Blog Coordinator

Knobe's X-Phi Page

X-Phi Grad Programs

« Announcement: Character as Moral Fiction | Main | Buffalo Philosophy of Mind Job »



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

Dmitri Gallow

Imagine that I gave people the following vignette:

"Bob is 48 years old. His wife recently left him and the court gave her full custody of Bob's two children. Bob now owns a sports car, dresses in hawaiian shirts and shorts, and spends every night out at some club, chatting up 20 somethings, most of whom find him and his hawaiian shirts repulsive (and don't keep their feelings private). When Bob thinks about the direction his life is going, he feels excited. Bob enjoys going out to the club in his sports car and chatting up 20 somethings, even though these encounters never result in any kind of physical or romantic relationship. When he reflects on the turn his life has taken, he thinks "It's like I'm young all over again! A chance to start from scratch!" Whenever Bob thinks about his life as a whole since the divorce, he feels great."

I bet you would have many, many people say that Bob is not happy. But I bet you'd have far fewer say that Bob is doing anything immoral. Probably, most of them would feel sorry for Bob. I would guess that those people would take the details of Bob's life into consideration, and make the well-justified inference that Bob is emotionally and psychologically reeling in the wake of his divorce. He might have deluded himself into thinking that his life is great, and that he's excited about the future; but that doesn't mean that Bob's life *is* great, or that he *is* excited about the future. He's just stuck in the denial stage of grief.

But none of that has anything to do with the concept of "happiness". It has everything to do with our background knowledge about people like Bob.

And I suspect that something similar is going on with the two Marias. Respondents have background knowledge about people like the attention-seeking, celebrity-chasing, drugged-out Maria. They know that people like that tend to be deeply insecure, unstable, and deluded. We know that people like that generally *aren't* happy, even if they tell themselves and the world that they are.

But put aside my alternative hypothesis. Because, even in the absence of competing explanations, I don't see how you can reasonably draw the conclusion that the difference between the two Marias which is *making* the difference to our happiness judgments is the difference in moral evaluation. There are *tons* of differences between these two people. Why couldn't it be that respondents tend to think that babies make people happier? Or that drugs tend to make people less happy? Or (as I would think) that LA sucks and anybody who lives there must, by necessity, be unhappy?

Moreover: even if we grant that it is the difference in moral evaluation which is making the difference in our happiness judgments, what, besides the fact that you happen to be interested in concepts and not beliefs, makes you think that these results must be due to something in the very *concept* of happiness, as opposed to widely held beliefs about happiness? Why couldn't it be that people accept something like the pop theory of karma, and generally believe that bad people won't be happy, without that belief of theirs being a conceptual truth?

In other words, even if the experimental results demonstrate that respondents don't believe that immoral people are happy (and I don't think the results come anywhere close to showing anything like that), what licenses the inference from that empirical result to the conclusion that this belief of theirs stems from the very *concept* of happiness?

Joshua Knobe


Thanks for these helpful comments. The issues you raise here are important ones, which we would love to discuss in further depth, but let me begin the conversation just by providing a little bit of additional information about our experimental results. (Since this additional information was not in the paper linked above, there is obviously no way that you could have already been aware of it.)

1. Psychologists often understand the concept of happiness in terms of certain psychological states. Specifically, it is often assumed that a person counts as happy if and only if she (a) has a lot of positive affect, (b) has little negative affect and (c) judges her life to be a good one. Our suggestion that value judgments are playing some kind of role can be seen as an alternative to this widely accepted view.
To test between the two hypotheses, we gave participants the case in which Maria lives a hollow and meaningless life and asked them about all three of the features that are usually taken to be constitutive of happiness. Participants overwhelmingly agreed that she had all three of these features (she had a lot of positive affect, little negative affect, and believed her life to be going well). Yet they did not agree with the claim that she was happy. This result suggests, at the very least, that there is something in people's ordinary concept of happiness that is not accounted for in the most widely accepted theory.
(The effect observed here is perhaps interestingly different from the one at work in your example of the man with a mid-life crisis. I suspect that people would not agree that this man had all the psychological features normally associated with happiness.)

2. Participants in the study about the good vs. bad Maria were asked not only about whether Maria was happy but also about whether she had a good life. The impact of condition on happiness attributions was then completely mediated by judgments about whether she had a good life.
In other words, the majority of participants thought that the wholesome mother had a good life and that she was happy. However, those participants who thought that she did not have a good life tended to conclude that she was not happy. Moreover, the entire difference between the two conditions in happiness judgments could be explained by the difference in good-life judgments. (None was due to, e.g., an independent effect of stereotypes about L.A.)

Dmitri Gallow


Thank you very much for your response.

With respect to your first point, I agree that the psychological theory does appear to be lacking (as a claim about the *concept* of happiness, as opposed to a claim about the psychological state of happiness, or a mere stipulative definition of a psychological kind of happiness). However, I should be able to reject that theory without accepting the theory that happiness is in some sense a normative concept.

With respect to the second point, I was probably far too flippant in my earlier comments. Let me try to be more careful: I take it that you demonstrated that there is a positive correlation between people's judgments that Maria has a good life and people's judgments that she's happy. I'm curious about the move from this correlation to the conclusion that people's good life judgments are responsible for their happiness judgments. Why shouldn't we infer instead that people's happiness judgments influence their good life judgments? Or perhaps there is some third factor (like a widely held background belief that lives like L.A. Maria's are both bad and unhappy) which is responsible for both the good-life judgments and the happiness judgments. You write that "the entire difference between the two conditions in happiness judgments could be explained by the difference in good-life judgments." I am guessing that by that you mean that the R-squared for the regression is high. But a high R-squared doesn't rule out confounders like these. And I don't see how they are being controlled for. (Although the link is not working for me, so I couldn't look at the experimental design.)

Joshua Knobe


Thanks for your gracious response. I think that you are getting at something very important here, and I'm glad we are getting a chance to pursue it.

First, a brief clarification about mediation. The basic logic behind the analysis goes roughly as follows. Suppose that you want to predict a participant's happiness judgments, and you were able to use either information about which condition the participant was in or information about whether she thought that Maria had a good life. In that case, the results show that either of these two pieces of information would be useful. But now suppose that you had *both* of these pieces of information. In that case, the results show that it would not be at all helpful to use the information about which condition the participant was in; it would only be helpful to use the information about the participant's good-life judgments. (The information about condition appears to be helpful only insofar as it predicts good-life judgments, which in turn predict happiness judgments.)

The other point you make, though, is a more fundamental one, which I do not think that our present experiments can adequately address. The suggestion is that one could generate an explanation of the pattern of results by putting together (a) a theory about people's concept of happiness and (b) a theory about people's substantive views about which sort of life is likely to lead to which sort of psychological state. Then the theory about the concept might end up saying that the concept is not normative in any way.

This strikes me as a very promising suggestion, and we are trying to find ways of addressing it experimentally. Obviously, the first step here would be to come up with something a little bit more concrete to put in place of (a) and (b) -- there is no real way of trying to explore this possibility in the abstract.

Joshua Knobe

One last thing: I apologize for the problem in finding the supplementary materials. The URL should have been:

Sven Nyholm

For further discussion of the videos, and the experiments, check out the parallel discussion over at pea soup:

Beau Williams

I fully understand everything that this video ia saying. However, there is one thing that troubles me and makes me question the objectivity. The narrator makes an allusion that the respondants ranked the level of happiness of the two Marias according to their lifestyle. (She referred to the first Maria as "wholesome")

That is not a valid assumption to make, however obvious it may otherwise appear. FIRST, the only reason the first Maria could be viewed as "wholesome" is because she was very clearly painted thatway - the respondent's judgement never entered the equation. That is very close to using two men in the video, one short and one tall, then stating that the respondents were judging their happiness by their height. Dude...they followed the choices you forced them in to. (As validation for this argument, you'll need to note that I rankedthe first Maria as happier for a different reason. The first Maria appeared to be in control of her own actions. No mention wasmade that she felt like a bad mom if she did'nt devote so muxh time to the kids -she did it on her own. And it seems that she could modify her focus and dedication any time in reaponse to circumstances. That cannot be said of the second Maria...her actions are not free decisions, rather they are presumed reactions necessary to acheive an identification she has found exciting. She does not have the same level of control, because any adjustment she makes will be dictated by the actions and choices of others. Thus, her happiness is not as genuine because what she really is experiencing is the relief that she hasn't failed. Her so-called "happiness" can be taken away at any time because it is generated by others. No onw can take away the happiness of the first Maria because her happiness is aelf-generated.

With that in mind, maybe what we need to consider is that how we define things like "wholesome" is really an instinctive expression of what makes us happy, healthy, etc., and not the other way around. In any case, assumptions like what I pointed out at the start will skew the interpretation of the results.

Jonathan Phillips

Hi Beau,

Thanks for your thoughtful comments. You make a couple of nice points.

First though, I should probably make it clear that in the original study, we did actually ask participants how good they thought Maria's life was. (This just wasn't included in the video because it was a little bit more complicated.) As you might expect, people thought that the Maria, who was described as a mom, was living a better life than the Maria who was living in LA. Interestingly, these judgments of Maria's life were actually better predictors of participants' judgments of happiness than just the information about which Maria they were assigned to read about. So, for example, if a person read about the Maria who was a mom, but didn't think she had a good life, then that person was much less likely to think that she was happy. (Josh says a little bit more about this kind of analysis in his above post to Dmitri.)

Your other point about having control over one's life is also a really interesting one. I hadn't thought about this before. In some subsequent studies we've run, we used completely different stories which might help to address the issue your raised. So, in one example, participants were told about a man named Mark who either spent his life volunteering in Africa or spent his time hopping around nightclubs in America. Both of the men ended up being equally dissatisfied with their lives. We found that participants judged the Mark who was volunteering in Africa to be happier than the Mark who spent his life in nightclubs. In contrast, participants judged them to be equally unhappy. So, at least in follow-up studies like this one, it doesn't seem like it's really the control the person has over their life which is creating the pattern of responses. Of course, that's not to say that the control wasn't playing a role in this original study (or in your own responses) -- I suppose that only further experiments will be able to address that question.

Thanks again for your thoughtful comments!

The comments to this entry are closed.

3QD Prize 2012: Wesley Buckwalter