Experimental Philosophy: Morality, Intentionality, and Values (Oh My!)

If you haven't been following the On-line Philosophy Conference (OPC), you should go over and check it out. It has just finished its third of four weeks, and there were some good papers in this round. I was particularly interested in the paper by Joshua Knobe and Erica Roedder titled "The Concept of Valuing: Experimental Studies," because it's experimental philosophy, and it's by Knobe, whose work is easily the most interesting in experimental philosophy.

Before I get into Knobe's work, and the Knobe and Roedder paper, let me tell you a story. When I was in graduate school, I went to see a talk (that sounds weird, doesn't it?) by a pretty well-known philosopher of mind. I won't mention his name, but as a hint, I will say that you can wear him with a dress shirt, and if you're wearing a tuxedo, you're probably wearing him in a bow. I've forgotten the exact topic, but it had something to do with the unity of conscious experience. The talk was filled with counterfactuals, hypotheticals, and little thought experiments. In other words, it was pretty much everything you'd expect from "conceptual analysis." By the end of the talk, it was all I could do to stay awake. I wasn't just sleepy, though; I was frustrated and sleepy, and those two states do not work well in combination. I was sleepy because conceptual analysis is, well, boring. I was frustrated because he was talking about an issue that I felt, and still feel, would be better addressed through empirical, rather than conceptual analysis. And that's not simply because I'm coming at it from the usual scientific perspective, which amounts to, "Yes, but where's the data?" It's also because the unity of consciousness is an empirical topic, and questions about it are empirical questions. So they should be asked and answered that way. When I stood up to ask a question about the potential empirical implications of his position, I probably looked like a raving narcoleptic, and I wouldn't be surprised if most of the philosophers in the room dismissed them. The speaker certainly didn't seem to find them very interesting. But I think they were important questions, nonetheless.

The point of my telling this story is that even before I had heard of experimental philosophy, I was a fan of it. I felt that there were many philosophical questions, especially those pertaining to problems of mind and psychology, that could best be answered by conducting experimental investigations. When I learned of experimental philosophy, I was very excited. Finally, philosophers who recognize that empirical investigation is essential. Then I started to read some experimental philosophy, at first by accident; I stumbled upon a Stich paper in the journal Cognitive Psychology, without realizing he was a philosopher, or that he was doing experimental philosophy. It turns out, they weren't doing quite what I had in mind. They were doing conceptual analysis, and attempting to answer questions that only arise out of conceptual analysis, but they were doing it with an n larger than 1. That's better than doing it with an n of 1, but it still doesn't feel like it's quite right.

The point of my telling you that is this: the one experimental philosopher I've read whose work feels like it is, or at least could be, an example of the empirical approach to philosophy that I've always envisioned, is Joshua Knobe. In fact, he's already used his experimental work to say something about problems that arose out of empirical psychology, and not conceptual analysis in philosophy (see this paper). So that's why I find Knobe's work to be the most interesting in experimental philosophy, and also why I was eager to read his OPC paper. Now I'm eager to talk about it, and since I have a blog, that's what I'm going to do.

Let me start (I say "let me start" in the 4th paragraph; that's funny) by describing some of Knobe's older work on intentionality. His most well-known experiment used the following scenarios (from this paper, p. 3):

Scenario 1: The vice-president of a company went to the chairman of the board and said, 'We are thinking of starting a new program. It will help us increase profits, but it will also harm the environment.'

The chairman of the board answered, '‘I don'’t care at all about harming the environment. I just want to make as much profit as I can. Let'’s start the new program.'

They started the new program. Sure enough, the environment was harmed.


Scenario 2: The vice-president of a company went to the chairman of the board and said, 'We are thinking of starting a new program. It will help us increase profits, and it will also help the environment.'

The chairman of the board answered, 'I don't care at all about helping the environment. I just want to make as much profit as I can. Let's start the new program.'

They started the new program. Sure enough, the environment was helped.

After reading one of these scenarios, participants were asked whether the chairman had intentionally harmed/helped the environment. Eight-two percent of the participants who read scenario 1 said that the chairman had intentionally harmed the environment, while 77% of the participants who read scenario 2 said that the chairman had not harmed the environment. Using this, and several follow up experiments (check out Knobe's homepage for links to his many papers on this topic), Knobe has argued that morality plays a role in our use of the concept of intentionality. I'll say more about these experiments in a moment, but for now, let's move on to the OPC paper.

The OPC paper is, as the title suggests, on valuing. It's a short paper, so you should probably just go read it yourself. Go on, I'll wait. OK, back? Since you've just read the paper, you know that it presents an argument against a view of the concept of valuing in the philosophical literature that Knobe and Roedder put thusly (p. 1):

[T]he concept of valuing can be defined in purely descriptive, non-normative terms.

They argue that, contrary to this view, "moral considerations actually play a role in the concept [of valuing]." They don't claim that moral features are the only, or even the primary features of our concept of valuing, but that, when other features are absent, moral features can determine whether we believe a person does or does not have a certain value. As evidence for their position, they present the results of two experiments. Since you've read the paper (you haven't? Go, now! I'll wait again), you don't need me to describe the experiments, but because I'm a nice guy, I'll do so anyway. In their first experiment, they use these two scenarios (p. 3-4):

Scenario 3: George lives in a culture in which most people are extremely racist. He thinks that the basic viewpoint of people in this culture is more or less correct. That is, he believes that he ought to be advancing the interests of people of his own race at the expense of people of other races.

Nonetheless, George sometimes feels a certain pull in the opposite direction. He often finds himself feeling guilty when he harms people of other races. And sometimes he ends up acting on these feelings and doing things that end up fostering racial equality.

George wishes he could change this aspect of himself. He wishes that he could stop feeling the pull of racial equality and just act to advance the interests of his own race.


Scenario 4: George lives in a culture in which most people believe in racial equality. He thinks that the basic viewpoint of people in this culture is more or less correct. That is, he believes that he ought to be advancing the interests of all people equally, regardless of their race.

Nonetheless, George sometimes feels a certain pull in the opposite direction. He often finds himself feeling guilty when he helps people of other races at the expense of his own. And sometimes he ends up acting on these feelings and doing things that end up fostering racial discrimination.

George wishes he could change this aspect of himself. He wishes that he could stop feeling the pull of racial discrimination and just act to advance the interests of all people equally, regardless of their race.

In the Scenario 3, George consciously harbors values that most people in our society would consider to be morally bad, but "feels a certain pull" towards values that we would consider morally good. In Scenario 4, the opposite is true. George consciously believes he values something we would consider morally good, but "feels a certain pull" towards values that we would consider morally bad. After reading the scenario, the participants were asked whether they agreed with these statement: "Despite his conscious beliefs, George actually values racial equality" (for Scenario 3) or "Despite his conscious beliefs, George actually values racial discrimination" (for Scenario 4). They rated their agreement on a 6-point scale (-3, disagree strongly, +3, agree strongly). Participants' were significantly more likely to agree with the statement for Scenario 3 than for Scenario 4. In other words, they were much more likely to believe that the man in Scenario 3 really did value something they thought to be morally good than they were to believe that the man in scenario 4 valued something they thought to be morally bad.

In their second experiment, Knobe and Roedder used this scenario:

Situation 5: Susan grew up in a religious family, but while she was in college, she started questioning her religious beliefs and eventually became an atheist.

She will be getting married in a few months to her longtime boyfriend. Recently, the subject of premarital sex has come up.

Susan definitely has a desire to have sex with her boyfriend, but whenever she thinks about doing so, she remembers what her church used to say about premarital sex and feels terribly guilty. As a result of these feelings, Susan has not had sex yet.

Because she is no longer religious, Susan believes there is nothing wrong with premarital sex.

They gave this scenario to two groups: members of a Mormon Bible study group, and people in Washington Square Park in New York. As you might expect, members of the Mormon Bible study believed that refraining from premarital sex was morally good, while the Washington Square Park participants didn't really care about premarital sex. And consistent with the results from their first experiment, those who thought that refraining from premarital sex was morally good were much more likely to believe that Susan valued it as well than those who didn't value premarital sex.

So, from these two experiments, it really does appear that, "people's intuitions about an agent'’s values depend in part on moral considerations" (p. 6). The commenter on the Knobe and Roedder paper, Antti Kauppinen, has a different interpretation, appealing to Donald Davidson's "principle of charity." I won't get into that, but you can read the commentary here. Instead, I'm going to argue that the studies don't tell us much about valuation.

Here is the problem I see with both experiments. One of many proper control conditions for these studies would involve a scenario like this:

Scenario 6: Susan lives in Atlanta, where most people are Braves fans. She thinks these people are basically correct. That is, she enthusiastically roots for the Braves, regardless of who they're playing.

Nonetheless, Susan sometimes feels a certain pull in a different direction. She often finds herself feeling like the Cubs are a better team to root for. And sometimes she ends up acting on these feelings and rooting for the Cubs.

Susan wishes she could change this aspect of herself. She wishes that she could stop feeling the pull to root for the Cubs and just root for the Braves.

You'd then give this scenario to people in Atlanta and people in Chicago, and ask them how much they astatementh the statment: "Despite her conscious beliefs, Susan is actually a Cubs fan." You might predict that Chicagoans (who would be Cubs fans, for the most part) would believe that Susan was really a Cubs fan, while Braves fans in Atlanta wouldn't. Presumably Cubs and Braves fans don't have moral reasons for valuing one team over the other (unless they live in Boston), so if this prediction was confirmed, it wouldn't be a result of moral considerations. OK, I know that's pretty silly, but I threw it together as I typed it. If you don't like it, substitute it with a scientist who's feeling a pull towards a rival theory. The point is, if you're going to argue that moral considerations at work, you have to test similar scenarios in which morality is not an issue. It may simply be that people are inclined to believe that people agree with them, and all they need is one piece of evidence to confirm that belief. In that case, it's just motivated cognitionbias my-side bias, an illusion of common ground, or something similar at work. Sure, that could still mean that moral considerations would be at play in intuitions about valuation, which might result in the same philosophical implications knobe and Roebberson believe their position does, but it could also mean that other considerations based on evidence of other common belief could influence those intuitions as well. The problem Knobe and Roedder's experiments is that they don't test any potential psychological mechanisms, and thus we don't know the extent, or the cause, of their results.

That problem's not unique to these studies, either. Knobe's intentionality studies (e.g., the one using Scenario 1 & 2, and those linked at Knobe's homepage) don't test different possible mechanisms, either. One might be able to explain the results of the experiment using Scenario 1 & 2 by reference to research on counterfactual thinking, for example. We know, for instance, that people are much more likely to counterfactually mutate negative events/outcomes. Counterfactual reasoning is also associated with attributions of blame and experiences of guilt and regret. Perhaps people are more likely to attribute intentionality to the chairman in Scenario 1 because they're more likely to reason about it counterfactually (simply because it involves a negative outcome; negative for moral reasons, for sure, but it would be more likely to be mutated if it were negative for any other reason), and thus to assign blame (which might require intentionality), whereas Scenario 2 involves a positive outcome that is not likely to be counterfactually mutated. If counterfactual reasoning is involved in that way, then the implications for our intuitions about intentionality is much different. It's still related to morality through blame, which wouldn't be very surprising, but morality would play an indirect role in the assigment of intentionality through the mutability of negative outcomes. That explanation might not work, but we don't know, because no psychological mechanisms have been tested. And thus, we don't really know how to interpret the data.

As I've arrived at the end of this post, I feel like I've been overly critical of Knobe's (and Roedder's) work. The truth is, I find his experimental designs to be ingenious, and his results often surprising, at least at first glance. And as I said earlier, I think his work is really the most promising in experimental philosophy. Still, there's one more step to take: connecting these experiments more directly with the larger literature on concepts and reasoning, and then testing the predictions of different possible psychological mechanisms in order to tease out the ways in which morality plays a role in intuitions about intentionality and values. Once researchers begin to do that, I'll be the biggest fan of experimental philosophy out there. And if that's not motivation enough, then this might be. Understanding the mechanisms involved will help us to develop a much greater understand our intuitions related to important philosophical issues. If understanding those intuitions is as important to philosophy as experimental, then such work would be of great importance to philosophy.

UPDATE: If you read this post before this update was added, you may have noticed some weird things going on. You can thank Blogger's spell check for that. Hopefully most of them are gone, now.