Using a novel paradigm to examine observational fear-learning across development

Abstract

Background:

Fear-learning through observing others begins early in life. Yet, most observational fear-learning research has focused on adults. The current study used a novel developmentally appropriate observational fear conditioning paradigm to examine differences in observational fear-learning among children, adolescents, and adults.

Method:

Thirty-six typically developing children, 41 typically developing adolescents, and 40 adults underwent differential observational fear conditioning followed by a direct exposure test. Skin conductance response (SCR) and self-reported fear were measured.

Results:

Successful differential observational fear-learning was demonstrated in all three age groups as indexed by SCR, yet developmental differences emerged. Children showed overall higher physiological arousal during acquisition compared to adolescents and adults. Additionally, children reported less differential fear and were less successful at reporting the conditioned stimulus–unconditioned stimulus contingency compared to adolescents and adults. Finally, adolescents tended to overgeneralize their fear compared with adults.

Conclusions:

This is the first study to compare observational fear-learning among children, adolescents, and adults. The novel task effectively induced observational fear-learning, particularly among adolescents and adults. Findings revealed developmental differences that have both theoretical and clinical implications.

1 |. INTRODUCTION

Although observational learning is an important way that children learn to discriminate danger from safety, most experimental research on observational fear-learning examines adults (Askew & Field, 2008; Olsson & Phelps, 2007). Moreover, the few such studies in children use both observational and verbal learning (Field et al., 2001). Most importantly, no studies examine adolescents or compare observational fear-learning across age groups. The aim of the current study is to test a new developmentally appropriate observational fear-learning paradigm and examine differences in observational fear-learning among children, adolescents, and adults.

Observational fear-learning has been demonstrated in several studies (Haaker et al., 2017; Olsson & Phelps, 2007). It may induce similar levels of differential physiological responding and engage similar brain regions as Pavlovian fear-learning (Lindström et al., 2018). Fear extinction through observation also may be more effective than Pavlovian fear extinction, carrying possible clinical implications (Golkar et al., 2013). Limited research extends this study across development.

Several studies examine vicarious fear-learning during development. Many studies use retrospective self-report to assess the development of pathological fear via vicarious learning. These studies report that 15.5%–42% of adults with specific phobias attributed the onset of their fears to vicarious learning during youth (King et al., 1998). These retrospective studies with adults cannot explain how children and adolescents acquire fear in real-time through observational learning. Prospective studies show that infants (12 months) and toddlers (15–20 months) learn fear and avoidance by watching their mother’s fearful facial expressions (Gerull & Rapee, 2002; Murray et al., 2008). Additionally, school-age children (7–10 years) report greater fear and show more avoidance of fearful stimuli following a vicarious fear-learning task, compared to children who do not complete the learning task (Askew et al., 2014; Askew & Field, 2007). Although progress has been made in understanding vicarious fear-learning in children, important methodological limitations remain. Namely, most paradigms use static images rather than live models and augment observation with explicit instructions; none measure autonomic responses.

A major need exists for research in adolescents given the salience of social factors in this age group (McLaughlin & King, 2015). Though some studies examine social learning and the intergenerational transmission of anxiety in adolescents, no studies investigate observational fear-learning in this age group (Eley et al., 2015; Haddad et al., 2011). Moreover, no study examines age differences in observational fear-learning across children, adolescents, and adults.

An effective observational fear-learning protocol for adults was recently published (Haaker et al., 2017). During observational fear acquisition, participants observe a learning model presented with two stimuli. The conditioned stimulus (CS+) is paired with an electrical stimulation (unconditioned stimulus [US]) delivered to the learning model. The second stimulus is never paired with the US (CS−). Next, participants view the stimuli to test the observational learning and to extinguish the fear. Because no electrical stimulation is delivered to participants, the same US can be used with youth. Hence, the main adaptation we implement in the current study is to replace the adult learning model with an adolescent.

We hypothesize that the observational fear-learning paradigm successfully induces differential fear-learning in children, adolescents, and adults. Additionally, we expect that participants’ fear responses to the direct presentation of the CSs during the direct exposure test (test) positively correlate with levels of fear elicited by observing the US and CS+ during acquisition. Finally, this is the first observational fear-learning study to include psychophysiology measurement in youth. Hence, age-related physiological differences are explored to extend data from direct fear-learning studies that have shown an overall higher level of arousal among younger children (Shechner et al., 2014, 2015).

2 |. METHODS

2.1 |. Preregistration

This study was preregistered on the Open Science Framework (https://osf.io/5dn7r).

2.2 |. Participants

Thirty-six typically developing children (age range: 8.00–13.00 years, M = 10.3, SD = 1.50; 41.7% female), 41 typically developing adolescents (age range: 13.00–18.00 years, M = 15.34, SD = 1.26; 56.1% female), and 40 adults (age range: 18.00–38.00 years, M = 23.94, SD = 4.455; 75% female) participated in this study. The procedure was approved by the local Institutional Review Board. Parents of participants under 18 years of age and adult participants signed consent forms. Participants under 18 years of age signed assent forms. Parents were told that their children may believe they would, but would never, in fact, receive an electric stimulation during experimentation. Consent and assent forms included the following sentence: “During the experiment, noninvasive, and nondangerous electrical stimulation may be utilized.” Participants were informed that they could leave at any time without penalty. One adolescent and four children aborted the study before the test due to fear; for one child, the experiment was aborted due to technical difficulties. These six participants were excluded from analyses. Participants were recruited through ads in print and digital media, social networking sites, and word-of-mouth. All participants received a modest gift certificate upon completion of their participation.

2.3 |. Instruments and measurements

2.3.1 |. Observational fear conditioning paradigm

The observational fear conditioning paradigm was designed for youth (see Supporting Information Material), based on a similar task in adults (Haaker et al., 2017). It contains a 7-min video of a 14-year-old male actor (learning model) undergoing the bell differential fear conditioning task (Shechner et al., 2015). Blue and yellow cartoon bells were presented on the learning model’s monitor and served as conditioned stimuli (CSs). Each bell was displayed for 8 s per presentation. One bell was repeatedly paired with an electrical stimulation to the learning model’s wrist (US) for 66% of presentations (CS+; threat cue). The US was administered at Second 7 of the CS+ presentation, lasted for 1 s, and coterminated with CS+ offset (see Figure 1). During reinforced CS+–US trials, the learning model contorted his face and squeezed his hand to convey mild discomfort (see Clip S1). During nonreinforced CS+ trials, the model did not physically react. The second bell was presented in the absence of the US (CS−; safety cue). Each CS was presented six times. Two videos were made to ensure counterbalancing the color of the threat and safety cues between participants. In both videos, the CS− was presented in the first trial. Before watching the video, participants were told: “You will now see a video of another person. This person may receive an electric stimulation to his hand. Please pay attention so that you may infer what is happening.” Importantly, no electric stimulation was administered to the learning model during the making of this paradigm. Therefore, to ensure correct timing of the learning model’s observable reaction, a visual cue was displayed on his monitor which was then removed during video editing (Adobe Premiere Pro).

FIGURE 1.

(a) Observational fear conditioning paradigm: Learning model during a CS+–US reinforced trial. The bell was displayed for 8 s, per presentation. During reinforced trials, the US was administered at second 7 following stimulus onset, lasted for 1 s, and coterminated with CS+ offset.; (b) Participant undergoing acquisition by watching the observational fear-learning paradigm. Skin conductance response (SCR) electrodes are attached to the participant’s left palm. (c) Step-by-step experimental procedure including the title and measures collected during each step. Self-reported fear was measured at four timepoints, (1) preacquisition, (2) postacquisition, (3) pretest, and (4) posttest. SCR was measured during two phases (1) acquisition and (2) direct-exposure test. CS, conditioned stimulus; US, unconditioned stimulus

The study consisted of two phases: Observational fear acquisition and a direct-exposure test (test). During acquisition, the observational fear-conditioning paradigm was used and applied via PsychoPy software (Peirce et al., 2019). During the test, the same colored bells (six blue and six yellow) were presented directly on the participant’s screen but were never paired with the US. The color of the threat and safety cues and their order were counterbalanced between participants. Skin conductance response (SCR) was measured during both phases by placing two isotonic gel electrodes on each participant’s left palm (on the hypothenar and thenar muscles) and was recorded at a sampling rate of 1,000 Hz, using an 8-Slot Bionex system (Mindware Technologies Ltd.). Additionally, self-reported fear toward each CS (i.e., “how afraid are you of this bell?”) was measured at four time points (1) preacquisition, (2) postacquisition, (3) pretest, and (4) posttest, using a Likert scale ranging from 1 to 10. Specifically, participants were presented with the blue and yellow bells, separately, and were asked, “how afraid are you of this bell?” This question assessed participant’s momentary subjective feeling toward the stimuli.

2.4 |. Procedure

Throughout the experiment, participants were seated in front of a 19-in. monitor in a soundproofed room. First, participants rated their level of fear to the CSs (preacquisition). Next, SCR electrodes were attached to participants’ left hand. Participants then underwent observational fear acquisition followed by an assessment of contingency awareness and rating of self-reported fear (postacquisition). Next, faux stimulation electrodes were placed on participants’ right wrist, and for a third time, participants rated their level of fear to the CSs (pretest). Participants were then told, “you will now be presented with identical images to those that the boy in the video saw,” and underwent the direct exposure test. Finally, participants rated their level of fear to the CSs for the last time (posttest). Of note, no electrical stimulation was administered throughout the experiment. Participants were debriefed at the end of the experiment about the faux stimulation electrodes. See Figure 1 for the step-by-step experimental procedure.

2.5 |. Data analysis

2.5.1 |. Contingency awareness

Following acquisition, CS–US contingency awareness was assessed using three questions: (1) An open question about what they understood from acquisition, (2) which bell(s) were presented with the electric stimulation, and (3) how many times did each bell appear with an electric stimulation, from 0% to 100%. Participants were coded as showing contingency awareness if they answered at least two questions correctly. The CS–US association was analyzed using χ².

2.5.2 |. Skin conductance response (SCR)

SCR was analyzed using MindWare data analysis software. The SCR to the CS+ and CS− was assessed as the difference between base-to-peak amplitude, between 500 ms and 7 s following stimulus onset. A square root transformation was applied for each participant response. SCR units are reported in square root microSiemens (√microSiemens). Repeated-measures analyses of variance (ANOVAs) were used to assess differences between the two phases (acquisition and test), the two stimuli (CS+ and CS−), and the six trials serving as within-subject factors, and the three age groups (children, adolescents, adults), serving as a between-subjects factor. Additionally, for acquisition, the average response, excluding the first trial to each CS, and the average response to the US, were calculated over all participants. A repeated-measures ANOVA was then used to compare the three stimuli by age group. Lastly, Pearson correlations were employed to examine the associations between the US and the CSs in both phases, and PROCESS 3.4 (Hayes, 2012) was used to assess the role of age as moderator for these associations.

2.5.3 |. Self-reported fear

For the four self-report timepoints (preacquisition, postacquisition, pretest, posttest), averages of the self-reported fear ratings toward the stimuli (CS+, CS−) were calculated. Repeated-measures ANOVAs were used to assess differences between the four timepoints with the two stimuli serving as within-subject factors, and the three age groups serving as a between-subjects factor. Finally, repeated-measures ANOVAs were conducted for each age group separately to evaluate differences per age group, and pairwise comparisons of means were assessed for stimulus type at each time point.

All analyses were conducted to examine the main effects and interactions of physiological and self-reported fear as a result of observational fear acquisition and the direct exposure test across development. All analyses were conducted in SPSS.

3 |. RESULTS

3.1 |. Contingency awareness

The χ² analysis of correct versus incorrect awareness of the CS–US association was significant between age groups, χ²(2) = 7.747, p = .021; 90% of adults, 95% of adolescents, and only 75% of children report correctly on CS+–US association.

As one of the aims of the study was to examine developmental differences in observational learning, we included all participants in the primary analysis reported in the manuscript. Nonetheless, we conducted a second analysis including only those participants who understood CS–US contingency and have now included the additional findings at the end of the results section.

3.2 |. SCR

A significant phase-by-stimulus-by-trial interaction emerged, F(5, 540) = 8.229, p < .001, η² = 0.071. See Figure 2 (top graph). During the first trial of fear acquisition, SCR to the CS− (M = 0.751, SD = 0.526) was higher than the CS+ (M = 0.462, SD = 0.502), t(110) = −5.989, p < .001. During the test, SCR to the CS+ was higher than the CS− in the second, (CS+, M = 0.709, SD = 0.680; CS−, M = 0.321, SD = 0.448), t(110) = 5.463, p < .001, and third trials (CS+, M = 0.537, SD = .612; CS−, M = 0.314, SD = 0.536), t(110) = 3.833, p < .001, after correcting for multiple comparisons. No differences between the CSs were found in the second half of the test, indicating successful fear extinction. Additionally, no phase-by-stimulus-by-trial-by-age group interaction emerged, F(5, 540) = 0.379, p = .956, η² = 0.007. Nonetheless, as this is the first experiment to compare across age groups, the graphs according to age have been displayed. See Figure 2 (bottom graphs).

FIGURE 2.

Top graph (a): Averaged over all age groups, skin conductance response (SCR) phase (acquisition; test) by stimulus (CS+; CS−) by six trials interaction. During the first trial of fear acquisition, SCR to the CS− was higher compared to the CS+, seemingly due to an order effect. During the test, SCR to the CS+ was higher compared to the CS− in the second and third trials, and no differences were found in the second half of the test. Error bars indicate standard error. Bottom graphs (b–d): By age group, SCR phase (acquisition; test) by stimulus (CS+; CS−) by six trials in adults (b), adolescents (c), and children (d). A phase-by-age group interaction emerged; on average, SCR was lower in acquisition compared to the test, among adults and adolescents, but not among children. Error bars indicate standard error. CS, conditioned stimulus; US, unconditioned stimulus

A phase-by-age group interaction emerged, F(2, 108) = 4.628, p = .012, η² = 0.079; SCR was lower in acquisition compared to the test, amongst adults, F(1, 39) = 26.578, p < .001, η² = 0.405 and adolescents, F(1, 39) = 15.432, p < .001, η² = 0.284 but not children, F(1, 30) = 0.190, p = .666, η² = 0.006. See Figure 2 (bottom graphs). Comparing the age groups in each phase separately, differences emerged during acquisition, F(2, 110) = 12.685, p < .001 but not the test, F(2, 110) = 2.053, p = .133. During acquisition, SCR among children was higher than adults and adolescents (all ps < .013).

To understand the role of the US, a repeated-measures ANOVA for three stimuli (CS+, CS−, and US) by age group, during acquisition showed a significant main effect of stimulus, F(2, 210) = 67.610, p < .001, η² = 0.392. Pairwise comparisons revealed that SCR was higher to the US than to both CSs (all ps < .001). No difference emerged between the CS+ and CS−, p = .061 (see Figure 3).

FIGURE 3.

Mean skin conductance response (SCR) during acquisition. Comparing the three stimuli (CS+, CS−, and US) during acquisition, a significant effect emerged. Pairwise comparisons revealed SCR to the US was higher than both the CS+ and the CS−. No difference emerged between the CS+ and CS−. Error bars indicate standard error. CS, conditioned stimulus; US, unconditioned stimulus

Correlation coefficients were computed between SCR to the US and each CS, during each phase. A positive correlation manifested between the US and the CS+ (r = .767) and the CS− (r = .643) during acquisition, and similarly, between the US and the CS+ (r = .678) and the CS− (r = .599) during the test. All correlations were significant (all ps < .001).

Using PROCESS 3.4, the moderating effect of age on these correlations was examined. Only the correlation between the US and the CS+ during the test was significantly moderated by age, p = .0179. All other US associations were not (all ps > .061). That this association increased with age, implies that the effect of learning was strongest among adults, followed by adolescents, followed by children (see Figure 4).

FIGURE 4.

Scatter plots by age group, between the skin conductance response (SCR) to the US during acquisition and the SCR to the CS+ during the test. The correlation between the SCR to the US during acquisition and the SCR to the CS+ during the test was moderated by age. Children’s association was weakest, followed by adolescents, followed by adults. CS, conditioned stimulus; US, unconditioned stimulus

3.3 |. Self-reported fear

A significant timepoint-by-stimulus-by-age group interaction emerged, F(6, 306) = 2.502, p = .022, η² = 0.047. Next, a significant timepoint-by-stimulus interaction was seen in adults, F(3, 102) = 11.986, p < .001, η² = 0.261, and adolescents, F(3, 117) = 2.816, p = .042, η² = 0.067, but not in children, F(3, 87) = 2.241, p = .089, η² = 0.072. Posthoc analyses revealed no differential fear toward the CSs in preacquisition (see Figure 1c, “pre-acquisition”) across all age groups (all ps > .250). During post-acquisition (see Figure 1c, “post-acquisition”), only adolescents reported more fear to the CS+ (M = 1.13, SD = 1.786) than the CS− (M = 0.68, SD = 1.385), t(39) = 2.421, p = .020, but not adults or children (all ps > .357). After placing the faux stimulation electrodes, that is pretest (see Figure 1c, “pretest”), self-reported fear to the CS+ was greater than the CS−, in adults, CS+, M = 2.21, SD = 3.381, CS−, M = 0.26, SD = 0.595; t (38) = 3.724, p = .001 and adolescents, CS+, M = 1.55, SD = 2.375, CS−, M = 0.93, SD = 1.859; t(39) = 2.502, p = .017, but not in children, p = .085. Similarly posttest, (see Figure 1c, “posttest”), self-reported fear to the CS+ was greater than the CS− in adults, CS+, M = 1.43, SD = 2.319, CS−, M = 0.55, SD = 1.484; t(39) = 3.520, p = .001 and adolescents, CS+, M = 1.23, SD = 2.370, CS-, M = 0.73, SD = 1.664; t(39) = 2.639, p = .012, but not in children p = .269 (see Figure 5).

FIGURE 5.

Self-reported fear across time points in adults (a), adolescents (b), and children (c). For self-reported fear across four time points (1) preacquisition, (2) postacquisition (before faux electrodes were attached), (3) pretest (after faux electrodes were attached), and (4) posttest, a timepoint-by-stimulus-by-age group interaction emerged. A significant timepoint-by-stimulus interaction was seen in adults and adolescents, but not in children. Posthoc analyses revealed no differential fear toward the conditioned stimuli (CSs) in preacquisition across all age groups. During postacquisition, only adolescents reported more fear to the CS+ than the CS−, but not adults or children. After placing the faux stimulation electrodes, that is, pretest, and after the test phase, that is, posttest, self-reported fear to the CS+ was greater than the CS−, in adults and adolescents, but not in children. Error bars indicate standard error

3.4 |. Analysis including only participants who understood CS–US contingency

A second analysis was performed including only participants who understood the CS–US contingency. All aforementioned significant results remained the same (see Supporting Information Material for a detailed report). Two additional SCR findings emerged: First, the repeated-measures ANOVA for three stimuli (CS+, CS−, and US) by age group during observational fear acquisition yielded a similar main effect for stimulus as before, F(2, 188) = 68.339, p < .001, η² = 0.421. Pairwise comparisons now revealed that SCR to the CS+ (M = 0.385, SD = .344) was higher than the CS− (M = 0.315, SD = 0.305), p = .015 during acquisition. Importantly, this finding is similar to findings from direct fear conditioning where participants show greater SCR to the CS+ compared to the CS− during acquisition. Second, during observational fear acquisition, a one-way ANOVA of SCR to the US revealed age group differences where children’s SCR to the US was higher compared to both adolescents (p = .029) and adults (p = .005), and no difference was found between the latter two age groups (p = .439).

4 |. DISCUSSION

Observational fear conditioning allows people to learn about danger and safety. Because anxiety disorders emerge early, when people are especially sensitive to their environments, studies of fear-learning through observation inform developmental and clinical research. The aims of this study were to validate a new observational fear-learning paradigm using an adolescent model and to examine possible developmental differences in observational fear-learning. Three major findings emerged. First, differential SCR confirmed that observational fear acquisition occurred in all age groups. Nonetheless, developmental differences in SCR emerged for the overall response between the experimental phases, and a moderating effect of age was found on learning. Second, developmental differences were found in self-report. Children reported less differential fear between the CSs and were less able to report the CS-US contingency, compared to adolescents and adults. Third, adolescents, but not adults, reported differential fear to the CSs even before the faux electrodes were attached, indicating fear overgeneralization.

Children, adolescents, and adults adapted their defensive responses through observation, as indicated by SCR. This learning evoked a higher SCR when participants were subsequently placed in a similar threatening context to the learning model. This extends to other research on adults, which finds the act of observing a conspecific undergoing a fear-learning task to elicit a physiological response (Golkar et al., 2013; Olsson & Phelps, 2007). Additionally, as in previous studies, participants’ physiological response to US delivery to the model predicted the ensuing differential SCR during the test. Our finding extends previous research by demonstrating that children and adolescents, like adults, show differential SCR during observational fear-learning. This is the first study to report SCR in youth during an observational fear-learning task. Taken together, these results suggest that across age groups, the human physiological defense system responds differently to danger versus safety cues based on observation.

Differences in SCR as a function of age did emerge when examining the overall response between the experimental phases. Adolescents and adults showed higher levels of arousal during the test, when a direct threat was suspected, compared with acquisition, where the threat was only observed. In contrast, children showed similar levels of heightened arousal across phases. Children’s tendency for fear overgeneralization may account for this age difference. Children’s overgeneralization could reflect misunderstanding, leading them to believe that they were in direct danger during acquisition. Another developmental finding emerged when only participants who understood CS–US contingency were analyzed. Children showed higher SCR to the US compared to the other two age groups, which indicates an overall greater physiological response to the observed threat. Developmental differences in SCR have been demonstrated in previous direct fear-learning studies, in which children showed higher levels of arousal and tended to overgeneralize their fear compared to adults (Schiele et al., 2016; Shechner et al., 2015).

Evaluating the US–CS+ association revealed age to moderate learning. Indeed, adults, followed by adolescents, followed by children showed a stronger association between their SCR to the US during acquisition and their SCR to the CS+ during the direct exposure test. A stronger association between physiological response to the US and the previously neutral, reinforced cue could reflect better learning. This interpretation is in line with the previous explanation of less learning in children than adolescents and adults.

Developmental differences were more apparent in the self-report than SCR data. Unlike adolescents and adults, children did not report differential fear towards threat and safety cues, indicating difficulty in appraising the associative differences between the stimuli. Additionally, although the means were in the right direction, children’s self-reported fear levels did not reach statistical significance, possibly due to the large variance seen in the younger age group. This lack of differential self-report amongst children is somewhat consistent with previous observational fear-learning studies (Field et al., 2001; Field, 2006). For example, one study found that a video of an adult model did not effectively induce fearful beliefs among children, yet differential self-report was achieved when verbal cues were included (Field et al., 2001). In contrast, in another study, children did report more fear after seeing novel images of animals paired with scared faces (Askew et al., 2014; Askew & Field, 2007). Our paradigm included only a video of a live model, but no explicit verbal instructions. Taken together, verbal cues and still images may be more developmentally appropriate than live videos for inducing a vicariously learned self-reported fear response. Although our study is the first to show that children portray a differential physiological fear response when watching a live learning model, future studies should consider including verbal information to enhance the self-reported fear responses as well.

This account receives further support when examining the CS–US contingency awareness by age group. A total of 25% of children, compared with only 5% of adolescents and only 10% of adults failed to report correctly on the CS–US association. Noticeably, this higher rate in children further highlights discrepant age-related patterns for implicit physiological responding and explicit cognitive understanding. One direct fear-learning study indeed demonstrated children to experience more difficulty reporting CS–US contingency awareness (Waters et al., 2009) but other studies fail to find such differences (Craske et al., 2008). The current paradigm shows an important developmental disparity between direct fear-learning and observational fear-learning in reporting CS–US contingency awareness.

Our last finding suggests that although adolescents successfully learned fear through observational learning, they tended to overgeneralize their fear as compared to adults. Specifically, adolescents rated the CS+ as significantly more aversive than the CS−, after acquisition, even before the faux electrodes were attached. In contrast, adults reported that they were not afraid of either CS following acquisition, and only after the electrodes were attached, did they report on greater differential fear. As this is the first study to compare adolescents and adults in observational fear-learning, no previous data exist against which to compare these results. Yet, our finding is consistent with previous direct fear-learning research showing that adolescents display greater fear and overgeneralization toward threat, compared to adults (Den et al., 2015; Lau et al., 2011). These studies demonstrate developmental differences in neural activation and in hypervigilance to threat. Taken together, the current and prior findings emphasize the need to monitor multiple data streams when assessing aspects of learning.

The current study has two main limitations regarding the experimental protocol and the stimuli order in the paradigm. First, this novel paradigm was based on a previous observational fear-learning protocol (Haaker et al., 2017), but a change was made regarding the placement-timing of the faux stimulation electrodes. Whereas in the previous protocol the faux electrodes were attached at the beginning of the experiment, in the current protocol, they were attached just before the direct exposure test. This decision was based on pilot data which initially adhered to Haaker’s design, but participants were confused between the SCR and the faux electrodes. The decision to place the faux electrodes just before the test generated both advantages and disadvantages. One advantage is that it mirrors a real-life observational fear-learning scenario in which an observer is not directly exposed to the threat during learning. Another advantage is that placing the electrodes just before the test may enhance the participant’s threat expectation by drawing attention to this shift. However, this may also be a disadvantage as some unrelated effects could be introduced. Future studies should examine how the timing of the electrode placement impacts the observational fear acquisition and subsequent direct exposure to the stimuli. A second limitation regards the stimuli order in the observational fear-learning paradigm. As previously described, two videos were made to ensure counterbalancing the color of the threat and safety cues between participants. In both videos, the CS− was always presented first, generating an order effect. Importantly, when the first trial was removed from analysis, the same results emerged.

In conclusion, this novel observational fear-learning paradigm induces consistent differential fear responses in adolescents and adults, as well as interesting developmental differences between the two age groups. This might support future studies comparing other aspects of development using various designs. Results in children were more mixed. Given the importance of examining this age group, future studies might consider augmenting observational fear-learning using alternative methods, including explicit verbal cues.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.