Incident experience predicts freezing-like responses in firefighters

Freezing is a defensive response to acute stress that is associated with coping and alterations in attentional processing. However, it remains unclear whether individuals in high risk professions, who are skilled at making rapid decisions in emergency situations, show altered threat-induced freezing. Here we investigated the effect of incident experience in a high risk profession on freezing. Additionally, we explored whether any effect of incident experience on freezing would be different for profession-related and -unrelated threat. Forty experienced and inexperienced firefighters were presented neutral, pleasant, related-unpleasant, and unrelated-unpleasant pictures in a passive viewing task. Postural sway and heart rate were assessed to determine freezing. Both postural and heart rate data evidenced reduced freezing upon unpleasant pictures in the experienced versus the inexperienced group. Relatedness of the unpleasant pictures did not modulate these effects. These findings indicate that higher incident experience relates to decreased threat-induced freezing, at least in a passive task context. This might suggest that primary defense responses are malleable through experience. Finally, these findings demonstrate the potential of using animal to human translational approaches to investigate defensive behaviors in relation to incident experience in high risk professions and stimulate future research on the role of freezing in resilience and coping.


Introduction
Freezing is a major defensive response to threat in animals, which is characterized by heart rate deceleration (bradycardia) and immobility [1][2][3][4][5]. In general, freezing is considered adaptive as it optimizes attentional processes and prepares the organism to actively cope with acute stress [5][6][7]. Accumulative evidence suggests that freezing, as measured via reductions in postural sway and heart rate, can be induced in human subjects in a laboratory setting using aversive stimuli in a passive viewing task [8][9][10][11][12][13].These findings are robust; passively viewing a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 chose this population as the target group for the current study given their experience with directing emergency activities, which requires rapid decision making in emergency situations. Twenty experienced and 20 inexperienced fire officers were preselected based on the number of big fires they have coordinated in their current rank (! 4 for the experienced firefighters, and 3 for the inexperienced firefighters). A big fire was defined as involving more than three fire engines. Thus, the initial total sample consisted of 40 fire officers. However, five fire officers who were tested had to be excluded because of technical errors (n = 3), or poor signal quality of the heart rate data (n = 2). Accordingly, data are reported from thirty-five firefighters consisting of experienced (n = 19) and inexperienced (n = 16) fire officers. For a description of the characteristics of the sample, see Table 1.
As firefighters are regularly screened for physical and mental health, we only included healthy firefighters. Fire officers with a bodyweight above 120 kilograms were excluded from participation due to the limits of the stabilometric platform for accurate measurements. Furthermore, we excluded fire officers who experienced a big fire in the three weeks before the assessment to avoid acute effects of an incident. All participants received information about the experiment and gave written informed consent. The study was approved by the Institutional Review Board: the Ethical Committee Behavioural Sciences (Ethische Commissie Gedragswetenschappen in Dutch).

Apparatus and material
Equipment for postural sway and heart rate measurements. A valid and reliable way to assess freezing-like behavior is by means of a stabilometric force plate, which quantifies spontaneous fluctuations in postural sway during threat exposure [8][9]21]. Participants performed a passive viewing task on a custom-made strain gauge force plate (dimensions: 500 mm x 500 mm; sample frequency: 200 Hz), which consisted of four sensors measuring forces in the vertical direction. The signals from these sensors were used to derive time series of the center of pressure (COP) in the anterior-posterior (AP) and the mediolateral (ML) direction. Python 2.7.8 was used for posturographic data recording. Simultaneously, heart rate data was acquired during the passive viewing task using an Arduino pulse sensor attached to the fingertip of the right index finger. The signal was registered at a sample frequency of 200 Hz using Python 2.7.8 and LabVIEW. The visual stimuli of the passive viewing task were presented 1 m in front of the participants at eye level on a 22-inch height adjustable screen using PsychoPy v1.81 software.
Passive viewing task. Participants were instructed to stand upright and remain stationary on the board, barefooted with their feet next to each other approximately 30 cm apart. We instructed them to equally distribute their body weight on both legs with their arms hanging relaxed beside their body. At the same time, they were asked to focus on the screen to watch a sequence of stimuli on the screen. Before the actual task, participants viewed a sequence of six trials in which letters instead of pictures were presented in order to get familiar with the task. The passive viewing task was presented subsequently, in which four blocks with 20 pictures of one type of stimulus category (neutral, pleasant, related-unpleasant, unrelated-unpleasant) were presented consecutively for three seconds without inter trial interval. The order of blocks and stimuli was counterbalanced. Blocks were separated by a black screen (five seconds) followed by a white fixation cross (two seconds).
As visual stimuli, four sets of pictures (neutral, pleasant, related-unpleasant, unrelatedunpleasant) were used from the International Affective Picture System (IAPS; Center for the Study of Emotion and Attention, 1999). Every set consisted of 20 pictures. The neutral pictures consisted of pictures of objects, such as clothespins and a mug. The pleasant pictures consisted of pictures of sports, such as waterskiing and skydiving. The related-unpleasant set included pictures that were related to the work of firefighters, such as burnt bodies and fires; whereas the unrelated-unpleasant pictures depicted threatening scenarios unrelated to the work of firefighters, such as weapons and snakes. The categorization of relatedness was confirmed in an independent sample of firefighters (n = 5 experienced; n = 5 inexperienced; coordinated more or less than 3 big fires respectively), that rated the related-unpleasant and unrelated-unpleasant stimuli on a scale ranging from 0 = unrelated to 10 = related to their profession as indicated by a main effect of relatedness (F (1,9)  Questionnaires. We included several self-report questionnaires to assess individual differences in trait anxiety, tonic immobility, and optimism, because these factors have been related to individual differences in automatic stress responses [22][23][24][25][26]. We included these measures to test whether any group differences in freezing would be due to differences on these measures rather than differences in incident experience. Trait anxiety was assessed using the Spielberger State-Trait Anxiety Inventory (STAI) [27]. The STAI demonstrated good reliability and validity [27]. Participants responded to 20 self-reported items on a four point scale to indicate how anxious they feel in general. The Tonic Immobility Scale (TIS) was used to retrospectively assess reactions during a previous, most frightening, event [28]. Ten self-report items were scored on a seven point scale. Factor analyses have demonstrated that the TIS comprises two subscales: Fear (TIS-Fear; three items) and Tonic Immobility (TIS-TI; seven items) [29]. In the current study, we focused on the TIS-TI only, as we have previously shown that this subscale is a more valid measure of threat-related responses [24].Optimism was assessed with the Life Orientation Test-Revised [25]. This is a brief questionnaire consisting of ten items. Participants had to indicate to what extent the items applied to them on a five point scale. Of the ten items, four items serve as fillers. The questionnaire demonstrates acceptable reliability [25].
Subjective immobility was assessed by asking the participants to rate the extent to which they felt paralyzed when watching the picture on a scale ranging from 0-10 (0 = not paralyzed at all to 10 = extremely paralyzed). Furthermore, participants rated all picture stimuli on a scale ranging from 0-10 in terms of valence and arousal (0 = extremely unpleasant to 10 = extremely pleasant; and 0 = not arousing at all to 10 = extremely arousing respectively).

Procedure
Prior to testing, we attached the fingertip pulse sensor to assess heart rate throughout the experiment. Participants were then instructed to sit down in front of a computer screen to watch a three-minute neutral film depicting nature scenes, in order to normalize heart rate. Participants were then instructed to remove their shoes and take their position on the center of the stabilometric platform (with their feet 30 cm apart) to perform the passive viewing task. Before the start of the task, standardized instructions were displayed on the monitor. After the passive viewing task, participants completed the questionnaires and stimulus ratings.

Data-and statistical analyses
The analyses for posturographic and heart rate data were conducted using MATLAB R2009b (The MathWorks, Natick, MA). Statistical analyses were performed using IBM SPSS Statistics 23 (IBM Corp., Armonk, NY). Demographic data. We used analyses of variance (ANOVA) to test for any differences on the group characteristics.
Freezing. Following previous work, freezing was defined as a reduction in postural sway and heart rate [8-9, 21, 26]. Posturographic data were bandpass filtered using a 3rd order Butterworth filter with 0.01-10 Hz as a cutoff frequency. The standard deviation in anterior-posterior direction (SD-AP) was then determined for each stimulus, after which a mean SD-AP was calculated per stimulus category, with lower values representing increased freezing to the relevant stimulus-type. To calculate the heart rate per stimulus category for each participant, we analyzed the plethysmography data in the frequency domain using Welch's periodogram method for each stimulus category condition separately [30]. The power spectral density distributions of the plethysmography data were calculated using a range from 0.7-2.4 Hz. Next, for each stimulus category, beats per minute (BPM) were calculated via the peak frequency multiplied by 60. Following previous work [21,26], these mean SD-AP and BPM per stimulus category were used as variables in the following statistical analyses.
Two separate mixed design ANOVAs of postural sway (SD-AP) and heart rate (BPM) data were run to test the effects of stimulus category and incident experience on freezing. Stimulus category (neutral, pleasant, related-unpleasant, unrelated-unpleasant) was entered as a withinsubject factor, group (experienced, inexperienced) as a between-subject factor, and SD-AP and BPM as dependent variables.
Subjective ratings. We ran three separate mixed design ANOVAs with stimulus category (neutral, pleasant, related-unpleasant, unrelated-unpleasant) as a within-subject factor, and group (experienced, inexperienced) as a between-subject factor to test the effects of stimulus category and incident experience on the subjective picture ratings in terms of immobility, valence, and arousal.
Importantly, threat-induced freezing was operationalized as freezing in response to unpleasant stimuli relative to neutral or pleasant stimuli in order to control for individual differences in postural sway and heart rate [8-9, 12, 21, 31-33]. Moreover, the pleasant stimulus category allows to control for arousal. Therefore, the crucial test is the group (experienced, inexperienced) x stimulus category interaction with pleasant versus (un)related pictures as the most relevant contrast for testing freezing behavior in experienced versus inexperienced firefighters. The crucial test for exploring the effects of profession-relatedness is the group (experienced, inexperienced) x stimulus category (related-unpleasant versus unrelated-unpleasant) interaction. For all analyses, significant interaction effects were followed up by simple (interaction) effects analyses. Multivariate test statistics were used in case of violation of sphericity. Alpha was set at .05. Table 1 presents the demographic data. Apart from incident experience (number of big fires), the groups did not differ on age and measures related to trait anxiety, tonic immobility, and optimism (all Fs < 3.0). Table 2 shows the mean data of the freezing measures in terms of postural sway (SD-AP) and heart rate (BPM). Both postural sway and heart rate data evidence a difference in freezing between the experienced and the inexperienced group as described below.
ANOVAs of BPM within each group separately suggested that the interaction effect was due to the presence of threat-induced reductions of heart rate in the inexperienced group Taken together, the postural sway and heart rate data evidence that threat-induced freezing was present in the inexperienced group but not in the experienced group. Specifically, the group difference in threat-induced freezing in terms of postural sway was mainly driven by the unrelated-unpleasant (versus pleasant/neutral) stimulus category, whereas the group difference in terms of heart rate was mainly driven by the related-unpleasant (versus pleasant) stimulus category. Crucially, the groups did not differ on the comparisons between related-versus unrelated-unpleasant pictures, indicating that relatedness of the unpleasant stimulus category did not modulate the effects of incident experience on threat-induced freezing. Fig 1 shows the group differences in freezing in terms of postural sway and heart rate for the most relevant contrasts (pleasant versus related-and unrelated-unpleasant pictures). Table 2 shows the mean data of the subjective ratings with respect to immobility, valence and arousal.

Subjective ratings of stimuli
Immobility. An ANOVA showed a significant main effect of stimulus category on immobility ratings (F (3,31)

Discussion
We found that inexperienced firefighters showed more freezing relative to experienced firefighters. Specifically, inexperienced firefighters demonstrated threat-induced freezing as reflected by reduced postural sway and heart rate in response to unpleasant pictures (relative to pleasant and neutral pictures), whereas the experienced firefighters did not show such response. Additionally, we did not observe any modulation by the profession-relatedness of the unpleasant pictures. These findings indicate that incident experience in firefighters was associated with reduced threat-induced freezing. Moreover, this reduction of freezing may not be restricted to profession-related threats, but possibly generalizes to other threats. Taken together, our findings may suggest that primary defense responses are malleable by training and experience. Compared to inexperienced firefighters, our current experienced firefighters have received more education and training within their profession. Additionally, experienced firefighters have more experience with aversive events and experience in dealing with emergency situations relative to inexperienced firefighters. There are at least two mechanisms that could explain a reduction of threat-induced freezing through experience in firefighters. First, experienced compared to inexperienced firefighters might show reduced threatinduced freezing due to an enhanced ability to control this primary defensive response. Although freezing to acute stress is adaptive, the ability to switch from immobility to active coping is important for adaptive responding [18][19]. This flexibility could be particularly important in high-risk professions, where optimal planning and decision making are necessary in emergency situations. Literature on controllability suggests that the ability to exert behavioral control over adverse events may enhance the ability to control this primary defensive response in future events [34][35][36][37]. In other words, experience of behavioral control has been shown to promote stress resilience that extends to other situations: even when one encounters an uncontrollable stressor later in time, the behavioral and neurochemical impact of the stressor is reduced via top-down inhibitory control [34,36]. Such ability to control primary defensive responses could be accompanied by the recruitment of the instrumental system, which allow flexible behavior in a dynamic environment [18][19]38]. By analogy, firefighters who are more experienced, have exercised behavioral control in stressful situations more frequently, and might therefore have more (perceived) capability of making rapid decisions under acute stress compared to inexperienced firefighters. Thus, one possibility is that experienced firefighters might have an enhanced ability to control primary defensive response, resulting in reduced threat-induced freezing. Future studies and more detailed time series analyses are needed to study effects of incident experience on inhibitory control on automatic defense behavior, as well as on the ability to switch between freezing and action.
Alternatively, experienced firefighters might show reduced threat-induced freezing due to faster processing of threatening stimuli. Rather than a passive state, freezing has been suggested to reflect an active state that promotes risk assessment in potentially threatening situations (5). Recent work has shown that under acute stress, freezing serves as a preparatory state for action [16,39]. Despite the use of a passive task context in the current study, in which active responses are not required, inexperienced firefighters might still demonstrate automatic preparatory responses upon the threatening stimuli. This is in line with the model of situational awareness, which suggests that automatically scanning for relevant clues in the environment plays an important role in gaining appropriate awareness of a situation in high-risk professions [40]. Thus, freezing in inexperienced firefighters could be crucial in real life situations as it facilitates adaptive responding to emergency situations. On the other hand, experienced firefighters might have developed an improved ability to process and integrate perceived information into decision making, which is reflected by reduced freezing in a passive task context that does not require action preparation. Future research is needed to disentangle the different hypotheses regarding the underlying processes of our observed effect of incident experience on threat-induced freezing. For instance, a worthwhile pursuit is to examine the effects of incident experience on freezing in passive versus active contexts.
Note that reduced threat-induced freezing was also found in patients with PTSD [26,41] and borderline personality disorder [42] using passive viewing paradigms. In these groups, reduced freezing was assumed to reflect increased hyperarousal and vigilance and increased automatic fight-flight responses, which might attenuate adaptive freezing behavior. In other words, reduced threat-induced freezing might indicate impaired risk assessment and increased active responsivity to potential threatening stimuli, mediating cognitive distortions and exacerbated threat perception. However, it is unlikely that reduced freezing in our experienced firefighters is the result of a similar mechanism. First, we have tested healthy firefighters only, who are regularly screened for physical and mental health. Second, inexperienced firefighters reported higher arousal levels than experienced firefighters during picture viewing. Third, patients with threat-related psychopathology appear to suffer from impeded "downregulation", as reflected by increased baseline or overall heart rate [42][43][44]. Our current heart rate data suggest that this was not the case for the experienced firefighters, as their overall heart rate was similar to (or if anything lower than) their inexperienced counterparts, rendering the inability to downregulate an unlikely explanation.
Our data did not provide any evidence for a modulatory effect of relatedness on freezing, although the profession-related pictures were rated as more related and also more aversive compared to the unrelated pictures across the whole sample. However, subjective ratings are more vulnerable to influences from awareness of experimenters' demands and do not have to correspond with implicit and objective measures per se. Although it has been suggested that appraisal of relevance plays a crucial role in affective processing [45], the observed effect of incident experience on freezing might relate to alterations of other, perhaps more fundamental defense mechanisms. Therefore, rather than a specific group difference in freezing for profession-related (versus -unrelated) threat, the observed effect of incident experience was general to threat that is relevant to survival. In any case, we need to be careful with drawing conclusions based on a nonsignificant effect of relatedness, and further research on the relevance of relatedness in experience and freezing is warranted.
One could argue that our finding of increased freezing in the inexperienced versus experienced firefighters reflects an alteration in the degree of freezing in the inexperienced (rather than the experienced) group. However, findings with affective picture viewing typically found freezing in response to unpleasant pictures in healthy participants without high risk professions [8][9], suggesting that our observed effect of incident experience reflects alterations of freezing in the experienced firefighters. However, the question remains whether the inexperienced firefighters would resemble healthy participants without high-risk profession in terms of the degree of freezing. Future studies should include healthy participants without being involved in high risk professions as a control group to establish that incident experience indeed reduces freezing. Also, it is important to note that our sample consisted of battalion chiefs and incident commanders, who coordinate rescue activities and are responsible for their team. Although the experienced and inexperienced firefighters differed in their experience with the coordination of big fires, they have all built up substantial experience with firefighting-related tasks during their career. It is possible that experience with firefighting-related tasks from earlier ranks might also alter threat-induced freezing, contributing to a smaller difference between the groups in terms of threat-induced freezing in the current study. Future studies are needed to examine the effects of incident experience on spontaneous automatic defense responses in firefighters with lower ranks and thus less responsibility over the success of an operation.
The difference in freezing between experienced and inexperienced firefighters might be explained by an a priori difference between the groups in terms of trait and resilience rather than by incident experience or training, and that self-selection (through career switches) could have played a role in our findings. However, our questionnaire data do not suggest that this is the case as there were no differences between the experienced and inexperienced firefighters in terms of trait anxiety, tonic immobility, and optimism. To account for such interpretational limitations, we suggest that prospective studies are needed to elucidate more rigorously the effect of incident experience on defensive reactions to threat.
In sum, we have shown that higher incident experience in firefighters is associated with reduced threat-induced freezing. We did not find a modulatory effect of profession-related (versus -unrelated) threat on freezing. Although further research is needed to elucidate the mechanisms underlying our current findings, and replications are warranted given our small sample size, these findings help to bridge animal and human studies on defensive behaviors and indicate that primary defense responses are malleable by experience or training. Accordingly, the present findings represent an important step in inspiring future research into defensive reactions in high risk professions. The current findings demonstrate the potential of adopting animal to human translational approaches to assess defensive responses in relation to incident experience in high risk professions. Finally, these findings could stimulate the development of new (translational) procedures to test and train risk assessment abilities, which is particularly relevant in high risk professions.