Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Aug 21.
Published in final edited form as: J Clin Psychol. 2025 Dec 23;82(3):350–358. doi: 10.1002/jclp.70078

Nonsuicidal Self-Injury (NSSI) in Virtual Reality: Effect of Blood-Viewing on Affect Controlling for Pain

Rachael E Dumas 1, Stephanie E Stacy 1, M Kati Lear 1, Kandice M Perry 1, Lara E Glenn 1, Daniela A Branson 1, Gabriella M Zeller 1, Kyle Summerfield 2, Carolyn M Pepper 1
PMCID: PMC13492349  NIHMSID: NIHMS2199449  PMID: 41431423

Abstract

Empirical evidence suggests engagement with nonsuicidal self-injury (NSSI) serves to regulate emotions, though there is controversy surrounding specific mechanisms by which this occurs, partially because of limited ability to observe the behavior within a laboratory setting. Although behavioral analogues (e.g., pressure, cold) have found support for the role of pain in affect regulation, they have neglected other potential processes, including viewing one’s wound and seeing blood. In a virtual reality paradigm, participants were randomly assigned to either a “see blood” (n = 35) or “no blood” (n = 29) condition following a negative mood induction to simulate the experience of self-cutting. Positive (PA) and negative affect (NA) were assessed at baseline, post-induction, and post-NSSI. Analyses failed to support differential changes in both NA and PA following engagement with NSSI between conditions, though NA increased significantly post-NSSI across conditions. Results from this study provide no evidence to suggest viewing blood alone (i.e., without pain) is an integral part of emotion regulation processes associated with NSSI. Unexpected mood worsening following virtual reality NSSI simulation raises questions about the utility of this paradigm for studying the affect regulation functions of NSSI above and beyond existing methods (e.g., EMA). However, qualitative evidence and follow-up assessments suggest it is a feasible and safe way to study NSSI within a laboratory setting. Future research is needed to adjust virtual reality procedures for NSSI to make the experience more realistic (e.g., adding pressure) and to clarify whether it can capture established affect regulation functions of NSSI.

Keywords: NSSI, virtual reality, affect regulation, negative affect, blood

Nonsuicidal self-injury (NSSI) is defined as deliberate self-harm enacted without suicidal intent (most often self-cutting; Cipriano et al., 2017). It is a growing clinical and public health concern, and prevalence rates are estimated to range from 7.5 to 46.5% across adolescent, young adult, and adult populations (Cipriano et al., 2017). NSSI has been associated with a range of psychopathology (e.g., depression, eating disorders, borderline personality disorder) and is one of the leading risk factors for suicide attempts—second only to suicide ideation (Franklin et al., 2017). Moreover, the presence of comorbid NSSI is associated with elevated psychiatric symptoms (e.g., Wang & Eaton, 2023). Given its broad association with risk for poor outcomes, researchers have long endeavored to elucidate the mechanisms driving the perceived need to engage in the behavior and the resulting function that the behavior serves.

There have been many proposed etiological models of NSSI (e.g., Hooley & Franklin, 2018; Selby & Joiner, 2009). One mechanism with consistent support across all proposed models is that engagement with NSSI serves to regulate difficult and intense emotions, with one meta-analysis suggesting that 63–78% of individuals who engage in NSSI report emotion regulation as a primary function (Taylor et al., 2018). Even within this established function of the behavior, there is controversy surrounding the mechanisms by which an individual experiences a temporary relief from their negative affect. Various possibilities have been proposed, including those surrounding the avoidance of emotional and cognitive stimuli (e.g., temporary relief from negative affect).

The specific causal mechanisms that drive decreased negative affect following NSSI remain understudied, in part due to methodological limitations to studying the behavior. Because of ethical concerns related to inducing self-injury within a laboratory setting, the majority of NSSI research has used self-report or clinician-rated accounts of self-injury and its functions. Such research is limited by a range of factors, including recall bias, impression management, motivation, and adoption of response sets (Ammerman et al., 2018).

Other studies have used standardized behavior analogues to attempt to recreate the experience of NSSI, primarily by administering a painful stimulus (e.g., cold, heat, pressure, shock; Bresin & Gordon, 2013; Denning et al., 2025). Although these studies have provided valuable support for how the experience of pain may gratify the self-reported functions of NSSI, they have neglected other processes. For example, in cutting, the most frequently used method of NSSI, reported by over 70% of individuals endorsing NSSI (Cipriano et al., 2017), the impact of examining one’s wound or seeing blood may serve emotion regulation functions. Prior self-report research has suggested that individuals often report that blood and the resulting wounds and scars associated with them are important to emotion regulation (e.g., Glenn & Klonsky, 2010; Gradin Franzén & Gottzén, 2011). Only two experimental studies investigating the role of blood have been conducted. In one study, the sight of blood led to a faster decrease in arousal among participants with borderline personality diagnoses (i.e., researcher administered blade stimulus; Naoum et al., 2016), and in another, a group that was randomized to see blood demonstrated increased positive affect compared to a group that received a sham injury (i.e., finger prick; Stacy et al., 2022).

Although research studying NSSI within a laboratory setting using standardized behavior analogues has provided valuable information about the functions of NSSI, it is also limited in its scope. It is yet unclear to what extent these stimuli relate to the actual experience of NSSI (e.g., ecological validity) or whether other factors separate from pain, such as seeing blood, may be essential components to affect regulation in NSSI. Thus, there is a clear and present need for research that 1) more closely approximates the actual experience of self-injury and 2) can control for potential confounding variables associated with pain (e.g., set response styles to pain) to directly test the role of seeing blood on NSSI processes and functions.

The use of virtual reality in NSSI research could allow for safe and ethical study of the process of self-injury, changes in thoughts and emotions immediately preceding and following the act, and the manipulation of external triggers (e.g., scripted social situations to evoke sadness or anger). Virtual reality paradigms have been implemented successfully and safely for other behaviors that are unethical and/or difficult to induce in the laboratory, including suicide attempts (e.g., Caulfield et al., 2022; Huang et al., 2021), posttraumatic stress disorder (e.g., Deng et al., 2019), and mood disorders (e.g., Fodor et al., 2018). Researchers have highlighted the usefulness of these techniques in elucidating causal processes for low-occurrence behaviors, which are otherwise difficult to establish using other methodology (e.g., ambulatory assessments, cross-sectional data). Such virtual reality paradigms could also be used clinically to evoke triggering emotions to teach alternative distress tolerance skills.

Although virtual reality laboratory paradigms for NSSI may reduce ecological validity through various procedural mechanisms (e.g., eliminating experience of pain, reducing choice in when/how to engage in NSSI), it may be particularly useful in providing researchers with the ability to observe and to control for specific aspects of self-injury within the laboratory to better understand theoretical processes relevant for subsequent intervention. In a now classic article, Mook (1983) argued that ecological validity is not necessary when the goal of the study is to test theoretical predictions to better understand underlying processes. In this spirit, we aim to study the emotion regulation processes of NSSI in a laboratory environment that does not produce physical sensations or pain and is limited to certain methods of NSSI (i.e., cutting), thereby omitting some aspects of the construct to allow us to more carefully test the impact of other processes.

Thus, the aims of the current study were twofold: to assess the feasibility of virtual reality for studying emotion regulation functions of NSSI cutting and to replicate and extend findings that suggest seeing blood impacted a subsequent increase in positive affect following an NSSI proxy condition (Stacy et al., 2022) while controlling for pain. For the feasibility aim, we expected to see reductions in negative affect and increases in positive affect following NSSI, previously established in both EMA and experimental research. We also expected that participants reporting cutting as a main form of NSSI who saw blood in the virtual reality scenario would describe the experience as more representative of their NSSI experiences, would engage with the virtual reality scenario for longer, and would cut themselves more times compared to those who do not see blood. Finally, based on previous findings that suggest blood plays an important role in the emotion regulation process associated with NSSI, we expected that individuals randomized to see blood would demonstrate larger magnitudes of decreased negative affect and increased positive affect after the self-injury scenario compared to those who do not see blood.

Methods

Participants

Participants were undergraduate students (n = 64) 18 years and older who reported engaging in NSSI primarily via cutting within the last three years. Eligible students were invited to enroll in the study based on responses to a mass-screening questionnaire and received course credit for participation. Students were excluded if they were under 18 years of age or had blood-injection phobia (or had ever fainted during an injection or blood draw). In a double-masked procedure, participants were randomized to a “see blood” condition (n = 35) or to a “no blood” condition (n = 29).

Participants (85.9% cisgender women) ranged in age from 18 to 25 (M = 19.6; SD = 1.5). The majority of the sample was White (92.2%), non-Hispanic (84.4%), and identified as heterosexual (51.6% heterosexual; 37.5% bisexual; 10.9% gay). Specific characteristics of NSSI within the sample included a median of 50 cutting episodes (range 2 – 400) and median of 5 distinct NSSI methods (versatility range 1 – 11), though all participants were selected because they reported cutting as a primary method. All participants indicated that affect regulation functions associated with NSSI (i.e., calming oneself down, releasing emotional pressure, reducing overwhelming emotions) were at least “somewhat relevant” to their experiences of NSSI, though the median for each item was “very relevant.” Randomized groups were similar with respect to gender identity, sexual orientation, race, age, and NSSI lifetime frequency and versatility.

Self-Report Measures

Inventory of Statements About Self-Injury (ISAS)

The ISAS (Klonsky & Glenn, 2009) was used to screen for individuals who reported self-injury via cutting within the last three years. The ISAS assesses the lifetime frequency of 18 NSSI behaviors (e.g., cutting, burning). Participants estimated the number of times they have performed each behavior. NSSI versatility was assessed by the number of NSSI behaviors reported by participants across their lifetime. The behavioral scales have shown stability over time (i.e., 0.75 1-year test-retest correlations for cutting; Glenn & Klonsky, 2011).

Positive and Negative Affect Schedule (PANAS)

The PANAS (Watson et al., 1988) was used to assess negative and positive affect at three distinct timepoints (i.e., baseline, post-mood induction, post-NSSI). The PANAS was shortened to include five negative affect (NA) items (i.e., afraid, ashamed, hostile, nervous, and upset) and five positive affect (PA) items (i.e., excited, enthusiastic, inspired, alert, and determined). PA items were added after data collection had begun (n = 47). Participants rated each item according to the extent they felt each item “at this moment” using a 5-point Likert scale (1 – “not at all;” 5 – “extremely”). Bresin and Gordon (2013) reported high Cronbach’s alphas ranging from 0.76 to 0.79. In the current study, there was generally high internal consistency for both NA (Cronbach’s α = 0.60 – 0.78) and PA (Cronbach’s α = 0.75 – 0.80).

Importance of Blood Questionnaire

The Importance of Blood Questionnaire was adapted from Glenn and Klonsky’s (2011) Brief Structured Interview for Non-Suicidal Self-Injury. It assesses whether participants deem the sight of blood important during NSSI. The questionnaire asks, “Is it important for you to see blood during a skin-cutting episode?” with participants responding “yes” or “no.” If “yes,” participants rated how important it is for them to see the blood (1 – “ not important;” 5 – “extremely important”), briefly described why it is important for them to see their blood, and answered additional questions about the importance of blood to self-injury. A total of 89.1% (57 of 64) participants identified seeing blood during NSSI as important, with the median rated importance 4 out of 5. There were no observed differences between conditions on the reported importance of seeing blood.

Follow-Up Interview

In an optional follow-up interview at the conclusion of the study, participants were asked a series of questions to assess their experience with NSSI in virtual reality, including “how similar was this to your experience when you self-injure?” and “during the simulation, were you looking at the wound?” These responses were recorded on a tape recorder and analyzed for qualitative themes to better understand participants’ experiences of NSSI during the virtual reality simulation. Responses to the similarity item were coded as 0 (not at all like self-injury experiences), 1 (somewhat similar), or 2 (very similar).

One-Week, Post-Study Follow-Up

An optional one-week follow-up was implemented to ask participants of their experiences of NSSI thoughts and behaviors following their participation, including whether participating in the study had increased urges and/or self-injurious behaviors, whether participating triggered an onset of self-injury that had originally stopped, and if participating in the study increased the frequency of self-injury.

Mood Induction

A mood induction task was adapted from Bresin and Gordon (2013). Participants listed four “attributes of the person you would ideally like to be” and four “attributes of the person you feel you should be.” Then, participants rated each attribute on a) how far they felt they were from possessing the attribute and b) how important they felt it was to possess it using a 7-point Likert scale (1 – “not at all;” 7 – “extremely”). The researcher calculated total scores to identify which attribute the participant felt was both furthest from attainment and most important. Finally, participants were asked to “choose a specific time in your life in which you were supposed to act with or be [chosen attribute] and were not.” Participants were encouraged to choose a situation that “caused great stress and made them upset, mad, nervous, or fearful” and rated the stressfulness of the situation using a 7-point Likert scale (1 – “not stressful at all;” 3 – “somewhat stressful;” 5 – “very stressful;” 7 – “the most stress I have experienced”). The participant was instructed to write about the event in detail for 10 minutes.

Virtual Reality

Virtual reality technology self-cutting environment was built in Unity3D, targeting the original HTC VIVE Virtual Reality headset. A front-mounted first-generation Leap Motion IR tracker was used for hand tracking. The application relied exclusively on hand tracking for all input and interactions, allowing the user to pick up a knife and to manipulate their position and orientation to simulate the desired actions. The marking effects were achieved through a combination of decals and particle effects.

Procedure

All procedures were approved by the University Institutional Review Board. Prior to the participants’ arrival to the lab, a research assistant randomized the condition of the virtual reality paradigm to “blood” or “no blood” by selecting a hidden box on the program so that both the researcher and the participant were blind to the condition prior to the experiment beginning. Upon arrival to the lab, students provided written consent for participation and completed online measures, including demographic information and a baseline PANAS (i.e., Time 1). Participants were then allowed time to acclimatize themselves to the virtual reality environment (i.e., adjusting skin tone, visual field, headset; independently exploring virtual space). Once participants verbally confirmed they were comfortable in the virtual space, participants completed the mood induction, immediately followed by another PANAS measurement (i.e., Time 2).

Participants were then given the following instructions before putting the virtual reality headset back on: “For this task, you have been randomly assigned to either a ‘see blood’ or ‘no blood’ condition. Regardless of the condition you’ve been assigned to, you will pick up the knife using your dominant hand and cut your arm with your palm facing up. You can cut your arm as many times as you would like. You can look at your wounds for as long as you want, but for no more than 5 minutes. When you are finished looking at your wounds, please look to the right and say, ‘I’m done.’ Then you can remove the headset.” If participants had no further questions, they were instructed to put the headset on and to put themselves back in the stressful situation they spent 10 minutes writing about. The researcher recorded the time spent in the simulation (i.e., from the time the individual picked up the knife and began cutting to the time they indicated they were done) and documented the number of cuts made.

Following the simulation, participants completed the final PANAS (i.e., Time 3), the blood questionnaire, and were debriefed (i.e., given a resource sheet with safety information). Participants could then elect to complete an optional follow-up feedback interview (n = 58). We were particularly concerned about participant well-being following the negative mood induction and simulated NSSI. At the end of the study, the researcher inquired into the participant’s safety and wellbeing and offered to watch a funny video and/or eat cookies with the participant prior to leaving the lab. If participants continued to express distress, the researcher offered to play a game. Following this activity, if participants continued to express feeling unsafe, trained graduate student researchers assessed safety and formulated a safety plan with the participant. All participants were also given a number of mental health and self-injury related resources at the end of the study. Two researchers were always present in the lab in case one needed to seek consultation from licensed clinical psychologists in the department training clinic, located down the hall from the lab. Part way through the study, we added an optional one-week follow-up interview to assess possible negative outcomes from participation in the study. Of 28 invited participants, 22 (78.6%) opted to participate. None reported increased urges to self-injure or actual NSSI following the study.

Analytic Plan

Main Analyses

An independent samples t-test was used as a manipulation check to assess potential group-level differences in the rated stressfulness of the chosen mood induction situation, time spent viewing wounds, and number of cuts made. Adequacy of the negative mood induction on changes in NA and PA from baseline (T1) to post-induction (T2) was examined using separate 2 (Condition) × 2 (Time) mixed-effects factorial analysis of variances (ANOVAs) with repeated measures on the final factor. Hypothesized differences in reaction to the manipulation of blood from post-induction (T2) to post-manipulation (T3) were assessed in separate 2 (Condition) × 2 (Time) mixed-effects ANOVA models for NA and PA. Differential responses to seeing blood were expected to manifest as significant interactions. Descriptive statistics for NA and PA across conditions and time points are provided in Table 1. Effect sizes for omnibus ANOVA effects are presented as partial eta squared (ηp2). Values of 0.01, 0.06, and 0.14 were used as interpretive benchmarks for small, medium, and large effects, respectively. All analyses were conducted using SPSS software.

Table 1.

Means and standard deviations for negative (NA) and positive affect (PA) by experimental condition.

n T1 T2 T3
NA
Blood 35 9.51 (2.64) 11.71 (3.20) 12.80 (4.19)
No-Blood 29 8.38 (2.23) 10.48 (3.23) 11.52 (4.01)
PA
Blood 25 9.68 (3.50) 7.96 (2.98) 8.08 (3.23)
No-Blood 22 8.59 (3.65) 8.32 (4.12) 8.50 (3.73)
Open in a new tab

A Priori Power Analyses

A power analysis was conducted for the mixed effects ANOVA models. Estimating a small effect for two conditions (blood and no blood) with three measurements of affect (baseline, post-mood induction, post-procedure), a sample of 74 is recommended to achieve adequate power. Ultimately, following a lengthy data collection halted by the COVID-19 pandemic, a total sample of N = 64 was recruited. Given that the virtual reality technology was becoming increasingly outdated, this sample, consisting of 35 participants in the “blood” condition and 29 participants in the “no blood” condition was deemed acceptable, albeit underpowered, to analyze condition differences.

Results

Manipulation Check

Event Stress and Engagement with Virtual Reality Paradigm

Group-level differences in the reported stressfulness of the participant-chosen mood induction situation, time spent viewing wounds, and number of cuts made were examined using independent samples t-tests. Analysis failed to support group-level differences in the reported stressfulness of events selected (p = 0.57), time spent viewing wounds (p = 0.053), and the number of cuts made (p = 0.28).

Mood Induction: Changes in Negative and Positive Affect

Two separate 2 (Condition) × 2 (Time) mixed-effects ANOVAs were conducted to assess expected changes in NA and PA from baseline (T1) to post-mood induction (T2). Analyses identified a main effect of time for NA (p < 0.001, ηp2 = 0.52), though this effect was not meaningfully different across the “blood” and “no blood” conditions (p = 0.85, ηp2 = 0.001). Further evaluation of scores indicated significant increases in NA from T1 (M = 9.00, SD = 2.53) to T2 (M = 11.16, SD = 3.25) for all participants (p < 0.001, Δ = 2.16). Analyses of changes in PA, however, identified a conditional effect of time (p = 0.04, ηp2 = 0.09), wherein participants in the “blood” condition evidenced a larger magnitude of decrease in PA post-mood induction (T1: M = 9.68, SD = 3.50; T2: M = 7.96 SD = 2.98) than did participants in the “no blood” condition (T1: M = 8.59, SD = 3.65; T2: M = 8.32, SD = 4.12). Collapsed analyses indicated significant decreases in PA from T1 (M = 9.17, SD = 3.57) to T2 (M = 8.12, SD = 3.52) for all participants (p = 0.004, Δ = −1.04). Results provide strong support for the efficacy of the mood manipulation, with evidence of increased NA and decreased PA across conditions.

Primary Analyses

Blood Viewing: Changes in Negative and Positive Affect

Two separate 2 (Condition) × 2 (Time) mixed-effects ANOVAs were conducted to assess expected changes in NA and PA from post-mood induction (T2) to post-self-injury (T3). Analyses identified a main effect of time for NA (p < 0.001, ηp2 = 0.10), though this effect was not meaningfully different between the “blood” and “no blood” conditions (p = 0.85). Further evaluation of scores indicated an unexpected increase in NA from T2 (M = 11.16, SD = 3.25) to T3 (M = 12.22, SD = 4.13) for all participants, collapsing across conditions (p < 0.01, Δ = 1.06). Analyses of changes in PA, however, provided no evidence of changes in PA from T2 (M = 8.13, SD = 3.52) to T3 (M = 8.28, SD = 3.44) across conditions (p = 0.55, Δ = 0.15). No between-by-within interactions were observed in NA and PA analyses (all p ≥ 0.93; See Figure 1).

Figure 1.

Figure 1

Open in a new tab

2 (Condition) × 2 (Time) interaction for change in NA and PA from post-mood manipulation to post-virtual reality NSSI engagement.

Note. Solid lines represent “blood” condition and dotted lines represent “no blood” condition.

Acceptability and Safety

Of 64 participants, 58 (90.6%) completed the optional interview immediately following the study. Across conditions, 63.8% (i.e., 37 out of 58) of interviewed participants reported that the experience was somewhat or very similar to their actual experience of NSSI. As expected, differences were observed across conditions, with a larger proportion of participants in the “blood” condition (i.e., 75.0%; 24 out of 32) reporting the experience was somewhat or very similar to their actual experience of NSSI than the “no blood” condition (50.0%; 13 out of 26; X2 (1, N = 58) = 3.88, p < .05). Common feedback regarding the similarity of the experience included the usefulness of the mood induction for approximating the emotional state of participants‘ experience while engaging in NSSI outside of the laboratory and a feeling of satisfaction or relief following engagement with the virtual NSSI. Common themes regarding dissimilarities included missing blood (i.e., “no blood” condition), lack of a sensation of touch (e.g., pressure, pain), the instrument used, location of injury, and the feeling of nervousness associated with having the researcher in the room.

Finally, of the 22 participants who completed the one-week post-study follow-up, 72.7% of participants (i.e., 16 out of 22) reported experiencing no urges to engage in self-injury following the study. Although 27.3% (i.e., 6 out of 22) of participants reported experiencing an urge to self-injure in the week following the study, only two of these participants attributed this urge to the study itself, and they reported that these urges abated in the days following the study. Importantly, at one-week follow-up, none of the respondents reported having engaged in NSSI as a result of the study or having relapsed.

Discussion

The current study aimed to replicate and extend findings from previous research that suggested manipulating the presence of blood in a behavioral analogue of NSSI (i.e., finger prick) produced differential effects on affect regulation so that participants randomized to see blood demonstrated a larger magnitude of increase in PA following the NSSI proxy (Stacy et al., 2022). We attempted to isolate the effect of blood by randomizing participants to either a “blood” or “no blood” condition following a negative mood induction while controlling for pain in a novel virtual reality behavioral analogue of NSSI. Analyses failed to support our hypothesis that participants randomized to the “blood” condition would demonstrate a larger magnitude of decrease in NA and increase in PA following engagement with NSSI in virtual reality. In fact, results suggested that across conditions, participants demonstrated a significant increase in NA and unchanged PA from the negative mood induction following engagement with NSSI in virtual reality, which contradicts our general hypothesis that engaging in NSSI in virtual reality would fulfill previously established affect regulation functions.

Interestingly, although manipulation of the presence of blood did not result in differential outcomes in affect regulation, 89.1% of the sample (57 of 64) reported that viewing blood was important to their experiences of NSSI. Findings from the current study provide no evidence to suggest that the sight of blood alone (i.e., without sensory touch, pressure, or pain) results in differential changes in affect regulation. Thus, it may be that blood serves a secondary function to pain so that rather than functioning to regulate affect directly, it serves to moderate and/or inform the response to pain (e.g., showing that NSSI was done “right” or that it is “real”). Similarly, the lack of any sensory information from the virtual blade may not have been sufficiently similar to actual NSSI to downregulate affect. Results of this study align with earlier findings demonstrating that painful stimulation leads to a larger reduction in negative affect than non-painful stimulation for people with a history of NSSI (Bresin & Gordon, 2013). Overall, these results failed to support the hypothesis that seeing blood is an essential part of affect regulation in NSSI and instead, lend support to the importance of pain. Future research may aim to parse apart nuances to changes in affect following pain using virtual reality to control for various aspects of the self-injury scenario.

Findings from the current study also failed to provide evidence to suggest that a virtual reality behavioral analogue of NSSI would replicate established affect-regulating functions of NSSI (i.e., relief from NA and increase in PA), instead demonstrating an increase in NA and unchanged PA following engagement with NSSI. Prominent theories of NSSI posit that affect regulation (e.g., relief from NA) is a core function of the behavior (e.g., Chapman et al., 2006; Hooley & Franklin, 2018). Cross-sectional research has consistently revealed that participants report affect regulation as a primary function of NSSI (see Taylor et al., 2018 for review), which may suggest that our virtual reality paradigm was unable to produce a similar functional effect within the laboratory. It is possible that negative mood induction may have elicited emotions that were too strong to be effectively downregulated using NSSI in virtual space. Similarly, other aspects of the experimental procedures, such as discomfort from the virtual-reality environment or being asked to perform a private behavior in public, directly contributed to the increases in negative affect observed across all conditions.

Emerging research in ecological momentary assessment (EMA) and other ambulatory assessments has produced mixed findings, with some studies finding a decrease in NA following engagement with NSSI (e.g., Armey et al., 2011; Andrewes et al., 2017a, 2017b) and others finding an increase in NA following NSSI (e.g., Houben et al., 2017). Notably, although a meta-analysis found evidence for a medium to large effect for a decrease of NA following NSSI in ambulatory assessments (Kuehn et al., 2022), the length of time an assessment of NA occurred following engagement with NSSI varied, ranging from 1 to 10 hours post-engagement. In fact, several studies suggest that NA may follow a curvilinear trajectory, with NA peaking shortly after NSSI (e.g., Andrewes et al., 2017; Armey et al., 2011). Thus, it is possible that our assessment of NSSI, which directly followed engagement with virtual reality NSSI, reflects the beginning of such a curvilinear pattern of responding to NA. Other research highlights that engagement with NSSI results in a very small decrease in NA and that this decrease is not meaningfully different from the decrease in NA that occurs after an individual has an intense urge to engage in NSSI and does not (Störkel et al., 2023), suggesting that individuals may be falsely and retrospectively associating a natural decrease in emotions to engagement with NSSI. This finding supports the Cognitive-Emotional Model of NSSI, which posits that outcome expectancies of emotion regulation, coupled with low self-efficacy to resist the behavior, drive one’s explanation for their NSSI behavior more than actual negative reinforcement of reduced negative affect following actual NSSI (Hasking et al, 2017).

Finally, it is possible that our combined assessment of negative affect via an aggregated scale may have obscured “successful,” albeit small, decreases in certain negative emotions (e.g., “nervous,” “upset”) with increases in others (e.g., “ashamed”). This may be partially in line with the Experiential Avoidance Model of NSSI that suggests individuals may only benefit temporarily from an imminent escape from a particular emotional state, which is then followed by an eventual increase in other negative emotions as a consequence of engaging in the behavior as experiential avoidance (Chapman et al., 2006). Together, recent findings from novel methods of studying NSSI, especially those that are able to measure affect more closely to the behavior itself (e.g., EMA) suggest that there is much to be explored regarding both the time-oriented nature of decreases in NA. Virtual reality paradigms may be particularly useful in further investigating whether immediate changes in affect following an NSSI event differ substantially from longer-term changes in affect expected without NSSI.

Despite these findings, qualitative feedback from follow-up protocols suggested that virtual reality is a feasible and safe way to study NSSI within a laboratory setting. A majority of our sample (i.e., 63.8%) reported that the experience was at least somewhat similar to their experiences of NSSI, despite approximately half the sample having been randomized to a condition where they do not see blood following virtual cutting. Consistent with our hypotheses, those who were randomized to see blood were more likely to describe the experience as more representative of their NSSI experiences, though they did not engage with the paradigm longer or cut themselves more times compared to those randomized to the “no blood” condition, which may be a result of other study limitations (e.g., being observed by the researcher, cutting different locations, no physical sensations). Moreover, at one week follow-up, 90.9% of respondents reported no negative effects from the study, and those with increased urges to self-injure reported that they were manageable.

Taken together, these participant reactions suggest that the virtual reality environment is a reasonable proxy for NSSI and is safe for participants. Virtual reality paradigms have been implemented successfully and safely for other behaviors that are unethical to induce in the laboratory. For example, several studies have implemented virtual reality paradigms to study suicide (e.g., Caulfield et al., 2022), and a longitudinal study has supported the safety and ethics of such a research method (Huang et al., 2021). Improving on the methodology in virtual reality paradigms that aim to study NSSI (e.g., implementing choice, improving graphics) while continuing to monitor the safety and ethics of the procedure may provide valuable information for controlled laboratory studies in NSSI that are able to elucidate causal relationships other methodology cannot.

Strengths and Limitations

Findings from this study should be interpreted in context with the following strengths and limitations. To our knowledge, this was the first study to successfully use a novel experimental paradigm using virtual reality to study NSSI. In addition, this was the first study to eliminate the potential confound of pain while assessing for the effect of viewing one’s blood on affect regulation functions of NSSI. Furthermore, participants indicated that the experience was similar to their actual experience of NSSI. Our sample also consisted of a balance of heterosexual and sexual minority (LGBQ+) individuals (i.e., 48.4% of the sample identifying as bisexual or gay), which represents a difficult to recruit and understudied population at particular risk for NSSI.

Among study limitations, data from our experimental design at times may have contained error as a result of glitches in increasingly outdated technology (e.g., virtual arm disappearing, blood “floating” above the arms). Due to delays in data collection associated with the COVID-19 pandemic, data collection spanned from 2017 – 2024. During this time, updates to study graphics were not made to provide consistency in experiences, but this also led to eventual lags in graphics as technology continued to advance. Although we do not believe this would be sufficient to significantly alter findings as these spurious events were randomly distributed across conditions, it may have negatively affected a closer approximation of the experience of NSSI for some participants due to delays in engaging in the paradigm after the mood induction (e.g., as a result of restarting the technology). Further, as graphics continue to improve, the enhanced realism of cutting-edge virtual reality paradigms may produce different results. At this point, however, it is unclear that virtual reality paradigms create less noise in the data than other experimental paradigms (e.g., standardized pain stimuli).

There were several other aspects of the graphics and the experimental design that may also limit the generalizability to self-injury in more naturalistic settings, such as the instrument used (i.e., a knife rather than more commonly used instruments such as pocketknives, razor blades, box cutters), the appearance of the cuts (i.e., appeared as unsteady, wavy cuts rather than straight), and the absence of any physical sensation (e.g., pain, pressure). Many participants also noted that they felt uncomfortable or self-conscious during the scenario, as they were aware they were being observed by the researcher and they self-injure in private. Moreover, participants were only allowed to cut themselves on their forearms, and a large proportion of our sample reported only cutting on their legs or other parts of their body (i.e., 53.1%, 34 out of 64 participants).

Future Directions

Future researchers should endeavor to improve upon the design of this experiment to continue to evaluate the usefulness of virtual reality to more closely approximate the experience of NSSI within the laboratory (e.g., implementing choice of instrument/location of NSSI, improving consistency in graphics, providing a proxy for pressure). Moreover, given research that suggests engagement with NSSI results in a very small decrease in NA that is not meaningfully different from the decrease in NA that occurs after an individual has an intense urge to engage in NSSI and does not (Störkel et al., 2023), researchers may wish to include additional measures to assess for both changes in affect (i.e., assessing affect at multiple timepoints following engagement with NSSI in virtual reality) and measures that may explain the expectancy that NSSI results in affect regulation (e.g., relieving perceptions of tension, increasing perceptions of control). Finally, in interviews after the experiment, some individuals described feelings of relief and satisfaction following the virtual injury. It is possible that a virtual reality paradigm is more effective for a subgroup of participants that could be identified in future research.

Conclusions

Analyses failed to support differential changes in both NA and PA following engagement with NSSI between conditions, though NA increased significantly post-NSSI across conditions. Results from this study provide no evidence to suggest viewing blood alone (i.e., without pain) is an integral part of emotion regulation processes associated with NSSI. It is also unclear whether virtual reality is a useful paradigm for studying the affect regulation functions of NSSI above and beyond existing methods (e.g., EMA), especially without a laboratory simulation of pain (e.g., pressure). However, qualitative evidence and follow-up procedures suggests it is a feasible and safe way to study NSSI within a laboratory setting. Future research is needed to adjust virtual reality procedures for NSSI accordingly to make the experience more realistic (e.g., pressure) and to clarify whether it can capture established affect regulation functions of NSSI.

Acknowledgments

This project was supported by a grant from the National Institute of General Medical Sciences (Grant P20GM103432) from the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

We have no known conflict of interest to disclose.

References

  1. Ammerman BA, Berman ME, & McCloskey MS. (2018). Assessing Non-Suicidal Self-Injury in the Laboratory. Archives of Suicide Research, 22(2), 193–223. 10.1080/13811118.2017.1319312 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Andrewes HE, Hulbert C, Cotton SM, Betts J, & Chanen AM. (2017a). An ecological momentary assessment investigation of complex and conflicting emotions in youth with borderline personality disorder. Psychiatry Research, 252, 102–110. 10.1016/j.psychres.2017.01.100 [DOI] [PubMed] [Google Scholar]
  3. Andrewes HE, Hulbert C, Cotton SM, Betts J, & Chanen AM. (2017b). Ecological momentary assessment of nonsuicidal self-injury in youth with borderline personality disorder. Personality Disorders: Theory, Research, and Treatment, 8(4), 357–365. 10.1037/per0000205 [DOI] [PubMed] [Google Scholar]
  4. Armey MF, Crowther JH, & Miller IW. (2011). Changes in Ecological Momentary Assessment Reported Affect Associated With Episodes of Nonsuicidal Self-Injury. Behavior Therapy, 42(4), 579–588. 10.1016/j.beth.2011.01.002 [DOI] [PubMed] [Google Scholar]
  5. Bresin K, & Gordon KH. (2013). Changes in negative affect following pain (vs. Nonpainful) stimulation in individuals with and without a history of nonsuicidal self-injury. Personality Disorders: Theory, Research, and Treatment, 4(1), 62–66. 10.1037/a0025736 [DOI] [PubMed] [Google Scholar]
  6. Caulfield NM, Karnick AT, & Capron DW. (2022). Exploring dissociation as a facilitator of suicide risk: A translational investigation using virtual reality. Journal of Affective Disorders, 297, 517–524. 10.1016/j.jad.2021.10.097 [DOI] [PubMed] [Google Scholar]
  7. Chapman AL, Gratz KL, & Brown MZ. (2006). Solving the puzzle of deliberate self-harm: The experiential avoidance model. Behaviour Research and Therapy, 44(3), 371–394. 10.1016/j.brat.2005.03.005 [DOI] [PubMed] [Google Scholar]
  8. Cipriano A, Cella S, & Cotrufo P. (2017). Nonsuicidal Self-injury: A Systematic Review. Frontiers in Psychology, 8, 1946. 10.3389/fpsyg.2017.01946 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Deng W, Hu D, Xu S, Liu X, Zhao J, Chen Q, Liu J, Zhang Z, Jiang W, Ma L, Hong X, Cheng S, Liu B, & Li X. (2019). The efficacy of virtual reality exposure therapy for PTSD symptoms: A systematic review and meta-analysis. Journal of Affective Disorders, 257, 698–709. 10.1016/j.jad.2019.07.086 [DOI] [PubMed] [Google Scholar]
  10. Fodor LA, Coteț CD, Cuijpers P, Szamoskozi Ștefan, David D, & Cristea IA. (2018). The effectiveness of virtual reality based interventions for symptoms of anxiety and depression: A meta-analysis. Scientific Reports, 8(1), 10323. 10.1038/s41598-018-28113-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Franklin JC, Ribeiro JD, Fox KR, Bentley KH, Kleiman EM, Huang X,Musacchio KM, Jaroszewski AC, Chang BP, & Nock MK. (2017). Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin, 143(2), 187–232. 10.1037/bul0000084 [DOI] [PubMed] [Google Scholar]
  12. Glenn CR, & Klonsky ED. (2010). The Role of seeing blood in non-suicidal self-injury. Journal of Clinical Psychology, 66(4), 466–473. 10.1002/jclp.20661 [DOI] [PubMed] [Google Scholar]
  13. Glenn CR, & Klonsky ED. (2011). One-Year Test-Retest Reliability of the Inventory of Statements about Self-Injury (ISAS). Assessment, 18(3), 375–378. 10.1177/1073191111411669 [DOI] [PubMed] [Google Scholar]
  14. Gradin Franzén A, & Gottzén L. (2011). The beauty of blood? Self-injury and ambivalence in an Internet community. Journal of Youth Studies, 14(3), 279–294. 10.1080/13676261.2010.533755 [DOI] [Google Scholar]
  15. Hasking P, Whitlock J, Voon D, & Rose A. (2016). A cognitive-emotional model of NSSI: using emotion regulation and cognitive processes to explain why people self-injure. Cognition and Emotion, 31(8), 1543–1556. 10.1080/02699931.2016.1241219 [DOI] [PubMed] [Google Scholar]
  16. Hooley JM, & Franklin JC. (2018). Why Do People Hurt Themselves? A New Conceptual Model of Nonsuicidal Self-Injury. Clinical Psychological Science, 6(3), 428–451. 10.1177/2167702617745641 [DOI] [Google Scholar]
  17. Houben M, Claes L, Vansteelandt K, Berens A, Sleuwaegen E, & Kuppens P. (2017). The emotion regulation function of nonsuicidal self-injury: A momentary assessment study in inpatients with borderline personality disorder features. Journal of Abnormal Psychology, 126(1), 89–95. 10.1037/abn0000229 [DOI] [PubMed] [Google Scholar]
  18. Huang X, Funsch KM, Park EC, Conway P, Franklin JC, & Ribeiro JD. (2021). Longitudinal studies support the safety and ethics of virtual reality suicide as a research method. Scientific Reports, 11(1), 9653. 10.1038/s41598-021-89152-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Klonsky ED, & Glenn CR. (2009). Assessing the Functions of Non-suicidal Self-injury: Psychometric Properties of the Inventory of Statements About Self-injury (ISAS). Journal of Psychopathology and Behavioral Assessment, 31(3), 215–219. 10.1007/s10862-008-9107-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kuehn KS, Dora J, Harned MS, Foster KT, Song F, Smith MR, & King KM. (2022). A meta-analysis on the affect regulation function of real-time self-injurious thoughts and behaviours. Nature Human Behaviour, 6(7), 964–974. 10.1038/s41562-022-01340-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Mook DG. (1983). In defense of external invalidity. American Psychologist, 38(4), 379–387. 10.1037/0003-066X.38.4.379 [DOI] [Google Scholar]
  22. Naoum J, Reitz S, Krause-Utz A, Kleindienst N, Willis F, Kuniss S, Baumgärtner U, Mancke F, Treede R-D, & Schmahl C. (2016). The role of seeing blood in non-suicidal self-injury in female patients with borderline personality disorder. Psychiatry Research, 246, 676–682. 10.1016/j.psychres.2016.10.066 [DOI] [PubMed] [Google Scholar]
  23. Selby EA, & Joiner TE. (2009). Cascades of Emotion: The Emergence of Borderline Personality Disorder from Emotional and Behavioral Dysregulation. Review of General Psychology, 13(3), 219–229. 10.1037/a0015687 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Stacy SE, Pepper CM, Clapp JD, & Reyna AH. (2022). The effects of blood in self-injurious cutting: Positive and negative affect regulation. Journal of Clinical Psychology, 78(5), 926–937. 10.1002/jclp.23267 [DOI] [PubMed] [Google Scholar]
  25. Störkel LM, Niedtfeld I, Schmahl C, & Hepp J. (2023). Does self-harm have the desired effect? Comparing non-suicidal self-injury to high-urge moments in an ambulatory assessment design. Behaviour Research and Therapy, 162, 104273. 10.1016/j.brat.2023.104273 [DOI] [PubMed] [Google Scholar]
  26. Taylor PJ, Jomar K, Dhingra K, Forrester R, Shahmalak U, & Dickson JM. (2018). A meta-analysis of the prevalence of different functions of non-suicidal self-injury. Journal of Affective Disorders, 227, 759–769. 10.1016/j.jad.2017.11.073 [DOI] [PubMed] [Google Scholar]
  27. Wang M, & Eaton NR. (2023). Linking non-suicidal self-injury to psychopathology: The utility of transdiagnostic and DSM-based models. Journal of affective disorders, 332, 55–63. 10.1016/j.jad.2023.03.075 [DOI] [PubMed] [Google Scholar]
  28. Watson D, Clark LA, & Tellegen A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. 10.1037/0022-3514.54.6.1063 [DOI] [PubMed] [Google Scholar]

RESOURCES