Joshua Knobe and Erica Roedder

Joshua Knobe (UNC-Chapel Hill) and Erica Roedder (New York University), “The Concept of Valuing: Experimental Studies,” with commentary by Antti Kauppinen (University of Helsinki).  Both the paper and the commentary can be found here.

Comments

[Neil]

Here’s a prima facie problem for the view: it seems that there are cases in which (if I understand Knobe and Roedder) I predict the response will go the other way. These are not cases concerning what values to attribute to an agent, but in which we ask whether an agent deviates from what’s widely taken to be a norm. Here are two:

Sexuality:

(a) John is married and has been for 12 years. John has been faithful throughout his marriage, but he often finds himself fantasising about having sex with men. Is John really homosexual?
(b) John has lived in a committed gay relationship for 12 years. John has been faithful throughout his relationship, but he often finds himself fantasising about having sex with women. Is John really heterosexual?

Race:

(a) George has 7 white grandparents and one African-American grandparent. George considered himself white, but some people think he’s really African-American. Is George really African-American?
(b) George has 7 African-American grandparents and one white grandparent. George considered himself African-American, but some people think he’s really white. Is George really white?

I predict that subjects would be more willing to say that John is really homosexual in vignette (a) than to say that he is really heterosexual in vignette (b); similarly that George is really African-American in vignette (a) than to say that he is really white in vignette (b) – that adherence to the norm has, as it were, a higher threshold than departures.

Of course, here I’m reporting my intuitions. I’m quite confident, but I could be wrong. If I’m right, this seems to conflict with the valuing claim, where adherence to the valued norm seems to have a lower threshold.

[Antti Kauppinen]

I think the sort of charity-based explanation I adopt in my response can handle both Josh and Erica's data and Neil's examples, at least the first one (I haven't thought about the second). I think Neil's intuition is plausible, at least if we add a little detail to the fantasizing bit - John spends a lot of time surfing the net for gay/straight porn, and so on. Here we have a prima facie case of what we could call a behaviour/attitude mismatch. This calls for explanation, since it's a deviation from the norms of practical rationality. Now, surely it's a part of folk wisdom that external sanctions often lead people to act in ways that don't express their true selves. Our social norms discourage homosexual behaviour. So it makes sense that someone who has gay inclinations might want to keep them in check. With heterosexual leanings, it's the other way around. On the face of it, it doesn't make sense why someone who's deep down a heterosexual would suppress his sexuality and remain in a gay relationship. (It would be more plausible that he's really bisexual.) So, if John B was really heterosexual, his relationship behaviour would be baffling, while we can easily understand John A's plight if he's really gay. Hence the asymmetry. Consequently, there's an inclination to conclude that John B isn't hetero. Of course, his fantasizing and surfing would be pretty odd if he wasn't bisexual (which is what I suppose would be the characterization favoured by most people in both cases, but especially the second). But being bisexual is being non-hetero, so that fits with the assumed pattern of response.

Incidentally, I suspect the more romantically inclined would read John B's case as one of a tragic love for the man to whom is committed and who he can't forsake, even though he's deep down a heterosexual. Here the suppression of hetero desires would make sense. And I wouldn't be surprised if those who do take sexual 'preference' or orientation to be a matter of valuing - a slightly odd position - like maybe the Mormon Bible study group, would after all deploy their own values in interpretation and conclude that both Johns are really heterosexual, though the first is misled by homosexuality-favouring popular culture to experience sinful temptations and the second is caught up in a abominable relationship, probably due to personal intervention by the Devil.

Finally, even if people did take sexual orientation to be a matter of valuing and thought being gay was bad, Neil's predicted responses wouldn't as such contradict the 'valuing the good' thesis. All it says is that people are *more likely* to attribute values they share, which isn't to say they're not prepared to attribute values they don't share when other considerations, such as the ones I point to above, are relevant.

[uriah]

Hey Erica and Joshua - exciting and imaginative work as usual. A few questions/worries (with apologies for length thereof):

First, since the work of Tversky and Kahneman, we've been aware of many instances of irrationality in performnace of cognitive tasks. Someone might say that your findings show that we are guilty of irrationality and incompetence in applying a non-normative concept of valuing, not that we are competently and rationally apply a normative concept of valuing. In many ways this would be the more “conservative” approach to the data.

One way to think of this “someone” is as proposing an alternative hypothesis to explain the data. This would imply a need for devising an experiment that would address the “confound.” But a more radical way to think of this troublemaker is as claiming that the experimental method is blind to the distinction between normatively positive (= competent) application of a normative concept from normatively negative (= incompetent) application of a non-normative concept, in the sense that it doesn't have the means to undo any confound with that form. Something else needs to be adduced to determine what competent performance of the task would be. In the case of Tversky, this was “something else” is provided by the laws of logic: they (rather than any experiment) tell you what competent performance in inductive reasoning should look like. In the present case, the natural thought is that conceptual analysis tells you what the concept is - whether it is normative or not - and thus what competent performance in the task of applying it would amount to. The experiments then come in and tell you whether we performed well or not.

Second, another troublemaker might come along and suggest that people use information about what really is valuable in deciding whether to apply the concept of valuing to someone simply by way of deploying a Davidsonesque principle of charity. Just as we try to interpret people's verbal behavior so that their beliefs come out true, so we try to interpret it so that their valuings come out correct. Again there are two versions of this objection: one that invites further experimentation to separate out even more factors, and another one that challenges the foundational soundness of the experimental method.

This connects to a wider question, which is a pet peeve of mine, regarding the experimental method's ability to distinguish (i) non-normative application of a normative concept from (ii) normative application of a non-normative concept. You go with (i), but I want to go with (ii). The thought is that there are normative factors that affect our application of the concept of valuing, but the concept being applied is purely descriptive. And here again there will be two versions of this worry, with the radical one claiming that the experimental method cannot distinguish (i) from (ii).

Thirdly, a student of mine once told me he thought the method you used in this particular study, though not necessarily the experimental method in general, was blind to the distinction between causal relations and constitutive relations. Although you can claim that there is a relation between the concept of valuing and what is really valuable, you haven't shown that the relation is constitutive rather than causal. This may well come to the same as the second point above, and in any case is in the ballpark. Although my student preferred the unradical version of the worry, a radical version can also be formulated as before.

[Antti Kauppinen]

(Uriah:) "another troublemaker might come along and suggest that people use information about what really is valuable in deciding whether to apply the concept of valuing to someone simply by way of deploying a Davidsonesque principle of charity".

Welcome to the club, Uriah! ;) This seems to be the spontaneous reaction of most people I know to Erica and Josh's data. I, too, develop the Davidsonian line of response at (too much) length in my commentary paper, available on this very site...

[uriah]

Mea culpa, Antti - I haven't read your commentary yet, but will do so soon. I'm curious to see how you develop this criticism.

[Yentz Mahogany]

Josh / Erica:

Let's assume that the methodology etc. is sound (though I take Uriah's first comment quite seriously).

If you insist on calling this "folk psychology", then I suppose that the results might seem normative. However, if you understand this to be a project in the analysis of folk interaction, then it may be understood as descriptive.

A value may be understood as a relational/relative fact, and not a entirely a dispositional one. We must seriously consider the possibility that valuing is a form of meaning, and hence, a social-psychological phenomenon which is the result of interaction; while attitudes are (ostensibly) purely dispositional, and reflect the psychological facts of the situation. Mere psychological states of a single individual involve attitudes, which are different from values.

In other words, while Kauppinen's (DCV) is consistent with (AT), it is insufficiently capable of capturing the nature of a value; for it necessitates only a single individual, which may be an error. So we'd replace Kauppinen's (DCV) with:

The Social-Descriptive Concept of Valuing (SDCV): The folk concept of valuing is the concept of having certain psychological states about someone else.

The outcome would be that one could never know one's own values, except in the context of one's observations of their own interactions with others -- those things which a person is willing to "take to the mat" in their relationships.

I've never been impressed with Davidson and don't see how the principle of charity can be gleaned from this one study alone. The Davidsonian interpretation differs from the one I offered here in that we can expect that a control group (x) who hold a negative view of the people under their examination (y), they will be happy to attribute with a negative bias. People aren't charitable unless they feel a disincentive to appear irrational in front of an experimentor, or if they see a halo around their subject.

[Jonathan Weinberg]

I believe that the charity-type move is what I asked you about at our session in Portland, Joshua -- it does indeed seem to be a popular response! I'll be curious to hear/read what you want to say in response to it. (I should note, in response to Yentz's anti-Davidsonian complaint, that one doesn't need anything even remotely like the entire vast machinery of the Davidsonian framework. All one needs is to have a bias towards attributing pro-attitudes to others based on the pro-attitudes that one has onself.

[Yentz Mahogany]

Hi Jon,

Charity, however, doesn't seem good enough. It may be sort of on the right track, but may not be sufficient.

[Joshua Knobe]

Thanks so much for all of these comments! They're really very helpful.

Here are a few quick thoughts in reply:

1. Uriah, Antti and Jonathan all suggest that our results might be explained in terms of the principle of charity. So perhaps it would be helpful to say a little bit about the difference between charity-based explanations and our own approach. The key difference is this: The principle of charity is supposed apply quite generally to all aspects of folk psychology, whereas our explanation suggests that there is something special about *the concept of valuing in particular* that makes it susceptible to the influence of moral considerations. That is to say, our explanation suggests that the concept of valuing differs in some way from other aspects of folk psychology.

2. Neil brings up some evidence that seems relevant here. He shows that people's use of various other folk-psychological concepts is not susceptible to the influence of moral considerations in the same way that their use of the concept of valuing is. This evidence suggests that, whatever it is that allows moral considerations to impact people's use of the concept of valuing, that same factor does not also influence people's application of other folk-psychological concepts.

3. But the real clincher comes from studies of people's use of the concepts of intention and intentional action. These studies show precisely the *opposite* effect. (People are actually more likely to say that someone has an 'intention' to perform an action when that action is bad.) Results like these indicate that it is a mistake to search for one single mechanism that explains all of the ways in which moral considerations affect the application of folk-psychological concepts. Different concepts work in different ways.

4. Uriah suggests that we ought to seek a standard of rightness that lies outside the bounds of the experimentally verifiable. This is an interesting point, but its status seems to depend on the purposes that animate our inquiry in the first place.
If our aim is to understand the fundamental nature of folk psychology, then it probably won't be helpful to look for standards that lie outside the realm of what can be verified experimentally. Instead, we should try to understand the underlying competence behind people's ordinary folk-psychological attributions.
By contrast, if our goal has nothing to do with folk psychology per se, it might be helpful to search for some standard against which the fundamental competence might be judged and found wanting. I would be curious, though, how one might go about finding such a standard. Do you have any suggestions?

[Jussi Suikkanen]

Joshua,

few quick points. About your 1. Wouldn't it be an advantange of the charity based explanation that it is a more general explanation about how we attribute mental states to others? At least scientific theories are usually to be preffered if they can account for more things.

About 3. If I remember your paper correctly, what the people were asked was were the actions in the relevant cases intentional and what they did in order to achieve such and such (at least in the Phil Studies paper). There is quite a gap from there to attributions of folk-psychological states like intentions with particular contents. I take it that according to the folk psychology there are plenty of intentional actions for which the agents don't have specific intentions (Mele's example: I have a standing intention to walk to campus each morning; I take my steps along the way intentionally; I don't have an intention to take a separate set), and there are also plenty of unintentional action for which agents' have the relevant intentions (lack of control cases usually). Given this it might turn out that the attribution of different intentions is explainable using the principle of charity.

[Yentz Mahogany]

Joshua,

You wrote: "...our explanation suggests that there is something special about *the concept of valuing in particular* that makes it susceptible to the influence of moral considerations."

According to a (probably controversial) meta-ethical doctrine of moral interactionism, all morality emerges from interaction and/or is directed towards interaction. It is a view with interesting philosophical and empirical implications, and may be interpreted to be a part of the doctrines of many prominent philosophers; but would require an essay to express.

Thus, any moral claim might necessarily have a second-order character, involving mental states about the mental states of others. The concept of value may also (arguably) have this character. If true, the relation between valuation and moral significance would seem far more obvious.

As I've indicated, all of this may be in the borderlands of folk psychological research, but still by far within the territory of the empirically verifiable (or cogently deducible, which is often all you can ask for in the social sciences).

You wrote: "(People are actually more likely to say that someone has an 'intention' to perform an action when that action is bad.)" But is it really this simple? I would expect that the friend of a person who does something bad shall attempt to put the best light upon the action done.

[Joshua Knobe]

Thanks Jussi; these are really interesting points.

With regard to 1: You are right to say that we should usually prefer more general explanations to more specific ones. The key question, then, is whether the more general explanation can actually be made to work in this case.

With regard to 3: You are right to say that I originally emphasized the effect on the use of 'intentionally' rather than the effect on the use of 'intend' or 'intention.' Although my first study showed a small effect for 'intention,' I didn't make a big deal of it and actually suggested that the real effect might only arise for 'intentionally.' But further research has not been kind to my hypothesis. Studies by David Pizarro, Eric Uhlmann, Thomas Nadelhoffer and Fiery Cushman have all shown that powerful effects can be found for the use of 'intend.' Their work shows pretty clearly that moral considerations can actually influence attributions of psychological attitudes.

[Yentz Mahogany]

Jussi,

Mike goes before the judges of American Idol. The judges tell him he is a horrible singer. Because of the 'fundamental attribution error' (in social psychology), Mike interprets this as the expression of the lousy aesthetic values and tastes of the judges. It turns out that Mike really is a horrible singer.

Was Mike trying to optimize agreement? Or was he resisting optimization in order to protect his own ego and related projects?

[Joshua Knobe]

Yentz,

These ideas sound promising, but it's a little hard to understand them in this condensed form. Do you think you might be able to explain at greater length precisely what you are getting at?

[Yentz Mahogany]

Joshua,

I'm basically saying that:

1. Moral interactionism is the view that a) all moral claims arise out of interaction, and/or b) all moral claims must be other-directed. A 'values-interactionism' would be similar, where a) all values are produced by social interaction, and/or b) all values are attributions of dispositions or attitudes to others.
2. An 'interaction' is not quite the same as folk psychology, since it depends on the existence, minds, feelings, etc. of two people (social psychology).
3. If (1) and (2), then moral claims are not quite in the domain of folk psychology; rather, they're about folk interactions, which involves a different set of concepts.
4. When we talk about the facts of a person's psychology, their dispositions, attitudes, etc., their claims about values may seem to be normative. By contrast, facts about an interaction, especially concerning the person's attribution of someone else's attitudes, do not appear to be normative, but descriptive (albeit a description through the warped glass of perception, which may be either charitable or malign).
5. Moral interactionism would be related to a kind of values-interactionism in that the former would be composed out of the latter. Bob could say, "X is bad", and then I would be able to interpret that they have a value, and if their value is strong enough, that they have a corresponding moral. But if Bob had merely been speaking to themselves, then there'd be no values involved, because none could be interpreted. There would only be Bob's attitudes.

The theoretical implication is that the last man on Earth would have no morals, values, or obligations except insofar as he thinks he is serving someone. The empirical expectation might be that people, when confronted with this philosophical argument, would agree with it.

Personally, though, I doubt they would.

[uriah]

Thanks for the thoughtful responses, Joshua. Continuing a bit:

On your 4, which is in response to my first point. You raise good questions. But however we try to home in on standards of competence, they must be different from generalizations about performance. How to produce standards of competence is a tricky point, but my claim is that it cannot be a purely a matter of generalizing over collected data. Consider the study of grammar. Martians land in the US and try to figure out the grammar of English. If they just record how people talk, and produce generalizations on the basis of the data they collected, they will get a descriptive theory of *grammar performance*, but they still wouldn't know what *grammar competence* is, and therefore what English grammar is. There are probably things they might do to try and get out of this predicament - perhaps ask the grammar experts, or study the performance of good grammar users such as lawyers, philosophers, and unimaginative novelists. But in any case they'll have to make allowances for a gap between performance and competence and realize that their data, unadulterated, reflect only on the former. And when the Martians try next to figure out what earthlings mean by “valuing,” they'll have to make the same allowances. They'll have to realize that by running experiments on people's willingness to apply the concept of valuing to other people, they are only gathering data on performance, and something extra would have to be done to get an accurate picture of competence. By analogy to what they did with grammar, they might ask the experts on the concept of valuing, or study the best appliers of the concept of valuing. The experts, I contend, are the conceptual analysts. As for the best users, they would have to be people who have common dealings with attribution of valuing and for whom quite a bit hangs on getting the application of the concept right. I'm not sure who fits this description, but it might be interesting to run the experiment with them as subjects. I predict you'll get the result that the normative effect is significantly smaller than in the population at large.

(The case of grammar is tricky, because there is some subtle interplay between performance and competence: standards of competence are to some extent “leashed” to widespread performance (in the sense that they have freedom from actual performance, but not unlimited freedom). There are other areas where the standards of competence aren't so leashed. Again, Tverky and Kahneman's work on inductive reasoning is relevant. The standards of competence they presupposed were taken straight out of logic textbooks. In this case, the standards are outside the experimental realm altogether. That said, it is more plausible that the case of valuing is more similar to that of grammar than to that of inductive reasoning.)

On your 1 and 3, which are in response to the Charity hypothesis. You make very interesting, and on the face of it plausible, claims here, and I'll have to think about them some more. But note, very quickly, that although Davidson thought the principle of Charity applied universally (and indeed transcendentally), someone else might say that Charity kicks in for some parts of our folk moral practice and not others. For example it might apply to the concept of valuing and not to that of intentional action… Of course, there would have to be an explanation of why that would be.

[tnadelhoffer]

Chris--of Mixing Memory fame--has an interesting post on his blog about Knobe and Roedder's paper. It's worth checking out. You can find it at:

http://mixingmemory.blogspot.com/2006/05/experimental-philosophy-morality.html#links

[Yentz Mahogany]

Uriah,

Your Martians could skip asking the experts and just ask ordinary person (x) about whether or not they think that language user (y) would be considered a competent speaker by the generalized populace (z). (Call this third-order sense of language the SP-language.) Then examine the statistically normal opinion on a given sample of the SP-language. This ranges over auditory-speaking performance, but through the filter of receiving-understanding (another kind of performance).

[Antti]

A couple of points in response to Josh's. I'm on the road, so this will be really brief.

First, in terms of the dialectic, if charity works elsewhere, the burden of proof is on Josh and Erica to show it doesn't work here, no? I think, obviously, that I've made a good case for it. Second, charity with respect to values is most directly relevant to attributions of valuing and attitudes that clearly involve valuing. Neil's cases are really about what's normal and what's not, so I can't see them as counterexamples. (The race case just seems to depend on the current practice of classifying 'races' - 'whites' have only white ancestors, while 'African-Americans' can have ancestors of any kind.) Point 3: charity requires by default maximizing agreement with respect to the interpretee's rationality, values, and beliefs, in that order. If there's a reason to believe the subject has a different starting point, the process will be more complex. Anyway, in the pollution case, maybe the rationality consideration takes priority over the morality one. That is, it is harder to make sense of the chairman bringing about bad effects unintentionally than him bringing about good effects unintentionally, because he'd be irrational in the former case. (This is very condensed; I'll have to think about this.)

[Joshua Knobe]

Chris,

This is truly an extraordinary post -- more the sort of thing one would expect to find in a journal article than the sort one expects from a blog. I would love to hear more about how you think these matters could be put to the test.

First, you suggest that the effect we found here could also be obtained in cases that were not specifically *moral*. But suppose that we actually did find the effect in non-moral contexts. We would then have to determine whether (1) the effect was simply due to a performance error [such as my-side bias] or (2) a more general notion of goodness was playing a role in the concept of valuing itself. Do you have any thoughts about how one might be able to adjudicate between these hypotheses?

Second, you suggest that the effect for intentional action might be due to our greater inclination to 'undo' bad events in counterfactual reasoning. Do you have any ideas about how to test this hypothesis against the hypothesis that moral considerations actually play a role in the concept of intentional action itself?

[Chris's post can be found at http://mixingmemory.blogspot.com/2006/05/experimental-philosophy-morality.html]

Uriah,

Thanks; this really helps to clarify the key ideas from your earlier comment.

It is common practice to distinguish between the fundamental criteria underlying people's judgments (their competence) and the various systems they use in determining whether or not a given case meets these criteria (their performance systems). When we observe people having a particular pattern of intuitions, it is always an open question whether this pattern is a reflection of their underlying competence or whether it simply reflects a problem in their performance systems. But, of course, this is simply an empirical question, and it can be addressed through systematic experiments.

What you seem to have in mind is something a bit different, a standard that lies entirely outside the minds of individual subjects. But I wonder if the notion you are after might also be amenable to empirical inquiry. You suggest that conceptual analysis can proceed by looking at the intuitions of people who are in an especially good position to understand the relevant concepts. So would you be more open to experimental approaches if the subjects were not just arbitrary people off the street but people who had some special claim to be good judges in the relevant domain?

Antti,

Some people try to explain the results for intentional action by positing a general mechanism that leads people to attribute mental states with morally *bad* contents, while others try to explain the results for valuing by positing a general mechanism that leads people to attribute mental states with morally *good* contents. But it seems unlikely that both of these explanations could be correct. So the overall pattern of results appears to lend some support to the hypothesis that at least one of the effects is due to something in the nature of our particular concepts. Does that sound right to you?

[Jonathan Weinberg]

I must confess that I'm still not seeing what's supposed to be problematic with the charity/"my side" bias explanation of this data. Plausibly, willingness to attribute a value to someone is correllated with willingness to attribute a corresponding desire to them; and a charity/"my side" bias will incline one to attribute desires (including hidden ones) to people consonant with one's own desires.

Let me also offer, fwiw, a "ditto" to Joshua's response to Uriah re: competence/performance. However, there seems to me to be some deep problems with trying to use the c/p distinction to adjudicate what is or isn't part of a _concept_. The c/p distinction applies most naturally to capacities or processes; but that doesn't seem to be the picture of concepts that y'all are working with.

[Gilbert Harman]

A possibly minor point (but maybe not) is that it is a mistake to identify a person's values (noun) with what the person values (verb). I might value your advice, but your advice isn't one of my values. I value listening to music, but listening to music isn't one of my values. One of my values is that people should be honest, but I don't value that people should be honest, I value their being honest.

[Joshua Knobe]

These comments have been really helpful. We now have a number of different theories on the table, and it will be interesting to think how to decide between them.

Here are a few quick thoughts:

1. The prototype theory does seem to follow in a fairly plausible way from certain generally accepted views about prototype concepts. After all, it is widely assumed that prototypes encode the most typical features of instances of a category, and since the objects of people's values are typically morally good, there is every reason to expect that this feature would be encoded in the prototype. It seems, moreover, that the actual pattern of data is exactly what one would expect if moral goodness were one feature of a prototype concept. Just as in so many other cases, one finds that a particular feature is not a necessary condition but that it does make a difference in cases where a given object has some of the relevant features but lacks others.

2. In the discussion thus far, no one has offered any reason to doubt this particular hypothesis. Instead, people have suggested that it might be possible to explain all of the data using an interesting alternative hypothesis. This alternative hypothesis says that moral considerations don't play any role in the underlying concept itself. Rather, moral considerations lead us to attribute additional mental states, including some mental states which actually are represented in the concept of valuing.

3. If someone makes a claim about *which* specific mental states people are attributing here, it will be very straightforward to test it. For example, suppose someone suggests that the effect is due to an attribution of unconscious beliefs. This hypothesis would be easily testable. We could simply add in sentences to both conditions that explicitly attribute unconscious beliefs to the agent. (That is, we would say in one condition that the agent has an unconscious belief in the importance of racial equality and in the other that the agent has an unconscious belief in the importance of racial discrimination.) Then, if we no longer found any difference between conditions, we would know that the original effect was due to attributions of unconscious beliefs. But if we did still find a difference, we would have reason to suspect that something else was afoot.

4. By contrast, if someone just says that people are attributing different mental states in the different conditions but doesn't say specifically what those mental states are, it's hard for me to see just how we could go about deciding between this hypothesis and the prototype hypothesis we offered in the paper. But the fact that I have trouble thinking of any way to decide between these two hypotheses doesn't mean that it isn't possible. I'd love to hear any suggestions about how one might go about arranging an experiment (even a thought experiment) that enables us to figure out which of these hypotheses is the better one.

[Antti Kauppinen]

Josh, having both feet on the ground again, here's a few comments on your latest points. (I'll get back to the earlier ones later.)

First, it is true that nobody has emphasized the problems with the moralized concept hypothesis, though the point has been made that theoretical parsimony would favour an explanation in terms of general mechanisms that we have to acknowledge anyway, if we have one. I think, in fact, that that's enough. But that's not to say it's the only reason to reject the thesis I termed MCV (the claim that some of the criteria of application of the concept of valuing are moral) in my commentary. I found it difficult to formulate just what the truth conditions of claims about valuing are according to you, and I'm not sure if I got it right. There's two obvious readings, call them subjective and objective. I think you mean the first. The subjective version would be something like this:

(SMTC) 'A values o' is true-for-B if enough of A's psychological features with respect to o are included in the prototype for valuing and B values o.

The objective one, in turn, would be something like this:

(OMTC) 'A values o' is true if enough of A's psychological features with respect to o are included in the prototype for valuing and o is in fact valuable.

Now, I understand OMTC; the only problem with it is that it's false. SMTC, in contrast, amounts to a peculiar subjectivism that I'm not sure if I even understand, especially given the oddity of 'true-for-B'. I don't think the folk are so confused. That would mean, among other things, that if I'm wondering whether Bess values clean air more than cheap gas, I would not have to focus on what I know about Bess but also on whether I find clean air more valuable than cheap gas. It would mean that if Jodie disagrees with me about the relative value of clean air and cheap gas, she neither agrees nor disagrees with me about Bess's values when she says "Bess doesn't value clean air more than cheap gas". Surely these consequences are most disagreeable, and thus positive reasons to reject the moralized concept hypothesis, or at least prefer an explanation of the data that doesn't attribute to the folk a view with such consequences. It goes without saying that the same problems arise for moralized versions of any other psychological concept.

Second, though I didn't want to harp on methodological points, I don't think the testability of alternative hypotheses is trivial either. Part of what I wanted to highlight with my discussion of charity and holism is that people read more into stories than they are explicitly told. This isn't something that could be eliminated; you can't put in a clause to the effect that "these are all the psychological states that George has." Rather, it's a deep fact about interpretation of texts and of people. The longer the vignette, the more possibilities you can close off, but also the more difficult it is to isolate just what aspect of the story people are responding to, to make certain they are bearing in mind all the considerations mentioned, and so on. And at the end of it, you'll get gut reactions that arguably don't tell us much about the folk's concept, though it may be of psychological (rather than philosophical) interest what processes lead to them. So it doesn't look all that simple to me - which isn't to say I wouldn't be interested in finding out how people would respond to the additional questions I suggest in my commentary!

Finally, as I point out in a footnote, I agree with Harman that we should be on guard against the verb/noun confusion with respect to 'values'. (The confusion renders Malle's study pretty much irrelevant.) In technical contexts, I would much prefer moral judgment, or ought-judgment or evaluative judgment if the question is more specific. In ordinary language, it seems the most natural terms for these would be finding or thinking something good or bad, or accepting something as one's duty, or something like that - I'm not a native speaker, obviously. Would people's gut reactions be the same if they were asked whether they agree with "George finds racial equality to be a good thing"? (I doubt if there would be a major difference, but maybe some.) How would "The story doesn't give enough information to decide whether George thinks racial equality is good or not" compete?

[Erica Roedder]

Hi all,

Thanks so much for the thoughtful comments. Regarding charity, Josh has already posted regarding a number of the problems/challenges we see for the thesis, but he didn't really mention one of the things we talked about--so I'll post it here instead.

It's important to realize that strictly speaking the hypothesis of charity *need not* be in tension with the hypothesis we give in the paper. At the same time, it *is* in tension with something suggested by the paper. Let me explain.

First, what is charity? We know what it is in Davidson's terms, but how do we understand it as a causal-psychological claim about the mind? For my money, charity is best understood as a high-level generalization providing an abstract characterization of a large set of inferences that individuals make. Across many discrete concepts, subjects are inclined to draw conclusions that fit (for our purposes) the following description: if X is good, then other people will have certain mental states (e.g., wanting, valuing, believing to be good) directed at X.

The important thing here is that this is a HIGH LEVEL generalization. Understood this way, the hypothesis we make in the paper—that being good is part of the prototype of valuing, and that this explains why subjects are inclined to move from a judgment that X is good to the judgment that the character values X— will indeed count as charity! This because the subject's inferences fit the pattern of inference outlined above, and that's all we mean when we explain a subject's response by appealing to charity. (Compare: Describing a substance as soluble v. describing the way that this substance breaks down in water.)

So far, I've pointed to how charity is compatible with our hypothesis. However, it's also importantly in tension with something suggested by the paper. As Josh pointed out above, we think the mechanism that "carries out" this inference, in the case of valuing, is rather unique. In the paper, our hypothesis is that the subject makes (charitable) attributions because 'goodness' is part of the prototype of valuing. So the mechanism which carries out the charitable pattern of inference is prototype-centered. On the other hand, we suspect (and this is only a suspicion) that for many other concepts in the philosophy of mind, the mechanism that "carries out" charity is *not* prototype-centered. For example, I suspect that 'goodness' is not part of the prototype of the concept of wanting--even though we might be biased, via *other* mechanisms, to be charitable when using this concept--i.e., we have a background belief that people want what's good.

In brief, I've suggested an understanding of charity as a high level generalization. On this understanding, our hypothesis can be understood as characterizing the mechanism which instantiates a type of charitable inference. In this sense, our hypothesis is *compatible* with charity. However, to simply attribute our results to charity, without saying more about the specific mechanisms involved, will blur the line between the (importantly) different ways charity is instantiated, and it is in tension with our belief that there's something *special* about the mechanism involved with the concept of valuing.

I hope this post helps sort out some of the ways that our hypothesis is compatible and incompatible with charity. Thanks again for the great response to our paper!