Joshua Knobe has run a series of experiments that suggest that our intuitions about whether or not an action was intentional may be influenced by whether or not the effects of the action were harmful/bad. Here’s a brief summary of one of those studies.
In a case where an exec sought to max company profits knowing that the actions would harm the environment, a majority of people judged that the exec intentionally harmed the environment. In a second case where the only difference was that the exec knew the actions would benefit the environment, a majority of people judged that the exec did not intentionally benefit the environment.
In his paper “The Concept of Intentional
Action” Knobe addresses some alternative explanations of the data.
One is offered by Alfred Mele. He argues that people who shift their judgements about intentionality based on the badness of the effects may have an explicit belief like:
(A) All blameworthy actions are intentional.
The idea is that they entertain the first exec scenario, and judge that he is doing something worthy of blame. With (A) they infer that the action is intentional. They have no corresponding judgement about blameworthiness to compel them to draw the inference in the case where the exec benefits the environment.
Knobe argues that Mele’s explanation fails and cites a study to control for Mele’s explanation. Persons were given a drunk driving scenario in which many judged that the harm caused was blameworthy, but that the action that directly caused the harm was not intentional. The subjects that judged this way were then given some of the original Knobe cases. Their judgements about intentionality still seemed to shift in much the same way. However, their answers in the drunk driving scenario suggest that they do not believe (A) – so Mele’s explanation fails. (Note: Knobe says Mele has recanted this explanation in light of the empirical data)
But there are a couple of hypotheses in the spirit of Mele’s that could do the work of Mele’s original hypothesis. Consider the following two theses.
(B) Most blameworthy actions are intentional.
(C) All blameworthy actions ultimately result from some intentional action.
If someone believed either (B) or (C) they might make some kind of inference to the conclusion that the exec in the harm case acted intentionally, but also judge that harms caused by drunk driving are blameworthy and unintentional.
(Posted from Android phone)
That’s an interesting proposal, but I’m not sure (B) and (C) are enough. First, (B) plus the details of the Harm Case aren’t enough to warrant the inference to harming the environment intentionally. Presumably a good explanation would have to cite the additional belief(s) that puts this case in the category of most. Second, (C) doesn’t seem sufficient either. (C) plus the fact that harming the environment (the side-effect) is blameworthy only warrants the conclusion that harming the environment was the result of some intentional action or other. But subjects judged that the harming of the environment itself was done intentionally. Wouldn’t we need to attribute more beliefs to explain their coming to that conclusion?
But maybe I’m missing something. It has been awhile since I’ve read the paper. In any event, it doesn’t seem like it would explain the data that has since been gathered by Machery and others which suggest this phenomenon doesn’t have much to do with blame at all. Or are you just trying to see what moves can be made here given only the data Knobe reports in this paper?
I was just trying to capture the Knobe data. I haven’t read the Machery stuff. Do you have a reference?
I’m now worried about (C) too.
It does still seem, however, like (B) would be enough. The inference would course be non-deductive.
Regarding the bit about going from harm to blameworthiness, I think Mele’s idea (although I haven’t read Mele’s original paper). Is that judgement about blame is made when presented with the first scenario, and he steps in at that point to explain where the intentionality judgement comes from. He does it with (A). But here’s how it could go with (B)
1. The execx action of harming the environment is blameworthy
2. Most blameworthy actions are intentional
3. Therefore. The exec’s action of harming the environment is intentional.
(Again, non-deductive of course)
Am I missing something on that one?
p.s. I’m curious how the Machery experiments were set up
The Machery paper came out in 2008 in Mind and Language. Knobe cites it in the book in his paper. It looks like the citation was added for the book for the footnote—the one in which he says that everyone has subsequently shown that he was wrong! 🙂
Machery’s trade-off hypothesis and the experiment he ran to test it are quite good, I think. He basically finds the asymmetry in judgments for cases in which there is no blameworthiness or moral character at all. His trade-off hypothesis is that people tend to think that a side-effect is done intentionally when it’s perceived as a cost needed to incur a gain—e.g. paying an extra dollar for the smoothie you want (when you don’t care about paying more) versus getting the smoothie in a commemorative cup (when you don’t care about the cup).
I see where you’re going with (B), but I’m still not sure about it. Even if the argument/inference is supposed to be non-deductive, I’m not sure it’s a good one. It really depends on how many “most” is in P2. And if it’s not a good inference and there’s no reason to think people are making this error, it seems like a bad explanation.
But I do think it’s a good move to press on Knobe here. It seems important that Mele needn’t attribute to people the belief that ALL blameworthy actions are intentional. But it does still seem like P2 has to be based on something like the belief that a great deal of them are. And then the plausibility of the explanation might hang on the plausibility of attributing such a belief to people.