How Wrong Can You Be?

How badly can you go wrong at one thing while succeeding at another? This paper pursues a version of this question that is provoked by Chapter 3 of Williams’s The Metaphysics of Representation: how badly can you go wrong at understanding why ordinary things in the world behave as they do, while succeeding at locking onto those things as objects of thought? Williams’s discussion entails that successes along these two dimensions go hand in hand. It entails that to count as thinking about ordinary things, a subject must also count as being sufficiently right about why they behave as they do. I think this is a wrong result. The paper is structured as follows. §1 summarizes the relevant part of Williams’s Chapter 3. §2 explains how this part of Williams’s proposal generates the result I have indicated, and why I think this result is bad. §3 considers some possible lines of reply. §4 attempts a more general diagnosis. Chapter 3 of The Metaphysics of Representation offers (among other things) a solution to an old puzzle for people theorizing about cognitive representation of ordinary things. Here is a toy version of this puzzle which will suffice for present purposes.1 Take a mundane situation in which we would like to be able to say that a subject is having a thought that attributes a commonplace property to an ordinary object. For example, suppose that subject S is looking at a green ball in a situation involving no causal or perceptual funny business, and thinking a thought she would express by saying ‘That is green’. We would like – from the theorist’s point of view – to make claims like this: S’s singular concept picks out that ball (the one she is currently pointing at). S’s general concept picks out the property of being green. S’s belief is true iff that ball is green. Now consider a permutation of this initial (partial) interpretation of S (Williams calls this a ‘twisted’ interpretation): i*) S’s singular concept picks out o, where (o is an object which exists iff the ball does, and which stands in relation R to the ball). ii*) S’s general concept picks out the property of being F, where a thing that stands in R to the ball is F iff the ball itself is green. iii*) S’s belief is true iff o is F. And now consider what reason there might be to prefer the that ball is green interpretation of S’s belief to the o is F interpretation. The interpretations bring the beliefs in as true relative to exactly the same possible ways the world might be: the world is such that the ball is green iff it is such that o is F. So if we want to keep (i)–(iii) over (i*)–(iii*), we have a problem: to find reasons for the preference, given that the interpretations demand the same things of the world if the belief is to be true. Williams’s solution to this puzzle provides a worked example of his recipe for generating assignments of representational contents. The recipe is built around Williams’s Substantive Radical Interpretation constraint on such assignments: Substantive Radical Interpretation (hereafter ‘SRI’): The correct interpretation is the interpretation which best rationalizes the subject’s dispositions to act in the light of her experiences (where ‘rationalize’ is to be read ‘substantively’, so that the correct interpretation comes out as the one that maximizes the extent to which the subject is responsive to genuine reasons) (26). SRI motivates the following four-step recipe for the assignment of representational contents to concepts (construed as vehicles of representation).2 Firstly, isolate a core of behavioural dispositions that may be regarded as constitutive of operation with the concept (in Williams’s terms, these dispositions constitute the concept’s ‘conceptual role’). Secondly, uncover constraints on substantive rationality for this general class of concept-deployment (in Williams’s terms, these are ‘first order’ constraints on what it takes for a subject’s deployments of this kind of concept to be in normative good order). Thirdly, argue that, given these constraints, the interpretation that best rationalizes the core behaviours will treat the concept as representing … X. Finally, conclude that, by SRI, the correct interpretation will treat the concept as representing X. Williams joins many others in taking it that, for the concepts associated with observational predicates like ‘green’ and ‘in motion’, the core behavioural dispositions include dispositions to treat the concept as applicable in response to an appropriate kind of perceptual experience (59). He also takes it that, for us, the ‘core’, conceptual-role-constituting behavioural dispositions for observational property concepts include dispositions to formulate explanations, and to use these explanations to arrive at beliefs that outrun what we observe (62, 64). This is the first stage of the SRI-based story for the concepts associated with observational predicates. At the second stage, Williams derives two constraints on rationalizing interpretation for the concepts associated with these behavioural dispositions3: Constraint 1: The rationality-maximizing interpretation will maximize the goodness of the subject’s explanations (the extent to which connections she treats as explanatory actually do explain the phenomena she observes). Constraint 2: The interpretation that maximizes the goodness of the subject’s explanations will maximize the extent to which her observational property concepts are assigned what are in fact potent properties (properties in virtue of the possession of which things behave as they do). Given these materials, Williams’s solution to the permutation puzzle unfolds as follows (see p. 64): 1 The correct interpretation of subject S is the one that best rationalizes her dispositions to apply observational concepts in response to experience; deploy these concepts in explanation; and use these explanations to arrive at further beliefs. [SRI applied to the case of our observational property concepts.] 2 The interpretation that best rationalizes the dispositions at 1 maximizes the goodness of the subject’s observational-property-deploying explanations. [Constraint 1] 3 The interpretation that maximizes the goodness of these explanations maximizes the extent to which observational property concepts are assigned what are in fact potent properties. [Constraint 2] So 4 The correct interpretation maximizes the extent to which S’s observational property concepts are assigned potent properties. [From 1–3] Now add a pair of metaphysical assumptions: 5 Being green is (as a matter of metaphysical fact) a potent property; being F is not 4 and 5 entail 6 The correct interpretation assigns being green to S’s concept rather than being F. So 7 In a case where the subject is disposed to believe in response to experience of a green ordinary thing, the correct interpretation treats this belief as a belief about the thing experienced. [From 6, given SRI] Note that, though the ‘twisted’ (i*)–(iii*) interpretation deviates in its assignments to both the subject’s singular concepts and her general concepts, the solution to the problem – the account of why the initial interpretation is to be preferred – is running off the ‘property’ side of things. The solution works by combining the property-side assumptions at 5 into the account of correctness of assignment of contents to observational property concepts at 4 to give the conclusion that S’s concept refers to being green rather than being F – this is 6. With the property-side work done, we can use the fact that S’s experience gives her at least some reason to ascribe this property to the initial interpretation’s value for , and no reason to ascribe this property to the twisted interpretation’s value for to generate the object-side result at 7. (Compare Williams’s way of putting the permutation challenge on p. 54: the twisted interpretation is generated by putting in two twists for each atomic thought, one on the interpretation of the singular concept and one on the interpretation of the general concept. Williams’s solution to the challenge gives us a reason to resist the twist to the interpretation of the general concept, and when we untwist the interpretation of the general concept, the singular concept untwists along with it.) The previous section explained Williams’s solution to the permutation puzzle. I shall now bring out an aspect of this solution that I think should at least give us pause for thought. Here is a scenario to get us started: Macroscopic Maxwell’s Demon Sally is an adult who has lived from birth in an environment where the course of macroscopic observable events is controlled by a hidden demon. The demon’s interventions are completely unsuspected by Sally, though their results are ubiquitous in her experience of the macroscopic world. To keep things simple, suppose that the kinds of events available for observation by Sally are severely restricted. Balls of various colours come in from the left. The demon hits them with its tennis racquet, sending them off in various directions at various speeds. Sally sees the balls moving around, but not the demon’s interventions. The ways the demon hits the balls so far have involved sufficient regularity that Sally has been led to form generalizations like ‘Red ones turn upwards’; ‘An orange one’s angle of deviation from its initial path depends on the time interval since the last green one’; and so on. She has refined these generalizations against her experience, and developed a theory which she uses to predict, and treats as explaining, the behaviours of balls of various colours. In fact, these behaviours depend on how the demon swings its racquet. And the demon swings its racquet according to its own whims: the colours of the balls play no role in determining the paths they will take. Consider Sally watching a red ball approach. , she thinks. And ask yourself this question: is Sally’s thought about the ball. I think the answer to this question is ‘Yes’, and would need quite a lot of convincing to the contrary. I conjecture that that’s probably your initial reaction too. It seems that Sally is thinking about the ball. Whatever has gone wrong for her has gone wrong with respect to her grasp of why things like the ball behave as they do, and this thing-that-has-gone-wrong has left her capacity to think thoughts about the ball intact. However, it is hard to see how Williams’s framework can deliver this result. To see why this is, we must first say a little more about Constraint 2 (Premiss 3 in Williams’s solution to the permutation puzzle). As it stands, Constraint 2 states a necessary condition connecting goodness of explanation and potent properties. But one apparent lesson of the Demon case is that whether potency of properties transmits to goodness of explanation depends not just on whether the explanation deploys property concepts which are concepts of potent properties, but whether the explanation fits these properties together in appropriate ways. If this apparent lesson is right, Constraint 2 must be modified accordingly: Constraint 2 (modified): The interpretation that maximizes the goodness of the subject’s explanations will maximize the extent to which the subject is treated as combining representations of what are in fact potent properties into explanatory beliefs that are appropriate to the ways these properties in fact determine the behaviour of objects that have them. Now, if Constraint 2 is modified in this way, we can see that Williams’s solution to the permutation puzzle at least does not deliver the verdict that I have suggested intuition demands in the Demon case. For in the Demon case, Sally’s explanations are wildly off relative to the actual causal potencies of the colour and motion properties. They are so wildly off that it is tempting to say that her deployments of observational concepts do not make it onto the ‘goodness of explanation’ scale. But if Sally’s deployments of observational concepts do not make it onto the goodness of explanation scale, Williams’s proposal gives us no reason to say that she is thinking about the ball and ascribing a colour to it, rather than ascribing a colour* to an object assigned to the ball under a permutation. And there is what looks like a reasonably good argument for a stronger claim. Williams’s story not only fails to deliver the verdict that Sally is thinking about the ball; it is in fact inconsistent with this verdict: 1 The correct interpretation treats Sally’s thought as about the ball. [Assumption] 2 The correct interpretation treats Sally’s thought as ascribing the property being red rather than the property F. (From 1 and the ‘two twists’ structure of the permutation challenge.) 3 An interpretation is correct only insofar as it maximizes the potency of the subject’s explanations (treats the subject as formulating explanations that in fact reflect potent relations between potent properties). 4 The correct interpretation treats Sally’s deployments of in a way that maximizes the potency of her because it is red explanations. [From 2, 3] But 5 Sally’s explanations are too wide ofthe mark to count as reflecting the potency of the properties with which she is in observational contact, so too wide of the mark to enable a genuinely potency-maximizing treatment of her because it is red explanations. Contradiction. So we have arrived at the bad result for Williams’s framework that I want to bring out. In this framework, a subject counts as thinking about the ordinary things with which her perceptual experiences put her in contact only if she counts as at least doing reasonably well at explaining why these things behave as they do. To avoid misunderstanding, I should stress that the challenge to Williams that I have raised in this section is not a sceptical challenge. The question I raised about Sally was not ‘How do we know we’re not in Sally’s situation?’ It was ‘Given her situation, should we say that Sally is thinking about the ball?’ I have suggested that the answer to this question should be ‘Yes’, and that, it least on the of things, Williams’s framework to saying are not to as is about to How might Williams while his solution to the permutation Here are with 1 in the argument from the previous that Sally’s thought is about the ball. This would the challenge as I have the Demon scenario is just one many such If this first is to of the scenario for good and the must be that no case with this structure is one where the subject counts as thinking about the objects she is And it is that not to To see us first that it is not a genuine that we are in Sally’s is sufficiently like a scenario that, as far as is it can be by is our preferred response to However, it is a observation that there have been in our when our situation has been in the following we have been (as it wildly as to why things or behave as they do. We thought that things are of that objects to the because each kind of thing has its that there is no at a that matter is that is one thing and If we this first response on Williams’s we have a between two with respect to who the behaviours of ordinary things in ways that we now know to be The first is to say that these people do from our point of as it onto the goodness of explanation scale. The second is to say that, from our point of the do not count as thinking about ordinary things. To the first we would need to argue that one or other of these is to be And the problem is not just about what it is for us to say about our It is also about what we are to say about Given the of in our of why ordinary objects behave as they do, it looks like of to to know that our explanations potencies in the properties of ordinary things. this first response is that if we make what looks like the right, – that we might just be wrong about why things behave as they do – we our right to to know that we are thinking about these things at 2, that an treatment of an treatment of If this is the Williams us an account of why the of interpretations works in one but not the The solution to the permutation puzzle in §2 takes it that when we untwist the interpretation of general concepts, the twist with respect to singular concepts is with If 2 is to be we need an account of why this does not also apply the other way Williams might 2 and the general concepts untwists the singular part of his solution to the permutation puzzle. But in that case, will us an account of why our ordinary singular concepts ordinary that 2 and 3 entail The would have to be that there are constraints on genuine explanations which are not by Sally’s deployments of If Sally is not in genuine explanations, 2 and 3 can be but 4 does not this will need an account of the between what Sally is doing and genuine attempts at explanation which does not the that our own deployments of observational concepts are genuinely might be that only some of our observational property concepts are by us in genuine that with respect to these concepts untwists the interpretation of ordinary singular and that the interpretation of ordinary itself untwists the interpretations for observational property concepts untwist that untwists the interpretation of and observational concepts come with no that our deployments of them are genuinely But of this would both an account of genuine explanatory of an observational concept, and a reason to think that there can be no scenario in which the subject is genuine that Sally’s deployments of observational property concepts do make it onto the scale. This will with Constraint I have that if there is a between potency of properties and goodness of explanation it be just that the interpretation that maximizes goodness of explanation maximizes the extent to which the property concepts the subject treats as explanatory are assigned potent properties. This would give us no reason to prefer an interpretation that treats a subject as explaining the course of her experience in ways that do why things behave as they do to one that treats the subject’s explanations as in genuinely potent properties, but how these properties to one If 5 is to be the will have to be that there is a version of Constraint 2 which is in Sally’s case, and which that explanations must potent properties in ways. the discussion so §1 out Williams’s solution to the permutation challenge. §2 the Demon case to argue that this solution Williams to what I suggested is an the that at locking onto ordinary things as objects of thought depends on sufficient in explaining the behaviours of these things in of the actual causal of actually potent properties. This section has more be about the of each of the in this But rather than about which might be the best for to keep Williams’s response to the permutation I shall by to the challenge I have So far I have on one of Williams’s But the discussion of this a general that I shall now to bring out. by to Williams’s Substantive Radical Interpretation The correct interpretation maximizes the extent to which the subject is treated as to reasons by her Now, why are we to be by SRI in the first I it that the is like we with the that a right account of the of states treats beliefs in by experience and rationalizing given we have at least the following and constraint on the of representational contents to and To count as in this we must go about things in a way that, as far as assignments to the subject of beliefs that are not by her experience do not her does the about role entail and given the role of a who this constraint is not genuinely in the of to work out what the subject at But an that to this constraint just is one that maximizes So if the initial about role is it with it the view of correctness of assignment of beliefs and belief the correct assignment maximizes the extent to which the subject is treated as beliefs that are by her I this to be a of thought in of a general approach. SRI results if we add that the relevant of is substantive In Chapter 2, Williams gives us an argument for this last The is with the that the correct interpretation might need to maximize only rationality – the extent to which the subject’s beliefs in a way that is appropriate given her experience Williams’s point is that the is to a for correctness of interpretations that will out wrong For example, consider interpretations I and which with respect to what they treat the subject as in response to her experience, but with respect to what they treat her as with respect to the world her I treats the subject as (in that the world is there when she is not treats S as to the of only the and of the world that she that the only constraint on correct interpretation is that we maximize the extent to which the subject is treated as beliefs that are by her experiences and that to various I and will be Now, Williams takes it that the wrong interpretations are … He for SRI on the that a right framework must entail the of wrong But given the role by the to the role of belief in the of thought, there is we can say about if you are to be an at you should Williams in moving to us that we can off some of for the rationality of a subject’s beliefs which are the beliefs form a which the that they are by her that the beliefs various and Williams is right that there are on being that go these – them as as there are interpretations that are out by substantive but not by contents to the subject’s beliefs in a way that maximizes her to the is not contents to her beliefs in a way that maximizes the extent to which she is treated as beliefs that are given her But in that case, given the initial point about in a way that maximizes only to is not genuinely the subject as and beliefs at So if the initial point about role us as far as the and it us the way to to the of thought as far as and but to Williams in SRI is this the general challenge to Williams’s framework that I want to bring out can be put like 1 right account of the of states treats beliefs in by experience and rationalizing given 2 The at 1 generates SRI – the that the correct interpretation of a subject is the interpretation that maximizes the extent to which the subject is treated as to reasons by her of this section so 3 SRI entails that that a subject as genuinely with respect to a given part of her also this part of her as involving the of beliefs that in being about version of the aspect of the SRI-based framework by discussion of the Demon 4 that may intact. to be by the Demon case, and to some extent by the discussion of the first in are so one of them has to The discussion of was on 3 and But own for what it is is that the is I think Williams is just right with respect to of what about what the view might in many The Metaphysics of Representation is a in how to be an if you want to be of the things about which I think is right are into 2 and 3. But if 2 and 3 are right, 4 is And this We need to say about the role of

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.