely affected patients with internalizing psychopathology.

Author Contributions

Annmarie MacNamara: conceptualization, formal analysis, funding acquisition, investigation, methodology, resources, writing – original draft.

Ethics Statement

Study procedures were in compliance with the Helsinki Declaration of 1975 (as revised in 1983) and were approved by the institutional review boards at the institutions where each study was conducted.

Consent

Informed consent was obtained from all participants included in the study.

Conflicts of Interest

The author declares no conflicts of interest.

Data Availability Statement

Several datasets are available in online repositories (e.g., OSF), as described in the original publications. Please contact the author for all other inquiries.