In a very nicely constructed recent paper (“Are ethical judgments intrinsically motivational? Lessons from ‘acquired sociopathy'”, Philosophical Psychology 16 (2003),) Adina Roskies offers a compelling argument against the position she dubs “motive internalism” (others have called it “belief internalism” or “judgment internalism”). Motive internalism states:
If an agent believes that it is right to A in circumstances C, then he is motivated to A in C.
Roskies’ strategy is simple enough: She points to examples of agents for whom motive internalism is false.
Roskies discusses findings from studies of those with injuries to the ventromedial (VM) prefrontal cortex, an area of the brain widely believed to influence perception, reasoning, and emotional control. These patients perform normally on most all cognitive metrics, but have difficulty conforming to social expectations, and in particular, they seem not to act in accordance with their own sincerely held moral judgments. Roskies notes that there is no evidence of a defect of reasoning or perception, nor do patients with VM injuries lose their comprehension of moral concepts. Those with VM damage report low emotional response and affect in those contexts that ordinarily produce characteristically moral emotions (guilt, etc.), and physiologically, the VM patients do not produce the skin-conductance response (SCR) thought to correlate with the emotional or affective responses associated with moral judgments. Nevertheless, the VM patients sincerely affirm commonplace moral judgments. Roskies therefore proposes that VM patients fail to be motivated by their own moral beliefs (not even to the minimal degree that those beliefs are overridden by other motives such that guilt, regret or other reactive attidues occur when they act contrary to those beliefs). VM patients, she concludes, are counterexamples to motive internalism.
Roskies’ paper is careful, and she addresses several objections that defenders of motive internalism might make to her thesis. But I’d like to consider an objection Roskies does not mention: Simply enough, the VM patients are not counterexamples to motive internalism because they do not actually possess the moral beliefs that (allegedly) fail to motivate them.
Now, I am far from an expert on belief attribution and the array of semantic, psychological, and epistemological issues it raises. But to my understanding, a wide spectrum of theories about the nature of belief would endorse the claim that an agent does not believe that P unless she behaves as a person who believes that P characteristically behaves, where the behavior in question is not only linguistic. Such a claim would be endorsed by ‘dispositionalist’ or ‘functionalist’ theories of belief, and need not contradict representationalist theories of belief either. Intuitively, it seems right to say that, e.g., a person who believes that water is deadly poison must behave as we would expect a person who believes that water is deadly poison would behave, to wit, she would probably suffer frequent dehydration and have a hysterical fear of rainstorms. (Caution is needed here: How such a person would behave is determined by a good deal else besides the belief in question, including desires and her other beliefs. But it seems at least in principle possible to isolate the role that a given belief plays in a person’s overall pattern of behavior.)
So in the case of the patients with VM damage, cannot a plausible case be made that they do not have the relevant moral beliefs, as evidenced by the fact that those beliefs do not motivate them? Could there be better evidence of their not believing these moral statements than the very fact that they do not behave (nor do they seem to have the causal psychological and physiological antecedents of those who do behave) in accordance with their supposed moral beliefs?
Roskies argues that defenders of Motive Internalism face a dilemma: very roughly, either MI is false or it’s trivial. I suspect she would say that your suggestion is impaled on the second horn. If the very fact that the agent lacks motivation is taken as evidence that he lacks the relevant belief, then MI is unfalsifiable, hence trivial. At any rate, that seems to be Roskies position.
Campbell,
Thanks for the remark. But I don’t see that my suggestion succumbs to the ‘triviality’ horn. Roskies goes through several possible versions of MI, some of which she thinks are trivial, but the version I gave:
If an agent believes that it is right to A in circumstances C, then he is motivated to A in C.
is the one she deems “substantive” rather than trivial.
And it seems to me fair game to ask what empirical evidence counts in favor of a person’s beleiving something, since Roskies helps herself to plenty of evidence regarding whether a belief motivates an agent to act upon it (without, so far as I can tell, taking up the prior question of what evidence suffices for attributing belief in the first place).
And I don’t see how the claim
“Being motivated to act as if a belief is true is prima facie evidence that a person has the belief”
fails a falsification test any more than any other claim relating effects as evidence of their causes. I’m not suggesting, as a behaviorist might, that the behavior is the belief. Nor am I making a conceptual claim regarding belief and motivation. It seems like a genuninely empirical and falsifiable claim that is not in fact falsified. Presumably, the mental state of belief is causally efficacious somehow, and if it’s not efficacious with respect to motives or behaviors, then isn’t this good reason not to attribute belief in those instances?
”
Hi,
As Roskies herself notes, the defenders of internalism often claim that their cherished thesis is a priori. So they might well admit that as such it is not empirically falsifiable any more than a true thesis in arithmetic is empirically falsifiable. So far, so good.
However, defenders of internalism would therefore not take this to show that their thesis is trivial. On the contrary, they could and should insist, many a priori truths are very interesting and far from obvious. Godel’s incompleteness theorem springs to mind, for example. On this view of the issues, we can have a debate about whether internalism is true, and that debate might well be interesting. But it will be a debate which employs whatever methodology is appropriate to a priori questions.
Roskies takes herself to be arguing against those who think the issue is well cast entirely in a priori terms, but if she were to accuse them of holding a trivial view simply because its not open to empirical falsification then she seems to assume what she sets out to show.
This is not to say that the cases she discusses are not philosophically interesting. The point instead is that some of the issues they raise are a priori. In particular, are the empirical facts about these subjects upon which we can all agree (that they lack certain motivational states) sufficient already to show that they do not, after all, have the relevant moral beliefs?
To her credit, Roskies does address this question, and argues that given that the subjects in these cases seem sincere, we should conclude that they do indeed believe what they say unless they no longer understand the meanings of moral terms. She then argues (persuasively to my mind) that the latter is unlikely.
However, there is another possibility. The subjects might understand the meanings of moral terms and be sincere in that they believe that they have the relevant moral beliefs, but wrong about their own states of mind. That is, one of them might e.g. believe that he believes stealing is wrong even though he doesn’t really believe it. This might betray a lack of transparency about the subject’s own states of mind rather than a failure to understand moral concepts or a lack of sincerity.
They key point is that if I say that ‘p’ but do not believe that p, I can still count as sincere so long as I at least genuinely believe that I believe that p. In that case, I do not intend to deceive my interlocutors and seem guilty of delusion rather than insincerity. Getting back to the issues raised by Cambell’s reply, this view of the nature of sincerity is itself arguably a priori. So resolving the correct interpretation of the cases may require a priori theorizing which is not amenable to empirical falsification.
Best,
Mike
I am a bit sceptical about whether mere dispositionalism or functionalism can provide a counter-argument against Roskies’ argument for judgment-internalism. This is because that view about belief ascriptions seems to fit her cases very neatly.
Think of one of the cases before the injury. In that case, we, in dispositional or functionalist spirit, can ascribe a belief that ‘such-and-such is right’ for the agent given the evidence of her being disposed to act in certain ways in certain circumstances. Now, after the injury, the agent still says that she has the same belief. However, we can no longer observe the disposition to act in certain ways from the agent’s actions. Would we then need to say that the agent has lost the relevant disposition and thus the belief? I do not see why we should. All we would need to say is that whatever mental state played the causal role of producing action previously is unable to do so now due to the disabling condition of the injury. That injury blocks the causal route running from the action producing moral belief to the action before the agent desires to act. If the injury could be repared we would expect that normal action and motivation to act would return, and this is because the disposition/moral belief remained the same throughout the injury.
The externalist can thus argue that the functional ‘black box’ or disposition to be motivated to act in certain ways is unaffected by the injury even though it is thereafter unable to affect anything. Thus, the agent could have the same belief without motivation.
I am trying to think of a good analogy. All I can see on my desk is a table lamp. Its’ function is to light up my desk and it is disposed to do so when I switch it on. However, its lamp may sometimes unscrew slightly and thus it won’t thereafter light up. It still has the same functional role and disposition though which it is then unable to carry out succesfully.
For the record though, I would like to defend some sort of weak judgment-internalism…
Michael:
Funny – I just wrote about this on Desert Landscape (based on a talk, not the paper, though):
http://www.arizonaphilosophy.com/?p=125
At the talk, Jesse Prinz offered basically the same objection as you, claiming that the VM patients don’t have the same moral concepts as we do.
As an internalist, my favorite response (as I wrote in my post) is that there’s a version of internalism that is unthreatened by the argument. This version is formulated as follows: necessarily, normally, if x is a moral judgments then x is motivating. The difference is the introduction of the normality condition. This version, or something close to it, was suggested by Jamie Dreier in a 1990 Ethics paper. It’s used to accommodate the moral judgments of the amoralist and the wicked. But it applies here too, because the moral judgments of the brain-damaged subjects are not normal in the relevant sense.
Hi,
I like the idea behind Jussi Suikkanen’s reply. The basic idea seems compatible with a fairly robust version of internalism, though.
For example, it could be that our moral judgements are at least partly constituted by plans (on Gibbard’s view, they are entirely constituted by such). Now suppose Jussi’s story works, in the sense that these planning states no longer have their normal causal powers due to the patients’ unfortunate condition in these cases. Perhaps the condition prevents background general plans from translating to proximate intentions when the time of action arrives, e.g.
Plausibly, even though it does not cause the agent to act (it is blocked), the relevant planning state should be classified as a motivational state (it has the ‘direction of fit’ of desire in some suitable way of cashing out that slippery metaphor). So it looks like moral judgment still essentially involves a motivational state.
Or at least, this line of argument seems not to undermine a version of internalism which allows for such “causal blockers.” We’d still need some independent reason to believe moral judgments are partly constituted by such planning states (or whatever), though, but that is just to say we need an argument for internalism.
[I realize this line of argument is incompatible with the original ‘they don’t have moral beliefs line’ I seemed to defend in my reply to Campbell. But I didn’t really mean to commit myself to that reply so much as suggest that its not so vulnerable to the triviality objection, at least]
– Mike
A quick thought about Uriah’s proposal that we take “necessarily, normally, if x is a moral judgments then x is motivating.” to be the appropriate internalist thesis.
My worry is this: suppose you grant that the brain damaged, the amoral, and the wicked are counter-examples to the stronger internalist principle: they have moral beliefs, but these beliefs don’t motivate. But you still want to insist on this weaker principle. Go to a world in which all the people there are brain damaged, amoral, or wicked; there the people make moral judgments, but it isn’t normal that the judgments motivate. (None of them do in that world.)
So on one understanding of the propoposal — one in which we take “normal” to mean something like “typically” or “on average” — it seems clearly false.
Is there another way of understanding the principle that is both plausible and not subject to this sort of counter-example?
I suppose one natural move to make in response to Kris’ worry, which I think is open to Uriah and Jamie, is to work with a normative conception of ‘normal’. I thought often internalism was formulated in terms of what would move an agent insofar as she’s rational. Is this kind of internalism thought to be uninteresting because it has too many exceptions built into it in advance? I thought that this was how Smith and Korsgaard understood the view. Why can’t the internalist say that these subjects are not even potential counterexamples due to the fact that these subjects suffer from a kind of incapacitation?
Kris:
That’s a nice problem you raise. It shows that for the normality version of internalism to be substantive and interesting, normality must not be construed in statistical terms.
I think Clayton is right to suggest a normative construal of normality. One natural way to make this work out is in terms of propert function. A version of internalism along these lines is developed by one of our grad students at Arizona, Matt Bedke. The resulting thesis is something like “If x is a moral jusdgment, then it is the proper function of x to motivate.”
In any case, the result we must obtain is that in the world in which everybody is amoral or brain-damaged, everybody is abnormal. Any construal of normality that will yield this result will be workable for the kind of response I have in mind.
“Go to a world in which all the people there are brain damaged, amoral, or wicked; there the people make moral judgments, but it isn’t normal that the judgments motivate. (None of them do in that world.)”
Jamie’s origianal understanding of the “normally” was not statistical, but I’m not sure it was normative. And I think it handles these sorts of cases. His idea was that one can have moral beliefs if one stood in the right relation to some normal cases in which people expressed their moral beliefs using the same terms you do, and when those people are normally motivated by their moral judgements. Thus the “Sadists” (an invention of Gideon Rosen’s, who act so as to promote the bad) will get their terms translated with our orthographically identical terms because their practice developed from a practice by the “normal” folks who were motivated.
In the examples that started the thread, the brain damaged can be their own normal people. But on a planet where everyone always had that sort of motivational defect, their terms would not mean what ours do. Or so I think the example would go.
I suggest that this form of internalism can be thought of as interestingly related to some of Burge’s suggestions about getting to use a word with a certain meaning by being in a community of speakers some of whom know enough to use it correctly enough. Other members of the same community can get credited with the same thoughts, even when their usage does not respect the relevant constitutive truths about the referents of their terms.
Hey, Mike. Good to have you on board!
Let me make one small clarification on my previous comment. I didn’t mean to claim that all unfalsifiable statements are trivial. Rather, I meant only to attribute that view to Roskies. (I probably didn’t make that clear enough.) That Roskies holds this view is suggested by passages such as the following:
PEA Soup Contributor Lesson #1: Don’t post something provocative, go on vacation, and then expect to keep up with everything that follows! So thanks to everyone for these remarks. I’ll do my best to do justice to them all.
Mike (and indirectly, Campbell)- Your remarks lead me to conclude that internalism is an a priori thesis, derived from the nature of belief (i.e., about what belief does within our psychological economy). But I’m willing to say that an a priori thesis can be a rejected on a posteriori grounds (as Roskies does) and is therefore open to potential falsification. And I would agree with your implicit claim that assessing whether A counts as evidence for B can require a priori theorizing about what B is.
Uriah, Kris, and Clayton (at least) – As Campbell hints at in his last comments, Roskies dismisses as ‘trivial’ any form of internalism less robust than the one she labels substantive (though I think the real criticism is that these are unfalsifiable rather than trivial in the sense of being tautological). The lesser forms would include those appealing to normality (whether statistical or normative), rationality, etc. I’m inclined to agree with her in that appeals of these sort then rest internalism not on the intrinsic motivational powers of belief, but on some futher presumably psychological factor (that may well operate in conjunction with belief to produce motivation).
Jussi- Your post captures the heart of my worry really.
“Think of one of the cases before the injury. In that case, we, in dispositional or functionalist spirit, can ascribe a belief that ‘such-and-such is right’ for the agent given the evidence of her being disposed to act in certain ways in certain circumstances. Now, after the injury, the agent still says that she has the same belief. However, we can no longer observe the disposition to act in certain ways from the agent’s actions. Would we then need to say that the agent has lost the relevant disposition and thus the belief? I do not see why we should. All we would need to say is that whatever mental state played the causal role of producing action previously is unable to do so now due to the disabling condition of the injury.”
But if the behavior and affect were present before (and absent after) the injury, and all that’s present afterward is the linguistic affirmation of belief, should we be so ready to ascribe the belief after the injury? As the saying goes, talk is cheap, and I would rank subjects’ verbal behavior as weaker evidence than either affect or actions. Of course, we would want an explanation of why subjects affirm what they do not believe (self-deception, inadequate attention to their own beliefs, a weak understanding of ‘believe’, perhaps). But if I’m asked (for example) what’s the better evidence about what people believe about some moral issue (e.g., the obligatoriness of acting to save others from preventable death or suffering) — their linguistic affirmations or their actions and emotions — I’d side with the latter.
Mike:
When I made my suggestion to Adina after her talk, she responded in the way you indicate, saying that interesting internalism is a necessity claim, not a normality claim. But I think we can usefully distinguish two versions of internalism:
(A) Normally, moral judgments motivate.
(B) Necessarily normally, moral judgments motivate.
I think the operator “necessarily normally” is an important one, and makes for genuine internal relations between the things it links. If so, (B) is a substantive internalism.
How to make sense of “necessarily normally” is not obvious, but I think it’s the kind of operator you get in Davidsonian claims about “the very idea,” for example. I’ve always thought that many philosophical claims are at bottom, in their plausible version, claims of necessary normality.
When I suggested this to Adina, she said she would have to think about it. She seemed to really mean it!
Michael,
you are right – talk is cheap. And, part of me wants to agree with you. But, then there is part of me that wants to say that we should attribute the relevant moral belief especially if we are functionalists or dispositionalists.
Many of the latter have also been materialists. So, they would say that the belief as a mental state just *is* the small bit of brain in certain state (whichever it happens to be) that plays the right causal role (perhaps inbetween observations about the naturalistic world and desires for action). If this is true, then the question of how we should describe Roskies’ cases I guess becomes empirical. Is it this bit of brain that is damaged in the injuries or some other that blocks the causal path in front of this state? Whatever the answer here is it is hard to see why we should necessarily say that it is the former.
Hi,
Campbell: Fair point, I see that you didn’t mean to endorse the triviality worry. I hadn’t realized how seriously to take the ‘At any rate…’ sentence at the end of your original post. People sometimes use sentences like that rather loosely to indicate fallibility rather than to put what came before inside a ‘So-and-so would say’ operator. I now think we agree.
Michael:
People working on the a priori sometimes distinguish positive dependence from negative dependence. Your point seems to be that internalism is not positively dependent on empirical evidence but it is negatively dependent on it. Fair point. But then the real issue becomes whether we can have a priori grounds for how best to interpret any given empirical case, and here we seem to agree.
With regard to the ‘Necessarily normally’ stuff, Mark Lance and Maggie Little have some very interesting things to say about this (the relevant papers are posted on Mark’s website) under the heading of ‘defeasible generalizations’. I won’t try to summarize them here as I won’t do justice to them without going on at some length. Suffice it to say that there idea is neither merely statistical nor is it the Davidsonian one.