Characterizing and measuring awe

The notion of awe can be traced back to the concept of “the sublime” in aesthetics, as articulated by Edmund Burke, Immanuel Kant, and others [6]. Contemporary scholarship has continued to examine the link between awe and the sublime, suggesting that these experiences share core features including perceptual vastness, cognitive accommodation, and complex states of simultaneous positive and negative emotion, including elements of fear [6–8]. Empirical work further supports the importance of subjective and objective features (e.g., height, scale, magnitude) in creating a sense of vastness in both sublime and awe experiences [9,10]. Awe as a distinct psychological construct was first systematically outlined in the seminal paper “Approaching awe, a moral, spiritual, and aesthetic emotion” by Keltner and Haidt. They articulate two necessary qualities of awe: (1) perceptual vastness and (2) a need for accommodation. First, awe involves contact with a perceptually vast stimulus and second, a need for accommodation on behalf of the experiencer, which entails the need for a cognitive shift in one’s understanding of these stimuli [2].

Since Keltner and Haidt’s seminal paper in 2003, researchers have better characterized various facets of awe and developed validated and empirical measures of the awe experience [11,12]. Other facets of awe include alterations to time (e.g., the sense that time slows down or speeds up; [10–12], small self (e.g., the sense that the self is diminished or shrinks), connectedness/oneness (e.g., union or communion with all living things or the universe as a whole), and physiological sensations (e.g., jaw dropping, chills, raised eyebrows). Each of the facets has empirical support, for instance, research has shown that vast objects such as nature [13,14] space [5,15], spiritual [16,17] or psychedelic [18,19] experience or grand theories [20], and relatively vast objects such as interpersonally significant experiences with loved ones (e.g., seeing your child walk for the first time; [21]) elicit awe. Research has also demonstrated the need for accommodation, for example awe relates to learning [22,23] and challenges or changes the way that participants viewed the world (e.g., experiencing something that they did not believe was possible; Campos et al., 2013 [24]). Though, it is important to note that operationalizing and eliciting vastness has been much easier for researchers than the assessment and elicitation of a need for accommodation. Awe has also been shown to be associated with alternations to time [15,22,25], a small sense of self [4,17,26,27], feelings of connectedness [26,28], and hallmark physiological sensations such as a dropped jaw, goosebumps, chills, and raised eyebrows [15,29–32]. Each of these refer to the acute subjective effects of awe.

Outcomes of Awe

The awe literature has produced three findings of particular interest regarding outcomes of experiencing awe. These include increases in: (1) agency detection, (2) prosociality, and (3) life-satisfaction. Agency detection refers to “the tendency to interpret events as the consequence of intentional and purpose-driven agents” (Valdesolo & Graham, 2014, pg. 170). Prosociality involves “…inclinations to share, care, and assist, which further enable individuals to function more effectively within social collectives” (Piff et al., 2015 pg. 883). Life-satisfaction, an overall cognitive assessment of how one’s life is going, is a commonly measured component of subjective well-being (Rudd et al., 2012). Table 1 provides a summary of the original studies on which the work in this paper was based.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Overview of Original Studies.

https://doi.org/10.1371/journal.pone.0356728.t001

The current study

In a series of 6 distinct studies, we attempt to replicate these outcomes associated with awe using two types of replications: conceptual vs. direct [33–35]. The term “conceptual replication” is used throughout the literature to mean repeating studies conceptually, with intentional changes to the methodology to see whether the effects hold under new conditions. On the other hand, a direct replication involves repeating studies conceptually and methodologically as closely as is possible to see if the same result can be reproduced. In the first series of studies in this paper, we conduct conceptual replications, which were designed to expand upon the work of previous authors by including novel hypotheses and multiple outcomes (see Table 1). In the case of direct replications, we consulted with and received materials from the original lead authors to conduct high-fidelity versions of their original work as closely as possible. For all replications the same analytic techniques were used to examine differences between three (ANOVAs) or two (t-tests) conditions, as well as the same partial eta squared (η_p²) effect size measure to allow for comparison across studies (see Table 2 for an overview of replications).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Overview of replications.

https://doi.org/10.1371/journal.pone.0356728.t002

Data collection for these studies took place between January 2018 and January 2019. Study 1 uses video manipulations to test all three outcomes (agency detection, prosocial behavior and life satisfaction). Study 2 follows the same methodology, using writing and recall manipulations, and finally Study 3 utilizes virtual reality to test the associations between awe and these three outcomes. Following (unexpected) replication failures in these conceptual replications, we conducted a second series of three studies which were direct replications of the three studies mentioned in the introduction, but with larger sample sizes. Data collection for these studies took place between February 2025 and April 2025. Study 4 is a direct replication of Study 1 from Valdesolo & Graham (2014) looking at awe and agency detection, Study 5 is a direct replication of Study 2 from Piff et al.‘s (2015) study examining awe and prosocial behavior, and Study 6 is a direct replication of Study 3 from Rudd et al.’s (2012) paper on awe and life satisfaction. In the subsequent sections, first the conceptual replications are described, followed by the direct replications.

Conceptual replications

In this present series of studies, we attempted to replicate these three findings—that awe increases (1) agency detection, (2) prosociality, and (3) well-being—and sought to investigate whether these three outcomes bore a relation to each other. Specifically, our initial hypothesis was that agency detection would mediate the relationship between the experience of awe and the outcomes of prosociality and well-being. Additionally, we were interested in the exploratory question of whether different aspects of awe in the awe experience scale (AWE-S; 9) – such as perception of vastness, need for accommodation, altered perception of time, self-diminishment, connectedness, or physiological reactions––related differentially to different outcomes.

We could not explore these hypotheses because foundational findings from the literature did not replicate under the current conditions in online and laboratory-based conceptual replication attempts of the bolded studies described in the previous section [3–5].The previous research reviewed above included, on average, 107 participants per study (ranging from 63 to 264), whereas the present research (except Study 3) includes anywhere from twice to ten times the number of participants (average of 598). We therefore exceed the 2.5 times sample size rule of thumb described by Simonsohn (2015) [36]. In most cases in the present studies, the emotion induction techniques and outcome measures are identical to those used in the previous research. Because the first two studies rely on online participants recruited via mTurk, Study 3 utilized a more immersive, laboratory-based awe induction technique to replicate findings.

It should be noted that although data for two of the six studies were collected via Amazon Mechanical Turk, concerns regarding participant non-naivety and response trustworthiness were addressed through multiple quality assurance procedures (e.g., attention checks, eligibility screening, and exclusion of low-quality responses), although these concerns remain an inherent limitation of online convenience samples. All data have been made publicly available on the Open Science Framework repository and can be accessed using the following identifier: DOI 10.17605/OSF.IO/VD26U. We report how we determined our sample size (above; Table 1), all data exclusions (if any), all manipulations, and all measures in the study.

Study 1: Video Manipulations

Study 1 study used videos to elicit awe, amusement, or no emotion (neutral). A battery of questionnaires was then administered. We expected the experimental awe condition would result in changes to supernatural beliefs, agency detection, prosocial beliefs, and well-being relative to the active and passive control conditions.

Participants

A total of 755 Participants were recruited using the online survey platform Amazon Mechanical Turk (M-Turk). Participants provided written informed consent and agreed that they were over 18 years of age. Upon completion of the survey, participants were compensated two dollars for their participation in the study. This study was approved by the first author’s Institutional Review Board at the time the conceptual replications were conducted (Long Island University, protocol #18/01–001).

Of the initial participants, 130 were removed from the study: 1 for failing to give consent to participate; 65 for answering “yes” when asked whether they had seen the video before; 5 for failing the attention check question of “Please write something you remember from the video” (if they didn’t respond or responded with an irrelevant answer). Lastly, 59 participants were removed for submitting an incomplete survey.

Participant Characteristics. In the Awe condition (N = 220), the mean age was 36.89 (SD = 10.95), 51.8% male, and 79.5% non-Hispanic white, 10.9% Black, and 5.9% Asian or Asian America. In the amusement condition (N = 178), the mean age was 35.83 (SD = 10.17), 55.1% male, and 77.0% non-Hispanic white, 8.4% black, and 7.9% Asian or Asian American. Finally, in the Neutral condition (N = 227), the mean age was 36.75 (SD = 10.70), 55.9% male, and 84.1% non-Hispanic white, 6.2% black, and 5.3% Asian or Asian American.

Measures

Internal reliabilities for all measures are available in the supplement (S1 Table). In Study 1, all scales and subscales were reliable (α > .70), except the Baby Scale of the Mind Perception Scale (α = .57).

Positive and Negative Affect. Items from the Positive and Negative Affect Scale developed by [37] measured both positive and negative emotion. For Negative Affect, participants were presented with statements such as “In this moment I feel anxious,” and then asked to evaluate the degree to which they felt the given emotion on a likert-type scale from ranging from 1 “Least intensely” −11 “Most intensely”. For positive Affect, participants were presented with statements such as “In this moment I feel happy,” and then asked to evaluate the degree to which they felt the given emotion. A single-item measure of awe was embedded into the PANAS following the same format: “In this moment I feel awe”.

Awe. The Awe Experience Scale (AWE-S; Yaden et al., 2019) measures various qualities associated with awe experiences and contains six subscales: vastness, accommodation, self-diminishment, connectedness, time alteration, and physiological reactions. Participants evaluated the degree to which they agreed with statements such as “I felt my sense of self shrink” from 1 “Strongly Disagree” to 7 “Strongly Agree”.

Intentional Pattern Perception. Intentional pattern perception was measured using a numerical judgment task adopted from past research by Valdesolo and Graham (2014). In this task which consisted of 10 trials, participants are asked to distinguish between random strings of numbers that are either computer generated, or human produced.

Supernatural Belief. Supernatural Belief was measured using questions adopted from past research by Valdesolo and Graham (2014). Given that supernatural control (i.e., the idea that the universe is controlled by God or a supernatural force) was the measure of interest in past research by Valdesolo and Graham (2014), the three-item index of belief in supernatural control [38] was also scored separately. These items were assessed on a 10-item likert-type scale.

Need for Cognitive Closure. Given that Valdesolo and Graham (2014) used the construct of Cognitive Closure to measure uncertainty tolerance, the Need for Closure-Short Form Scale developed by [39] was used in this study to measure uncertainty tolerance on a likert-type scale ranging from 1 “Strongly Disagree” −6 “Strongly Agree”. Valdesolo and Graham (2014) focused on the ambiguity subscale of the Need for Closure scale in their past work. For this reason, intolerance of ambiguity was scored separately.

Mind Perception. Gray et al.‘s (2007) Mind Perception Scale [40] was used to measure mind perception in this study. The scale contains 48 questions, with 6 questions organized around a given target (e.g., adult human, baby, dead woman, dog, God, robot, superman, and tree). Each of the items is evaluated on a likert-type scale ranging from 0 “Not at all” to 6 “Very much”.

Prosocial Tendencies. Given that Piff and colleagues (2015) assessed awe and its relationship to prosocial behavior, Carlo et al.‘s (2003) Prosocial Tendencies Measure [41] was used to measure prosocial behavior in this study. Items such as “I can help others best when people are watching me” were evaluated on a likert-type scale from 1 “Describes me not at all” to 5 “Describes me greatly”.

Life Satisfaction. Given that Rudd, Vohs, and Aker (2012) used The Satisfaction with Life Scale (SWLS; Diener et al., 1985) [42] this scale was used to measure life satisfaction using a 7-item likert-type scale ranging from 1 “Strongly disagree” to 7 “Strongly Agree”.

Procedure

The survey was created using Qualtrics survey software and distributed on M-Turk, an online survey platform. Once they had given their consent to participate and filled out demographic information, participants were randomly assigned to one of the three video conditions: the awe video (experimental), the amusement video (control), or the neutral affect video (control). After watching the video, the rest of the survey was the same for all participant and all participants received the survey in the same order.

Study 1 Discussion

This online study randomized participants (N = 625) into one of three emotion conditions––awe (n = 220), amusement (n = 178), or neutral (n = 227)––who then viewed short video clips based on those used by Valdesolo & Graham, (2014) and Piff and colleagues (2015). The awe induction successfully induced awe, as measured by both a single item awe measure (in the PANAS) and a multifactorial measure of awe (the AWE-S), compared to the neutral or amusement conditions. But this did not produce any significant effects in most of the other outcome measures, including supernatural beliefs, intentional pattern perception, uncertainty tolerance, prosocial behavior, and mind perception. Life satisfaction was marginally significantly associated with awe, but only compared to the neutral condition and not the active control.

Under the present conditions, we did not replicate the finding reported in Valdesolo and Graham (2014), Piff and colleagues (2015); and Rudd and colleagues (2012), though it should be noted that they used a few different forms of emotion inductions, especially video but also writing exercises. With Study 1 using video, Study 2 would re-assess using writing exercises.

Study 2: Writing Manipulations

Study 2 used writing prompts to elicit awe (experimental), pride (control), or neutral (control). The same battery of questionnaires was then administered. According to past research, we hypothesized that the awe manipulation should result in changes to supernatural beliefs, agency detection, prosocial beliefs, and well-being.

Participants

A total of 744 participants were recruited using the online survey platform M-Turk. Participants provided written informed consent and agreed that they were over 18 years of age. This study was approved by the first author’s Institutional Review Board.

Of the participants, 174 were removed from the study: 21 participants were removed for failing to respond to the writing task or responding with an irrelevant answer (e.g., writing a random string of letters, copy and pasting text from elsewhere on the survey or internet into the response box); 29 participants were removed for failing the attention check by answering incorrectly when they were asked if they were paying attention to the survey. Lastly, 124 participants were removed for submitting an incomplete survey.

Participant Characteristics. In the Awe condition (n = 184), the mean age was 35.06 (SD = 9.71), 64.1% male, and 82.1% non-Hispanic white, 10.9% Black, and 2.2% Asian or Asian America. In the general positivity condition (n = 189), the mean age was 34.38 (SD = 10.31), 55.1% male, and 81.0% non-hispanic white, 10.9% black, and 6.3% Asian or Asian American. Finally, in the Neutral condition (n = 197), the mean age was 36.08 (SD = 11.18), 58.2% male, and 83.2% non-hispanic white, 6.6% black, and 5.6% Asian or Asian American.

Measures

The same measures in Study 1 were used. These include a single item awe measure embedded within the PANAS in addition to the AWE-S (which includes vastness, need for accommodation, self-loss, connectedness, time alteration, physiological changes). Additionally, supernatural beliefs, intentional pattern perception, uncertainty intolerance, prosocial behavior, mind perception, and satisfaction with life outcomes were also assessed. Reliabilities for these measures in this sample can be seen in the supplemental materials (S1 Table); however, all the scales and subscales were found to be reliable (α > .70), with the exception of the Baby (α = .52) and Dog (α = .66) subscales of the Mind Perception Scale.

Procedure

Participants were randomly assigned to conditions. After the writing prompt, the survey was identical across conditions.

Study 2 discussion

This study (N = 570) randomly assigned participants into one of three emotion induction conditions––awe (n = 184), general positivity (n = 189), or neutral (n = 197) ––and then asked them to complete a recall writing task. This writing task successfully induced awe, as measured by a single awe item on the PANAS scale and the AWE-S, as compared to the general positivity and neutral conditions. But, contrary to expectations and recent research (Piff et al., 2015; Rudd et al., 2012; Valdesolo & Graham, 2014), no significant effects were found on any of the outcome measures, including supernatural beliefs, intentional pattern perception, uncertainty tolerance, prosocial behavior, mind perception, and life satisfaction. It could be the case that videos and writing may manipulate awe but may do so too subtly to produce other effects. Other, more immersive techniques may provide stronger subjective effects. Furthermore, it may be that in-person, laboratory studies could amplify the effects of these manipulations. These issues are addressed in Study 3.

Study 3: Virtual reality manipulation

This next study attempted to use virtual reality to elicit awe in a laboratory setting, with a virtual reality version of Google Earth. In the experimental condition, participants were floating above the building in which the study was taking place and slowly floated upwards so that they could see the city, the state, the country, and, eventually, the entire world. Unlike the past two manipulations, this study tried to create a sense of vastness perceptually and cognitively using a virtual reality program. A battery of questionnaires was then administered, with some additions, as we were hoping to test moderators related to beliefs about the world (measured by a “primal world belief” scale) and feelings of connectedness with other people (using the inclusion of others in the self measure). We expected the awe condition (experimental), as compared to an active control and a neutral control, would result in changes to supernatural beliefs, agency detection, prosocial beliefs, and well-being.

Participants

A total of 100 participants were recruited at a top American university from the North East using the SONA system to give undergraduate students course credit for participating in studies. Participants provided written informed consent and agreed that they were over 18 years of age. This study was approved by the Institutional Review Board at the University of Pennsylvania.

Of the participants, 9 were removed from the study for incomplete data.

Participant Characteristics. In the Awe condition (n = 34), the mean age was 19.91 (SD = 1.11), 29.4% male, and 44.1% non-Hispanic white, 8.8% Black, and 38.2% Asian or Asian America. In the active control condition (n = 30), the mean age was 19.80 (SD = 1.32), 36.7% male, and 44.7% non-Hispanic white, 16.7% black, and 23.3% Asian or Asian American. Finally, in the Neutral condition (n = 27), the mean age was 36.08 (SD = 11.18), 58.2% male, and 48.1% non-Hispanic white, 3.7% black, and 40.7% Asian or Asian American.

Measures

Reliability of measures used in this study are available in the supplement (S2 Table). All scales and subscales used in this study were reliable (α > .70), except for the Enticing world belief subscale (α = .69) of the primal world belief measure.

Mood Direction. Items from the Positive and Negative Affect Scale developed by Watson, Clark, and Tellegen (1988) measured both positive and negative emotion. For Negative Affect, participants were presented with statements such as “In this moment I feel anxious,” and then asked to evaluate the degree to which they felt the given emotion. As in the previous studies, a single-item measure of awe was embedded within the PANAS.

Awe. The Awe Experience Scale (AWE-S; Yaden et al., 2019) was administered. This scale measures various qualities associated with awe experience and contains six subscales: vastness, accommodation, self-diminishment, connectedness, time alteration, and physiological reactions.

Presence. The sense of presence, or of feeling ‘present’ in the virtual environment, were measured by the ITC-Sense of Presence Inventory [46]. Sample item “I had a sense of “being there” in the virtual space.”

Inclusion of Other in Self. The Inclusion of Other in Self (IOS; Aron et al., 1992) [47] item was administered to measure the degree to which participants felt connected to other people. The item features two circles, one marked ‘self’ and one marked ‘other’ that are more or less overlapping. Participants indicate how connected to others they feel by indicating less or more overlapping circles.

Intentional Pattern Perception. Intentional Pattern Perception was measured using the same numerical judgment task from Study 1 adopted from past research by Valdesolo and Graham (2014).

Primal World Beliefs. The Primals Inventory [48] is a measure of beliefs about the world as a whole. In this case, four sub-scales were used: Interconnectedness (e.g., “The world is a place where everything is completely interconnected”), Safe (e.g., “I tend to see the world as pretty safe”), Enticing (“No matter where we are, incredible beauty is always around us”), and Alive (“What happens in the world in meant to happen”).

Supernatural Belief. Supernatural Belief was measured using questions adopted from past research by Valdesolo and Graham (2014). Given that supernatural control (i.e., the idea that the universe is controlled by God or a supernatural force) was the measure of interest in past research by Valdesolo and Graham (2014), the three-item index of belief in supernatural control [38] was also scored separately.

Prosocial Behavior. We used Piff and colleagues (2015) task measure of prosocial behavior in which a research assistant pretends to drop several pens on the floor and then counts how many pens the participant helps pick up.

Life Satisfaction. Given that Rudd, Vohs, and Aker (2012) used The Satisfaction with Life Scale (SWLS) developed by Diener et al. (1985), we used SWLS to measure life satisfaction.

Procedure

After receiving the experimental manipulation, participants filled out the same questionnaire.

Study 3 Discussion

This study randomly assigned participants into one of three emotion induction conditions and assessed feelings of awe as well as other outcomes. The virtual reality induction successfully induced awe in the experimental and active control conditions, as measured by a single awe item on the PANAS scale and most sub-scales of the AWE-S, as compared to the neutral control condition. No significant effects were found on any of the outcome measures, including supernatural beliefs, intentional pattern perception, primal world beliefs, prosocial behavior, moral circle, inclusion of other in the self, and life satisfaction. This was true across all three groups as well as between only the experimental and neutral conditions. This study used a slightly more immersive induction technique (virtual reality as opposed to writing or a video) and was conducted in-person in a laboratory setting. It may be the case that these studies were still too subtle manipulations of awe to impact outcome measures. It could also be the case that previous studies on awe found significant effects through random variation in smaller sample sizes.

Direct replications

The conceptual replications had several notable limitations. First, the battery of outcome measures was delivered in the same order each time to all participants across all three conceptual replication studies. Therefore, order effects may explain null results especially for outcome measures which appeared toward the end of the battery. Second, the potentially fleeting nature of awe may have made it impossible to capture an outcome effect in so large of a battery of outcomes measures in general. Third, some of our manipulations and outcome measures were inspired by, but not exactly the same as, the three studies we based our replications on [3–5].

To address these limitations, we next conducted a series of direct replications of the studies mentioned in Table 1. We chose studies from past research that could be replicated using online methods in order to collect larger sample sizes. Rather than presenting multiple outcome measures following awe induction, we directly replicated the three studies using three distinct samples that contained the same exact materials and presented in the same exact way as the original studies.

For these direct replication studies, the target sample was increased to 2.5 times the original [36]. Finally, though part of the analyses of the original papers, mediation analyses could not be conducted in any of the direct replications in our studies as we found no difference in outcomes between conditions across replications. This set of studies was approved by Johns Hopkin’s University’s Institutional Review Board. These studies are presented below. Preregistration for the series of direct replications is available at: https://doi.org/10.17605/OSF.IO/S9MCN and all data associated with this study have been made available at: https://osf.io/458qc/.

Study 4: Direct replication of video manipulation

Study 4 in our study was a designed as a direct replication of Valdesolo and Graham’s (2014; Study 1) video manipulation to elicit awe, amusement, or no emotion (neutral). Several authors (M.G., D.B.Y., along an original author, Valdesolo) have previously conducted a pre-registered direct replication using an in-person college campus sample as the authors did in the original study, similarly finding that the awe manipulation was successful, but outcomes from the original study did not replicate (Yaden et al., in press). To increase statistical power and sample size, an online Prolific sample was chosen rather than recruiting participants from a college campus. We acknowledge that this represents a methodological deviation from the original study. However, the participants in our study viewed the exact same video stimuli that the original study participants viewed on iPads, and the measures, procedures, and analytic approach were otherwise matched to those used by the original authors. Consistent with our definition of direct replication (p.7), we classify this study as a direct replication while recognizing that the change may have influenced the findings.

Participants

A total of 312 participants were recruited using the online survey platform Prolific. Participants provided written informed consent and agreed that they were over 18 years of age.

Of the initial 312 participants, 12 were removed from the study: 3 for failing 1 of the 2 attention checks built into the survey. Lastly, 9 participants were removed for submitting an incomplete survey. A total of 300 participants remained. This is 2.5 times the 120 participants recruited in the original study.

Participant Characteristics. In the Awe condition (N = 108), the mean age was 42.83 (SD = 13.99), 58% female, and 82% non-Hispanic white, 7% Black, and 6% Asian or Asian American. In the amusement condition (N = 101), the mean age was 39.55 (SD = 13.24), 64% female, and 83% non-Hispanic white, 11% black, and 3% Asian or Asian American. Finally, in the Neutral condition (N = 91), the mean age was 40.46 (SD = 14.30), 62% female, and 79% non-Hispanic white, 8% black, and 10% Asian or Asian American.

Measures

Supernatural control [38]. This 3-item measure asks participants to rate the extent to which they believe that various supernatural forces exert control over events.

Belief in God [49]. This single item is a general measure of belief in God.

Other supernatural beliefs [50]. These are four items related to other supernatural beliefs (curses, ghosts, miracles, and angels). These items were included in the original study, but their analysis was not reported. We include them here in order to exactly replicate the original methods but make no predictions about the effect of condition on these items.

Emotional states. An 8-item measure of self-reported emotional states, including a manipulation check that asked participants to indicate how much awe they felt during the video. This was an ad-hoc scale created by the authors of the original study.

Procedure

The survey was created using Qualtrics survey software and distributed on Prolific, an online survey platform. Once participants had given their consent to participate, they were randomly assigned to one of the three video conditions: the awe video (experimental), the amusement video (control), or the neutral affect video (control). After watching the video, the rest of the survey was the same for all participants.

Study 4 Discussion

This online study randomized participants (N = 300) into one of three emotion conditions––awe (n = 108), amusement (n = 101), or neutral (n = 91)––who then viewed short video clips used by Valdesolo & Graham, (2014). The awe induction successfully induced awe, as measured by a single item awe measure but this did not produce any significant effects in most of the other outcome measures, including supernatural control or belief in God. When the Awe condition was compared to the neutral condition alone there was a significant difference in supernatural control between conditions, though this difference did not hold when compared to the active control condition. This pattern of findings is virtually identical to those found by Yaden et. al. (in press) which conducted a preregistered direct replication (PDR) of this same exact study – using an in-person (N = 299) sample.

Study 5: Direct Replication of Writing Manipulation

Our Study 5 was a direct replication of Piff et al.‘s (2015) Study 2 which used writing and recall prompts to elicit awe (experimental), pride (control), or neutral (control). Like the original study we expected that participants would report a significantly increased small sense of self in the awe condition and that this small self would mediate changes in ethical decision making.

Participants

A total of 188 participants were recruited using the online survey platform Prolific. Participants provided written informed consent and agreed that they were over 18 years of age.

Of the 212 participants recruited, 24 were removed from the study: 1 participant was removed for failing to provide consent; 2 participants were removed for failing one of the two attention checks. Lastly, 21 participants were removed for submitting an incomplete survey. A total of 188 participants remained, which is 2.5 times the 75 participants recruited in the original study.

Participant Characteristics. In the Awe condition (n = 55), the mean age was 39.96 (SD = 13.43), 49.1% male, and 42% non-Hispanic white, 10.9% Black, 7.3% Asian or Asian America, and 1.8% Multiracial. In the Pride condition (n = 58), the mean age was 40.31 (SD = 14.29), 43.1% male, and 45% non-Hispanic white, 10.3% black, and 6.9% Asian or Asian American, and 5.1% Multiracial. Finally, in the Neutral condition (n = 75), the mean age was 39.31 (SD = 11.99), 22.7% male, and 57% non-Hispanic white, 4% black, 8.0% Asian or Asian American, and 5.3% Multiracial.

Measures

Emotion. Participants were asked to evaluate the degree to which they felt the following emotions on a 7-point Likert-type scale (1 = not at all, 7 = extremely): Anger, Awe, Disgust, Fear, Pride, Sadness, and Happiness.

Small Self (Shiota et al., 2007). To measure the small self, participants were asked to evaluate their endorsement of the following prompt on a 7-point Likert-type scale (1 = not at all true, 7 = very true): “I felt the presence of something greater than myself”.

Ethical Decision Making Task (Detert et al., 2008). Participants were presented with eight hypothetical scenarios describing self-interested acts that violated social norms and standards of behavior such as:

“You’ve waited in line for 10 min to buy a coffee and muffin at Starbucks. When you are a couple of blocks away, you realize that the clerk gave you change for $20 rather than for the $10 you gave him. You savor your coffee, muffin, and free $10”.

For each of the scenarios participants were asked to evaluate how likely they would be to engage in the behavior described on a 7-point Likert-type scale (1 = not at all likely, 7 = highly likely).

Procedure

Participants were randomly assigned to the awe, pride, or neutral writing task. After the writing prompt participants completed measures of emotion states, followed by the small self item. Participants were then informed that they would complete an unrelated study of filler items which asked about their beliefs about the world. Finally, the ethical decision-making task was administered.

Study 5 Discussion

This study (N = 187) randomly assigned participants into one of three emotion induction conditions––awe (n = 55), pride (n = 58), or neutral (n = 75) ––and then asked them to complete a recall writing task. This writing task successfully induced awe, as measured by a single awe item. Like the original study, we also found that ratings on a single-item small self differed between groups, significantly higher in the Awe condition. Finally, unlike the original, we did not find any significant differences between conditions on the ethical decision-making task. Though we can say that we were able to partially replicate Piff et al.‘s (2015) study, it is worth noting that a small sense has been conceptualized as a facet rather than an outcome of awe experience [11,12]. Therefore, it could be argued that the small self-rating in this study serves more as a further manipulation check of a specific facet of awe, rather than as an outcome. On the other hand, the prosocial behavior outcome of interest in this study, ethical decision making, was not significantly different between conditions.

Study 6: Direct Replication of Story Task

Our study 6 was a direct replication of Rudd et al.‘s (2012; Study 3) study which assigned participants to an awe (experimental) or neutral (control) condition. This manipulation asked participants to imagine themselves as the subjects of a story and then measured the impact of awe on perceived time availability as well as the preference for experiential over material goods. Like the original, we expected the awe condition to induce an expanded feeling of time as well as a preference for experiential (over material) goods.

Participants

A total of 268 participants were recruited using the online survey platform Prolific. Participants provided written informed consent and agreed that they were over 18 years of age.

Of the 268 participants recruited, 5 were removed from the study. One participant was removed for failing one of two attention checks and 4 were removed for submitting a blank survey. A total of 263 participants remained, which is 2.5 times the original 105 participant recruited for the original study.

Participant Characteristics. In the Awe condition (n = 142), the mean age was 38.70 (SD = 12.96), 38.7% male, and 83.8% non-Hispanic white, 3.5% Black, 4.2% Asian or Asian American, and 3.5% Multiracial. In the Neutral condition (n = 121), the mean age was 38.64 (SD = 13.10), 42.1% male, and 83.5% non-Hispanic white, 5.0% black, 6.6% Asian or Asian American, and 0.8% Multiracial.

Measures

Emotion. Participants were asked to rate their current feelings of the following emotions on a 7-point Likert type scale (1 = not at all, 7 = very much): calmness, anxiety, relaxation, worry, awe, sadness, boredom, and fear.

Perceived Time Availability. Perceived time availability was measured using a 4-item index created for the purposes of the original study which contained items such as “I have lots of time in which I can get things done,” “Time is expanded”, “Time is boundless” and “Time is slipping away” (which was a revere-scored item. Participants were asked to evaluate these prompts on a 7-tiem Likert-type scale.

Life Satisfaction (Kahneman et. al, 2006). Participants were asked to fill out a single-item measure of momentary life satisfaction (“All things considered, how satisfied are you with your life as a whole, right now?”) from 1 = not at all satisfied with life to 7 = extremely satisfied with life.

Experiential vs. Material Goods. Participants were asked to make hypothetical choices between material goods (such as a watch) and experiential goods (such as Broadway show tickets) said to be equivalent in price.

Procedure

Participants were told that they would be completing several unrelated surveys. First participants were instructed to read the story task (to which they were randomly assigned an awe story or a neutral story). Next, participants completed filler items followed by a perceived time availability index. The momentary life satisfaction item was administered next, followed by the material vs. experiential goods task. Finally, single-item measures of various emotion states were given.

Study 6 Discussion

This online study randomized participants (N = 300) into one of two emotion conditions––awe (n = 108), amusement (n = 101), or neutral (n = 91)––who then were asked to read a story and imagine themselves either climbing the Eiffel tower or climbing an unnamed tower [5]. Unlike the other studies in which awe was successfully induced but predicted outcomes did not change, here the awe induction did not successfully induce awe (and outcomes again did not change). Since the awe induction was unsuccessful, it is unclear whether the lack of effects on the outcomes was due to the lack of such an effect or due to the failure of the awe induction.