Blog Coordinator

X-Phi Grad Programs

« The proto-reliabilist hypothesis | Main | Registration Open: Experimental Philosophy as Applied Philosophy »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Wesley Buckwalter

Personally, I am pretty convinced by these data, and some prior results in moral psychology, that entailment doesn't capture the way people think about obligation and ability. That doesn't mean the two aren't still psychologically connected of course, just that there are exceptions. For this reason, I reject philosophical arguments for entailment that depend on our moral psychology. But I was wondering:

Are there arguments people find especially compelling for entailment that don't depend on intuitions or our moral psychology as premises? Given that this principle seems to have loomed pretty large in moral philosophy, it might be helpful if people could share arguments they've found particularly convincing concerning other ways you might go about demonstrating this entailment relation. It would also be helpful if people have some ideas about why commonsense views about something so basic like "ought" and "can"shouldn't be really diagnostic for theorizing about these things.

Michael Hannon

I gave these vignettes to my class of undergrads, and similarly found they tended to agree with the “ought” attribution in the high blame version, while disagreeing in the low blame version.

After further discussion, however, pretty much everyone who initially agreed that "Adams ought to meet Brown at noon" clarified their response in the following way: they agreed with the "ought" attribution because they meant something like "Adams ought not to have stayed at his house so late, and ought to have left in time to meet Brown at noon." After clarification, most of them agreed that, strictly speaking, it is not the case that Adams ought to meet Brown at noon (now that it's impossible). Rather, they thought Adams ought to have met Brown at noon, which is why they agreed with the "ought" attribution. So many of the people who seemed to reject the entailment from "ought" to "can" in fact did not reject that entailment, but rather interpreted the ought statement in a different way.

Moti Mizrahi

I'm glad to see that OIC is getting more attention from experimental philosophers. But I have to say that the vignettes quoted above are very (very) similar to the ones I have used in my own experimental work on OIC:

Unfortunately, there is no mention of my work in the Cognition paper.

Also, how do you (or the authors of the Cognition paper) suppose the "refutation" is supposed to work? After all, p isn't refuted by the fact that people don't think that p without the addition of some controversial assumptions (

John Turri

Hi Wesley,

Great point. I agree that the most interesting question moving forward is figuring out how judgments about ability and obligation are related, given that the evidence makes entailment an exceedingly unlikely candidate.

I'm not aware of any compelling argument for entailment, so hopefully others who are will chime in.

John Turri

Hi Michael,

That's very interesting evidence worth having in view, thanks for sharing!

I had a question about your interpretation of the outcome of the class discussion. How do you distinguish among students (a) merely clarifying an earlier statement, (b) reversing their opinion on an earlier statement, and (c) changing what they say, without changing their opinion, in response to social cues?

Michael Hannon

Admittedly this was just a casual conversation and I wasn't rigorously trying to control for those, but in a weak attempt to do so, I started by asking them to clarify why they thought Adams ought to meet Brown at noon, rather than starting by trying to interpret their responses myself and asking if they agreed. Once several students said things indicating they meant (something like) "Adams ought to have been there at noon (by leaving earlier)", I then tried to clearly articulate that view, and, once I did, several of the students who made the ought attribution in "high blame" said (something like) "That's exactly what I meant". Admittedly, people can be wrong about what they thought they meant, but I think we should take at least some of them at their word about this.

John Turri

Thanks, Michael, that's a fair point. I wasn't suggesting that you should have been controlling for those things in that setting, as that would be an unduly burdensome standard. When you say that you tried to "clearly articulate that view," was that just with respect to the content of the "ought" attribution, or did it extend to the more general issue of why that would be the correct attribution (in virtue of, say, something like OIC)?

John Alexander

I have only read this post. My issue is with asking what the case is at 11:45, not 11:30 when the choice to leave or stay was made. At 11:45 the conditions have changed in the 'low-blame case' such that the obligation created by the promise is no longer applicable because low-blame Adam would still want to be there at noon and will explain why he is late, etc. (Austin's excusing conditions). Had low-case Adam known that there would be traffic, I suspect that he would have left earlier. When he left for lunch he did not think he would be held-up in traffic.

John Turri

Thanks, John A., I think it's worth considering that objection. In order to evaluate it, I was wondering if you could say what it means for a "obligation" to be "no longer applicable." One way to understand that is we wouldn't hold Adams accountable for failing the obligation (i.e. we won't blame him). Another way to understand it is that the obligation stops existing. You mention an "excusing condition," which seems to lead naturally to the first reading — we excuse the failure. Is that what you had in mind?

Michael Hannon

Hi John T.,

At first I tried to summarize and articulate back to them what I took to be the content of the "ought" attribution (without mentioning the OIC issue), and then after they agreed with the clarification, I told them about the more general issue of OIC, and explained why I was interested in asking them about their thoughts.

A quick aside: a few students strongly agreed with the statement that Adam ought to meet Brown in the "low blame" condition, while also strongly disagreeing with the claim that Adam ought to in the "high blame" condition. When I asked about this, they said it was because it sounded odd to say someone ought to do something they have no desire to do.


There is an ambiguity about what the promise was and what the parameters of its execution were. If I promise to meet you for lunch, and the parameter is at noon, but I arrive at 12:30, have I not met my promise? Well I failed to meet a parameter, but the parameter wasn't part of the content of the promise. I think promises always have parameters that the participants feel they understand. Thus, at 12:30 a person can still keep the promise to meet someone for lunch at noon (i.e, with a noon parameter). The ambiguity is whether the promise had a parameter or whether the promise content included the clause "at noon".

John Turri

Hi Landini,

The possibility for ambiguity is certainly a relevant consideration. I believe that the research team would grant this and, indeed, that they were sensitive to it. That's part of the motivation for how they designed their second experiment, which I did not describe in detail in the original post. One important thing they did was ask people to rate (i) "at 11:45 AM, Adams ought to keep his promise" and (ii) "Adams can keep his promise" in the very same context. They found that responses to these two statements did not significantly correlate. Assuming that participants understood "keep his promise" similarly for the two statements — which seems likely — we can now avoid the concern about ambiguity interfering with our ability to interpret their responses.

Of course, it is theoretically possible that participants filled in the details of "keep his promise" differently for the two statements. But I think this would need to be supported by evidence before it's taken seriously.

John Alexander

John T. Thanks for the reply. I tend to think it is a combination of the two. People make promises within existing conditions that are known, or knowable, to all parties. Adam knowingly and freely promises Brown to meet for lunch under conditions C. Low-blame Adam has determined that it will take 30 minutes to reach the restaurant based on his knowledge of past traffic conditions. He leaves in time, but the traffic conditions are now such that had he known about them he would have 1) left earlier and/or 2) contacted the other person etc. Because the conditions have changed through no fault of Adam's, it seems to me that the obligation is no longer binding other than to explain and justify why he was late. It is no longer binding because he cannot make it by noon, even if he wants too. Brown should accept his explanation and excuse him for being late. In an important sense, the original obligation is still in force because it explains why Adam needs to provide an explanation to Brown for why he was late and why Brown should accept it. Brown should realize that Adam could not keep the agreement, therefore he should not be blamed. Because of the role that 'promising and excusing' play in our moral language and lives it seems that 'ought implies can' applies to the original decision-making context, not to latter (11:45) one. Does not this type of thing happen all the time in our daily lives?

The difference can be seen by this example: imagine that Adam wants to meet Brown for lunch as promised, but on the way he decides to stop for a drink with Joe which makes him late for his lunch with Brown. In this case, Brown would be justified in not accepting the excuse because Adam is late because of a decision he controlled in deciding to stop for a drink. This is a situation that Adam can, and does, control. This also applies to Adams deciding that he does not want to meet Brown for lunch as promised. Adam is in control of the situation in these examples so he can do as he chooses, but not the one at 11:45; he still wants to be there at noon, but now knows he cannot be there.

It would be interesting to ask people what Adam should do if he has a cellphone and knows he will be late. Most will say, I think, that Adams should call Brown because this is something he can, and should, do within the obligation established by the agreement.

Hope this clarifies my point.

Wesley Buckwalter

Interesting results, Michael! Out of curiosity, what did you tell your students was the content of the ought attribution? It might also be interesting to ask your students if the agent had a moral obligation, broke their promise, or should be excused for doing so. For me I always wonder, if there is no obligation existing then what is it that they need to be excused from?

I also worry that the manner of questioning you seem to be taking to reveal people's actual views actually encourages a specific kind of bias to influence people's judgments. I have in mind the biases of blame and excuse validation, whereby we indicate that existing rules were/not broken in an attempt to blame/exculpate agents. This has more to do with evaluating agents rather than identifying the presence of moral rules. They are so easy to mix up.

Landini that's a really interesting analysis of promises. I just asked my family about this, sick of waiting around for me to do things, and they say showing up at 12:30 definitely breaks my promise to do something at 12:00. One data point!


This sort of prompt is in the tradition of recent (past 25 years or so) cognitive economics studies, many of which are open to the same kinds of question. One kind is whether the prompt's scope is limited in the way the experimenters apparently need it to be in order to come to the conclusion they want.

To be specific, look at the work the word "unexpectedly" needs to do in the "low blame" version. To get to the "ought doesn't imply can" conclusion of the experimenters, that word must screen off Adams' responsibility for car maintenance and his knowledge of the probability his car might break down (which can factor in to the appropriateness of the promise). But even if your car is a wreck it can still break down "unexpectedly". And given that well-maintained cars break down quite rarely in recent years, a given breakdown can (absent other information) provide evidence of poor maintenance. This sort of factor might, even subconsciously, influence a quick judgment of likely responsibility, especially in 'fast thought" mode.

So -- flashy headline, but this stuff is really hard to do conclusively and I don't see it in this case.

John Turri

Hi Skef,

Thanks for sharing your thoughts. The researchers don't draw that conclusion based on the results from the "low blame" version here. So I'm not seeing how your point could tend to undermine their interpretation.


Ah, ok yes, I misunderstood. Their conclusion actually rests on this statement “At eleven forty-five, it is still true that Adams ought to meet Brown at noon.” that virtually no one would ever ask outside of thought-experiment-land but differs from a question that at least *might* be asked only by tense ("to have met"). And what you get in the survey are answers that would correspond well to the more natural question. So: maybe many people don't think ought implies can or many people have trouble with trick reading comprehension questions. And we already know the latter from standardized testing.

Still dubious!

Wesley Buckwalter

Hey Josh, thanks for sharing your intuitions. Personally I think the promise means he has an obligation despite changing circumstances, so it's great to chat with someone who has a different view. I was just having trouble reconciling these two claims where you write one, "Because the conditions have changed through no fault of Adam's, it seems to me that the obligation is no longer binding...because he cannot make it by noon, even if he wants too," and then later two "the original obligation is still in force because it explains why Adam needs to provide an explanation to Brown" I agree with you about the second part regarding forgiveness and blamelessness. But what is an in-force but non-binding obligation?


Contemplating this, here's what seems like a better test: Have a set of statements made at 11:45 by different folks including "Smith" saying "You ought to go to meet Brown at noon now." Ask the participant which statements she agrees with. If the issue is really in the concepts of "ought" and "can", that answer should track the answer in the study (right?). If it doesn't, it seems much more likely that the participants aren't tracking all the logical factors the experimenters are attributing to their reasoning, which makes it much less likely that the concepts are at issue.

John Turri

Hi Skef,

I agree that statements of that form occur very infrequently, and we should be mindful of questions potentially seeming weird in a way that affects performance. But, in the present case, it doesn't strike me as in any way infelicitous or difficult to understand. And it definitely doesn't seem like a trick.

Aside from that, I suspect that the researchers were perhaps sacrificing a bit of naturalness in order to narrow down the potential for temporal ambiguity. As you can see from some of the comments on this thread, that's something other researchers can be expected to ask about.

Ultimately, this sort of concern is fully addressed only by testing different wordings.


If part of the premise of experimental philosophy is that we should model philosophical work more on the social sciences, then we should also be concerned that we won't wind up addressing issues of wording by trying different combinations, because studies that give the expected result aren't likely to be published.

John Turri

Hi Moti,

I was really gratified to see that work of yours, thanks so much for linking to it here. It should definitely be included in any literature review on the topic moving forward.

There seems to be a very simple and innocent possible explanation for why the vignettes ended up being similar: there is a common cause in Walter Sinnott-Armstrong's 1984 paper on OIC. Sinnott-Armstrong is a co-author on this paper, you cite his earlier paper in yours, and he proposes similar cases in that earlier paper.

OIC proponents have said that an entailment from "ought" to "can" is a commonsense moral principle or reflected in the meaning of moral language. Then someone looks and notices that patterns in commonsense moral judgment don't reflect any such entailment or meaning. Could you please say what the controversial assumptions here are supposed to be?

John Alexander

John T. Where I thought traffic came into play in low-blame I cannot explain. But having his car break down unexpectedly has the same affect. Sorry for the mis-read.

I have a question: is not saying that 'a implies b' a weaker claim than 'a entails b?' If so, cannot one reject the entailment and maintain the implication? How did your participants understand the relationship between ought and can?

John Turri

Hi, John A. The research team focused on proposals that OIC "is true not only universally, but also necessarily, analytically, or conceptually." Their results undermine these views.

How do people actually understand the relationship between obligation and ability? I think that's the money question moving forward. In my view, the findings discussed above, and other recent results (including Moti Mizrahi's, linked in the comments above), rule out any form of conceptual entailment and contain some interesting leads on how the concepts might actually relate to one another. I expect it to be an exciting area over the next few years.

Moti Mizrahi

Hi John (A.),

I think you may have misunderstood my comment. I was not trying to speculate about what had transpired when the authors of the Cognition paper were writing it. Instead, I was reporting the following facts:

1. The Cognition paper was received 1 July 2015, revised 12 January 2016, accepted 18 January 2016, and was available online 2 February 2016.

2. My paper was available as forthcoming on PhilPapers since December 2014 and published November 2015.

A quick search of “ought implies can” turns up three papers of mine on OIC ( ) on the first page, including the experimental one. There is also my reply to comments contributed to the special issue of Methode in which my experimental work on OIC appeared ( ).

On Social Media, two of the authors of the Cognition paper told me that they simply missed my paper and I believe them. However, as far as I can tell, giving credit to the originator of an idea/experiment/argument is still the norm in academia.

The comments to this entry are closed.