Trauma film viewing and intrusive memories: Relationship between salivary alpha amylase, endocannabinoids, and cortisol

The endogenous cannabinoid (ECB) system is a small molecule lipid signalling system that is involved in stress response activation and is associated with PTSD, but it is unclear whether salivary ECBs are part of the sympathetic nervous system response to stress. We conducted an adapted trauma film paradigm, where participants completed a cold pressor test (or control) while watching a 10-minute trauma film. We also collected saliva and hair samples and tested them for ECBs, cortisol, and salivary alpha amylase (sAA). As hypothesised, there were significant positive correlations between sAA activity and salivary ECB levels, particularly 2-arachidonoyl glycerol (2-AG), though ECBs were not correlated with sAA stress reactivity. Participants who had a significant cortisol response to the trauma film/stressor reported less intrusive memories, which were also less distressing and less vivid. This effect was moderated by arachidonoyl ethanolamide (AEA), where decreases in AEA post-stress were associated with more intrusive memories in cortisol non-responders only. This study provides new evidence for the role of ECBs in the sympathetic nervous system.


Introduction
The endogenous cannabinoid (ECB) system is a small molecule lipid signalling system that is known to regulate the function of central and peripheral nervous systems involved in behaviour, emotion, learning, and memory (Kano et al., 2009;Ligresti et al., 2016).The ECB system is composed of cannabinoid receptorspresent throughout both the central and peripheral nervous systems (Howlett et al., 2002) as well as discrete ligand molecules including arachidonoyl ethanolamide (AEA; Devane et al., 1992) and 2-arachidonoyl glycerol (2-AG; Sugiura et al., 1995), which bind to the cannabinoid type 1 receptor, and N-acyl ethanolamides such as oleoylethanolamide (OEA) and palmitoylethanolamide (PEA) which do not bind to the cannabinoid 1 receptor but exert indirect effects on the ECB system (Ligresti et al., 2016).
Measuring stress reactivity and intrusive memory symptomology concurrently is an ideal way to understand how disorders such as PTSD occur (Ney et al., 2019;Pitman et al., 2012).This is because higher stress responses of the sympathetic and central nervous systems are known to predict higher frequency of negative emotional and intrusive memories (Nicholson et al., 2014;Roozendaal and McGaugh, 2011;Schultebraucks et al., 2019), and activity within these systems tend to be a biomarker of PTSD symptomology long-term (Yehuda, 2009).Trauma film paradigms provide an experimental method for simulating trauma and subsequent intrusive memory development in healthy humans and has led to innovative new treatments for PTSD (Holmes and Bourne, 2008;Iyadurai et al., 2018); however, viewing a trauma film by itself does not reliably increase biological stress reactivity (Otte et al., 2016).Simulating the biological conditions that lead to PTSD in an experimental setting will allow better quality diagnostic and treatment tools to be developed.Recent research has suggested that increased frequency of intrusive memories of the film content are reported by participants who undergo a stress task prior to trauma film viewing (Hilberdink et al., 2022), although one study recently suggested that this only occurs if there is a corresponding increase in cortisol following stress (Schultebraucks et al., 2019).This finding aligns with our understanding that negative and intrusive memories are consolidated as a result of cortisol reactivity following stress (Roozendaal and McGaugh, 2011) -if participants do not have a cortisol response after film viewing, they will not necessarily report intrusive memories (Chou et al., 2014).Integrating stress induction to the trauma film paradigm should allow more precise investigation of the role of stress hormones in the development of intrusive memories following analogue trauma.
To date, research has focused on the relationship between ECBs circulating in the bloodstream and human stress and memory.We recently showed that ECBs are reliably quantifiable in saliva (Ney et al., 2020), and that salivary ECBs show acute stress reactivity (Ney et al., 2021;Ney et al., 2023).In the latter study, salivary ECBs were not correlated with blood ECB levels and showed immediate increases following stress (Ney et al., 2021), suggesting that salivary ECBs do not originate from the blood stream and are instead part of the sympathetic nervous system response to stress and are likely synthesised locally within the saliva glands 'on demand', which is a known characteristic of the broader ECB system (Kano et al., 2009;Ligresti et al., 2016;Ohno-Shosaku et al., 2001).Salivary ECBs, metabolic enzymes, and cannabinoid receptors are known to be present in the parotid and submandibular glands (Busch et al., 2004;Dall'Aglio et al., 2010;Matias et al., 2012), and it has been shown that blockade of cannabinoid receptors in the submandibular gland results in inhibition of norepinephrine-stimulated saliva excretion (Andreis et al., 2022;Prestifilippo et al., 2006Prestifilippo et al., , 2013)).However, direct correlation between the salivary ECB and sympathetic nervous system responses to stress have not been tested.Similarly, it is not known whether salivary ECBs are markers of wider human physiological functioning and whether their measurement can be useful for understanding human behaviour, memory, and psychopathology.Measurement of salivary ECBs may present a cheaper and less-invasive way to understand the ECB system.
In the current study, we conducted an adapted trauma film paradigm, where participants completed a cold-pressor task (or control) while watching a 10-minute trauma film.Participants completed the stress task simultaneously while watching the film (rather than before or after) to emulate a real-world stressor, where stress induction occurs at the same time as the traumatic event.Saliva samples were collected immediately before, after, and 25-minutes post trauma film viewing and were measured for ECBs AEA, 2-AG, OEA, and PEA, as well as stress markers cortisol, progesterone, and salivary alpha amylase (sAA), which is a validated marker for the norephinepherine response to stress (Nater and Rohleder, 2009).We also measured hair samples for AEA, 2-AG, OEA, PEA, cortisol, and progesterone as a retrospective measure of these circulating analytes in participants.We predicted that ECB responses to stress would be correlated with sAA activity and that ECB levels would be associated with the frequency of intrusive memories of the trauma film reported by participants.We hypothesised that our stress manipulation would induce a stronger stress response relative to control and that participants in the stress group would report significantly more intrusive memories compared to the control group.We also planned to test whether any group effects were moderated by cortisol responsiveness, given recent evidence that intrusive memories only increase following stress if there is an associated cortisol stress response (Schultebraucks et al., 2019).

Participants
There were 74 participants who volunteered to participate in the experiment (57 females, M age = 23.041,SD age = 7.82).Participants were either first-year undergraduate students who earned course credit for participating or were recruited via a community board and earned AUD $25 for completing the experiment and a further AUD$20 for completing all four diary entries.The study was approved by the Queensland University of Technology (QUT) Human Research Ethics Committee (approval number 4298) and experiments conducted in accordance with the National Statement on Ethical Conduct in Human Research.Written informed consent was obtained from participants.

Questionnaires.
To provide a subjective measure of distress associated with the experiment, participants rated the trauma film and also completed the Subjective Units of Distress Scale (SUDS; Wolpe, 1969) at three timepoints in the experiment.The trauma film rating was a 9-point Likert scale where participants rated how unpleasant ("0 = most pleasant" to "9 = most unpleasant") and intense ("0 = least intense" to "9 = most intense") the film clips were and how aroused ("0 = very calm" to "9 = highest arousal/excitement") the participant felt while watching the film clips.The SUDS is a single response on a scale ranging from "0 = totally relaxed" to "100 = highest distress/fear/anxiety/discomfort that you have ever felt".To assess individual differences that might affect the formation of intrusive memories, there were three Likert scale questionnaires in the experiment: the state-trait anxiety inventory (STAI; Spielberger et al., 1983), the depression, anxiety and stress scales 21 (DASS-21; Lovibond and Lovibond, 2018), and the intolerance of uncertainty scale (IUS-12;Carleton et al., 2007).The STAI has 40 items and assesses trait and state anxiety, the DASS-21 has 21 items and assesses depression, anxiety and stress levels over the past week, and the IUS-12 is the shortened, 12-item version of the full 27-item questionnaire, which assesses intolerance of uncertainty.For the STAI, ratings range from "1 = not at all" to "4 = very much so".For the DASS-21, ratings range from "0 = did not apply to me at all" to 3 = "applied to me very much, or most of the time".For the IUS-12, ratings range from "1 = not at all characteristic of me" to "5 = entirely characteristic of me".The questionnaires were included to account for individual differences that might affect the formation of intrusive memories.Finally, participants completed an intrusive memory questionnaire during the experiment and on the four evenings following the experiment.This survey consists of three questions, which assess the number of intrusive memories relating to the film that the participant experienced, the content of the intrusive memories (i.e., which film clip the memory related to), and how vivid and distressing the intrusive memories were on average, via a 5-item Likert scale ranging from "Not at all" to "Extremely".

Trauma Film.
Intrusive memories can be reliably induced by using a variety of film clips of vehicular accidents and interpersonal violence (James et al., 2016).Sexual trauma has been used in some previous studies (Schultebraucks et al., 2019) but was not included as the trauma film in the current study was designed to only be distressing enough to induce intrusions that would subside over a week (Lau-Zhu et al., 2018).Hence, the 10-minute trauma film in the current study consisted of seven clips depicting interpersonal and accidental trauma.Three clips depicted interpersonal trauma from commercial movies: "Hostel" (2005), directed by Eli Roth (duration: 2 min 05 sec); "Irreversible" (2002), directed by Gaspar Noé (duration: 1 min 16 sec); and "Drive" (2011), directed by Nicolas Winding Refn (duration: 1 min 55 sec)."Hostel" and "Irreversible" have been used in previous studies (Franke et al., 2021; M.P. O'Donohue et al.Herzog et al., 2022;Wegerer et al., 2013).Two clips depicted accidental trauma from public awareness advertisements: "Cow" (2009), a clip directed by Peter Watkins-Hughes depicting a car accident caused by texting while driving (duration: 1 min 10 sec); and the clip "kitchen safety" (2007) from Workplace Safety and Insurance Board (htt ps://www.youtube.com/watch?v=-t_fUg0i9sI, duration: 0 min 25 sec), which was also used by Dibbets et al. (2018); Dibbets et al., (2018).The other two clips were news media clips covering a mugging incident (https://www.youtube.com/watch?v=YogZxzGXdrs, duration: 0 min 36 sec) and an animal attack (�https://www.youtube.com/watch?v=lkuxmBv4scM, duration: 1 min 02 sec).The clips were arranged in the same order for all participants.The trauma film was presented via a desktop computer and headphones.

Intrusion Provocation Task
The Intrusive Provocation Task (IPT) provokes and assesses intrusive memories during the experiment by presenting blurred stills of scenes from the trauma film (Lang et al., 2009;Malik et al., 2014).Each still was a moment preceding the most intense scene in the clip.Using Microsoft PowerPoint, one still from each film clip was displayed for 2 seconds, with a 3 second interval between each scene where the screen was black.After all stills had been presented, a black screen was presented for 2 minutes, during which participants could think freely, followed by a thank you message.Throughout the task, participants completed a tally of how many intrusive memories were experienced and after the task they completed the intrusive memory diary with reference to the IPT.

Biological Samples.
Salivary samples were collected at three timepoints in the experiment.These were collected using the passive drool method (1 mL) and were placed in a − 20 • C freezer within 1.5 hours of collection and kept frozen until analysis.These samples were then analysed via liquid chromatography-mass spectrometry using previously described methods 35 at QUT Proteomics facility, which is equipped with a Sciex QTRAP 6500.Hair samples were optional for participants and were collected at the end of the experiment.Three centimetres, reflecting approximately 3 months' worth of hair growth, was collected from the scalp by the experimenter and stored at room temperature until analysis.Analysis was conducted at QUT Proteomics facility using the Sciex QTRAP 6500 and with a method adapted from Ney, Felmingham, et al. (2021).Both hair and saliva methods used stable isotope dilution and were used to measure AEA, 2-AG, OEA, PEA, and cortisol concentrations.Peaks were recorded in Analyst (Sciex) and scored in Skyline daily 21.1.9.34.Values for salivary analytes are expressed in picograms per millilitre of saliva (pg/mL), whereas hair analytes are expressed in picograms per milligram of hair (pg/mg).
Salivary alpha amylase (sAA) enzymic activity was analysed using a kinetic enzyme immunoassay (1-1902, Salimetrics) according to the manufacturer's instructions.Samples were analysed using a Varioskan LUX plate reader at 405 nm.sAA activity was expressed in units per millilitre (U/mL).A laboratory dilution error resulted in loss of 33 data points that precluded sAA analysis in the correlation tests (i.e., due to inconsistent activity levels between samples), leaving data from 63 participants for these analyses (i.e., 33 data points across 11 participants were incorrect, meaning all data points for 11 participants).However, since only the absolute levels of sAA activity was incorrect in these 33 datapoints, it was still possible to categorise participants as sAA responders or non-responders (see Statistical Analyses below).

Procedure
Participants completed a Qualtrics survey prior to the experiment, which excluded those who were pregnant or had pathological diagnoses, neurological, cardiac, epileptic conditions, and/or daily or almost daily exposures to violent TV shows, movies, and video games.All participants completed the experiment between 12 pm and 5 pm, to control for diurnal variation of cortisol and sAA (De Lacerda et al., 1973;Nater et al., 2007).At the beginning of the experiment, the first salivary sample and the first SUDS rating were collected.Then, the cold pressor test and trauma film paradigm commenced (for a similar method, see Hilberdink et al., 2022).This manipulation was intended to increase participant stress levels in the stress group via integration of the cold-pressor test into the trauma film viewing.Participants were randomly allocated to the control or experimental (stress) group.In the cold pressor test, participants submerged their right hand up to and including their wrist in either room temperature water (control group) or 0-4 • C ice water (stress group) at four intervals while the film was paused: before the film (for 45 seconds), after the first three clips (for 30 seconds), after the fourth clip (for 1 minute), and after the fifth and sixth clips (for 45 seconds).Variability of the time that participants had submerged in the water was intended to reduce the predictability of the procedure, as in Smeets et al. (2012), where variability of ice water timing was used to increase the stress level of participants.After removing their hand from the water, participants were allowed to quickly dry their hand before the experimenter continued the film.
After the trauma film, the second salivary sample and the second SUDS rating were collected.Then, participants completed the trauma film valence rating, followed by the questionnaires.After this, the experimenter defined intrusive memories for the participant and explained the diary procedure.Participants then completed the IPT, followed by the intrusive memory questionnaire.Finally, the third salivary sample and the third SUDS rating were collected, and a hair sample was obtained from consenting participants.To assess intrusive memories on the four evenings following the experiment, participants were sent an online version of the intrusive memory questionnaire via an automated email and text message.

Statistical Analyses
Demographics were analysed between groups using one-way ANOVAs and chi-square tests of independence.Univariate and repeated measures ANOVAs were conducted to compare subjective ratings, biological responses, and intrusive memory frequency, vividness, and distress between stress and control groups.Bivariate correlations were conducted to assess the associations between the biological outcome measures.While tests were conducted comparing each specific timepoint (e.g., sAA time 1 with AEA time 1), changes in pre-to poststress induction were also compared (e.g., sAA time 2-time 1 with AEA time 2-time 1).In this way, we were able to test the relationship between endocannabinoid and sAA stress reactivity in addition to the general relationship between these biomarkers.For all analyses, biomarkers were square root transformed to improve normality.
Two multivariate analyses of variance (MANOVA) were used to test the associations between biological measures and intrusive memories.Specifically, in model 1, the average number of reported intrusive memories (square root transformed), the average vividness rating of intrusive memories, and the average distress rating of intrusive memories were the dependent variables, whereas Condition (stress versus control), Cortisol Responder and sAA Responder were added as the predictor variables.Cortisol and sAA responding was determined as a minimum 15.5% increase in cortisol or sAA level from pre-to post-stress (Miller et al., 2013).In model 2, the dependent variables were the same as model 1, though the predictor variables were Condition (stress versus control), Cortisol Responder, change in AEA level post-stress, and change in 2-AG level post-stress.The change in AEA and 2-AG levels were calculated as the difference between time 1 and time 2 AEA and 2-AG levels.Intrusive memory frequency was square root transformed to improve normality.All analyses were performed in SPSS v27 for Windows.

Biological responses to stress and film viewing
Two (Condition: Stress, Control) × three (Time: baseline, post, 20min post) repeated measures ANOVAs were conducted separately for each biological marker.There was no significant effect of Time for AEA: = 3.99, p =.021, η p 2 =.06.These significant effects reflected decreases in 2-AG, OEA, and PEA after baseline, indicating no effect of stress, and an increase in cortisol concentrations 20-min post stress, which indicates a cortisol stress response (see Fig. 1).Higher SUDS ratings 20-min poststress were associated with lower levels of AEA: r(68) = − .29,p =.018 and 2-AG: r(68) = − .30,p =.013, though no other significant effects between SUDS ratings and their corresponding salivary biomarkers were found, and there were no significant relationships between change in AEA and 2-AG pre-to post-stress and change in SUDS ratings pre-to post-stress (all ps >.1).sAA enzymic activity was not significantly increased following the stressor, F(2114) = 2.24, p =.111, η p 2 =.04, though there was a significant, positive relationship between pre-to post-stress SUDS rating and pre-to post-stress sAA activity r(58) =.39, p =.003, such that higher sAA activity following the stressor was associated with a higher SUDS distress rating.There were no Condition × Time interactions for AEA: F( 2136

Intrusive Memory Data
Of the 74 participants who completed the experiment, there were 72, 68, 66, and 68 participants who completed the first, second, third, and fourth diary entries, respectively.There were 62 participants (83.78%) who completed all four diary entries.A significant effect of Time was observed on the number of intrusive memories (see Fig. 3a): F(4,57) = 39.33,p <.001, η p 2 =.73, with the number of intrusive memories decreasing over time.There was not a significant effect of Condition: M diff =.03, F(1,60) =.004, p =.95, η p 2 <.001, and there was not a significant interaction between Condition and Time: F(4,57) = 1.12, p =.36, η p 2 =.07.See Fig. 3b for a distribution of the prevalence, across time, of each type of intrusive memory related to the film; the "fire hydrant" scene from the movie "Hostel" (2005) was the most prevalent memory for both groups.
A significant effect of Time was observed on the vividness of intrusive memories (see Fig. 4a): F(4,57) = 41.07,p <.001, η p 2 =.74, with vividness decreasing over time.There was not a significant effect of Condition: M diff =.15, F(1,60) =.66, p =.42, η p 2 =.01, and there was not a significant interaction between Condition and Time: F(4,57) = 1.15, p =.34, η p 2 =.08.A significant effect of Time was observed on distress from intrusive memories (see Fig. 4b): F(4,57) = 30.03,p <.001, η p 2 =.68, with distress decreasing over time.There was also a significant interaction between Condition and Time: F(4,57) = 2.57, p =.047, η p 2 =.15.This effect was mainly driven by the effect of Time, with post-hoc simple effect tests revealing only a trending difference between the Conditions at Day 2 (p =.088), with no other potential differences between the Conditions in reported distress at any other time point.There was not a significant effect of Condition: M diff =.21, F(1,60) = 1.39, p =.24, η p 2 =.02.In summary, there were the expected effects of Time on SUDS ratings and on the prevalence, distress, and vividness of intrusive memories.However, there were no effects of Condition on any of these variables.

Association Between Biological Data and Intrusive Memories
Previous studies have shown that experimental stressor effects will only elicit more intrusive memories if there is an associated biological stress response (Schultebraucks et al., 2019).Therefore, we classified participants as Responders or Non-Responders for cortisol and sAA.While there were slightly more cortisol responders in the stress condition (49% of stress group were responders compared to 31% of control participants), this difference did not reach significance, χ 2 (df=1, N=70) = 2.14, p =.143.There was no difference between sAA responders across stress and control conditions (50% versus 53% respectively, p =.814).
A multivariate ANOVA was conducted for the combined in-lab and at-home reports of intrusive memory frequency, intrusive memory distress, and intrusive memory vividness.Independent variables included Group, Cortisol Responder, and sAA Responder, as well as their interactions.The multivariate results revealed a significant main effect of Cortisol Responder, F(3,56) = 5.78, p =.002, η p 2 =.24, though no other effects were significant (lowest p value =.165).Within this significant multivariate test (Fig. 5), Cortisol Responders had significantly lower distress to intrusive memories (p =.001, Fig. 5a), lower vividness of intrusive memories (p =.037, Fig. 5b) and lower total number of reported intrusive memories (p =.005, Fig. 5c).Notably, the Condition x Cortisol Responder interaction was not significant (p =.762), suggesting that Cortisol Response determined these intrusive memory outcomes regardless of whether participants were assigned to the stress or control groups.
Following from this analysis, we were interested in whether these effects were moderated by changes in AEA and 2-AG concentrations following stress induction, given that endocannabinoids (AEA and 2-AG specifically) are known to modulate the HPA response to stress.We ran a further multivariate ANOVA with the combined in-lab and at-home reports of intrusive memory frequency, intrusive memory distress, and intrusive memory vividness as dependent variables, and Cortisol Responder, Group, change in AEA concentration following stress and change in 2-AG concentration following stress, as well as their interactions, as the predictor variables.In this analysis, Cortisol Responder still significantly predicted the intrusive memory multivariate solution, F(3,50) = 6.30, p =.001, η p 2 =.27, and this effect was moderated by change in AEA, F(3,50) = 2.90, p =.044, η p 2 =.15, but not in 2-AG (p =.680).Again, no other significant effects were found (highest p =.105).The significant multivariate interaction was only relevant for intrusive memory frequency (p =.088, η p 2 =.06), with distress (p =.517, η p 2 =.01) and vividness (p =.295, η p 2 =.02) showing no effect, though the significant multivariate solution suggests that overall evidence intrusive memory frequency, vividness, and distress was indicative of a Cortisol Responder x AEA change interaction.This interaction (Fig. 5d) revealed that change in AEA after stress did not affect the reported number of intrusive memories of participants who showed a significant Cortisol Response, but higher AEA response following the stressor was associated with lower number of intrusive memories for Cortisol non-Responders.No significant relationships were found between hair biomarkers and intrusive memory frequency, vividness, or distress.

Discussion
In the current study, we tested the effect of stress induction on endocannabinoid saliva reactivity and intrusive memories of violent film clips.The stress induction was not successful as there were no differences between groups in any stress reactivity measure or in the number, distress, or vividness of the reported intrusive memories.However, there were robust effects of cortisol responsiveness, with cortisol responders reporting significantly fewer intrusive memories, as well as lower vividness and distress of intrusive memories, irrespective of whether participants were assigned to the stress or control group.The relationship between cortisol responsivity and the number of reported intrusive memories was moderated by change in AEA post-stress, with decreases in AEA post-stress associated with more intrusive memories in cortisol non-responders, but not cortisol responders.Finally, as hypothesised, there were significant positive correlations between salivary alpha amylase (sAA) and salivary endocannabinoids, but there were no significant associations between stress-related change in sAA enzymic activity and endocannabinoids.This study further clarifies the role of endocannabinoids in the sympathetic nervous system and reinforce the importance of controlling for cortisol responsiveness following stress induction in laboratory trauma film paradigms.
While previous studies have shown that stress reactivity and related biological responses are positively associated with intrusive memory frequency (Cheung et al., 2015;Hilberdink et al., 2022;Nicholson et al., 2014), viewing traumatic imagery or film does not necessarily invoke a biological stress response (Otte et al., 2016).Recently, Hilberdink et al. (2022) showed that inclusion of a socially evaluated cold-pressor task prior to trauma film viewing significantly increased the number of reported intrusive memories relative to control, and that this was positively associated with the biological stress responses measured in saliva.
In the current study, we incorporated the cold-pressor task into the trauma film paradigm such that participants engaged in both activities at the same time.However, we found that there was no increase in any measure of stress reactivity in the stress condition, and no related change in the frequency, vividness, or distress of reported intrusive memories.There were two key differences between our study and Hilberdink et al. (2022).Firstly, our stress task did not include a social evaluation component, which has been shown to produce significantly higher stress reactivity compared to the cold-pressor alone (Smeets et al., 2012).It is possible that viewing a trauma film is stressful enough such that participants did not feel additionally stressed by the presence  There was a significant main effect of Time, where intrusive memories became less vivid over time.There were no other significant effects.Panel B: Distress of intrusive memories.There was a significant main effect of Time, seemingly due to intrusive memories being less distressful in the days following the experiment itself.There was also a significant Time x Condition interaction, but none of the simple effects tests were significant.Error bars are standard error. of another mild stressor (the cold-pressor).Alternatively, Hilberdink et al. scheduled their stress task prior to trauma film viewing, whereas we conducted ours at the same time as the trauma film task.This may have further reduced the perceived stressfulness of the cold-pressor.Our decision to coincide stress induction with film viewing was based on the idea thatin most casesa real-world traumatic experience (emulated by the trauma film) is accompanied by a stress response.In real life, a stress response does not usually occur prior to a traumatic experience.Our study aligns more closely with the findings of Schultebraucks et al. (2019), who found that increased numbers of intrusive memories were associated with the stress group only among cortisol responders to the stress induction task.However, in contrast to this study, we found that cortisol responders reported significantly fewer intrusive memories that were both less distressing and less vivid compared to those reported by cortisol non-responders.This effect was consistent regardless of whether participants were in the stress group or non-stress group, supporting the notion that most participants who experienced a biological stress response were already substantially stressed from viewing the trauma film.
It is unclear why we observed an opposite effect to that reported by previous studies, though it can be explained from a theoretical perspective.Firstly, cortisol is known to selectively consolidate negative memories (Roozendaal et al., 2009), resulting in better consolidation of these emotional memories.In contrast, intrusive memories reflect a disturbance in emotional memory consolidation and are theorised to be the result of poorly consolidated memories of negative experiences (Brewin et al., 2010).This theory is supported by observations of chronically low cortisol concentrations in patients with PTSD (Yehuda, 2009), which is theorised to contribute to the ongoing impairment in healthy consolidation of the trauma memories.The differences in stress timing between our and previous studies' may have also contributed to the discrepant cortisol effects on the number of intrusive memories.Specifically, the fact that our stressor coincided with the trauma film viewing may possibly have affected intrusive memory consolidation differently to studies that have induced stress prior to film viewing.Future studies will need to explore this issue in more depth.sAA enzymic responsivity did not influence intrusive memories reported by our participants.
Our group has previously shown that endocannabinoids are reliably detectable in saliva (Ney et al., 2020) and are stress reactive but not correlated with blood levels of endocannabinoids (Ney, Stone, et al., 2021).This work suggests that salivary endocannabinoids may be part of the sympathetic nervous system and synthesised on-demand in the saliva glands, much the same as in other peripheral and central nervous system tissue and organs (Kano et al., 2009;Ligresti et al., 2016).The current study therefore also aimed to test whether endocannabinoids in saliva are positively correlated with sAA enzymic activitya known proxy of plasma norepinephrine (Nater and Rohleder, 2009).The data in the present study confirmed this hypothesis, which provides further support that salivary endocannabinoids are part of the sympathetic nervous system and even may be additional proxy biomarkers for peripheral norepinephrine stress reactivity, though this possibility is yet to be directly investigated.
Our data also revealed that post-stress elevations in salivary AEA moderated the effect of cortisol reactivity on the number of intrusive memories reported by participants.The direction of this effect suggested that negative or no change to AEA level immediately following stress was associated with more intrusive memories in participants who did not have a cortisol response.Conversely, participants who had a strong AEA increase following stress induction reported an equivalent number of intrusive memories, regardless of their cortisol responsivity.This is the first time that endocannabinoids have been tested in relation to experimental intrusive memories to our knowledge and interpretation of the finding is difficult due to the lack of knowledge concerning the exact biology of endocannabinoids in saliva, particularly given evidence from our laboratory suggests that salivary endocannabinoids do not correlate with blood endocannabinoid levels (Ney, Stone, et al., 2021).However, the most straightforward explanation of this data is that AEA may be involved in the regulation of cortisol effects on intrusive memories.Previous studies have found that sympathetic nervous system activity interacts with cortisol to predict intrusive memory frequency (Nicholson et al., 2014), which is a proposal based on the existing literature on the mechanisms underpinning negative memory consolidation (Roozendaal et al., 2009).Further, endocannabinoids have also been shown to tightly regulate cortisol responses (Balsevich et al., 2017;Hill and Tasker, 2012), and even may serve as an emotional buffer following negative experiences (Alvares et al., 2008;Morena and Campolongo, 2014).Our data suggests that some form of sympathetic or HPA stress response was required to reduce intrusive memory experiencing, though the nature of this interaction should be considered exploratory and needs to be explored in future research.
This study was limited by data loss of sAA analyses, which occurred due to a technician error during sample preparation.This prevented sAA being compared to endocannabinoid data due to inconsistencies in sAA enzymic activity level.Further, stress reactivity to the cold-pressor task was not observed, implying that a more stressful manipulation was necessary to accurately assess the intended outcomes of the study.Critically, high salivary endocannabinoid concentrations were observed in the baseline saliva sample.This is not the first time that we have seen this effect (Ney, Stone, et al., 2021), and it is highly possible that elevated baseline levels are due to the first sample routinely being contaminated by recent food, phlegm, blood, or other matter that is not present in subsequent samples.Future research that tests whether an additional baseline sample is necessary to establish true baseline salivary endocannabinoid levels is essential prior to further research in this area.We also did not measure the intensity of salivation during the study.While it is known that ECBs are involved in the regulation of salivation (Andreis et al., 2022;Prestifilippo et al., 2006Prestifilippo et al., , 2013)), it is not to our knowledge known whether the intensity of salivary excretion affects ECB levels in saliva.Future studies will need to assess this issue and interpret our results here retrospectively.Finally, our study did not collect data on some factors that can affect stress reactivity and intrusive memory frequency.These factors include childhood trauma (see Suzuki et al., 2014), menstrual cycle phase (Felmingham et al., 2012;Soni et al., 2013), and hormonal contraception (Laird et al., 2019).We also did not collect sympathetic data such as heart rate or blood pressure, which could have improved our knowledge of the relationship between ECB and sympathetic nervous system activity.Future studies should examine the relationship between salivary ECBs and these physiological sympathetic measures for deeper insight into the relationship between ECBs and the sympathetic nervous system.
In conclusion, we found that addition of the cold-pressor task to a trauma film did not increase stress reactivity to the trauma film and did not result in increased frequency of intrusive memories in healthy participants.However, this lack of effect was robustly accounted for by cortisol responsivity to both the stress task and the trauma film.Endocannabinoids measured in saliva were reliably positively associated with salivary alpha amylase and change in AEA following stress moderated the effect of cortisol responding on the number of the reported intrusive memories.These findings contribute to a growing literature on the role of salivary endocannabinoids in human behaviour and biology, and also reveal boundary conditions for stress-induced intrusive memory consolidation in humans.

Funding
This work was funded by an Early-Mid Career Enabler grant awarded to LJN by the Queensland University of Technology.

Fig. 3 .
Fig. 3. Panel A: Number of intrusive memories.There was a significant main effect of Time, where participants experienced less intrusive memories over time.No other effects were significant.Panel B: Prevalence of each sub-category of intrusive memory (i.e., each film clip in the trauma film), summed across participants.The "fire hydrant" scene from the movie "Irreversible" was the most prevalent intrusive memory for both the control and stress groups.Film clips are listed in the same order (going from top to bottom) as they were presented in the trauma film.Error bars are standard error.

Fig. 4 .
Fig. 4.Panel A: Vividness of intrusive memories.There was a significant main effect of Time, where intrusive memories became less vivid over time.There were no other significant effects.Panel B: Distress of intrusive memories.There was a significant main effect of Time, seemingly due to intrusive memories being less distressful in the days following the experiment itself.There was also a significant Time x Condition interaction, but none of the simple effects tests were significant.Error bars are standard error.

Fig. 5 .
Fig. 5. Panel A: Distress of intrusive memories.There was a significant main effect of Cortisol Responder, where intrusive memories were less distressing in Cortisol Responders.Panel B: Vividness of intrusive memories.There was a significant main effect of Cortisol Responder, where intrusive memories were less vivid in Cortisol Responders.Panel C: Average number of reported intrusive memories (square root transformed).There was a significant main effect of Cortisol Responder, where intrusive memories were less frequent in Cortisol Responders.Panel D: Simple slopes moderation of Cortisol Responder effect on average number of intrusive memories by change in AEA pre-to post-stress induction.Decreased or no change in AEA following stress was associated with higher number of intrusive memories in Cortisol non-Responders.Decreased AEA = 1 standard deviation of AEA change below the mean, No change AEA = mean AEA change from pre-to post-stress, and Increased AEA = 1 standard deviation of AEA change above the mean.Error bars are standard error.

Table 1
Demographics and biomarker demographics between Stress and Control participant groups.