1.
Mike Albert's address is P. O. Box 535, Bedford, MA 01730, USA.
2.
For the time being, and as a service to our readers and contributors, we have a directory called `soft' hanging from our directory sorites at the node http://www.csic.es/sorites. The directory contains some of the non-commercial software we are referring to, such as archivers or 8-to-7 encoders (or 7-to-8 decoders).
3.
The procedure is as follows. Suppose you have a file called `dilemmas.wp5' in your directory c:\articles, and you want to submit it to SORITES. At your DOS prompt you change to your directory c:\articles. We assume your WordPerfect files are in directory c:\WP51. At the DOS prompt you give the command `\wp51\convert'; when prompted you reply `dilemmas.wp5' as your input file whatever you want as the output file -- suppose your answer is `dilemmas.ker'; when prompted for a kind of conversion you choose 1, then 6. Then you launch you communications program, log into your local host, upload your file c:\articles\dilemmas.ker using any available transmission protocol (such as Kermit, e.g.). And, last, you enter your e_mail service, start an e_mail to to sorites@ifs.csic.es and include your just uploaded dilemmas.ker file into the body of the message. (What command serves to that effect depends on the e_mail software available; consult your local host administrators.)
4.
Final form of this paper has benefited very much from the minutious reading and discussing of a first draft done by the editor of this journal, Lorenzo Peña, as well as from a discussion group in the Spanish Institute for Advanced Studies (CSIC), Madrid (Spain), formed by Carlos Thiebaut, Cristina Lafont and Lorenzo Peña. Thanks to all of them. Working extensively on this paper, which is a draft of the line of argument of my dissertation, was possible thanks to a grant of the Studienstiftung des deutschen Volkes.
5.
Cf. Goodman, N.: Fact, Fiction, and Forecast [in the following FFF], Hassocks
6.
p.95: «The entrenchment of a predicate results from the actual projection not merely of that predicate alone but also of all predicates coextensive with it. In a sense, not the word itself but the class it selects is what becomes entrenched, and to speak of the entrenchment of a predicate is to speak elliptically of the entrenchment of the extension [=reference, A.M.] of that predicate.»
7.
Exceptions to this can be found in the works of J.Leplin concerning his concept of «methodological realism» (see fn47) and S.Blackburn Reason and Prediction, Cambridge MA 1973, ch 4, who gives a realist account of Goodman's paradox.
8.
This has been demonstrated by interpretations of this theory given by H.K.Wettstein «Demonstrative Reference and Definite Descriptions» in: Philosophical Studies 40 (1981), 241-57, «Has Semantics Rested on A Mistake?», in: Journal of Philosophy 83 (1986), 185-209; «Cognitive Significance Without Cognitive Content», in: Almog, J. &al. (eds.): Themes from Kaplan, N.Y. 1989, 421-454, «Turning the Tables on Frege or How is it That «Hesperus is Hesperus» is Trivial?», in: Tomberlin, J.E. (ed.): Philosophical Perspectives 3: Philosophy of Mind and Action Theory, Atascadero (Cal.) 1989, 317-39, and N.U.Salmon («How Not to Derive Essentialism From the Theory of Reference», in: Journal of Philosophy 76 (1979), S. 703-725, as well as Reference and Essence, Princeton 1981 and «Reference and Information Content: Names and Descriptions», in: Gabbay, D./Guenthner, F. (Eds.): Handbook of Philosophical Logic, Vol. IV: Topics in the Philosophy of Language, Dordrecht 1987.
9.
The chief example of this attitude seems to be Putnam, although he as well should count as one among the theorists named before.
10.
This is to say, I haste to add, that I neither pretend to give a solution (because there is none) nor to abund in the theory of identity in modal logic. These are two defects which I want to be clear about from the beginning; they are due to the general character of the theses I want to put forward: they should be valid, I think, for every account of identity through possible worlds, because they do not concern the concrete structure of an assumption of sammeness of kind as such but its place and unavoidability in certain practices. I suppose that the most natural reading of the following results from the assumption of a modified Kripke-semantics for possible world like the one proposed by Deutsch in «Semantics for Natural Kind Terms», in: Canadian Journal of Philosophy 23/3 (1993), 389-412, and his improvements in «Semantic Analysis of Natural Kind Terms», in: Topoi 13 (1994), 25-30. However, as I said, the concern of this paper is less in semantics proper than in pragmatics.
11.
We can find witnesses for this suspicion on both sides. Thus H.K.Wettstein thinks that you simply miss the point of the theory of direct reference if you look for it exclusively in its aptness to formalize metaphysical speculation or in its contributions to the clarification of the interpretation of modal discourse. In «Turning the tables on Frege or How is it that «Hesperus is Hesperus» is trivial» he expresses this view as follows: «If one sees the modal arguments as at the core of the anti-Fregean approach, as I do not, one might conclude that intellectually mediated reference [i.e. the determination of extension by intension, A.M.] is not what the anti-Fregean revolution is about» (p.336, my italics), but, as we could add, in the theory of interpretation for modal logic. In «Cognitive Significance without Cognitive Content» (in: Almog, J./Perry, J./Wettstein, H. (eds.): Themes from Kaplan, N.Y./Oxford 1989, 421-54) he considers to be the «lesson of the anti-Fregean revolution» the insight that «linguistic contact with things --reference, that is-- does not presuppose epistemic contact with them» (454).
12.
This is Føllesdal's term who introduced it in his dissertation Referential Opacity and Modal Logic (Harvard 1961) and explained its use further in the articles «Quantification into Causal Contexts», in: Cohen/Wartofsky (eds.): Boston Studies in the Philosophy of Science, Bd. II, N.Y. 1965, 263-74, reappeared in: Linsky, L. (ed.): Reference and Modality, Oxford 1971, 52-62, «Knowledge, Identity and Existence», in: Theoria 33 (1967), 1-27, «Interpretation of Quantifiers», in: Rootselaar, B. van/Staal, J.F. (eds.): Logic, Methodology and Philosophy of Science, Amsterdam 1968, 271-81, «Quine on modality», in: Davidson, D./Hintikka, J.(eds.): Words and Objections: Essays in Honour of W.V. Quine, Dordrecht 1968, 147-57, «Situation Semantics and the `Slingshot' Argument», in: Erkenntnis 19 (1983), 91-8, «Essentialism and Reference», in: Hahn, L.E./Schilpp, P.A. (eds.): The Philosophy of W.V. Quine, LaSalle 1985, 97-113.
13.
This is, as everybody knows, Saul Kripke's term, who explained it and the premises for its application mainly in «Identity and Necessity» (in: Munitz, M. (ed.): Identity and Individuation, N.Y. 1971, S.135-64) and «Naming and Necessity» (mit Addenda) (in: Harman, G./Davidson, D. (eds.): Semantics of Natural Language, Dordrecht 1972, S.253-355 bzw. S.764-9).
14.
In a certain sense one can see this, at least in Føllesdal's case as a consequent application of Quine's critique of the analytic-synthetic distinction like the one pronounced in «Carnap and Logical Truth» (in: Hahn, L.E./Schilpp, P.A. (eds.): The Philosophy of Rudolf Carnap, LaSalle 1963, pp.385-406) where he says about definitions, which he cosiders to be the candidate of whose analysis we can most probably hope to get a notion on analyticity which does not coincide with logical truth: «Definitions (...) can be either legislative or discursive in their inception. But this difference is in practice left unindicated, and wisely; for it is a distinction between particular acts of definition (...) So conceived, conventionality is a passing trait, significant on the moving front of science but useless in classifying the sentences behind the lines. It is a trait of events and not of sentences.» (p.395)
15.
For reasons that, as I hope, will become clear in the following, I depart here to a certain extent from the «orthodoxy» of direct reference theory, because I want to make a more general use of its results without an essentialistic commitment from the outset. This is why I do not refer to «microstructures» or «object-identity» but rather introduce contextually an unspecified notion of «relevant sameness» which is evidently much broader than e.g. Putnam's «same» (sc. «The Meaning of `Meaning»', in: Putnam, H.: Philosophical Papers 2. Mind, Language and Reality, Cambridge MA, 1975, pp.215-71) or most of the other conceptions which have been developed in the framework of this theory (e.g. the writings of Salmon, Deutsch mentioned above). I consider it sufficient for the following to suppose some «sameness-in-use-relation» accepted by the users of singular or general terms in certain practices which are linked to inductive method and hypothetical reasoning. Each of these practices, as well as each discipline, will have its own specification of this relation of the form: «A is the same substance as B iff ...», «A is the same (historical,...) individual as B iff ...» etc., where «...» is probably an interpretation-condition drawing on admissible model-classes (`physically', `chemically', `historically' or otherwise admissible). Thus I am not necessarily referring only to «rigid designators» in the classical sense of unqualified identity, but to designators which are to be understood as rigid within each admissible model class. In that sense, substitutivity or identity seems to me to be a structure to be aimed at in the (a priori) evaluation of admissibility but not to be ontologically presupposed.
16.
In «Essentialism and Reference», in: Schilpp, P./ Hahn, L. (eds.): The Philosophy of W.V. Quine, LaSalle 31988, pp.97-113.
17.
For this kind of shortcut see Devitt «Against Direct Reference».
18.
This is a liberal allusion to Putnam's term «one-criterion-words» (cf. «Is Semantics possible?», in: Philosophical Papers 2. Mind, Language and Reality, CambridgeMA 1975, pp. 139-152) as to denote the class of general terms having necessary and sufficient conditions for their application or are «defined terms» in the strict sense of «definition» mentioned above in the text.
19.
Cf. Goodman, Nelson: Fact, Fiction, and Forecast, Hassocks 31979 [in the following FFF], p.45, fn.9. One could even adventure (see note 1) the hypothesis that a treatment valid for «natural kind words» should be expected to be valid for dispositional predicates as well: both types of expression are supposed to be counterfactual-supporting and -demanding: to explain the application of a dispositional predicate you have to invoke sooner or later a counterfactual condition, which is structurally the same when you demand that a kind-word refer «to the same things in all possible worlds». Both can only be introduced by reference to a part of their supposed total extension and have defeasible application-conditions, i.e. are supposed to function even when not associated with an exhaustive ncessary and sufficient condition for application. The best explanation of their use, i.e. to determine whether a given individual is or is not a such-and-such/has or has not such-and-such disposition is in both cases intimately tied to the best theoretical account available (this has been argued by W.K. Essler and R.Trapp in «Some Ways of Operationally Introducing Dispositional Predicates with Regard to Scientific and Ordinary Practice», Synthese 34 (1977), 371-96 and by Essler in «Some Remarks Concerning Partial Definitions in Empirical Sciences», Pacific Philosophical Quarterly 61 (1980), 455-62). I leave matters as confused and provisional as this because a thorough examination would demand its own place. However, see fn24 for some more details and section VI. for some speculations.
20.
Reference and Essence, Princeton 1981, Appendix II.
21.
«Semantics for Natural Kind Terms», Canadian Journal of Philosophy 23/3 (1993), p.404/405, where he shows that in a proper model-theoretic analysis of natural kind terms (his system NK) «the rule of necessitation [that is: {{phi}}|={{nec}}{{phi}}, A.M.] fails» (405). The important consequence this has for the usefulness of an «orthodox» reference-theoretic account (i.e., one making essential use of the notion of «rigidity» to model the behaviour of natural kind terms) of empirical classifications he stresses in «Semantical Analysis of Natural Kind Terms» (in: Topoi 13, 25-30) where he concludes: «It seems to me that the semantical concepts of rigidity and nondescriptionality are secondary to that of an important property.» (p.30)
22.
The similarity that can be sensed here between the so-called «model-theoretic argument» given by Putnam and the conditions that give rise to the theory of direct reference is, in my opinion, not casual. It shows that Putnam's argument, as given in Reason, Truth, and History, ch.2 and the proof in the appendix to the book, and its various variants, far from making him a «renegate» to realism (as M.Devitt would have it in «Realism and the Renegate Putnam», in Nous 17 (1983), pp.291-301) or committing him to transcendent idealism, shows (assuming that his reference theory is the core of his realist point of view) how realism is demanded for by paradoxical conditions within our practices when they are described in the traditional, semanticist way: the need for a new approach to reference is prompted rather than risked by the model-theoretic argument, it seems to me.
23.
The decisive steps to answer these questions would be some account of the representativity of a sample as much as a general account of what it is and how we know or suppose that some specification is sufficiently exact in the introductory situation. But this is far too complicated to be treated in this article.
24.
Thus the alleged «non-descriptivity» of natural kind terms would not be, as is often suggested, a result of some capacity of language to refer without any descriptive context but rather one of the continuous possibility of revision and conceptual change: there are not too few, but too many possible descriptions of the extension as to guarantee by this criterion referential transparency.
25.
This has been argued by Goosens and later by Deutsch (see below).
26.
«Semantics for Natural Kind Terms», in: Canadian Journal of Philosophy 23/3 (1993), pp.389-412.
27.
«Underlying trait terms», in: Schwartz, S.P. (ed.): Naming, Necessity, and Natural Kinds, Ithaca/London 1977, pp.133-54. Quine uses a similar term in connection with his explanation of the functioning and purpose of dispositional predicates («Necessary Truth», in: The Ways of Paradox, Cambridge, MA, 1975, 68-76) and clarifies their close resemblance with natural kind terms in «Natural Kinds» (in: Ontological Relativity and other essays, NY 1969, 114-38)). It would be worthwile investigating further Quine's conceptions and to compare them with what has been said in natural kind term reference theory. This is so because, following Quine's arguments one can see without difficulty a parallelism between dispositional predicates and kind terms and the evident importance of both in scientific practice, i.e. their epistemological import. Some indications may suffice to justify this claim: Quine calls (in «Necessary Truth») the counterfactual conditional-discourse underlying the use of dispositional predicates as indispensable for imputing dispositions on a domain and, above all, for the innerscientific practices of prediction and formulation (and interpretation) of hypotheses (p.73, 69), and describes its general epistemological structure as follows: «In general, when we say `If x were treated thus and so, it would do such and such', we are attributing to x some theoretical explanatory trait or cluster of traits.» (ibid., my italics). This attribution has the following status respectively function within a given corpus of knowledge: «the [disposition-, A.M.] term has been a promissory note which one might hope eventually to redeem in terms of an explicit account of the working mechanism.» (p.72, my italics) This suggests that the hypothesis to the effect of some «working mechanism» or a «submicroscopic structure» (ibid.) in the case of chemistry, in general of an «explanatory trait» (ibid.) is less to be interpreted as a serious hypothesis about the furniture of the world in itself than as a provisory, hypothetical and confirmable ontological posit with pending justification: «In the necessity constructions that impute dispositions, the generality lies along some known or posited explanatory trait. (...) They turn, still, on generality. But they turn on theory, too, precisely because they fix upon explanatory traits for their domains of generality.» (74) The acceptance and the concrete content and structure of these «promissory notes of common traits» is thus shown to vary from epoch to epoch, depending on the accessible «underlying theories» about what is possibly to be counted as a component of such a «common trait»: «What kind of account of a mechanism might pass as explanatory depends somewhat, of course, upon the general situation in science.» (72) This means that the use of dispositional classifications is comparably weak and relative to other, more fundamental or at least already approved and accepted classifications which in turn are seen to determine ontology. This is so because until there is no lawlike statement (or, according to the discipline) something functionally equivalent to it, the assumption of some such «ontological hypothesis» is nothing more and nothing less than a hypothesis with uncertain justification. Now, such general statements of law are known not to be inductively confirmed in any direct way (cf. e.g. W.K. Essler: Induktive Logik, Freiburg 1970, chap. V.4.); they are thus best be seen as belonging to the (contextual) a priori part of a theory as a whole which is rather than a consequence, a precondition of the investigation in the structure and content of the world.
28.
This is stressed by Putnam in «Possibility and Necessity» (in: Putnam, H.: Philosophical Papers 3. Realism and Reason, pp.46-69) where he remarks: «the `essence' that science discovers is better thought of as a sort of paradigm that other applications of the concept (...) must resemble than as a necessary and sufficient condition good in all possible worlds. This should have been apparent already from Quine's criticism of the analytic-synthetic distinction.» (p.64) That is: if you want to design a theory against the underdetermination-problems stemming from this criticism, this theory should not imply theorems to the effect of reproducing the very target of this criticism. So goes only half the way when he affirms in the same context: «saying that `Water is H2O', or any such sentece, is `true in all possible worlds' seems an oversimplification» (p.63); it is simply just as inadequate as saying that some such sentence is `analytic' and subject to the same criticism. It is, in other words, if theorem of a theory, part of an inoperative theory. In case of being an axiom, one should consequently look for ways to avoid it.
29.
Cf. «Essentialism and Reference», in: Hahn, E./Schilpp, P.A. (eds.): The Philosophy of W.V. Quine, LaSalle 31988, pp.97-113.
30.
This idea goes back to Quine's article «Natural Kinds» (in: Ontological Relativity and other Essays, N.Y. 1969, pp.112-38), where he treats dispositional terms, kind terms, counterfactual idiom, similarity grades and simplicity as a problem-cluster, for which he suggests that a clarification of one of the problems would have immediate consequences for the treatment of the others. However, I have the impression that Quine sees this problem-cluster as a sort of residual sphere of «second order» intensional talk which will be superseded as extensionalist approaches get better. This does not seem to be the case, for the approaches of Føllesdal and Kripke to some of the named problems does not make use of intensions in any suspicious way; quite on the contrary, Kripke's model-theoretic semantics of the modalities converts the whole idiom in a perfectly extensionalist language. And it was exactly there where the necessity for a distinction between kind terms and general terms arose. So it seems that the problem remains under extensionalist treatment.
31.
This is, of course, not to say that there are grades of truth. There are supposed to be grades of acceptability, measured by some measure-function (usually supposed to be some modification of a probability calculus), but not of truth, for what is to be accepted is the statement in question, i.e. true sentence. And this taking to be true of some determinate sentence is considered to be more or less rational, according to the output of the canonical method.
32.
In: Goodman, Nelson: Problems and Projects [in the following PP], Indianapolis 1972, pp.363-6. A precision of this argument resulting from the subsequent discussion with Carnap about this article can be found in «On Infirmities of Confirmation Theory» (PP, pp.367-70). The most famous version of the problem is probably the one in FFF, chapter 3 («The New Riddle of Induction»).
33.
Cf. Putnam, H.: Representation and Reality, Cambridge MA, 1988, p.13 where he remarks on the occasion of interpreting the changes in the specification of the concept `electron' within a «story of successive changes of beliefs about the same objects» (namely Bohr's various descriptions of them): «to treat all (...) occurences of `electron' [within this process, A.M.] as synonymous as is involved in his [Bohr's, A.M.] decision to treat later research programs in the story as extensions of the earlier ones (...) plays a central role in theory evaluation. In fact, treating `electron' as preserving at least its reference intact through all this theory change and treating Bohr's 1934 as a genuine successor to his 1900 theory is virtually the same decision».
34.
This point is, to my knowledge, due to W.Lenzen (Theorien der Bestätigung wissenschaftlicher Hypothesen, Stuttgart 1974, p.174ff., esp.183, fn5) That is to say, the new predicates construed by a definition like the one for «grue» («corrupted» we might call them, following W.K. Essler's terminology in «Corrupted Predicates and Empiricism», in: Erkenntnis 12 (1978), pp.181-7) do trivially coincide with the «normal» ones in case that the second clause (after the `or') in DGRUE is false because of a factual truth like
35.
This rule is the following (cf. F.v.Kutschera: «Goodman on Induction», in. Erkenntnis 12 (1978), pp.189-207): Take the two principles induction (a) and (b) and any predicate F and any set A of objects such that A={a1, ..., an} is the set e.g. of all objects tested for F until t (or, more general, a non-empty real subset of all objects in the supposed universe such that there is an ai{{epsilon}}A with i{{epsilon}}1, ...,n), that is, the available data. Then, this is Goodman's argument, there is a predicate F' such that F'(aj){{biimplication}}F(aj) for j=1, ...,n and F'(ak){{biimplication}}¬F(ak) for all k=1, ...,n. Formally this is F'x:=df (x{{epsilon}}A&Fx)V(#172;x{{epsilon}}A¬Fx. The contradiction arises for a simultaneous application of (a) or (b) for an+1 with respect to F and F' which seems justified, for Fx=F'x for A, which gives F'an+1=¬Fan+1 and Fan+1.
36.
To get an impression of the impact caused by it, may it suffice to recommend the excellent collection of essays on Goodman's paradox provided by D.Stalker: Grue! (Chicago/LaSalle 1994), especially the exhaustive annotated bibliography of texts in English on the problem contained in it.
37.
Goodman himself says that in view of this problem the aim has to be to reach a «dichotomy of predicates» (FFF, p.80). The insufficiency for an answer to this question of the resources given to us by past behaviour is stressed by him in his critique to Hume when he says: `Hume overlooks the fact that some regularities do and others do not establish such habits; that predictions based on some regularities are valid, while predictions based on others are not. (...) To say that valid predictions are those based on past regularities, without being able to say which regularities, is thus quite pointless. Regularities are where you find them, and you can find them anywhere.' (FFF, S.82) This obviously applies mutatis mutandis to descriptions of «induction-regularities» found in our culture (be they or not reached by reflexion: the question in point is whether they have normative import or not). What is demanded is a general procedure to distinguish two types of predicates in the structure of which one could find some set of interpretation-theoretic presuppositions of valid inductive inferences (as opposed to invalid ones); this is obviously impossible if one is limited to particular and contingent descriptions of the set of all valid inductions.
38.
Cf. Hacking, I.: «On Kripke's and Goodman's Uses of `Grue»', in: Philosophy 63 (1993), 269-95.
39.
A case where it might be reasonable to keep or construct a corrupted predicate could be, for example, the case of some «objective» change that would, however, for its particularity, not call for a change of theory, but nevertheles for a modification in the homogeneity-supposition. Imagine for example human population of a certain specific genetic structure inhabiting an area with active volcanoes. One day one of them erupts and this eruption causes testable changes in the genetic material of the children of the members of the population that survived the catastrophe. In that case it would be irrational to expect the predicate of genetic theory, say «to have characteristics F, G, and H in the genes», which was coextensional with «being a (geno-)typical inhabitant of area V» (and even theoretically more exact) before the eruption to be projectible afterwards. That means that it would be irrational, knowing the «rule» of corruption, to go on using the genetic predicate of before the eruption because it used to be perfectly projectible; rather one should take the «rule of corruption» in account and add it to the background knowledge (in the example this would be exactly a «Goodmanian» disjunction of the type: «someone is (geno-)typically V-ish iff he/she either has characteristics F, G, and H in the genes and was born before the eruption or has characteristics K, L, M and was born after the eruption»). One could even go on with this example and think of the possibility that this variation is not of dominant character and thus disappears, say, after the seventh generation so that the first predicate gets fully re-applyable (when all members of the variant-population have died). For the time between these two events it would, nevertheless, be inappriopiate.
40.
This follows from the definitions given above: If we take A as e.g. the class of all experiential sentences up to now and assume that there is no more individual than those contained in A, then Fx=F'x (see fn30). This is quite obvious, because the individual expression needed to define either «grue» on the basis of «green» or «green» on the basis of «grue» does, in that case, not refer. Thus the condition «not drawn until t» is trivially true because there is no more thing to be drawn and the condition in the second part of the disjunction of the definiens is, because of the (in classical logic) trivial falseness of «drawn after t», also false. Therefore the definition of «grue» (i.e., in general, of F') is satisfied, in this case, by all things that are green (i.e. F).
41.
For example we could imagine (in the case of Goodman's `grue') that there is only a finite number of additional things in our universe which does not exceed the number of green things -n; then the average homogeneity of the class of all grue things could never reach the average homogeneity of the class of green things because n does not become sufficiently little to be neglected. Otherwise the set of all grue things becomes (in the limit) almost indistinguishable from the simple (inductive) complement-class to green (pace the n green things in it, but if n is very much lesser than the total of all grue things in the limit, then there is an almost zero-possibility to get something green out of the class of grue things).
42.
This has been argued against certain attempts to dismiss Goodman's problem on grounds of «natural» categorization-systems which provide us with a conflict-blocking overhypothesis by J. Ullian in «More on `Grue' and Grue», in: The Philosophical Review 70 (1961), pp. 386-9.
43.
This is in fact the same as what Putnam says in his «model theoretic argument», in Reason, Truth and History, Cambridge MA 1981, ch.2. For a fine and very clarifying account of the structure of the argument see Hallett, M.: «Putnam and the Skolem Paradox», in: Clark, P./Hale, B. (eds.): Readiing Putnam, Oxford 1994, pp.66-97.
44.
It is never total, though: we can, with a change in «colour theory», most probably expect that our former predicates for colours in general will all be corrupted with the condition «Something is of colour x iff of colour x until the new theory was accepted or of colour y afterwards» or something like that. To suppose projectibility in a continuous sense we should then have to wait for a «unified colour theory». But in this section I am talking about the presuppositions we make from within an induction-based practice, and this is essential for the acceptability of realist assumptions, I think.
45.
This is, as mentioned before, a hint to the normative character of a possible reconstruction why we do not always get confused by Goodmanian predicates. The elucidation asked for is, as far as I can see, how to make clear why corrupted predicates must not occur in certain practices, and not a general answer to the sceptic, that is, a discovery of something that is the case that makes them not occur in fact. An assumption raised by such a reflexion is of intrinsic normative character because for the reasons given it is neither plausible nor even desirable to exclude Goodmanian predicates a priori. The only answer to the problem raised by corrupted predicates asked for is one from within the field where they actually cause trouble. If that is right, then the only thing we have to make clear is what exactly we do if we exclude them and what are the assumptions we have to make to be able to do so. This is the structure of the answer to the question how it is that we do not always stumble over corrupted predicates and consequently err in our inductive behaviour. These assumptions may have ontological import of a general sort (like the differentiation between sign, interpretation and object) and,, in virtue of that structure, exclude certain interpretative strategies as unapt to serve this aim (render certain interpretation theories wrong for an account of this behaviour and the contribution of language to the success of general behaviour), but one must not forget that this does not «prove» them to be «true». They are part of a rationality strategy seen from inside. From a participant perspective in the mentioned practices we certainly assume the existence and independence of our objects of investigation from the outcomes of the investigation, that is, we are and have to be «internal realists». However, this does not in the least mean that the ontology supposed in these practices has to be seen as any more priviledged than are these practices themselves in our conception of ourselves. This is to my mind the reason for the steady insistence on «explanatory relevance» of a common trait, for this is a case where the privilege of being worth to be pursued -- explaining, that is -- is almost too evident to be stressed. Especially it does not justify a claim to the effect that this is the world and the normativity integral to these assumptions has not to be misunderstood in the sense that, biewed from the outside (possibilitated by e.g. an alternative account of the domain) we have to hold on stubbornly to some set of categories.
46.
Kripke, S.A.: Wittgenstein on Rules and Private Language, Cambridge MA 1982, esp. p.20.
47.
Cf. FFF, p. 96: «Somewhat like Kant, we are saying that inductive validity depends not only on what is presented but also upon how it is organized; but the organization we point to is effected by the use of language and is not attributed to anything inevitable or immutable in the nature of human cognition.»
48.
In: Journal of Philosophy 59 (1962), S.647-58.
49.
Cf. Putnam, Hilary: Representation and Reality (Cambridge MA 1988), S.46.
50.
«semantics for Natural Kind Terms»
51.
. For a survey of theories about rights, see Morton E. Winston's anthology, The Philosophy of Human Rights (Belmont, California: Wadsworth, 1989).
52.
. Pannomial Fragments in The Works of Jeremy Bentham, vol. 3 (New York: Russell and Russell, 1962) 221.
53.
. After Virtue: A Study in Moral Theory (Notre Dame, Indiana: University of Notre Dame Press, 1981) 67.
54.
. I do think that there is a correct theory, and I discuss it in my own courses. I just do not feel that arguing for it is the best way to promote the solution of important social problems.
55.
What I have in mind is something like the «satisfaction» polls done periodically by the Gallup organization. V., for example, George Gallup, Jr., The Gallup Poll: Public Opinion 1993 (Wilmington, Delaware: Scholarly Resources Inc., 1994) 223-224.
56.
. Karl Marx, Economic and Philosophical Manuscripts in T.B. Bottomore, trans. and ed., Karl Marx: Early Writings (New York, McGraw-Hill, 1964) 124.
57.
I am greatly indebted to Lorenzo Peña (CSIC, Spain) for his helpful and careful comments on an early draft. My thanks to Josep Corbí and Tobías Grimaltos (Universidad de Valencia, Spain) for discussions about the first version of the paper.
Research for this paper has been funded by the Spanish Government's DGICYT as part of the project PB93-0683. My thanks to this institution for its generous help and encouragement.
58.
L. Laudan, Science and Values, Berkeley, Los Angeles: Univ. of Calif. Press, 1984 (hereafter SV).
59.
SV 59-60. Methodological rules for Laudan are hypothetical imperatives which relate a strategy to a goal according to this pattern: `If one's goal is y, then one ought to do x.' The accuracy of a methodological rule consists in the degree of success it has showed in attaining the goal at issue. Therefore they have to be tested against the historical record. See his «Progress or Rationality? The Prospects for Normative Naturalism», Amer. Philos. Quarterly 24 (1987): 19-31. A thorough discussion of this view can be found in G. Doppelt («The Naturalist Conception of Methodological Standards in Science: A Critique», Philosophy of Science 57 (1990): 1-19). Laudan's reply is in «Normative Naturalism», id., 44-59.
60.
K. Popper, The Logic of Scientific Discovery (London: Hutchinson, 1959), chs. VII y *VIII; W.V. Quine, «Simple Theories of a Complex World», in his The Ways of Paradox and Other Essays (New York: Random House, 1966), 242-46; E. Sober, Reconstructing the Past: Parsimony, Evolution and Inference, (Cambridge, MA: MIT Press, 1988). For a naturalistic approach that links simplicity with more basic epistemic values see B. Ellis, Truth and Objectivity (Oxford: Basil Blackwell, 1990), ch. 8.
61.
A classical paper on IBE is G. Harman, «The Inference to the Best Explanation», Philosophical Review 74 (1965): 88-95. A much more recent essay that underwrites the importance of this kind of inference in human knowledge is P. Lipton, Inference to the Best Explanation (London: Routledge, 1991). IBE connecting success with truth is fully developed in J. Leplin (ed.), Scientific Realism (Berkeley, CA: Univ. of Calif. Press, 1984).
62.
H. Putnam, Meaning and the Moral Sciences (London: Routledge & Kegan Paul, 1978) p. 25. In «Structural Realism: The Best of Both Worlds?», Dialectica 43 (1989): 99-124, J. Worrall maintains that pessimistic induction was clearly stated by Poincaré.
63.
«Some people have assumed that my aim was something like exactness or precision; or even applicability: that I hoped to find a numerical function which can be applied to theories and which tells us, in numerical terms, what their verisimilitude is (or at least their truth content; or perhaps their degree of corroboration). In fact, nothing can be further removed from my aims. I do not think that degrees of verisimilitude, or a measure of truth content, or falsity content (or, say, degree of corroboration, or even logical probability) can ever be numerically determined, except in certain limited cases (such as 0 and 1).» K.R. Popper, Objective Knowledge: An Evolutionary Approach (Oxford: Oxford U.P. 1972), p. 58.
64.
K.R. Popper, op. cit., ch. 9.
65.
See R. Boyd, «Metaphor and Theory Change» and E. McMullin, «Metaphor in Science», in Metaphor and Thought, A. Ortony (ed.), (Cambridge: Cambridge U.P., 1979). A more recent attempt -- a hybrid causal and description theory of reference, as is defined by its author -- can be found in D. Cummiskey, «Reference Failure and Scientific Realism», Brit. J. Phil. Sci. 43 (1992), 21-40. A well known for causal theories of reference is Naming, Necessity and Natural Kinds, P. Schwartz (ed.) (Cornell U.P., 1977). For an account of approximate truth distinguished from verisimilitude, probability and mere vagueness, see T. Weston, «Approximate Truth and Scientific Realism», Phil. of Science 59 (1992): 53-74.
66.
L. Laudan, Progress and Its Problems (Berkeley, CA: Univ. of Calif. Press, 1977).
67.
SV, 120 (words in italics have been added). Laudan suggests that a connection between approximate truth and success may exist: «I must stress again that I am not denying that there may be a connection between approximate truth and predictive success. I am observing only that the realists, until they show us what that connection is, should be more reticent than they are about claiming that realism can explain the success of science.» (SV, 119, footnote 21).
68.
See above footnote 9.
69.
A. Rosenberg and C.L. Hardin, «In Defense of Convergent Realism», Philosophy of Science 49 (1982): 604-15; D. Cummiskey, «Reference Failure and Scientific Realism: a Response to the Meta-induction», British Journal for the Philosophy of Science 43 (1992): 21-40; J.W. McAllister, «Scientific Realism and the Criteria for Theory-Choice», Erkenntnis 38 (1993): 203-22; E. McMullin, «A Case for Scientific Realism», in J. Leplin (ed.), op. cit., 8-40; J. Leplin, «Truth and Scientific Progress», in id., 193-217.
70.
E. McMullin, op. cit., p. 17.
71.
See the chapters 10 and 19 of his Reason and the Search for Knowledge, (Dordrecht: D. Reidel, 1984).
72.
Progress and Its Problems, ch. 4, footnote 2.
73.
Vann McGee, «A Counterexample to Modus Ponens», The Journal of Philosophy, Vol. LXXXII (September, 1985), 462-471; and Ernest W. Adams, «Modus Tollens Revisited», Analysis 48.3 (June, l988), 121-127.
74.
J. L. Austin, «Ifs and Cans», Philosophical Papers (Oxford: Oxford University Press, 1961)
75.
See St. Leeds: «Theories of Reference and Truth», in: Erkenntnis 13 (1978), pp.111-130; P. Horwich: «Three Forms of Realism», in: Synthese 51 (1982), pp.181-201; M. Williams: «Do We (Epistemologists) Need a Theory of Truth?», in: Philosophical Topics 14 (1986), pp.223-242.
76.
See W.v.O. Quine: Word and Object, Cambridge 1960; also: Ontological Relativity and Other Essays, N.Y. 1969.
77.
See Tarski: «The Establishment of Scientfic Semantics», in: Logic, Semantics, Metamathematics, NY 1956; also: M. Devitt: Realism and Truth, Oxford 1984; for a physicalist version, see H. Field's article «Tarski's Theory of Truth», in: The Journal of Philosophy 64/13 (1972), pp.347-375.
78.
L. Wittgenstein, «Vermischte Bemerkungen», in: Uber Gewissheit, Werkausgabe, vol.8, Francfort 1984, p.463.
79.
From a radically epistemic position as the one of Rorty it is always argued, against the defenders of a concept of truth as «rational acceptability under ideal conditions», that they are not sufficiently consequent with their own position because they do not eliminate «reality» or the presupposition of a «shared objective world» from their theories, since this presupposition, according to Rorty, cannot be more than a «residue of the theory of correspondence», that is, «it would make sense only if what is true is determined in some way by such a world» (in: «Sind Aussage universelle Geltungsansprüche?», p.10-11, Manuscript version.)
80.
The antifallibilism that is implicit in the epistemic conception of truth as «rational acceptability under ideal conditions» is pointed out by Putnam himself when he states in «Realism and Reason» (in: Meaning and the Moral Sciences, London 1978, pp.123-140): «The supposition that even an `ideal' theory might really be false appears to collapse into unintelligibility.» (p.126) A more detailed exposition of this conception of truth can be found in H. Putnam: Reason, Truth and History, Cambridge 1981, pp.54ff.; also: J. Habermas: «Wahrheitstheorien», in: Vorstudien und Ergänzungen zur Theorie des kommunikativen Handelns, Francfort 1984, pp.127-186; K.-O. Apel: «Fallibilismus, Konsenstheorie der Warheit und Letztbegründung», in: Forum für Philosophie (ed.): Philosophie und Begründung, Francfort 1986, pp.116-211.
81.
D. Davidson: «The Structure and Content of Truth», in: The Journal of Philosophy 87/6 (1990), p.307. A more detailed exposition of such an argument can be found in C. Wright: Truth and Objectivity, Cambridge, MA. 1992, p.37ff, especially p.45.
82.
A. Wellmer: «Wahrheit, Kontingenz, Moderne», in: Endspiele: Die unversöhnliche Moderne, Francfort 1993, p.158.
83.
See fn.7.
84.
In this respect, Davidson remarks in his «A Coherence Theory of Truth and Knowledge» (in: E. LePore (ed.): Truth and Interpretation. Perspectives on the Philosophy of D. Davidson, Oxford 1987): «What Convention T (...) reveals is that the truth of an utterance depends on just two things: what the words as spoken mean, and how the world is arranged.» (p.309)
85.
In what follows I will refer basically to Habermas's article «Wahrheitstheorien» [WT] (in: op. cit., pp.127-183).
86.
F.P. Ramsey: «Facts and Propositions» (1927), in: The Foundations of Mathematics, London/N.Y. 1931.
87.
Rorty distinguishes in his article «Pragmatism, Davidson and Truth» (in: E. LePore (ed.): Truth and Interpretation, Oxford 1986, pp.333-355) along with the disquotational use of the predicate `true' two other uses of the same: the endorsing use -- through which we assent or approve what is said by someone -- and the cautionary use -- through which we question the truth of what is said by someone. Returning to this distinction in his article «Universality and Truth» (1993), Rorty considers that the cautionary use -- that is, the use whereby we contrast «true» to «justified» -- is the only use that cannot be eliminated from our linguistic practices, since, in his mind, the other two uses can be easily paraphrased in terms that do not require the predicate `true'.
88.
J. Habermas, «Was heißt Universalpragmatik?», in: Vorstudien und Ergänzungen zur Theorie des kommunikativen Handelns, Francfort 1984, pp.388-89.
89.
This does not imply, of course, denying Davidson's thesis that truth is a primitive concept in the sense that it is undefinable, but only that it is possible to explain aspects of its use in the context of the revision of our beliefs that shed light on the meaning of this concept in its internal relation to other concepts, for example.
90.
Although the conditions for knowledge that I point out here are usually attributed to Plato (in: Theatetus 201 and, maybe, also in Meno 98), my recapitulation of these conditions follows (with slight variations) the one given by A.J. Ayer: The problem of Knowledge, London 1956, p.35, and R.M. Chisholm: Perceiving: a Philosophical Study, New York 1957, p.16.
91.
In this context I do not consider the difficulties pointed out by E. Gettier in his article «Is justified true belief knowledge?» (in: Analysis 23/6 (1963), pp.121-123) because they are intended to show the incomplete nature of these conditions whereas my argument is exclusively based on the unquestionable irreducibility of the same.
92.
Unless we understand the expression «good reasons» as an «achievement word» (G. Ryle), that is, taking for «good» not those reasons that «could be considered by all as being convincing» but only those that are actually correct; this second use, nonetheless, obviously presupposes already the truth as a condition, that is, would be the result of taking together 2) and 3).
93.
See H. Putnam: Reason, Truth and History, Cambridge 1981, pp. 54ff.
94.
This criticism is elaborated from different perspectives in the following writings of Wellmer: Ethik und Dialog [ED], Suhrkamp, Frankfurt 1986, pp.51-113; «Was ist eine pragmatische Bedeutungstheorie?» [WB], in: A. Honneth, T. McCarthy, et. al. (eds): Zwischenbetrachtungen. Im Prozess der Aufklärung, Suhrkamp, Frankfurt 1989, pp.318-372; «Wahrheit, Kontigenz, Moderne» [WKM], in: Endspiele: Die unversöhnliche Moderne, Frankfurt 1993, pp.157-177.
95.
In: G. Evans/J. McDowell (eds.): Truth and Meaning, Oxford 1976, pp.67-137.
96.
See footnote 6. Putnam has recently rejected (see «Comments and Replies», in: P. Clark/B. Hale (eds.): Reading Putnam, Cambridge, MA 1994, pp.242-295) his own conception of truth as «rational acceptability under ideal conditions». The only aspect that he maintains of the epistemical position is the intuition that a philosophically relevant explanation of the predicate `true' has to analyze our use of this predicate in its internal relation with concepts such as «rational acceptability», «epistemic conditions», etc. But now he rejects the veritable concession he had previously made to the epistemic position, namely, «the idea that truth could never be totally recognition-transcendent» (p.243). To explain this rejection Putnam appeals precisely to fallibilism: «Not only is truth not always recognizable by using anything that could be called a decision procedure, even under the best epistemic conditions; it is obvious that, in the case of empirical statements, decision as to truth are generally defeasable (and so are decisions as to whether one's epistemic position is good enough to decide on the truth of a statement)». (p.289, my emphasis) Given that this change of position is a recent one it remains to be seen if this argument will make Putnam, by his own logic, recognize that the very idea of «an `ideal' theory», that is, of a theory that could not be false cannot be sustained.
97.
That an antifallibilist interpretation of the opposition knowledge/error cannot be extracted from our use of such a pair of concepts is shown by the fact that it is neither contradictory nor problematic to say «I believed that I knew it». My belief that I know something can turn out to be as wrong as any other belief.
98.
This opposition can be understood both in the sense of the opposition «exist/does not exist» (relative to the reference of the terms) and in the sense of the opposition «is the case/is not the case» (relative to the truth of statements).
99.
To insist on the realist meaning of the concept of truth does not require adopting any concrete position in relation to the question of our epistemic access to the world. To that extent, the explanation of rational acceptability given by the discourse theory of Habermas, in itself -- that is, in so far as it merely gives an answer to the epistemical question -- is immune to these realist considerations. This can be seen in the central intuition of Habermas's discourse theory in relation to rational acceptability, namely: that «the satisfaction or non-satisfaction of truth conditions can only be stated through the argumentative cashing in of the corresponding validity claim.» (Die Neue Unübersichtlichkeit, Frankfurt 1985, p.228, my emphasis.) Undoubtedly, this discursive conception of rational acceptability is more convincing that any position of epistemic (or metaphysical) realism that has to appeal, in order to explain rational acceptability, to a correspondence or a causal relation between our beliefs and the «world in itself».
100.
In spite of the fact that Habermas introduces the formal concepts of world expressly as a correlate to the universal validity claims and even in contrast to relativist positions, such as Rorty's, he indicates that: «in the pragmatics of every use of language there is included the supposition of a shared objective world» (in: Nachmetaphysisches Denken, p.178). In order to defend such a position one would have to specify where o by means of what is this supposition anchored in every (cognitive) use of language. Such specification could be attained by means of a theory of reference that showed such a supposition as one of the inevitable normative presuppositions tied to the activity of referring that is proper to the cognitive use of language (as opposed to, say, the fictional use of language) as well as through a clarification of the realist sense of the concept of «truth», in which is shown the important normative function of such a supposition in our practices of revision and testing of the rational acceptability of our knowledge.
101.
This paper has been translated by by Miguel E. Vatter.
Trans. B. Jowett, The Dialogues of Plato, vol. 1 (New York: Random House, 1920), 807.
Republic 420. Op. cit. 683.
A classic example is Karl Popper. V. his The Open Society and Its Enemies, vol. 1, The Spell of Plato (London: George Routledge & Sons, 1945).