Mental Imagery of Suicide and Non-Suicidal Self-Injury: A Meta-Analysis and Systematic Review

1. Introduction

Self-injurious thoughts and behaviors (SITBs), or thoughts and behaviors related to harming oneself, are serious public health concerns. More than 700,000 people die by suicide each year (Fox et al., 2018; WHO, 2011), a rate which is only increasing (Centers for Disease Control and Prevention, 2020; Naghavi, 2019). Even more prevalent is non-suicidal self-injury (NSSI), or harming oneself without intent to die, which occurs in 4–5.9% of adults (Plener et al., 2016) and 15–18% of adolescents (Swannell et al., 2014). Research on suicide and NSSI has predominately focused on the pathway from suicidal ideation to suicidal behavior (Kleiman, 2020) and from thoughts of NSSI to engagement in NSSI (e.g., Turner et al., 2019). Along with serving as risk factors for suicidal and NSSI behavior, however, suicidal and non-suicidal self-harm cognitions are impairing and distressing in their own right (Hepp et al., 2020; Kleiman, 2020; Oppenheimer et al., 2022). As such, there have been recent calls to prioritize the study of suicidal and NSSI cognitions (Kleiman, 2020; Oppenheimer et al., 2022) in hopes of preventing suicidal and NSSI behavior and concomitant suffering and impairment.

Self-harm cognitions have been broadly conceptualized as verbal thoughts about harming oneself, with the vast majority of research on, and clinical assessment of, suicidal ideation and NSSI cognitions focused on verbal mental sentences about suicide or NSSI. Definitions of suicidal ideation and NSSI cognitions describe them as thoughts of killing oneself (in the case of suicidal ideation) or injuring oneself without suicidal intent (in the case of NSSI cognitions) and gold standard measures of suicidal ideation and NSSI cognitions ask, “Have you ever had thoughts of killing yourself?” and “Have you ever had thoughts of purposely hurting yourself without wanting to die?” (item 1 on the Self-Injurious Thoughts and Behaviors Interview-Revised (SITBI-R); Fox et al., 2020 and item 51 on the SITBI; Nock et al., 2007 respectively). Many individuals, however, experience their cognitions in the form of mental imagery, or seeing in the mind’s eye (Pearson et al., 2015), and many cognitive processes implicated in psychopathology take the form of mental imagery. Flashbacks associated with Post-Traumatic Stress Disorder (PTSD) (Brewin, Gregory, Lipton, & Burgess, 2010; Homes et al., 2007; Brett & Ostroff, 1985), mental images of feared situations or stimuli in social anxiety and specific phobias (Alfano et al., 2008; Hirsch, Clark, Mathews, & Williams, 2003; Hackmann, Clark, & McManus, 2000), appearance related mental imagery in eating disorders (Osman, Cooper, Hackmann, Veale, 2004), and intrusive mental imagery-based memories in depression (Deeprose & Holmes, 2010; Reynolds & Brewin, 1998) are all widely established risk factors for corresponding disorders (Hagenaars & Holmes, 2012; Heyes, Lau, & Holmes, 2013; Holmes et al., 2016). Only recently, however, have researchers begun to study NSSI and suicidal mental imagery, or images in the mind’s eye related to harming or killing oneself.

Maladaptive cognitions that take the form of mental imagery may be even more impairing than verbal thoughts given the strong ties of mental imagery with emotion, perception, and motivation. Mental imagery is often described as an emotional amplifier such that relative to verbal thoughts, negative mental imagery elicits greater negative affect and positive imagery elicits greater positive affect (see Holmes & Mathews, 2010 for a review). There also is extensive evidence that mental imagery shares overlap with real perception, suggesting that mentally imagining an event may be closer to really viewing or experiencing that event relative to verbal thoughts (Ganis et al., 2004; Kosslyn et al., 2001; Mathews et al., 2013; Pearson et al., 2015). Finally, mental imagery appears to have a strong motivational component. Individuals rate imagined events as more likely to occur than those that they have not imagined (Gregory et al., 1982; Pham & Taylor, 1999) and mentally imagining an action may make it more likely that an individual will actually engage in that action (e.g., Jeannerod, 1995; Libby et al., 2007; Renner et al., 2017). Taken together, this would suggest that suicidal and NSSI mental imagery may result in greater negative affect, be more similar to viewing or experiencing suicidal or NSSI behavior, and may motivate engagement in suicidal or NSSI behavior to a greater extent than verbal thoughts about NSSI and suicide. If true, standard measures of NSSI and suicide risk may be missing a critical piece of NSSI and suicidal cognitions particularly associated with risk. Directly assessing suicidal and NSSI mental imagery could therefore improve risk assessment while also providing a fruitful target for intervention.

Here, we provide the first systematic review or meta-analysis synthesizing what is currently known regarding suicidal and NSSI mental imagery. We provide meta-analytic estimates for the prevalence of suicidal and NSSI mental imagery, and qualitatively review the current literature on the content and characteristics of suicidal and NSSI mental imagery, and associations between suicidal mental imagery and suicidal behavior and between NSSI mental imagery and engagement in NSSI. We also qualitatively review what is known regarding how to intervene on suicidal and NSSI mental imagery. Given the nascent nature of this literature, meta-analytic evidence of the prevalence of NSSI and suicidal mental imagery and a review of this nature will increase confidence in the notion that NSSI and suicidal mental imagery are common, impairing, and worth additional research inquiry. In addition, this comprehensive overview is critical in identifying gaps in the existing literature and key areas in which additional research and clinical innovation is needed. Finally, we offer conclusions regarding how mental imagery should be incorporated into assessment and treatment for suicide and NSSI.

2. Method

This project was registered in PROSPERO (CRD42021282021) and followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines.

3. Results

4. Discussion

The present systematic and meta-analytic review is the first to evaluate and summarize the empirical body of literature on suicidal and NSSI mental imagery. The goal of this review was to establish meta-analytic estimates of the prevalence of suicidal and NSSI mental imagery. It further aimed to qualitatively review the literature on the content and characteristics of self-harm mental imagery and the associations between self-harm mental imagery and risk for self-harm behaviors, as well as discuss the literature on interventions for self-harm mental imagery.

Our meta-analytic review finds self-harm mental imagery to be highly prevalent. Even though the vast majority of research on self-harm cognitions has focused on verbal thoughts and that existing gold standard measures of suicide and NSSI risk exclusively assess verbal thoughts of self-harm, over three-quarters of individuals from clinical samples reported that they had experienced mental imagery related to self-harm. Prevalence rates were consistently high regardless of whether studies reported on mental images of suicide or NSSI and whether rates were from samples with a history of suicide attempts, suicidal ideation, or NSSI. However, rates of self-harm mental imagery did appear to be highest when individuals were experiencing high levels of distress, for example, when individuals actually engaged in self-harm behaviors. Importantly, these prevalence rates were pulled from clinical samples, and diverse clinical presentations at that; it therefore remains unknown what the prevalence of self-harm mental imagery is in the general population or among specific clinical (e.g., among individuals with depression) or demographic (e.g., among individuals of diverse gender identities) groups. In addition, few studies directly compared rates of mental imagery and verbal thoughts about suicide and NSSI. Findings from the few studies that have, are mixed. One study found that self-harm mental imagery may be more common than suicidal verbal thoughts (Lawrence et al., 2022) while two others found the opposite (Lawrence et al., 2021; McEvoy et al., 2017), though all three studies found that most individuals report both mental imagery and verbal thoughts about self-harm.

Regarding what we know about the content and characteristics of self-harm mental imagery, self-harm mental imagery often depicts actual self-harm engagement or means of self-harm engagement (Cloos et al., 2020; Crane et al., 2012; Haines et al., 1995; Holmes et al., 2007; Ng et al., 2016) and is perceived to be highly vivid, realistic, and preoccupying (Crane, Barnhofer, & Holmes, 2012; Holaday & Brausch, 2015; Holmes et al., 2007). Furthermore, results indicate that self-harm mental imagery may be especially vivid, realistic, and preoccupying when mental health distress is high (Hales et al., 2011; Schultebraucks et al., 2020). It makes sense that individuals find self-harm mental imagery to be intense and detailed if individuals are imagining actually engaging in self-harm. Mental imagery shares largely overlapping neural substrates with perception (Ganis et al., 2004; Pearson, 2019); as such, mental imagining self-harm may be experienced similarly to past experiences of actually engaging in self-harm or rehearsal of future engagement in self-harm.

Although the vast majority of individuals rated their self-harm mental imagery as at least moderately vivid, realistic, and preoccupying, there were mixed findings regarding whether individuals rated self-harm mental imagery as more distressing or as more comforting and positive (Hales et al., 2011; Holmes et al., 2007; Ng et al., 2016). This may reflect a risk or protective factor. On the one hand, individuals who rate their self-harm mental imagery as more distressing may be more likely to avoid self-harm behavior as they have learned that self-harm is associated with increased negative emotion. On the other hand, the increase in distress experienced when mentally imagining self-harm could lead individuals to more immediately need a means of regulating distress, which could lead to self-harm behavior. Similarly, experiencing self-harm mental imagery as comforting could reinforce self-harm as an emotion regulation strategy and lead individuals to habituate to pain associated with self-harm, or could lower distress to the extent that self-harm behavior is not needed to regulate distress.

A next step may be to decipher whether there are within-person or between-person differences in how comforting or distressing NSSI and suicidal mental imagery is. For example, it may be that some individuals experience this mental imagery as comforting and others as distressing, which may put different people at risk for self-harm behavior. Alternatively, it could be that there are specific mental images or certain moments of time in which mental imagery is more or less comforting and more or less distressing for a given individual, which also may impact their proximal risk for self-harm behavior. This line of inquiry is especially important given what is currently known regarding functions of NSSI. If some individuals engage in self-harm as a means of emotion regulation (Andover & Morris, 2014), for instance, NSSI mental imagery could either serve a similar regulatory function (increasing comfort) or could increase the need for regulation (increasing distress). Interventions could also then be tailored based on whether individuals experience their mental images of suicide and NSSI as more comforting or distressing.

Regarding potential differences in content and characteristics of self-harm mental imagery and verbal thought, few studies have directly compared the two. Mental imagery is, however, consistently rated as more emotionally laden and “more real” than verbal thoughts in other contexts (Holmes & Mathews, 2010). This suggests that self-harm mental imagery may be even more maladaptive or perhaps more strongly linked with risk for self-harm behavior than self-harm verbal thoughts in part because it is more vivid, realistic, and linked with a stronger emotional response. There is some evidence that more vivid and realistic mental imagery is associated with increased risk for maladaptive behavior in related fields. More vivid mental imagery is associated with higher intensity substance use (e.g., May et al., 2004) and food (e.g., Harvey et al., 2005; Tiggemann & Kemps, 2005) cravings and perhaps even consumption (Zorjan & Schienle, 2022). This would need to be studied in relation to self-harm directly, but it is possible that self-harm mental imagery is associated with especially high risk for self-harm behavior when it is perceived as vivid and realistic.

The reviewed empirical studies suggest that suicidal mental imagery may in fact be associated with a higher likelihood of suicidal behavior (Schultebraucks et al., 2020), even when controlling for verbal thoughts about suicide (Lawrence et al., 2021; Lawrence et al., 2022). Findings are more mixed regarding whether experiencing NSSI mental imagery is associated with increased risk for NSSI. Although Cloos and colleagues (2020) found that mentally imagining NSSI may be associated with engagement in NSSI, Hasking and colleagues (2018) did not find a significant association.

Some limited research has attempted to identify potential mechanisms linking self-harm mental imagery with self-harm behavior. Experimental studies identified within this systematic review, however, were largely limited to those focused on NSSI mental imagery. Findings from experimental studies investigating NSSI mental imagery suggest at least partial support for escape conditioning in which imagining oneself engaging in NSSI behavior lowers both physiological arousal and subjective negative affect (Brain et al., 1998; Brain et al., 2002; Haines et al., 1995). As such, it is possible that naturally occurring NSSI mental imagery also serves these functions, potentially negatively reinforcing NSSI through emotion regulation. However, whether NSSI mental imagery increases or decreases the likelihood of NSSI behavior remains unknown due to a lack of prospective assessments at the timescale in which NSSI risk unfolds (i.e., real-time ecological momentary assessment). It is possible that by achieving the desired emotional regulatory effects through mentally imagining NSSI behavior, some may observe reductions in the urge to engage in NSSI behavior. Yet, it also is possible that imagining NSSI behavior serves to make more salient the desired emotional outcomes resulting from NSSI behavior, thereby increasing approach motivation. Indeed, some evidence for this causal relationship is found in the craving literature, where imagining consumption tends to increase desire for, rather than satiate desire for, actual consumption (Devos et al., 2022). Although the reviewed experimental studies were conducted to test reinforcement theories of NSSI behavior itself, by using mentally imagined NSSI as proxy, findings provide insight into the physiological correlates of NSSI mental imagery. Future studies are needed to replicate and extend these findings and to more fully investigate experimentally induced suicidal mental imagery.

When interpreting the association between self-harm mental imagery and engagement in self-harm behavior, and the results of the extant experimental studies, it is useful to consider existing self-harm theories. Joiner’s concept of acquired capability for suicide, within the interpersonal theory of suicide (Joiner, 2005; Van Orden et al., 2010), is that acquired capability, or habituation to the pain and fear associated with suicide, is necessary for engagement in suicidal behavior. One way that individuals may effectively habituate to this pain and fear may be through repeated exposure to suicidal and NSSI mental imagery. In line with this conjecture, mental imagery results in habituation in other contexts (e.g., imaginal exposure to treat anxiety disorders; Maples-Keller & Rauch, 2020). Studies experimentally inducing NSSI mental imagery generally also found decreases in negative affect and physiological arousal as you would expect during habituation. It is important to note, however, in the only study to directly evaluate the relation between suicidal mental imagery and acquired capability for suicide, the association was not significant (Holaday & Brausch, 2015). Given evidence suggesting that the measure Holaday and Brausch (2015) used to assess acquired capability for suicide (i.e., the Acquired Capability for Suicide Scale) may have poor psychometric properties (Rogers et al., 2021), and that this finding has yet to be replicated, future prospective research using alternative measures of acquired capability are needed to evaluate this hypothesis more thoroughly.

A second relevant theory of self-harm is the integrated motivational-volitional theory of suicidal behavior (IMV; O’Connor & Kirtley, 2018). According to the IMV model, mental imagery may serve as a ‘volitional moderator’ or factor that increases the likelihood of transitioning from suicidal ideation to suicidal behavior. O’Connor and Kirtley posit that this is due to mental imagery serving as a form of cognitive rehearsal for the behavior. Indeed, the concept that mentally imagining an action increases the likelihood of engaging in that action has been demonstrated in other fields (e.g., Ji et al., 2021; Koehler, 1991; Libby et al., 2007). For example, mentally imagining engaging in behavioral activation has been shown to be associated with a higher likelihood of actual engagement in behavioral activation (e.g., Ji et al., 2021; Renner et al., 2017). This may be because, as noted above, mental imagery is similar to real perception. Mentally imagining self-harm thus may act as mental rehearsal, preparing individuals for engagement in self-harm. However, this remains to be tested experimentally.

Supplementary Material