With this observation, Kripke asserts, Wittgenstein has raised a new type of philosophical skepticism regarding what it means to grasp a rule -- and indeed, what it is to make meaningful use of language. The skeptic argues that there is no fact about an individual that could constitute his or her meaning that he or she is following rule R rather than rule R' (WRPL 39). If one is to answer the skeptic, one must show what fact it is about the mental state of the person apparently following rule R that actually constitutes his or her following rule R rather than rule R' (WRPL 11). The problem the skeptic raises, according to Kripke, is a normative one in that there is nothing about that person's present mental state that determines what he or she ought to do in the future (WRPL 56): just because the person thinks he or she is following rule R now is no guarantee that he or she will follow rule R in future situations requiring the kind of response typically generated by R.

It might be argued that the normative question is answered just in case the person follows the appropriate rule correctly. But to this response Kripke objects that -- once again, if we are to try to justify this (apparently) correct rule-following strictly in terms of the person's mental state -- we would then need to posit a rule for interpreting the original rule, and a further rule for interpreting that rule, and so on ad infinitum (WRPL 81). In short, there apparently is no non-regressible internal fact about the purported rule-follower that can account for his or her following the given rule should an appropriate situation come up in the future.

In sum, then, the skeptical critique asks, "When a person's behavior appears to indicate that he or she is following rule R, how do we know that he or she is actually following rule R and not some rule R' which would produce the same output given the same input? Indeed, how does the person him- or herself know that it is rule R and not rule R' that is now being followed, even if at all appropriate times in the past he or she was in fact following rule R?" Because we cannot answer this question by looking at facts internal to a person, Kripke concludes, it is the case that

It is important to stop here to note what Kripke's argument is not. It is not an argument about epistemology. It does not, in other words, depend on the objection that the attribution of rule-following (or, by extension, any attribution of a determinate internal content to another person) is rendered impossible by poor or insufficient evidence by which to justify the particular attribution. The skeptical critique thus is not a variant of Hume's critique of inductive inference, though Kripke does note that Hume's critique is "analogous" to the skeptical critique (WRPL 63). It is worth mentioning, in fact, that at the epistemological level, nearly any explanatory device -- including that offered by Kripke -- is vulnerable to charges of evidential underdetermination. Inductive uncertainty cuts both ways here, in other words; as Chomsky points out in his response to Kripke, it bears on explanatory devices generally (KL 237).

Nor does Kripke's argument represent a behaviorist attempt to refute the reality of internal states generally. As Kripke points out, his argument does not assume -- nor in fact does it require -- a dismissal of the "inner," however that concept may be defined (WRPL 14, 44). To be sure, Kripke does assert that hostility to the inner "is to be argued as a conclusion" (WRPL14). But this hostility does not consist in denial of the existence of inner states; rather it consists in the denial of their explanatory priority and force. It is for that reason that Kripke frames his difference with Chomsky's linguistics in terms of a reevaluated notion of competence that would discard the idea that talk of internal rule-following asserts facts or explains behavior (WRPL note 22 at 31).

To repeat, then, the skeptical critique is directed toward the normative problem Kripke sees growing out of efforts to trace "a certain nexus from past to future" (WRPL 62), and it rejects the suggestion that facts about an individual's internal states can provide the grounding needed to settle the problem.

2. COLLECTIVE FACT AS SOLUTION

After concluding that the normative problem of (apparent) rule-following behavior cannot be solved by looking at facts about the individual, Kripke offers a "skeptical solution."

Kripke begins by stating that we must shift the focus of our explanatory search beyond the individual and redirect it toward "a wider community" (WRPL 89). If we do so, Kripke argues, we will come to acknowledge that an individual can be said to have mastered (or to possess) a given concept if his or her "particular responses agree with those of the community" (WRPL 92). Accordingly, our attribution of concepts or rules to individuals should call on "no special `state' of their minds," but should consist instead in our taking them into the appropriate community provided they do not exclude themselves by sufficiently deviant behavior (WRPL 95). Their behaviors and responses must, in other words, accord with the appropriate regularities observed among the other members of the community. Kripke's main point here is that an individual's agreement with the judgments or responses of the given community is "essential" if the given rules and concepts are to be properly ascribed to that person (WRPL 96). Perhaps most important of all, Kripke states that this community-characterizing regularity of response "must be taken as a brute fact" (WRPL 98).

The two explicit claims Kripke's skeptical solution seems to make regarding the explanation of apparent cases of rule-following, then, are:

There is a third claim that Kripke does not explicitly make, but that I believe is implicit in his attempt to explain the futural force of norms. This is:

As we will see, this "transcendence condition," as I will call it, is the pivot on which normative explanation balances, and one with important implications for the viability of the skeptical solution. I will explore this issue in depth at a later point in this paper. For now, though, I will look at the first two of Kripke's claims in turn.

2.1. The Agreement Condition

The first claim -- that the correct interpretation of a rule is a matter of agreement with the judgments of others -- states what I will call the agreement condition. It is offered as an alternative to the infinite regress that, as we have seen, Kripke believes is entailed by the explanatory appeal to facts internal to an individual. The agreement condition would relocate the explanatory device from internal, individual states to the collective phenomenon of behavioral regularity. Under the agreement condition, Jones's announcing the sum "125" given the inputs 68 and 57 would no longer be considered the product of a mental algorithm, but instead would be understood as a response consistent with those of others in Jones's community. With this move, Kripke replaces the explanatory turn to attributed mental operations with a turn to Wittgenstein's concepts of "agreement in judgment" and "forms of life."

Kripke defines a form of life (Wittgenstein's Lebensform) as "a set of responses in which we agree, and the way they interweave with our activities" (WRPL 96). The form of life is, in other words, a set of relevant agreements in judgment proper to a given group. The important point here is that forms of life are a matter of sameness of behavior in that Kripke takes forms of life to mean that "our actual community is (roughly) uniform in its practices with respect to [a given operation such as] addition" (WRPL 91). Kripke locates the center of gravity of forms of life in something called "the community," understood as a more-or-less uniform collective effect. If any instance of behavior is to be explained or described, it therefore will have to be explained or described in terms of its conformity to behaviors observed throughout the community. The emphasis here thus shifts from (attributed) facts internal to individuals to a collective fact that is observed as a set of consistent or regular effects enacted by members of a group.

That is the primary sense in which the notion of forms of life appears in WRPL. As Chomsky points out, Kripke employs the term in a second sense as well. In this second sense, which Chomsky characterizes as "metaphorical" (KL 232), Kripke concedes that forms of life may encompass "highly species-specific constraints" on behavior (WRPL note 77 at 97; also quoted in KL 232). In fact this second sense of forms of life -- whether or not it actually is metaphorical -- represents an interesting broadening of the original definition, since it implies or can be construed to imply that membership in a species, and the (presumably) built-in behavioral constraints that entails, constitutes de facto agreement with the right kinds of responsive regularities. If so, this would introduce a notion of community so broad as to allow nearly unlimited deviations within its scope and still qualify them as representing community-conforming regularities. Indeed, one might well wonder whether under this broad construction of form of life any activity engaged in by any member of a species could, in the end, qualify that member as a "member of the community." But then this would seem to empty the idea of agreement -- the basic criterion of community membership, and hence participation in a form of life -- of any useful content.

For his part, Chomsky notes that acceptance of the notion of species-specific constraints would in fact come down to a corresponding acceptance of the "Cartesian conception of `other minds'" (KL 234). This would effectively shift the criterion for determining participation in a form of life from community-relative behavioral regularities to an inference regarding the makeup of others, i.e., that their use of language, apparent exercise of will, creative performances, etc. are evidence enough to attribute to them a mind like enough to one's own as to confer recognition of personhood upon them. Note, though, that Chomsky cautions that the attributions made on the basis of such inferences are hypothetical, especially when considered in the unscientific, everyday contexts in which they typically are drawn, and that drawing them does not answer the potential objection that there (still) is no internal fact of the matter to consider in cases of apparent rule-following (KL 236). It is, in effect, a working hypothesis along the lines of Dennett's "intentional stance" -- something that serves well under the ad hoc conditions of everyday life but may not, and indeed even need not, hold up under scientific scrutiny. In any event, as Huen acknowledges in his commentary, if we restrict ourselves to what is given in Kripke's argument, then Chomsky is correct to assert that the notion of form of life does indeed encompass the notion of personhood implicit in the Cartesian stance (Huen Section 1.2(2) "Reply").

Chomsky's suggested alternative aside, it would appear that as Kripke presents it, the agreement criterion is circular. This is because if we understand form of life to designate socially rather than biologically shared patterns of activity, then presumably what constitutes a form of life (i.e., the behavioral agreements in terms of which such patterns are established) can only be based on that form of life. If this is so, we would simply be presupposing the criterion we are hoping to reach. But what if, instead of characterizing the agreement criterion as circular, we were instead to interpret its (apparent) circularity as evidence of its irreducibility? That in fact is what lies behind Kripke's second claim.

2.2. The ¥Explanatory Baseline Condition

The second claim -- that the behavioral regularities representing agreements in judgment are brute facts -- states what I will call the explanatory baseline condition. By "brute fact" I understand Kripke to mean "explanatory primitive."

The explanatory baseline condition holds that the behavioral regularities observable in given communities are irreducible as explanatory devices. For the skeptical solution, in other words, there is nothing to explain beyond the "brute fact" of agreement of responses that distinguishes members of the community. Under the explanatory baseline condition, reference to facts about individuals other than that they do or do not conform to the relevant behavioral regularities explains nothing of interest, since the explanatory status of such facts would be derivative at best, and misleading otherwise. There is (or is not) the fact of agreement, and the explanatory chain terminates there.

I earlier characterized Kripke's position as a variety of strong methodological collectivism, and now we can see why. For the skeptical solution demands not only that we factor in collective phenomena (agreements in judgment) to account for apparent rule-following behavior, but that we factor out any explanatory reference to internal phenomena whatsoever. The result is indeed the "hostility to the inner" that Kripke promised, and this hostility is nothing if not emblematic of strong methodological collectivism.

But Kripke's strong stance is, I believe, the source of the weakness of his position. For as I will show next, the skeptical solution's insistence that agreements in judgment are explanatory primitives -- its explanatory baseline condition, in other words-- is inconsistent with the transcendence condition implicit in any account of norms.

3. THE TRANSCENDENCE CONDITION

I have claimed above that the third condition implicit in the solution to the normative problem of apparent rule-following is that agreements in judgment must transcend any given instance in which they are embodied. I called this the "transcendence condition." I would now like to look at what I mean by this, and at what implications the transcendence condition carries for the skeptical solution specifically and for methodological collectivism generally.

We can introduce the transcendence condition by posing the following questions: Can agreements in judgment be reducible to the instances in which they are attributed? Or must they, as norms, be transcendent in relation to any given instances? Must they not form a basis -- a replicable basis -- on which our future agreements can agree? And if so, can they be taken to be explanatory primitives, or do they prompt us to say more -- about reproducibility, about causality, and ultimately, about states of affairs of individual psychology? There will indeed be more to say about these last three items, but for now it will be enough to say that the transcendence condition, simply put, is this: In order to fulfill their normative function, agreements in judgment must be applicable beyond any given instance of agreement. They must, in other words, transcend any given instance in which they are enacted.

3.1. Agreements in Judgment as Composites

I would like to suggest that for purposes of analysis, an agreement in judgment can be thought of as a composite phenomenon made up of two components: a normative content, and an exemplary embodiment. An agreement in judgment is thus

(1) a publicly observable event, action, or state of affairs that
(2) embodies certain information about how something should be done.

I believe that the first stipulation -- the public observability condition, we might say -- is not only consistent with Kripke's description of agreements in judgment, but even is required by it. Indeed, the entire notion of a community characterized by behavioral regularities would be incoherent without the condition that these regularities be accorded some kind of public status. This is underlined by the notion that such regularities can somehow be tested. If I read Kripke correctly, such tests would somehow have to put the individual response before the community or its representative(s), thus rendering the former publicly accessible and therefore subject to judgment.

The public observability condition would further entail that the agreement in judgment have the status of a concrete example. Under this interpretation, an agreement in judgment would be a normative model the function of which is to constrain or regulate behavior. As a model, an agreement in judgment would constitute a certain pattern in conformity to which members of the relevant group will attempt to conduct themselves in current or future situations. Such a pattern -- and it is here that we would find the information referred to in the second stipulation -- would serve as a regulating constraint on behavior or response. But in order to do so -- in order that its regulatory function should have any force -- it will have to be reproducible across similar cases. Put as explicitly as possible, an agreement in judgment -- that is, a relevant, replicable precedent functioning as a standard against which present and future activity can be measured -- must, by definition, be applicable, mutatis mutandis, in future situations. It must, by virtue of its normative content, project a certain influence from present to future. This requires that its normative content not be reducible to any given instances in which it is embodied. It will, in other words, have to overflow the particular situation in which it is enacted.

3.2. The Separability of Normative Content

What I am getting at here is that there must be a certain separability of content and embodiment. To hold instead that normative contents are not separable -- that is, that they are embodied in the act and that such embodiment is a primitive -- is to deny de facto that they are or can be applicable to other, future situations. For if they were inseparable from the particular instances in which they are embodied, they would not be anything other than contingent on a particular situation. And a purely immanent model cannot provide a basis for explaining how such norms can regulate behavior in a multiplicity of instances.

The separability of normative contents from the instances of their enactments becomes particularly important when the former are understood to underwrite competence. Consider that a competent actor is someone who can apply skills (or in more Kripkean terms, perhaps, conforms to the appropriate constraints) in diverse yet relevantly similar situations. Such skills, and the norms that help constitute them, represent a capacity that must generate the correct responses whenever called upon. For this reason, I think it is important to keep in mind that agreements in judgment make sense only if they are understood to represent potentialities that can enter into the production of behavior. But I will look at this causal function in more detail later.

A word of qualification is in order here. The normative content of a particular agreement in judgment may be applicable or relevant only in relatively narrow, well-delineated situations. This is because many competencies are applicable only in the context of more or less well-defined practical domains -- that is, fields of work, intellectual or professional disciplines, organized undertakings, etc., that are distinguished by particular methods, objects, materials, and perhaps established institutions as well. (Alternately, it is possible to think of a practical domain in Kripkean or Wittgensteinian terms as representing a particular form of life to which a particular set of agreements of judgment are appropriate.)

Upon analysis, then, agreements in judgment reveal themselves to consist of reproducible normative contents embodied in particular examples. With this understanding in place, we can begin to show how their normative force can carry over to future instances. We can begin to do so by looking at how agreements in judgment can be reproduced. This is, I believe, the crucial problem, and one in relation to which the skeptical solution breaks down.

4. THE PROBLEM OF REPRODUCTION

The problem of explaining how behavioral regularities are reproduced (the problem of reproduction, for short) is the problem of accounting for how behavioral regularities -- Kripke's agreements in judgment -- can be enacted and re-enacted by diverse members of the same community in diverse yet appropriately similar situations. The problem of reproduction is essentially this: Given that we are to characterize people's actions in terms of agreements in judgment or regularities of behavior, how can we account for the fact that members of a group do seem to instantiate behaviors that, taken in the aggregate, exhibit a certain consistency or regularity? How can it be that the behavior exhibited by person A in a given situation is also reasonably approximated by person B in a relevantly similar situation? Put another way, if we accept someone into the community on the basis of his or her agreement in judgment in a given area, how is it that we can get him or her to agree, and thus to be eligible to be considered a part of the community, in the first place? How, in other words, can he or she reproduce the responses that others would produce under the appropriate conditions? The problem of reproduction thus is the problem of explaining how different individuals can in fact come to the agreements in response that are supposed to undergird instances of (apparent) rule-following behavior.

It seems to me that the skeptical solution, if it is to be a solution, must be able to answer the questions posed above. What this comes down to is that the skeptical solution must explain, and not simply describe, the individual fact of Jones's supposed conformity to group responses. How did Jones get so that he responds with "125" when asked to sum 68 and 57? An answer to this question would build an explanatory bridge, as it were, to get us from the individual fact -- that Jones is behaving in a way consistent with other members of his or her group given the same situation -- to the collective fact, which is that members of Jones's group tend to act in the same manner given the proper circumstances. The skeptical solution must, then, tell us why Jones's behavior can reproduce the regularities observed among his group.

But it does not. The explanatory baseline condition, with its claim that agreements in judgment are brute facts, effectively rules the problem out of bounds. But it seems to me that this is begging a question that must be answered; referring to behavioral regularities as primitive simply takes for granted that which must be explained, and that won't do here. If we are to find an answer, we must look instead to the transcendence condition and its implications.

The problem of reproduction is, in effect, the other side of the transcendence condition. For if the transcendence condition holds that an agreement in judgment implies a normative content that is not reducible to any given instances in which it is enacted, then the problem of reproduction can be seen as concerned with how transcendent normative contents can be communicated to, and consequently internalized by, individuals. Implicit in this view is an understanding of normative contents as having a causal role in the production of behavior, and it is to a consideration of this causal role that I now turn.

4.1 Agreements in Judgment as Causes of Behavior

The problem of accounting for the reproduction of behavioral regularities is similar to what Stephen Turner in The Social Theory of Practices (STP) calls the problem of transmission. To be sure, Turner addresses the problem of transmission in the somewhat different context of the theory of socially-constructed practices. Accordingly, there are important differences between the problem of transmission and the problem of reproduction. But looking at these differences will, I believe, tell us something important about the nature of forms of life and the agreements in judgment of which they are comprised.

Turner's analysis of the transmission of practices is mostly concerned with the related problems of location (where exactly are "practices" located?) and sameness (how does one ensure that those who can be said to share a set of practices in fact do internalize the same thing?) (STP 61-62). Turner's analysis therefore is directed toward the causal claims made for "practices," i.e., that "practices" somehow are the causal component operative behind observable behavioral regularities, and that this causal component is the same for all of those to whom a given practice or set of practices can be attributed. No such direct and explicit causal claims are made by Kripke on behalf of forms of life and their constituent agreements in judgment. Thus there is no claim, correlated to the normative claim that explicitly is put forward, that forms of life or agreements in judgment are causal objects within individuals that are (somehow) the same in all of those individuals whose behavior appears to be regulated by them. The skeptical solution's rejection of the explanatory validity of the "inner" would, in any case, constitute a strong denial that the problem of sameness (in the sense that Turner understands it) is an issue at all. Indeed, Kripke might argue, the only sameness that must be accounted for is the sameness of behavior that is exhibited when an agreement in judgment is in fact displayed.

On the other hand, one might plausibly argue (and I believe it is the case) that the skeptical solution's appeal to agreements in judgment does have causal implications. For it would seem that a normative constraint would just carry a causal force entering into the production of a given behavior. Put another way:

Given points (1) and (2), I think point (3) follows, with the result that normative contents must be thought of as causal constituents of instances of behavior. My reasoning is that (1) there has to be some way of accounting for how different individuals can produce similar responses in similar circumstances, given that (2) they are guided by norms that transcend any given instance of their being put to use. This must be because (3) the content of these norms is a constituent of the states entering into the production of the behaviors in question. What we need to show, then, is how these norms i>cause people to act in a certain way.

5. EMULATIVE ACQUISITION

I would like to suggest one process that can account for how at least some behavioral regularities (i.e., agreements in judgment) can be reproduced. This process is one of emulative acquisition. Emulation, basically, is a process through which one acts by copying the actions of another. This may seem like a simple enough proposition, but the mechanisms underlying it are not. In fact the chain joining the exemplary agreement in judgment to its reproduction in a new situation can be traced through no less than the following four phases:

5.1. Demonstration and Observation

Emulative acquisition begins when one person observes another person demonstrating a skill or activity of some sort. This second person serves as a model, and his or her activity functions as a structured, exemplary response that the observer subsequently will try to reproduce.

I stated earlier that a model can be thought of as a publicly observable pattern. Consider now how the activity comprising a given performance configures the observer's stimulus field as a certain pattern. This pattern is made up of a focus -- consisting of the most important aspects of the activity -- and a periphery. Much of the organization of the stimulus field into focus and periphery probably can be attributed to the observer's cognitive-perceptual processes, among which variations of attentiveness most likely play a prominent role. Carol Kates in Pragmatics and Semantics (PS chapter 8) suggests that the purely contingent features of any given object, event, or state of affairs present to an observer are "bracketed out" -- or, put bluntly, ignored -- through a process analogous to Husserl's phenomenological reduction. The resulting selection provides the basis for a generalized conceptualization made up of features that can be expected to be relatively invariant across cases.

And this is where the transcendence condition comes in. What the transcendence condition requires is generalization, and that is what the observer's perceptual-cognitive processes would serve to bring about. For absent purely contingent aspects, the generalized conceptualization would embody a content that, once internalized, in principle could be brought to bear in diverse yet relevantly similar situations, as needed.

As suggested above, such generalization probably is in large part a function of the observer's attentiveness. But it is important to bear in mind that the attentive selection of observed features is as much a matter of the recognition of mind-independent facts as it is a (loosely speaking) mental act of construction. In fact it seems reasonable to believe that there is a large extent to which the configuration of the stimulus field serves to pre-select the salient features that will impinge upon the observer. An action taking place within a given setting would constitute a focal point to which the observer's attention would be drawn; it would count as a salient feature of the general environmental configuration in which it is located. The rest of the setting would stand as a kind of backdrop against which the main, attention-drawing event would take place. It is precisely this prestructuring of the stimulus field that would provide the basis for ignoring or otherwise leaving out any purely contingent aspects of the activity.

5.2. Internalization

The observer of an exemplary performance, then, confronts the stimulus field as a given pattern structured by salient features. Observers may internalize these patterns by way of the brain's pattern-recognizing functions, which give rise to correlated internal patterns, or content states (see Churchland chapter 9 and Purves chapter 8) within the observer. These correlated content states, which most likely are "stored" in the brain as synaptic weights disposing the brain to instantiate a given pattern in the event of occurrence of the appropriate internal or external stimuli, function as conceptual contents, or mental prototypes, which contain information relevant to replicating the activity. Such contents or prototypes in effect encode generalized information about the observed performance.

In claiming that internal content states are correlated to the external features to which they are related, I am not claiming that our mental states mirror the external environment. What I am claiming instead is that our internal states are related to, and in some complex way dependent upon, the features of the objects, events, and states of affairs they are directed toward. I am assuming that it is the function of our content states to (somehow) provide a reasonably accurate and reliable map of the external conditions they represent, and this requires that they be constrained in significant ways by the objects, events, and states of affairs represented. It does not follow from this assumption that there must be a one-to-one, causal and (therefore) infallible relation between external features and internal contents. Rather, the mechanism of representation is assumed to be a complex one in which multiple perceptual modalities must be synthesized, short term memory must function properly, and allowances made for proprioceptive and other internal stimuli. The result of such a process would be an internal state correlated to the situation it is "about" on the basis of general features inhering in or derived from that situation. This is all I mean by "internalization."

It would be in such internal states that normative contents would be embodied. Consider that the action demonstrated for the purpose of learning shows not only how something is done, but how it ought to be done. There would be, in other words, a normative dimension bundled into the more purely mechanical information imparted by the demonstration. For analytical purposes one might be able to separate out this normative aspect from the purely mechanical aspect, but it is difficult to see how this would in fact be accomplished by the observer -- granting that he or she would even want to make such a separation. Certainly he or she would assume that, all things being equal, what he or she is seeing demonstrated is what he or she ought to be doing in the given situation.

5.3. Causation

What the attentive observer acquires (internalizes) in watching the exemplary performance, then, is a content state carrying information on the normative as well as the mechanical aspects of the activity observed. This internal state further serves to play a causal role in response, given the proper circumstances.

Content states of the sort I am considering play a causal role in behavior by virtue of the fact that they are representational structures that guide the performances to which they are correlated. As Fred Dretske has suggested (Dretske 94), a content state is a cause of an instance of behavior to the extent that the behavior is undertaken on the basis of what the content state represents. The content state underwrites the behavior and, by virtue of what it represents, serves to ensure that a performance or activity takes the shape it does when it does. Let me amplify this point.

I have been using the expression "content state" as a general term to stand for an internal state that includes normative as well as cognitive or doxastic content. But whether we are considering internal states carrying information about what one thinks is the case or about what one thinks one ought to do, the causal relation to the performance is the essentially the same. The constraints contained in norms, no less than the environmental information contained in beliefs, provide reasons why a performance takes one form rather than another. One will act in a certain way not only because one takes the situation to be a certain way (rather than another), but because one thinks one ought to act in a certain way rather than in another. In narrowing the range of possible responses, in offering a kind of template for the response selected, a normative content plays a determining role in fitting the performance to its surrounding environment.

In the foregoing sections, I have attempted to provide one account of how agreements in judgment could be reproduced. To be sure, that account is conjectural, and is restricted to one simple type of learning. But it does attempt to explain the highly problematic matter of how behavioral regularities can be reproduced by individual members of a group. I believe something like it must be operating if we are to explain any of a number of behavioral regularities characterizing given social groups; in all likelihood, the model must be expanded to account for other forms of learning, including those with explicit reliance on and transmission of rules.

It is natural to ask here whether emulative internalization, with its emphasis on experiential learning, could account for language learning. I do not claim that it can. Thus I am not attempting to offer an empirical alternative to Chomsky's rationalist theory of language acquisition (as has, for example, Kates in PS). I do think, however, that something like the process I have described can and probably does account for some aspects of vocabulary acquisition either directly (through, e.g., ostension) or indirectly (through the acquisition of cognitive prototypes underwriting linguistic reference). Such experientially-derived learning of these aspects of language is not ruled out by Chomsky. But the (limited) purpose of my introducing this account has not been to postulate a theory of language learning, but rather to provide the explanation for normative reproduction that Kripke's account lacks, but nonetheless seems to call for.

The general point, then, is that if we want to try to explain how it is that people come to agree, we cannot keep the explanatory baseline condition, which holds that the behavioral regularities that "are" agreements in judgment stand as explanatory primitives. On the contrary, explaining reproduction means making reference to underlying states within individuals -- something that is not compatible with the explanatory baseline condition.

There is one other matter that I think deserves attention.

5.4. Normative Judgments versus Dispositions to Act

In a perceptive move, Kripke rules out a dispositional response to the skeptical paradox because, he states, a dispositional account of behavior is descriptive and does not engage the question of normative constraints on behavior (WRPL 37). Kripke is right on the mark with this. But I would add something here: the difference between the dispositional account, which tells us what will happen given certain conditions, and the normative account, which tells us what ought to happen given certain conditions, is the evaluative and/or discriminative moment that characterizes the latter.

This is to say that when it is a matter of applying or conforming to norms, we must at some point determine what response would be relevant, given the circumstances. This I believe is an essential aspect of normative behavior: one must decide (or choose, or even acquiesce in) what is appropriate, i.e., what one ought to do. There is, in other words, a discretionary aspect to norm-following. I would argue that this is an essential feature of normatively informed behavior, and that it is only by virtue of it that we can speak coherently of agreements in judgment as involving judgment at all. But we will miss this discretionary aspect entirely if we do not make reference to internal states.

It seems to me, in fact, that bringing given internal contents to bear in certain situations is the essence of judgment. "Judgment" usually refers to an activity involving cognitive or doxastic states -- internal states, in other words. (The same is true of the German Urteil, which of course is the term Wittgenstein originally used.) In fact Lycan's "crude" definition of judgment captures what we might think of as the salient feature of judgment: "to judge is to host and manipulate a representation, an inner formula of some sort" (Lycan 55). Thus it would seem that if we are to attempt to explain judgments, we would at some crucial point have to talk about facts within individuals. For that reason, it is striking to see the expression "agreement in judgment" in the context of an argument that would all but eliminate recourse to psychological explanation.

6. METHODOLOGICAL INDIVIDUALISM, NONETHELESS

In offering a position informed by emulative internalization to account for behavioral regularities, I have shifted the primary focus of explanation away from collective facts and back to individual states. In effect, I have inverted Kripke's inversion of explanatory criteria. In doing so, I have, of course, rejected Kripke's methodological collectivism in favor of an account that represents a variety of methodological individualism. But I believe that if we do not perform this inversion of Kripke's inversion, we will find ourselves in a position just as untenable as that which Kripke rejects, and for the same reasons. That is to say that, with the acceptance of agreements in judgment (or behavioral regularities) as explanatory, brute primitives, we fall into a skeptical paradox similar to the one Kripke claims to find in explanations based on facts about states within individuals.

This paradox is brought into sharp relief by the reproduction problem. In effect, without somehow accounting for the reproduction of agreement in any appropriate case, we cannot account for how a present agreement, embodied in a present performance, can determine that a future performance will embody the appropriate agreement. The same problematic, normative nexus from past or present to future that Kripke cited when faced with explanation based on internal facts is not eliminated when explanatory devices are inverted, but rather is displaced. Unless we can come up with an account of the way diverse individuals in diverse yet relevantly similar (or appropriate) circumstances can replicate the "correct" agreement in judgment, we become subject to the same criticisms Kripke leveled at individualistic accounts: appealing to the idea that one's agreeing in judgment in the appropriate situation is an explanatory primitive may be "irrefutable," but does indeed seem "desperate" and in any case leaves the allegedly primitive state "mysterious" (WRPL 51).

Without a convincing (or indeed, any) account of how these agreements in judgment can be reproduced, we are left asking the same question the skeptical critic directed to toward rule-following explanations: how can one account for the future normative force of past or present agreements in judgment? How, in other words, can one ensure that, given the appropriate conditions, the community will be in agreement in the future? As I have argued above, in order to function normatively, agreements in judgment must contain a transcendent normative content that can play a causal role in the future production of behavior. And as I also have argued, this situation means that we must, in discussing agreements in judgment, refer ultimately to facts about individuals and their internal states.

7. CONCLUSION

Like Chomsky, then, I do not think we can come up with a satisfying explanation of behavior -- whether considered individually or as a given regularity across cases -- without referring to states about (internal to) individuals. At the same time, I accept Kripke's conviction that observable modes of behavior serve to ground the habits attributable to any given individual. What I found inadequate in his account was the neglect of any explanation of how modes of behavior can be assimilated by those whom we would "take into the community."

It might seem that I have split the difference between Kripke and Chomsky. If this is the appearance, it may be because Kripke and Chomsky are proponents of the "strong" versions of their respective sides. One consequence of the strong collectivist viewpoint I have rejected is, as Huen puts it, that "the notion of norm cannot be purely based on some fact of an individual" (Huen Section 1.2(2) "Reply"). But the key here is "purely" this qualifying adverb opens quite a loophole in the argument -- one big enough to drive a methodological individualist through, as it were. On the other hand, a purely individual normative fact would presuppose a correspondingly strong claim for the individual -- perhaps something deriving from some form of rationalism, which of course is a fundamental element in Chomsky's overall program. But what is a "non-purely" ("impurely"?) individual normative fact? Can it be something that is external and yet at the same time is internalized? That in fact is what I have argued above.

Daniel Barbiero

Click here to go the COMMON BIBLIOGRAPHY

Click here to see Works Cited in this Essay only.