Naturalizing Idealizations: Pragmatism and the Interpretivist Strategy.
Bjørn Torgrim Ramberg, CSMN, University of Oslo
b.t.ramberg@csmn.uio.no
This paper appears in Contemporary Pragmatism, no 2, vol. 1(2004), pp 1-63.
Let us say, with Quine, Davidson and Dennett among others, that a person's language and
psychological attitudes have their identities fixed with the theories generated by an idealized
interpreter of that person (Quine, 1960; Davidson, 1984, 1986a, 1986b, 1989a, 1989b, 1990a;
Dennett, 1978, 1987a, 1991a). A reason for saying this is that it will help us see how the
capacities to entertain attitudes and to communicate linguistically can be natural capacities,
capacities we may happily attribute to creatures who fall squarely within the scope of
evolutionary biology. This, at any rate, is Rorty's principal reason. The interpretivist strategy
permits us, Rorty suggests, to give an account of persons which introduces
no breaks in the hierarchy of increasingly complex adjustments to novel stimulation—the
hierarchy which has amoebae adjusting themselves to changed water temperature at the
bottom, bees dancing and chess players check-mating in the middle, and political
revolutions at the top. (Rorty 1991b, 109)
How does it do this? I will develop an answer emphasizing the naturalistic motivations of the
interpretivist strategy, an answer that is also intended to draw out and situate some of the
commitments underpinning the view of philosophy that Rorty has worked out over the last thirty
five years (1967, 1979, 1982, 1989, 1991a, 1996, 1998b).1 While this combination of
1
Though he would probably have resisted the radical contextualism I propose in sections IV and
V, as I develop the interpretive strategy I take myself in the main to be following Davidson. I
will provide very little in the way of explicit textual defence of the readings I impose on him, and
on the works of Rorty and Dennett. Since the assimilation I imply is controversial, this may
seem an odd omission. However, if the burden I assume in this paper can be sustained, it
suggests that a Rortyan reading of Davidson and Dennett has the virtue of motivating aspects of
Naturalizing Idealizations. Page 2.
constructive polemic and metaphilosophical commentary makes for a long paper, the view of
Rorty's pragmatic philosophy that I want in this way to make vivid can be stated briefly. Rorty's
thought represents a dialectical transformation of naturalism. As he brings naturalism to bear
fully on the project of philosophical reflection itself, Rorty finds himself fundamentally changing
the requirements we impose upon our thinking whenever we seek to assume a naturalistic
philosophical stance toward some subject matter.2 To appreciate the naturalizing capacity of the
interpretivist strategy is to understand how Rorty's naturalistic critique of philosophy alters the
nature of naturalism itself.
There is an ulterior purpose behind this indirect approach to Rorty's pragmatism. Rorty laments
the tendencies toward "decadent scholasticism" of contemporary professionalized philosophy.
(Rorty 1993a, 1995a) The response he advocates has been, on the whole, the complete
overthrow of the vocabularies in which much contemporary professionalized philosophy is
carried out. The vocabularies of epistemology, of content-oriented philosophies of mind and
language, the vocabulary of meta-ethics, these are all, to Rorty’s ear, corrupt beyond redemption.
However, it seems to me that Rorty would give up nothing of substantive importance, and indeed
be better placed to reclaim some useful fortifications from which to combat the decadent
scholasticism that he often astutely exposes, if he were less revolutionary inclined toward the
historically developed vessels within which much philosophy is presently conducted. Rorty
their views that critics have found unsatisfactory. It is not my concern here to defend exegetical
claims of a more categorical nature.
2
This diagnosis suggests that Rorty’s is a highly dynamic project, with an impetus for constant
revision of its own terms built in right at the core. By the same token, it provides a
methodological hypothesis for an account of the changes in Rorty’s views over the last 30 years.
Both his early physicalism and a somewhat more lasting tendency to think of rationality in
algorithmic terms (and therefore to be hastily dismissive of the notion) may be fall-out from
attempts to naturalize philosophical reflection in terms that later came to be undermined by that
very endeavour. (Cf., Rorty,1994, 126.)
Naturalizing Idealizations. Page 3.
could afford, for instance, to be less reticent than he has so far been about invoking notions like
'rationality' and 'norms of reason'. Now, the differences in the respective depictions of the
concept of rationality in Rorty (1989) and Rorty (1996) indicate movement in the direction I
would urge; the former is dismissive (eg., 1989, 47), whereas the latter cautiously suggests that
'rationality' is a reconstructible notion that can be made to do useful work. I think the point
generalizes to pretty much all the terms that in philosophical contexts arouse Rorty’s deep
suspicions; “knowledge,” “truth,” “content,” and the like. As I see it, the best remedy for
decadent philosophy is to conduct an aggressive campaign of pragmatizing reappropriation of
just those terms that traditionally have been employed to express ahistoricist and essentialist
conceptions of reflection. If this essay were successful it would lend illustrative support to this
claim, and help pave the way for an invigorated, assertive, historically self-conscious brand of
pragmatist philosophical reflection that I like to think of as Revisionist Rortyanism.3
I. Pragmatic Redescription versus Philosophy of Mind.
A distinctive feature of the interpretivist strategy as it has been developed after Quine (1960), is
that it aims for naturalization without taking the route through nomological or conceptual
reduction. Where some see only three alternatives—some form of reduction, outright
elimination, or a retreat to dualism—the post-Quinean interpretivist claims to mark out a fourth
possibility.
The coherence of this possibility can certainly be doubted. Fodor, for example, persistently
argues that the interpretivist's theory of the attitudes, with its inevitably ensuing holistic
individuation, is really a coy form of eliminativism.4 For Fodor, the honest position to take, if
3
From here on, I will often use ‘naturalistic pragmatism’, or simply ‘pragmatism’ (and related
forms), to refer to this view. .
4
See Fodor (1987, chapter 3) and (1994, 5-7); and Fodor and LePore (1992, 1993a, 1993b).
Compare Fodor's remark: "If aboutness is real, it must be really something else." (1987, 97)
Naturalizing Idealizations. Page 4.
you must be an interpretivist about the attitudes, is that of Quine (1960).5 Kim, to take another
example, has no less persistently argued that a commitment to monism, to physical predicates as
the proper constituents of basic laws, and to the reality of the attitudes, makes the reduction of
the mental to the physical inexorable.6 Such arguments are typically rooted in firmly intuited
constraints on what it really is to consider something as real, intuitions that yield the
metaphysical conviction that naturalism and reductivism (or eliminativism) are inseparable.7
Pragmatists will treat such ontological intuitions as symptoms to be examined, not as foundation
for argument. They will see them principally as expressions of commitments to particular
vocabularies.8 The attempt to settle what the reifications of a vocabulary really are, in terms of
Dretske has a similar view of the options (1988, 1995).
5
Quine, of course, softens his stance in later treatments of the attitudes. See Quine (1991) for a
view not very different from Davidson’s on the ontological status of the attitudes.
6
See Kim (1993), particularly essays 14 and 17. For Davidson’s defence of anomalous monism,
see Davidson (1993a).
7
Reformist Rortyans follow (the later) Carnap and decline explicit argumentation about what it
really is to consider something to be real. By contrast, Fodor’s fundamental problem with
interpretivism is exactly that our explananda are rendered insufficiently real, at least in a sense
of “real” allegedly required by Granny’s touch-stone intuition, tutored—as in Granny’s
grandson’s case—by a clear-eyed appreciation of the ontological presuppositions of our idea of
what it is for scientific explanation to succeed. For Kim, to take another example, non-reductive
physicalists fail to perceive the incoherence of their conjunctive position because they do not see
that, fundamentally, what there is, is properties; if states and events are individuated by their
properties, then, for any physicalist, a non-physical property of some event must be the very
property it is just because it supervenes on exactly those physical properties upon which it does
in fact supervene. This allows us quite plausibly to conclude that a minimum commitment of
physicalism is that non-physical predicates refer only in so far as they are nomologically related
to physical predicates. Similarly, Frank Jackson (forthcoming) takes the view that physicalism
commits us to the view that assertions framed in non-physical terms would be made true by
states of affairs characterizable in the terms of physical ones.
8
This is a central notion for Rorty, as Brandom (2000b) stresses. I will have more to say about
it in section III.
Naturalizing Idealizations. Page 5.
some other, ontologically legitimizing vocabulary, is itself at odds with a naturalistic view of
thought. Pragmatists do not want to say that the mental is really something physical or material.
Nor, though, do they want to say that, really, it is something non-material or non-physical.
Naturalistic pragmatists are proposing ways to describe ourselves as thinkers and agents that
make the philosophical contrast between mind and matter seem to be without any particular
ontological point. Perhaps one might signal this sort of attitude by calling oneself a nonreductive physicalist (Rorty, 1987). My strong suspicion, however, is that it is not very helpful
to try to spell out the antidualistic commitments of a pragmatized naturalism in terms of its
relation to physicalism. 'Physicalism'—in all its varieties with their attendant conceptual
distinctions—is burdened with the connotations of a dichotomous folk-ontology, one that has
been hypostatized in the terms of art of the kind of philosophical vocabulary to which naturalistic
pragmatists are busy working up alternatives.
Indeed, our notion of mind and the vocabulary in which it is embedded well illustrates how
philosophical analysis and 'intuition', providing mutual support and reinforcement, can entrench
a particular set of problems and make them appear mandatory. Unfortunately, though, it could
also be taken to bear out the anti-pragmatist point that "mere coherence" is not enough; we need
a touch-stone against which to test the truth of even the most reflectively equilibrated beliefs. If
not a priori reflection, then empirical science may provide just such a touch-stone—so long as
we believe that science can aim to articulate a description of the world warranted by criteria that
are demonstrably truth-indicative (eg., Haack, 1993). Demonstrably truth-indicative criteria, we
realize, are ones that normative epistemology will show we have good reason to believe point us
toward the way the world is, in the way we have good reason to believe that a compass will point
us toward the Magnetic North Pole. If we fail to perceive the conceptual connection between the
very idea of justification—or assertoric warrant—and a distinct truth-norm (eg., Haack, 1994;
Wright, 1992), a connection which allows us to draw a distinction between genuine, objective
Naturalizing Idealizations. Page 6.
warrant and mere assertion-games, then we are stuck with parochial coherence as our only
measure. The result is a kind of idealism without the innocence, a jaded ironism with no
recourse to rational means of settling theoretical (or practical) conflict. A charge against the
pragmatic view I defend is precisely that this is just where it leaves us (Haack, 1994).9
What the pragmatist suggests, however, is that this very construal of enquiry and of warrant and
of truth is forced on us by the assumptions embedded in an entrenched vocabulary of mind. This
vocabulary leaves much of philosophy preoccupied with conceptual problems the various
proposed solutions to which generally float quite free of the practical and theoretical problems
that engage us as the 21st Century gets under way. The pragmatic philosopher treats such
conceptual problems as points of leverage for vocabulary shifts. Tracing questions posed in the
vocabulary of mind back to the assumptions that make them appear compelling, pragmatic
philosophy is self-consciously historicist. This is not, it is important to note, to reduce
philosophy to the telling of the history of philosophy. It is to oppose a conception of philosophy
that treats the history of the subject as a more or less valuable heuristic aid to reflection. The key
historicizing move of the pragmatist is to temporalize meaning, and so to treat content in sociogenetic terms. This move is what makes advancement in philosophical understanding
inseparable from the telling and retelling of reconstructive histories of the problems we are trying
to understand.10 The pragmatist will, accordingly, offer genealogies of philosophical problems,
9
For Rorty’s scepticism toward the idea of a truth-norm as deployed by Haack and by Wright
respectively, see Rorty, 1995b, 1998c.
10
Hegel, Nietzsche, Heidegger, and Gadamer make versions of this move the focal point of their
thought, and this is what makes these thinkers such attractive tool-boxes for Rortyans. Allen
(1993) demonstrates the power of this approach in his historical account of the role of the
concept of truth in philosophy. McDowell (1994) and Brandom (1994) are recent works which
fully integrate this semantic historicism into groundbreaking displays of constructive philosophy.
Confronting issues that are still often posed and debated in ahistorical terms, as if the questions
were posed sub specie aeterni, Brandom and McDowell both show up the intellectual poverty of
ahistoricism.
Naturalizing Idealizations. Page 7.
genealogies which aim to redescribe our philosophical urges and inclinations in such a way that
we can extricate those theoretical aims we may want to stand by from what has appeared to be
mandatory frameworks for their articulation.11
Now, the interpretive strategy plays a crucial role in this effort, because, pragmatists hopefully
believe, it will allow us to precipitate out a vocabulary of agents and thinkers from the
vocabulary structured around that pair of intimate antonyms, 'mind' and 'matter'. Pragmatists
hope that this will, eventually, undercut the governing intuition of Philosophy of Mind, the
conviction that the kinds we capture with psychological ascriptions just could not in themselves,
at least not straightforwardly, be natural states of natural creatures. Pragmatists do not believe
that our practice of psychological ascriptions leads us inexorably to the mind-body problem.
Rather, they see in 'mind' the vestiges of 'soul', and hypothesize that the real problem is actually
our deeply-rooted attachment to this ancestral notion—and our concomitant commitment to the
idea that the relation of 'mind' to its conceptual counterpart is a central philosophical difficulty.
It is this attachment that makes it appear prima facie mysterious how the vocabulary by which
we are able to treat some things as agents could capture a way that some natural creatures (and,
perhaps, artificial systems) are in the world. What the pragmatist polemic takes aim at, then, is
this attachment, this deep-rooted commitment. This, for the pragmatist, is what philosophically
motivates the interpretivist strategy.
The pragmatist is not claiming to solve the mind-body problem, nor to dissolve it. Nor is the
problem being diagnosed as illusory, as a product of some form of conceptual confusion,
linguistic mistake or general lack of semantic alertness. The pragmatist takes the mind-body
problem to be real, but transient. It is a problem we will come to see as idle once we have
11
This strategy has long been explicit in Rorty; cf., Rorty (1967).
Naturalizing Idealizations. Page 8.
developed better ways of conceiving ourselves and our relations to our surroundings, once we
have developed, that is, better vocabularies. These vocabularies will be better in the specific
sense that they will enable us to treat certain items as agents without sticking us with
dichotomous schemes of fundamental ontological kinds, the kind of kinds whose relation one to
the other cannot but become immediately problematic. The interpretivist strategy is attractive
because it holds out the promise of just this kind of improvement in our conception of the
capacities that make us persons.
Dennett may be used to illustrate the approach I have in mind. Contrast Dennett's attitude to the
attitudes with his attitude to consciousness; to put it crudely, the attitudes survive (1987a, 1991a)
while consciousness must go (1991b). Why is this? For Dennett, a principal tool in the
campaign against intuitions of mind and the reifications that philosophy spins from it is natural
science. Then, has neuroscience discovered the underlying states that we might plausibly take
belief-talk really to refer to? Have neuroscientists determined that, really, there is no such thing
as consciousness? This, we know, is hardly Dennett's point (cf., 1981c, 93; 1991b, Part III,
Appendix A). Dennett is motivated by the diagnosis that the folk-notion of consciousness keeps
us wedded to a set of interwoven descriptions of mind and self that inhibit our susceptibility to
the naturalizing influence of science on our self-image. This set of descriptions is what we
gesture at with the notion of the subjective. The sense that the notion of the subjective is a rich
and bona fide mine of philosophical problems and insights is an explicit target of Dennett's
seditious account of mind.12 Dennett's view is that the linguistic practices in which our notion of
consciousness is embedded (the vocabulary from which the philosophical invention 'qualia' takes
its intuitive power—see Dennett, 1988; 1991b), are practices we would do well, if we want to
naturalize our conception of ourselves, to alter. But this, any pragmatist knows, we can do only
12
Also of Davidson’s. See Davidson, 1989a, 1989b, 1991b. Rorty (1994) endorses the point.
Naturalizing Idealizations. Page 9.
in so far as we are able make satisfying alternative descriptions available. As I read Dennett
(1991b), his ambitious account of consciousness exploits the tensions at the interface of science
and common sense in order to alter our sense of what is obvious about awareness, of what are the
base-line home truths of our experience of being what we are. Seeking to instil changes in the
linguistic habits that make up the vocabulary of the phenomenology of experience, Dennett
hopes to make descriptions arising from natural science mesh more easily with our sense of what
matters about us as persons.13 The vocabulary of the attitudes, by contrast, is in Dennett's view
redeemed by the interpretivist strategy. Why? Because, like Rorty, he takes it that this account
shows how it is that attitude ascriptions can be useful to the point of indispensability to natural
creatures like us.14
I am about to offer a version of the interpretivist strategy that will enable me to make explicit its
intimate connection with a pragmatist conception of rationality and of philosophy. I will be
stressing two closely related aspects of this intimate connection. Pragmatism serves
interpretivism, in so far as an effective defence of the interpretivist strategy against common
objections will appeal to a pragmatic conception of rationality. Interpretivism serves
13
On this reading, Dennett's conception of philosophical strategy and of how philosophical
practice may contribute to our betterment is similar to Rorty's. The differences between these
two philosophers are best explained with reference to their respective therapeutic aims. Rorty’s
attempts to affect vocabulary change are explicitly grounded in a commitment to democratic
liberal politics. Dennett's stem from the over-riding goal of making it possible for us to adopt a
self-conception consistent with the picture of human beings that is emerging from natural science
without moral or spiritual impoverishment. Both philosophers, I believe, would regard these
projects as related.
14
I hasten to grant the truth of the suspicion that my reading of Dennett may be anachronistic; as
a matter of exegesis, it may be a distortion to see the formulation of the intentional stance,
particularly as developed in 1978 and 1981a, as informed by the ontological tolerance which is
present in Dennett’s recent writings. Thus it may be that the pragmatic response to the charge of
instrumentalism that I eventually offer in the final section is one that Dennett at that time neither
would nor could have availed himself of. For my purposes in the present paper, I can afford to
be agnostic on this point.
Naturalizing Idealizations. Page 10.
pragmatism, in so far as the strategy becomes, in the context of the conception, a tool for
naturalization. A crucial question, then, is this. How can we tell whether the interpretivist
strategy will actually do the job the pragmatist wants it to do? This question is not yet the
question of how the strategy can be an instrument of naturalization; the latter issue is
metaphilosophical, and turns on our assessment of the conceptual resources that interpretivists
employ in formulating the strategy. The prior question asks simply how we may assess the
interpretivist's claim to be giving an account of content, whether that account is regarded as
naturalistic or not: does the strategy produces a convincing account of what we, the folk,
recognize—or, better, can be brought to recognize—as key features of the lives of persons?
The states with which the interpretivist is concerned—the states we invoke when describing
creatures as agents and thinkers—are anchored in our attributive practices of run-of-the-mill
interpretation and psychological explanation, and these practices provide the measure of
plausibility. Consequently, an important kind of argument against interpretivists is one which
drives in a wedge between what a given version of the approach says about the individuation and
attribution of psychological states and the kinds of relations these may stand in to each other, on
the one hand, and the individuative and attributive practices embedded in our ordinary use of
folk-psychological terms, on the other. Much of what follows will address this sort of argument.
In particular, I will examine claims to the effect that the interpretivist strategy is incompatible
with the degree to which and the manner in which the irrational and the non-rational enter into
our accounts of persons.
II. The interpretive strategy and two reasons why it seems so implausible.
Consider the ideal interpreter, IDA, let us say. IDA is a thoroughly theoretical being, whose
essence it is to implement a specified methodology of interpretation. In so doing, IDA is
purported to provide a model for a certain kind of ability or competence that we actual
Naturalizing Idealizations. Page 11.
interpreters appear to have. However, the methodology in question has precious little to do with
the actual methods of field linguists or translation-manual constructors. The point of this
methodology is to make manifest a way to view the sorts of concepts that we apply essentially in
descriptions of agents and thinkers. The relation of the methodology of ideal interpretation to the
actual capacities of actual interpreters is captured in the following question: could we, if we
possessed the knowledge about some person expressed in IDA's theory, plausibly be said to
understand that person? The issue here is how the interpretivist's proposed account of the nature
and point of psychological attitudes and linguistic meaning, as expressed in the constraints on
ideal interpretation, is tested against folk-psychological practice. In so far as IDA appears
capable of coming up in a given case with attributions and ascriptions that harmonize with those
of actual interpreters, this provides support for the view of the nature and point of these
attributions and ascriptions that the interpretive strategy is devised to make explicit.
The interpretive strategy is intended, then, to tell us something about how we should think about
what it is we are doing when we engage in psychological attribution and semantical ascription.
It should be noted that on this construal of its theoretical point, IDA's methodology has no
particular normative implications at all, even implicitly, for us actual interpreters, eager, as we
ever are, to improve our understanding of our fellows. It may turn out for some characterization
of IDA that the conclusions drawn on the basis of the evidence we allow end up diverging from
what we should want, intuitively, to say about the subject of the interpretation. In that event, and
to that extent, the relevant specification of ideal interpretation would lose its point. It would
cease to play a useful role in our attempt to illuminate the vocabulary of thought and action.15
15
Davidson typically does not put the point exactly as I have done here. Still, I think what I
have just said captures the role of Davidson’s notion of radical interpretation, pace Fodor and
LePore (1992, 1993a). Fodor and LePore attack Davidson for assuming that the methodology of
radical interpretation will work. Searching through Davidson’s writings, they find proffered
therein no reason to believe that radical interpretation is possible. Davidson’s reply (1993b, 77-
Naturalizing Idealizations. Page 12.
IDA will be idealized in several ways, of which the following are among the more conspicuous.
For one thing, IDA will be cognitively idealized; IDA’s ability to construct and modify explicit
theories in light of evidence, and to assess their relative empirical merit, their adequacy to the
evidence, is unencumbered by the contingent characteristics that keep actual theorizers from
contemplating in principle available alternatives. Further, the evidential base for IDA’s
theorizing is one no actual interpreter could ever rely on. Not only will IDA observe everything
subjects of interpretation do, including, of course, the noises they make, and the environing
conditions of all this activity; IDA will also have access to the behavioural dispositions of
interpreted subjects. That is to say, for purposes of theory-construction IDA is assumed to be
able to appeal to the truth-values of counter-factual conditionals of a kind that actual interpreters
would have to treat as untested predictions. Finally, IDA is ideal in being without
preconceptions, both as to the semantic value of particular vocables, movements or inscriptions
84; 1995) is to the effect that they are reporting on the results of a wild goose chase, one they are
misled into embarking upon because they seriously misconstrue the dialectic of the
interpretivist’s strategy. In particular, they badly misdiagnose the relation between ideal
interpretation and actual interpretation, as I have just set it out. What an interpretivist is
committed to demonstrating, is that the explicit methodology of ideal interpretation (whether
Davidson’s radical interpretation or some other idealization) tends to end up with just the stateand meaning-attributions we ourselves think are appropriate in given circumstances. Put another
way; we may innocuously assume that radical interpretation is possible, as long as we do not
simply assume that radical interpretation is right interpretation. Interpretivists may frame this
issue in different ways, depending on context and purpose. Coming at it, as it were, from the
left, one may determine that the specified methodology of ideal interpretation will yield some
interpretation or other, and then set out to investigate whether the interpretation thus arrived at is
one the folk would approve of. Approaching, alternatively, from the right, we may—less
transparently, perhaps—assume that the outcome of ideal interpretation must be just what we
would want it to be, and then go on to wonder whether thus successful interpretation, as
governed by principles and constraints specified by the theory, is in fact possible. Perhaps it is
fair to say that Davidson has at different times come at the issue from either direction. What he
has not done— what an interpretivist cannot do—is assume both that ideal interpretation (as per
specified constraints and principles) is possible, and that ideal interpretation yields the right
interpretation. Someone who made both these assumptions simultaneously would have no need
for arguments.
Naturalizing Idealizations. Page 13.
produced by the subjects, and as to the particular details of the subjects' intentional relations to
the world.16 This last point we might put by saying that IDA has no initial view of the particulars
of the pattern of truth-preferences that are distinctive of some arbitrary subject of interpretation.17
What IDA must have, however, is a view of certain general features of any such pattern of
preferences; IDA must operate with certain desiderata that any set of attitude-ascriptions should
conform to. Otherwise, the idealized observational access to the subject's behavioural
dispositions and their contexts would do no good, because nothing would constrain the
inferences IDA may draw from that evidence. There would be nothing in particular that the
“evidence” could be counted as evidence for, and so it would not be evidence at all.18
The central task for the interpretivist is to make explicit the empirical methodological constraints
under which IDA is to deliver her specifications of meanings and attitudes. Specifically, the
interpretivist must characterize those general features of truth-preference patterns that allow IDA
to see observed events as evidence for particular theories of meaning and belief. This
16
Davidson follows Quine and characterizes this last idealization— the ignorance-condition, as
we may call it—with the adjective “radical”. I think it is useful to emphasize also other
dimensions of idealization involved in the construct which embodies the methodology at the core
of the interpretivist’s position. Hence my relabelling of what is essentially Davidson’s construct.
17
By changing the crucial meta-attitude to be identified by the radical interpreter from that of
holding-true to preferring-true, Davidson (1990a) explicitly unifies decision-theory and truththeory in a single interpretational enterprise. Since attitudes of all sorts can be nailed down in
terms of truth-preferences between pairs of indicative sentences, the formal constraints we
impose by modelling a theory of meaning on a Tarskian truth-theory now force structure not just
on the semantic theory, but on the ideal interpreter’s account of everything she ends up taking as
actions. A Tarskian truth-theory gives shape not just to the interpretation of language, but to
interpretation more generally; it makes possible both interpretations of words and individuation
and attribution of attitudes. See also Davidson 1980b, 1995.
18
The claim so far is only that without some general constraints on the pattern of preferences
that IDA observes, observation would be useless for purposes of theory-construction. I shall
arrive at a somewhat less bland claim in Section VII; in the absence of such desiderata IDA
would not only not be able to construe observed behaviour as evidence for a theory, she would
not be observing behaviour, of any sort, at all.
Naturalizing Idealizations. Page 14.
characterization is what displays the view of the vocabulary of thought and action that the
interpretivist recommends. It must, on the one hand, serve the naturalizing motivation for the
pragmatist's deployment of the interpretivist strategy, while securing, on the other, convincing
results when put to the test by means of IDA. An initial characterization might be: IDA must
structure her descriptions of the actual and possible events that serve as evidence in accordance
with the pattern of reason. The suggestion here, familiar from the writings of Dennett (1987) and
Davidson (1984), is that a subject’s perspective on the world revealed by interpretation
inevitably emerges as a rational one. The point of the suggestion is this. What it is to be a belief
or other psychological attitude is to be a state in a network of states that allows us to see a
significant segment of the behaviour of some creature as manifesting a rational orientation to its
environment. According to this position, attitude attribution discloses a point of view on the
world, the particular nature of which is traced by those ascriptions.19 By the terms of this point
of view, some subset of its occupier's causal transactions with her environment are seen to serve
intelligible purposes. That intelligibility is what gives the attitude-scheme its value—to serve our
predictive needs, as Dennett (1981a, 1981b, 1991a) emphasises, perhaps predictive interests of a
particular sort, as Davidson (1991a) hints. I will be revisiting the important connection between
interest and content ascriptions at several points throughout the paper.
19
This way of characterizing what IDA reveals is intended to emphasize the gratuitousness of
the idea that the interpretivist strategy leaves out of account, or may place one at odds with, a
bona fide first-person perspective. For a vivid exposure of the superstitions that give rise to the
gratuitous idea, see Bilgrami (1992, particularly 49ff and 225ff). “It is true,” however, as
Bilgrami notes,
that an externalism may be insensitive to the right constraints, in which case externalism
would indeed be guilty of the charge of failing to capture the agent’s point of view and,
therefore, failing to get right what the agent really believes. (237)
The right kind of sensitivity is ensured by the demand that all concept-fixing and contentassignment is governed holistically by the totality of such assignments given by some
interpretation.
Naturalizing Idealizations. Page 15.
The essence of this view is that in so far as we are dealing with creatures (or machines, or what
have you) qua agents, the better theory simply is the one that better rationalizes behaviour. Of
the many theories that could be made to account for the evidence, the optimal theory for IDA
has the subject(s) less beset by irrationalities than do alternative theories.20 This is to assert
an unrepentantly rationalistic version of the methodological constraint on ideal interpretation,
one we might therefore label the Rationality Maxim. It will be important to keep in mind, as we
assess objections to interpretivism, that RM has what we may call global scope. That is to say,
IDA relies on RM to choose between candidates for total theories—or, in anticipation of a later
distinction, for total accounts. Just because it constrains theory-choice holistically, RM governs
the interpretation of any particular utterance or movement only in an indirect, mediated way.
The kind of rationality-judgements we will require IDA to be guided by are going to be over-all
judgements of the global state of subjects captured or characterized by various candidate theories
or accounts.
At any moment or stage of interpretation, then, RM constrains the simultaneous attribution of the
entire gamut of intentional attitudes: according to it, IDA will endeavour to make a subject, at a
time, believe what is true and cherish what is good, dread the terrible and yearn for the lovely
(cf., eg., Davidson 1970, 222). The demand imposed by RM is not only a demand for
consistency among a subject's beliefs and attitudes, and for coherence among the subject's means
of describing the world. Rationalizing a person by RM, IDA will seek to have the subject prefer
true the right sentences—that is, just those sentences which, as IDA interprets them, the subject
ought to prefer—and to prefer them, moreover, for the right reasons. Aiming for global
rationality will not single out a class of attitudes, such as beliefs regarding matters of fact, rather
20
Speaking more strictly, we should refer to the set of optimal theories, to allow for different but
empirically equivalent ways of describing the evidence. Complications are introduced in Section
IV.
Naturalizing Idealizations. Page 16.
than, say, matters of method or matters of value. There is no fact-value gap nor truth-method
gap in ideal interpretation. And since noises are speech only when situated in a general context
of agency, having subjects prefer true the right sentences IDA must also have them do the right
thing. In short: applying RM, IDA insists, as far as possible, on her subject's cognitive and moral
perfection.21
21
Bilgrami offers the following constraint as the corner-stone of his brand of externalism:
(C): When fixing an externally determined concept of an agent, one must do so by
looking into indexically formulated utterances of the agent which express indexical
contents containing that concept and then picking that external determinant for the
concept which is in consonance with other contents that have been fixed for the agent.
(1992, 5)
Constraint (C) emphasizes the hermeneutic dependence of any concept-fixing clause offered by
the interpreter on the attribution of contents to the subject of interpretation. This is the heart of
the difference between Bilgrami’s view of content and those associated with Kripke and Burge,
his principal targets. It forms the basis for Bilgrami’s trenchant critique of “orthodox”
externalisms, views which are not species of the interpretivist strategy. This polemical
orientation is one reason why the relation between Bilgrami’s proposal and the issue I frame in
terms of the contrast between humanitarians and rationalists (Section III) is not transparent.
Bilgrami takes (C) as an alternative to the humanitarian proposal that we seek to minimize
inexplicable error. Certainly, Bilgrami’s Constraint (C) differs from Grandy’s proposal at least
in so far as it guides the interpreter’s selection between alternative descriptions of the objects to
which some (indexical) concepts of the interpreted agent is being linked, even in cases where no
contemplated alternative would result in the imputation of a false belief. (Bilgrami 1992, 7-8,
237) Bilgrami is surely right to insist that the role of the interpretation-constraint cannot be
limited to guiding or restricting the attribution of error. Nevertheless, it seems to me that it is
possible to raise the issues I will pursue (sections III and IV) within the context set by (C).
Bilgrami, following the line of thought of Quine (“occasion sentences”) and Davidson (“the
simplest and most basic cases”), takes utterances containing indexicals as entry points for
interpretation. He also insists against orthodox causal externalism that even indexically-based
assignments are holistically constrained; “there are no unmediated causal links” between
environment and contents (9), no content-establishing causal links unmediated by other contentattributions. Precisely this important point, however, requires us to say something about the
nature of the pattern that the interpreter is to make explicit beyond its internal coherence. The
notion of ‘consonance’ invoked in (C) is to be understood in terms of formal and material
inference relations; the constraint is free of empirical psychology. Constraint (C) guarantees the
coherence of the pattern of inference relations constituted by the beliefs composed of the
concepts attributed to an agent; it thus emphasizes the demand that we find the subject’s means
for describing the world coherent with her view of the world. RM also demands this. But I do
not see that (C) by itself tells us—as RM does, and as principles invoking agreement, truth, or
explicable error have all been intended to do—how to stabilize attributions of attitude-contents
Naturalizing Idealizations. Page 17.
A natural and frequent objection is that this way of characterizing the methodology of the
interpreter must be wrong, since people patently are not impeccably rational, as this injunction to
IDA appears to be presupposing. We need, this line of criticism has it, to let various well-known
impediments to such perfection come to expression in IDA's methodology. Different versions of
this challenge have been directed at interpretivists since Grandy (1971) criticized Quine's (1960)
deployment, in his account of radical translation, of Wilson's (1959) principle of charity. This
kind of objection is directed at the content of RM, and not (so it seems) at interpretivism per se.
After all, interpretivism might fly with a more modest (or, as I shall say, with a weaker)
methodological principle at its core. In Sections III and IV below, I attempt to defang a careful
statement of this position—which, paraphrasing Grandy, I will call the humanitarian position—
as offered by Føllesdal (1982a). After elaborating the humanitarian view, I argue for two distinct
claims. Firstly (Section III), interpretivism cannot depend on a psychologically qualified maxim
with a weaker rationality-demand than that of RM and still serve to explicate the point and
function of our vocabulary of psychological and semantic ascriptions. If we were to adopt the
humanitarian position we in fact give up on interpretivism as a naturalizing strategy. In making
this point, I will elaborate and make use of the Rortyan notion of a vocabulary. Secondly
(Section IV), contrary to arguments underwriting humanitarian claims, RM-based ideal
interpretation can indeed be squared with the considered preferences of actual interpreters among
competing ascriptions. Accommodating RM to such preferences yields, I argue, a view of
meaning which is radically contextualist, and attractive to pragmatists on independent grounds.
In Section V, I emphasize the distinctively Nietzschean nature of this notion of the contextuality
of content by contrasting it with the 'locality-of-meaning' thesis advanced by Bilgrami (1992).
on the basis of the available evidence firmly enough to allow a reasonably determinate fixing of
holistically determined concepts. In the dialectic I set up (Section III), the question of how to
achieve that stability is just what distinguishes the rationalist and the humanitarian positions.
Naturalizing Idealizations. Page 18.
Even if humanitarian objections can be met or deflected, a further worry is highly salient in the
present context. It is hard to see how an unabashed appeal to “the norms of reason” of the sort
issued by way of RM to our ideal interpreter could sustain any serious naturalistic ambition.22 I
hope to make this less difficult by elaborating, first, the pragmatic nature of the conception of
reason that informs the interpretive strategy; and second, a pragmatic conception of what
naturalization demands. My point of departure (in Section VI) is an objection directed explicitly
at the interpretive strategy itself. However we formulate the maxim that guides IDA—whether
charitably (RM) or humanistically—we still secure some degree or other of rationality for
psyche-endowed creatures by philosophical fiat. But surely, the objection goes, how good we
are at thinking, and how well we act, must be empirical questions. Fodor (1987) has claimed
this, though with more passion than argument. Stich (1981, 1990) has pressed the point by
drawing on the plausibility of certain kinds of research programmes in cognitive psychology. I
argue that the interpretivist begs no question of empirical import with respect to the quality of
our cognitive capacities. Continuing to use Stich as a foil, I go on to suggest that doubts about
the interpretivist's commitment to naturalism arise in part from a conception of rationality—and
thus of the claims the interpretivist makes about rationality—that interpretivists need not, and
should not, buy into. Then, in Section VII, I claim that the antireductivism of the interpretive
strategy is incompatible with naturalism only on certain metaphysical assumptions. These
assumptions are directly challenged by Rortyan pragmatism. For the pragmatist, irreducibility
emerges not as a reflection of a metaphysical gap, but as an ontologically innocuous reflection of
the divergent human interests that vocabularies serve. What needs naturalizing, I suggest, is not
this or that descriptive practice, but philosophy. In the concluding section (VIII), I draw some
consequences of my Nietzschean version of interpretivism for our conception of philosophical
22
This is a worry Rorty has expressed in conversation; it is exactly the worry that I, on behalf of
Reformist Rortyans, should like to alleviate.
Naturalizing Idealizations. Page 19.
reflection. The conclusion I offer runs against the impression that Rorty sometimes creates; we
can stake a claim for the irreducible distinctiveness of reason and for philosophical reflection
without betraying Rorty's pragmatic naturalizing of philosophy.
Naturalizing Idealizations. Page 20.
III. Why pragmatists are not humanitarians.
Ideal interpretation minimizes the unintelligibility of the agent. Thus stating the obvious, this
flexible formulation may seem to suggest a final convergence of various proposed versions of
principles of humanity and charity (see e.g., Evnine 1991).23 It seems compatible, for example,
with the much fuller version of the empirical constraint on interpretation that Føllesdal (1982a)
has given. Føllesdal provides a clear, characteristically circumspect examination of the status of
rationality as a constraint on interpretation, and suggests a formulation that well captures the
point of humanistic qualifications of charity. When ascribing attitudes to persons, Føllesdal
suggests,
on the basis of observation of what [they do and say], do not try to maximize...rationality
or...agreement with yourself, but use all your knowledge about how beliefs and attitudes
are formed under the influence of causal factors, reflection, and so forth, and in particular
your knowledge about [their] past experience, [their] various personality traits, such as
credulity, alertness, reflectiveness etc. Ascribe to [them] the beliefs and attitudes you
would expect [them] to have on the basis of this whole theory of [persons] in general and
[individuals] in particular. (1982a, 315-16)
It might look like Føllesdal's catalogue of explanatory avenues could serve as a specification of
the general demand to minimize unintelligibility. However, this apparent convergence of
humanity and charity is illusory. The issue between rationalists and humanitarians remains
sharp.
Rationalists (I stipulate) endorse RM; what counts as justifying the attribution of a content and
an attitude toward it is to assign to that state a location in a pattern of intentional states in just
23
Charitable proposals typically suggest maximizing agreement or truth. (See Føllesdal (1975)
for incisive criticism of this idea.) Humanitarian counterproposals typically insist on
modifications which are alleged to bring the attributive theory in line with what, intuitively, is
the perspective of the agent under interpretation.
Naturalizing Idealizations. Page 21.
such a way that the pattern as a whole minimizes the irrationality of the agent. That, says the
rationalist, just is what it is to capture the first-person point of view of some agent. It may
certainly happen in a particular case that imputation of a bad attitude (a false belief, a strange
desire) represents a better interpretation than does a good one, but that is because the
preservation of local truth or goodness sometimes detracts from the global preservation of virtue
for which IDA, according to the rationalist, must strive.
Humanitarians, by contrast, do not think that we make an agent's perspective on the world
manifest by minimizing the irrationality of the subject. They want to introduce constraints on
interpretation that modify or qualify the injunction to minimize irrationality; they believe that the
justification of an attribution may be secured by reference to psycho- and socio-biographies, or
to physiological theories and histories. With Føllesdal, they think that empirical theories of how
actual human beings typically (mis)perceive, (fail to) reason, and (inefficiently) act—theories
embodying causal generalizations that may be invoked in support of non-rationalizing
attributions—can serve to make agents intelligible.
In taking this view Føllesdal stresses that theories of persons must be based on the recognition of
"rationality as a second-order disposition" (1982a, 316), and so he endorses the view that action
explanation is constitutively related to rationality. As he says,
just as our theory of explanation of action must have room for...deviant phenomena, so on
the other hand the classification of something as deviant, as rationalization, as repression,
sublimation, etc., is possible only on the basis of such a theory of how actions should be
explained. (1982a, 310)
Nevertheless, Føllesdal takes this dependence of causal explanations on rationalization to be
compatible with the view that explanations that do not serve over-all rationalization, but are
grounded in empirical psychological theories, may be invoked in justification of an
Naturalizing Idealizations. Page 22.
interpretation.
The intuitions underlying the humanitarian view of ideal interpretation seem compelling. Since
we, as a matter of undeniable empirical fact, fall far short of our ideals of wisdom,
circumspection, integrity, insight, and so on, this ought to be reflected methodologically in our
conception of ideal interpretation. Granted, interpretation of persons must avoid that admittedly
ill-defined limit where we have too much causal explanation and insufficient rationalizing going
on to speak coherently of psychological states, but within this boundary causal constraints are
not only possible, they are virtually self-evidently required. What we demand of a theory
assigning psychological states and semantic values is that it captures an agent's perspective of the
world. Since much thought and action is governed by irrational and non-rational influences, a
methodology of interpretation that construes us as though we were perfectly rational is less likely
to produce the right theories than is one that explicitly takes our common short-comings into
account. When we try to articulate some agent's point of view on the world, generalizations that
bear on the nature and formation of that point of view are clearly relevant. They must therefore
be built into the methodology of the interpreter, modifying the assumption of rationality.
Nothing in this picture seems explicitly to challenge the idea that the items assigned by the
theory are in fact individuated by the theory. The humanitarian seems entitled to the claim that
he is offering a version of the interpretive strategy. Why should we reject it? If we assign IDA a
principle weaker than RM, such as Føllesdal's, we will foil the interpretivist's aspirations to offer
an account of agency that is at once both non-reductive and naturalistic. The remainder of the
present section is my attempt to make good on this claim.
The naturalizing potential of the interpretivist strategy rests in significant part on what Davidson
calls "a bland monism." (Davidson 1970, 214) It is monistic, because it denies the dualist's
Naturalizing Idealizations. Page 23.
thought that there are two ontological kinds; mental and physical. It is bland in a somewhat
peculiar sense; it also denies the reductivist or eliminativist thought that there is one ontological
kind of a sort to which our various ways of talking may stand in questionable relationship. The
pragmatist thus takes the lesson of Davidson's (1970) argument for anomalous monism to be that
we need not worry about the ontological priority of kinds of description, but only about their
relative utility for specific purposes. Indeed, the naturalistic pragmatist encourages us to retreat
altogether from ontology, advocating a view of language that simply leaves no room for it; the
world causes our noises to mean what they do—by way of the complicated patterns of similarityjudgements that we endlessly interacting noise-makers are disposed to produce.24 Reference, on
this view, comes dirt cheap; a greater or lesser capacity for connecting us with what is really out
there will not be what distinguishes one descriptive practice from another. We may, I suppose,
still think of philosophical reflection as an attempt to illuminate what there is; but this cannot be
construed as a matter of gauging the relative referential success of various descriptive practices.
It becomes, rather, a matter of providing characterizations of the interests we have in referring to
items of this or that sort.25
It is with respect to differences of such descriptive interests that we distinguish
24
Cf., Davidson 1982, 1986a, 1989c, 1991a, 1992, for the evolution of this view. Jennings, in
the preface to his remarkable account of disjunction, expresses a commitment to pragmatic
naturalism in the philosophy of language thus:
If an instructively oversimple slogan were to emerge from my efforts and be offered as
amicular advice to discourse generation researchers, along the lines of the earlier ‘Don’t
ask for the meaning; ask for the use’, it would be ‘Mainly we emit sounds’. (1994, ix)
25
Even commentators with great sympathy for Davidson’s views (eg., Farrell, 1994) think
Rorty’s retreat from ontology is a retreat from the constraints of the world. I hope my reading
makes evident how wrongheaded this is. Rorty, following Davidson, takes thought to be a
natural capacity of some worldly creatures. It is only in a world filled with the kinds of things
we generally think and talk about that thinking and talking could emerge as natural coping
strategies.
Naturalizing Idealizations. Page 24.
vocabularies.26When I insist on distancing the notion of a vocabulary from the concept of
translation, it is not because, at least not just because, I happen to have fastened on a certain kind
of use of ‘vocabulary’ as paradigmatic. Whatever other senses we can plausibly give the notion
of a vocabulary, the one that I characterize, which has nothing in particular to do with
translation, is in any case central for many of Rorty’s purposes. Brandom (2000) is absolutely
right to suggest that for Rorty, a principal virtue of the ‘vocabulary’ vocabulary (as Brandom
dubs it), is that it provides a way of designating discursive bodies that completely incorporates
Quine’s dissolution of any principled distinction between semantical and empirical
commitments, as well as Davidson’s devastation of the thought that the idea of a conceptual
scheme is a philosophically interesting or fruitful one. What motivates Rorty’s use of the
concept of a vocabulary, is his thought that it may bring us closer to a philosophical vocabulary
within which we may still the ontological urge, the urge that leads us to engage in projects of
26
The concept is ubiquitous in Rorty’s writings (see Brandom 2000b). I can think of only two
places, however, where Rorty considers the individuation of vocabularies (1989, 7fn; 1998a).
Rorty (1989, 7fn) and Brandom (2000b) regard a vocabulary as something that is suitable for
translation. In what follows, I diverge. Certainly there is a sense of ‘vocabulary’ which fits this
characterization, for example when we talk contrastingly of the vocabularies of Aristotle,
Newton and Einstein. But I think that even in these cases, the sense of ‘translation’ is derivative.
When we, as I do in this paper, speak of the vocabulary of mind, of the attitudes, of psychology,
biology or physics, or when we speak of the vocabulary of norms, or of virtue, or of rights, or of
New Age Spiritualism or Mahayana Buddhism, it sounds to me odd to say that we are thereby
picking out suitable objects of translation. Certainly, we can translate things uttered in such
vocabularies, but that is because they are utterances, and so bits of language, and hence
translatable. Indeed, I don’t find it the least bit odd to think that successful translation may be a
sign of a shared vocabulary—just imagine a foreign New-Ager equipped with a pocket
dictionary touring hip desert towns in Arizona. Conversely, we can be brought to realize that we
are encountering some vocabulary we do not know when translation bogs down unexpectedly.
What we then require is only in a derivative sense translation. Think of the bilingual professor
translating from her copy of Kritik der Reinen Vernunft being asked by the eager but Kantless
student to translate what she just said into “ordinary” English. What is really being asked for,
what we really need in such cases, is an interpretation of the relevant practice, one which
explains the norms governing the use of the terms by making clear to us what the constitutive
commitments are—and thus telling us what the practice is.
Naturalizing Idealizations. Page 25.
ontological legitimation. The concept serves this purpose precisely in so far as it allows us to
pick out discursive structures in a manner that precludes any attempt to restore an ontologically
potent form of the distinction between what we talk about and how we talk about it. I worry that
to think of inter-vocabularic relations principally in terms of translation is to think in a way
which may place all but the most self-consciously Quinean among us at odds with this purpose.
The point of any vocabulary can be explicated only relative to the specific goals, needs and
interests of its users or potential users. As is the case with other kinds of tools, what makes a
vocabulary the particular vocabulary it is just is the particular manner in which it serves the
needs and interests it serves. However, the relation between vocabularies and their uses differs
from the relation between tools and their purposes in an important respect. Just as vocabularies
cannot be individuated independently of the interests they serve, so these interests cannot be
stated without employing the vocabulary. When we articulate the goals or purposes that give
point to a vocabulary, then, we are offering an individuating characterization of that vocabulary,
and making such a proposal is not distinct from providing a general description of the kinds of
objects to which the vocabulary refers.27
When we claim to be characterizing a vocabulary, we thereby claim to be giving a basic account
of some set of concepts. That is to say, we claim to be offering reasons for thinking that the
interests we invoke, the concepts we analyze, and the manner of the analysis, all are linked in
such a way that to use a different kind of concept would, eo ipso, be to serve different kinds of
interests. Claiming to offer a basic account, in this sense, is not to rule out the possibility of
there being—or coming to be—systematic conceptual relations between the vocabulary one thus
specifies and other vocabularies. Rather, it is to insist that such conceptual relations will not
27
To say this, it is not necessary for me to deny that we can sometimes characterize those
interests in another vocabulary. But such characterizations are parasitical, in the sense that we
rely on the vocabulary we are thereby evaluating to identify the interests we characterize.
Naturalizing Idealizations. Page 26.
provide a way for us to keep the interests as is and drop the concepts, in favour of those of some
other vocabulary. If we decide to say, for example in a case where one explanatory paradigm
replaces another in some area of enquiry, that this is actually what has happened, the conclusion
to draw would be that our earlier conception of the vocabulary in question stood in need of
revision; we had not fully grasped what we were talking about. What we were working with was
a pseudo-vocabulary; in so far as we had obtained a philosophical analysis of the vocabulary, it
was one in which interest and concept turned out, in retrospect, not to be well matched.
Vocabularies are as enduring as interests are, which means that some will be highly transient,
and others may be impossible for us to get by without. Like interests, they may be nested,
contested, and individuated at cross-purposes. Further, we must not suppose that intellectual
history will yield categorical diagnoses; emerging conceptual connections between vocabularies
may lead to better, perhaps more comprehensive, accounts of vocabularies and interests, or they
may indicate changes in interest, or themselves cause changes in interest. What may appear to
one historian as the emergence of a better characterization of a vocabulary will to another appear
as the abandoning of a set of goals in favour of another set. Such messiness tends to increase as
historical distance decreases, approaching the chaotic at the limit constituted by the present.
Specifying interests, moreover, is itself an interest-governed enterprise—when we invoke
vocabularies in our descriptions of social or intellectual evolution, no perspective is possible that
is not laden with normative commitments. Similarly, any philosophical characterization of a
vocabulary, staking a claim for the basic nature of some set of concepts, will involve a stipulative
element. It will embody a proposal for conceiving of our interests in a certain way, a plea for
seeing them that way and for assigning them a certain weight. The notion I am characterizing is
essentially a hermeneutic one—vocabularies are never neutrally described, and they are never
fully given.
Naturalizing Idealizations. Page 27.
Quite evidently, then, this heuristic notion of a vocabulary, pegged to the notion of interest,
needs to be handled with some care. Nevertheless, it will presently serve a useful purpose. It
provides exactly the right perspective on the interpretive strategy; this strategy is an attempt to
make a case for a characterization of a vocabulary. As such it offers an account of a set of
concepts, links the analysis of the concepts to certain interests, and holds the account thus
offered to be a basic one, in the sense I have just characterized. Such a project cannot
accommodate humanitarian modifications of ideal interpretation, as I will now argue.
What is distinctive, Davidson proposes, about "accounts of intentional behaviour" is that they
"operate in a conceptual framework removed from the direct reach of physical law by describing
both cause and effect, reason and action, as aspects of a portrait of a human agent." (1970, 225)
Now, this is a claim that the interpretivist strategy is designed to preserve. As a constitutive
account of a vocabulary of action, it aims to portray the rules governing the concepts of that
vocabulary just so as to ensure the removal from law that Davidson speaks of. The interpretivist
strategy does exactly this when it offers us a view of these concepts whereby the very feature
that gives them purchase on persons, free agents (as we redundantly say), at the same time
renders them unsuitable as predicates of empirical law. A point of portraying concepts as
governed holistically by rationality-considerations is to deprive those concepts of the particular
kind of stability which empirical theorizing requires of its predicates; to the extent that some
putative empirical generalization links psychological concepts in a way that is at odds with the
norms governing them, to that extent the content of the generalization itself grows wobbly. This
is just the feature of the concepts of the vocabulary that allows us to see ourselves and others as
agents. What makes the vocabulary that Davidson aims to characterize the vocabulary it is, is its
constitutive relation to agency.
Naturalizing Idealizations. Page 28.
Hence, when Davidson concludes that "[t]here cannot be tight connections between the realms
[of the mental and the physical] if each is to retain allegiance to its proper source of evidence"
(1970, 222), he is not just expressing a theoretical observation, he is expressing the very point of
the rationality-constraint in ideal interpretation. That constraint is the centre-piece in a proposal
which purports to make sense of agency by linking it constitutively to concepts that are identified
exactly so as to cut across bodies of empirical, nomological generalization. The crucial point
here is that this tight connection between particular interests and particular kinds of norms for
application of concepts is what allows us to speak of a distinct vocabulary. It is only by virtue of
its claim to offer an account of a distinct vocabulary, one incorporating the essential concepts of
thought and action, that the interpretivist strategy can hope to provide a basic account of those
concepts. This, in turn, is exactly what enables pragmatists to say that there is no further
question of what intentional states are than what the interpretive strategy reveals.
It is this claim to be offering an account of a distinct vocabulary that the humanitarian version of
the interpretivist strategy scuttles. On Føllesdal's view, holistic theories of persons
hermeneutically balance causal psychological hypotheses and rationalizing interpretations in an
attempt to account for all the behavioural evidence there is. Now, it is true that the balance has
to be tilted toward rationalizations, otherwise, Føllesdal insists, any talk of psychological states
loses its point. But within the theory, given the tilt, causal explanations are not subsidiary to, or
derived from, or dependent for their meaningfulness on, rationalizing hypotheses in any sense
other than that all elements of such a theory depend for their content on each other. This Quinean
holistic interdependence does not prioritize any element over another, and so it is equally true
that in Føllesdal's conception, while rationalizing interpretations must dominate the theory, they
also depend for their content on the strictly causal explanations the theory invokes. The
problem, however, is that the formulation of particular empirical generalizations of the latter sort
presupposes that we have a more or less firm, more or less independent grip on the concepts
Naturalizing Idealizations. Page 29.
designating the kinds we thus link. But ideal interpretation is supposed to offer an account of
what such a grip consists of, with respect to concepts describing thought and action.
If we imagine that we could step back from the characterization of IDA and ask what the items
that interpretation reveals really are, then Føllesdal's humanitarian proposal may tempt us. For
then we could imagine that both rationalizing accounts and empirical theorizing are providing us
with indications, serving as evidence for the nature of the complex states we are trying to
diagnose. But the naturalistic pretensions of the interpretive strategy are based on a refusal to
allow a gap for ontology between vocabularies and their denotata. The interpretivist thinks that
the only answer to the question of what content-states really are is an account of the vocabulary
in which content-states are assigned. Once the question is allowed whether a vocabulary is
adequate to the items it invokes, then the interpretivist loses this answer. The alternative is to
regard the account of ideal interpretation as constitutive of the concepts applied, and hold that
there is nothing more to be said about the relation between the nature of the members of the
extensions of those concepts and the concepts themselves than what IDA tells us. If, however,
we then go on to accept that IDA may invoke empirical, non-rationalizing generalizations in
support of her theory-choice, we are giving up on our aspirations to offer, by way of IDA, a
basic account. For now we abandon the idea that the vocabulary of action is distinct from the
vocabulary (or vocabularies) of empirical law. And nobody could be misled into thinking that
the interests embodied in a vocabulary of nomological generalization could be characterized by
offering a methodology of ideal interpretation. In this case, the interpretivist strategy would not
have succeeded in characterizing the vocabulary of agency and thought after all—it would
characterize what I called above a pseudo-vocabulary. Once that is made apparent, the question
of what thought and action might really be looms once more, to be answered, perhaps, in terms
shaped by the interests that find expression in the pursuit of particular kinds of empirical theory.
Naturalizing Idealizations. Page 30.
To serve the pragmatist, the interpretive strategy must deliver a constitutive description of the
concepts of action and thought. This means that we must not build into our account of the nature
of these concepts and the interest they serve a reliance on generalizations that depend, as
empirical generalizations do, on the availability in principle of a prior identification of the kind
of states we are trying to characterize. If these considerations are sound, we have a conditional
result: if the interpretivist strategy is to have a hope of meeting both its non-reductive aspirations
as well as its naturalistic ones, it is going to have to be on the basis of RM. But of course RM
may not be defensible. It may be that it simply cannot account for the grip on the concepts of
thought and action that we language-users constantly display. To show that it can, it is necessary
to disarm examples geared to demonstrating that ideal interpretation by RM gives implausible
results. What we need is not just an argument against some particular alleged counter-example.
What we need is a general strategy directed at the root of anti-rationalist examples, which allows
us to nip the humanitarian impulse even before counter-examples may bud from it. To provide
such a strategy is the aim of the next section.
IV. On Føllesdal on urges and attitudes: how causal generalizations rationalize persons.
It will be helpful to separate two kinds of insight that typically are presumed to militate strongly
against RM-based interpretivism and in favour of the humanitarian version. The first might be
put as the claim that what can reasonably be imputed to persons must be informed by what we
know about their modes of access to the world. Such knowledge may demand, for example, the
ascription of erroneous beliefs to persons in a way that appears to be at odds with the
requirements of RM. The second arises from the indisputable point that persons are in fact less
than fully rational in their believing and desiring; we do not always believe or desire what we
should believe or desire in light of other things we hold to be true and good. These I shall deal
with in turn.
Naturalizing Idealizations. Page 31.
The first kind of consideration is emphasized by Grandy (1973).28 Grandy points out that what
we know about the way that the world impinges on us must necessarily constrain attributions to
persons of views of how things are; "the causal theory of belief, " he claims, "accords much
better with the principle of humanity than with the principle of charity."29 What seems clearly
right about Grandy's discussion of interpretation is the claim that "it is better to attribute to [a
subject] an explicable falsehood than a mysterious truth." (Grandy, 445) Without impugning
Grandy's treatment of his philosophical target, we may observe that this consideration does no
damage to the view that RM governs the theorizing activity of the idealized interpreter. For one
thing, even a rather vague injunction to maximize agreement or truth (by the light of the
interpreter) applies holistically, and may in conjunction with the empirical evidence suffice to
account for error attribution. Here the global scope of RM creates the necessary leeway.30 A
more revealing point, however, is that the main worry motivating Grandy's principle of humanity
is in fact pre-empted in the very specification of RM-based ideal interpretation. Grandy is
concerned to keep Quine's radical translator from attributing to subjects beliefs which, although
true, it is highly implausible that they would have; beliefs which could not be a part of their
perspective on things. The explicit point of ideal interpretation, however—as opposed, perhaps,
to the more limited aim of the constructor of translation manuals—is to provide a rationalizing
description precisely of some field of causal relations on which supervenes the attitudes and the
28
The same point is behind Føllesdal’s claim that “in a satisfactory theory of meaning there
seems to be no way of avoiding the study of sensory experience.” (1975, 40) Still, for Føllesdal,
the connection between an account of sensory stimulation and a theory of meaning is certainly
not direct: “Even though the impingements on my sensory surfaces may remain the same, what I
experience may come to differ, as my beliefs and theories concerning the world change” (1982b,
559). Davidson (1990b) goes further when he doubts, in criticism of Quine, that a theory of
sensory stimulation would have any role at all to play in the project of IDA.
29
Grandy’s principle of humanity says that “we have, as a pragmatic constraint on translation,
the condition that the imputed pattern of relations among beliefs, desires, and the world be as
similar to our own as possible.” (1973, 443)
30
Several people have made this point, eg, Evnine (1991), Malpas (1992).
Naturalizing Idealizations. Page 32.
agency of some subject. It proceeds on the basis of a (revisable) view of what this field is, that
is, of what the objects, events and relations are that fall under the scope of the theory. That view
will involve among other things a theory of the nature of the subject's sensory connections to her
environment. This causal theory, in turn, provides one constraint on IDA's comprehensive
theory of perceptual salience, and it will generally (but not categorically) preclude assigning to
the subject perceptual beliefs about objects with which she could not, by the theory, be enjoying
any sensory connection. This means that IDA's choice of ascriptions is indeed constrained by
causal theory. But there is no reason to think of this constraint as in any way qualifying or
modifying the rationality-demand of RM. Rather, it contributes importantly to a prima facie
delineation of the scope of the theory, by providing an initial fix on some of the items that should
be accounted for, or that may be invoked, in the ascriptive theory delivered by IDA.
The second kind of consideration is the one I want to dwell on. It is nicely brought out in
Føllesdal's discussion of the rationality-constraint. Føllesdal's point, on behalf of the
humanitarian view, is that the right interpretation of persons in many cases will rely on empirical
generalizations subsuming psychological states whose warrant is independent of their
consequences for global rationality-assessments. RM-optimality cannot drive interpretation,
since we can imagine cases where the strict application of RM gives a clearly less felicitous
result than the application of a humanitarian principle. If this is the case, then we must have
criteria other than rationality-maximization guiding IDA in selecting the theory that settles
psychological and semantic ascription. Ergo, a weaker reading of the demand to minimize
unintelligibility would be required than that delivered by RM. Admissible explanations of bad
attitudes cannot be confined, as the rationalist would have it, simply to accounts that make them
intelligible as misfirings of rational strategies and thought-patterns. While the humanitarian and
the rationalist do agree that explicable error is preferable to inexplicable truths, the contrast
between the views now emerges as a difference regarding what sort of an account of error to
Naturalizing Idealizations. Page 33.
allow, in the context of interpretation, as explanation.
Føllesdal makes his point by means of an example he attributes to Patrick Suppes. (Føllesdal
1982, 310) A young pupil with an attractive instructor very frequently comes up to the teacher
after class, to ask questions concerning schoolwork. The sincere first-person account of this
behaviour depicts it as a sustained attempt to obtain answers to questions regarding matters
academic, an attempt motivated by a desire to learn. Undeniably, though, given what we know
about persons in the throes of early puberty, it is easy to further specify the circumstances of the
case in such a way as to make the temptation to go beyond the first-person account positively
irresistible. We may soon find ourselves explaining the behaviour not by invoking the student's
professed desire to learn, but in terms of "urges adduced by the psychologist." (1982, 314)
Føllesdal elaborates this example to stress two points in particular. The first, positive, point is
that "whenever [persons] experience [themselves] as carrying out [actions], what [they] do
should be explained in conformity with the pattern of reason explanation."31 (1982, 313) The
second point is the one at issue; the case shows, Føllesdal believes, that we must reject the
"normative methodological hypothesis" that in interpreting persons "we should always try to
make [them] come out as rational as possible." (1982, 314)
31
The principle captured in the first point is precisely not intended to exclude further or
complimentary psychological explanations involving non-rationalizing descriptions of the
events. I read it not as a proscription at all, but simply as another embodiment of Føllesdal's
commitment to the thesis that all action-explanation rests constitutively on rationalizing
interpretation. For the principle is, I think, as near-analytic as the present context allows: since
the state "experiencing oneself as performing an action" takes its content, like all other mental
states, from its place in the hermeneutic circle of third-person interpretation, it is hard to see how
we could fail to conform. Once IDA has attributed a description of an event under which the
subject regards it as an action, she will have applied the pattern of reason explanation. The
relation (within the theory) between such a description and the attribution of the relevant selfunderstanding to the subject is intimate indeed.
Naturalizing Idealizations. Page 34.
Now, if we take this hypothesis to deny actual interpreters the right to support their efforts to
figure out what is going on in the minds of their fellows by way of non-rationalizing causal
generalizations, then I agree; we must reject it. If, by contrast, we take this hypothesis to be the
claim that ideal interpretation is governed by RM, then we must insist on it. This distinction is
essential. Certainly, as a principle alleged to be constitutive of a vocabulary, RM must be seen to
be compatible with our actual interpretive practices. However, this does not immediately
preclude endorsing humanity as a methodological ideal for actual interpreters—as a formulation
of the strategies actual interpreters ought to follow. What we must show is that what actual
interpreters thereby would achieve, is something IDA secures by virtue of RM. To show this is
to give non-rationalizing causal generalizations of the kind Føllesdal urges good interpreters to
invoke an integral place in a vocabulary the constitutive purpose of which is to reveal rational
agency. Sticking by RM, the interpretivist must argue that the kind of causal psychological
generalizations that we rely on to support actual interpretations can have that supportive role
only in so far as they operate, in the ideal, in the service of rationalization. If we could show
this, we would entitle ourselves to maintain what is the critical claim: psychological concepts are
the concepts they are by virtue of their making evident in behaviour patterns that conform to the
norms of reason.
Consider again the case of the pining pubescent. Føllesdal distinguishes various schematic
possibilities regarding the relative explanatory power of the first-person account and the account
couched in terms of urges. Even in the case where the first-person reasons offered "were neither
sufficient nor necessary to explain [the] behaviour [and] the urges themselves were sufficient,"
(1982, 315) the reason-explanation must be part of our account of the event, otherwise we would
not be talking about an action at all. But in this case a satisfactory explanation would also have
to include the causal psychological factors, the teen-age drives, those hormonally triggered
urges. The implied case against RM here is that it would have us exclude the urge-explanation,
Naturalizing Idealizations. Page 35.
given that a rationalizing reason-explanation is readily available. But this would, ex hypothesi,
be to opt for a bad explanation. Taking the pubescent to be acting on an acknowledged and
reasonable desire may yield an account of what the student is doing that is, by common sense,
inferior to one which has the student acting in self-deception, on the basis of unacknowledged
urges.
I do not want to quibble about this common-sense verdict. Certainly there are cases where we
quite reasonably take someone to be acting irrationally in the sense here at issue; that is, by
taking them to be acting in blindness to their own actual motives. It is true, too, that such
interpretations often will be supported by non-rationalizing generalizations. Let us stipulate that
Føllesdal's pubescent provides us with just such a case. What, in the face of such concessions,
can be said for RM?
Føllesdal implies that if we were to choose between the first-person account and the
psychologist's story, we would be choosing between a rationalizing action-explanation and a
non-rationalizing causal account of the behaviour. But there is another way to characterize the
options. For what the psychologist does is to privilege another action-explanation, along the
following lines; the pupil found the teacher very attractive, and, giving high priority to sexrelated ends, designed a way to deepen and extend their personal contact. It would appear that
viewing pubescents as weighing these kinds of ends heavily is a good way to make sense of
much otherwise puzzling behaviour. And 'making sense' in this instance means 'making
reasonable'. The effort to rationalize is still in view here. Nevertheless, RM is in trouble, in so
far as urges are invoked to justify this alternative action-description; the urge-explanation
attributes a certain propensity to rank desires on the basis of physiological properties. It trades
exactly on the kind of generalizations subsuming intentional states that appear to be thoroughly
empirical in nature, and that therefore may yield ascriptions which conflict with the norms we
Naturalizing Idealizations. Page 36.
have supposed, by RM, to constitutively govern the attributions of attitudes.
One way to dodge the problem is this. In our dealings with others, tracing in their behaviour the
pattern of free, deliberate action is only one interest among many. There is a varying, but rarely
insignificant, degree to which we care about people simply as objects. We should not assume
that the features of concepts that allow agency to emerge are features of the terms by which we
most efficiently describe and predict people for any purpose whatsoever. This raises the
possibility that our ordinary dealings with people is conducted in a hybrid of vocabularies, where
interests that may be at cross-purposes, or even directly antagonistic, find expression in different
ways. Our common-sense intuitions, therefore, may be intuitions that express habitual ways of
striking compromises between these interests. So it may be that when we invoke common sense
against RM we are smuggling in interests at odds with those that RM is expressly designed to
capture—namely the interests we have that depend on our treating our fellows (and thus also
ourselves) as autonomous and rational, i.e., as agents.
Whatever the plausibility may be of these considerations, they do not hold out much promise of a
persuasive response on behalf of RM to the concerns that motivate the humanitarian position.
This is because they straightforwardly deny an important humanitarian claim about the examples
invoked; the examples show, it is alleged, that causal generalizations play an indispensable role
in settling the content of the intentional states of the interpreted subjects. We could reply that
the very notion of content thus being invoked is itself a hybrid notion, hostage to interests at odds
with those connected with agency. This response, however, carries with it an air of stipulation so
strong that we would be entitled to suspect that the subject is being changed. Certainly it would
be an unattractive solution to anyone with pragmatist sensibilities; it is, after all, our ordinary
attributions of intentional states, whatever work they do, that we are in the business of
illuminating.
Naturalizing Idealizations. Page 37.
Granting, then, the point of the humanitarian examples, I want now to propose what I take to be a
more fruitful line of response. Accommodated within the framework set by RM, cases such as
urge-explanations will turn out have interesting implications for the nature of thought,
implications that a naturalistic pragmatist should find independently attractive. Urgeexplanation, as any explanation of a non-rationalizing kind that purports to ground attributions,
may be seen as a way of coping with a fundamental tension in any interpretation of intentional
systems; that between cohesion and scope. Roughly, the greater the number and variety of
events that a single theory has to account for, the greater is the likelihood of anomalies and
inconsistencies. When the strains are great, indeterminacy may increase up to a point where we
begin to lose a clear sense of the contents being attributed. A possible response for IDA is to
give up the unity of the theory—and with it, to some degree, the unity of the person—by
allowing separate theories to account for different chunks of evidence. One imagines that these
evidential chunks might overlap to a very large extent. The theories would differ, however, at
least in what they discount as anomalous. They have, we might say, different focal points,
around which meaning and belief are coherently rendered, focal points that will make them
useful for different purposes. Each such theory would holistically deliver a compliment of
attitudes and fix the interpretation of the agent's concepts, in accordance with RM. Such theories
can also be brought together, in an account of a person, but when they are, the mode of their
interaction would perforce have to be rendered in arational, causal terms.32
Nevertheless, we can easily imagine that the construction of such causal structures in the
theorizing of IDA is governed by RM, that is, as constrained by global rationality-requirements
32
The model here is Davidson’s treatment of mental division (cf Davidson, 1974, 1985a,
1985b). Cavell (1993) provides an interesting explication of Davidson on irrationality, and puts
the Davidsonian notion of mental division to creative use. Without blaming her for the particular
proposals I make here, I want to acknowledge a significant debt to Cavell’s account of psychoanalytic concepts in Davidsonian terms.
Naturalizing Idealizations. Page 38.
directed at the person being interpreted. There is, for example, the general demand that we
bring all relevant considerations at our disposal to bear in our deliberations. Guided by such a
norm, IDA would aim to minimize the number of consistent sets required to parcel out the
ineliminable inconsistencies in any total set of action-descriptions, semantical interpretations and
psychological attributions.33 Here a demand for intra-theoretic consistency works in conjunction
with a demand to keep at a minimum any non-rational inter-theoretic interaction in accounting
for the sayings and doings of a single subject. Together, these demands—norms of reason—
constrain the structure of the total account of the person arrived at by IDA.
What I have just now done, is introduce the idea of interpretational types. We actual interpreters
rely on such types all the time, of course, often for ill in the form of pernicious stereotypes. My
suggestion is, however, that even in ideal interpretation something akin to stereotypes is
indispensable. Unlike us actual interpreters, IDA will have an ideally flexible range of such
types, around which the various theories that make up the elements of the account of a person
will be organized. For IDA, these types will be precipitated out as a result of the pressure
imposed by the norms of RM on the behaviour of creatures like us.
Having introduced this notion of interpretational types, we must face the questions of what it is
that is supposed to fix these types, to give them application to particular agents, and to allow
IDA to keep track of their roles in her account of some agent. These questions point to the space
that has now emerged for the kind of causal non-rationalizing generalizations that Føllesdal
suggests must be brought to bear in interpretation. What such bodies of generalization do is
designate an interpretational type, a particular perspective on the subject of interpretation, and
anchor it to other identifiable features of agents. Take the pubescent-type, the type that we and
33
Ray Jennings has suggested that paralogical notions provide formalizations of just the kinds of
constraints I have in mind here.
Naturalizing Idealizations. Page 39.
the psychologist invoke when over-riding the first-person account of Føllesdal's pupil; to impose
this type on a person is to structure a theory around a set of attitudes and intentionally
characterized propensities, to give this set explanatory priority, and to discount, in that instance,
evidence against it. In the case of the pining pubescent, where hormones are invoked, the type is
defined in physiological terms. But this is not essential—generalizations linking intentional
states of different kinds may similarly be called upon to support non-rationalizing attributions.
In this case, too, as for example in psychoanalytic theory, we introduce interpretational types that
cut across persons and are linked by causal generalizations to kinds of attitudes or dispositions.
What the psychologist does by treating the young student as an instance of a type imposing a
certain value-hierarchy, IDA will do to us all. The point here is perfectly general, applicable not
just to young piners in the throes of hormonal urges they do not understand. To impose the type
on some person, is to secure a set of attitudes at the heart of an interpretation, around which
much, though not all, of the subject's behaviour can be usefully interpreted—in Suppes's
imagined case as puberty-behaviour. What governs the interpretation here is still RM. What
makes the behaviour the behaviour of a person transcending the type, are the causal
generalizations invoked to identify the person as a person of that type. These generalizations will
specify the attitudes and propensities that make up the core of the type, and link them to nonintentional properties, or intentional properties given by a theory not structured around that type.
Accounts of persons now appear as networks of causally interrelated rationalizing theories of
this kind, with each theory being geared to a type that gets projected onto the biological creature.
Because each theory's particular focal point will be specified through generalizations that
subsume intentional descriptions without also expressing the norms for their application, such
ascriptive theories appear to be grounded in genuine empirical generalizations.
No-one, however, should leave with the impression that such interpretational types represent
Naturalizing Idealizations. Page 40.
empirically-discovered bodies of generalizations capable of delivering predictions of intentional
states independently of RM. Take, for example, our reliance on hormonal states in imputing
thoughts to pubescents. In the unlikely event that we were to develop ways of describing our
pubescents in terms of easily diagnosable kinds yielding greater over-all RM-conformity than
psycho-hormonal generalizations allow us to find in their behaviour, then psychological
explanations in terms of hormonally induced urges would just die off. To that limited extent the
issue between humanitarians and rationalists is perhaps an empirical one; the rationalist would
predict that when there are systematic changes in the predicates that we invoke in causal
generalizations subsuming psychological states, application of the new terms will generally yield
greater RM-conformity than the ones being replaced.34 Freudian theory is the paradigmatic
example. The causal structures of the psychoanalytic soul rendered Fin-de-Siècle neurotics
more rational, and thereby seriously modified the range of available responses to their neuroses;
the liberating potential of psychoanalysis lies precisely in this fact. The rationalist would further
hold that to show how a particular change in theory yields an increase in RM-conformity, would
also be to explain that change. But that is not an empirical claim.
The role of psychological generalizations that are not expressions of the norms of reason, and
therefore not constitutive of the concepts by which we ascribe thought and agency, is
nevertheless to implement RM. By way of such generalizations, IDA both defines the
interpretational types (cross- or sub-persons, if we like), and explicates their interrelations and
conditions of application. In Føllesdal's example, we find that the imputation of a certain desirestructure is justified by (indirect) reference to hormonal states. In other cases, we find
behaviour-explanations invoking causal relations between differently focussed attributive
theories—psychoanalytic models will provide specific examples. Explanation of action may
34
Ideally speaking, that is. But the ceteris paribus clauses hedging this prediction are made
rich indeed by socio-political interference.
Naturalizing Idealizations. Page 41.
invoke a particular theory with its defining focus, or it may invoke (causal) relations between
interpretational types imposed on a single agent. In the latter case some degree or other of
irrationality will be explicitly attributed to the agent. By building an account of a person
consisting of distinct theories of sub- or cross-personal types, each with a specified warp, or a
particular focal point, IDA accommodates the tensions in her evidential base by dividing it into
more coherent subsystems. The central point is that the operative principle for IDA is still to
make as much behaviour as possible as reasonable as possible. However, I claim, this principle
may be upheld, when we are dealing with persons, not just by minimizing irrationality and error
in a single theory accounting for all behavioural evidence. There is another axis of
accommodation of the tensions that this evidence, as a body, inevitably produces; we may limit
the scope of a theory, gerrymandering the evidence it is required to account for. We now treat
the total body of evidence, the person as a whole, as subject to bundle of variously restricted, but
causally related interpretative theories, each with a high degree of consistency and cohesion. We
thus preserve the predictive and explanatory power of each theory, but we do so by sacrificing
the psychological unity of persons.
Is this, however, really a sacrifice? Perhaps, in so far as the model of IDA's account of a person
now emerging suggests an indeterminacy to meaning and belief over and beyond the ones
familiar from Davidson's account of radical interpretation (1984a). The added indeterminacy is
significant; shifting between the theories that enter into IDA's account of an agent, we are not
merely shifting from a concern with one action to another, or some region of an agent's
psychology to another. We may find ourselves, as we move from one theory to another,
individuating actions differently.35 Not only may different theories account for the agent's
behaviour differently in the sense that they may identify some action by different descriptions.
35
This is the kind of indeterminacy that Dennett (1991a) stresses. Unlike Bilgrami, I think
semantics and intentional psychology is rife with this kind of individuative indeterminacy.
Naturalizing Idealizations. Page 42.
Theories may differ even in what they characterize as an action. The clearest examples here are
again probably psychoanalytic ones, where apparently or superficially viewed non-intentional
behaviours may be redescribed in terms of the heretofore opaque intentions of a hidden or
disguised locus of agency. But it can hardly be disputed that run-of-the mill folk-psychological
practice also provides examples of context-determined shifts in the lines we draw between the
intentional and the non-intentional behaviour of some person. The salient fact about this form
of indeterminacy, however, is that it resolves into the context-bound nature of content. Unlike
the indeterminacies of logical permutation or reference, or of the doxastic and semantic division
of labour within an attributive theory, the substantive indeterminacy of the identity of actions
disappears as soon as we allow a context of interest to trigger a particular theory (or set of
equivalent theories) amongst the possible theories embedded in IDA's account of a subject.
Precisely because this strong form of indeterminacy leaves indeterminate something that makes a
difference to practice—to how we respond to an agent, say—it is not a genuine indeterminacy. It
only appears that way when regarded from a perspective that abstracts away from the varying
contexts of interest and purpose that settle what we mean and think.
What is really at stake here, what my contextualist account of ideal interpretation puts under
great strain, is the possibility of the reification of mental content. On the model of ideal
interpretation I have proposed, the interpreter does not eliminate anomaly in behaviour. Rather
the interpreter produces a set of devices, alternative theories, which allows us selectively to
displace anomaly, deviance from norms of reason, and thus insulate behaviours or behaviourpatterns on which we may want for particular purposes to focus. The prevalence of conflict
within the evidential base constituted by the actual behaviour of any entity of sufficient
behavioural complexity to count as a person is universal. In the crucible of RM, such conflict
forces upon IDA the strategy of interpreting differently circumscribed subdivisions of subjects,
on pain of the dissipation of thought in a fog of indeterminacy. A consequence of this is that the
Naturalizing Idealizations. Page 43.
patterns of reason traced by interpretation become multiply ambiguous. Reasonably determinate
thought emerges only when an agent is interpreted as an agent of some kind, that is, in some
context, for some purpose. Hence, ideal interpretation settles content only relative to contexts
specified in terms of some subset of the various purposes, aims and interests we may have in
approaching a subject as an agent. Such defining contexts may be just what come to expression
in our characterization of the range of interpretational types by which we make sense of persons.
The aim of this section was to suggest how we can come to regard causal explanations as
drawing their content from the application of RM, and how prima facie conflicts between RM
and causal explanations disappear when we distinguish between normative principles for actual
interpreters and the vocabulary-constitutive principles of ideal interpretation. The key move of
the argument was the introduction of interpretational types as locus of the attributive theories of
IDA. Causal generalizations specify the relations between various cross- and sub-personal types,
and between such types and types specified in other terms, e.g., physiological ones. What
remains fixed is that we explain and predict what persons do by rationalizing their behaviour,
because it is only as rationalized that they act at all. What we have discovered, however, is that
this very commitment dissolves the notion of mental content into a process of alternative and
alternating rationalizing descriptions, each representing some purpose-relative perspective on a
person, a locus of agency. From the perspective of the pragmatist, campaigning for
naturalization of our conception of persons by overcoming the metaphor of inner space and the
reifications associated with the concept of mind, this should be a happy thought. Predicates
designating mental states characterize aspects of agents in contexts of interaction with others in a
shared world.
V. Bilgrami versus Nietzsche on the contextuality of content.
It now appears that notions invoked in descriptions of ideal interpretation such as 'the perspective
Naturalizing Idealizations. Page 44.
on the world,' 'the total theory of an agent,' or 'the totality of the behavioural evidence,' are
misleading. They are misleading at least in so far as they suggest that there is a single, general
perspective, defined by a general interest in agency as such, from which determinate thoughtattributions and action-descriptions emerge. There is no such perspective, nor, hence, is there
such a thing as the perspective on the world of IDA's subject. It is better to think of the
idealization toward which these terms gesture as consisting in IDA's ability to form
simultaneously an indefinite range of interpretational perspectives on some one subject, each of
which may constitute its evidence differently. For IDA, with her account of an agent consisting
of a set of causally related non-equivalent theories, there is no saying what the subject thinks or
does in general; looking simultaneously through the various theories that go into an account of a
subject, IDA would induce in herself an utterly blurring astigmatism. Determinate ascriptions of
content and descriptions of action come only when the subject is regarded through one lens or
another, that is to say, for some purpose or other. Thought and action emerge, as particular,
interest-governed interpretative perspectives on behaviour are actualized. What our thoughts are
and which actions we perform depend not only on what we do and what goes on in us and what
the world is like, but quite literally on the particular perspectives from which we actually come
to be regarded as engaging with the world. A fortiori, it depends on there being particular
perspectives; I take the actual interaction of interpreters to be a condition of intentionality.36
The claims about content-attribution which underlie this conclusion bear some resemblance to
one of the key commitments of the version of externalism that Bilgrami defends, the thesis of
'the locality of content.' I have emphasized the consequence of my view that the very identity of
content-states becomes radically contextualized, dependent on particular explanatory situations
and aims in a way that makes it interest relative and dependent on actual interpretation. Bilgrami
36
As does Davidson, though he may not have been attracted to my way of arriving at the view.
Naturalizing Idealizations. Page 45.
does not accept this (238-41), but he has argued for a similar claim forcefully and at length. It
will be worthwhile, therefore, to contrast my version of the thesis of the interest-dependence and
context-relativity of meaning with the locality-claim advanced in Bilgrami's appealing account of
content.37
Bilgrami distinguishes what he calls the aggregative level from the local level of concepts.
(1992, 10-13; 42ff) He suggests that meaning-theory, characterizing the former level, produces
clauses that together specify, "all the beliefs an agent associates with each of his terms." (143)
The concepts of such a theory, Bilgrami takes it, are useless for action-explanation. Explanation
happens at the local level, and does not, at least not directly, employ concepts as specified by
meaning-theory.
Rather, what happens is that in citing beliefs (and desires) of agents at the level of the
explanation of some piece of behaviour, we are distilling some beliefs out of the
aggregate of beliefs summarized in any given clause or clauses at the meaning-theoretic
level, and we are citing contents which are composed of concepts that are to be
understood in terms of these selected beliefs. (143)
Bilgrami presents this view as a way to meet the claim that has shaped a significant part of
Fodor's intellectual life, namely the charge that holists deprive content of explanatory power.
Fodor's complaint (1987, 1994; Fodor and LePore, 1992) can be brought out thus. The
explanatory point of a remark such as, "Because he wanted to get to Bombay, and believed that
Bombay was the flight's destination" in response to the question, "why did Bilgrami board that
plain?" trades on a regular connection between beliefs and desires of a certain kind and actions
of a certain kind. For such regularities really to be regularities, we require identity of the
37
Particularly given Rorty’s view, quoted on the dustcover of Bilgrami’s book, that, “Akeel
Bilgrami has taken a giant further step along the path broken by Quine and lengthened by
Davidson—the path to a radically naturalistic theory of meaning.”
Naturalizing Idealizations. Page 46.
predicates invoked across applications. But, since for Bilgrami (by his own admission—cf., 143)
'Bombay' is associated with what is probably a unique and unstable set of beliefs, the scope of
the generalization shrinks to the point of vanishing and the explanation evaporates. If we
endorse a holistic account of concepts, we get no explanation by attributing to Bilgrami
'Bombay'-thoughts; the concept of Bombay invoked here is useless, since, like Larkin’s ladinglist, it applied only to one man once. So holists, Fodor is convinced, have no right to feel
satisfied by psychological explanation—and what is worse; if holism is true, nor does Fodor.
Bilgrami counters with the point that the requisite generality is preserved in local contexts, where
only explanatorily relevant beliefs are invoked in the specification of the concepts that make up
the contents attributed. To the worry that we now have two different notions of content going
Bilgrami replies,
There are two notions of concepts: aggregative and local. But there is only one notion of
content because the aggregate level of concepts does not compose any contents at all.
They are trumped-up posits, only there to acknowledge a larger pool of resources from
which local concepts (which do go into contents) are selected....the only level at which
any serious work is done by the notion of content is at the local level. (144)
My problem with this reply is not what it asserts. I find Bilgrami's notion of the locality of
content congenial, and the non-reductive externalism in which it is embedded persuasive. The
problem I have is that I am not clear on the motivation for the reply, as an answer to someone in
Fodor's predicament; you need to say a lot more, it seems to me, or you need to say less,
depending on who you are out to convince. I opt for less, and am not able to see why the
following short reply to Fodor's worry will not do. Fodor thinks to himself, "When people want
to go to Bombay and believe a certain flight will take them there, then, ceteris paribus, they get
on the plane. Bilgrami desires and believes thus. That's why he got on the plane." Fodor feels
satisfied. And why should he not? What really does the explanation here? Fodor's concepts; the
Naturalizing Idealizations. Page 47.
generalization he relies on has application to all those to whom Fodor, by the evidence available
to him, attributes the requisite belief-desire combination—in his terms, at that time. Period. If
you are a holist and an anti-reductivist, and a naturalist to boot, this should do—you will be
entirely unmoved by further worries about whether I (or Fodor) mean what I (or he) meant
yesterday by those same terms, or whether, if we both agree on the explanation, we really aren't
using terms differently. That may happen of course, but the only grounds we might have for
thinking so is that the assumption of agreement leads to unexpected and puzzling uses of terms
elsewhere; the hermeneutics of ascription is open-ended. But what the pragmatic naturalists will
not allow any room for is a notion of meaning that is not fully exhausted by a rationalizing
holistic characterization of speech and other behaviour. Such a behaviour-transcendent notion of
meaning treats manifest behaviour as kind of indicator of underlying states. It opens the door for
sceptical worries about identity of contents across agents and time-slices of agents of a sort that
may remain pressing in spite of our successful prediction of speech and other action. It should
be dismissed.
This will undoubtedly seem, from Fodor's point of view, like sheer, brick-headed point-missing;
if you think reduction—or at least the clearly perceived possibility of reduction— is required for
the legitimation of causal explanation, then the above method for reinflating the scope of
explanatory generalizations will be entirely beside the point. So the short answer certainly will
not make Fodor more favourably disposed toward holism. Bilgrami, however, is no reductivist.
And what I do not see is why, if the short answer does not work, Bilgrami's does better. This is
because I cannot, even with Bilgrami's careful guidance, see my way clear to believing that we
may individuate the items that are contextually selected and activated through the local concepts,
which for Bilgrami compose explanatory content, in a manner that avoids complicity with the
full compliment of attributions expressed by theories of meaning at the aggregative level. If we
may not so individuate them, then Bilgrami's answer to Fodor is spoiled, collapsing into the
Naturalizing Idealizations. Page 48.
dismissive one I just offered.
The version of the thesis of interest- and context-dependency of meaning that I propose is in any
case not intended to provide an answer to Fodor's worries about holistic individuation of the
denotata of psychological predicates (though the paper as a whole is intended to help those not
yet firmly set in their metaphysical ways be less compelled by the kind of assumptions which
yield Fodorian worries). My concern is with the nature and justification of the generalizations
that may be invoked in ideal interpretation, and only indirectly with the problem of their scope.
But my difficulties with the two-level account, offered as a response to Fodor, is rooted in an
attachment expressing itself also in the difference between Bilgrami's locality-thesis and my
contextualism. My doubts are rooted in what may be called whole-hog holism. Whole-hog
holists think, as I still think, that any attribution of content and specification of a concept in ideal
interpretation involves projections giving maximally complete descriptions of the evidence—
under some specification. So in fixing concepts locally in terms of selected beliefs, we make
implicit reference to some such maximally complete description. The claim I stand by differs
from Bilgrami's; as a whole-hog holist, I am made a contextualist by the fact that I deny that
ideal interpretation yields any single coherent theory at all of all the beliefs associated with each
of an agent’s terms, no matter how metaphysically light-weight we make the status of such a
theory. The resulting contextualism provides me with an opportunity to accommodate the use of
what looks like straightforward empirical generalizations within what is in essence a normdriven enterprise. This contextualization yields refraction into multiple theories also at the level
of the kind of transient, super-fine-grained concepts that are the deliverances of theories of
meaning. Context, on the model I suggest, does not select doxastic items from meaning-theory,
it selects entire theories of meaning and belief.
No maximally complete description of the evidence is a definitive account of all the evidence
Naturalizing Idealizations. Page 49.
there is. Still, the notion of maximal completeness is necessary, it seems to me, to forestall a
crippling indeterminacy. However, this constraint operates, as I have stressed, within the context
of particular interpretive aims and interest. Theories of meaning may well be trumped-up
postulates, abstractions of available resources, but even such theories, such characterizations, are
selected among by IDA in settling some context of agency or other.
Since my thesis combines whole-hog holism with contextualism of content, it is in one way more
radical than Bilgrami's. It is Nietzschean, in a quite distinct sense in which Bilgrami's is not; it
denies that contextually determined meaning and belief refer back to any sort of coherently
specifiable totality of ascriptions at all.38 IDA's account, as I called it, is an abstraction; it
specifies thought and meaning only in relation to particular contexts. Still, though this
Nietzschean version of interpretivism is undoubtedly too...well, Nietzschean (some, though not
I, would say "anti-realist") for Bilgrami's tastes, the view I propose is not immediately
incompatible with the two-level structure that Bilgrami posits. Furthermore, it should be
emphasized that to contextualize meaning by depriving intentional systems of unity in the
manner I suggest, is not to deprive content of unity in Bilgrami's sense (3-4; 15-16). The point
of ascribing content to states of agents, lies in the ability such ascription secures for us to offer a
distinctive kind of characterization of behaviour. The separation of contents ascribed in
interpretation from psychologically explanatory contents is a disastrous move; unity of content in
Bilgrami's sense is essential. My Nietzschean perspectivizing of content does not damage this
unity, since context-relativity of meaning and thought pervades both semantic attribution and
psychological explanation. What we think and mean explains what we do. What there is not—
38
"No, facts," Nietzsche notoriously proclaims, "are precisely what there is not, only
interpretations." (1968a, 481) With this and other formulations of his perspectivist view of truth
(eg, 1968b, III (12)), I take Nietzsche not to be proclaiming a wild-eyed scepticism or nihilistic
relativism, but to be registering an anti-representationalist's complaint against the very idea of
ontology. Much the same could be said about my perspectivist view of agency.
Naturalizing Idealizations. Page 50.
even as an abstract, rarefied posit—is a unified subjectivity of which what we think, mean and do
are manifestations, and in terms of which our behaviour might be non-interest-relatively
categorized and explained.
VI. Pragmatizing Reason: Stich versus the interpretivist strategy.
Though I have relied on Føllesdal and Bilgrami for contrastive force in staking out my position,
the disagreements I have focussed on refer back to significant common ground. I want now to
address objections to the interpretivist strategy as such, levelled from a necessarily more distant
perspective. These objections put pressure on the naturalistic aspirations that inform my version
of the interpretivist strategy. They will provide an opportunity to elaborate explicitly the notion
of reason that RM invokes, and thus, I hope, to assuage the worries that Rorty has about the antinaturalistic implications of the language I rely on.
Stich has been a vocal critic of interpretivism, culminating in (1990) where he invokes a range of
empirical evidence for the systematic cognitive failures of human beings.39 This, Stich claims, is
incompatible with the interpretivist's commitment to the impeccable rationality of agents. Where
Føllesdal relies on the common sense embedded in actual attributions, Stich relies on science to
impugn RM. But Stich would have no truck with Føllesdal's humanitarian ersatz. He takes
exception to the idea that any normative principle could be constitutive of psychological states.
To see this, we must distinguish two elements in his critique. The first is the claim, documented
by the experiments of empirical researchers, that human beings evince systematic defects in their
reasoning.40 People just are not as rational as the interpretivist makes them out to be. So if the
interpretivist is making an empirical claim, it is false. It should be clear by now that this point by
39
See also Fodor and LePore (1993a), and Dennett’s reply (1993, 215 ff).
40
See Stich (1990, 4-9) for a quick overview of some of the results of this research.
Naturalizing Idealizations. Page 51.
itself does no damage to the interpretivist position, since it does not entail that people do not act
irrationally. The second element, however, is for Stich the crucial one. What Stich finds
objectionable is that the interpretivist is not making an empirical claim at all. If the rationality of
human beings were the sort of thing that could be established by a priori philosophical
argument, this "would make nonsense of the empirical exploration of reasoning and its foibles"
(Stich, 1990, 18). Furthermore, Stich claims, making the possibility of intentional-state
attribution hinge on the rationality of the subject also threatens the one dimension of
epistemology that he sees much point to. "It would...turn the effort to articulate and defend a
normative theory of cognition into an arcane and academic exercise of no particular practical
importance." These two allegations I will address in this section and the following one.
I am about to argue that the interpretivist strategy does not have the first of these two alleged,
and allegedly unwelcome, consequences. But what could be said of the second? Here I think the
interpretivist should bite the bullet, welcome the imputation and follow Rorty in making a virtue
of necessity. Furthermore, understanding the rationale for this response will help us bring out
the sense in which the interpretivist strategy serves naturalism. Before pursuing this point,
however, I return to the first allegation. The question is; does the interpretivist strategy run afoul
of the sort of empirical research into human cognition that diagnoses, in folk-psychological
terms, systematic propensities to err or sustain cognitive illusions?
Empirical evidence appears to show that in certain replicatable situations, faced with specific
kinds of cognitive tasks, human subjects deviate in systematic ways from what appears plainly to
be the desired result. Are interpretivists forced to take a line on this research that places them at
odds with what scientists say, and what we folk should want to say, about these cases, and the
manner in which the results are established? We may note, initially, that some disagreement
persists among theoreticians of cognitive psychology about how to characterize the results of
Naturalizing Idealizations. Page 52.
empirical research into human cognitive processing (see e.g., Manktelow and Over, 1993). So
for example Gigerenzer (1993) argues that standard cases rest on interpretations of the tasks that
are not mandatory; the appearance of cognitive illusions can be made to go away by paying
attention to the interpretation that researchers implicitly are placing on, e.g., the laws of
probability that underwrite their diagnosis.41 Johnson-Laird and Byrne (1993), arguing the other
side, spend several pages rehearsing what they take to be Stich's case against those who doubt
that empirical results can tell us how (ir)rational we are. One lesson a philosopher might be
forgiven for drawing from the persistence of such argument is that amongst cognitive
psychologists, as with Quinean radical translators, one theorist's evidence for irrationality is
another theorist's evidence for the poverty of the former's model.
Let us, however, set this lesson aside, and not engage in what Johnson-Laird and Byrne call
"heroic denial" (192) of the phenomena. It is difficult to feel anything but sympathy for their
efforts to clear away the aprioristic or transcendental arguments of armchair thinkers to
substantive conclusions about how good human beings are at cognitive processing. It is not,
after all, as if there is no historical precedent for the kind of philosophical legislation that
Johnson-Laird and Byrne impute to the "proponents of rationality." It is not good exegesis
however, to lump interpretivists like Dennett and Davidson in with this kind of aprioristic,
legislative philosophical practice. For, indeed, no question relevant to the viability of the
41
Gigerenzer, referring to the research of Tversky and Kahneman, discusses our tendency to
commit what is known as the “conjunction fallacy”. Provided with information about the
hypothetical but now nevertheless quite famous Linda, subjects go on to rank for their
probability two possible descriptions of Linda. One alternative is a conjunction (Linda is a bank
teller and active in the feminist movement), and the other is one constituent of that conjunction
(Linda is a bank teller). It turns out—sadly, one might think—that most of us are inclined to
assign a greater probability to the conjunction. Worse, it seems that most of us, while happy
enough to acknowledge that a conjunction cannot be more probable than one of its conjuncts,
find it very hard to counter-act our tendency to reason in this way. Having the error pointed out
to us in a particular case, does not seem to help us much in our next encounter with a case of this
kind. (Gigerenzer, 284)
Naturalizing Idealizations. Page 53.
interpretivist strategy is settled if we grant, as we surely should, that experiments do in fact
reveal stable and persistent patterns of cognitive error. Nobody thinks people do not sometimes
reason erroneously, and nobody ought seriously to entertain a theory of intentional states that
requires us to deny that we can investigate the kinds of reasoning-mistakes that people are prone
to making. The issue is what kinds of patterns it is we thus uncover, and how we ought to
characterize them.42 The key point is that the kinds of systematically occurring mistakes and
persistent cognitive illusions that researchers report need not be regarded as indicative of
irrationally maintained cognitive strategies, even if experimenters have isolated task-structures in
relation to which those strategies lead us into error.
What is implicitly at stake here is the scope of the normative diagnosis, as well as the source of
its normative force. A person may be convicted of irrationality only on the basis of a judgement
of how his or her states of mind are related to one another (cf., Davidson, 1985b; Stich, 1990,
Chapter 2). A response to some cognitive task may be taken as indicative of irrationality only in
conjunction with other states attributed (perhaps implicitly) to the subject. But then, if we are
judging the rationality of the person's cognitive performance, considerations other than the
validity of the particular piece of reasoning in question may be drawn into the picture. The
global scope of IDA might well show it to be the case that these mistakes systematically occur,
in suitably contrived circumstances, as the result of cognitive procedures that in fact are optimal,
in a perfectly good sense. They may be procedures we should have deemed it irrational for
creatures like us to abandon or modify (had we had a choice—which we often do not).43
42
Dennett makes a version of this point in response to Stich (1981a) and elsewhere. (Dennett,
1981a, 1981b) Dennett’s view is that to talk sharply about such phenomena, we need to retreat
from the intentional stance to the design stance. On my model, RM gives content also to our
diagnoses of irrationalities and cognitive error.
43
Stich seems to me exactly right about this—that rationality judgements are comparative
judgements, and presuppose a context of specific goals, purposes and empirical limitations.
Naturalizing Idealizations. Page 54.
Imagine that a cognitive procedure, one that in a certain type of case consistently brings subjects
into a state of cognitive illusion, is triggered by a definite set of parameters. Though we can
empirically locate kinds of cases where the procedure and the set of parameters together cause a
cognitive misfiring, this does not in itself show that we ought to modify either. The conjunction
of judgements which expresses the cognitive illusions, however assailable they appear when
regarded within a narrowly circumscribed context, do not in themselves support an inference to a
global judgement that the subjects are not thinking as they ought to think. Errors and illusions
certainly represent failures on the part of agents to live up to norms. However, these infractions
must be viewed in the context of all the evidence available to IDA, evidence that, from the point
of view given by the experimental context, would constitute collateral information. And it may
well be that, in the light of such “collateral” information we actual interpreters would come to
see the suspect judgements as results of a cognitive procedure we should deem it unwise to
abandon. In that case, we may no longer be entirely comfortable with the claim that those
mistaken judgements represent errors that the subject categorically ought to endeavour to put
herself in a position to avoid. The global perspective of RM leaves IDA room to conclude that
the norms in relation to which error is diagnosed may be trumped by metanorms, norms guiding
a person’s choice of cognitive strategy.
While experiments pinpointing the circumstances under which we tend to make certain kinds of
error yield powerful constraints on proposed models for human cognition, it is not easy to see
how such research can directly support firm judgements about the extent to which we, as
creatures, live up to the ideal of rationality. The reason is that it turns out to be hard to agree on
how to settle in advance the question of what kind of cognitive strategy the subject ought, in a
given case, to employ. We could of course stipulate that a rational cognitive strategy relative to
any given kind of task is one that consistently (and with maximum relative efficiency) yields
Naturalizing Idealizations. Page 55.
what we agree to be the right result for that kind of task. (Setting aside, as we have been, the
difficulty that arguments like Gigerenzer's may raise for particular conceptions of the right
result.) By that standard we are, indisputably, less than fully rational. However, we should then
also have to say that this is a kind of failure of rationality that it may be quite irrational for a
natural creature to seek to eliminate. The diminishing marginal disutility of certain kinds of
mistakes is such that the cost of developing or exercising cognitive strategies that would ensure
their elimination is entirely out of proportion to the gain (if any) to be had from being thus error
free.
The point I have just made is an instance of a familiar and perfectly general point about the
rationality of real (and thus finite) creatures. But of course, those who claim the evidence shows
that human beings systematically display irrationality will have an immediate retort. All this
establishes, they will point out, is that human irrationality can be explained, something that
certainly was not in dispute. That we as a kind are irrational in the particular ways we seem to
be may reveal something about the constraints under which nature has tweeked our cognitive
strategies and capacities.
The point to notice, however, is that the explanation of the phenomena offered is one that
rationalizes the apparent instances of irrationality in just the sense that RM requires; it renders
them the natural outcome of the application of procedures to which creatures like us quite
possibly ought to subscribe. (Or, as the case might be, by showing them to be the result of
cognitive mechanisms we ought to be pleased that evolution has endowed us with.) The
explanation does this by showing that the procedures or mechanisms are good for creatures like
us. We might press the point by saying that how rational it is rational for us to be depends on the
contingencies of our creature needs and interests and on the features of the environment within
which we pursue their satisfaction. Looking for a more felicitous, less paradoxical way to put it,
Naturalizing Idealizations. Page 56.
we should say that the idea of pure rationality, conceived as explicatable in terms of formal
principles, is an idea that for the purposes of rationalizing interpretation of behaviour is without
categorical normative force. When we as actual interpreters assess our cognitive practices for
their rationality, any genuine critical force our judgements may have derives in the final instance
from other aspects of our practices that we are at that moment, in the form of substantive
normative principles, implicitly privileging as measures of worthiness. When IDA judges
according to RM, what is required is an assessment of the contribution of particular aspects of
practice or behaviour to over-all agent goodness—an ideal interpreter is, as Dennett (1981c)
suggests, in this particular and substantial respect an ideal representative of ourselves.
Interpretivists are routinely chastised for refusing to come clean about what exactly it is they
attribute to us all when they make rationality a condition of having mental states. This, we can
now see, is because there is on their view nothing, a priori, to come clean about, except that to
be rational is a very, very good and important thing to be. Interpretivists are staunchly antireductivist with regard to the notion of rationality that IDA implements; when we empirically
investigate human cognitive capacities and strategies, we might discover all kinds of interesting
tendencies and results—but there is no fixable, explicatable notion of rationality against which
we can measure such findings and draw conclusions about the degree and distribution of
rationality of human beings as a kind. Indeed, there is nothing in the pragmatist's interpretive
strategy that suggests we could not come to adjust our assessments of rationality as a result of
empirical study of our cognitive capacities.
We must reject the interpretation of the interpretive strategy that sees it as an a priori
philosophical argument to a substantive conclusion about the quality or value of our cognitive
procedures. What underwrites the connection between rationality and psychological attitudes is
itself a species of naturalism; our conception of what we ought to be doing in the way of
Naturalizing Idealizations. Page 57.
reasoning leads us, as Dennett once put it, "eventually to a consideration of what we in fact do."
(1981c, 98). It must be emphasized, however, that such considerations of our actual practices
afford us no basis for a reductive account of rationality. Any gloss—or analysis—of 'rationality'
represents some particular application of our cognitive practices to themselves. Whatever
normative force such a particular application has, inevitably derives from attachments to aspects
of our actual cognitive practices. These attachments, in turn, can be rooted nowhere but in
experience, in the interaction of our creature need and interest with the environment in which we
function. Perhaps one day it will be unnecessary to add that this does not mean that these
practices cannot be meaningfully criticised or reformed—it implies only that they cannot be
assessed wholesale, by some standard not of our own experiential devising.
Let us now turn to Stich's charge that the interpretivist's strategy of linking meaning and
intentional states to the notion of rationality pre-empts the point of "a normative theory of
cognition". Stich, a self-styled "normative pluralist" about reason, still wants, it appears, to be
able to come up with a monistic evaluation schema for cognitive strategies, in terms of which,
perhaps, the legitimacy of a plurality of incompatible strategies may be established, relative to
the needs and interests and environments in which creatures do their cognizing. A burden,
infamously, of Stich's argument is that 'truth' and 'rationality' will not play any explanatory role
in this evaluation schema.44 With respect to a representationalist conception of truth, Stich
presses the question: "If that is indeed what it is for a belief to be true, do you really care
whether your beliefs are true?" (1990, 22) He arrives at "a consistently negative answer," and
concludes, "There is nothing special or important about having true beliefs." (1990, 24) At this
point, however, the proponent of the interpretivist strategy might accept the conditional in Stich's
44
“Notorious” may be the better word; see eg., Haack (1994). In what one with a polite
euphemism might term a “spirited” paper, Haack lumps together Stich and Rorty under the
catchy epithet “vulgar pragmatism”.
Naturalizing Idealizations. Page 58.
rhetorical question but retain another option. In the spirit of pragmatic naturalism, the
interpretivist should reply that we might surely stick with 'truth' and reject the representationalist
psycho-semantic analysis. The result is that Stich's modus ponens becomes the pragmatist's
modus tollens.45 Because pragmatic naturalists will view 'true' and 'rational' as approbations like
'good' or 'beautiful', they will see them as flexible, fragmentation-proof sorts of notions—notions
employed primarily to signal our approval of such cognitive and linguistic practices as we come
to endorse in light of our changing and multifarious collection of particular aims and our
evolving conception of what we want to be like. Here the interpretivist out-pragmatizes Stich's
pragmatic theory of cognition. The question the pragmatist asks is, in effect: why should we feel
compelled to monism at the meta-epistemic level? One answer is that it is a presupposition of
the brand of epistemology to which Stich remains explicitly committed. But the pragmatist's
reply to that, surely, is, "then so much for epistemology." Nor is this an entirely flippant answer.
One might offer it while indeed agreeing with a very great deal of the substance of Stich's
diagnostic arguments against analytic and anti-sceptical epistemology, and also with the spirit of
his normative cognitive pluralism. In particular, one may agree that there is no substantive
content to the idea of "intrinsic epistemic virtue," (1990, 24) doubting along with Stich whether
anything can be made of the thought that there are ways of reasoning that are inherently good
irrespective of our contingent needs and interests. From here, however, a pragmatist should go
on to doubt whether anything that is at all useful to the conduct of particular inquiries could be
gleaned at the level of abstraction at which "a normative theory of cognition" must be pitched.46
45
Rorty (1992, 1993b) and Davidson (1991) are explicit here, invoking Putnam’s naturalistfallacy argument against the very project of providing an analysis of truth. Dennett (1981c)
makes exactly this point when he refuses, in reply to Stich (1981), to commit himself to any
particular analysis of ‘rationality’ on the grounds that it is “a pre-theoretical notion”: ”I want,”
says Dennett, “to use ‘rational’ as a general-purpose term of cognitive approval—which requires
maintaining only conditional and revisable allegiances between rationality, so considered, and
the proposed (or even universally acclaimed) methods of getting ahead, cognitively, in the
world.” (Dennett 1981c, 97)
46
Cf., Williams (1991) for an elaborate argument against the coherence of the project of
Naturalizing Idealizations. Page 59.
Stich does not seem to doubt this. Instead, having rejected, with due pathos, the idea that 'truth'
and 'rationality' are concepts one may invoke to specify or explain what virtuous cognition
amounts to, Stich then seeks an alternative conceptual framework within which to pursue such an
explanation, precisely in the hope of resuscitating normative epistemology recast now as a quasiempirical discipline (1990, 28). Embracing a full-blown, full-blooded naturalistic pragmatism,
however, the interpretivist ought to reject this very project. The full-blown pragmatist will
accept Stich's dim view of the prospects for normative epistemology based on a proper
understanding of the content of 'truth' and 'rationality' but will not base this view on the claim
that it sometimes is not good to be rational, or that truth, when properly analyzed, turns out to be
something we should not obviously require of our beliefs. The pragmatist's scepticism toward
this sort of epistemological project arises instead from the view that 'rationality' and 'truth' are
not notions that have substantive normative content at all independently of our concrete
evaluations of particular instances of cognitive virtue, evaluations which must remain, as Stich
also insists, interest and purpose relative.
We Rortyans eagerly nod when Stich tells us that "appeals to rationality, justification, and the
rest" cannot serve in any substantive sense "as final arbiters in our effort to choose among
competing strategies of inquiry." (1990, 21) But we must dissent when Stich goes on to say that
we are thereby "in effect, denying that rationality or justification have any intrinsic or ultimate
value." The point, rather, is that nothing can serve as final arbiters in our effort to choose
among competing cognitive strategies. Nevertheless, truth, rationality, and justification could no
more fail to be intrinsically valuable than the good could fail to be intrinsically choice-worthy.
What ensures this is also what guarantees that analysis of these concepts which aims to abstract
away from malleable interest and contingent commitment, is normatively sterile—or self-
doubting or legitimizing “knowledge of the external world.”
Naturalizing Idealizations. Page 60.
deceptive.47 And it is, importantly, what ensures that no question of empirical substance
regarding the nature of our cognitive lives is begged by those who, pursuing the interpretivist
strategy, treat RM-governed ideal interpretation as making explicit the nature of the concepts
employed in folk-psychological ascription.
Let us take stock. Interpretivists are not, contrary to Stich's first allegation, concerned to explain
away the findings of cognitive psychology. How prone we are, as a kind, to making various
sorts of cognitive mistakes is certainly an empirical question, as is the extent of our ability to
learn to overcome such tendencies, or to compensate for them. So, too, are the extent of, and the
causes of, variation in these regards amongst members of our kind. With respect to these
empirical issues, the interpretivist construes rationality as a second-order category; particular
kinds of error of reasoning do not by themselves indicate any particular degree of global
irrationality. The global rationality-judgements of IDA express a view not only of the relation
between psychological states and processes, but also of the relation between these and the
constraints and needs and interests that provide the context in which these states are formed and
in which such processes operate. Such global rationality-judgements are the ones on which IDA
is instructed to rely when evaluating her candidate theories and accounts. And such judgements
are simply not settled by the specific patterns of error that research psychologists reveal. There
is, then, no conflict between the project of empirical investigation of particular cognitive
mechanisms and the commitment of the interpretivist strategy to RM. Indeed, the interpretivist
would claim, what gives us a firm grip on the patterns of error diagnosed by the psychologist,
what gives us confidence in the identifications of the intentional states on which the formulation
of any such diagnosis relies, is precisely their compatibility with RM as a globally regulative
principle. The patterns of error traced by empirical cognitive psychology owe what sharpness
47
Though this is not to suggest that such analysis may not have value, and indirect normative
implications, in particular contexts, where particular ends and interests are at stake.
Naturalizing Idealizations. Page 61.
they have to the possibility that just such errors may, from the global perspective of IDA, be
good errors to make for creatures like us.48 Nothing that empirical cognitive psychology could
uncover would, unaided by metaphysical commitment, be capable of damaging this claim.
It emerged in Section IV that action-explanations may turn on non-rationalizing generalizations
subsuming the kind of intentional states that we suppose to have caused the action. Neither from
this, however, nor from the viability of empirical tracking of cognitive error-patterns, does it
follow that we can empirically determine the extent to which we as a psychological kind are or
fail to be globally rational in the sense required for the application of RM. The latter possibility
is what the interpretivist must reject, as a possibility that is ruled out by the strong constraint
expressed by RM. This rejection issues from the interpretivist's conception of the rationalityjudgements on which we make ideal interpretation turn. Such judgements are, to condense the
matter, expressions of a dynamic, evolving cognitive meta-practice of idealizing projection of
what we actually find ourselves to be doing in the way of thinking and desiring.
Stich's second allegation was this. If we accept the interpretivist's conception of what it is to have
psychological attitudes then we can no longer envisage an enterprise that aims to be normative
with respect to our epistemic practices in general. This accusation, I have argued, is fair, and
trades on a point the interpretivist should be pleased to concede. Conceiving of reason as the
pragmatic naturalist does, any characterization of rationality or of warrant sufficiently abstract to
appear philosophical will, by virtue of this fact, be normatively impotent. It will not tell us how
to acquire fewer false beliefs, or desire better things, or act more wisely. To interpretivist ears,
Stich's proclamation of the fragmentation of reason sounds like the final disillusion of a wouldbe essentialist, a reluctant lament for the fragmentation of epistemology. The latter is something
48
I have not relied on this point in my response to Stich since it assumes the point Stich places at
issue, namely how we should conceive of the identity conditions of psychological states.
Naturalizing Idealizations. Page 62.
that naturalistic pragmatists, particularly "vulgar" ones, should not feel inclined to grieve.49
VII. Naturalism and reduction
Stich, we have seen, thinks interpretivists are rushing in where all but scientists should fear to
tread.50 In this he is representative of a large group of philosophers of psychology. These are
theorists who have taken to heart the Quinean view that the way to bring about the naturalization
of some domain is to bring it under the scope of natural science. This commitment also comes to
expression in Stich's view that interpretivism must be wrong, because it entails a manner of
individuation of psychological and semantic states that renders them scientifically quite
pointless.51 The conceptual limits to irrationality that Stich believes fall out of the interpretivist
strategy are "profoundly uninteresting" (Stich 1990, 51). "It is," Stich thinks,
an observer-relative, situation-sensitive constraint that marks no natural or
theoretically significant boundary...Plainly, the demarcation between states that
are intentionally describable and states that are not is going to be vague...it will
49
See Haack (1994) for a set of criticisms levelled at Stich from the perspective of someone who
wants to defend the idea of epistemology.
50
Davidson anticipates this reaction to his brand of non-reductive naturalism: “Do we,” he asks,
“by declaring that there are no (strict) psychophysical laws, poach on the empirical preserves of
science-a form of hubris against which philosophers are often warned?” (Davidson 1970, 216)
51
Here is Stich (1992):
The literature strongly suggests that those who want a naturalistic account of mental
representation want something like a definition—a set of necessary and sufficient
conditions —couched in terms that are unproblematically acceptable in the physical or
biological sciences. (260)
Stich disparages this reductivist impulse, in a recognizably pragmatic spirit, distancing himself
from projects like those of Fodor and Dretske. His pragmatism remains Quinean, though, in a
sense that is shared neither by Davidson nor Rorty. This is because Stich’s sceptical attitude
toward the interpretivist strategy depends on treating science as settling what we are entitled to
say there is. For the Rortyan pragmatist, there is no particular vocabulary, not science, not
metaphysics, which has a special, legitimating role by virtue of a capacity to settle what in
general there is.
Naturalizing Idealizations. Page 63.
not be stable, or objective, or sharp. (1990, 51-2)
This distinction "is not one that divides nature at its joints."52 The predicates characterized by
the interpretivist strategy are thus useless to the cognitive psychologist. Stich might be willing to
grant that the interpretivist strategy may plausibly be said to catch central features of the
vocabulary of the folk-psychological attitudes; but then, we should conclude, so much the worse
for folk psychology.
52
Stich attempts to make vivid our appreciation of this point with a thought-experiment; a
sequence of brains each (but the first, namely Stich’s) of which is a computational duplicate of
the former, except for one of the sentences in the “belief box”. Considering the owners of the
brains, Stich thinks that
when we attempt to describe these people in intentional terms (in a given context),
we will be forced to divide them up into two radically different groups. The ones
relatively close to me have intentionally characterizable states, the ones very far
away do not. If the computational paradigm in psychology is on the right track,
then this distinction, mandated by the chauvinistic principle of humanity, is
without any psychological significance. (Stich 1990, 53)
It is probably best to take Stich here to be speaking in the spirit of explication, and not argument.
Still, it does seem remarkable that the passage invites us to assume that the syntactically
characterized objects at a functionally defined brain-space settle which (if any) folkpsychological intentional states the creature whose organ the brain is might be in. With regard to
Stich’s sequence, what the rationality-maxim in fact would mandate, were we tempted to follow
Fodor and insist on identifying folk-states with computational states, is that we refrain from
identifying types of the former with types of the latter across brains—who is interpretable by
whom would then just not be something Stich’s case tells us anything about. But, of course, the
interpretivist has no incentive at all to follow Fodor this way. On the contrary, as far as the
interpretivist is concerned, her best buddy may well have a “belief box” that differs radically
from her own, with entirely different syntactic objects in it. So also with “the principles that
govern how these inscriptions interact with one another” (1990, 53); for the interpretivist,
questions about how, and on what, your soul-mate’s brain performs its computations simply are
not at issue when you marvel at the astounding frequency with which you find yourself
articulating each other’s thoughts. Attitudes are fixed by an ideal interpreter which attributes
them to persons depending on when they do whatever they can be observed to do (including
what noises they make); undoubtedly, people’s brains play an essential part in this. But it does
not follow from this that the brain and its states are the proper subjects of belief attributions. In
the context of a polemic against the interpretivist, Stich’s invitation is completely tendentious.
That he issues it, is an expression of the hold of the assumption he shares with Fodor and
countless others: the naturalization of folk-psychological states requires us to find ways of
characterizing those states in terms of predicates licensed by scientific theory.
Naturalizing Idealizations. Page 64.
A similar concern is expressed by Fodor and LePore (1993b) when they argue, in intended
reductio, that for all that the interpretivist could tell, it may be that we do not have beliefs at all,
but simply schmeliefs. Now, schmeliefs are,
…propositional attitudes exactly like beliefs in their functional roles, their
qualitative contents (if any), and their satisfaction conditions, except that they are
not analytically constrained by the principles of charity. To make matters worse,
it might be supposed that it is nomologically necessary that schmeliefs are
mostly true (mostly rational, or whatever)...Then, ceteris paribus, the only
difference between a creature's having beliefs and its having schmeliefs would be
that, in the latter case, there are logically possible world in which what the
creature has are mostly false, and in the former case there aren't. It might thus be
really quite difficult to tell beliefs and schmeliefs apart. (1993b, 75)
Here Fodor and LePore gently mock the interpretivist for characterizing predicates that will not
serve in a science of behaviour; for such a science it just could not matter whether a cognitive
state is a belief or a schmelief. And while Fodor—with Granny looking over his shoulder—is
unwilling to give the interpretivist the concept of belief (etc.), the point here is really the same as
the one Stich relies on: making intentional states out to be intrinsically normative is to disqualify
them as entities that could figure in any genuinely scientific account of human behaviour, since
such an account does not need—indeed, could not be sensitive to—the distinction the
interpretivist wants to draw between normatively constituted intentional states and other
(putative) cognitive states.
These objections highlight the worry that interpretivism cannot satisfy the demands of
naturalism. Since the interpretive strategy renders the vocabulary of thought and agency in terms
irreducible to predicates that will allow a nomic account of human behaviour, it must be rejected.
Naturalizing Idealizations. Page 65.
From this perspective, if you agree with the interpretivist that the strategy illuminates the
concepts of folk-psychological practice, then this simply shows that folk-psychological states are
not to be taken seriously (Stich). If, by contrast, it is your credo that these states are to be taken
seriously, then, from the same perspective, it follows that the interpretivist must be simply wrong
about the concepts of folk psychology (Fodor). In either case, you are taking it that the
ontological fate of the reifications of folk psychology is separable from questions of what we as
actual interpreters achieve by employing them and why we want to achieve those things—you
are taking it that there is a substantive ontological fact here to be settled, one way or the other, by
the success or failure of reductive proposals. On this perspective, the significance we ought to
afford the vocabulary of agency—its "ontological status"—is a function of our ability to link it
up with a vocabulary of science. It could in principle be that in spite of its utility this vocabulary
is actually ontologically inadequate. It could come to stand revealed, by philosophy, as invalid.53
For the pragmatic naturalist, the argument runs in the other direction; the irrelevance of the
prospects of reduction to the run-of-the-mill purposes and interests served by our vocabulary of
agency suggests that the naturalisation of this vocabulary has little to do with the supposed
philosophical validity that reduction is alleged to provide. Consider the kind of dissatisfaction
that Dennett's version of interpretivism often provokes. Reading Dennett, we quickly form the
impression that to have beliefs and desires is to be predictable from the intentional stance. One
might go on to think that this makes the vocabulary of beliefs and desires, in Dennett's words
(referring to frog psychology), "a practically useful but theoretically gratuitous short-cut..."
(1987c, 109). And then, impressed with Dennett's explicit disavowal of any principled
philosophical distinction between frog-psychological states and human-psychological states, one
might think that folk-psychology is simply a place-holder for a more enlightened, empirically
53
The kind of worry is expressed by Fodor (1987) and Dretske (1988). Cf., Stich (1991).
Naturalizing Idealizations. Page 66.
adequate conceptualization.
Certainly Dennett has flirted with this view. And even when he explicitly retreats from it
(1987b, 1991a) his critics often try to pin him to it. The intentional stance seems to Dennett's
critics to make at once both too much and too little of the attitudes. The sheer, contingent fact of
predictive success just seems too feeble a basis for a claim to realism of any sort; it is a fact that
cries out for explanation, and it is here, among the terms of possible explanation, that the
ontological action is. Such explanation might provide terms for a grounded realism toward the
attitudes, or it might display the ontological shabbiness of the vocabulary of folk psychology.
But Dennett's strategy claims for itself the right to endorse the attitudes while insulating them
from the success or failure of this kind of explanatory descent. For Dennett, it is enough that
folk-psychological explanation works, that it gets us what we want. For his critics, this is
irresponsible; while folk psychology may be here to stay, as long as this is just because no better
means of prediction actually happens to come along, this is not ontologically reassuring. The
thought that if we were to develop better predictive schemes then that would spell the end of folk
psychology, that thought seems just too irrealistic—instrumentalistic as the charge typically has
it—to be the sort of thought we want to have about our beliefs and desires.54 What makes
Dennett's views so unsatisfactory to such readers is that he simply dismisses the thought that
realists and eliminativists alike so clearly intuit: that the ontological status of the attitudes must
depend on the fate of attempts to characterize them by means of the predicates of an account that
actually explains, in other terms, the predictive success folk psychology appears to provide for
its user-group.
It is easy, perhaps, in thus objecting to interpretivism in the guise of Dennett's intentional stance,
54
See Dennett (1987b) for some regrets about “instrumentalism” and other labels, a lament he
carries further in (1993).
Naturalizing Idealizations. Page 67.
to lose sight of the fact that the fate of the vocabulary of intentional states is not, on the
pragmatist's view, confined to the question of which predictive strategy is most reliable, or
detailed, or elegant or precise or accurate. As much at issue is the question of what it is that is to
be predicted. What we folk (psychologists) care about, typically, is not how people move
various parts of their bodies, but what it is that we do by so moving them. And, again typically,
whatever predicates we settle on in our descriptions of bodily movements, these are predicates
agents can satisfy by moving their bodies in slightly, perhaps very, different ways. Such
differences we generally want the predicates of our folk-psychological vocabulary to be
insensitive to. What makes different movements instances of the same type of action, are the
interests that give applicability to the predicates explicated by ideal interpretation.55
In all cases, some interest(s) will give point to our typology, and in all cases, "multiple
realizability" of kinds of behaviour in physical movement would seem to prevail. There are no
such things as brute psychological regularities because there are no such things as brute bits of
behaviour. The point isn't merely that we only care to predict when we have some motive, or
that some of the things people do matter more to us than others—though this is undoubtedly so.
The point is that we cannot predict, indeed that there is nothing to predict, except in so far as
we care about some things rather than others, in so far, that is, as we have predictive interests of
one kind or another. Psychological explanation and prediction is, necessarily, of behaviour of
this or that kind, and the kinds here refer us ineluctably back to need and interest.
55
Similarly with empirical psychology: perhaps, as Dennett suggests, the job of psychology is to
formulate regularities that serve in explanation of “the reliability with which ‘intelligent’
organisms can cope with their environments and thus prolong their lives.” (1981b, 64) Then this
explanatory interest would be what ultimately settles what counts, for some behaviour, as being
an instance of that behaviour. Perhaps there are other ways of characterizing the explanatory
interest of cognitive psychology that are no less plausible. Such a different explanatory interest
might be sensitive to slightly different ranges of differences and similarities, and so classify
behaviours differently.
Naturalizing Idealizations. Page 68.
Dennett (1991a) reveals his pragmatist stripes when he defends the integrity of folk psychology
precisely by arguing the irreducibility of the types of this vocabulary to the predicates of some
other vocabulary. Asserting the reality of the patterns we trace with intentional-state ascriptions,
Dennett does not so much retreat from instrumentalism as take the edge of it by arguing that no
other instrument will do for these purposes. He denies, by implication, that the predictive aims
of folk psychology are specifiable in terms that transcend the vocabulary, and against which it
could, as a strategy, come up short. Once we follow Rorty and bring the individuation of the
very items of prediction under the scope of the vocabulary-constituting interests, instrumentalism
ceases to be the thin end of the eliminativist wedge.
If the argument in Section IV has merit, the identification of actions is not only interestdependent in a general way; the nature of these interests is such as to make the identity of
intentional states (and thus actions) dependent on actual contexts of interaction. There is no
fixing the elements of the subjective perspective of an agent on the world as such. To see an
item as an agent, then, is not only to see the item as autonomous with respect to the categories of
empirical law. It is also to see that item as possessing a nature beyond what any determinate
attribution of thought will make explicit; where agents are concerned there is, to paraphrase
Heidegger, always more being than theory. I suggest that this is a constitutive feature of the
vocabulary of agency—i.e., a part of what it is to consider some item as an agent. This is a way
to articulate the Nietzschean element that I made explicit in Section V. To endorse it is to
preclude the possibility that any vocabulary of empirical theory could ever do the job for which
we rely on the ascription of intentional states. But if reductive legitimation of this vocabulary is
ruled out, how can the vocabulary of agency realize a naturalistic purpose of any kind?
Reduction, says the pragmatist, is a meta-tool of science; a way of systematically extending the
Naturalizing Idealizations. Page 69.
domain of some set of tools for handling the explanatory tasks that scientists confront.
Naturalization, by contrast, is a goal of philosophy; it is the elimination of metaphysical gaps
between the characteristic features by which we deal with agents and thinkers, on the one side,
and the characteristic features by reference to which we empirically generalize over the causal
relations between objects and events, on the other. It is only in the context of a certain
metaphysics that the scientific tool becomes a philosophical one, an instrument of legislative
ontology. This is the metaphysics of scientism. It treats the gap as a datum, and it takes natural
science (or some sub-set of it) to be the philosophically fundamental account of what kinds of
items we may, in a respectable voice, say that there are in the world. Identifying the natural with
the science-side of the gap and the unnatural with the psychological side, scientistic philosophers
like Fodor and Stich set out to either redeem or reject the latter in terms of the former. Given the
assumptions, this is what naturalism demands.
The pragmatic naturalist, by contrast, treats the gap itself, that which transforms reduction into a
philosophical project, as a symptom of dysfunction in our philosophical vocabulary. Pragmatic
naturalism does not aim at conceptual reduction, but at a transformation of those conceptual
structures we rely on to sustain our sense of a metaphysical gap between those items we catch in
our vocabulary of thought and agency, and those items we describe in our vocabularies of causal
regularities.56 It is in the context of this metaphilosophical project that the interpretive strategy
56
By this characterization, McDowell (1994) is a pragmatic naturalist. I do not think he would
object to this; indeed, in this paragraph I take myself to be following McDowell’s lead. It is a
central lesson of Mind and World (see particularly lecture IV) that if we are to “reconcile
reason and nature” (1994, 86), we must exactly challenge those ways of thinking that make it
appear as if reconciliation must take the form of reduction. The differences between
McDowell’s metaphilosophical stance and Rorty’s are smaller than McDowell’s appropriation of
Kant might suggest. McDowell takes a much more optimistic view than does Rorty about how
much of the vocabulary of modern philosophy can (and should) be successfully reformed
through a naturalistic transformation of the vocabulary of mind; their therapeutic aims, however,
are shared.
Naturalizing Idealizations. Page 70.
as wielded by Dennett and endorsed by Rorty emerges as a naturalizing one. It is not merely
non-reductive, it is anti-reductionist; it seeks to free us from those philosophical perceptions that
transform reductive enterprises into tests for ontological legitimacy.
VIII. Naturalized philosophy
We may get a clearer sense of the philosophical context in which interpretivism functions by
considering the following provocative remark of Davidson's: "I can imagine a science concerned
with people and purged of 'folk psychology', but I cannot think in what its interest would
consist." (1987, 447) This stands in striking contrast to the sentiments of scientistic
philosophers. Is Davidson suggesting that a cognitive science as conceived by the Churchlands,
or by Stich—or, for that matter, by Dennett—is inherently without interest; that it could be of no
value? This would be an absurd view to take, and thus an absurd attribution. The point of the
remark is not that this would be an uninteresting science, but that such a science, however
interesting, would not illuminate the philosophical issues that Davidson takes himself to be
addressing; it answers to different interests. It would be wrong to think Davidson means
merely that such a science would not be relevant to his particular concerns, however. His remark
surely is intended normatively, expressing a conception of what philosophical concerns are, of
what the interests are that philosophical reflection should be responsive to.
What conception might lie behind the thought that a science of behaviour "purged of 'folk
psychology'" is philosophically irrelevant? It is a conception that ties philosophy to an interest in
practice. The conception, however, is not simply a matter of being responsive to the demand that
theory must be made relevant to our practical concerns, of resonating to Marx's final imperative
in the Theses on Feuerbach. The relation is constitutive, not imperative. A part of the
philosophical context that gives point to the interpretivist strategy is the claim that behaviour
Naturalizing Idealizations. Page 71.
emerges as purposive behaviour only in the vocabulary of folk-psychology; it is only by the
terms of this vocabulary that (some) events emerge as instances of motivated action. The
constitutive point of the vocabulary is to show up agency. We have seen some ramifications of
this. One, which I emphasized in Section III, is that the vocabulary will be structured around
concepts that insulate the members of their extensions from nomic generalizations. A second is
that the vocabulary yields determinate characterizations of agency only as it unfolds; no room is
left for the idea of action as a manifestation of an underlying subjectivity (sections IV, V). A
third ramification is this. For the pragmatist, as we have seen, attempts to reflect upon what
there is are not distinct from reflection upon the nature of our vocabularies. Because we
illuminate our vocabularies by giving explicit expression to the interests we take them to serve,
philosophy itself, even at its most abstract, becomes wedded to the vocabulary of action. Any
attempt to reflect upon the nature of things of some kind brings us to the question why we
(should) care about that kind of thing, and this question will immediately throw us back into
the vocabulary of agency.
This makes it evident why a science of human behaviour that gives up "the vocabulary of folk
psychology" would be philosophically uninteresting. This should not, clearly, be taken to mean
that there are not difficult questions philosophers may ask about what we do when we do
science—science of human behaviour and other topics—nor that individual sciences cannot pose
their own peculiar philosophical questions, nor that philosophers may not contribute fruitfully to
the reductive enterprises of science. But for anyone who conceives of philosophy as having an
ineliminably practical interest—for anyone who thinks that our attempts as philosophers to
reflect on what there is and how things are inexorably refer us back to a context which also
involves questions of what we should value and what we should strive to become—to leave
behind the vocabulary of agency is not finally to find a way to solve (or dissolve) philosophical
questions about creatures with psyche. Rather, what we will then have found is a way to sever
Naturalizing Idealizations. Page 72.
any tie between our topic and human praxis. For pragmatists, it is by their relation to human
practice that philosophical questions take such content and point as they have.57
I have just proposed a view of philosophy that emphasizes the distinctiveness of the vocabulary
of intentional states, of agency, and which ties philosophy as an enterprise to that distinctiveness.
This may seem to place me at odds with Rorty (1997), who invokes the lack of sharp
individuation-criteria for vocabularies to doubt that the "gulfs" between intentional psychology
and physics and between (as his example has it) biology and physics, respectively, are "not
equally wide, and of the same sort." (?) In both cases, we are just distinguishing vocabularies on
the basis of distinct purposes. For Rorty, no vocabulary, or division of vocabularies, is
philosophically special or privileged. There is an important truth to this, but I think its
significance may be slanted by Rorty's fear of reason. The truth is that there is no other measure
for critical evaluation of what we do or want than other things we do or want; there is no critique
or justification that transcends the contingencies of need and interest, contingencies that give our
vocabularies their shape. Recognizing this, however, does not force us to give up the idea that
philosophy has a constitutive relation to the norms of reason. To insist on this relation, in the
context of the interpretivist strategy, is just another way of stressing the point that philosophy is
57
This point is perhaps easier to read into Davidson’s writings than into Dennett’s, but I should
not think Dennett would be in serious disagreement with what I have just said about the point of
folk psychology. Dennett remarks on the inescapable nature of the intentional stance in (1981a,
27). In (1991a) he also makes clear that he regards the individuation of the kinds of behaviour
predicted from the intentional stance as interest relative. I should, however, like to extend this
point also to non-intentionally characterized behaviour—indeed to predicates that serve
predictive regularities in general. Since I don’t see what compels us to treat the idea of a basic
law as anything but an abstract idealization, I do not see why we should believe that there must
be a level of regularities where all questions of implementation become otiose—where, that is to
say, all that could possibly be offered by way of explanation of apparent constant conjunctions is
to say, with Fodor, that “God made it that way.” (Fodor, 1991) Indeed, Fodor’s quip is highly
apposite, since the idea of a finished physics and the idea of omniscience—the one (set of)
coherent account(s) that accounts for all there is to be accounted for— are mutually supportive
notions; doubting the point of one would seem to leave the other in serious trouble.
Naturalizing Idealizations. Page 73.
reflection on praxis.
The Reformist Rortyan claims, on behalf of philosophy, that when we invoke norms of reason
we are drawing on interestingly distinctive explanatory resources. Rorty, however, balks at such
apparently privilege-dispensing talk. He makes his point through a commentary on Davidson's
"Mental Events" (1970), polemically aimed at McDowell (1994). Rorty rightly suggests that
Davidson's notions of heteronomic and homonomic generalizations presume an independent
ability to individuate vocabularies, and therefore cannot be relied on in an argument for the
distinctiveness of vocabularies. Nor is this Davidson's strategy, though; I take him to be using
the concepts to express the conclusions he draws from what he proposes are the distinctive
features of our psychological vocabulary. Nevertheless, if the explanatory purposes for which
we deploy biological concepts can be met only in so far as we build teleology into our
descriptions, then those concepts, as well, will constitute a vocabulary distinct from that of nonteleological science. And then, as Rorty says, the differences between the "gulfs" are "of the
same sort". This is right, of course—at a certain level of abstraction. That is to say, the
differences between the vocabularies are the same in so far as both bio-physical and psychophysical generalizations come out heteronomic (as do psycho-biological ones). If this is so, then
Rorty is right when he suggests that the particular contrast Davidson (1991a) wants to highlight
between the concepts of the intentional vocabulary and those of other vocabularies is not
illuminated by the distinction between heteronomic and homonomic generalizations; not
illuminated, that is, by the distinction between vocabularies as such. But this is because that
distinction is exactly blind to the different sort of differences that we may want to invoke in
justification of proposed distinctions between vocabularies. It does not preclude that there may
also be interesting differences between these inter-vocabularic relations. Davidson's intent, I
suspect, is exactly that we should see an interesting difference between the sort of conceptual
features that may distinguish the biological or the geological from each other or from the
Naturalizing Idealizations. Page 74.
chemical or the physical, and the sort of conceptual features that make the psychological distinct
from all of these. This possibility is certainly available even if we treat both kinds of differences
as constitutive of vocabularies. This would explain why Davidson (1991a), as Rorty points out,
revokes his earlier claim (1970) that the distinctiveness of intentional concepts arises from the
indeterminacy of translation. It also means it would be wrong to conclude that Davidson could
not be urging us to find an interesting difference between psychology and non-intentional
sciences generally. The difference in question may be billed as meta-vocabularic: a difference
amongst the sort of differences that we, using the 'vocabulary' vocabulary, can rely on to
distinguish vocabularies. Once philosophers assume, as Rorty urges, and as I have tried to do,
the 'vocabulary' vocabulary, this is precisely what philosophical accounts of things will seek to
illuminate.
A reason, one might suppose, why Rorty appears less eager than Davidson to emphasize the
distinctiveness and indispensability of the vocabulary of agency, is that he is a great deal less
dismayed than Davidson about the prospect of leaving philosophy behind. This supposition,
however, would be mistaken—or at least misleading. Rorty strives to naturalize our conception
of philosophical reflection by thinking of it as an adaptive activity of natural creatures. We
should, he urges, learn to think of ourselves in terms such that there no longer appears to be
anything conceptually or philosophically mysterious about our being embodied thinkers, or
agents in a world of causes. The interpretivist strategy naturalizes precisely in so far as it frees
us from worries about the "ontological status" of the kinds that constitute the denotata of our
various ways of describing things. While Rorty's naturalism is in important ways Quinian, it is
not Quine's naturalism, nor that of Quine's more scientistic descendants. By resisting the
scientistic urge that informs both realism and eliminativism, the pragmatic naturalist insists that
questions of what sort of predictive vocabulary to apply when, and to what—or whom, are
questions that by their nature will not be contained within the scope of theoretical criteria of
Naturalizing Idealizations. Page 75.
theory-choice. As questions of vocabulary choice, such questions resist methodological
resolution. Neither mounting scientific knowledge nor the increasingly sophisticated theoretical
superstructure of methodology raised upon it by philosophy of science will, all by itself, tell us
under what aspects we should care about things.
Nor, however, and equally importantly, is this the kind of question that philosophy should be
attempting to derive answers to by settling questions of ontology. Such decisions, which are
normative and clearly decisions of fundamental importance, retain for pragmatists an
ineliminably practical element—in the sense that they cannot be extricated, by abstraction, from
what are essentially experimentally derived considerations of what we think we want to be like,
what we want our practices to become. But what pragmatic naturalists with one hand take away
from philosophy—the idea of ontology (whether as metaphysics or natural science) as a
substantive enquiry into the legitimacy of vocabularies—they return with the other; we are left
with a conception of philosophy as aiding our practical and ethical deliberations, our
experimentations, by imaginatively providing alternatives to what begins to look like conceptual
hang-ups and fixed ideas ('intuitions'), and depicting altered self-conceptions for us to try out.
On this view, the job of a philosopher is to make vivid how our practices might change if we
were to describe things—particularly human beings—in altered vocabularies, or if we extend
particular vocabularies into new domains. This intellectual practice is not so much a pursuit of
truth as it is a pursuit of alternative perspectives on the relevance to each other of various ways
of making truth-claims. It is exemplified by the pragmatic naturalist's promotion of the
interpretive strategy.
The interpretivist strategy undermines (as I have argued in sections III, IV and V) the reification
of mental content and of subjecthood. At the same time, the strategy also frees the notion of
reason from the transcendental aspirations in which it has been embedded (as I try to show in
Naturalizing Idealizations. Page 76.
sections VI, and VII), and makes a notion of reason available for a pragmatized conception of
philosophy. These two aspects of the naturalizing effects of the strategy are related. Both follow
from a characterization of a vocabulary of reflection that aims to extricate our notion of agency
and personhood from the dualistic, dichotomizing elements in the conception of subject and
object that have come to be dominant in the modern stage of the narrative that Plato launched.
These elements are what condition the opposition between reason and contingent creaturely
need, and they are what makes 'ontology'—the reductive reconnection of metaphysically ranked
vocabularies—appear both as a domain of substantive enquiry and as a pressing task. Some of
these elements are, to our detriment, still powerfully entrenched in our common vocabulary of
the mental. They are no less active in the tough-minded resolve of contemporary physicalism
than in the species-aggrandizing conceits of the early dualists of the modern era. Although they
are still shaping conceptions of philosophical problems and of the tasks of philosophy, these
elements are not presuppositions of philosophical reflection. In seeking to give them the slip,
Rorty is engaging in the distinctively philosophical project of providing a reasoned view of better
ways of being human.
Naturalizing Idealizations. Page 77.
Notes:
Naturalizing Idealizations. Page 78.
References:
Allen, Barry, 1993. Truth in Philosophy. Cambridge, MA: Harvard University Press.
Barrett, Roger and Gibson, Roger (eds.), 1990. Perspectives on Quine. Oxford: Blackwell.
Bilgrami, Akeel, 1992. Belief and Meaning: The Unity and Locality of Mental Content.
Oxford: Blackwell.
Brandom, Bob, 1994. Making It Explicit: Reasoning, Representing, and Discursive
Commitment. Cambridge, MA: Harvard University Press.
...., (ed.), 2000a. Rorty and His Critics. Oxford: Blackwell. "The Vocabularies of
Pragmatism: Synthesizing Naturalism and Historicism."
…., 2000b. “Vocabularies of Pragmatism: Synthesizing Naturalism and Historicism.” In 2000a.
Brandl, Johannes and Gombocz, Wolfgang (eds.), 1989. The Mind of Donald Davidson. Grazer
Philosophische Studien, 36. Amsterdam: Rodopi.
Cavell, Marcia, 1993. The Psychoanalytic Mind: From Freud to Philosophy. Cambridge,
MA: Harvard University Press.
Cohen, L.J. et al. (eds.), 1982. Logic, Methodology, and Philosophy of Science. Amsterdam.
Dahlbom, Bo (ed.), 1994. Dennett and his Critics. Oxford: Blackwell.
.
Davidson, Donald, 1970. "Mental Events." In Foster and Swanson. Page references to
Davidson, 1980a.
...., 1974: "Paradoxes of Irrationality." In Wollheim.
...., 1980a: Essays on Action and Events. Oxford: Oxford University Press.
...., 1980b: "Towards a Unified Theory of Meaning and Action." Grazer Philosophische
Studien, 11.
...., 1984: Enquiries into Truth and Interpretation. Oxford: Oxford University Press.
...., 1985a: "Deception and Division." In LePore and McLaughlin.
...., 1985b: "Incoherence and Irrationality." Dialectica, 39 (4).
...., 1985c: "Rational Animals." In LePore and McLaughlin.
...., 1986a: "A Coherence Theory of Truth and Knowledge." In LePore. Reprinted, with
"Afterthoughts, 1987," in Malichowski.
...., 1986b: "A Nice Derangement of Epitaphs." In LePore.
...., 1987: "Knowing One's Own Mind." In Proceedings of the American Philosophical
Association.
...., 1988a: "Epistemology and Truth." In Proceedings of the Fourth Panamerican
Philosophy Conference, Cordoba, Argentina.
...., 1989a: "The Myth of the Subjective." In Krausz.
...., 1989b: "What is Present to the Mind?" In Brandl and Gombocz.
...., 1989c: "The Conditions of Thought," in Brandl and Gomboczi.
...., 1990a: "The Structure and Content of Truth." The Journal of Philosophy, 87 (6).
...., 1990b: "Meaning, Truth, and Evidence." In Barrett and Gibson.
Naturalizing Idealizations. Page 79.
...., 1991a: "Three Varieties of Knowledge." In Phillips Griffiths.
...., 1991b: "Epistemology Externalized." Dialectica, 45 (?).
...., 1992: "The Second Person," in French, Uehling, and Wettstein.
...., 1993a: "Thinking Causes." In Heil and Mele.
...., 1993b: "Replies to Essays." In Stoecker.
...., 1995: "Could There Be a Science of Rationality?" International Journal of Philosophical
Studies, 3 (1).
Dennett, Daniel C., 1978. Brainstorms: Philosophical Essays on Mind and Psychology.
Montgomery, VT: Bradford Books.
...., 1981a. "True Believers." In Heath. Page references to Dennett 1987a.
...., 1981b. "Three Kinds of Intentional Psychology." In Healy. Page references to Dennett,
1987a.
...., 1981c. "Making Sense of Ourselves." Philosophical Topics, 12. Page references to Dennett
1987a.
...., 1987a. The Intentional Stance. Cambridge, MA: MIT Press.
...., 1987b. "Instrumentalism Reconsidered." In Dennett, 1987a.
...., 1987c. "When Frogs (and Others) Make Mistakes." In Dennett, 1987a.
...., 1987d. "Midterm Examination: Compare and Contrast." In Dennett, 1987a.
...., 1988. "Quining Qualia." In Marcel and Bisiach.
...., 1991a. "Real Patterns." The Journal of Philosophy, 88 (1).
...., 1991b. Consciousness Explained. New York: Little Brown.
...., 1993. "Back From the Drawing Board." Replies to essays, in Dahlbom.
Dretske, Fred, 1988. Explaining Behaviour: Reasons in a World of Causes. Cambridge, MA:
MIT Press.
...., 1995. Naturalizing the Mind. Cambridge, MA: MIT Press.
Evnine, Simon, 1991. Donald Davidson. Palo Alto: Stanford University Press.
Farrell, Frank B. 1994. Subjectivity, Realism, and Postmodernism—the Recovery of the
World. Cambridge: Cambridge University Press.
Fodor, Jerry, 1987. Psychosemantics. Cambridge, MA: MIT Press.
...., 1994. The Elm and the Expert: Mentalese and its Semantics. Cambridge, MA: MIT
Press.
Fodor, Jerry, and LePore, Ernest, 1992. Holism: A Shoppers Guide. Oxford: Blackwell.
...., 1993a. "Is Radical Interpretation Possible?" In Stoecker.
...., 1993b. "Is Intentional Ascription Intrinsically Normative?" In Dahlbom.
Foster, Lawrence, and Swanson, J.W., 1970. Experience and Theory. Amherst: University of
Massachusetts Press, London: Duckworth.
French, Peter, Uehling, Theodore, and Wettstein, Howard (eds.), 1992. Midwest Studies in
Philosophy: The Wittgenstein Legacy. Series vol 17. Minneapolis: University of
Naturalizing Idealizations. Page 80.
Minnesota Press.
Føllesdal, Dagfinn, 1975. "Meaning and Experience." In Guttenplan.
...., 1979. "Hermeneutics and the Hypothetico-Deductive Method." Dialectica, 33 (3-4).
...., 1982a. "The Status of Rationality Assumptions in Interpretation and in the Explanation of
Action." Dialectica, 36 (4).
...., 1982b. "Intentionality and Behaviorism." In Cohen, Pfeiffer, Podewski.
Gigerenzer, G., 1993. "The Bounded Rationality of Probabilistic Mental Models." In Manktelow
and Over.
Grandy, Richard, 1971. "Reference, Meaning, and Belief." The Journal of Philosophy, 70
(14).
Guttenplan, Samuel, 1975. Mind and Language. Oxford: Oxford University Press.
Haack, Susan, 1993. Evidence and Enquiry: Towards Reconstruction in Epistemology.
Oxford: Blackwell.
...., 1994. "Vulgar Pragmatism: Not an Edifying Prospect." In Saatkamp.
Healy, R. (ed.), 1981. Reduction, Time, and Reality. Cambridge: Cambridge University Press.
Heath, A.F. (ed.), 1981. Scientific Explanation. Oxford: Oxford University Press.
Heil, John and Mele, Alfred (eds.), 1993. Mental Causation. Oxford: Oxford University
Press.
Hurley, Susan and Shute, Stephen (eds.), 1993. On Human Rights: The 1993 Oxford Amnesty
Lectures. New York: Basic Books.
Jackson, Frank, (forthcoming). The 1995 John Locke Lectures.
Jennings, Ray, 1994. The Genealogy of Disjunction. Oxford: Oxford University Press.
Kim, Jaegwon, 1993. Supervenience and Mind. Cambridge: Cambridge University Press.
Krausz, Michael (ed.), 1989. Relativism. Bloomington, Indiana: University of Indiana Press.
LePore, Ernest (ed.), 1986. Truth and Interpretation. Oxford: Blackwell.
LePore, Ernest and McLaughlin, Brian (eds.), 1985. Actions and Events. Oxford: Blackwell.
McDowell, John, 1994. Mind and World. Cambridge, MA: Harvard University Press.
Marcel, A. and Bisiach E. (eds.), 1988. Consciousness in Contemporary Science. Oxford:
Naturalizing Idealizations. Page 81.
Oxford University Press.
Malichowski, Alan (ed.), 1990. Reading Rorty. Oxford: Blackwell.
Malpas, Jeffrey E., 1992. Donald Davidson and the Mirror of Meaning. Cambridge:
Cambridge University Press.
Manktelow, K.A., and Over, D.E. (eds.), 1993. Rationality: Psychological and Philosophical
Perspectives. London: Routledge.
Nietzsche, Friedrich, 1968a. The Will to Power. Trans. Kaufman and Hollingdale. New York:
Vintage.
...., 1968b. On the Genealogy of Morals. Trans. Kaufman and Hollingdale. New York:
Vintage.
Phillips Griffiths, A. (ed.), 1991. A.J. Ayer Memorial Essays. Cambridge: Cambridge
University Press.
Preyer, Gerhard, Frank Siebelt and Alexander Ulfig (eds.), 1994. Language, Mind and
Epistemology: On Donald Davidson's Philosophy. Dordrecht: Kluwer Academic
Publishers.
Quine, W.V., 1960. Word and Object. Cambridge, MA: MIT Press.
...., 1991. Pursuit of Truth. 2nd ed.. Cambridge, MA: Harvard University Press.
.
Rorty, Richard, 1967. "Introduction." In Rorty (ed.), The Linguistic Turn. Chicago: University
of Chicago Press.
...., 1979. Philosophy and the Mirror of Nature. Princeton, NJ: Princeton University Press.
...., 1982. Consequences of Pragmatism. Minneapolis: University of Minnesota Press.
...., 1986. "Pragmatism, Davidson, and Truth." In LePore. Reprinted in Rorty, 1991.
...., 1987. "Non-reductive Physicalism." Reprinted in Rorty, 1991a.
...., 1989. Contingency, Irony, and Solidarity. Cambridge: Cambridge University Press
...., 1991a. Objectivity, Relativism, and Truth. Cambridge: Cambridge University
Press.
...., 1991b. "Inquiry as Recontextualization: An Anti-Dualist Account of Interpretation." In
Rorty, 1991a.
...., 1992. "Putnam on Truth." Philosophy and Phenomenological Research. 52 (2).
...., 1993a. "Human Rights, Rationality and Sentimentality." In Hurley and Shute.
...., 1993b. "Putnam and the Relativist Menace." The Journal of Philosophy. 90 (9).
...., 1993c. "Holism, Intrinsicality, and the Ambition of Transcendence." In Dahlbom.
...., 1994. "Consciousness, Intentionality, and the Philosophy of Mind." In Warner and Szubka.
...., 1995a. "Philosophy and the Future." In Saatkamp.
...., 1995b. "Reply to Haack." In Saatkamp.
...., 1996. Pragmatism as Anti-Authoritarianism. Ms.
...., 1998a. "McDowell, Davidson, and Spontaneity." Philosophy and Phenomenological
Research, LVIII, 2.
Naturalizing Idealizations. Page 82.
….,1998b. Truth and Progress. Cambridge: Cambridge University Press.
….,1998c. "Is Truth a Goal of Inquiry? Davidson versus Wright." In 1998a.
Saatkamp, Herman J. (ed.), 1995. Rorty and Pragmatism. Nashville, TN: Vanderbilt
University Press.
Stich, Stephen, 1981. "Dennett on Intentional Systems." Philosophical Topics, 12.
...., 1990. The Fragmentation of Reason: Preface to a Pragmatic Theory of Cognitive
Evaluation. Cambridge, MA: MIT Press.
...., 1991. "What is a Theory of Mental Representation?" Mind. 101.
Stoecker, Ralf (ed.), 1993. Reflecting Davidson. Berlin: De Gruyter.
Warner, Richard, and Szubka, Tadeus, 1994. The Mind-Body Problem: A Guide to the
Current Debate. Oxford: Blackwell.
Wilson, N. L., 1959. "Substances without Substrata." Review of Metaphysics, 12.
Wollheim, Richard (ed.), 1974. Freud: A Collection of Critical Essays. Garden City, NY:
Anchor Books
Wright, Crispin, 1992. Truth and Objectivity. Cambridge, MA: Harvard University Press.