The effect of collective trauma on craving for cannabis and tobacco

Abstract

Exposure to collective traumatic events is associated with increased substance use. We tested if experimentally priming thoughts of collective trauma increases substance craving, while examining the potential anxiety-buffering roles of self-affirmation, connection to a national/cultural group, and attachment security (mainly attachment anxiety). In two pre-registered experiments, priming collective traumatic events (the October-7th terror attack in Israel) increased cannabis craving for moderate-high risk problematic cannabis users (Study 1), and tobacco craving among daily smokers (Study 2). Attachment anxiety predicted heightened craving in both studies but did not moderate the priming effects. Self-affirmation and group identification had no impact. October-7 prime increased death thought accessibility in Study 1 but did not mediate the prime effect on craving. Findings highlight the causal effects of reminders of collective trauma on craving for cannabis and tobacco. We discuss the studies’ limitations and possible implications for understanding substance use as a defense against existential threats.

Introduction

Research has shown that collective existential threats and exposure to traumatic events can lead to various mental health problems and problematic substance use (e.g. DiMaggio et al., 2009; Eliashar et al., 2024; Lifshin et al., 2026). According to Terror Management Theory (TMT; Greenberg et al., 1986), these events remind people of their biological vulnerability and fragility and motivate them to defend themselves from this terrifying awareness by first suppressing the threatening thoughts from consciousness (i.e. “proximal defense”), and then by attempting to gain or sustain a compensatory sense of psychological security (i.e. distal defense; e.g. Arndt et al., 1997; Pyszczynski et al., 1999). Smoking a cigarette or a joint may help temporarily remove fears and death-related thoughts from consciousness (e.g. Arndt et al., 2013; Nagar and Rabinovitz, 2015). Thus, TMT may particularly help explain the psychodynamic mechanisms by which people turn to tobacco or cannabis as a self-medication in response to collective trauma. TMT also suggests that core anxiety buffers – attachment security, self-esteem, cultural worldviews – may help people feel protected from existential threats and that disruption of these buffers might heighten death-related concerns and intensify defensive responses. However, there is no clear indication from prior research that these buffers or their disruption moderate the effects of death reminders on substance craving. In the current study we wanted to experimentally test the effects of priming reminders of a collective trauma on craving for cannabis and tobacco, while considering the potential anxiety-buffering roles of self-esteem validation (e.g. Schmeichel and Martens, 2005), connection to a national group/cultural worldview, and attachment security (confidence in one’s lovability and that support would be available when needed). Understanding the causal relationship between collective trauma and substance craving, as well as the potential moderators involved, can advance the scientific existential social psychological study of addictive behaviors.

Terror management theory (Greenberg et al., 1986; Solomon et al., 2015) asserts that people rely on three basic death-anxiety buffers: investment in cultural worldviews, self-esteem, and attachment security (e.g. Hart et al., 2005). Cultural worldviews, such as religious, national and political identities providing people with order, meaning, value and, importantly, hopes for literal and symbolic death transcendence. By living up to the standards of cultural worldviews people obtain and maintain self-esteem, and feel accepted, loved and deserving of the protection and immortality that their cultural worldviews offer (e.g. Solomon et al., 2015). In addition, interactions with supportive others in times of need contribute to the formation of a sense of attachment security, which is a building block of resilience, well-being, and mental health, and can reduce existential concerns by fortifying feelings of social connectedness and belonging (e.g. Mikulincer, 2019).

Empirical research in the framework of TMT has demonstrated that experimentally priming death-related thoughts (mortality salience) increase attempts to validate cultural worldviews (e.g. Arndt et al., 1997), to invest in self-esteem strivings (e.g. Taubman-Ben-Ari et al., 1999), and to seek proximity from security providers (e.g. Taubman-Ben-Ari et al., 2002). There is also evidence that strengthening anxiety buffers by either enhancing/affirming self-esteem, priming mental representations of attachment security, or validating cultural worldviews tends to reduce death-related concerns and to inhibit activation of other defenses following death reminders (e.g. Arndt et al., 1997; Greenberg et al., 1992; Harmon-Jones et al., 1997; Mikulincer and Florian, 2000). And, importantly, threats to the validity of anxiety buffers (e.g. worldview threat, self-relevant negative feedback, rejection/separation primes) increase the availability of death-related thoughts to consciousness (e.g. Hayes et al., 2008; Mikulincer et al., 2002; Schimel et al., 2007).

Studies in TMT have also demonstrated that substances like tobacco, cannabis, or alcohol can serve as death-anxiety buffers. Laboratory and field experiments found that after being reminded of death, people are more likely to consume alcoholic rather than non-alcoholic beverages (e.g. Ein-Dor et al., 2014; Wisman et al., 2015), to show increased tobacco and cannabis craving (e.g. Arndt et al., 2013; Nagar and Rabinovitz, 2015) and to defend cannabis related cultural worldviews (Hayes and Rafferty, 2020). Cannabis consumption also reduced death-related thoughts after mortality salience (Nagar and Rabinovitz, 2015). Although these different substances may have different physiological effects and anxiety buffering capacities, they all allow to “escape” from self-awareness after thinking about death (e.g. Wisman et al., 2015).

Research also shows consistent associations between disruption of specific components of the anxiety-buffer system and addictive behavior. For example, disruption of the sense of attachment security in the form of attachment anxiety (i.e. worries about one’s lovability) has been consistently associated with increased substance abuse (e.g. Fairbairn et al., 2018; Schindler, 2019) and behavioral addictions (e.g. D’Arienzo et al., 2019; Niu et al., 2023). Several studies have also found that threats to one’s ingroup-identity and cultural trauma are associated with substance abuse (e.g. Kaya et al., 2016). Lifshin et al. (2026) found that Israeli soldiers exposed to life threatening combat situations during the Swords of Iron war, attachment anxiety and holding oppositional attitudes to the dominant political worldviews predicted more post-traumatic stress symptoms, which, in turn, resulted in heightened problematic substance use and behavioral addictions. Wisman et al. (2015) found that death primes (vs control primes) increased alcohol consumption in a nightclub only among participants reporting low self-esteem. However, the relationship between self-esteem and substance abuse is less consistent in the literature (e.g. Schroeder et al., 1993; Uba et al., 2013; Vasquez et al., 2011).

Collective trauma refers to psychological reactions to a traumatic event that affects an entire society. In contrast to personal trauma, memory of collective trauma can persist beyond the time and space in which the events occurred (e.g. Hirschberger, 2018). Research indicates that reminders of collective trauma can increase death-related concerns and elicit similar terror management responses (e.g. Landau et al., 2004; Pyszczynski et al., 2003; Wohl et al., 2010). Non-experimental evidence indicates that collective traumatic events are associated with increased substance use. Studies show that October-7 attack in Israel (e.g. Mashiach and Davidovitch, 2024) was associated with increased population levels of post-traumatic stress (e.g. Levi-Belz et al., 2024; Refaeli et al., 2025) and substance use (e.g. Eliashar et al., 2024; Feingold et al., 2024).

However, although studies indicate that collective trauma increases DTA (e.g. Landau et al., 2004), and that reminders of death can increase substance craving, as far as we know, no study has experimentally tested the effect of these reminders (collective-trauma prime) on substance craving or use. Some studies unrelated to TMT have found that reminders of personal trauma can elicit craving for substances, especially among people with post-traumatic stress disorder (e.g. Coffey et al., 2010; Romero-Sanchiz et al., 2022; for review see e.g. DeGrace et al., 2022). But these studies did not test the possible effects of collective trauma reminders in the general population. We are also unaware of experiments testing the moderating role of anxiety buffers or their disruption on substance craving or use after a collective-trauma prime. Thus, research is needed to establish evidence of the effects of collective trauma and anxiety buffers on substance craving and use.

In the current research, we aimed to experimentally test the hypotheses that priming thoughts of a collective trauma increases craving for cannabis and tobacco among frequent users, while considering the potential moderating roles of self-esteem validation, connection to a national group/cultural worldview, and attachment security. We focused on the collective trauma of the October-7 2023 attack in Israel, and conducted two studies in 2025 among Israeli Jews. We included measures of attachment orientations (anxiety, avoidance) and importance of Israeli national identity, and randomly assigned participants to read about the October-7 attack or a control topic (dental pain). We also tested if a self-affirmation (e.g. Sherman and Cohen, 2006) manipulation (vs control) would provide people with psychological security and moderate the hypothesized effects of the October-7 prime on craving. Our outcome measures were craving for smoking cannabis in Study 1 and for smoking tobacco in Study 2. Importantly, we assessed the level of problematic cannabis use or tobacco cigarette dependency, as we expected the effects of collective-trauma reminders on craving to occur especially among those who report higher levels of problematic use or dependency. Finally, we also examined whether the effect of the collective-trauma prime on craving is mediated by increased death thought accessibility (DTA), such that higher DTA after the October-7 prime would be associated with higher craving.

Transparency and openness

All hypotheses, method and statistical analyses were pre-registered (Study 1: Appendix 1 in the Supplemental File; Study 2: https://aspredicted.org/rsw3-kfnp.pdf). Materials, data and code are available at: https://doi.org/10.17605/OSF.IO/Q9GJY.

Study 1

Study 1 was conducted to test the effects of reminders of the October-7 attack (vs a control condition) on DTA and cannabis craving. We also considered the moderating role of problematic cannabis use using a standardized measure (Humeniuk et al., 2010), as well as self-affirmation (e.g. Schmeichel and Martens, 2005; Sherman and Cohen, 2006), attachment orientations (anxiety, avoidance), and importance of Israeli national identity (i.e. connection to their culture).

Our predictions were: (1) The October-7 prime (vs the control prime) would increase cannabis craving especially among heavier users who are at higher risk for problematic use; (2) this effect would be weaker after self-affirmation (vs control condition); (3) the effect of the October-7 prime would be stronger among participants with higher attachment anxiety and lower importance of Israeli identity (disconnection from their cultural worldview); (4) Although DTA might not always mediate terror management responses, as measuring it might bring thoughts of death back to consciousness (e.g. Hayes and Schimel, 2018), we tested if DTA may mediate the effect described in Prediction 1, such that participants with higher levels of DTA after the October-7 prime would report more cannabis craving. Beyond these predictions, we also examined political view, religiosity, age and gender as additional potential covariates.

Method

Participants

This research was approved by the Institutional Review Board (IRB) at Reichman University, protocol number: P_2025010. All participants signed informed consents. All materials were anonymized and did not included identifiable data.

Participants were adult Jewish Israelis (ages 21–60) recruited online through a paid panel (iPanel) for monetary compensation. Following our preregistration (Supplemental Appendix 1), we removed responses from individuals who did not regularly smoke cannabis, had missing data, non-serious responses to the article manipulation checks or affirmation and control condition, or other technical difficulties. Responses from 249 participants were analyzed (134 women and 115 men; M age = 35.97, SD age = 7.83).

An a-priori power analysis using G*Power (Faul et al., 2007) indicated that detecting a small-medium effect size (f² = 0.05) at 80% power in a hierarchical regression with 5 tested predictors and 9 total predictors (including covariates) would require 263 participants. A sensitivity analysis indicated that our sample of N = 249 was sufficient at detecting the targeted effect size even in the most inclusive study design (f² = 0.05).

Procedure and materials

After signing informed consents, participants completed a screening question for frequency of cannabis use in the past 3 months (Humeniuk et al., 2010). Only participants who reported using cannabis at least 1–3 times a month were allowed to proceed. ¹ Participants then completed questions about their age and gender, Jewish ethnicity, religiosity (1 = secular, 7 = religious), importance of national identity (1 = not important at all, 7 = very important), and political view (1 = right wing, 4 = center, 7 = left wing). Attachment orientations were assessed using the 12-item Experiences in Close Relationships Scale (Lafontaine et al., 2016) tapping attachment anxiety (6 items, α = 0.83), and attachment avoidance (6 items, α = 0.81).

Participants were then randomly assigned to read an article about the October-7 attack (n = 126) or about a control topic – dental pain (n = 123). The October-7 article summarized the events and aftermath of the attack and included three recognizable pictures. The control article discussed dental pain, matched on number of words and pictures. We included two short open-ended questions about the article and the emotions it aroused to ensure that participants had read it and to strengthen the priming effect. This was followed by a short delay and distraction task about leisure activities, in keeping with past TMT research (e.g. Greenberg et al., 1994).

Participants were then randomly assigned to a self-affirmation task (n = 117) or control task (n = 132). In the self-affirmation task, following past research (e.g. Schmeichel and Martens, 2005; Sherman and Cohen, 2006), participants were first asked to rank 11 listed values from most important to least important. Then they were asked to write about the value that they ranked as most important to themselves and to describe a time when this value was personally important to them and allowed them to feel good about themselves. Participants in the control task condition were asked to rank values that may be important to the average person and explain why the value that they ranked as number 5 could be important for the average person.

Death thought accessibility was then measured using a word-completion task (Greenberg et al., 1994) previously used among Israelis (e.g. Mikulincer and Florian, 2000). The task included 20 Hebrew word fragments that participants were asked to complete with the first word that came to mind by filling in one missing letter. Nine of the 20 Hebrew fragments could be completed with either neutral or death-related Hebrew words. The possible death-related words were words for death, mourning, cadaver, grave, murder, skeleton, body, killing, dying. The dependent measure was the sum of death-related words completed.

Craving for cannabis was measured using 7 items of the Marijuana Craving Questionnaire (Heishman et al., 2009). Items tapped participants’ current craving for cannabis (e.g. “I need to smoke Marijuana right now”; 1 = strongly disagree, 7 = strongly agree; α = 0.91). A mean craving score was computed with higher scores indicating higher cannabis craving.

Self-reported problematic substance use during the past 3 months was measured with the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST 3.1; Humeniuk et al., 2010). Questions related to the frequency of non-medically use, craving, and maladaptive consequences of cannabis. Scores were computed by summing up the items in each of the substances in accordance with the scale protocol. Considering studies suggesting a higher cutoff of the ASSIST moderate-risk category (8 instead of 4; Davis et al., 2009) we computed the categories as: low risk = 1–7 (n = 73); moderate risk = 8–26 (n = 156); and high risk = 27–39 (n = 20). However, to better differentiate between the moderate risk participants, and also include enough people in the high-risk category, we lowered the cutoff between the moderate-risk and the high-risk category (moderate risk = 8–20, n = 116; high risk = 21–39, n = 60). This allowed us to have a larger number of higher risk users (n = 60, approximately 25% of the sample).

At the end of the survey participants answered questions about their attention and experience. They were then debriefed about the purpose of the study.

Results

Preliminary analyses

To ensure that random assignment into experimental conditions worked, we conducted a series of preliminary analyses. Confirming this, a series of 2 × 2 ANOVAs for priming condition (October-7, dental pain) and self-affirmation task (yes, no) indicated no differences in participants’ age, religiosity, importance of national identity, political view, attachment anxiety or avoidance, and ASSIST cannabis scores, all Fs < 1.97, ps > 0.162. There were also no significant differences between the four conditions in the modified ASSIST cannabis categories scores, χ²s < 1.02, ps > 0.313, and gender, χ²s < 3.34, ps > 0.067.

Although the relationship between gender and priming condition was marginally statistically significant, p = 0.067, with more women (n = 75) than men (n = 51) in the October-7 prime condition, gender was not related to craving or DTA (rs < 0.06, ps > 0.363) and including it in the analyses did not affect the results.

Zero-order associations

Means, SDs, and Pearson correlations are presented in Supplemental Table S1 (Appendix 2). These correlations indicated that craving was significantly associated only with attachment anxiety, r = 0.22, p < 0.001 and the continuous ASSIST cannabis score, r = 0.43, p < 0.001, all other rs < 0.09, ps > 0.203. ASSIST modified categories were also associated with craving, F(2, 246) = 28.13, p < 0.001, η² = .186 (all categories differed at p < 0.002).

As expected, the October-7 prime condition increased DTA compared to the control condition, F(1, 245) = 3.90, p = .050, ηp² = 0.016, as participants after the October-7 condition revealed higher levels of DTA (M = 1.48, SD = 1.26) than after the control condition (M = 1.19, SD = 1.04). There was no significant effect for self-affirmation or the interaction, Fs < 0.12, ps > 0.736. However, DTA was not significantly associated with craving or any other measure rs < 0.08, ps > 0.247. Notably, the self-affirmation condition did not significantly relate to any outcome. ²

Moderation analyses

To test for the interactive effects of the October-7 priming and ASSIST cannabis categories on craving, we conducted a hierarchical regression analysis in PROCESS (Hayes, 2022, model 1). Following our preregistration (Supplemental Appendix 1), we included only statistically significant predictors in the model predicting craving. Initial analyses indicated that self-affirmation and national identification were not related to craving or moderate the effect of the October-7 prime. Thus the priming condition (October-7 = 1, dental pain = 0) and the ASSIST categories (low-risk = 1, moderate risk = 2, high-risk = 3; defined as a sequential categorical variable) were entered as predictors, attachment anxiety was entered as a covariate (mean centered; initial analyses indicated no interactions), and cannabis craving as the outcome.

The total regression model was statistically significant, F(6,242) = 12.70, p < 0.001, R² = 0.24. Attachment anxiety was a statistically significant covariate, b = 0.20, t = 3.04, p = 0.003. ASSIST categories, for the contrast of low risk versus high and moderate risk, statistically significantly predicted craving, b = 1.26, t = 4.12, p < 0.001. Importantly, there was a two-way interaction between the priming condition and ASSIST scores contrasting low-risk and moderate risk with high-risk, b = 1.20, t = 2.69, p = 0.008 (Figure 1). Other effects were not statistically significant, ts < 0.74, ps > 0.459.

Figure 1.

The interactive effect of collective trauma prime (October-7 vs dental pain) and the modified ASSIST cannabis categories (low-risk, moderate-risk, moderate-high-risk) on cannabis craving in study 1 (N = 249).

As shown in Figure 1, the October-7 trauma prime increased cannabis craving for participants in the (modified) high-risk category, b = 0.76, t = 2.09, p = 0.038, as participants primed with the October-7 attack reported higher craving (M = 4.70, SE = 0.36) than those in the dental pain control condition (M = 3.94, SE = 0.36). However, the October-7 prime did not increase craving compared to the control condition among moderate-risk users (M = 3.20 vs M = 3.72, SE = 0.26), t = 1.69, p = 0.093, or among low-risk users (M = 2.53 vs M = 2.46, SE = 0.33), t = 0.21, p = 0.836.

Mediation Analysis

Tests for the possible mediating role of DTA were conducted in PROCESS (Hayes, 2022, models 8 and 15). The priming condition and the ASSIST categories were entered as predictors and DTA (mean centered) as the mediator. Indirect effects were estimated with bootstrapping using 10,000 resamples. These analyses did not produce any statistically significant indirect effects on cannabis craving through DTA (all 95% Confidence Intervals included zero).

Exploratory analysis

We also conducted exploratory analyses with DTA as a predictor variable, considering that increased DTA may have acted as a double death prime and reintroduced the need for craving. The details are reported in Appendix 3. In brief, there was a statistically significant DTA × ASSIST binary × attachment anxiety interaction, in which among high-moderate risk users who are also high in attachment anxiety, higher DTA (+1 SD) predicted more craving compared to low levels of DTA. However, there was no difference in craving as a function of DTA among participants low in attachment anxiety or among low/moderate risk cannabis users (Supplemental Figure S1).

Discussion

The results of this study provided partial support for our hypothesis. Among participants in the modified-high risk for problematic cannabis use, the October-7 prime (vs the dental pain prime) increased cannabis craving. Attachment anxiety was also related to heightened craving as predicted. However, counter to the pre-registered hypothesis, self-affirmation did not affect the results, and there were no moderation effects by attachment anxiety or national identity. Although the October-7 prime condition increased DTA, it did not mediate the interactive effect of the collective trauma prime and problematic cannabis use on cannabis craving. Nevertheless, exploratory analyses indicated that high levels of DTA were related to more craving among high-attachment anxiety participants at moderate-high risk for problematic cannabis use.

These findings suggest that reminders of collective trauma can increase cannabis craving among moderate-high risk users. Attachment anxiety was also related to craving as an additive but not moderating factor of the collective-trauma prime effect. The fact that this effect was not moderated by anxiety buffers may be attributed to the relatively low number of participants in the moderate-high risk use category, which limited the statistical power for testing higher order interactions within this subsample. The null findings regarding the moderation effects of self-affirmation can also be interpreted as unsuccessful manipulation, perhaps due to the online environment that participants were in. Alternatively, findings that no anxiety buffer moderated the collective-trauma prime effect may suggest that craving for cannabis is more of a proximal rather than distal terror management response, and therefore other distal defenses such as attachment security, self-esteem or connection to the cultural worldview might not moderate it. Although we did find that high DTA did predict craving for high attachment-anxious, moderate-high risk users, it did not directly mediate the effect of the collective-trauma prime. This may also suggest that the mediation effect did not occur because it brought thoughts of death back to consciousness (e.g. Hayes and Schimel, 2018), initiating another proximal defense.

Study 2

Study 2 was conducted to conceptually replicate Study 1′s findings, this time focusing on craving for tobacco cigarettes (nicotine). We switched to tobacco from cannabis considering that in Study 1, the effects of the October-7 prime occurred only among moderate-high risk users, and that these users are relatively harder to sample. A larger sample of daily cigarette smokers provided us with more statistical power to test for the potential moderating roles of self-affirmation, attachment anxiety, connection to cultural worldview, and self-esteem. Considering that DTA did not mediate the effect in Study 1, we changed the design so that craving would be treated as a proximal terror management defense. To do this we measured craving immediately after the October-7, without a delay.

As in Study 1, we predicted that the October-7 prime (vs control) would increase craving, and that self-affirmation (vs control), attachment anxiety, self-esteem, and importance of national identity may moderate the effect. We also tested if the effect is moderated by cigarette dependency scores.

Method

Participants

The study was approved by the Institutional Review Board (IRB) at Reichman University, protocol number: P_2025010. All participants signed informed consents. All materials were anonymized and did not include identifiable data.

Participants were adult Jewish Israelis (ages 18–60) who participated through Israeli Midgam Panel for monetary compensation. As detailed in our preregistration, we excluded participants who did not smoke cigarettes daily, who reported smoking electronic cigarettes, or who had procedural errors. Responses from 281 participants were analyzed (143 women and 137 men and 1 non-binary; M age = 39.02, SD age = 10.78). ³

An a-priori power analysis indicated that detecting a small-medium effect size (f² = 0.05) at 80% power in a hierarchical regression analysis with 6 tested predictors and 9 total predictors would require 263 participants. A sensitivity analysis indicated that our sample of N = 281 was sufficient at detecting the targeted effect size even in the most inclusive study design (f² = 0.05).

Procedure and materials

The procedure and materials were similar to Study 1 with several exceptions. Participants completed two screening questions regarding daily tobacco cigarette or electronic cigarettes smoking (Do you smoke tobacco cigarettes daily?, Yes/ No; Do you smoke electronic cigarettes daily?, Yes/ No), years of smoking, and the demographic questions, the importance of national identity question, and the ECR-12 scale (α = 0.82 for anxiety, α = 0.80 for avoidance) described in Study 1. We also included a single item measure of self-esteem (“I have high self-esteem”; 1 = Not very true of me, 9 = Very true of me; Robins et al., 2001).

The self-affirmation and control tasks as well as the October-7 and dental-pain primes were identical to those described in Study 1. Participants were first randomly assigned to either the self-affirmation task (n = 145) or control task (n = 136). After this manipulation, participants in each self-affirmation condition (yes, no) were randomly assigned to read either an article about the October-7 attack (n = 146) or about dental pain (n = 135), as in Study 1.

Tobacco craving was measured immediately after the articles with the 10-item brief questionnaire on smoking urges (Cox et al., 2001). Scores were measured on a 0–100 sliding scale. A mean craving score was computed with higher scores indicating higher tobacco craving (α = 0.94). Finally, we included the 5-item self-reported Cigarette Dependency Scale (CDS-5; Etter et al., 2003). Scores were computed in accordance with the scale protocol (α = 0.82).

Results

Preliminary analyses

A series of 2 × 2 ANOVAs for priming (October-7, dental pain) and self-affirmation (yes, no) indicated no differences in participants’ age, religiosity, importance of national identity, political view, self-esteem, years of smoking, and cigarette dependency, all Fs < 2.78, ps > 0.097. Chi-Square tests indicated that there was also no difference between the four experimental conditions in gender distribution, χ²s < 2.70, ps > 0.257. However, a significant main effect for priming condition was found on attachment anxiety, F(1, 277) = 4.76, p = 0.030, ηp² = 0.017. The main effect for self-affirmation on attachment anxiety was also significant, F(1, 277) = 5.44, p = 0.020, ηp² = .019. There were also marginally statistically significant differences between the October-7 prime and dental-pain prime in importance of national identity, F(1, 277) = 3.59, p = 0.059, ηp² = 0.013, and religiosity, F(1, 277) = 3.43, p = 0.065, ηp² = 0.012. We therefore included these variables as moderators and covariates in the main analyses (only statistically significant covariates were included as preregistered).

Zero-order associations

Means, SDs, and Pearson correlations are presented in Supplemental Table S2 (Appendix 4). Craving related to cigarette dependency, r = 0.44, p < 0.001, and attachment anxiety, r = 0.44, p < 0.001, as well as to political view, r = −0.18, p = 0.003, and religiosity, r = −0.13, p = 0.021, but not to other variables, |rs| < 0.11, ps > 0.057.

Moderation analyses

Initial hierarchical regression analyses indicated that only the October-7 prime condition, attachment anxiety, and cigarette dependency significantly predicted craving. As in Study 1, the self-affirmation manipulation had no significant effect and was not included in the model. We therefore conducted a hierarchical regression with the prime condition (October-7 vs control) and the mean centered attachment anxiety and cigarette dependency scores.

The first regression step, was statistically significantly, F(3,277) = 32.32, p < 0.001, R² = 0.259, with October-7 prime (vs dental pain), b = 9.30, t = 3.35, p < 0.001, attachment anxiety, b = 4.03, t = 3.89, p < 0.001, and cigarette dependency, b = 11.57, t = 7.91, p < 0.001, predicting heightened craving for tobacco. The second step was also statistically significant, ΔF(3,274) = 2.79, p = 0.041, ΔR² = 0.022. A PROCESS follow up yielded a statistically significant attachment anxiety × cigarette dependency interaction, b = 2.82, t = 2.76, p = 0.006, in which attachment anxiety predicted craving more strongly when cigarette dependency was high (+1 SD), b = 15.35, t = 7.61, p < 0.001, than when cigarette dependency was low (−1 SD), b = 7.90, t = 3.97, p < 0.001. No other interactions were significant, ts < 1.07, ps > 0.286. All main effects remained statistically significant, ts > 2.37, ps < 0.018. Figure 2 depicts the means of craving at the different levels of October-7 versus control prime, attachment anxiety and cigarette dependency.

Figure 2.

Main effects of collective trauma prime (October-7 vs Dental Pain), attachment anxiety (high vs low) and cigarette dependency (high vs low) on tobacco craving in study 2 (N = 281).

Discussion

The results indicated that there were additive effects for the October-7 prime, attachment anxiety and cigarette dependency on craving for tobacco. As in Study 1, the effect of the October-7 prime was not moderated by any anxiety buffer, including the self-affirmation manipulation, attachment anxiety, self-esteem, national identification, or cigarette dependency. Only attachment anxiety was moderated by cigarette dependency, as high attachment anxiety predicted craving especially among participants with high dependancy. These findings suggest that collective-trauma prime can increase motivation for tobacco use, as a proximal defense, which might not be moderated by distal terror management defenses.

General discussion

The current research provided an experimental test of the hypothesis that memories of collective trauma can increase craving for cannabis and tobacco. The results showed that priming thoughts of the October-7 terror attack increased craving for cannabis among moderate-high risk cannabis users (Study 1) and for tobacco among daily cigarette smokers (Study 2). These effects were not moderated by self-affirmation manipulation. In line with our theory and past research, attachment anxiety predicted craving in both Studies but did not moderate the collective-trauma prime effects. In Study 1, October-7 prime also increased DTA, but it did not mediate the effect of collective-trauma prime on cannabis craving. Nevertheless, exploratory analyses indicated that DTA predicted more craving among high attachment-anxious moderate-high risk cannabis users.

At the macro level, these findings support an existential social psychological model for substance use and possibly addiction. They suggest that, among high-risk users, craving for substances may be activated as a defensive response to collective trauma and other existential threats. These findings are in line with both TMT and the self-medication model of substance use (e.g. DeGrace et al., 2022; Khantzian, 1997). The findings that attachment anxiety affected craving are also consistent with past research (e.g. Fairbairn et al., 2018; Schindler, 2019).

At a more micro level, from a TMT perspective, anxiety buffers (e.g. self-esteem, attachment security, ingroup identification) did not moderate the collective-trauma prime effects on craving for cannabis or tobacco. This finding suggests that craving in response to collective trauma may operate more as a proximal defense, aimed at immediately removing the (death-related) threat from conscious awareness rather than a more distal response aimed at gaining a sense of security and immortality – senses offered by attachment security, cultural worldviews, or self-esteem (for a review see e.g. Kosloff et al., 2019). Future experimental studies could test this assumption by comparing craving responses measured immediately after exposure to collective existential threat (vs control) or after a delay period. Nevertheless, although both tobacco smoking and cannabis smoking may serve as proximal regulatory function, (as shown in Study 2 and in previous research, e.g. Arndt et al., 2013), possibly taking attention away from the threat and providing immediate relief (e.g. Wisman et al., 2015), cannabis smoking can, at least in some cultures, also serve a more “distal defense,” which requires a delay (e.g. Hayes and Rafferty, 2020; Study 1).

The current studies have several important limitations to consider. First, despite having enough statistical power in Study 1, because the effects were found only among participants who are at moderate-high risk for problematic cannabis (n = 60), we might not have had enough power to detect higher order interactions in the relevant subsample. Hence we should be careful in interpreting the lack of moderation, at least in Study 1. A larger sample in Study 1 may have provided more power to detect even a small mediation effect by DTA. However, the DTA measure in Study 1 might have been problematic as it introduces a double death prime (e.g. Hayes and Schimel, 2018), which may hinder the ability to clearly detect the effect of collective-trauma prime on craving. For this reason, we did not include it in Study 2. Nonetheless, more research is needed to understand the role of DTA in the link between collective trauma and substance craving.

The self-affirmation manipulation did not work in our studies. This might have been due to technical issues and the online nature of the study. Perhaps if it was properly manipulated (e.g. in a physical setting), it might have had a stronger effect. Future research may also attempt to manipulate a self-esteem boost or attachment security prime to test their moderation effects. Although we did measure attachment anxiety (in both studies) and self-esteem (in Study 2), experimental manipulations of these constructs are needed to fully understand their moderation capacity. In the current studies, we also did not find any significant effect for importance of national identity. This might also relate to the fact that it was measured using only a 1-item scale, and perhaps also to the specific nature of the collective trauma prime. Because the October-7 trauma might also undermine the perception of Israel as a powerful and secure state (e.g. Levi-Belz et al., 2024), having a strong Israeli identity might have not provided psychological security in this case. More research is needed to explore this possibility, using better measures of connection to the national group and cultural worldview (e.g. Alfasi and Besser, 2024; Roccas et al., 2006). Finally, because this study addresses the highly specific and intense events of the Israel-Hamas war, caution is needed regarding generalizability of these findings to other types of collective trauma, and across different populations. Future research may investigate if other collective traumatic events, such as economic crises or natural disasters, might also increase substance craving among regular users.

Despite these limitations, this research may have several important implications for future research on addiction. The finding that craving for substances such as tobacco and cannabis was causally affected by a reminder of collective trauma is a clear testament to the important role that collective trauma and other existential threats might have on addiction. As such, more research should be directed to understanding this effect and the ways in which it may be moderated. Although attachment anxiety did not moderate the effects of trauma priming on substance craving, it did have a large additive effect. Perhaps strengthening attachment security may prove to be a promising route for future research.

Considering that, at least in Study 2, the effects of the collective-trauma prime on craving occurred without delay, as a proximal defense, future research may also investigate the possible mediating role of conscious emotions, like fear, in mediating the observed effect (e.g. Coffey et al., 2010; DeGrace et al., 2022; Tull et al., 2013). There may also be important differences between tobacco and cannabis smoking. For instance, depending on cultural norms, cannabis smoking may constitute a more distal worldview defense than tobacco (e.g. Hayes and Rafferty, 2020). Future research may also test the role that consumption of different substances may play in managing collective trauma and existential concerns, as was done with mortality salience primes and cannabis smoking (Nagar and Rabinovitz, 2015). Indeed, consumption of cannabis or alcohol may have more long-term effects than tobacco smoking. Finally, studies may test if these effects can be generalized to other substances (e.g. opiates) and behavioral addictions (e.g. gambling, pornography consumption) to further understand their causes (e.g. Lifshin et al., 2026). Our hope is that the current research may inspire such efforts.

Supplemental Material

sj-docx-1-hpq-10.1177_13591053261450563 – Supplemental material for The effect of collective trauma on craving for cannabis and tobacco

Supplemental material, sj-docx-1-hpq-10.1177_13591053261450563 for The effect of collective trauma on craving for cannabis and tobacco by Vera Skvirsky, Uri Lifshin, Maayan Yacubovitz, Hila Avnit, Tslil Dvir, Maayan Marcus, Maya Regev, Rotem Flor and Mario Mikulincer in Journal of Health Psychology