We have considered Schnall, Haidt, Clore, & Jordan (2008) as evidence for the idea that moral intuitions rely on the Affect Heuristic (as Sinnott-Armstrong et al (2010) propose). Whenever we encounter potential evidence, we should ask two questions of it. First, is it really evidence? Second, is it sufficient to justify us in accepting the claim we take it to be evidence for?
This recording is also available on stream (no ads; search enabled).
On this course you will be evaluating quite a lot of scientific evidence. As this is not something you are required to be familiar with doing before taking the course, I shall go through the process of evaluation quite slowly for the first time.
This includes me, your lecturer. Always evaluate the evidence for yourself.
When faced with a potential piece of evidence, there are three questions you should always ask:
- Has the study been replicated?
- Are there similar studies? If so, are the findings convergent?
- Has the study featured in a review? If so, does the review broadly support the findings of this study?
In the case of Schnall et al. (2008), I originally suggested (in the recording and the first version of these notes) that we answer as follows:
- No, afaik this study has not been replicated. (This is fine; the only concern is when a study has been replicated and the replication failed.1)
- Yes, there are similar studies (e.g. Eskine, Kacinik, & Prinz, 2011); yes, these findings are convergent with those of Schnall et al. (2008).
- Yes, the study has featured in at least one review (Chapman & Anderson, 2013, p. 313). Yes, this review does broadly support the findings of Schnall et al. (2008).
However, since recording this section, I learned that Ugazio, Lamm, & Singer (2012, p. Experiment 1a) report a failed partial replication of Schnall et al. (2008) (thank you Ollie!). Since these authors did not distinguish between high and low private body consciousness, the failure does not appear to be informative and does not undermine the main conclusion (whereas a further update, below, does undermine it).
The review mentioned in (3) provides strong support for the broad conclusion:
‘To date, almost all of the studies that have manipulated disgust or cleanliness have reported effects on moral judgment. These findings strengthen the case for a causal relationship between disgust and moral judgment, by showing that experimentally evoked disgust—or cleanliness, its opposite—can influence moral cognition’ (Chapman & Anderson, 2013, p. 313).
At this point, it seems there is little doubt that we are right to take the findings of Schnall, Haidt, et al. (2008) as evidence. This is what I originally concluded, and what I say in the recording (‘overwhelmingly yes’ at 10:42). However, since then I realised that a meta-analysis by Landy & Goodwin (2015) draws the opposite conclusion,2 as does a recent study (Jylkkä, Härkönen, & Hyönä, 2021; thank you Julina!). Authoritative commentaries by Giner-Sorolla, Kupfer, & Sabo (2018, pp. 261–2) and Piazza, Landy, Chakroff, Young, & Wasserman (2018) conclude that the available evidence is not strong.3 If I were recording the lecture today, I would not be quite so bold. Overall we appear to have only weak evidence.
But there is a further question we should ask before accepting the Hypothesis about the Affect Heuristic.
Previously (in Moral Intuitions and Emotions: Evidence) we considered supporting the Hypothesis that the Affect Heuristic is true by appeal to evidence for the correctness of one of its predictions. But this way of supporting the hypothesis has two weaknesses:
- it is post-hoc (the evidence for the prediction existed before the prediction was generated); and
- for all we know other predictions of the hypothesis may be falsified.
Neither weakness means that the evidence entirely fails to support the Hypothesis that the Affect Heuristic provides a correct account of moral intuitions. But these weaknesses do indicate that we require more robust support for the Hypothesis.
In searching for more robust support, we should consider the most successful arguments for heuristics (in reasoning generally, not in ethics specifically), and use these arguments as a model for what we would need to establish the Hypothesis about the Affect Heuristic.
Schnall, Haidt, et al. (2008) do provide good evidence.
We have sufficient evidence to conclude that feelings do influence moral intuitions (although this point will get further consideration in PS: Does emotion influence moral judgment or merely motivate morally relevant action?).
But this evidence is not by itself sufficient to justify accepting the hypothesis that the Affect Heuristic provides a correct account of moral intuitions.
A different (but related) Affect Heurstic has also be postulated to explain how people make judgements about risky things are: The more dread you feel when imagining an event, the more risky you should judge it is (see Pachur, Hertwig, & Steinmann, 2012, which is discussed in The Affect Heuristic and Risk: A Case Study).
According to Sinnott-Armstrong et al. (2010, p. 256), moral intuitions are ‘strong, stable, immediate moral beliefs.’
A replication can be more or less direct; that is, it may adhere very closely to the original experiment, or it may include varations in the stimuli, subjects and settings. Very indirect replications are sometimes called conceptual replications.
Some of the same authors pubilshed another study in the same year (Schnall, Benton, & Harvey, 2008) which an attempt to replicate has quite convincingly indicated that the effect is not powerful enough to have been discovered by the original study (Johnson, Cheung, & Donnellan, 2014). My recommendation is not to consider studies where there is an informative failure to replicate. ↩
McAuliffe (2019) also provides a review, but this is less nuanced. There are philosophical discussions, offering interestingly different perspectives, in May (2014), May (2018) and Kumar (2016). We will consider Kumar (2016) later in the context of dual process theories. ↩