Reasons or Rationalisations


Cillian McHugh


July 28, 2020

This post provides a brief overview of a paper where we tested alternative explanations of moral dumbfounding proposed by critics (McHugh et al. 2020).

(full text available here)


I’m not sure how I ended up studying moral dumbfounding1, but it has intrigued me since I first read about it in a chapter by Prinz (2005). I admit that the first time I read the scenarios, I found myself dumbfounded. This meant that I was puzzled by critics of the phenomenon arguing that dumbfounding might not be a real phenomenon.

Challenges to Moral Dumbfounding

Arguments against the dumbfounding narrative include:

Evidence for Judgements based on Principles or Reasons

The strongest challenge to moral dumbfounding came from a set of studies by Royzman, Kim, and Leeman (2015), which tested moral dumbfounding directly, using the Julie and Mark dilemma which reads:

Julie and Mark, who are brother and sister, are travelling together in France. They are both on summer vacation from college. One night they are staying alone in a cabin near the beach. They decide that it would be interesting and fun if they tried making love. At very least it would be a new experience for each of them. Julie was already taking birth control pills, but Mark uses a condom too, just to be safe. They both enjoy it, but they decide not to do it again. They keep that night as a special secret between them, which makes them feel even closer to each other.

  • Royzman et al. (2015) identified two principles that appear to be guiding judgements
    • The Harm Principle
    • The Norm Principle
  • Participants who endorse either principle have a reason to judge the behaviour of Julie and Mark as wrong (and therefore cannot be dumbfounded)

Exclusion of Participants

Royzman et al. (2015) excluded participants who endorsed either principle from analysis, and found that dumbfounding effectively disappeared.

Harm-based reasons

Participants were presented with the following two questions (emphasis added here):

  • “are you able to believe that Julie and Mark’s having sex with each other will not negatively affect the quality of their relationship or how they feel about each other later on?”
  • “are you able to believe that Julie and Mark’s having sex with each other will have no bad consequences for them personally and/or for those close to them?”

If participants responded “no” to either question they were excluded from analysis.

Norm-based reasons

Participants were presented with the following two statements and asked to select the one they agrred with most (emphasis added here):

  1. “violating an established moral norm just for fun or personal enjoyment is wrong only in situations where someone is harmed as a result, but is acceptable otherwise.”
  2. “violating an established moral norm just for fun or personal enjoyment is inherently wrong even in situations where no one is harmed as a result.”

Participants who selected the second statement were excluded from analysis.

Reasons or Rationalizations

Royzman et al. (2015) excluded participants based on their endorsing of either principle. We argue that endorsing provides a poor measure of the degree to which participants judgements can be attributed to a given principle. We propose that if a participant’s judgement is based on a given principle, the participant should be able to articulate the principle independently of a prompt, and they should apply that principle across different contexts. We contcuted 3 studies to address this.

Three Exclusion Criteria

We adopted the same method as Royzman et al. (2015), and excluded participants whose judgements could be attributed to either the harm principle or the norm principle. However rather than relying on endorsing alone, we developed additional, and more rigorous, exclusion criteria. We compared rates of dumbfounding based on these different criteria. We also tested the relative accuracy of the exclusion criteria (based on false exclusions).

First we assessed whether participants articulated either principle. We provided an open-ended response question and asked participants to provide reasons for their judgements. The reasons provided were coded for mention of either principle.

Second we assessed whether participants applied the harm principle across different contexts. We asked participants 3 questions about their judgements of behaviours/activities that could potentially lead to harm2

Across three studies we assessed rates of dumbfounding based on these different exclusion criteria.

  • Endorsing (Royzman et al., 2015, all studies)
  • Articulating (all studies)
  • Applying (Studies 2 and 3)

Measuring Dumbfounding

Dumbfounding was measured using the critical slide which read as follows:

“Julie and Mark’s actions did not harm anyone or negatively affect anyone. How can there be anything wrong with what they did?”
1. There is nothing wrong
2. It’s wrong but I can’t think of a reason
3. It’s wrong and I can provide a valid reason

(The selecting of option 2, the admission of not having reasons, was taken to be a dumbfounded response)

Study Overview

  • Study 1: Articulating either principle
  • Study 2: Articulating the norm principle or Articulating and Applying the harm principle
  • Study 3: Replication of Study 2

Study 1 Participants

  • 110 (60 female, 49 male, 1 other; Mage = 32.44, min = 18, max = 69, SD = 11.28) (University students/alumni: N = 52 (35 female, 17 male; Mage = 25.71, min = 18, max = 38, SD = 3.8; MTurk: N = 52 (25 female, 32 male, 1 other; Mage = 38.47, min = 19, max = 69, SD = 12.34)

Study 2 Participants:

  • 111 (67 female, 44 male, 0 other; Mage = 34.23, min = 19, max = 74, SD = 11.42) (University students/alumni: N = 50 (31 female, 19 male; Mage = 28.32, min = 19, max = 48, SD = 6.65; MTurk: N = 50 (36 female, 25 male, 0 other; Mage = 39.08, min = 20, max = 74, SD = 12.25)

Study 3 Participants:

  • 502 (287 female, 212 male, 3 other; Mage = 39.05, min = 18, max = 81, SD = 12.46)


Rates of Dumbfounding depending on Exclusion type

On the graphs below you can see the frequency of each response to the critical slide for each study. The furthest left bars represent the full sample. The responses for each of the sub-samples following the relevant exclusion criterion are displayed and labelled separately.

The response we are intersted in is the red bar denoting a dumbfounded response. Percentages of the full sample are displayed within the plot. Percentages of the relevant sub-samples are displayed in parenthesis below the count.

Study 1:

Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.

Study 1 - Responses to critical slide for the full sample and for each exclusion criterion

Study 2:

Study 2 - Responses to critical slide for the full sample and for each exclusion criterion

Study 3:

Study 3 - Responses to critical slide for the full sample and for each exclusion criterion

As we can see above, dumbfounded responding is found in the full sample for each study.

Replicating the finding by Royzman et al. (2015), if we exclude participants who endorse either principle, rates of dumbfounding are negligible.

However, if we also account for whether people articulate or apply either principle, dumbfounded responding is observed. So whether or not dumbfounding is real, depends on which exclusion criterion is more accurate.

Relative Accuracy of Exclusion Criteria

While we do not have a direct measure of the relative accuracy of the different exclusion criteria, we can assess the relative rates of false exclusions. All exclusions are based on attributing participants’ judgements to either principle. In this case, the judgements should be consistent with the relevant principle. As such, participants who selected “There is nothing wrong” (blue bars above) should not be excluded from analysis. Any participant who selected “There is nothing wrong” and was excluded, was falsely excluded.

Below we have subsetted the participants who selected “There is nothing wrong” across each study, and plot the rates of rates of false exclusion based on each exclusion criterion. The percentage of participants who selected “There is nothing wrong” is displayed within the plot, and the percentage of the full sample is displayed in parenthesis below the count.

Exclusion of participants who selected ‘There is nothing wrong’

As can be seen from the above, rates of false exclusion are much higher when endorsing is the only exclusion criterion. Some false exclusion remains for each of the other exclusion criteria, however it is a considerable improvement. Based on this we conclude that these revised criteria are more robust, and our studies provide evidence that Moral Dumbfounding is indeed real.


Moral dumbfounding cannot easily be explained away by attributing participants’ judgements to endorsed principles. Accounting for whether people articulate or apply these principles provides a more robust measure of inclusion/exclusion. Using this more robust measure, dumbfounded was responding is consistently observed across three studies.



Gray, Kurt James, Chelsea Schein, and Adrian F. Ward. 2014. “The Myth of Harmless Wrongs in Moral Cognition: Automatic Dyadic Completion from Sin to Suffering.” Journal of Experimental Psychology: General 143 (4): 1600–1615.
Haidt, Jonathan. 2001. “The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgment.” Psychological Review 108 (4): 814–34.
Haidt, Jonathan, Fredrik Björklund, and Scott Murphy. 2000. “Moral Dumbfounding: When Intuition Finds No Reason.” Unpublished Manuscript, University of Virginia.
Jacobson, Daniel. 2012. “Moral Dumbfounding and Moral Stupefaction.” In Oxford Studies in Normative Ethics, 2:289.
McHugh, Cillian, Marek McGann, Eric R. Igou, and Elaine L. Kinsella. 2017. “Searching for Moral Dumbfounding: Identifying Measurable Indicators of Moral Dumbfounding.” Collabra: Psychology 3 (1): 1–24.
———. 2020. “Reasons or Rationalizations: The Role of Principles in the Moral Dumbfounding Paradigm.” Journal of Behavioral Decision Making 33 (3): 376–92.
Prinz, Jesse J. 2005. “Passionate Thoughts: The Emotional Embodiment of Moral Concepts.” In Grounding Cognition: The Role of Perception and Action in Memory, Language, and Thinking, edited by Diane Pecher and Rolf A. Zwaan, 93–114. Cambridge University Press.
Royzman, Edward B., Kwanwoo Kim, and Robert F. Leeman. 2015. “The Curious Tale of Julie and Mark: Unraveling the Moral Dumbfounding Effect.” Judgment and Decision Making 10 (4): 296–313.
Sneddon, Andrew. 2007. “A Social Model of Moral Dumbfounding: Implications for Studying Moral Reasoning and Moral Judgment.” Philosophical Psychology 20 (6): 731–48.


  1. Moral dumbfounding occurs when people stubbornly maintain a moral judgement, even though they can provide no reason to support their judgements (Haidt 2001; Haidt, Björklund, and Murphy 2000; Prinz 2005; McHugh et al. 2017). Dumbfounded responses may include: (a) an admission of not having a reason for a judgement, (b) the use of an unsupported declarations (“It’s just wrong!”) to defend a judgement.↩︎

  2. “How would you rate the behavior of two people who engage in an activity that could potentially result in harmful consequences for either of them?”; “Do you think boxing is wrong?”; “Do you think playing contact team sports (e.g. rugby; ice-hockey; American football) is wrong?”↩︎