Neuroscientific evidence for pain being a classically conditioned response to trauma- and pain-related cues in humans : PAIN

Journal Logo

Research Paper

Neuroscientific evidence for pain being a classically conditioned response to trauma- and pain-related cues in humans

Franke, Laila K.a,*; Miedl, Stephan F.a; Danböck, Sarah K.a; Grill, Markusa; Liedlgruber, Michaela; Kronbichler, Martinb,c; Flor, Hertad,e; Wilhelm, Frank H.a

Author Information
doi: 10.1097/j.pain.0000000000002621

1. Introduction

Chronic pain and posttraumatic stress disorder (PTSD) cooccur at striking frequency,1,83,90,93 but mechanisms explaining this comorbidity are poorly understood.66 Re-experiencing the traumatic event as recurrent, involuntary, and intrusive distressing memories is a core symptom of PTSD.4 Of interest, patients with PTSD re-experience the traumatic event not only as distressing visual intrusions (eg, the perpetrator's face) but also often (eg, 49% of patients with PTSD53) as painful sensations.25,34,82 Possibly, re-experiencing the traumatic event could thus be a factor common to both PTSD and chronic pain.

One way to understand re-experiencing is through classical conditioning. Following conditioning principles, originally neutral stimuli (conditioned stimuli [CS]) become associated with a biologically relevant unconditioned stimulus (US) through temporospatial proximity. Subsequently, CS usually elicit similar responses to the US (ie, conditioned responses [CRs]). Previously, we showed that CS not only elicited affective and physiological fear CRs but also intrusive memories as CRs.25,28,39,87,98 Studies investigating intrusions as CRs have exclusively used aversive film clips as US. As such, perhaps unsurprisingly, CRs mostly remained within the visual modality. The inclusion of painful US might lead to CRs in the somatosensory modality. Thus, in this study, we tested whether pain-signaling CS elicit pain-CRs, just as aversive footage–signaling CS have shown to produce fear and visual intrusions as CRs.28,76,87,97

Most clinicians endorse the idea that pain can be a CR and believe that there is strong evidence for this idea.56 Furthermore, it has been theorized that pain can occur as a CR that maintains itself beyond the initial source of nociception has subsided.27,61,67,95 However, experimental evidence for the idea of pain as a CR is scant. Research has mainly evaluated how conditioning modulates pain perception. Results have supported conditioned hyperalgesia, i.e., the phenomenon that, after conditioning, individuals perceive nociceptive stimuli as more painful if preceded by pain-signaling CS.5,36,42,55 On the contrary, studies have also suggested conditioned allodynia, ie, that, after conditioning, individuals perceive stimuli as painful at a lower threshold if these stimuli are preceded by pain-signaling CS.54,92 Although evidence suggests that pain-signaling CS can either amplify pain or render ambiguous nociceptive stimulation to be perceived as painful more easily, whether such CS also elicit pain in the absence of any nociceptive stimulation remains an open question.

This study investigated whether pain-signaling CS elicit pain in the absence of nociceptive stimulation. Because we ultimately aimed to understand chronic pain after psychological trauma, we investigated pain conditioning within a negative affective context (vs a neutral affective context). Therefore, we used painful stimulation and aversive film clips (trauma films41) as US, resulting in a 2 (pain/no pain) × 2 (aversive/neutral film) within-subject trauma pain–conditioning paradigm (TPC). We hypothesized that, after classical conditioning, (I) pain can occur in response to pain-signaling CS in the absence of nociceptive stimulation. During conditioning and memory-triggering task (MTT), we captured pain in response to CS by self-report and 2 established multivariate functional magnetic resonance imaging (fMRI)-based markers of pain: the neurological pain signature (NPS)96 and the stimulus intensity–independent pain signature (SIIPS).101 We tested whether pain was stronger to pain-signaling CS vs no-pain–signaling CS and whether this was modulated by CS' affective context. In addition, we expected (II) that pain occurs spontaneously in daily life in the form of pain intrusions and that (III) individuals showing greater pain-CRs in response to experimental-CS show a greater probability of experiencing such pain intrusions.

2. Methods

2.1. Participants

We tested 74 healthy women aged between 18 and 35 years. We tested only women because previous research has suggested that men respond differently to aversive film clips than women100 and were unable to include a sufficiently sized group of men because of monetary time constraints. Exclusion criteria were medication use (except for oral contraceptives), cardiovascular, neuroendocrinological, and pain-related disorders, plaster allergies, reports of current mental disorders, and absence of self-reported physical and psychological resilience. Furthermore, participants reporting a high (more than 2-3 times a week) consumption of extremely violent media were excluded to reduce the odds of including participants for whom exposure to aversive film scenes would potentially elicit insufficient distress. Functional magnetic resonance imaging–specific exclusion criteria were pregnancy, ferromagnetic implants, other nonremovable metal objects, claustrophobia, and very high (>35 kg/m2) or too low (<18 kg/m2) body mass index. Over the course of the study, 10 participants were excluded from the analysis because of movement >3 mm in X, Y, or Z direction during fMRI sessions (N = 3 in the first session; of them, 2 were excluded from the second session), suspected brain anomaly (N = 1), dropout due to aversiveness of trauma analogue situation (N = 3), and technical difficulties (N = 3), therefore leaving a final sample of N = 64 for the first-session acquisition analyses, and N = 65 for the second-session MTT analyses. The study was approved by the Ethics Committee of the Paris-Lodron-University of Salzburg (GZ 7/2019). All participants provided written informed consent before participation and were reimbursed either with course credit or 80 Euro.

Current depression, anxiety, and stress symptomatology were assessed with the long version of the Depression Anxiety Stress Scales (DASS,37 German version70); trait anxiety was additionally assessed with the State Trait Anxiety Inventory (STAI,85 German version49). Somatization symptoms were measured with the Screening for Somatoform Symptoms (SOMS)-7.79 To characterize the sample regarding potential PTSD symptomatology, we used the revised Impact of Event Scale (IES-R,38 German version57) applied to the most distressing life event. As detailed in Table 1, DASS, STAI-T, SOMS-7, and IES-R were within normal ranges.

Table 1 - Sample characteristics.
M SD Range in sample
Age (y) 21.94 3.10 18; 31
DASS—D 3.51 4.08 0; 15
DASS—A 2.59 3.37 0; 15
DASS—S 5.74 5.66 0; 22
STAI-T 38.22 8.54 21; 61
SOMS-7 ICD-10 somatization index 0.92 1.35 0; 6
IES-R −3.17 1.65 −4.44; 1.59
Values indicating nonclinical relevance are <10 for DASS—D, < 8 for DASS—A, <15 for DASS—S, <39 to 40 for STAI-T,40 and <0 for IES-R. Average ICD-10 somatization scores within the general German population are M = 1.1 (SD = 1.7).72
DASS—D/A/S, Depression-Anxiety-Stress-Scales; IES-R, Impact-of-Events Scale (revised version); SOMS-7, screening for somatoform symptoms (7-item version); STAI-T, state trait anxiety inventory.

2.2. Stimuli

2.2.1. Unconditioned stimuli

2.2.1.1. Painful stimulation

Electrocutaneous stimulation was delivered using a reusable concentric surface electrode with 7 mm diameter and a platinum pin WASP electrode (Brainbox Ltd, Specialty Developments, Cardiff, Wales) that was attached to participants' inner side of the left calf and was controlled by a Digitimer DS7A constant current stimulator (Digitimer Ltd, Hertfordshire, England). To produce stable stimulus intensities,68 electrocutaneous stimulation was applied in 7-pulse trains with a duration of 988 milliseconds each and an interstimulus interval of 400 to 1300 milliseconds, resulting in a total stimulation time of 12.5 seconds. The first pulse train always coincided with film onset.

Stimulation intensity was determined individually using an established stepwise calibration procedure.75 Stimulation started at 0.2 mA and was incremented stepwise by 0.2 mA until participants verbally reported their perceptual threshold (sensation just noticeable). Stimulation intensity was then further incremented by 0.2 mA per trial until participants reached their pain threshold (sensation is painful). Subsequently, stimulation intensity was increased stepwise by 5% of the individually determined pain threshold until participants reported their pain tolerance (maximum pain tolerance), where maximum pain tolerance was defined as the moment shortly before one would want to tear the electrode off the calf, similarly to when one wants to drop a coffee cup that is too hot. If participants did not reach their pain tolerance after 3 of these intensity-increasing trials, the current stimulation intensity was increased by 15% until pain tolerance was reached.

During TPC, the applied electrical stimulation intensity corresponded to the intensity at 30% between the individual pain threshold and pain tolerance. On a scale ranging from 0 (no sensation) to 5 (painful) and 10 (maximal bearable), individually calibrated stimulus intensity had to be rated between 6 and 7. This calibrated stimulus intensity was used for the TPC. If the rating was too low or too high, stimulus intensity was adjusted by 0.1 mA and tested repeatedly until the desired rating was achieved.

2.2.1.2. Film clips

Four film clips with a 16-second duration were extracted from commercial movies. The 2 aversive film clips depicted severe interpersonal violence scenes (sexual assault in a tunnel; physical assault with an oxygen bottle) from the movie Irreversible.71 The 2 neutral film clips depicted ordinary human social interactions with no aversive content (beach walk; basketball game) and were extracted from Coach Carter18 and Mr. Jones, respectively26 Aversive and neutral film clips were previously validated for use in trauma film studies and were matched in content, valence, and arousal.4

2.2.2. Conditioned stimuli

Conditioned stimuli were 4 images resembling contextual elements of each film clip: a basketball (basketball game scene), a pier (beach scene), a tunnel (sexual assault scene), and an oxygen bottle (physical assault scene). Each CS had a duration of 4 seconds and immediately preceded US.

2.3. Procedures

2.3.1. Screening and questionnaires

After receiving written information about the study's procedure and an informed consent form, participants filled out an initial online screening form checking for inclusion and exclusion criteria. When meeting inclusion criteria, we scheduled individual phone calls to explain the study procedure in detail to participants and answer potential questions. One week before the first fMRI session, participants filled out further online questionnaires concerning trait (STAI-T) and demographic variables.

2.3.2. Functional magnetic resonance imaging session 1: trauma pain conditioning paradigm

2.3.2.1. Before the trauma pain conditioning

Functional magnetic resonance imaging sessions took place at the Christian-Doppler Clinic in Salzburg. On arrival at the clinic, participants answered a set of questionnaires (DASS, SOMS-7) to indicate whether they felt mentally and physically stable. Note that current psychological and physical resilience was an inclusion criterion in this study to prevent participants from developing intrusive memories of the aversive analogue trauma situation that would require potential clinical attention (eg, too intense and persistent). After providing written informed consent, participants underwent the pain calibration procedure while seated on the MRI table (see Section 2.2.1.1. Stimuli—Painful Stimulation). After determining the individual stimulation intensity, we acquired an eight-minute long resting-state fMRI sequence. Among others, this resting state was intended to accustom participants to the MRI environment.

2.3.2.2. Trauma pain conditioning

Before the acquisition phase, each of the 4 CS was presented 4 times for habituation. After all habituation trials, participants were instructed to bring back to mind how they felt while viewing each CS. Then, while seeing a screenshot of each CS, participants rated each of the 4 CS for pain and valence.

After habituation, participants were instructed that pictures (CS) might be followed by film clips or electrical stimulation. As displayed in Figure 1A, each CS was presented with a different US: pain + aversive film, no pain + aversive film, pain + neutral film, and no pain + neutral film (50% reinforcement rate). Film and pain US started together, ie, there was no onset asynchrony between both stimuli. The pairing of neutral and aversive film clips with pain vs no pain was counterbalanced between participants. Similar to the procedure after habituation, after all acquisition trials, participants were asked to bring back to mind how they felt while viewing each CS during acquisition and, while seeing a screenshot of the CS, rate each CS for pain and valence. After CS ratings, participants also rated each US for pain and valence while seeing a screenshot of the US film. See Figure 1A for an overview of the TPC paradigm. After the TPC paradigm, we acquired another 8-minute resting-state fMRI sequence (results to be presented elsewhere).

F1
Figure 1.:
Experimental design. On day 1 (Panel A), participants underwent the trauma pain conditioning (TPC) paradigm, a classical conditioning procedure with a 2 (pain/no-pain stimulation) × 2 (aversive film/neutral film) within-subject design. The TPC started with a habituation phase, where each CS was presented 4 times without US. After habituation, participants rated the CS for pain and valence. During acquisition, each CS was presented 8 times with US (ie, 8 reinforced trials) and 8 times without US (ie, 8 unreinforced trials; not displayed in the Figure for easier display of the study design). After all acquisition trials, participants again rated the CS for pain and valence. During habituation and acquisition, stimuli were presented in a pseudorandom order (not more than 2 consecutive stimuli of the same type), with ITIs ranging between 12 and 16 seconds. Twenty-four hours later (Panel B), participants returned to the laboratory and were re-exposed to CS from TPC during a memory-triggering task (MTT). During MTT, each CS was presented 3 times without US, with ITIs ranging between 6 and 18 seconds in a block lasting in total 52 seconds. Immediately after each CS-block, participants rated the CS-block on pain. Each CS-block was presented twice: once in a first run (early MTT) and once in a second run (late MTT). Within each run, CS-block presentations were randomized. In addition, participants also reported pain intrusions for 7 days (starting after acquisition) during daily life in a diary using a smartphone application (Panel C). A, aversive film clip; CS, conditioned stimulus; ITI, intertrial interval; nP, no pain stimulation; N, neutral film clip; P, pain stimulation; s, seconds; US, unconditioned stimulus.

2.3.3. Functional magnetic resonance imaging session 2, 24 hours after trauma pain conditioning: memory-triggering task

As in the first fMRI session, participants indicated whether they felt mentally and physically stable for participation. We also repeated the pain calibration procedure and, once participants were lying in the scanner, showed the aversive film clip that on day 1 had not been associated with pain for that individual. Then, we instructed participants that pictures (CS) might be followed by film clips or electrical stimulation. These steps were taken to keep US expectancy high during an MTT.

As can be seen in Figure 1B, during MTT, participants were re-exposed to each CS from the TPC paradigm in blocks. After each block, participants indicated subjective pain experienced during the respective block. After MTT, we acquired a high-resolution structural scan.

2.4. Measures and outcomes

2.4.1. Measuring pain

Self-report is the gold standard for assessing the experience of pain.20 However, self-report by itself is not a perfect measure because it is subject to limitations in self-perception, metacognition, and high interperson and intraperson variability.16,33,91 In this vein, additional measures tracking nociception (ie, the neural process of encoding noxious stimuli40,60) such as the NPS96 have been deemed welcome additional measures to assess pain.78 The NPS predicts pain level based on fMRI activity mostly within noxious stimuli intensity–encoding regions and has more than 90% sensitivity and specificity in predicting acute pain relative to other salient states, including anticipated pain, pain recall, threat cues,96 general (picture-induced) negative emotion,19 and observed pain.47 Considering that the NPS was developed for nociceptive pain and does not respond to several forms of psychological pain modulation,46,101 it only quantifies a subset of various brain processes contributing to pain. To also capture brain processes contributing to pain beyond nociception, Woo et al.101 developed a second multivariate brain measure related to pain, the SIIPS. This signature complements the NPS by predicting pain independent of nociception and mediating effects of several psychological manipulations (eg, cued expectancy and perceived control) on pain.101 To capture these processes, we assessed pain additionally with SIIPS responses.

Note that both the NPS and SIIPS are based on relative activity levels within regions of interest that are specified as weights that can be positive (indicating more predicted pain with greater activity) or negative (indicating less predicted pain with greater activity). The NPS includes positive weights in regions such as the dorsal anterior cingulate cortex (dACC), insula, secondary somatosensory cortex (S2), and thalamus and negative weights in structures deactivated by pain, such as the ventral medial prefrontal cortex (vmPFC) and occipital cortex. Because the current investigation aimed to evaluate NPS responses to visual stimuli, known to activate structures such as the occipital cortex that yield negative NPS weights,47 the inclusion of negative NPS weights would likely not accurately reflect nociceptive pain (eg, participants may show low NPS responses to nociceptive stimulation during CS presentations because the visual stimulation counteracts the expected pain-driven deactivation in occipital regions). As such, we conducted main analyses with the previously published positive NPS weight mask (NPS-pos) (for an overview of analyses performed on a manually edited NPS mask excluding occipital regions only, see Supplement 2, available at https://links.lww.com/PAIN/B592).23,52 The SIIPS, on the contrary, includes positive weights not only in regions that receive nociceptive input, including the operculum, insula, and cingulate cortex, but also in frontal regions associated with higher-level cognitive processes such as the dorsomedial PFC (dmPFC). At the same time, the SIIPS includes negative weights in regions such as the hippocampus and basolateral amygdala.101 Because these regions are activated during formation or expression of CRs,72,88 we excluded negative weights from the SIIPS and performed analyses on SIIPS values based only on positive-weighted structures (SIIPS-pos).

Taken together, the primary measures of pain in this study were self-report, NPS-pos, and SIIPS-pos responses. Secondarily, to examine which component of pain is conditioned, we also examined negative emotions (through self-report and picture-induced negative affect signature (PINES)19 responses) and arousal.

2.4.2. Conditioned (and unconditioned) response acquisition

2.4.2.1. Pain

Participants rated how strongly (0 = not strong at all, 10 = maximal bearable) they experienced pain sensations (self-reported pain) during each CS and US presentation. Furthermore, on a neural level, we extracted NPS-pos and SIIPS-pos responses to unreinforced CS trials and all US trials.

2.4.2.2. Negative emotions

Participants rated as how pleasant (0 = very pleasant, 10 = very unpleasant) they perceived each CS and US during the experiment. On a neural level, we extracted participants' PINES19 responses to unreinforced CS trials and all US trials.

2.4.2.3. Arousal

Participants rated how arousing (0 = not arousing at all, 10 = extremely arousing) they perceived each US during the experiment. On a physiological level, we acquired skin conductance responses (SCRs) for all CS and US trials.

2.4.3. Conditioned response retention (memory-triggering task)

2.4.3.1. Pain

Participants indicated the percentage of time (0 = 0% of the time, 10 = 100% of the time) they experienced painful bodily sensations during the CS block (self-reported pain). Note that this pain rating differed from that during TPC (ie, the analogue traumatic event) because we assumed that conditioned pain responses during MTT would rather reflect persistent CRs that could, according to our theory of conditioned intrusions,25,28,97 be conceived of as pain intrusions. Therefore, to capture this potential intrusive memory aspect of persistent pain CRs, we approximated pain ratings during MTT to previously established measures of conditioned (visual) intrusions.87,97 Furthermore, we also extracted NPS-pos and SIIPS-pos responses to each CS presentation within the early and the late MTT blocks.

2.4.3.2. Negative emotions and arousal

To reduce subject burden and prevent too rapid extinction of CRs, at MTT, we acquired only the most essential pain ratings. We acquired PINES responses and SCRs to each CS trial during early and late MTT blocks to capture retention of negative emotions and arousal to CS during MTT.

2.4.4. Conditioned response retention (daily-life pain intrusions)

Congruent with the empirically supported assumption that intrusive memories are a form of CR to trauma-related cues,28 we expected that participants would report pain intrusions (next to film intrusions) after conditioning. For this purpose, after TPC, we instructed participants to register intrusive memories for 7 consecutive days through an e-diary application PsyDiary for smartphones.69 We defined intrusions broadly as spontaneously occurring memories in the form of pictures, bodily sensations, sounds, feelings, or thoughts regarding the pain stimulation or film clips and sudden recurring physical sensations, thoughts, or feelings experienced while watching the film clips or experiencing the painful stimulation. We informed participants of the possibility that pain intrusions and film intrusions could mingle and instructed participants to decide whether their memory concerned primarily film clips or painful sensations. Furthermore, we instructed participants to register every intrusion in the e-diary app on occurrence, ie, in an event-based manner. Participants were also asked to report associated distress (0 = not at all distressing to 100 = extremely distressing) and strength of pain sensations (0 = not strong at all, 10 = maximal bearable) on a visual analogue scale. Finally, participants were also instructed to provide a short content description of the occurring intrusion and potentially occurring triggers in the e-diary app.

To monitor participants' compliance, we sent text message reminders for a questionnaire at 10 pm each day, where we explicitly assessed whether participants reported all intrusions throughout the day. If participants indicated noncompliance, they were asked to retrospectively estimate the true number and distress of intrusions, and we substituted the event-based intrusion score by the retrospectively estimated intrusion score (2% of data).

2.5. Apparatus and physiological recordings

Stimulus presentation and behavioral data acquisition were controlled by E-Prime 2.0 (Psychology Software Tools, Inc, Pittsburgh, PA). Skin conductance (SC, in μS) was measured using Ag/AgCl electrodes filled with isotonic electrode paste11; electrodes were placed on the lower palm of the left hand. Recording of SC data was performed with a sampling rate of 1000 Hz using Polybench 1.22 (TMSi-Twente Medical Systems International, EJ Oldenzaal, Netherlands), a Porti 32-channel-amplifier (TMSi), and an SC amplifier (Becker Meditec, Karlsruhe, Germany). ANSLAB 2.6 was used for SC analysis.9,99

2.6. Data analysis

2.6.1. Functional magnetic resonance imaging preprocessing and first-level analysis

Functional magnetic resonance imaging data preprocessing and analysis were performed using SPM12 (Wellcome Department of Cognitive Neurology, London, United Kingdom). First, functional images were corrected for geometric distortions by using the FieldMap toolbox, realigned, unwarped, and slice time–corrected to the onset of the first slice. Structural images were segmented and normalized to MNI standard stereotactic space. The resulting parameters were then used for normalization of the previously coregistered functional images, which were resampled to isotropic 3 mm3 voxels and smoothed with a 6-mm full width at half maximum Gaussian kernel. Acquisition and MTT session data were analyzed in an event-based manner. In the participant-specific first-level model, each event was convolved by a canonical hemodynamic response function. For acquisition analyses, regressors for the first-level model were CS responses during habituation (4 seconds), reinforced and unreinforced CS responses during acquisition (4 seconds), and US responses (16 seconds). For MTT analyses, first-level regressors were CS responses (4 seconds), and regressors of no interest were ratings (10 seconds). In both first-level models, we also added the 6 rigid body movement parameters determined from realignment from the respective session as covariates of no interest.

2.6.2. Neurological pain signature-pos, stimulus intensity–independent pain signature-pos, and picture-induced negative affect signature analyses

The expression strength of the NPS-pos, SIIPS-pos, and PINES patterns for each of the 4 US and CS conditions was determined by calculating the dot product of the respective vectorized activation images with the NPS-pos, SIIPS-pos, and PINES pattern weights. As explained in Section 2.4.1., we used a mask containing only positive weights for NPS analyses52 and only averaged over positive weighted structures to determine SIIPS responses.101

For US NPS-pos, SIIPS-pos, and PINES responses, activation images were individuals' first-level US-response images of each of the 4 conditions (ie, USpain + aversive, USnopain + aversive, USpain + neutral, and USnopain + neutral), including a total of 4 trials each. For NPS-pos, SIIPS-pos, and PINES responses to CS during habituation and acquisition, activation images were individuals' first-level CS-response images (ie, CSpain + aversive, CSnopain + aversive, CSpain + neutral, and CSnopain + neutral), respectively, averaged over the last 3 CS trials from habituation, 2 unreinforced CS trials from the first acquisition halve (acq1), and 2 unreinforced CS trials from the second acquisition halve (acq2). Similarly, for NPS-pos, SIIPS-pos, and PINES responses to CS during MTT, activation images were individuals' first-level CS response images, and these were averaged separately over the 3 CS trials of the early MTT block and over the 3 CS trials of the late MTT block. This returned continuous scalar values for each US and CS condition from habituation, (early and late) acquisition, and (early and late) MTT.

Higher NPS-pos values indicate higher nociceptive pain96; higher SIIPS-pos values indicate greater stimulus-independent pain101; and higher PINES values indicate greater picture-induced negative emotion.19

2.6.3. Second-level whole-brain analyses

Unreinforced CS responses of each condition were entered into a second-level random effects model. We calculated a full factorial design with pain (pain vs no pain) and film (aversive vs neutral) as within-participant factors for both acquisition and MTT sessions. For all analyses, the threshold was set to P < 0.05 corrected for multiple comparisons (based on the false discovery rate [FDR] and whole-brain level) and a cluster size of k ≥ 5.

2.6.4. Skin conductance response analyses

For US analyses, SCRs were quantified by subtracting the mean baseline skin conductance level (SCL, −3 to 0 seconds relative to the CS onset) from the maximum SCL during each of the four 16-second film clips, considering only the first presentation of each film clip. Similarly, for CR analyses, for each of the 4 CS conditions, we subtracted the mean pre-CS SCL (−3 to 0 seconds relative to the CS onset) from the maximum CS SCL (0-5 seconds relative to the CS onset). Skin conductance responses were normalized using the natural logarithm of 1 + SCR. Owing to technical problems, data from 2 participants were missing for SCR analyses.

2.6.5. Content and trigger classification of pain intrusions

For descriptive purposes, we categorized the content of pain intrusions and associated triggers. Based on theoretical expectations, the first author and an experienced graduate student classified pain intrusions based on participants' content description (1 = stinging/tingling pain sensations; 2 = other painful sensations; 3 = nonpainful bodily sensations; 4 = a combination of painful and nonpainful sensations; and 5 = others). Similarly, pain intrusion triggers were classified according to the sensory modality (1 = audiovisual; 2 = painful sensations; 3 = nonpainful bodily sensations; 4 = thoughts; 5= a combination of sensations; and 6 = others). Initial interrater reliabilities were k = 0.68 for pain intrusion content classification and k = 0.60 for trigger classification. Disagreements in classifications were discussed until a consensus was reached.

2.6.6. Statistical analyses

All statistical analyses were performed in RStudio81 in R.77 For manipulation checks and main analyses, we fitted Bayesian multilevel regression models (BMLMs)29,31 through the brms package using Stan in RStudio.14,17 Rating data (ie, self-reported pain, valence, arousal) were fitted with ordinal (cumulative) BMLMs.15 Standardized NPS-pos, SIIPS-pos, and PINES responses were fitted with a Gaussian distribution; standardized SCR data were fitted with a skewed normal distribution to account for the left-skewness of the response variable.14

All models contained repeated measurements over participants (over US conditions in manipulation checks; and CS conditions in acquisition/MTT analyses). Thus, to account for the dependency between observations over participants, responses by the same person were modelled with varying intercepts. Furthermore, the predictors pain and film were dummy coded (pain: no pain = 0, pain = 1; film: neutral = 0, aversive = 1). Because we expected that the effects of pain and film on unconditioned responses (URs) and CRs could vary between participants, we added varying slopes for pain and film in all models to account for this potential variability.6 For acquisition and MTT analyses, we also added varying slopes for time in all models. For an overview of fitted models, see Supplement 1 (available at https://links.lww.com/PAIN/B592).

2.6.6.1. Manipulation checks: unconditioned responses

To check whether pain and film manipulations were successful, we tested whether these experimental factors significantly influenced self-reported pain, NPS-pos and SIIPS-pos responses (pain), self-reported unpleasantness of the film clips and PINES responses (negative emotions), and self-reported arousal and SCRs (arousal) to the 4 film clips. For this purpose, we entered pain and film, as well as the interaction between pain × film as predictors into BMLMs.

2.6.6.2. Main analyses: conditioned response acquisition and retention
2.6.6.2.1. Acquisition

We examined to what extent participants acquired CRs by testing whether the outcome variables indexing (A) pain determined by (A.I) self-reported pain, (A.II) NPS-pos responses, and (A.III) SIIPS-pos responses; variables indexing (B) negative emotions determined by (B.I) self-reported valence and (B.II) PINES responses, and (C) SCRs indexing arousal increased significantly more to CSpain than CSnopain from habituation to acquisition phases. For this purpose, we added the factors pain and time into BMLMs and tested for pain × time interactions. Time yielded 2 levels in self-reported pain and valence (habituation = 0, acq1 + acq2 = 1) and 3 levels in NPS-pos, SIIPS-pos, PINES, and SCR analyses (habituation = 0, acq1 = 1, acq2 = 2) (this difference in levels is due to participants rating only CS after habituation and at the end of acq2; imaging data and SCRs were available for both acq1 and acq2). Because we expected that acquisition of CRs could depend on the affective context induced by our film conditions (aversive vs neutral), we also tested for pain × time × film interactions.

Significant interactions were followed-up by post hoc BMLMs within each study phase (habituation and acquisition phases) and film conditions (aversive and neutral). To account for baseline CS-valence differences suggested by analyses indicating that participants already rated CS signaling the aversive film clip as more unpleasant than CS signaling the neutral film clip during habituation over both pain conditions, as well as to control for general response tendencies, we corrected acquisition CRs for habituation CRs (ie, acq minus hab). In this way, we used habituation-corrected acquisition CRs in all post hoc analyses performed on the acquisition phase CRs.

2.6.6.2.2. Memory-triggering task

We examined to what extent participants retained CRs by testing whether the outcome variables (A.I.) self-reported pain, (A.II.) NPS-pos responses, and (A.III.) SIIPS-pos responses, as well as (B) PINES and (C) SCRs, remained significantly stronger to CSpain vs CSnopain during MTT. Because the MTT procedure was functionally similar to an extinction procedure, we considered the possibility that participants showed CRs in the first but not in the second MTT block. As such, we added a time factor (early MTT = 0, late MTT = 1) and tested for pain × time interactions. Furthermore, we checked whether the retention of CRs depended on the affective context with pain × time × film interactions. Significant interactions were followed by post hoc analyses.

2.6.6.3. Relationship between self-reported pain and multivariate brain markers of pain (neurological pain signature–pos, stimulus intensity–independent pain signature–pos)

We tested whether higher NPS-pos and SIIPS-pos responses were associated with increased self-reported pain at acquisition and MTT by fitting BMLMs15 where self-reported pain reports were (separately) predicted by NPS-pos/SIIPS-pos responses. To test whether the relationship between self-reported pain and NPS-pos/SIIPS-pos responses was modulated by film and pain, we added these factors into the model and tested for NPS-pos × pain × film interactions. Significant interactions were followed up by post hoc tests. Note that in line with our main analyses, we used habituation-corrected acquisition-CRs in the acquisition analysis. NPS-pos/SIIPS-pos responses were grand mean–centered and standardized before being entered into the BLML. Finally, we added random intercepts for subjects and random slopes for pain and film.

2.6.6.4. Relationship between pain-conditioned responses (self-reported pain and neurological pain signature–pos) and daily-life pain intrusions

As a secondary analysis, we examined whether acquiring and retaining stronger self-reported pain-CRs and NPS-pos–CRs influenced pain intrusions during daily life. In light of our primary analyses showing that the existence of SIIPS-pos–CRs were statistically uncertain, we did not include this variable in this secondary analysis. Because persistent PTSD is primarily linked to intrusions perceived as distressing,58 we aimed to obtain a more clinically relevant variable by weighting pain intrusions by their distress, ie, operationalizing pain intrusions as intrusion load (ie, daily intrusion number × average distress; equivalent to the sum of daily intrusive distress). Furthermore, we computed differential CR scores to CSpain vs CSnopain within each film condition during acquisition and MTT. Specifically, for the neutral film condition, we computed difference scores for self-reported pain and NPS-pos responses between CRs to CSpain and CSnopain within the neutral film condition. For the aversive film condition, we computed difference scores between CRs to CSpain in the aversive film condition and CSnopain within the neutral film condition because the latter condition figured as a more neutral control than the aversive film or no pain condition. For acquisition analyses, we used habituation-corrected pain-CR scores over both acquisition trials. For MTT analyses, we also used scores over both early and late MTT.

Considering that daily-life intrusions were measured over days, responses by the same person were modelled with varying intercepts. To account for the inflation of zero intrusions in the data, daily-life intrusions were fitted with a hurdle lognormal distribution. With this approach, we fitted data in 2 parts where we (A) estimated the probability of not experiencing (ie, zero) vs experiencing (ie, non-zero) intrusions (hurdle part, modeled as a Bernoulli distribution) and (B) estimated the amount of (ie, severity of) intrusions > 0 (lognormal part, modeled with a lognormal distribution).50

We thus fitted separate models for the predictors (I) self-reported pain and (II) NPS-pos responses during acquisition and MTT. In these models, we further added day (ie, experimental day on which intrusive memory was registered), as well as interactions between differential pain-CRs and day to model potential effects of pain-CRs on the persistence or decay of pain intrusions during daily life.86 Predictors were centered and standardized before being entered in BMLMs: differential pain-CR scores were centered to their respective means, and day was centered on the first 24-hour day after TPC, ie, on the second experimental day. Because we expected that the effect of day on pain intrusions (ie, the decay of intrusions) could vary between participants, we added a varying slope for the effect of day.

2.6.6.5. Bayesian multilevel regression model summary statistics

For a summary of model parameters, we reported regression coefficients and 95% credible intervals (CIs; ie, Bayesian confidence intervals). Based on CIs, we can state that there is a 95% probability that the respective parameter falls within this interval, given the evidence provided by the data, priors, and model assumptions. Effects were considered significantly different from zero if the estimate's 95% CIs did not include zero (this would indicate statistical significance on a 5% level). We used the weak or noninformative default priors of brms as priors, which have only negligible influence on the obtained results.13,14 We reported Bayesian R2 as our measure for effect sizes.30 All Bayesian models converged according to common algorithms-agnostic94 and algorithm-specific diagnostics.7 There were no divergent transitions, Rhat < 1.01 and ESS > 400, for all relevant parameters.

3. Results

Results are organized so that a first section (3.1 Manipulation Checks) tests whether pain and film (aversive/neutral) manipulations successfully elicited pain, negative emotions, and arousal; a second and a third section (3.2 CR Acquisition, 3.3 CR Retention) detail to what extent participants acquired and retained pain-CRs (self-reported pain, NPS-pos, SIIPS-pos, and pain intrusions), as well as negative emotion-CRs and SCRs within aversive/neutral film conditions; a fourth section (3.4 Relationship Between Self-Reported Pain and Multivariate Brain Markers of Pain [NPS-pos and SIIPS-pos]) shows the relationship between self-reported pain, NPS-pos, and SIIPS-pos responses at acquisition and MTT; and finally, a fifth section (3.5 Relationship Between Laboratory-Based Pain-CRs [Self-Reported Pain, NPS-pos] And Daily-Life Pain Intrusions) tests the relationship between pain-CRs and daily-life pain intrusions.

3.1. Manipulation checks: unconditioned response acquisition

3.1.1. Pain

3.1.1.1. Self-report

Participants reported experiencing more severe pain to the US with pain stimulation (USpain) than the US with no-pain stimulation (USnopain) (b = 7.20, 95%-CI = [5.54, 9.17]). Furthermore, participants also reported experiencing more pain to the aversive-film US (USaversive) than the neutral-film US (USneutral) (b = 4.60, 95% CI = [3.34, 6.13]). Post hoc analyses on the significant pain × film interaction (b = −3.96, 95% CI = [−5.38, −2.67]) within each film condition indicated that the difference between self-reported pain to the USpain and USnopain was more prominent within the neutral film (b = 7.87, 95% CI = [1.71, 11.83]) than within the aversive film condition (b = 2.99, 95% CI = [1.68, 5.18]). See Figure 2A (left) for details.

F2
Figure 2.:
Fitted (noncentered and nonstandardized) values of regressions estimating unconditioned responses (URs) to each US condition. Error bars represent 95% CIs, and asterisks mark statistical significance. Panel (A) depicts pain-URs: self-reported pain (0 = not painful, 10 = maximal bearable), NPS-pos, and SIIPS responses. Panel (B) depicts negative emotion–URs: valence ratings (0 = very pleasant, 10 = very unpleasant) and PINES responses. Panel (C) depicts arousal-CRs: self-reported arousal (0 = not arousing at all, 10 = very arousing) and SCRs measured in microSiemens (μS). acq, acquisition with trials averaged over both acquisition halves; acq1, first acquisition half; acq2, second acquisition half; hab, habituation; NPS-pos, neurological pain signature responses based on positive weights only; PINES, picture-induced negative emotion signature; SCRs, skin conductance responses.
3.1.1.2. Neurological pain signature–pos

In line with the results on self-reported pain, NPS-pos responses suggested greater nociceptive signaling in the USpain than in the USnopain conditions (b = 0.48, 95% CI = [0.31, 0.65]). Similarly, nociceptive signaling was greater in the USaversive than in the USneutral conditions (b = 0.37, 95% CI = [0.23, 0.50]). A nonsignificant interaction between pain × film suggested that the effect of pain on NPS-pos responses did not differ between neutral and aversive film conditions (b = −0.10, 95% CI = [−0.28, 0.08]). See Figure 2B (middle) for details.

3.1.1.3. Stimulus intensity–independent pain signature–pos

Stimulus intensity–independent pain signature–pos responses suggested greater pain in the USpain than in the USnopain conditions (b = 0.33, 95% CI = [0.12, 0.54]). Similarly, SIIPS-pos responses were greater in the USaversive than in the USneutral conditions (b = 0.38, 95% CI = [0.21, 0.54]). A nonsignificant interaction between pain × film suggested that the effect of pain on SIIPS-pos responses did not differ between neutral and aversive film conditions (b = 0.02, 95% CI = [−0.21, 0.24]). See Figure 2B (right) for details.

3.1.2. Negative emotions

3.1.2.1. Self-report

Participants perceived USaversive as more unpleasant than USneutral (b = 6.94, 95% CI = [5.66, 8.36]). Furthermore, participants also perceived the USpain as more unpleasant than the USnopain conditions (b = 1.88, 95% CI = [1.17, 2.64]). Post hoc analyses on the interaction between pain × film (b = −2.04, 95%-CI = [−3.05, −1.08]) suggested that while participants rated the USaversive as equally unpleasant regardless of whether they concomitantly received painful stimulation (b = −0.11, 95% CI = [−1.20, 0.57]), they rated the USneutral as more unpleasant when receiving painful stimulation than when receiving no painful stimulation (b = −2.02, 95% CI = [1.19, 3.15]). See Figure 2B (left) for details.

3.1.2.2. Picture-induced negative affect signature

Participants showed greater PINES responses to USaversive than USneutral conditions (b = 0.66, 95% CI = [0.55, 0.77]). Picture-induced negative affect signature responses did not significantly differ between USpain and USnopain conditions (b = 0.01, 95% CI = [−0.11, 0.11]), and results suggested no significant interactions between pain and film (b = −0.06, 95% CI = [−0.22, 0.10]). See Figure 2B (right) for details.

3.1.3. Arousal

3.1.3.1. Self-report

Self-reported arousal was higher to the USaversive than to the USneutral conditions (b = 4.88, 95% CI = [3.80, 6.10]) and higher to USpain than to the USnopain (b = 1.39, 95% CI = [0.70, 2.11]) conditions. Furthermore, the significant interaction between film and pain (b = −1.94, 95% CI = [−2.78, −0.91]) suggested that whereas participants reported similar arousal to the USaversive regardless of whether they concomitantly received painful stimulation (b = −0.43, 95% CI = [−1.18, 0.28]), participants rated the USneutral conditions as more arousing when this US was paired with painful stimulation vs no painful stimulation (b = 1.65, 95% CI = [0.80, 2.85]). See Figure 2C (left) for details.

3.1.3.2. Skin conductance responses

Participants showed greater SCRs to the USaversive than to the USneutral conditions (b = 0.59, 95% CI = [0.49, 0.69]) and greater SCRs to the USpain than to the USnopain conditions (b = 0.16, 95% CI = [0.06, 0.26]). The interaction between pain × film was nonsignificant (b = −0.09, 95% CI = [−0.22, 0.05]). See Figure 2C (right) for details.

3.2. Conditioned response acquisition

3.2.1. Pain

3.2.1.1. Self-report

Results on self-reported pain to CS revealed a significant pain × time interaction (see Table 2A.I for regression coefficients and 95% CIs). As displayed in Figure 3A (left) and corroborated by separate post hoc analyses on self-reported pain to CS during each study phase (habituation/acquisition), this interaction suggested that participants reported more severe pain to CSpain than CSnopain during acquisition (b =2.29, 95% CI = [1.51, 3.08]). During habituation, the difference in self-reported pain responses to CSpain vs CSnopain was nonsignificant (b =1.22, 95% CI = [−0.20, 2.66]). Results further suggested that the factor film (ie, CS signaling aversive film scenes vs neutral film scenes) modulated the pain × time interaction (Table 2A.I). Post hoc analyses on self-reported pain within each study phase revealed that the pain × film interaction was significant during acquisition (b = −0.89, 95% CI = [−1.67, −0.09]), but nonsignificant during habituation (b = −1.15, 95% CI = [−2.73, 0.31]). Follow-up analyses on the pain × film interaction within each film condition during acquisition suggested that participants reported more pain to CSpain than CSnopain not only within the neutral film condition (b = 1.34, 95% CI = [0.61, 2.17]; see upper panel of left Fig. 3A) but also within the aversive film condition (b = 0.87, 95% CI = [0.13, 1.63]; see lower panel of left Fig. 3A).

Table 2 - Effects of pain, film, and time on (A) pain responses (self-reported pain, NPS-pos, and SIIPS-pos responses), (B) negative emotion responses (self-reported valence and PINES responses), and (C) SCRs to CS during acquisition.
A—Pain B—Negative emotions C—arousal
A.I—self-reported pain B.I—self-reported valence
b 95% CI b 95% CI
Pain 0.59 −0.33, 1.53 0.73 −0.22, 1.69
Film 0.74 −0.19, 1.65 3.62 2.65, 4.69
Time_acq12 −0.39 −1.45, 0.65 −1.09 1.84,0.37
Pain × film −0.51 −1.76, 0.75 −1.37 2.49,0.24
Pain × time_acq12 3.85 2.49, 5.24 2.65 −1.53, 3.86
Film × time_acq12 2.60 1.27, 3.92 2.09 0.90, 3.29
Pain × film × time_acq12 −2.73 4.46,1.01 −2.25 3.66,0.86
R2 (95% CI) 0.58 0.52, 0.63 0.73 0.66, 0.79
A.II—NPS-pos A.III—SIIPS-pos B.II—PINES SCRs
b 95% CI b 95% CI b 95% CI b 95% CI
Pain −0.02 −0.17, 0.13 0.06 −0.22, 0.35 0.10 −0.15, 0.35 0.05 −0.09, 0.21
Film 0.06 −0.09, 0.22 0.15 −0.15, 0.45 0.01 −0.22, 0.33 0.06 −0.07, 0.20
Time_acq1 −0.10 −0.39, 0.19 −0.09 −0.39, 0.22 −0.24 −0.52, 0.03 0.26 0.08, 0.45
Time_acq2 −0.12 −0.46, 0.21 −0.03 −0.37, 0.32 −0.50 −0.80, −0.19 0.03 −0.14, 0.21
Pain × film −0.01 −0.22, 0.20 −0.18 −0.58, 0.24 −0.09 −0.43, 0.25 −0.10 −0.30, 0.08
Pain × time_acq1 0.46 0.09, 0.82 0.35 −0.06, 0.76 −0.05 −0.41, 0.31 0.19 −0.05, 0.43
Pain × time_acq2 0.53 0.13, 0.95 0.34 −0.08, 0.75 0.65 0.29, 1.01 0.17 −0.05, 0.38
Film × time_acq1 0.02 −0.41, 0.44 −0.02 −0.45, 0.41 0.18 −0.20, 0.57 0.06 −0.19, 0.30
Film × time_acq2 0.16 −0.27, 0.59 0.15 −0.31, 0.60 0.57 0.20, 0.92 0.07 −0.14, 0.27
Pain × film × time_acq1 −0.45 −1.00, 0.12 −0.34 0.93, 0.26 −0.25 −0.81, 0.32 −0.18 −0.51, 0.15
Pain × film × time_acq2 −0.64 −1.20, −0.10 −0.35 −0.96, 0.25 −0.88 −1.32, −0.44 −0.08 −0.36, 0.21
R2 (95% CI) 0.83 0.78, 0.87 0.34 0.24, 0.45 0.61 0.49, 0.71 0.49 0.34, 0.61
Coefficients are considered significantly different from zero if the corresponding 95% CI does not contain zero and are highlighted in bold. The factor time yields 2 levels for analyses on self-reported pain (habituation and acq12 = full acquisition [rating was at the end of acquisition]) and 3 levels for NPS-pos, SIIPS-pos, PINES, and SCR analyses (habituation, acq1 = first acquisition half, acq2 = second acquisition half). Time was centered on habituation scores across analyses. For improved readability, we do not display intercepts.
acq, acquisition; CS, conditioned stimuli; NPS-pos, neurological pain signature responses based on positive weights only; PINES, picture-induced negative affect signature; SCRs, skin conductance responses; SIIPS-pos, stimulus intensity–independent pain signature-1 based on positive weights only.

F3
Figure 3.:
Fitted (noncentered and nonstandardized) values of regressions estimating conditioned responses (CRs) to each CS condition during acquisition. Error bars represent 95% CIs, and asterisks mark statistical significance. Panel (A) depicts pain-CRs: self-reported pain (0 = not painful, 10 = maximal bearable), NPS-pos, and SIIPS responses. Panel (B) depicts negative emotion–CRs: valence ratings (0 = very pleasant, 10 = very unpleasant) and PINES responses. Panel (C) depicts SCRs measured in microSiemens (μS). acq, acquisition with trials averaged over both acquisition halves; acq1, first acquisition half; acq2, second acquisition half; hab, habituation; NPS-pos, neurological pain signature responses based on positive weights only; PINES, picture-induced negative affect signature; SCRs, skin conductance responses; SIIPS-pos, stimulus intensity–independent pain signature based on positive weights only.
3.2.1.2. Neurological pain signature–pos

Results on NPS-pos responses to CS revealed significant pain × time interactions (ie, pain × time_acq1 and pain × time_acq2 interactions, see Table 2A.II for regression coefficients and 95% CIs). As displayed in Figure 3A (middle) and corroborated by separate post hoc analyses on NPS-pos responses to CS within each study time point (habituation/acq1 + acq2), these interactions suggested that participants showed greater NPS-pos responses to CSpain than CSnopain over both acquisition halves (b = 0.38, 95% CI = [0.16, 0.61]) but not during habituation (b = −0.02, 95% CI = [−0.16, 0.12]). Nonsignificant pain × time (acq1 vs acq2) interactions suggested that this main effect of pain held across both acq1 and acq2 (b = 0.02, 95% CI = [−0.42, 0.46]).

Film modulated the pain × time_acq2 interaction (Table 2A.II). Post hoc analyses on NPS-responses revealed a pain × film interaction during acq2 (b = −0.50, 95% CI = [−0.98, −0.04]), which was associated with some uncertainty during acq1 (b = −0.35, 95% CI = [−0.79, 0.07]). Follow-up post hoc analyses on the pain × film interaction during acquisition phases suggested that participants showed greater NPS-pos responses to CSpain than CSnopain within the neutral film condition (acq1: b = 0.36, 95% CI = [0.05, 0.66], acq2: b = 0.42, 95% CI = [0.05, 0.77]; see upper panel of Fig. 3A, middle plot) but not within the aversive film condition (acq1: b = 0.01, 95% CI = [−0.31, 0.32], acq2: b = −0.08, 95% CI = [−0.43, 0.28]; see lower panel of Fig. 3A, middle plot). During habituation, the pain × film interaction was nonsignificant (b = −0.01, 95% CI = [−0.19, 0.17]).

3.2.1.3. Stimulus intensity–independent pain signature–pos

Stimulus intensity–independent pain signature–pos results largely mirrored those obtained in NPS-pos analyses but were associated with more statistical uncertainty and were nonsignificant. See Figure 3A.III and Table2A.III for regression coefficients and 95% CIs.

3.2.2. Negative emotions

3.2.2.1. Self-report

Analyses suggested a significant interaction between pain × film × time (Table 2B.I). As can be seen from Figure 3B (left) and suggested by post hoc tests within each film condition, this interaction concerned participants' ratings to neutral film CS: after acquisition, participants rated the neutral film signaling CSpain as more unpleasant than the neutral film signaling CSnopain (b = 1.91, 95% CI = [1.15, 2.69]). During habituation, this difference was nonsignificant (b = 0.46, 95% CI = [−0.29, 1.29]). Within the aversive film condition, valence ratings did not significantly differ between CSpain and CSnopain conditions within the aversive film condition at habituation (b = −0.61, 95% CI = [−0.61, 1.29]) or at acquisition (b = 0.14, 95% CI = [−0.43, 0.68]).

Note, however, that, possibly because CS depicted contextual cues from our film clips (oxygen bottle and tunnel for aversive film clips/basketball and pier for neutral film clips), participants already perceived aversive film CS as more unpleasant during habituation in both no pain (b = 2.71, 95% CI = [1.79, 3.96]) and pain conditions (b = 2.14, 95% CI = [1.27, 3.23]).

3.2.2.2. Picture-induced negative affect signature

We observed a significant interaction between pain × film × time_acq2 (Table 2B.II). As suggested by Figure 3B (right) and post hoc tests within each film condition and acquisition halve, PINES responses were enhanced to CSpain vs CSnopain within the neutral film condition in the second half of the acquisition (b = 0.56, 95% CI = [0.22, 0.89]); this difference between CSpain vs CSnopain within the neutral film condition was, however, nonsignificant within the first acquisition half (b = −0.04, 95% CI = [−0.36, 0.27]). Furthermore, PINES responses did not significantly differ between CSpain and CSnopain conditions within the aversive film condition in the first (b = −0.26, 95% CI = [−0.61, 0.10]) and second half (b = −0.20, 95% CI = [−0.50, 0.11]) of the acquisition.

3.2.3. Arousal (skin conductance responses)

Skin conductance responses to CS were generally greater during acq1 but not significantly modulated by pain or film during acquisition, see Table 2C and Figure 3C.

3.3. Conditioned response retention

3.3.1. Pain

3.3.1.1. Self-reported pain

Results on self-reported pain to CS during MTT revealed that participants reported more pain when re-exposed to CSpain than when exposed to CSnopain, and this effect was modulated by film (see left Fig. 4A). As suggested by post hoc tests within each film condition, participants still reported more pain to the CSpain than to the CSnopain within the neutral film condition (b = 2.49, 95% CI = [1.18, 3.90]), and such differential pain responses were nonsignificant within the aversive film condition (b = 0.11, 95% CI = [−1.01, 1.21]). These effects were similar between early and late MTT blocks, as indicated by nonsignificant interactions between pain, time, and film. See Table 3A.I for regression coefficients and corresponding 95% CIs.

F4
Figure 4.:
Fitted (noncentered and nonstandardized) values of regressions estimating conditioned responses (CRs) to each CS condition during MTT. Error bars represent 95% CIs, and asterisks mark statistical significance. Panel (A) depicts pain-CRs: self-reported pain (0 = not painful, 10 = maximal bearable) left and NPS-pos responses right. Panel (B) depicts negative emotion–CRs: PINES responses. Panel (C) depicts SCRs measured in microSiemens (μS). MTT, memory-triggering task; MTT1, early MTT; MTT2, late MTT; NPS-pos, neurological pain signature responses based on positive weights only; PINES, picture-induced negative affect signature; SCRs, skin conductance responses; SIIPS-pos, stimulus intensity–independent pain signature based on positive weights only.
Table 3 - Effects of pain, film, and time on (A) pain responses (self-reported pain and neurological pain signature-pos responses), (B) negative emotion responses (PINES responses), and (C) SCRs to CS during MTT.
A—Pain B—Negative emotion (PINES)
A.I—Self-reported pain
b 95% CI b 95% CI
Pain 2.55 1.36, 3.80 0.09 −0.25, 0.43
Film 2.26 1.02, 3.51 −0.10 −0.41, 0.21
Time_MTT2 −0.70 −1.63, 0.24 −0.17 −0.50, −0.15
Pain × film −2.41 3.72,1.16 0.04 −0.41, 0.21
Pain × time_MTT2 0.28 −0.92, 1.46 0.09 −0.37, 0.53
Film × time_MTT2 0.07 −1.18, 1.31 0.29 0.16, 0.73
Pain × film × time_mtt2 −0.06 −1.68, 1.57 −0.23 −0.85, 0.42
R2 (95% CI) 0.77 0.72, 0.81 0.20 0.10, 0.33
A.II—NPS-pos A.III—SIIPS-pos C—Arousal (SCRs)
b 95% CI b 95% CI b 95% CI
Pain 0.15 −0.16, 0.45 0.13 −0.19, 0.44 0.12 −0.04, 0.29
Film −0.15 −0.45, 0.15 −0.16 −0.48, 0.16 0.07 −0.09, 0.24
Time_MTT2 0.09 −0.21, 0.39 0.18 −0.14, 0.49 0.01 −0.16, 0.19
Pain × film 0.23 −0.18, 0.65 0.16 −0.29, 0.59 −0.13 −0.35, 0.09
Pain × time_MTT2 −0.39 −0.85, 0.09 −0.43 −0.90, 0.03 −0.05 −0.27, 0.16
Film × time_MTT2 0.10 −0.33, 0.523 0.15 −0.21, 0.61 −0.01 −0.23, 0.21
Pain × film × time_MTT2 0.14 −0.51, 0.81 0.08 −0.59, 0.76 0.08 −0.20, 0.37
R2 (95% CI) 0.23 0.09, 0.35 0.21 0.11, 0.33 0.01 0.01, 0.03
Coefficients are considered significantly different from zero if the corresponding 95% CI does not contain zero and are highlighted in bold. The factor time was centered on MTT1. For improved readability, we do not display intercepts.
CS, conditioned stimuli; MTT, memory-triggering task; MTT1, early MTT; MTT2, late MTT; NPS-pos, neurological pain signature responses based on positive weights only; PINES, picture-induced negative affect signature; SCRs, skin conductance responses; SIIPS-pos, stimulus intensity–independent pain signature-1 based on positive weights only.

3.3.1.2. Neurological pain signature–pos

We found no significant differences between NPS-pos responses to CSpain and CSnopain. This held across film conditions and both MTT blocks, as suggested by nonsignificant pain × film, pain × time, and pain × film × time interactions (Table 3A.II). Yet, considering the pattern of results displayed in Figure 4A (middle plot) and the uncertainty associated with the pain × time interaction (95% CIs = −0.85, 0.09), we examined in an exploratory fashion whether NPS-pos responses differed between CSpain and CSnopain within the aversive film condition. NPS-pos responses during early MTT were significantly enhanced to CSpain vs CSnopain during the aversive film condition (b = 0.38, 95% CI = [0.09, 0.66]). The difference between CSpain vs CSnopain within the aversive film condition was nonsignificant during late MTT (b = 0.13, 95% CI = [−0.25, 0.51]) or within the neutral film condition during early (b = 0.14, 95% CI = [−0.14, 0.44]) or late (b = 0.13, 95% CI = [−0.92, 0.87]) MTT.

3.3.1.3. Stimulus intensity–independent pain signature–pos

During MTT, SIIPS-pos results again largely reflected NPS-pos analyses but were associated with more statistical uncertainty and were nonsignificant, see Figure 4A (right plot) and Table 3A.III for details. As in NPS-pos analyses, given the uncertainty associated with the pain × time interaction (95% CIs = −0.90, 0.03), we explored whether SIIPS-pos responses differed between CSpain vs CSnopain during the aversive film condition. Although SIIPS-pos responses during early MTT were descriptively enhanced to CSpain vs CSnopain during the aversive film condition, this difference was nonsignificant as the corresponding 95% CI contained zero (b = 0.28, 95% CI = [0.02, 0.58]).

3.3.2. Negative emotion (picture-induced negative affect signature)

We observed no significant effects of pain, film, time, or any interactions between factors on PINES responses to CS during MTT. See Table 3B and Figure 4B for details.

3.3.3. Arousal (skin conductance responses)

There were no significant effects of pain, film, time, or any interactions between factors on SCRs to CS during MTT See Table 3C and Figure 4C for details.

3.3.4. Daily-life pain intrusions

Over the 7 days after TPC, participants reported on average 1.50 (SD = 2.23) spontaneously occurring pain intrusions during daily life. Participants rated these pain intrusions only as mildly distressing (M = 21.22, SD = 19.45) and strong (M = 24.37, SD = 23.01).

Mirroring the pain induced by the electrocutaneous stimulation on the left calf during the TPC paradigm, participants described pain intrusions during daily life as not only stinging or tingling sensations (48.2%; eg, stinging in my left calf, just like in the MRI, but weaker) but also as other nonspecific painful bodily sensations (9.6%; eg, pain in left calf while playing volleyball) and nonpainful bodily sensations (9.6%; eg, leg cramp). Some pain intrusions also consisted of a combination of painful or nonpainful sensations (13.2%; burning sensation on calf and cramping) or were not further specified (10.3%).

Some participants (37.2%) reported a trigger for their pain intrusions: these included audiovisual input (34.4%; eg, beach film scene, seeing the diary application on my smartphone), painful bodily sensations (12.5%; eg, epilating legs), nonpainful bodily sensations (12.5%; eg, changing position while lying on the couch), thoughts (25%; eg, thinking about the study), and a combination between sensations (12.5%) and other cues (3.1%).

3.4. Relationship between self-reported pain and multivariate brain markers of pain (neurological pain signature–pos, stimulus intensity–independent pain signature–pos)

3.4.1. Acquisition

3.4.1.1. Neurological pain signature–pos

Analyses suggested an interaction between NPS-pos and pain (b = 0.64, 95%-CI = [−0.02, 1.29]; note that because confidence intervals contained zero, these results are associated with some uncertainty): follow-up post hoc analyses within each pain condition suggested that NPS-pos responses and self-reported pain to CSpain were positively associated (b = 0.60, 95% CI = [0.18, 1.04]), indicating that individuals who reported more severe pain at acquisition to CSpain also exhibited higher NPS-pos responses to CSpain. Conversely, the relationship between NPS-pos responses and self-reported pain to CSnopain was still positive but nonsignificant (b = 0.16, 95% CI = [−0.05, 0.39]). See Figure 5 for details. During habituation, the association between NPS-pos responses and self-reported pain was also nonsignificant (b = 0.02, 95% CI = [−0.08, 0.14]) and not modulated by the factor pain (b = 0.01, 95% CI = [−0.15, 0.15]).

F5
Figure 5.:
Fitted (noncentered, nonstandardized) values on analyses predicting the relationship between self-reported pain and NPS-pos responses to CSpain vs CSnopain during acquisition. Shaded areas represent 95% credible intervals. For illustrative purposes and better appreciation, plots depict nonmean-centered NPS-pos estimates. Both self-reported pain and NPS-pos responses were corrected for habituation scores. NPS-pos, neurological pain signature with positive weights only.
3.4.1.2. Stimulus intensity–independent pain signature–pos

Analyses revealed a positive yet uncertain and nonsignificant association between SIIPS-pos responses and self-reported pain (b = 0.13, 95% CI = [−0.11, 0.37]); this was the case for both CSpain and CSnopain conditions, as suggested by a nonsignificant interaction between SIIPS-pos and pain (b = 0.54, 95% CI = [−0.15, 1.24]).

3.4.2. Memory-triggering task

3.4.2.1. Neurological pain signature–pos

At MTT, there was no longer a significant association between NPS-pos responses and self-reported pain (b = 0.36, 95% CI = [−0.27, 1.00]), and this held for both CSpain and CSnopain, as suggested by a nonsignificant interaction between NPS-pos responses and pain (b = 0.17, 95% CI = [−0.58, 0.91]).

3.4.2.2. Stimulus intensity–independent pain signature–pos

Finally, we found no significant association between SIIPS-pos responses and self-reported pain at MTT (b = 0.51, 95% CI = [−0.13, 1.16]), and this held for both CSpain and CSnopain, as suggested by a nonsignificant interaction between NPS-pos responses and pain (b = −0.19, 95% CI = [−0.95, 0.57]).

3.5. Relationship between pain-conditioned responses (self-reported pain, neurological pain signature–pos) and daily-life pain intrusions

3.5.1. Pain-conditioned responses within neutral film condition

3.5.1.1. Self-reported pain

Participants with greater differential self-reported pain-CRs to CSpain vs CSnopain within the neutral film condition during acquisition and MTT showed greater pain intrusion severity than participants with lower differential self-reported pain-CRs (acquisition: b = 0.56, 95% CI = [0.17, 0.96]; MTT: b = 0.43, 95% CI = [0.02, 0.85], see Fig. 6A.I and B.I); this held over days, as suggested by the nonsignificant interaction between self-reported pain-CRs and day (acquisition: b = −0.14, 95% CI = [−0.33, 0.04]; MTT: b = −0.17, 95% CI = [−0.37, 0.03]). For probability of pain intrusion absence during daily life, results suggested no significant effects of the magnitude of differential self-reported pain-CRs during acquisition or MTT (acquisition: b = −0.46, 95% CI = [−1.21, 0.25]; MTT: b = −0.11, 95% CI = [−0.83, 0.60]) or interactions between self-reported pain-CRs and day (acquisition: b = 0.16, 95% CI = [−0.09, 0.43]; MTT: b = 0.01, 95% CI = [−0.23, 0.25]).

F6
Figure 6.:
Relationship between spontaneous pain intrusions over days and differential (I) self-reported pain-CRs and (II) NPS-pos–CRs to CSpain vs CSnopain during acquisition (A) and MTT (B). Lines in plots A.I and B.I depict fitted values of regressions' lognormal part (ie, estimating the amount of pain intrusions [pain intrusion severity]; Lines in panel A.II and B.II depict fitted values of the model's hurdle part (ie, estimating the probability of zero pain intrusions [pain intrusion absence]). Shaded areas represent 95% credible intervals. acq, acquisition; CS, conditioned stimulus; CS_AP, CS paired with aversive film + painful stimulation; CS_NnP, CS paired with neutral film clip + no pain stimulation; CS_NP, CS paired with neutral film + painful stimulation; MTT, memory-triggering task; NPS-pos, neurological pain signature responses based on positive weights only.
3.5.1.2. Neurological pain signature–pos responses

We found no significant effects of differential NPS-pos–CRs to CSpain vs CSnopain within the neutral film condition during acquisition or MTT (acquisition: b = 0.11, 95%-CI = [−0.34, 0.57]; MTT: b = −0.11, 95% CI = [−0.57, 0.34]) or interactions between NPS-pos–CRs and day (acquisition: b = −0.07, 95% CI = [−0.32, 0.18]; MTT: b = −0.01, 95% CI = [−0.23, 0.23]) on the severity of pain intrusions during daily life. However, results suggested that participants who acquired greater differential NPS-pos–CRs at acquisition showed a lower probability of pain intrusion absence during daily life (b = −0.84, 95% CI = [−1.68, −0.08]), which was modulated by day (b = 0.33, 95% CI = [0.05, 0.62]): Participants with greater NPS-pos–CRs at acquisition mostly showed a lower probability of pain intrusion absence during the first days after TPC, which dissipated over the course of days. Conversely, participants with lower NPS-pos–CRs at acquisition showed a constant high probability of pain intrusion absence over days, see Figure 6A.II for details. Although following the same pattern as during acquisition, during MTT, the effects of NPS-pos–CRs (b = −0.22, 95%-CI = [−0.94, 0.54]) and interaction effect between NPS-pos–CRs and day (b = 0.19, 95% CI = [−0.07, 0.46]) on pain intrusion absence during daily life were nonsignificant.

3.5.2. Pain-conditioned responses within aversive film condition

3.5.2.1. Self-reported pain

Results on the relationship between differential self-reported pain-CRs within the aversive film condition and daily-life pain intrusions largely yielded an identical pattern as results within the neutral film condition but were associated with uncertainty because zero was included in the respective 95% CIs. Over acquisition and MTT analyses, no significant effects of self-reported pain-CRs (acquisition: b = 0.41, 95% CI = [−0.08, 0.88]; MTT: b = 0.03, 95% CI = [−0.42, 0.47]) or significant interaction effects between self-reported pain-CRs and day (acquisition: b = −0.10, 95%-CI = [−0.26, 0.07]; MTT: b = −0.01, 95%-CI = [−0.42, 0.47]) on pain intrusion severity during daily life were found. Similarly, results suggested no significant main effect of self-reported pain-CRs (acquisition: b = 0.01, 95% CI = [−0.73, 0.75]; MTT: b = 0.26, 95% CI = [−0.45, 1.03]), or significant interaction effects between self-reported pain-CRs and day (acquisition: b = −0.04, 95% CI = [−0.30, 0.22]; MTT: b = −0.19, 95% CI = [−0.42, 0.04]) on daily-life pain intrusion absence during daily life were observed.

3.5.2.2. Neurological pain signature–pos responses

For pain intrusion severity during daily life, analyses on differential NPS-pos–CRs within the aversive film condition over acquisition and MTT suggested no significant main effect of NPS-pos–CRs (acquisition: b = −0.40, 95% CI = [−0.86, 0.09]; MTT: b = 0.15, 95% CI = [−0.44, 0.75]) or significant interaction effects between NPS-pos–CRs and day (acquisition: b = 0.08, 95% CI = [−0.11, 0.27]; MTT: b = −0.11, 95% CI = [−0.30, 0.09]). Of importance, although, results on pain intrusion absence during daily life suggested that whereas NPS-pos–CRs during acquisition did not significantly influence subsequent pain intrusion absence during daily life (b = −0.37, 95% CI = [−1.14, 0.39]), during MTT, NPS-pos–CRs significantly predicted the probability of pain intrusion absence (b = −1.10, 95%-CI = [−2.20, −0.15]), in that participants with stronger differential NPS-pos–CRs to CSpain during the aversive film condition at MTT showed a lower probability of pain intrusion absence during daily life (Fig. 6B.II). These effects held over days (acquisition: b = 0.20, 95% CI = [−0.06, 0.46]; MTT: b = 0.22, 95% CI = [−0.07, 0.53]).

3.5.3. Summary

Overall, significant relationships between pain-CRs and pain intrusions primarily emerged within conditions where participants effectively showed significant differential pain responses to CSpain vs CSnopain (see Sections 3.2.1. and 3.3.1). Specifically, participants who acquired and retained stronger differential self-reported-pain-CRs to CSpain within a neutral affective film context showed greater pain intrusion severity during daily life. When self-reported pain-CRs to CSpain were acquired within the aversive film context, this relationship was nonsignificant but followed a similar pattern. Moreover, NPS-pos–CRs also predicted pain intrusions during daily life: participants who acquired stronger differential NPS-pos–CRs to the CSpain within the neutral film condition showed a lower probability for pain intrusion absence during daily life. Furthermore, during MTT, participants who retained stronger differential NPS-pos–CRs to CSpain within the aversive film, but not within the neutral film condition, showed a lower probability for pain intrusion absence during daily life.

3.6. Exploratory whole-brain analyses

To allow full appreciation of pain conditioning acquisition and retention, we computed an exploratory whole-brain analysis for the effects of pain (CSpain > CSnopain and CSpain < CSnopain). During acquisition, over both neutral film and aversive film conditions, the contrast CSpain > CSnopain revealed no significant activations. The interaction contrast between pain and film (see Supplement Table S3, available at https://links.lww.com/PAIN/B592) further revealed activations in widespread regions involved in salience and pain anticipation or processing including bilateral middle temporal and occipital gyri, bilateral rolandic operculum, bilateral anterior insula, and right supplementary motor area (SMA).8,22,63 When examining the pain contrast separately within the aversive film and neutral film conditions, it became apparent that the main effect of pain in salience and pain-processing regions exclusively originated within the neutral film condition (Fig. 7B). During early and late MTT, there were no statistically significant activation patterns at FDR-corrected P < 0.05.

F7
Figure 7.:
Activation clusters revealed by the whole-brain analyses. Panel (A) = Interaction pain × film during acquisition irrespective of film (F test); Panel (B) = pain > no pain within neutral film condition at acquisition (t test). All clusters were extracted at a threshold of FDR-corrected P < 0.05; k ≥ 5.

4. Discussion

This study investigated whether pain occurs as a CR to cues signaling but not inflicting physical pain. Data largely supported the hypothesis that participants acquire conditioned pain responses (pain-CRs): during acquisition, cues signaling pain (CSpain) elicited more self-reported pain and NPS-pos responses than cues signaling no pain (CSnopain). This was moderated by affective context: conditioned pain responses were weaker (self-reported pain-CRs) or absent (NPS-pos–CRs) during the aversive film condition. Critically, on re-exposure to CS during MTT 24 hours later, participants retained self-reported pain-CRs within the neutral film condition and NPS-pos–CRs within the aversive film condition. Finally, participants showing stronger self-reported pain-CRs and NPS-pos–CRs showed a greater probability and severity of pain intrusions during daily life compared with participants with weaker self-reported pain-CRs and NPS-pos–CRs.

The finding that cues associated with pain merely on a temporal and contextual level61 can elicit pain-CRs extends studies showing that classical conditioning endows CS with the ability to amplify the perception of pain in response to nociceptive stimuli42,43 or reduce pain thresholds to ambiguous at-pain threshold stimuli.54,61,92 Our study experimentally supports that pain may occur as a CR32,56 and complements research showing that, after conditioning, pain-signaling CS elicit a wide range of pain-related fears (eg, fear of movement), which may contribute to the transition from acute to chronic pain.61 Although the distinction between pain-related-fear CRs and pain-CRs needs scrutiny in future research, in this study, we demonstrate that pain-signaling CS not only elicit pain-related fears but also pain itself. Then, we detail the possible dimensions of these pain-CRs.

First, greater pain to CSpain vs CSnopain immediately after the initial learning phase and 24 hours later during MTT suggests that pain experienced in response to a CS is consciously perceived and reported. An important question is to what extent individuals were able to separate CS representations from US representations when rating CS on pain. Especially after the acquisition phase, where 50% of the CS were followed by US, it is possible that participants retrospectively rated the CS-US compound, rather than the CS. However, we believe that this reflects the very nature of conditioning, where an association of the CS and the US is formed in the brain and the CS turns into this compound.51 Agreeingly, the finding that participants continued rating CSpain as more painful than CSnopain even after a series of CS-no US trials during MTT suggests that pain ratings emerged from the CSpain triggering USpain representations (ie, CRs) even in the absence of the USpain.

Second, our data suggest that pain experienced in response to a CS also has a sensory component. Specifically, and mirroring self-reported pain results, participants showed greater NPS-pos responses to CSpain vs CSnopain. In line, whole-brain analyses also revealed widespread activations in regions involved in salience and pain processing8,22,63 including bilateral anterior insula, bilateral rolandic operculum, and right supplementary motor area to CSpain > CSnopain. Furthermore, CSpain > CSnopain revealed activations in occipital regions, which could relate to enhanced sensory processing of threat-related cues.3,59,65

Third, our findings suggest that pain experienced in response to a CS yields a strong negative emotional component. During acquisition, participants reported more unpleasantness and showed greater PINES responses to CSpain vs CSnopain even if the associated affective context was neutral. Of interest, the temporally more fine-grained PINES responses suggest that conditioned negative emotions to pain emerged only later during acquisition. Given that conditioned NPS-pos responses emerged immediately during the first acquisition half, acquisition of the sensory component of pain-CRs might be faster than acquisition of their affective component. Overall, our results support that both sensory and affective components of pain were (at least transiently) conditioned and thereby suggest that pain-CRs reflect pain as an unpleasant sensory and emotional experience.74

Overarchingly, how can we describe the type of pain experienced during a CS, and how is this type of pain different from nociceptive stimulus–evoked pain? The fact that pain-signaling CS elicited not only pain perception but also changes in brain activation patterns signalling nociception during acquisition and MTT suggests that conditioned pain might qualitatively resemble evoked pain, although less intense. The most striking difference may lay in that, as opposed to evoked pain responses, pain-CRs emerged without any nociceptive input. Of importance, the finding that nociceptive input does not seem to be necessary to elicit a pain experience with a sensory and negative emotional component supports the idea that there is no direct one-to-one relationship between nociception and pain. Tentatively, we might thus describe pain-CRs as instances of nociplastic pain with a strong negative emotional component, ie, pain that arises from altered nociception despite no clear evidence of actual or threatened tissue damage causing the activation of peripheral nociceptors or evidence for disease or lesion of the somatosensory system causing the pain.40,44,45

Whether individuals displayed enhanced self-reported pain and NPS-pos responses to CSpain depended on the associated affective background induced by film clips: during both acquisition and MTT, self-reported pain-CRs clearly emerged only within the neutral film condition. Pain conditioning may have been less evident within the aversive film condition because participants not only reported pain to the CSpain but to some extent also to the CSnopain (and USnopain). One possibility is that pain ratings to the aversive film or no pain condition resulted from memory biases or participants generalizing pain from the aversive film or pain to the aversive film or no pain trials. Alternatively, considering that film clips depicted pain in others, participants' pain responses to CSnopain within the aversive film condition could reflect mirroring of another person's pain experience.48 This would reflect the multidimensional nature of pain and support that there is no direct one-to-one relationship between nociception and pain, ie, that pain is also an emotional experience that may occur despite no actual tissue damage is causing activation of peripheral nociceptors.74

Resembling self-reported pain-CRs, NPS-pos and whole-brain differences to CSpain vs CSnopain during acquisition emerged only within the neutral film condition. A likely explanation for this is that the widespread neural activation associated with the highly aversive film64 masked the effects of the painful stimulation. By contrast, during MTT, participants showed greater NPS-pos responses to CSpain vs CSnopain only within the aversive film condition (please note that these results are based on exploratory post hoc tests and thus need to be considered with care). This could suggest that in the background of a highly aversive affective acquisition context, individuals may show no conditioned sensory pain responses in the here-and-now danger situation but do so after the danger has subsided. In line, evolutionary accounts, as well as clinical/empirical observations, suggest that even (or especially) when individuals narrow their attention or shutdown during psychological trauma, they subsequently can show strong reactions (eg, enhanced fear/arousal/intrusions) to stimuli present during the traumatic event.12,21,28,35 In sum, our results may implicate that sensory pain-CRs extending beyond the immediate acquisition phase are more likely if pain conditioning occurred under an aversive than a neutral affective context. Ultimately, this may explain some of the particularly high prevalence of chronic pain in PTSD.1,80,83,84

Finally, our results support that pain intrusion may occur in daily life and participants who demonstrated stronger self-reported pain-CRs and NPS-pos-CRs during acquisition, and partly also during MTT, showed greater probability and severity of pain intrusions. Participants described pain sensations during intrusions similar to those experienced during the experimental pain induction. As expected from a healthy sample in an analogue-trauma situation, participants rated pain intrusions as relatively mild in magnitude. Our experimental findings resonate with clinical observations in PTSD patients, including, for instance, a midwife who, after an unsuccessful resuscitation, kept feeling a painful strain in the hand she used for resuscitating (personal communication) or torture survivors describing pain arising from trauma memories.89 Noteworthy, although we have primarily discussed pain-CRs from the point of view of PTSD research,53 we believe that conditioned pain responses might explain persistent pain not only in PTSD and in patients with a background of childhood traumatization but may also be a promising mechanism explaining part of the etiology of nociplastic pain in other chronic pain conditions.

4.1. Limitations and future directions

A possible limitation of this study is that participants rated CS at the end of acquisition rather than immediately after or during each CS trial. We opted for retrospective ratings because they do not interfere with the evolvement of learning and/or elicit motor cognitive demands that could interfere with neurophysiological measures of pain-CRs.51 However, the fact that we acquired only 1 rating for each CS at the end of acquisition raises the question whether there may have been recall biases. This could imply a lower reliability of the self-reported pain measure and prevents modelling learning rates of pain-CRs. Furthermore, retrospective ratings forbid a direct mapping of self-reported pain-CRs and trial-by-trial acquired NPS-pos–CRs. As such, we cannot be totally sure whether NPS-pos responses indeed reflect participants' pain experience. Noteworthy, although, self-reported pain responses during acquisition significantly correlated with NPS-pos responses measured during each unreinforced CS presentation. Although we may only indirectly infer pain from NPS-pos responses,20 the overlap between retrospective CS pain ratings and trial-by-trial NPS-pos responses at acquisition support that pain ratings indeed reflected pain experienced during CS trials. Future research needs to investigate this (eg, with trial-by-trial CS ratings). Second, the use of maximal bearable (vs. unbearable) pain as the upper limit of the pain ratings might have focused participants' attention more on tolerability than intensity of pain and could have biased the pain ratings. Third, owing to the visual modality of stimuli, we focused only on positive NPS weights.96 Thus, we may not have fully captured brain processes contributing to nociceptive pain. Investigating NPS to CS from other modalities, for instance, somatosensory cues,22,62,73 is therefore a necessary future step in follow-up studies. Overall, SIIPS-pos largely mirrored NPS-pos results but were associated with more statistical uncertainty. This could suggest that extranociceptive brain processes quantified by the SIIPS are not critically involved in pain-CRs or that we have not ideally captured extranociceptive brain processes contributing to pain-CRs.

Our results are nuanced by pain-emotion interactions. We believe that this depicts the intertwined pain-emotion relationship and underlines the need for considering the affective context under which pain conditioning occurs. In this vein, our findings question to what extent pain and emotion can be separated as constructs: For instance, does pain just comprise an emotional component (eg, fear), or do pain and emotion have the same underlying mechanisms, overlapping in such a manner that their distinction is superfluous? Extending this to the construct of pain-related fear: how are pain-CRs different to pain-related-fear CRs? Fear and pain are both potent signals to adaptive behaviors (eg, fight-freeze-flight responses) and modulated by factors such as cognitions and expectations. Considering a recent perspective on this,33 we tentatively suggest that pain-related fear-CRs and pain-CRs differ mostly on a sensory level in that pain-CRs elicit nociceptive signaling in the brain, are consciously experienced as painful, and can consequently be sensed by and referred to localizations within the body. Although emotions seem to have differential somatic referents in the body, it might be the pattern of somatosensory referents and their associated neural manifestation that differentiates between painful and emotional experiences.33 Clearly, the pain-emotion relationship remains poorly understood and needs to be scrutinized in future pain conditioning research.

5. Conclusion

Study results suggest that classical conditioning could be a direct mechanism underlying pain experiences: The fact that participants acquired and, to some extent, also maintained pain-CRs implies that temporocontextually associating CS with nociceptive stimulation can endow these cues to elicit pain in the absence of noxious stimulation. Furthermore, stronger pain-CRs seem to enhance the probability and severity of experiencing spontaneous pain intrusions during daily life. Experiencing pain as a CR to cues associated with pain could thus underly the commonly observed pain re-experiencing in PTSD. Ultimately, pain re-experiencing through conditioning processes may perpetuate pain long beyond tissue healing and thereby constitute a mechanism explaining the high comorbidity between chronic pain and PTSD. In this case, behavioral interventions such as exposure-based therapies targeting extinction of CRs and cognitive interventions such as trigger discrimination training24 may prove particularly beneficial in pain management.10

Conflict of interest statement

The authors have no conflicts of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at https://links.lww.com/PAIN/B592.

Supplemental video content

A video abstract associated with this article can be found at https://links.lww.com/PAIN/B593.

Acknowledgments

The authors thank Dr. Wager for sharing the NPS pattern weights. The authors also thank Michael Martini, Johanna Lohse, Julia Lackner, and Natalie Stöckel for their help in recruitment of participants, data collection, and preliminary analyses. L. K. Franke was supported by the Doctoral School Cognitive Neuroscience at Paris Lodron University of Salzburg. S. K. Danböck was supported by the Doctoral College Imaging the Mind (FWF; W1233-B; Project Principal Investigator: F. H. Wilhelm). H. Flor was supported by a grant of the Deutsche Forschungsgemeinschaft (SFB1158/B03 to Frauke Nees and H. Flor).

References

[1]. Afari N, Ahumada SM, Wright LJ, Mostoufi S, Golnari G, Reis V, Cuneo JG. Psychological trauma and functional somatic syndromes: a systematic review and meta-analysis. Psychosom Med 2014;76:2–11.
[2]. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.), 2013. Available at: https://doi.org/10.1176/appi.books.9780890425596.
[3]. Antov MI, Plog E, Bierwirth P, Keil A, Stockhorst U. Visuocortical tuning to a threat-related feature persists after extinction and consolidation of conditioned fear. Sci Rep 2020;10:3926.
[4]. Arnaudova I, Hagenaars MA. Lights … action: comparison of trauma films for use in the trauma film paradigm. Behav Res Ther 2017;93:67–77.
[5]. Atlas LY, Wager TD. How expectations shape pain. Neurosci Lett 2012;520:140–8.
[6]. Barr DJ, Levy R, Scheepers C, Tily HJ. Random effects structure for confirmatory hypothesis testing: keep it maximal. J Mem Lang 2013;68:255–78.
[7]. Betancourt M. A conceptual introduction to Hamiltonian Monte Carlo. arXiv 2017. doi: https://doi.org/10.48550/arXiv.1701.02434.
[8]. Biggs EE, Timmers I, Meulders A, Vlaeyen JWS, Goebel R, Kaas AL. The neural correlates of pain-related fear: a meta-analysis comparing fear conditioning studies using painful and non-painful stimuli. Neurosci Biobehav Rev 2020;119:52–65.
[9]. Blechert J, Peyk P, Liedlgruber M, Wilhelm FH. ANSLAB: integrated multichannel peripheral biosignal processing in psychophysiological science. Behav Res Methods 2016;48:1528–45.
[10]. Bosco MA, Gallinati JL, Clark ME. Conceptualizing and treating comorbid chronic pain and PTSD. Pain Res Treat 2013;2013:174728–10.
[11]. Boucsein W, Fowles DC, Grimnes S, Ben-Shakhar G, Roth WT, Dawson ME, Filion DL. Publication recommendations for electrodermal measurements. Psychophysiology 2012;49:1017–34.
[12]. Brewin CR, Holmes EA. Psychological theories of posttraumatic stress disorder. Clin Psychol Rev 2003;23:339–76.
[13]. Bürkner PC. Advanced Bayesian multilevel modeling with the R package brms. R J 2018;10:395–411.
[14]. Bürkner PC. brms: an R package for Bayesian multilevel models using Stan. J Stat Softw 2017;80:1–28.
[15]. Bürkner P-C, Vuorre M. Ordinal regression models in psychology: a tutorial. Adv Methods Practices Psychol Sci 2019;2:77–101.
[16]. de C Williams AC, Davies HT, Chadury Y. Simple pain rating scales hide complex idiosyncratic meanings. PAIN 2000;85:457–63.
[17]. Carpenter B, Gelman A, Hoffman MD, Lee D, Goodrich B, Betancourt M, Brubaker M, Guo J, Li P, Riddell A. Stan: A Probabilistic Programming Language. Journal of Statistical Software 2020;76(1):1–32. https://doi.org/10.18637/jss.v076.i01
[18]. Carter T. Coach Carter. Paramount Pictures: USA, 2005.
[19]. Chang LJ, Gianaros PJ, Manuck SB, Krishnan A, Wager TD. A sensitive and specific neural signature for picture-induced negative affect. PLoS Biol 2015;13:1–28.
[20]. Davis KD, Flor H, Greely HT, Iannetti GD, MacKey S, Ploner M, Pustilnik A, Tracey I, Treede RD, Wager TD. Brain imaging tests for chronic pain: medical, legal and ethical issues and recommendations. Nat Rev Neurol 2017;13:624–38.
[21]. Defrin R, Schreiber S, Ginzburg K. Paradoxical pain perception in posttraumatic stress disorder: the unique role of anxiety and dissociation. J Pain 2015;16:961–70.
[22]. Diesch E, Flor H. Alteration in the response properties of primary somatosensory cortex related to differential aversive Pavlovian conditioning. PAIN 2007;131:171–80.
[23]. Duff EP, Moultrie F, van der Vaart M, Goksan S, Abos A, Fitzgibbon SP, Baxter L, Wager TD, Slater R. Inferring pain experience in infants using quantitative whole-brain functional MRI signatures: a cross-sectional, observational study. Lancet Digital Health 2020;2:e458–67.
[24]. Ehlers A, Clark DM. UKPMC Funders Group Posttraumatic stress disorder: the development of effective psychological treatments. Psychology 2011;62:11–18.
[25]. Ehlers A, Hackmann A, Steil R, Clohessy S, Wenninger K, Winter H. The nature of intrusive memories after trauma: the warning signal hypothesis. Behav Res Ther 2002;40:995–1002.
[26]. Figgis M. Mr. Jones. Rastar Productions: USA, 1993.
[27]. Flor H, Birbaumer N. Acquisition of chronic pain. Psychophysiological mechanisms. APS J 1994;3:119–27.
[28]. Franke LK, Rattel JA, Miedl SF, Danböck SK, Bürkner P-C, Wilhelm FH. Intrusive memories as conditioned responses to trauma cues: an empirically supported concept? Behav Res Ther 2021;143:103848.
[29]. Gelman A, Carlin JB, Stern HS, Dunson DB, Vehtari A, Rubin DB. Bayesian data analysis. Chapman and Hall/CRC, 2013. DOI: https://doi.org/10.1201/b16018
[30]. Gelman A, Goodrich B, Gabry J, Vehtari A. R-squared for Bayesian regression models. Am Stat 2019;73:307–9.
[31]. Gelman A, Hill J. Data analysis using regression and multilevel/hierarchical models. Cambridge: Cambridge University Press, 2006.
[32]. Gentry WD, Bernal GAA. Chronic pain. In: Williams R, Gentry WD, editors. Behavioral approaches to medical treatment. Cambridge: Ballinger, 1977. p. 173–82.
[33]. Gilam G, Gross JJ, Wager TD, Keefe FJ, Mackey SC. What is the relationship between pain and emotion? Bridging constructs and communities. Neuron 2020;107:17–21.
[34]. Hackmann A, Ehlers A, Speckens A, Clark DM. Characteristics and content of intrusive memories in PTSD and their changes with treatment. J Trauma Stress 2004;17:231–40.
[35]. Hagenaars MA, van Minnen A, Holmes EA, Brewin CR, Hoogduin KAL. The effect of hypnotically induced somatoform dissociation on the development of intrusions after an aversive film. Cogn Emot 2008;22:944–63.
[36]. Harvie DS, Meulders A, Madden VJ, Hillier SL, Peto DK, Brinkworth R, Moseley GL. When touch predicts pain: predictive tactile cues modulate perceived intensity of painful stimulation independent of expectancy. Scand J Pain 2016;11:11–18.
[37]. Henry JD, Crawford JR. The short-form version of the depression anxiety stress scales (DASS-21): construct validity and normative data in a large non-clinical sample. Br J Clin Psychol 2005;44:227–39.
[38]. Horowitz M, Wilner N, Alvarez W. Impact of event scale: a measure of subjective stress. Psychosomatic Med 1979;41:209–18.
[39]. De Houwer J. Revisiting classical conditioning as a model for anxiety disorders: a conceptual analysis and brief review. Behav Res Ther 2020;127:103558.
[40]. IASP. IASP: taxonomy. International Association for the Study of Pain, 2021. Available at: https://www.iasp-pain.org/terminology?navItemNumber=576#Pain. Accessed January 2021.
[41]. James EL, Lau-Zhu A, Clark IA, Visser RM, Hagenaars MA, Holmes EA. The trauma film paradigm as an experimental psychopathology model of psychological trauma: intrusive memories and beyond. Clin Psychol Rev 2016;47:106–42.
[42]. Jepma M, Koban L, van Doorn J, Jones M, Wager TD. Behavioural and neural evidence for self-reinforcing expectancy effects on pain. Nat Hum Behav 2018;2:838–55.
[43]. Jepma M, Wager TD. Conceptual conditioning. Psychol Sci 2015;26:1728–39.
[44]. Kosek E, Clauw D, Nijs J, Baron R, Gilron I, Harris RE, Mico JA, Rice ASC, Sterling M. Chronic nociplastic pain affecting the musculoskeletal system: clinical criteria and grading system. PAIN 2021;162:2629–34.
[45]. Kosek E, Cohen M, Baron R, Gebhart GF, Mico JA, Rice ASC, Rief W, Sluka AK. Do we need a third mechanistic descriptor for chronic pain states? PAIN 2016;157:1382–6.
[46]. Kragel PA, Koban L, Barrett LF, Wager TD. Representation, pattern information, and brain signatures: from neurons to neuroimaging. Neuron 2018;99:257–73.
[47]. Krishnan A, Woo CW, Chang LJ, Ruzic L, Gu X, López-Solà M, Jackson PL, Pujol J, Fan J, Wager TD. Somatic and vicarious pain are represented by dissociable multivariate brain patterns. eLife 2016;5:1–42.
[48]. Lamm C, Decety J, Singer T. Meta-analytic evidence for common and distinct neural networks associated with directly experienced pain and empathy for pain. NeuroImage 2011;54:2492–502.
[49]. Laux L, Glanzmann P, Schaffner P, Spielberger C. Das State-Trait Angstinventar. Weinheim (GER): Beltz, 1981.
[50]. Ning Li N, Elashoff DA, Robbins WA, Lin Xun L. A hierarchical zero-inflated log-normal model for skewed responses. Stat Methods Med Res 2011;20:175–89.
[51]. Lonsdorf TB, Menz MM, Andreatta M, Fullana MA, Golkar A, Haaker J, Heitland I, Hermann A, Kuhn M, Kruse O, Meir Drexler S, Meulders A, Nees F, Pittig A, Richter J, Römer S, Shiban Y, Schmitz A, Straube B, Vervliet B, Wendt J, Baas JMP, Merz CJ. Don't fear ‘fear conditioning’: methodological considerations for the design and analysis of studies on human fear acquisition, extinction, and return of fear. Neurosci Biobehavioral Rev 2017;77:247–85.
[52]. López-Solà M, Woo CW, Pujol J, Deus J, Harrison BJ, Monfort J, Wager TD. Towards a neurophysiological signature for fibromyalgia. PAIN 2017;158:34–47.
[53]. Macdonald B, Salomons TV, Meteyard L, Whalley MG. Prevalence of pain flashbacks in post-traumatic stress disorder arising from exposure to multiple traumas or childhood traumatization. Can J Pain 2018;2:48–56.
[54]. Madden VJ, Bellan V, Russek LN, Camfferman D, Vlaeyen JWS, Moseley GL. Pain by association? Experimental modulation of human pain thresholds using classical conditioning. J Pain 2016;17:1105–15.
[55]. Madden VJ, Harvie DS, Parker R, Jensen KB, Vlaeyen JWS, Moseley GL, Stanton TR, Read M. Can pain or hyperalgesia be a classically conditioned response in humans? A systematic review and meta-analysis. Pain Med 2016;17:1094–111.
[56]. Madden VJ, Moseley GL. Do clinicians think that pain can be a classically conditioned response to a non-noxious stimulus? Man Ther 2016;22:165–73.
[57]. Maercker A, Schützwohl M. Erfassung von psychischen Belastungsfolgen: die Impact of Event Skala-revidierte Version (IES-R) [Assessment of post-traumatic stress reactions: the Impact of Event Scale-Revised (IES-R)]. Diagnostica 1998;44:130–41.
[58]. Marks EH, Franklin AR, Zoellner LA, Marks EH, Franklin AR, Zoellner LA. Can't get it out of my mind : a systematic review of predictors of intrusive memories of distressing events can't get it out of my mind: a systematic review of predictors of intrusive memories of distressing events. Psychol Bull 2018;144:584–640.
[59]. McGann JP. Associative learning and sensory neuroplasticity: how does it happen and what is it good for? Learn Mem 2015;22:567–76.
[60]. Merskey H, et al. Pain terms: a list with definitions and notes on usage. Recommended by the IASP Subcommittee on Taxonomy. PAIN 1979;6:249.
[61]. Meulders A. Fear in the context of pain: lessons learned from 100 years of fear conditioning research. Behav Res Ther 2020;131:103635.
[62]. Meulders A, Vansteenwegen D, Vlaeyen JW. The acquisition of fear of movement-related pain and associative learning: a novel pain-relevant human fear conditioning paradigm. PAIN 2011;152:2460–9.
[63]. Michelle Welman FHS, Smit AE, Jongen JLM, Tibboel D, van der Geest JN, Holstege JC. Pain experience is somatotopically organized and overlaps with pain anticipation in the human cerebellum. Cerebellum 2018;17:447–60.
[64]. Miedl SF, Wegerer M, Kerschbaum H, Blechert J, Wilhelm FH. Neural activity during traumatic film viewing is linked to endogenous estradiol and hormonal contraception. Psychoneuroendocrinology 2018;87:20–6.
[65]. Moratti S, Keil A. Not what you expect: experience but not expectancy predicts conditioned responses in human visual and supplementary cortex. Cereb Cortex 2009;19:2803–9.
[66]. Morgan L, Aldington D. Comorbid chronic pain and post-traumatic stress disorder in UK veterans: a lot of theory but not enough evidence. Br J Pain 2020;14:256–62.
[67]. Moseley GL, Vlaeyen JW. Beyond nociception: the imprecision hypothesis of chronic pain. PAIN 2015;156:35–8.
[68]. Mouraux A, Marot E, Legrain V. Short trains of intra-epidermal electrical stimulation to elicit reliable behavioral and electrophysiological responses to the selective activation of nociceptors in humans. Neurosci Lett 2014;561:69–73.
[69]. MultiMediaTechnology. smarthealthcheck. 2013. Available at: https://www.smarthealth.at/project/psydiary-2/. Accessed June 28, 2021.
[70]. Nilges P, Essau C. Die Depressions-Angst-Stress-Skalen: der DASS – ein Screeningverfahren nicht nur für Schmerzpatienten. Schmerz 2015;10:649–57.
[71]. Noé G. Irreversible. StudioCanal: France, 2002.
[72]. Oh JP, Han JH. A critical role of hippocampus for formation of remote cued fear memory. Mol Brain 2020;13:112.
[73]. De Peuter S, Van Diest I, Vansteenwegen D, Van Den Bergh O, Vlaeyen JWS. Understanding fear of pain in chronic pain: interoceptive fear conditioning as a novel approach. Eur J Pain 2011;15:889–94.
[74]. Raja SN, Carr DB, Cohen M, Finnerup NB, Flor H, Gibson S, Keefe FJ, Mogil JS, Ringkamp M, Sluka KA, Song XJ, Stevens B, Sullivan MD, Tutelman PR, Ushida T, Vader K. The revised International Association for the Study of Pain definition of pain: concepts, challenges, and compromises. PAIN 2020;161:1976–82.
[75]. Rance M, Ruttorf M, Nees F, Schad LR, Flor H. Neurofeedback of the difference in activation of the anterior cingulate cortex and posterior insular cortex: two functionally connected areas in the processing of pain. Front Behav Neurosci 2014;8:357.
[76]. Rattel JA, Wegerer M, Miedl SF, Blechert J, Grünberger LM, Craske MG, Wilhelm FH. Peritraumatic unconditioned and conditioned responding explains sex differences in intrusions after analogue trauma. Behav Res Ther 2019;116:19–29.
[77]. RCore Team. R: a language and environment for statistical computing. Foundation for Statistical Computing, 2019. Available at: http://www.r-project.org/index.html. Accessed January 2021.
[78]. Reddan MC, Wager TD. Modeling pain using fMRI: from regions to Biomarkers. Neurosci Bull 2018;34:208–15.
[79]. Rief W, Hiller W. A new approach to the assessment of the treatment effects of somatoform disorders. Psychosomatics 2003;44:492–8.
[80]. Roy-Byrne P, Smith WR, Goldberg J, Afari N, Buchwald D. Post-traumatic stress disorder among patients with chronic pain and chronic fatigue. Psychol Med 2004;34:363–8.
[81]. RStudio Team. RStudio: Integrated development for R [Computer software], 2020. Available at: http://www.rstudio.com/. Accessed at 01.07.2021.
[82]. Sharp TJ, Harvey AG. Chronic pain and posttraumatic stress disorder: mutual maintenance? Clin Psychol Rev 2001;21:857–77.
[83]. Siqveland J, Hussain A, Lindstrøm JC, Ruud T, Hauff E. Prevalence of posttraumatic stress disorder in persons with chronic pain: a meta-analysis. Front Psychiatry 2017;8:164.
[84]. Siqveland J, Ruud T, Hauff E. Post-traumatic stress disorder moderates the relationship between trauma exposure and chronic pain. Eur J Psychotraumatol 2017;8:1375337.
[85]. Spielberger CD, Gorsuch RL, Lushene RE. Manual for the state-trait anxiety inventory. Palo Alto, CA: Consulting Psychologists Press, 1970.
[86]. Steil R, Ehlers A. Dysfunctional meaning of posttraumatic intrusions in chronic PTSD. Behav Res Ther 2000;38:537–58.
[87]. Streb M, Conway MA, Michael T. Conditioned responses to trauma reminders: how durable are they over time and does memory integration reduce them? J Behav Ther Exp Psychiatry 2017;57:88–95.
[88]. Sun Y, Gooch H, Sah P. Fear conditioning and the basolateral amygdala. F1000Res 2020;9:F1000. Faculty Rev-53.
[89]. Taylor B, Carswell K, Williams AC. The interaction of persistent pain and post-traumatic re-experiencing: a qualitative study in torture survivors. J Pain Symptom Manage 2013;46:546–55.
[90]. Toblin RL, Mack KA, Perveen G, Paulozzi LJ. A population-based survey of chronic pain and its treatment with prescription drugs. PAIN 2011;152:1249–55.
[91]. Tracey I, Woolf CJ, Andrews NA. Composite pain Biomarker signatures for objective assessment and effective treatment. Neuron 2019;101:783–800.
[92]. Traxler J, Madden VJ, Moseley GL, Vlaeyen JWS. Modulating pain thresholds through classical conditioning. PeerJ 2019;7:e6486.
[93]. Tsur N, Defrin R, Lahav Y, Solomon Z. The traumatized body: long-term PTSD and its implications for the orientation towards bodily signals. Psychiatry Res 2018;261:281–9.
[94]. Vehtari A, Gelman A, Simpson D, Carpenter B, Bürkner P-C. Rank-normalization, folding, and localization: an improved Rhat for assessing convergence of MCMC. Bayesian Anal 2020;16:667–718.
[95]. Vlaeyen JWS, Crombez G. Behavioral conceptualization and treatment of chronic pain. Annu Rev Clin Psychol 2020;16:187–212.
[96]. Wager TD, Atlas LY, Lindquist MA, Roy M, Woo CW, Kross E. An fMRI-based neurologic signature of physical pain. N Engl J Med 2013;368:1388–97.
[97]. Wegerer M, Blechert J, Kerschbaum H, Wilhelm FH. Relationship between fear conditionability and aversive memories: evidence from a novel conditioned-intrusion paradigm. PLoS One 2013;8:e79025.
[98]. Wegerer M, Blechert J, Wilhelm FH. Emotionales Lernen: ein naturalistisches experimentelles Paradigma zur Untersuchung von Angsterwerb und Extinktion mittels aversiver Filme. Z Psychiatrie Psychol Psychotherapie 2013;61:93–103.
[99]. Wilhelm FH, Peyk P. ANSLAB: autonomic nervous system laboratory (Version 4.0), 2005. Available at: www.sprweb.org. Accessed January 2021.
[100]. Wilhelm FH, Rattel JA, Wegerer M, Liedlgruber M, Schweighofer S, Kreibig SD, Kolodyazhniy V, Blechert J. Attend or defend? Sex differences in behavioral, autonomic, and respiratory response patterns to emotion–eliciting films. Biol Psychol 2017;130:30–40.
[101]. Woo CW, Schmidt L, Krishnan A, Jepma M, Roy M, Lindquist MA, Atlas LY, Wager TD. Quantifying cerebral contributions to pain beyond nociception. Nat Commun 2017;8:14211.
Keywords:

Classical conditioning; Pain conditioning; Intrusive memories; Trauma film; Somatic symptom disorder; Posttraumatic stress disorder; Chronic pain

Supplemental Digital Content

© 2022 International Association for the Study of Pain