The Role of Affect Valence and Pain in Differentiating Young Women With and Without Histories of Nonsuicidal Self‐Injury

ABSTRACT

Introduction

Nonsuicidal self‐injury (NSSI) often helps regulate affect, yet there is conflicting research on the immediate affective outcomes of pain and NSSI. People also engage in NSSI for intrapersonal and interpersonal reasons. It is unclear whether affective shifts post‐pain task differentiate individuals with and without NSSI histories, and are influenced by interpersonal or intrapersonal motives for NSSI. The present study examined the effect of pain on affective shifts, and to investigate motives for NSSI as moderators.

Methods

Participants were 134 women (n = 77 with recent/recurrent NSSI), that completed measures of NSSI and reported their positive and negative affect pre‐ and post‐pain‐task.

Results

Our findings suggest that participants without a history of NSSI reported greater decreases in positive affect pre‐to‐post task. NSSI history did not predict fluctuations in negative affect after controlling for stress. Namely, at low levels of stress, participants reported greater decreases in negative affect following the pain task. Finally, interpersonal motives for NSSI predicted increased positive affect following the endurance of pain.

Conclusion

Findings suggest that affect changes in response to pain may not be differentially associated with NSSI history; however, among people with a history of NSSI, interpersonal motives may predict changes in positive affect following a pain task.

1. Introduction

Nonsuicidal self‐injury (NSSI) refers to “the deliberate destruction of one's own bodily tissue without suicidal intent and for reasons not socially sanctioned” (Bentley, Nock, and Barlow ²⁰¹⁴, 638). Prevalence of NSSI is elevated among adolescent and emerging adult samples (17.2% and 13.4%, respectively), relative to adult samples (5.5%; Swannell et al. ²⁰¹⁴). While NSSI is distinct from suicide attempts, it is a potent risk factor for suicide (Cipriano, Cella, and Cotrufo ²⁰¹⁷). Consequently, efforts to understand the processes that maintain NSSI over time, and thereby highlight potential intervention targets, are of critical public health importance.

One way to understand the maintenance of NSSI over time is to examine the functions it serves (i.e., reinforcement contingencies for the behavior). Most models of NSSI (e.g., Chapman, Gratz, and Brown ²⁰⁰⁶; Muehlenkamp et al. ²⁰¹³; Nock and Prinstein ²⁰⁰⁴) posit that behaviors are reinforced by reducing distressing emotions. Theories suggest that the physical pain associated with NSSI may relieve the distress that often prompts NSSI (Selby et al. 2019). For example, the emotional cascade model (Selby et al. 2009) suggests that physical pain serves to distract from emotional pain, negatively reinforcing the NSSI. Alternatively, the pain offset theory proposed by Franklin et al. (2013) posits that all individuals, regardless of NSSI history, experience decreased negative affect following pain offset. Indeed, people report positive emotions following the cessation of a pain task (Franklin et al. 2013). Similarly, the Benefits and Barriers Model (Hooley and Franklin 2018) emphasizes the positive reinforcement of NSSI through inducement of positive emotions. Research underscores the role that emotion regulation serves in reinforcing and maintaining NSSI. Taken together, literature suggests that physical pain and both positive and negative emotions are potential contributors to the maintenance of NSSI.

Given the theoretical literature noted above, and to better understand reinforcement processes in NSSI, studies have examined the temporal relations between pain and affect among individuals with NSSI histories by administering pain‐provoking stimuli. In general, studies found that administering pain preceded decreased self‐reported negative affect and arousal (Bresin and Gordon 2013; Russ et al. ¹⁹⁹²) and startle eyeblink reactivity (Franklin et al. 2013) for individuals regardless of NSSI history. These effects were stronger following negative emotion inductions, especially for shame and guilt (Bresin and Gordon 2013; Inbar et al. ²⁰¹³; Schoenleber, Berenbaum, and Motl ²⁰¹⁴). However, there are mixed findings within this literature on NSSI, pain, and emotions, warranting further investigation.

One remaining and important question is whether people with NSSI histories have unique patterns of emotional responses to pain. Most studies have found that individuals with histories of NSSI display greater reductions in negative arousal and negative affect in response to pain compared to controls (Russ et al. 1992; Weinberg and Klonsky ²⁰¹²; Bresin and Gordon ²⁰¹³; Franklin et al. ²⁰¹³). Franklin et al. (2013) reported increased positive affect (operationalized as increased startle postauricular activity) following pain; however, this was not contingent on NSSI history. Further, Schoenleber, Berenbaum, and Motl (2014) reported that only participants without a history of NSSI reported reductions in positive affect following a pain task. Taken together, these studies support the generalization of negative affect relief following pain, regardless of NSSI history. Notwithstanding, some studies have found greater reductions in negative affect in people with NSSI compared people without. Additionally, there is less consistent evidence for whether there are increases in positive affect following pain (e.g., Franklin et al. ²⁰¹³), or whether this is only the case for those with NSSI histories (e.g., Schoenleber, Berenbaum, and Motl ²⁰¹⁴). Numerous factors may contribute to these inconsistent findings, including methodological challenges such as small sample sizes or assessing participants with less versus more recent NSSI history, focusing exclusively on populations with BPD, and/or conflating positive affect with the absence or reversal of negative affect rather than considering it a separate construct. Therefore, additional research is needed to elucidate patterns of affect and pain in the context of NSSI, with attention to differences in affect valence.

It is also possible that the emotional consequences of pain for people with NSSI is influenced by their motives for NSSI. People report engaging in NSSI for a number of reasons, and these are commonly organized into intrapersonal motives (e.g., to regulate or escape unwanted emotions; or to feel something, even if it's pain) and interpersonal motives (e.g., to communicate to others; Klonsky and Glenn ²⁰⁰⁹). Namely, Klonsky and Glenn (2009) suggest that intrapersonal motives include reinforcement contingencies related to affect regulation, self‐punishment, and marking distress, while interpersonal motives include peer bonding, sensation seeking, and communicating interpersonal boundaries. It stands to reason that people endorse motives most related to their typical experiences and functions of NSSI. For example, those who endorse greater levels of the intrapersonal emotion regulation motive may be more likely to experience decreased negative affect when they engage in NSSI. In contrast, given that shame ¹ and social comparisons ² may be considered social processes; that shame in particular reduces after pain (Schoenleber, Berenbaum, and Motl ²⁰¹⁴); and that self‐criticism is associated with increased positive affect following a pain task (Fox et al. 2017), greater interpersonal motives for NSSI may be associated with greater positive affect and less negative affect post‐pain task. To date, few studies have investigated NSSI motives and the relation between their relative endorsement and the actual emotional consequences of NSSI.

Given the mixed literature regarding pain and affect in individuals with and without histories of NSSI, the purpose of the present study was twofold. First, we examined changes in positive and negative affect following a pain provoking task. Second, as an exploratory aim within the subsample of individuals with NSSI, we examined whether certain NSSI motives (e.g., intrapersonal or interpersonal) moderated affect changes following the NSSI pain task. We hypothesized that, consistent with prior literature, (1a) a pain‐provoking task would result in significantly reduced negative affect and increased positive affect for all participants, and (1b) that reductions in negative affect and increases in positive affect would be stronger in participants with histories of NSSI. Given the exploratory nature of our second aim and that extant literature suggests that both interpersonal and intrapersonal functions may be associated with differential patterns of changes in affect post‐pain, we had no a priori hypothesis regarding which NSSI motive(s) would be associated with changes in affect.

2. Methods

2.1. Participants

Participants (N = 134, M _age = 20.79 years, SD = 2.94) were college students recruited for a larger study examining the effects of peer feedback on emotions and pain responses among individuals with and without recent, recurrent NSSI. Eligibility for the larger study was as follows: (1) assigned female at birth, ³ (2) between 18 and 35 years of age, and (3) be enrolled in a college course in the past year. In addition, participants in the NSSI group (n = 77) must have experienced a minimum of two lifetime instances of NSSI and either NSSI urges or behaviors within the past year or reported past‐year thoughts/urges for NSSI but reported five or more instances of NSSI in their lifetime. Participants in the no‐NSSI group (n = 57) denied any history of NSSI. Participants could endorse all relevant racial and ethnic identities, and 70.9% identified as White, 18.7% as Asian or Southeast Asian, 6.7% as Black, 5.2% as Hispanic or Latino, 3.7% as multiracial, and 2.2% as Native American/American Indian. In terms of sexual orientation, 53.7% of participants reported their sexual orientation as straight, 23.9% as bisexual, 7.5% as lesbian/gay, 6.0% as pansexual, 3.7% as other, and 3.0% as asexual.

2.2. Procedures

All procedures were approved by an Institutional Review Board and all participants provided informed consent. Participants were recruited through a university research subject pool and via advertisements posted in the surrounding community. Following a phone screen and diagnostic interview to determine eligibility, participants attended a separate laboratory session during which they completed a battery of questionnaires before a pressure algometer task. During this task, a pressure algometer exerted a constant force on the finger between the first and second knuckle on participants' dominant hand, producing a slow building pain. Participants were randomly assigned to use their index or middle finger during the task, the order of which was counterbalanced across participants (Forgione and Barber 1971). This finger pressure algometer consisted of a 30‐cm‐long hinge connected to a 40 cm long × 8 cm wide base. The algometer was calibrated to apply a constant 2 kg of pressure at a 1 cm × 2 mm Lucite edge (Schoenleber, Berenbaum, and Motl ²⁰¹⁴). This pain task has been used in previous NSSI research (see Hooley et al. ²⁰¹⁰; Schoenleber, Berenbaum, and Motl ²⁰¹⁴). Consistent with past use of this task, participants indicated when they first began to experience pain by stating aloud the word “pain,” and to indicate “stop” to terminate the task when they were no longer willing to endure pain. If the participant had not terminated a trial after 10 min (Schoenleber, Berenbaum, and Motl ²⁰¹⁴), the assessor ended the task. Participants' pain threshold was the time between trial initiation and participant‐reported “pain” (Hooley et al. 2010). Participants' pain tolerance was calculated as the difference between their termination of the trial (their “stop” point) and their pain threshold (Hooley et al. 2010). Following the pain task, participants completed procedures for the larger study. Those who completed the study were compensated with course credits or monetary compensation ($35), according to their preference.

2.3. Measures

2.3.1. Positive and Negative Affect Schedule (PANAS)

Affect was assessed via the PANAS (Watson, Clark, and Tellegen ¹⁹⁸⁸) immediately before and after the pain task. Participants rated the extent to which they were currently experiencing a series of 20 affect states on a scale from 1 (very slightly or not at all) to 5 (extremely). This scale contains 10 positive adjectives (positive affect scale) and 10 negative adjectives (negative affect scale). Internal consistency for both affect scales were adequate immediately before the pain task (ωs = 0.69–0.84). However, internal consistency was variable immediately after the pain task, with adequate internal consistency for positive affect (ω = 0.83) and poor for negative affect (ω = 0.38). Given the low internal consistency of the negative affect composite, analyses were also completed for each negative affect item.

2.3.2. Inventory of Statements About Self‐Injury (ISAS)

NSSI characteristics and motives for NSSI were assessed using the Inventory of Statements About Self‐Injury (ISAS) (Klonsky and Glenn 2009). Motives were assessed by asking participants to rate the relevance of a number of statements that begin with, “When I self‐harm, I am…” from 0 (not relevant) to 2 (very relevant). These generate 13 subscales of motives, which are distilled into interpersonal and intrapersonal motives (Klonsky and Glenn 2009; Klonsky et al. ²⁰¹⁵). Internal consistencies for the subscales were adequate (ωs = 0.71–0.85).

2.3.3. Depression Anxiety Stress Scale (DASS‐21)

The Depression Anxiety Stress Scale (DASS)‐21 (Lovibond and Lovibond 1995) is a 21‐item measure of depression, anxiety, and stress symptoms. Participants were asked to respond to items (e.g., I found it difficult to relax) on a scale from 0 (did not apply to me at all) to 3 (applied to me very much or most of the time). Higher DASS‐21 Stress scores indicate greater sensitivity to levels of nonspecific arousal. Internal consistency for the present study was adequate (ω = 0.82).

2.4. Data Analytic Plan

We first examined data missingness, scatterplots, and the distributions of our primary study variables. For Aim 1 (i.e., that the pain task would result in reduced negative affect for all participants, and that membership in the NSSI group would enhance this effect), we conducted a Group (NSSI vs. no‐NSSI) × Time (pre‐ vs. post‐pain task) repeated measures analysis of covariance (RM‐ANCOVA) with the negative affect composite as the dependent variables. We also conducted a Group (NSSI vs. no‐NSSI) × Time (pre‐ vs. post‐pain task) RM‐ANCOVA with positive affect as the dependent variable to examine the effects of the pain task on positive emotions. Given the poor internal consistency of the PANAS negative affect composite, supplemental analyses were completed to examine differences across the 10 negative affect items. For our exploratory aim (i.e., that self‐reported NSSI motives would predict post‐pain affect shifts), we used multiple regression analyses to examine NSSI motives (i.e., intrapersonal and intrapersonal) to predict Time 2 (i.e., post‐task) affect scores, controlling for Time 1 (i.e., pre‐task) affect scores in participants with histories of NSSI. Time 1 affect was entered in Step 1 and ISAS motives were added in Step 2 for each model. Finally, we conducted sensitivity analyses with DASS‐21 Stress as a covariate in both the positive and negative affect models to examine whether the findings related to NSSI group status were better accounted for by stress.

3. Results

Table 1 presents correlations and descriptive statistics for variables of interest. Table 2 presents descriptive statistics and independent samples t‐tests for baseline DASS‐21 Stress, affect at Time 1 and 2 between NSSI groups.

TABLE 1.

Descriptive statistics and correlations.

Variables	1.	2.	3.	4.	5.	6.	7.
1. T1 PANAS‐PA	19.44 (5.58)
2. T2 PANAS‐PA	0.74***	19.22 (5.72)
3. T1 PANAS‐NA	0.72***	0.54***	14.31 (3.99)
4. T2 PANAS‐NA	0.67***	0.75***	0.69***	13.90 (3.00)
5. ISAS interpersonal a	0.06	0.23*	0.09	0.19	7.07 (5.79)
6. ISAS intrapersonal a	0.05	0.18	0.07	0.11	0.23	15.34 (5.16)
7. NSSI history b	−0.13	−0.05	0.04	−0.15	—	—	77 (57.5%)

Note: M (SD) or n (%) are on the diagonal. There are no correlations between NSSI history and NSSI motives as only participants with a history of NSSI provided motives.

Abbreviations: DSHI, Deliberate Self‐Harm Inventory; ISAS, Inventory of Statements About Self‐Injury; NA, negative affect; NSSI, nonsuicidal self‐injury; PA, positive affect; PANAS, Positive and Negative Affect Schedule; T1, Time 1; T2, Time 2.

^a Descriptive statistics and correlations are only presented for participants with a history of NSSI.

^b Variable is contrast coded such that participants without a history of NSSI are coded as 0 and participants with a history of NSSI are coded as 1.

^* p < 0.05.

^*** p < 0.001.

TABLE 2.

Affect differences pre‐ and post‐pain task between participants with and without a history of NSSI.

Variables	NSSI group (n = 77)	No NSSI group (n = 57)	t	Cohen's d	p
	M (SD)	M (SD)
DASS‐21 stress	8.39 (4.54)	5.16 (4.08)	−4.20	0.75	< 0.001
PANAS‐PA (T1)	19.21 (5.77)	20.47 (6.23)	1.21	0.21	0.227
PANAS‐PA (T2)	18.99 (5.84)	18.81 (6.00)	−0.17	0.03	0.862
PANAS‐NA (T1)	14.27 (3.96)	14.40 (3.93)	0.19	0.03	0.850
PANAS‐NA (T2)	13.60 (2.89)	14.19 (3.48)	1.08	0.19	0.281

Abbreviations: PA, positive affect; PANAS, Positive and Negative Affect Scale; NA, negative affect; NSSI, nonsuicidal self‐injury.

3.1. Aim 1: Examine the Effect of Pain Exposure on Reported Negative and Positive Affect (Full Sample)

Table 3 presents results from RM‐MANOVAs which revealed a significant effect of time (pre‐ vs. post‐pain) on positive affect, F(1,132) = 6.59, ${\eta }_p^2$ = 0.048, p = 0.011, such that participants' positive affect decreased following the pain task. Changes in positive affect did not vary as a function of NSSI history, F(1,132) = 0.31, ${\eta }_p^2$ = 0.002, p = 0.577. There were no significant interaction effects between time and NSSI history on positive affect, F(1,132) = 3.87, ${\eta }_p^2$ = 0.028, p = 0.051. However, sensitivity analyses revealed that after covarying for DASS‐21 Stress, there was a significant effect of DASS‐21 Stress on positive affect (p < 0.001; see Table 4). Moreover, there was a significant interaction effect between NSSI history and time (p = 0.036), such that participants that presented without a history of NSSI demonstrated steeper declines in positive affect following the pain task when controlling for baseline DASS‐21 Stress.

TABLE 3.

RM‐ANOVA examining the effects of NSSI history and time on affect.

Predictors	F (df)	${\eta }_p^2$	p
DV = positive affect
Intercept	1593.92 (1,132)	0.93	< 0.001
Time	6.59 (1,132)	0.05	0.011
NSSI history	0.31 (1,132)	< 0.01	0.577
Time × NSSI history	3.87 (1,132)	0.03	0.051
DV = negative affect
Intercept	2382.37 (1,132)	0.95	< 0.001
Time	3.62 (1,132)	0.03	0.059
NSSI history	0.39 (1,132)	< 0.01	0.531
Time × NSSI history	1.00 (1,132)	0.01	0.320

Note: Significant findings are bolded.

Abbreviation: NSSI, nonsuicidal self‐injury.

TABLE 4.

RM‐ANCOVA examining the effects of NSSI history and time on affect.

Predictors	F (df)	${\eta }_p^2$	p
DV = positive affect
Intercept	20,652.19 (1,127)	0.75	< 0.001
Time	1.00 (1,127)	0.01	0.319
NSSI history	4.81 (1,127)	0.04	0.030
DASS‐21 stress	13.71 (1,127)	0.10	< 0.001
Time × NSSI history	4.50 (1,127)	0.03	0.036
Time × DASS‐21 stress	0.09 (1,127)	< 0.01	0.759
DV = negative affect
Intercept	11,847.19 (1,127)	0.82	< 0.001
Time	0.79 (1,127)	0.01	0.375
NSSI history	4.37 (1,127)	0.03	0.039
DASS‐21 stress	11.29 (1,127)	0.08	0.001
Time × NSSI history	0.03 (1,127)	< 0.01	0.875
Time × DASS‐21 stress	5.34 (1,127)	0.04	0.022

Note: Significant findings are bolded.

Abbreviations: DASS‐21, Depression Anxiety and Stress Scale; NSSI, nonsuicidal self‐injury.

Contrary to our findings for positive affect, there was no significant effect of time on negative affect, F(1,132) = 3.62, ${\eta }_p^2$ = 0.027, p = 0.059. Moreover, negative affect did not vary as a function of NSSI history, F(1,132) = 0.39, ${\eta }_p^2$ = 0.003, p = 0.531, nor was there a significant interaction effect between time and NSSI history, F(1,132) = 1.00, ${\eta }_p^2$ = 0.007, p = 0.320. Table S1 presents supplemental analyses for each PANAS negative affect item. Results suggest that there were significant effects of time on high arousal negative affect items (e.g., upset, scared, jittery, and nervous; ps < 0.044), such that scores on these items declined following the pain task. However, given that participants with a history of NSSI reported greater baseline irritability (p = 0.029), there was a significant interaction effect between time and NSSI history (p = 0.026), such that participants with a history of NSSI demonstrated steeper declines in irritability pre‐ to post‐pain induction. There were no significant main or interaction effects of time on other forms of negative affect (e.g., distress, guilty, hostile, ashamed, and afraid; ps > 0.108).

Of note, covarying for DASS‐21 Stress, the pattern of effects were slightly different, such that NSSI history was associated with greater negative affect pre‐pain task after controlling for DASS‐21 Stress (p = 0.039; see Table 4). When dissecting the negative affect composite into single affective components, patterns of findings surprisingly did not reveal any significant effects of NSSI history (ps > 0.121) with regard to specific negative affect components (see Table S1). Complementary to the primary analyses, participants with lower DASS‐21 stress presented with greater negative affect before the pain task. There was a significant interaction effect between time and DASS‐21 Stress, such that participants that reported low levels of stress demonstrated steeper declines in feeling nervous, distressed, scared, hostile, and afraid following the pain task (ps < 0.041).

3.2. Aim 2: Explore the Effect of NSSI Motives on Post‐Pain Negative and Positive Affect (NSSI Group Only)

Table 5 presents results from hierarchical linear regression models with regressed positive and negative affect at Time 1 and ISAS motives onto Time 2 positive and negative affect. All omnibus regression models were significant in predicting positive and negative affect following the pain task at each step (Fs > 20.79, ps < 0.001). Time 1 negative affect predicted negative affect at Time 2 in Steps 1 and 2 (bs > 0.48, ps < 0.001). In Step 2, interpersonal and intrapersonal motives for NSSI did not predict negative affect at Time 2 (bs < 0.06, ps > 0.190). Predictors accounted for 48.6% of the variance in negative affect at Time 2 and neither motive significantly accounted for additional variance in the model (ΔR ² = 0.016, p = 0.356).

TABLE 5.

Regression models of NSSI motives predicting affect post‐pain task.

Predictors	R 2	B	SE	t	p
DV = PANAS PA (T2)
Step 1	0.585	F(1,68) = 95.76, p < 0.001
Intercept		4.29	1.59	2.69	0.009
PANAS PA (T1)		0.77	0.08	9.79	< 0.001
Step 2	0.629	F(3,66) = 37.23, p < 0.001
Intercept		1.60	1.99	0.81	0.423
PANAS PA (T1)		0.76	0.08	9.97	< 0.001
ISAS interpersonal functions		0.17	0.08	2.12	0.038
ISAS intrapersonal functions		0.12	0.09	1.28	0.206
DV = PANAS NA (T2)
Step 1	0.470	F(1,68) = 60.23, p < 0.001
Intercept		6.57	0.96	6.83	< 0.001
PANAS NA (T1)		0.50	0.06	7.76	< 0.001
Step 2	0.486	F(3,66) = 20.80, p < 0.001
Intercept		6.69	1.20	5.01	< 0.001
PANAS NA (T1)		0.49	0.06	7.60	< 0.001
ISAS interpersonal functions		0.06	0.05	1.30	0.200
ISAS intrapersonal functions		0.02	0.05	0.33	0.740

Note: Significant findings are bolded after Benjamini–Hochberg correction.

Abbreviations: DV, dependent variable; ISAS, Inventory of Statements About Self‐Injury; NA, negative affect; PA, positive affect; PANAS, Positive and Negative Affect Scale; T1, Time 1; T2, Time 2.

In the positive affect model, positive affect at Time 1 predicted Time 2 positive affect in Steps 1 and 2 (bs > 0.75, ps < 0.001). In Step 2, ISAS interpersonal motives significantly predicted positive affect at Time 2 (b = 0.17, p = 0.038), but ISAS intrapersonal motives did not (b = 0.12, p = 0.206). Predictors accounted for 62.9% of the variance in positive affect in Time 2 and ISAS interpersonal motives accounted for a significantly more variance in positive affect at Time 2 (ΔR ² = 0.044, p = 0.025).

4. Discussion

The present study aimed to replicate and extend existing research on the effects of pain on affect among those with and without NSSI histories and explore the role of NSSI motives in predicting post‐pain affect. In line with past research (e.g., Franklin et al. ²⁰¹³), we predicted that a standardized painful task in the laboratory would be associated with decreased negative affect across all participants, regardless of NSSI history. However, this finding is qualified by one's level of stress, such that participants that reported lower baseline stress demonstrated greater decreases in negative affect over time. We further hypothesized that this effect would be amplified in those with NSSI histories (vs. without). Furthermore, we extended past research by examining whether post‐pain affect was associated with specific NSSI motives. Our findings provided additional evidence for the effect of pain on affect, and preliminary support for the potential for unique affective consequences of pain depending on specific NSSI motives.

Consistent with past empirical research (Franklin et al. 2013), we predicted that the pressure algometer pain task would result in decreased negative affect across all participants, regardless of NSSI history. Of note, we did not detect any effect of pain exposure on overall negative affect in our full sample, counter to past meta‐analytic evidence (Bresin, Kling, and Verona ²⁰¹⁸). This may due to our reliance on self‐reported measures of negative affect (vs. postauricular reactivity; Franklin et al. ²⁰¹³), or because of the poor internal consistency within the post‐pain self‐report affect measure. Accordingly, when we examined specific negative emotions post‐pain task, we found that predominantly high arousal negative emotions such as feeling afraid, hostile, scared, distressed and nervous significantly declined follow the pain induction in participants regardless of NSSI history, which is indeed consistent with findings from Franklin et al. (2013). Taken together our findings suggest that in terms of negative affect, that shifts in negative affect are not differentially associated with NSSI history, but rather elevations in nonspecific arousal.

The pain task was associated with shifts in reported positive affect as well. Specifically, the pain task was associated with decreased reported positive affect for participants without a history of NSSI. This is inconsistent with work suggesting that after pain, people may exhibit increased positive emotions, regardless of NSSI history (Franklin et al. 2013). Yet this should be considered in the context of our study procedures. This task occurred at the beginning of the session and participants may simply have been dreading the remainder of the study, for instance. These findings also partially align with past work demonstrating a decrease in self‐reported positive affect following an NSSI pain task (Bresin et al. 2010). Given that positive affect is less studied overall than negative affect, and often less consistently measured across studies, the literature on positive affect and NSSI is varied and contradictory. People may experience habituation to pain and an erosion of barriers to NSSI with repeated acts of this behavior (e.g., Franklin et al. ²⁰¹³; Gordon et al. ²⁰¹⁰; Hooley and Franklin ²⁰¹⁸). This suggests that for people who do not have a history of NSSI, that decreases in positive affect may be a barrier to the initiation of NSSI behavior.

Results provided mixed support regarding the role of specific NSSI motives on affect post‐pain task. In contrast with our expectations, intrapersonal motives did not predict affective shifts. It may be that intrapersonal motives (such as affect regulation) are so commonly endorsed (see Klonsky and Glenn ²⁰⁰⁹; Nock and Prinstein ²⁰⁰⁴) that they do not work well as a means of separating people and predicting differential experiences. Future studies should consider other methods of examining motives for NSSI to determine if they may have an impact on the experience of pain during and post‐NSSI.

Contrary to our findings with intrapersonal motives, interpersonal motives predicted greater positive affect post‐pain task among those with a history of NSSI. It is worth mentioning here that the context of the laboratory study was social in nature, with an experimenter providing guidance throughout the paradigm. Thus, people who are particularly motivated by social support (including possibly those individuals who report engaging in NSSI to influence or communicate with others) may have found participating in this study to be an especially positive experience. Previous work has found that self‐criticism, a trait that is perhaps most conceptually related to intrapersonally‐oriented self‐punishment motives, is related to increased positive mood during a pain task, in both NSSI samples (Fox et al. 2017) and controls (Fox et al. 2019). Yet self‐criticism at its core may also be considered to be a somewhat interpersonal phenomenon as well, given that self‐criticism and shame are closely associated with perceived social rank and reassurance‐seeking, including in samples with high rates of NSSI (Gilbert et al. 2010). Thus, although unexpected, it is possible that those individuals who are reporting elevated interpersonal motives are an overlapping sample with those reporting elevated self‐criticism. Of note, the current study only examined post‐task mood ratings, so it is likely that immediately reinforcing positive mood shifts may in fact be underestimated in the present study. Given Fox et al. (2019)'s findings regarding the differences between mood ratings during and post‐task, it does appear that mood changes are fleeting and timing is important in capturing precise emotional experiences over the course of an NSSI act. Therefore, future studies should use multiple ratings throughout the duration of an NSSI pain task to further examine patterns of affective shifts, which can facilitate our understanding of factors that contribute to NSSI engagement and persistence over time.

Internal consistency for the PANAS negative affect scale demonstrated adequate internal consistency before the pain task but scattered negative affect responses following the pain task. While not the primary focus of our paper, this unexpected finding warrants discussion. Previous literature suggests that anxiety symptoms decreased in clients with BPD and NSSI following a cold pressor pain test (Russ et al. 1992). Consistent with this previous finding negative affect decreased following an eyeblink test following a painful shock in participants with NSSI histories, but after controlling for emotional reactivity this difference disappeared (Franklin et al. 2013). The shifts in specific negative affect experiences in our study (e.g., feeling nervous, jittery, upset, and scared) are common affective phenotypes in people with anxiety. Thus, it is possible that pain may have unique effects on anticipatory, higher arousal negative affect. Consequently, these types of negative affect responses may not correlate with other lower arousal negative affect experiences (e.g., feeling guilty or ashamed) that are oppositional or less affected by painful stimuli, resulting in poor internal consistency. Alternatively, it is possible that pain leads to changes in the ability to differentiate among affective states, or emotion differentiation. Indeed, the dynamic model of affect proposes that stress and pain can lead to inverse relationships between positive and negative affect systems (Davis, Zautra, and Smith ²⁰⁰⁴), and it is possible that this theory could be expanded to suggest that pain may lead to greater ability to distinguish between affect components more generally. Given that poor emotion differentiation is associated with greater NSSI (Zaki et al. 2013), it is possible that the increased emotional granularity precipitated by pain may be another method by which NSSI is reinforced for some individuals. Future research is needed in this area.

Several limitations warrant consideration. First, we used a fixed‐order within‐person manipulation to examine baseline versus post‐pain affective experiences. Thus, it is impossible to tease apart the effects of time in the study versus pain. Second, we did not examine the effect of pain in the context of negative emotions. Given that NSSI often occurs following a time of stress, this lab‐based pain paradigm may not have been as ecologically valid. Third, our measurement of emotion was limited. Despite its validity, the PANAS (Watson, Clark, and Tellegen ¹⁹⁸⁸) does not assess positive emotions that may be associated with NSSI (e.g., “calm,” “relieved,” and “satisfied”). Further, internal consistencies for the negative affect scale were low. Analyses were therefore also conducted on the items separately, which evidenced disparate results between high‐ versus low‐arousal negative emotions. While we did conduct sensitivity analyses controlling for stress, our patterns of findings may be different when controlling for other factors related to negative affect (e.g., emotion regulation, emotional reactivity, and interpersonal difficulties) or specific diagnostic presentations (e.g., borderline personality disorder and major depression). We also relied on self‐reported emotions. Given that past self‐report research has shown decreased post‐task positive emotion (Bresin et al. 2010), whereas psychophysiological indices revealed increased post‐task positive emotion (Franklin et al. 2013), it may be important to adopt multi‐method approaches (Franklin et al. 2010). Future studies should incorporate such an approach to directly compare results using different methods of measurement. Fourth, although we compared people with a presence of NSSI history to those without, it is possible that the increased severity of psychopathology in general may have driven this finding. This concern is somewhat addressed by the robustness of the pattern of effects to controlling for general stress levels, however. Finally, we focused on motives as a person‐level moderator, whereas it may be that within‐person contextual influences are more important in driving responses to pain. Future studies may benefit from multi‐method and intensive longitudinal assessment to improve our understanding of the plurality of experiences of NSSI, pain, and emotion.

5. Conclusions

These limitations notwithstanding, this study makes important contributions to our understanding of the emotional consequences of pain in the context of NSSI. This study extends extant work by focusing on a sample of individuals with recent, recurrent NSSI. This innovation is critical as individuals with recent NSSI differ from those with distal NSSI behaviors (e.g., Whitlock et al. ²⁰¹⁵), and evidence suggests that any pain processing differences between those with vs. without an NSSI history start to diminish after NSSI cessation (Ludäscher et al. 2009). Results demonstrate that pain reduces positive emotions, yet this may not be a relevant contingency for people prone to NSSI who do not generally have high positive affect. Moreover, people who engage in NSSI for interpersonal motives may face the unique contingencies of greater post‐pain positive affect to grapple with. Thus, consideration of the emotional context of pain is an important avenue for future research and intervention.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.