This text was first published at xphiblog.com on February 28, 2019.
Discussions of moral luck usually start by presenting a pair of agents who engage in the same behavior but bring about very different outcomes. Drunk driving is the usual example. One driver – the lucky driver – arrives home without harming anyone. The second driver – the unlucky driver – hits a passerby. The question is then posed: are they equally blameworthy? Much ink has been spilled on that question (and rightly so). But an interesting issue arises even before we get there, namely, what’s going on with our attributions of luck. It seems odd to call the second driver unlucky. An accident caused by drunk driving seems to be the very opposite of a case in which a bad outcome is simply due to luck. What drives this intuition?
Philosophical accounts of luck often point to features such as lack of control, modal fragility and low probabilities as central to luck attributions. We can fill in the details in the case above in such a way as to have all three features present. And yet, it still seems unintuitive to claim that the accident was due to (bad) luck.
In a new paper, I argue that this is because the folk concept of luck is sensitive to normative considerations. In particular, it is influenced by a normative evaluation of an agent’s action and its relation to the ensuing outcome. Roughly, luck attributions are sensitive to whether the valence of the action matches the valence of the outcome. The idea is that when the valences do not match, we are more inclined to attribute luck (explaining why it seems fitting to describe the first driver as lucky, for it’s a case of bad action/good outcome). And similarly, we are less likely to attribute luck when the valences do match (e.g., bad action/bad outcome, as with the “unlucky” driver).
I tested this hypothesis across five different studies. In one study, I manipulated both the valence of the action and the valence of the outcome, and measured luck attributions. Here is an example of one vignette.
Negligent Action
About to perform a complicated procedure, a surgeon forgets to wash his hands. As a result, the chances of a failed [successful] procedure rise [drop] to about 30%.
As a matter of fact, the procedure fails [succeeds].
Virtuous Action
About to perform a complicated procedure, a surgeon takes special precautions, reviewing each part of the procedure carefully. As a result, the chances of a successful [failed] procedure rise [drop] to about 30%.
As a matter of fact, the procedure succeeds [fails].
Participants indicated their agreement with the following statement, “It was due to luck that the procedure failed [succeeded]” using a 7-point Likert scale ranging from “disagree” to “agree”.
Here are the results:

The results followed the predicted pattern: luck attributions were highly sensitive to whether the valence of the outcome matched the valence of the action. (It’s worth saying that this effect remained significant after controlling for judgments about subjective probabilities, modal fragility, causality, and lack of control).
In a different study, the perceived valence of the action was not manipulated across conditions but rather depended on the moral views of the participants themselves. Participants read a story about a university president faced with the task of deciding whether or not to cancel an upcoming talk by a controversial speaker. The perceived valence of the president’s action, and hence the normative relation to the outcome (success or failure at creating a positive environment at the university), thus varied with individual differences in judgments about what the president should do.
Here are the results:

Luck attributions differed significantly among participants with different moral views responding to the same scenario. For example, when the president decided to let the speaker give the talk and the decision led to a good outcome, participants who disagreed with the decision judged the outcome as lucky. Those who judged the president’s action as morally right, however, did not attribute the success to luck.
It thus seems that normative considerations are an important element in our folk notion of luck. That is to say, describing the first driver as lucky already involves a normative evaluation of her action and the ensuing outcome. And our refusal to attribute luck to the second driver can be partly explained by the fact that we are not inclined to attribute luck when bad actions bring about bad outcomes.
Any thoughts you might have would be very much appreciated!
Literature
Attie-Picker, Mario (2021): “Is the Folk Concept of Luck Normative?,” Synthese 198, 1481–1515. (Link)