Acessibilidade / Reportar erro

Memory as acquaintance with the past: some Lessons from Russell, 1912-1914

Abstracts

Russell's theory of memory as acquaintance with the past seems to square uneasily with his definition of acquaintance as the converse of the relation of presentation of an object to a subject. We show how the two views can be made to cohere under a suitable construal of 'presentation', which has the additional appeal of bringing Russell's theory of memory closer to contemporary views on direct reference and object-dependent thinking than is usually acknowledged. The drawback is that memory as acquaintance with the past falls short of fulfilling Russell's requirement that knowledge by acquaintance be discriminating knowledge - a shortcoming shared by contemporary externalist accounts of knowledge from memory.

Bertrand Russell; Acquaintance; Memory; Russell's Principle; Externalism


A teoria russelliana da memória como contato (acquaintance) com o passado parece coadunar-se mal com definição do contato (acquaintance) como a conversa da relação de apresentação de um objeto a um sujeito. Mostramos como as duas concepções podem ser conciliadas mediante uma interpretação apropriada de 'apresentação', que tem a vantagem adicional de salientar uma proximidade maior que a usualmente reconhecida entre a teoria da memória de Russell e idéias contemporâneas sobre referência direta e pensamentos dependentes de objeto. O preço a pagar é o reconhecimento de que a memória como contato (acquaintance) com o passado não satisfaz o requisito russelliano de que o conhecimento por contato deva ser discriminativo - uma limitação que é compartilhada pelas abordagens externalistas contemporâneas do conhecimento derivado da memória.

Bertrand Russell; Conhecimento por contato; Memória; Princípio de Russell; Externalismo


ARTIGOS

Departamento de Filosofia da UFRGS. paulo.faria@ufrgs.br

ABSTRACT

Russell's theory of memory as acquaintance with the past seems to square uneasily with his definition of acquaintance as the converse of the relation of presentation of an object to a subject. We show how the two views can be made to cohere under a suitable construal of 'presentation', which has the additional appeal of bringing Russell's theory of memory closer to contemporary views on direct reference and object-dependent thinking than is usually acknowledged. The drawback is that memory as acquaintance with the past falls short of fulfilling Russell's requirement that knowledge by acquaintance be discriminating knowledge - a shortcoming shared by contemporary externalist accounts of knowledge from memory.

Keywords: Bertrand Russell, Acquaintance, Memory, Russell's Principle, Externalism

RESUMO

A teoria russelliana da memória como contato (acquaintance) com o passado parece coadunar-se mal com definição do contato (acquaintance) como a conversa da relação de apresentação de um objeto a um sujeito. Mostramos como as duas concepções podem ser conciliadas mediante uma interpretação apropriada de 'apresentação', que tem a vantagem adicional de salientar uma proximidade maior que a usualmente reconhecida entre a teoria da memória de Russell e idéias contemporâneas sobre referência direta e pensamentos dependentes de objeto. O preço a pagar é o reconhecimento de que a memória como contato (acquaintance) com o passado não satisfaz o requisito russelliano de que o conhecimento por contato deva ser discriminativo - uma limitação que é compartilhada pelas abordagens externalistas contemporâneas do conhecimento derivado da memória.

Palavras-Chave Bertrand Russell, Conhecimento por contato, Memória, Princípio de Russell, Externalismo

I

The theory of descriptions introduced in 'On Denoting' was, first and foremost, an analysis of the logical form of propositions containing what Russell called 'denoting phrases' - most noteworthy among them, to be sure, definite descriptions.

But Russell's groundbreaking paper contains at least three further layers of articulation of the theory, which one might think of as concentric circles disposed around that logical core, the theory of incomplete symbols.

In the innermost circle Russell provides, by way of confirmation of the proposed analysis of descriptions, solutions to his three "puzzles" about (respectively) substitutivity, the Excluded Middle, and negative existential statements.

In the intermediate circle, formed by the so-called 'Gray's Elegy Argument', Russell mounts a dubiously effective assault on Frege's doctrine of sense and reference, which turns out to be a devastatingly effective criticism of his own previous theory of denoting concepts, as presented in The Principles of Mathematics (1903).1 1 In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case.

Then the outermost circle is formed by the epistemological remarks which surface at the beginning, and then again at the end of the paper, thus providing what one may call its frame. (The frame was to be examined in some further detail in 'Knowledge by Acquaintance and Knowledge by Description' (1911), and in the corresponding chapter of The Problems of Philosophy (1912); in the three Monist papers 'On the Nature of Acquaintance' (1914), extracted from the unfinished Theory of Knowledge of 1913; and of course in the remainder of that ill-fated book, mercilessly murdered in the cradle by Wittgenstein.)

This paper is about the frame of 'On Denoting'. Its main theme is an all too often neglected aspect of Russell's notion of acquaintance: namely that - Russell's emphasis on presence notwithstanding - acquaintance can be of what is no more. In other words, there is no necessary connection between being acquainted with, and being in the presence of, an object. (That is, as will be shown, the gist of Russell's conception of memory in the period with which I will be concerned here.)

That there might be anything like being acquainted with an absent object is easily neglected because, to begin with, Russell himself introduces acquaintance as the converse of the relation of presentation: 'That is, to say that S has acquaintance with O is essentially the same as to say that O is presented to S' (RUSSELL, 1911: 152).

And then there is what one might call the rhetoric of presence, which to my mind reaches its highest pitch in the 1918 lectures 'On The Philosophy of Logical Atomism', where (to give a single example) one finds Russell explaining the concept of a particular as that of an entity 'which has that so rt of self-subsistence that used to belong to substance, except that it usually only persists through a very short time, so far as our experience goes' (RUSSELL, 1918-19: 202).

Such pronouncements encouraged a view of Russellian acquaintance (hence, of the related conception of what is involved in entertaining a Russellian singular proposition) which leaves no place for the idea of acquaintance with an absent object. That consequence is particularly unfortunate in that it makes no sense - indeed, makes sheer nonsense - of Russell's 1912-1914 conception of memory, according to which in memory we are acquainted with past objects and events. Indeed, the prevailing view in the literature seems to be that Russell's early theory of memory, as presented in The Problems of Philosophy, in 'On the Nature of Acquaintance' and so forth was obviously untenable.2 2 See, e.g. Pears 1975. And, to be sure, by the time of The Analysis of Mind (1921) a very different picture was already in place, but that will not be my concern here.

As I see things, that is all seriously misguided. Accordingly, my main claims in this paper are (I state them in the order of presentation, which I take to be more or less their order of increasing interest, hence of disputability): (1) that the view is misguided; (2) that it is prompted by an unfortunate amalgamation (partly encouraged by Russell himself) of acquaintance and presence; (3) that there is a way to construe Russell's characterization of acquaintance as the converse of presentation which makes it consistent with his claim that an object of acquaintance may be 'in the present, in the past, or not in time at all' (RUSSELL, 1914: 127); (4) that the proposed interpretation brings Russell's views (in the relevant period) closer to contemporary views on direct reference and object-dependent thinking than is usually acknowledged; and (5) that a drawback of any such reading is likely to be the acknowledgement that memory as acquaintance with past objects and events falls short of fulfilling a requirement imposed by Russell himself on knowledge by acquaintance: namely, that it be (what knowledge by description alone should not, by Russell's lights, be granted to be) discriminating knowledge.

II

As I remarked at the start, the Theory of Descriptions is chiefly (although, as I will be insisting, not just) an analysis of the logical form of propositions containing 'denoting phrases' such as (in Russell's famous list) 'a man, some man, any man, every man, all men, the present King of England, the present King of France, the center of mass of the solar system at the first instant of the twentieth century, the revolution of the earth round the sun, the revolution of the sun round the earth' (RUSSELL, 1905: 415). Much of what I'll have to say will focus on the kind of denoting phrases exemplified by the last five entries in Russell's list: that is, on definite descriptions. Although the focus is not particularly original, I have two excuses for choosing it when aiming at a discussion of Russell's conception of memory as acquaintance with the past. First, I want to make use of the fact that in showing definite descriptions not to be singular terms but rather predicates of a sort (more specifically quantifiers, that is second-order predicates, at any rate expressions of logical generality), Russell manages to explain how their meaningfulness, as opposed to that of genuine singular terms, is not answerable to whether something exists; second, that the explanation implies a conception of the semantics of singular terms - hence of the requirements for singling out a particular object in thought - which contains the seeds of much contemporary thinking about direct reference, singular thought, and object-dependence.3 3 I follow Evans (1982, 1) and Neale (1990, 15) in using 'singular terms' interchangeably with 'referential expressions', thus excluding descriptions by stipulation. The stipulation, admittedly at variance with current usage in the philosophy of language, is meant to stress the logical gist of Russell's repudiation of 'the unity of the intuitive category of referential expressions' (Evans 1982: 42): namely, that descriptions are quantifiers rather than referential devices. To be sure, this holds only of descriptions in attributive position, and in sticking to the stipulation I don't mean to be taking a stand on the issue whether these are the only cases of logico-semantic (as opposed, say, to merely pragmatic) significance. So-called referential uses of definite descriptions, however they should be accounted for, fall beyond the scope of Russell's theory anyway, and are accordingly left out of the present discussion.

In a word, treating descriptions as quantifiers Russell relocated them from the logical category of singular terms to that of expressions whose extension is determined as full or empty according to the satisfaction of at least one predicate. Now the counterpart of that relocation is the acknowledgment of a class of logically proper names, and of a primitive mode of designation, irreducible to 'knowledge by description'. In Principia Mathematica, Russell introduces that mode of designation through the notion of direct representation:

Whenever the grammatical subject of a proposition can be supposed not to exist without rendering the proposition meaningless, it is plain that the grammatical subject is not a proper name, i.e., not a name directly representing some object. (WHITEHEAD & RUSSELL, 1910: 66)

That way of accounting for the fundamental distinction between singular terms and descriptions implies that if there are any genuine singular terms - hence, if there is any singular proposition - then there are object-dependent propositions: propositions such that, should there be no such thing as the object it is about, the proposition itself would be impossible.

And that's how the logical distinction between descriptions and singular terms necessitates its epistemological counterpart, the distinction between the two varieties of 'knowledge of things' - by acquaintance and by description. The latter distinction is irreducible if so is the former.

Now there is admittedly an awful lot in the epistemological background of the Theory of Descriptions which virtually nobody is prepared to accept without qualification anymore. Indeed, for many a contemporary analytical philosopher the very idea of a mode of cognition like Russellian acquaintance would seem to have been erased from the logical space of admissible possibilities as a result of the "assault upon immediacy" led by the likes of Wittgenstein, Quine, or Sellars, which precipitated the decline of empiricism in the analytic tradition. Wittgenstein's criticisms of ostensive definition and the "privacy" of experience; Quine's denouncement of the "dogmas" of analyticity and reductionism; Sellars's demolition of the "myth of the Given": all that conspired to promote the emergence, in the half-century following the end of World War II, of a philosophical culture deeply hostile to the metaphysics of experience which underwrites Russell's conception of knowledge by acquaintance.

I think the tide has been changing of late, in Russell's favor - witness the current debates about non-conceptual content, or the spread of "externalism" in semantics and the philosophy of mind; or, even more to the point, the variety of recent proposals to restore the very idea of knowledge by acquaintance to philosophical respectability (e.g. Mark Johnston, Jim Pryor, Michael Martin).4 4 See Martin (2001) and his forthcoming book on perceptual appearances. Mark Johnston has been dealing with acquaintance under the head of 'the manifest', the topic of his forthcoming book to bear that title. Jim Pryor has lectured on his 'epistemic theory of acquaintance' at a number of places, yet no printed statement of the theory is, to the best of my knowledge, available for the time being. But that is not my concern here. What matters to me at this point is to stress that Russell's rather unfashionable ways with sense-data and other private objects of acquaintance are anything but an external appendage to his always fashionable philosophy of logic - that there is, to use a phrase he deeply disliked (and which became, if only for that very reason, a favorite of his rebellious pupil Wittgenstein) an "internal relation" linking the logical distinction between singular terms and descriptions and the epistemological distinction between the two corresponding ways of knowing an object.

But I am claiming more for Russell's acquaintance-based epistemology. Specifically, I'd like to draw attention to the fact that, having introduced the idea of singular reference in terms which imply what we nowadays think of as object-dependence, Russell saw clearly something which (courtesy of over three decades of debates about direct reference, rigidity, content and context) we are now used to. I mean the dependence of thought-contents on the satisfaction of conditions over which the thinking subject lacks control - which, moreover, may entirely evade his epistemic access.5 5 The point here is that what makes it the case that an object-dependent thought is about a rather than about its phenomenologically indiscernible "twin" a* is a matter of the subject's being properly related, both causally and epistemically, to the object. Russell's notion of acquaintance was meant to encompass both the causal and the epistemic requirements, the latter taken in its strictest possible form as a requirement of discriminating knowledge; hence the reading of the so-called Russell's Principle (that a subject cannot entertain a thought about something unless he knows what it is he is thinking about) as requiring that 'the subject must have a capacity to distinguish the object of his judgment from all other objects' (EVANS, 1982: 89, my emphasis). This is arguably too strong a condition to be imposed by any plausible account of object-dependent thinking; moreover, it is certainly something content externalists should know better than grant to their critics. For all that, I open-eyedly proceed, throughout this paper, as if a relaxed version of Russell's Principle were not available; in particular, I say nothing about the advantages of substituting reliable discrimination for incorrigible discrimination (say, along the lines sketched long ago by Martin Davies and embraced since by Neale and others); or of taking the quantifier in "all other objects" to range over a set of contextually relevant alternatives, rather than over an unrestricted domain. I have allowed myself so to proceed for two reasons, of which the first is purely historical: seeing why Russell could not find a place in conceptual space for anything like what Burge and others call 'partial understanding' enables one to see the rationale for his verdict about the outcome of lacking discriminative knowledge (in the strongest possible sense) of an object: namely that no singular thought about that object is available to the subject (see the Bismarck example in the following paragraph, and then the brief account of McDowell's criticism of Russell in section IV below.) My second reason bears on the current status of the debate surrounding the so-called Memory Argument against content externalism (on which see section V below): arguably, much of what is at stake in assessing the force of Boghossian's attempted reductio of compatibilism hinges on the plausibility of building one or another form of a discrimination requirement into the account of what it takes to know the content of one's own mental states.

Accordingly, Russell was, to the best of my knowledge, the first philosopher ever to have described in any detail the striking situation where a subject is in a position to know that there is a possible thought satisfying a particular description, yet is unable to think that thought, due to lack of the appropriate cognitive access to the constituents of its content (of the proposition, in Russell's peculiar sense of the word, which is thereby taken as true). The idea makes its appearance in the course of Russell's brief excursus on the privacy of experience in 'Knowledge by Acquaintance and Knowledge by Description' (and in the corresponding chapter of The Problems of Philosophy). There we are invited to think of Prince Otto Bismarck, know to us as, among other things, 'the first Chancellor of the German Empire', and to consider how we are able to do that. And the answer is, in brief: not the way Bismarck, and only Bismarck himself, could do it. This is how Russell puts it:

[...] when we say something about Bismarck, we should like, if we could, to make a judgment which Bismarck alone can make, namely, the judgment of which he himself is a constituent. In this we are necessarily defeated, since the actual Bismarck is unknown to us. But we know that there is an object B called Bismarck, and that B was an astute diplomatist. We can thus describe the proposition we should like to affirm, namely 'B was an astute diplomatist', where B is the object which was Bismarck. What enables us to communicate in spite of the varying descriptions we employ is that we know that there is a true proposition concerning the actual Bismarck, and that, however we may vary the description (so long as the description is correct), the proposition described is still the same. This proposition, which is described and is known to be true, is what interests us; but we are not acquainted with the proposition itself, and do not know it, though we know it to be true.6 6 Russell 1911: 158 (1912: 57). A related possibility is discussed, in the framework of a 'direct reference' view of names, by Keith Donnellan: we may sometimes be in a position to know that a certain sentence expresses a truth without knowing the truth that it expresses (DONNELLAN, 1977). As for Russell, the problem for Donnellan is raised by cases where we lack the mode of access to the relevant object which alone would enable us to think what in each case we fail to think. More on that below.

To be sure, many would feel inclined to think of such predicament as one more unfortunate consequence of Russell's commitment to such "creatures of darkness" as private objects, private meanings and other beetles in the mind's box (see e.g. DIAMOND, 2000). I see things otherwise. Actually, I will claim that getting rid of "the myth of the inner", of the privacy of the mental, of immediacy and presence, and going external, public and historical has only made things worse. For the public objects to which we refer in a shared, historical world are, on a plausible view about the constitution of thought-contents (which goes under the labels 'externalism' or 'anti-individualism') as apt to evade the reach of our attempts at meaning them as Russellian individuals were. And if I am right there's no "private language argument" which could possibly be of any help here.

My aim is just to state the problem; I don't know the solution, if any there is. But the presentation of the puzzle may well provide us with some insight into an all-too-often neglected aspect of Russell's philosophy, which I would claim should be deemed an integral part of his legacy, and hence of what it is ours to inherit - that is to accept or relinquish, in either way to come to grips with - in his philosophy. I mean Russell's deep-seated uneasiness with time and history.7 7 See, for a particularly striking statement, Russell (1904).

III

I claimed that the distinction between Russell's two kinds of knowledge of things (by acquaintance and by description) is as irreducible as that between singular terms and descriptions. In other words, what I called the 'frame' of 'On Denoting' bears a more intimate relation than meets the eye to the picture itself.8 8 To be sure, the main elements of the frame were an integral part of Russell's philosophy long before 'On Denoting'. The idea of a direct, immediate, non-conceptual mode of cognition, and of its priority over judgment (which reaches its most striking formulation in the Principle of Acquaintance discussed in Russell 1911) was part and parcel of the Moore-Russell anti-idealistic picture of judgment as grounded on receptivity.

No mention is made in 'On Denoting', though, of time; accordingly, the reader is left with no hint as to the respective roles of acquaintance and description in our knowledge of the past.

The frame is composed of two paragraphs placed at the beginning and at the end of the paper in nice symmetry: they are, respectively, the second and the penultimate paragraphs in the text.

The second paragraph follows the general statement, which opens the paper, of the problem of giving a correct account of denoting phrases - a problem Russell claims to have solved with the theory about to be presented. Before proceeding to the presentation, though, we have our attention drawn to the significance for epistemology (in addition to 'logic and mathematics') of the theory. The significance lies in its bearing on the distinction between the two kinds of knowledge of things: 'The distinction between acquaintance and knowledge about is the distinction between the things we have presentations of, and the things we reach only by means of denoting phrases' (RUSSELL, 1905: 415). No distinction between kinds of acquaintance is introduced, other than that between acquaintance in perception and acquaintance in thought.9 9 The latter had been playing a crucial role in Russell's account of logical and mathematical knowledge since his 'revolt against idealism'. Compare this often-quoted statement from the Preface to The Principles of Mathematics: 'The discussion of indefinables - which forms the chief part of philosophical logic - is the endeavour to see clearly, and to make others see clearly, the entities concerned, in order that the mind may have that kind of acquaintance with them which it has with redness or the taste of pineapple' (RUSSELL, 1903: xv). A paradigm case of inaccessible objects, which can only be known by description, is introduced: not very surprisingly, it turns out to be other people's minds - 'a very important instance', says Russell, of entities which 'there seems to be no reason to believe we are ever acquainted with' (ibid.).

In the next to last paragraph, which forms the second half of the frame, the Principle of Acquaintance is presented as a result of the theory of incomplete symbols. What that means is hardly disputable: Russell sees the new account of denoting phrases, which replaces the highly unstable theory of 'denoting concepts' put forward in The Principles of Mathematics, as inheriting from its predecessor the role of providing a logical foundation for the principle of the priority of acquaintance over conceptual thinking: 'Thus in every proposition that we can apprehend (i.e. not only in those whose truth or falsehood we can think about), all the constituents are really entities with which we have immediate acquaintance.' (1905: 427). The principle dictates, as will be argued at length in 'Knowledge by Acquaintance and Knowledge by Description', and illustrated there by the Bismarck case, the acknowledgment of a variety of cases of inaccessible thoughts. In such cases, 'although we can form propositional functions C(x) which must hold of such and such a material particle, or of So-and-so's mind, yet we are not acquainted with the propositions which affirm these things that we know to be true, because we cannot apprehend the actual entities concerned.' (Ibid.)

I think it is fair to say that the cases Russell has in view, both here and in the later, more extended discussions of the epistemological import of the Theory of Descriptions, are usually cases of what one might call standing inaccessibility. It is certainly no accident that a modal verb occurs in the sentence just quoted from 'On Denoting', as part of Russell's first sketchy account of the very possibility of such cases. Ditto for the Bismarck case: in the attempt to make the judgment 'which Bismarck alone can make' we are 'necessarily defeated, since the actual Bismarck is unknown to us' (RUSSELL, 1911: 158. My italics.) Such cases illuminate the significance of the Theory of Descriptions for the account of the structure of empirical knowledge, which was to become a main concern of Russell's in the years following the completion of Principia Mathematica. As he states near the end of 'Knowledge by Acquaintance and Knowledge by Description': 'We have acquaintance with sense-data, with many universals, and possibly with ourselves, but not with physical objects or other minds' (1911: 167). And in the corresponding chapter of The Problems of Philosophy: 'The chief importance of knowledge by description is that it enables us to pass beyond the limits of our private experience. In spite of the fact that we can only know truths which are wholly composed of terms which we have experienced in acquaintance, we can yet have knowledge by description of things which we have never experienced' (RUSSELL, 1912: 59).10 10 We have here in a nutshell the Russellian project (to be pursued further by Carnap, Goodman and others) of a 'logical construction of the world'. The project is described in Russell's Harvard Lectures of 1914 Our Knowledge of the External World as a Field for Scientific Method in Philosophy and in the important paper, published that same year, 'The Relation of Sense-Data to Physics' (reprinted, along with 'Knowledge by Acquaintance and Knowledge by Description', in the collection Mysticism and Logic and Other Essays).

Yet, knowledge by description will prove just as useful in cases of contingent inaccessibility, whether temporary or definitive. In particular, it may be all that one's left with in cases of lost access. (If only, that is, one's left with some definite description to make for the loss of acquaintance.)

In the remainder of this paper I will concentrate on cases of lost access. They will mainly concern, predictably enough, memory failures, certainly including but not restricted to forgetting. (Actually, the cases I'm most interest in are those where the loss is not in any intuitive sense describable as a case of forgetting.) As promised, I'll take my clue from a vindication of Russell's view of memory as acquaintance with the past.11 11 This step is not going to be particularly hard to take. A pretty good job has already been done in Martin 2001, to which the reader is referred. This is not a point of merely historical interest. As I will try to show, Russell's view is both plausible (way more, to my mind, than any version of the still widely accepted view of memory as representation) and attractive in that it nicely matches a related view, itself originating in Russell's work, of the semantics of singular terms. I mean, of course, the very idea of direct reference.

IV

I said at the beginning that the prevailing rejection of Russell's 1912-1914 theory of memory is prompted by a tacit assimilation - to some extent, sad to say, encouraged by Russell himself - of acquaintance and presence. Accordingly, the very idea that one could be acquainted in Russell's sense with an absent object has tended to drop out of sight all but entirely.

In the beginning of this tale of neglect (which I won't tell here, just gesture at) there is, of course, the informal definition of acquaintance as the converse of 'the relation of object and subject which constitutes presentation' (RUSSELL, 1911: 152). The prevalence of examples of acquaintance involving either sense-perception or its intellectual counterpart (what in 'On Denoting' was called 'acquaintance in thought') has encouraged the view that there can only be acquaintance with what is, so to speak, 'before the mind', currently presented to the subject, present.

The resulting picture of Russellian acquaintance is illustrated by some of John McDowell's writing on Russell. (I make a point of stressing that I choose McDowell as a scapegoat out of respect: if such a keen and attentive reader of other philosophers can be found guilty of misreading here, the possibility that Russell himself should be guilty of more than his usual share of obscurity - which, yes, is a lot to say - is better taken seriously).

Let a couple of examples suffice. In McDowell (1970) Russell (along with Wittgenstein) is saddled with a form of Plato's problem about false identity statements. Take two arbitrary objects, a and b; four possibilities are open: I know both; I know a but not b; I know b but not a; I know neither. In the first case, I cannot possibly mistake a for b (since I know both); in all the other cases, I am debarred from judging (either truly or falsely) that a and b are the same through lack of knowledge of at least one of them (Theaetetus 188ac). The principle underlying the argument as Plato presents it turns out to be a predecessor of what Gareth Evans called 'Russell's Principle', namely, that 'it seems scarcely possible to believe that we can make a judgment or entertain a supposition without knowing what it is that we are judging or supposing about' (RUSSELL, 1911: 159).12 12 The principle is explicitly appealed to by Plato at a later stage in the Thaetetus, as a step in an argument designed to show that the condition to be satisfied by a true judgment in order for it to qualify as knowledge could not possibly be the ability to discriminate the object of the judgment from other objects, for the plain reason that such ability is a requirement of any judgment (whether true or false), hence nothing that could possibly be added to a true judgment so as to turn it into knowledge. See Theaetetus 208e-209d, and McDowell's comment in McDowell 1973: 255-256. The principle is deemed by Russell to be 'the chief reason' for accepting the Principle of Acquaintance. In McDowell's interpretation, the "blind spot" which would prevent both Plato and the logical atomists from adopting the obvious "Fregean" way out of the paradox (I may think of different objects under the same mode of presentation; I may think of a single object under different modes of presentation) is a paradigm case of 'captivity by a picture'. The picture here is that of acquaintance as a 'transparent' mode of access to its object, unhindered by the vagaries of conceiving - of knowing "under a description". Now, the crucial point here is that, on the side of the object, what is supposed to ensure such transparency is full givenness: a Platonic-Russellian object of acquaintance would have to be wholly given to the inspecting gaze - no hidden aspects, no back sides (no wonder, one feels like adding, Russell should have considered surfaces as serious candidates to the role of sense-data!). Again in McDowell 1986 Russell is found guilty of systematically refusing to accept the possibility of empty singular terms (and of the corresponding illusion of entertaining a singular thought) out of a tendency to evade that most anti-Cartesian consequence of the very idea of an object-dependent thought: what one might call, imitating Kripke, the riskiness of meaning.13 13 To the extent that being able to mean a particular object is matter of being properly related to that object - e.g. of partaking in a unique and non-disrupted historical chain of uses of a designator for that object - a speaker's atempt to mean an object is vulneable to failure through ungroundedness: that is the sense in which meaning something is apt to be a 'risky' business. I am here extending to subsentential components the notion of groundedness as it features in the explanation of the semantical paradoxes (actually, ungroundedness of statements is often, if by no means exclusively, derived from ungroundedness of subsentential components, most notably singular terms). Anyway, here's Kripke: 'an adequate theory must allow our statements involving the notion of truth to be risky; they risk being paradoxical if the empirical facts are extremely (and unexpectedly) unfavorable. There can be no syntactic or semantic "sieve" that will winnow out the "bad" from the "good" ones' (1975: 692). Accordingly, the technique of incomplete symbols would provide Russell with an all-purpose safety device: short of being able to certify that the purported object of reference is fully given, beyond reasonable doubt, as only objects of acquaintance can be, it's not a genuine singular proposition (hence, not any thought about any particular object) that one is entertaining. The Theory of Descriptions degenerates (with the reconception of most ordinary singular terms as disguised descriptions) in the service of a fantasy of having the sphere of thought safe from the threat of failure, of exposure to a perhaps uncooperative world; and of avoiding the risk of the world's failing to supply an object for there to be a thought about it by telling the world to get lost and procuring a surrogate for the missing object.14 14 Lest my brutally sketched presentation be dismissed as a hyperbolic parody of McDowell's talk about a 'favor from the world', I hasten to note that (to give one single example) the rhetorically much austerer Neale shares essentially the same diagnosis: 'Bent on exorcising the possibility of someone entertaining an object-dependent proposition when, in reality, there is no object which the purported proposition is about, Russell maintained that only a sense-datum could be trusted to be the "subject" of such a proposition. It is by now pretty generally agreed that this position is untenable, and I shall not spend any time addressing any of the familiar objections to it' (NEALE, 1990: 18). No mention is made of Russell's ever having considered the possibility of anything like being acquainted with a (now) absent object.

Accordingly, McDowell talks, as of an improvement upon Russell's views, of 'the possibility of liberalizing the notion of acquaintance outside the case of perceptually presented objects' (MCDOWELL, 1986: 232). He goes on to suggest - quite pertinently, as I have already hinted at - that the historical chain of uses of a proper name as described by Kripke might be thought of as a mode of acquaintance with the past on that liberalized view; and deplores (again rightly to my mind) Gareth Evans's resistance to such idea, surmising that it was grounded 'perhaps on the basis of an excessive individualism' (1986: 232).

Promising as that sounds, I want to submit that one should grant Russell with having at least noticed the difficulty - even if, again, Russell himself is to blame for the near invisibility of his attempt to step out of the "metaphysics of presence" of which he's supposed to have been a hostage.15 15 In what follows I describe, as best as I can, Russell's attempt to dissever acquaintance and presence to consciousness. It should go without saying that the attempt, even if successful, is predicated on a thoroughly individualistic picture of the mind, with the consequence that nothing beyond the subject's own past can possibly fall in the scope of what Russell is prepared to count as retained acquaintance. To that extent, the acquaintance theory of episodic memory (the only kind of memory Russell deals with) falls radically short of providing the mode of acquaintance with the past that McDowell has in view: a thought about, say, Aristotle will still be deemed a general thought, to be dealt with, on a par with thoughts about nonexistent objects, by the theory of descriptions. I thank Maite Ezcurdia for urging me to acknowledge the extent to which McDowell's proposal remains apposite even granted my plea on Russell's behalf.

For, as I said at the start, there is after all a way of construing Russell's characterization of acquaintance as the converse of presentation which makes it consistent with his later claim that an object of acquaintance may be 'in the present, in the past, or not in time at all' (RUSSELL, 1914: 127).

To see how that might be, one should begin by taking into account the fact that the 'converse of presentation' picture was from the very beginning qualified by Russell's expressed qualms about the 'associations and natural extensions' of the word 'presentation': 'To begin with, as in most cognitive words, it is natural to say that I am acquainted with an object even at moments when it is not actually before my mind, provided it has been before my mind, and will be again whenever occasion arises' (RUSSELL, 1911: 152).

Start with ordinary language: we make acquaintance of people and places and things and we don't say we have lost such acquaintance when they are no longer present. We may come to lose it, but absence is not a sufficient condition of such loss. What Russell brings into play here is the idea of retained acquaintance - I borrow the phrase from Michael Martin (2001). Nothing much is said in 'Knowledge by Acquaintance and Knowledge by Description' about the conditions of such retention, apart from the intriguing suggestion that acquaintance in absentia stands to "live" acquaintance as a standing cognitive state stands to an occurrent one - that is, at any rate, how I read the claim that the sense in which we retain acquaintance of an object which is no longer present is 'the same sense in which I am said to know that 2+2=4 even when I am thinking of something else' (Ibid. My italics.)

It's only in The Problems of Philosophy, published the following year, that retained acquaintance is called by its name, and a sketch is offered of Russell's reply to the question, how that is possible at all. The discussion of acquaintance in the chapter which bears the title of the Aristotelian Society paper starts, as befits an exposition aimed at a non-professional audience, with sense-data - all mention of acquaintance with abstract objects studiedly postponed, givenness in perception providing here the archetype of presentation. But then the first extension beyond sense-data which Russell now introduces is acquaintance by memory (1912: 48).16 16 That there is such a thing as direct awareness of what one has experienced Russell deems 'obvious'; in spite of which he goes on to sketch an argument to the effect that immediate knowledge by memory (acquaintance with the past) is the source of all knowledge concerning the past: 'without it, there could be no knowledge of the past by inference, since we should never know that there was anything past to be inferred' (1912: 49).

The most striking feature of the theory of memory as acquaintance with the past is the uncompromising rejection of the traditional, and still largely prevailing, conception of memory as representation: specifically, as the present image of a past object or event. Russell is careful to acknowledge the role of images in recalling, but he is adamant that such images cannot possibly constitute the memory itself. The sheer fact that 'the image is in the present whereas what is remembered is known to be in the past' (1912:115) should suffice to show that; but Russell goes on to appeal to the fact that we often know, with some degree of certainty, how accurate our present image is of the remembered object - a psychological fact ostensibly involving a comparison between both which would be impossible unless the remembered object, as distinct from the image, were somehow 'before the mind': 'Thus the essence of memory is not constituted by the image, but by having immediately before the mind an object which is recognized as past' (Ibid.)17 17 The same point recurs in the 'Definitions and Methodological Principles in Theory of Knowledge', published in The Monist in 1914 in sequence to the three papers 'On the Nature of Acquaintance'; taken together, these constitute the first four chapters of the 1913 Theory of Knowledge. 'Being in the past' writes Russell, 'is not an intrinsic property of an object, but a relation to the subject; thus memory will have to be distinguished from sense and imagination as a different relation to objects, not as the same relation to different objects'. (RUSSELL, 1984: 56). Russell goes on to claim that 'memory is excluded if we say that the acquaintance we are concerned with must not be of an object given as past' ( Ibid. My italics.)

In the second of the three Monist papers 'On the Nature of Acquaintance' the rejection of the representation theory of memory is linked to Russell's criticism of the 'veil of ideas': the picture of empirical knowledge which posits ideas (mental representations) as the interface between mind and world. The criticism was launched in 'Knowledge by Acquaintance and Knowledge by Description', where the doctrine of ideas is exposed as resting upon (a) 'a failure to form a right theory of descriptions', combined with (b) 'a dislike of relations' (1911: 160). The former shortcoming accounts for the illusion that all our cognitive commerce with the world is mediated by concepts - in Russell's terms, that all cognition is achieved under a description. The latter (as much a legacy of the idealistic tradition as the former) accounts for the view that, since judging is a mental act, the constituents of a judgment must be in the mind - in other words, that being about Julius Caesar is an intrinsic property of my judgment that Julius Caesar was assassinated. If we appreciate the point of Russell's presentation - in the next to last paragraph of 'On Denoting' - of the Principle of Acquaintance as a result of the Theory of Descriptions, I think the significance of the criticism is clear: Russellian acquaintance is direct cognitive access to objects (of perception, memory and so forth), much as Russellian (logically proper) naming is direct reference.18 18 The criticism of the 'veil of ideas', if taken seriously, should have erased from our accounts the picture of Russell as a "Cartesian" thinker, in the (historically inaccurate) sense in which that adjective has been most often used in the analytic tradition. For an enlightened version of the charge of "Cartesianism" see McDowell 1986: 236-247. And I believe the criticism sheds light on both Russell's hostility to Frege on Sinn and Bedeutung and on his view that the present image of a past object cannot be a memory of the object.

That images play the role of Fregean senses in the theory of memory is argued at some length in Theory of Knowledge, Chapter VI ('On the Experience of Time'), where Russell raises the question: 'Does our knowledge of the past involve acquaintance with past objects, or can it be accounted for on the supposition that only knowledge by description is involved in our knowledge of the past?' (RUSSELL, 1984: 71). If the former, then some propositions of the form 'This is past', where the demonstrative refers to an object of acquaintance, must be true; if the latter, then all knowledge of the past is expressible in propositions to the effect that entities satisfying some propositional functions existed in the past. Now, says Russell, the view that all knowledge of the past is knowledge by description might be maintained by introducing images: 'it might be said that we have images which we know to be more or less like objects of past experience, but that the simplest knowledge we have concerning such objects is their resemblance to images' (Ibid.) The most basic kind of cognition concerning the past would then take the form 'this resembles something-in-the-past' where the demonstrative refers to an image, and 'something' is a bound variable.

Unlike such knowledge by description of the past, immediate memory is 'a two-term relation of subject and object, involving acquaintance, and such as to give rise to the knowledge that the object is in the past' (1984: 70). The definition is not intended to settle the issue whether there is such a thing as immediate memory. However, Russell argues, it seems highly unlikely that there should be no such thing: 'It is indubitable that we have knowledge of the past, and it would seem, although this is not logically demonstrable, that such knowledge arises from acquaintance with past objects in a way enabling us to know that they are past' (1984: 70). The idea here seems to be that we would not be able to infer from our knowledge of the present any proposition concerning the past (not even as a hypothesis) unless we were currently conscious of something past as past.

Now what can 'knowing something past as past' look like? Russell is, to say the least, rather laconic about that, but one promising suggestion is to be gathered, it seems to me, from his uncompromising rejection of the assimilation of memories to images.

I think we can see why images, even those of past objects and events, cannot possibly constitute the memories of such objects and events. To be sure, entertaining an image is as much an intentional act as are perceiving or remembering. Yet an essential feature of both perception and remembrance is clearly not a necessary condition for entertaining an image.

An object of perception is essentially located in time and space: it is perceived as present (in time) and as occupying a place in what one might call the "neighborhood" of the perceiving subject: the region of space comprising all points accessible in principle to the subject. The clause 'in principle' is to be taken strictly - think, for a trite example, a long extinct star (I see it there, I see it there now). Whenever I perceive an object, in other words, I locate it in a time which is present (my present) and in a space which I am in principle able to traverse (and if as a matter of fact I cannot, it's only what Russell used to call "our medical limitations" which are to blame).

Likewise, remembered objects are essentially located - even if vaguely so; for even when there is uncertainty about their location, they are de iure placed sometime (in the past) and somewhere. (And in the case of episodic memory - the only kind of memory Russell deals with - in some place where I, the subject of the remembrance, have been.)

Images, on the other hand, are not essentially placed in a spatiotemporal framework; moreover, as long as we take them as mere images, they essentially lack spatiotemporal location. In that respect, images resemble fictional objects. Like fictional objects, images are insulated: the space of imagery has, so to speak, no outside. And it is, for that very reason, essentially (I feel like saying 'ontologically', since this is not about epistemic limitations) incomplete. The reader of Joyce's Ulysses who would ask what happened to Molly Bloom the day after June 16, 1904 would not know what a novel is - there's no day after for Molly Bloom. Or think of the visitor in the Louvre who wanted to have a look at the back of the Gioconda. That lady has no back - which doesn't mean she lacks anything.

Be that as it may, the important point is that acquaintance is not wedded to presence. And that should be enough to, first, acquit Russell from the charge of indulging in the "metaphysics of presence"; and then proceed to ask how dear the acquittal comes.

The crucial question now is whether acquaintance with the past could possibly accomplish this critical task ascribed by Russell to knowledge by acquaintance as opposed to knowledge by description: to ensure the satisfaction of 'Russell's Principle'. According to Russell's Principle, remember, it is a necessary condition for a subject S to think about an object a that S be in a position to tell a from other objects.19 19 And, on Russell's construal, this amounts to being in a position to incorrigibly discriminate that object 'from all other things' (see note 6 above).

The principle made its appearance, significantly, in the context of the sole sketch of an argument ever given by Russell for the Principle of Acquaintance (introduced in 'On Denoting', as we saw, as a result of the analysis of denoting phrases as incomplete symbols): we should think that every proposition that we understand must be wholly composed of constituents we are acquainted with, for 'it seems scarcely possible to believe that we can make a judgment or entertain a supposition without knowing what it is that we are judging or supposing about' (RUSSELL, 1912: 58).

That is not much, but I think an explicit argument, not unlikely to resemble what Russell may have had in mind, can actually be derived from the Theory of Descriptions; and that it brings us very close to vindicate Russell's claim that the Principle of Acquaintance is a 'result' of that theory. Trouble is, I regret to say, it is not easy to see how the argument can be made to apply to recalled, as opposed to currently perceived, objects without overstepping the limits of Russellian acquaintance even in its extended form - a difficulty which will presently bring me to my last and thorniest point.

The idea is that knowing an object by description (knowing that such-and-such propositional function is satisfied by a single object) is not a sufficient condition to tell that object from every other thing, no matter how exhaustive the description. For an object a such that (F0a ? F1 1 In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case. a ? ... ? F?a) might have a "twin" a* such that (F0a* ? F1 1 In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case. a* ? ... ? F?a*): the twins a and a* would be descriptively indiscernible although numerically distinct, as Kant's waterdrops.20 20 See the Critique of Pure Reason, A 273 / B 328. My original example was that of incongruous counterparts - say the left and the right hand as they feature in Kant's 1770 Dissertation. To which it was rightly objected that different descriptions can be given of the right and the left hands by specifying a coordinate system with the right multiplicity. That much is granted, yet it misses Kant's point, as nothing in the resulting pair of descriptions would single out one of them as a description of the right rather than the left hand, or vice versa (See SEVERO, 2005 for a thorough and illuminating discussion of the issue). For all that, the example was certainly a distracting one, which I am glad to let go. And since, by assumption, we would only be able to think of a (and, likewise, of a*) as the x such that (F0x ? F1 1 In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case. x ? ... ? F?x), nothing in the contents of our thoughts would make it the case that it was a, rather than a*, we were thinking of. The next step would consist in assuming (as Strawson invites us to do in Chapter I of Individuals) that every object in the universe has at least one descriptively indiscernible counterpart. In such circumstances, we would not be able to entertain any singular thought. And since, by assumption, no other identifying procedure would be available to us, that intrinsic indeterminacy in the contents of our thoughts (the trademark of their logical generality) would not admit of supplementation by any such non-intellective acts of determination as, say, the conversio ad phantasmata of Thomist epistemology. We would then be liable to find ourselves in the plight of thinking without quite knowing what we were thinking of.

When, on the other hand, we are able to locate in space the objects of our thoughts, the question 'Which x such that (F0x ? F1 1 In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case. x ? ... ? F?x)?' can be thoroughly settled by ostension: 'This one!'

No analogous argument seems to apply to the case of remembered objects, though. For suppose the past object a has a qualitatively identical twin a*. And suppose I remember a and now I come upon a*.21 21 The scenario I am appealing to was introduced by Gareth Evans in The Varieties of Reference; his example was that of two indiscernible steel balls: see EVANS 1982: 90. Demonstrative presentation is of no avail to the task of deciding whether a = a*. In the preceding scenario, where the twins were simultaneously presented to the subject, the 'Which?' question was settled by pointing to one of them and saying 'This one'. Now, presented with a single object and having nothing besides the memory of its indiscernible twin, the subject is left with two demonstratives - 'this' and 'that' - and an identity problem.

The difficulty here is analogous to that which besets the externalist conception of mental content: given two phenomenologically indiscernible though actually distinct objects (kinds, substances), there is no telling "from the first person perspective" alone whether or not they are the same.

V

Externalism is, in a very rough sketch, the thesis that the contents of mental states are, at least in part, determined by the relations the subject stands in to his environment. In other words, content is a relational rather than an intrinsic property of mental states. For the purposes at hand, the salient feature of that thesis is the consequence that mental content is identified with respect to circumstances which may not be (which, possibly, very often are not) epistemically accessible to the subject herself. The challenge for externalists is then to show how that consequence could be reconciled with "first person authority": i.e., with the intuitive assumption that each person has a privileged access to the contents of her own thoughts.

The standard externalist response to that challenge is a view put forward, in different shades, by Tyler Burge, Donald Davidson, Sidney Shoemaker, and others. According to that view, the propositional content of the "first order" thought p is wholly incorporated into the self-ascription (the "second order") thought 'I think that p' - whatever that "first order" content may be.22 22 See DAVIDSON, 1984, 1987; BURGE, 1988; SHOEMAKER, 1988; HEIL, 1988. Thanks to that incorporation, such self-ascriptions ('cogito-like thoughts', as Burge calls them), display the property of self-verification: my thinking that I am thinking that p makes it true that I am thinking that p.

The incorporation of the "first order" thought ('This glass is full of water') into the self-ascription (the "second-order" thought 'I am thinking that this glass is full of water') is deemed to be a constitutive relation, which owes nothing to the contingencies of mental causation (contra HEIL, 1988): as Burge, Davidson and Shoemaker variously stress, first person authority is grounded in normative features of intentionality - in particular, in the constitutive role of self-knowledge in rationality. The idea here is that ascribing rationality to a subject necessarily amounts to ascribing her a prima facie privileged access to the contents of her propositional attitudes: the ability to bring one's own judgments to critical examination is a constitutive feature of rationality; now the exercise of that ability requires, of a subject who thinks about any object whatever, that she knows it is that object, and not any other thing, she is thinking of.

Ascriptions of privileged access are, however, defeasible - as are, quite generally, ascriptions of rational capacities and attitudes. Moreover, what Burge, Davidson, Shoemaker et alii describe as a constitutive relation between a "first order" thought and the corresponding "second order" self-ascription is restricted (this is an essential feature of every 'cogito-like though') to present time - more specifically, to thoughts whose characteristic expression is the first person, present tense, indicative mood in its assertorical use.23 23 That remarks contains (among other things) the key to unravel the vexing logico-philosophical problem known, after Wittgenstein, as 'Moore's Paradox'. Only in that privileged case is a judgment of the form 'S thinks that p' a sufficient condition of its own truth. In all other cases, differences in perspective (as between persons, times, or moods) make for a wide range of possibilities of misattribution.24 24 As Davidson writes: 'Neither speaker nor hearer knows in a special or mysterious way what the speaker's words mean; and both can be wrong. But there is a difference. The speaker, after bending whatever knowledge and craft he can to the task of saying what his words mean, cannot improve on the following sort of statement: 'My utterance of "Wagner died happy" is true if and only if Wagner died happy'. An interpreter has no reason to assume this will be his best way of stating the truth conditions of the speaker's utterance' (1984: 13).

And, as with the Burge-Davidson-Shoemaker view of first-person authority, so with Russell's Principle of Acquaintance: it also seems to work only in the first person present tense case. Memory as retained acquaintance, in a word, is not memory enough, as it fails to accomplish the task ascribed to knowledge by acquaintance: to ensure the satisfaction of Russell's Principle.25 25 That holds, to be sure, of Russellian acquaintance, and of the restrictions it imposes on episodic memory. I myself would think that to the extent that acquaintance is retained, one should be able to single out a recalled object by using a memory demonstrative ('That man I met yesterday'), and that insisting on the possibility of disruption of the pertinent anaphoric chain (the one linking the present memory demonstrative 'that man' to the past perceptual demonstrative 'this man') is tantamount to insisting, about the perceptual case, on the possibilty of hallucination. But then I am also prepared to endorse the 'partial understanding' account of Twin Earth style scenarios, and the liberalized notion of acquaintance recommended by McDowell: in other words, I embrace anti-individualism. For all that, here I speak as Russell's counsel: I'm pleading for a fair hearing.

And that's how we come to lose acquaintance - or what acquaintance should enable us to do: discriminate the objects of our thoughts - even while we forget nothing. Memory is vulnerable to other forms of loss besides forgetfulness: the one discussed here would not spare even the indelible recollections of Don Isidoro Funes, el memorioso, in Borges's tale.

Suppose I pass a number of times by the window of an antique shop, and each time I stop to admire a Chinese porcelain bowl displayed in pride of place among other objects; and each time I entertain such thoughts as 'This bowl would look perfect in my living room, on the mahogany cabinet' or 'Here's a perfect birthday gift for my mother'.26 26 This is variant of an example of Kevin Falvey's (see FALVEY, 2003: 223-229), itself a successor of Gareth Evans's thought experiment about the two steel balls (see above). In each of these occurrences, the object of my thinking is plainly the bowl I'm looking at in the shop's window. But now suppose the antique dealer has a supply of such bowls, all exactly alike, which he has been selling regularly (he bought the set in an auction from a insolvent merchant from Hong Kong); and suppose each time it was another (qualitatively indiscernible though numerically different) bowl which I beheld. And here I am, wholly ignorant of these successive substitutions, sitting by myself at home and thinking: 'When I draw next month's salary, the first thing I'll do is buy that Chinese bowl'. In that occurrence, the expression 'that Chinese bowl' lacks a reference: there's no particular bowl I'm thinking of - which is another way of saying that I am under the illusion of entertaining a singular thought.27 27 On what is involved in such illusion, see EVANS, 1982: 46 and MCDOWELL, 1986. In Russellian terms, 'that Chinese bowl', in my current usage, is not, appearances notwithstanding, a singular term but a definite description (something like 'the bowl I have been watching') which, through failure of uniqueness, denotes nothing.28 28 In urging this analysis, I am of course building an assumption of uniqueness into the description of the successive thought episodes elicited by the view of what was on display at the shop's window. I willingly grant that, should unicity be of no concern to the prospective buyer, the story might as well be read as involving a case of deferred ostension, where 'that Chinese bowl' would be short for something like 'a bowl of that kind'. Then it might help readers to imagine that what the subject was after was, say, a bowl alleged to have belonged to a famous person, or the one which once was owned by his grandmother and had since been sold - in brief, an object with a particular individuating history. But then, as I am unable now (at home) to pick out any particular Chinese bowl I have been watching, I am by the same token unable to reiterate any of the thoughts I entertained while window-shopping (even though I can reiterate, in an attempt at expressing those thoughts, the verbal construction which might, in each case, have expressed them) - and that is not due to my having forgotten anything at all. Rather, a necessary condition of my referring to any of the objects I previously thought of is no longer satisfied (FALVEY, 2003: 223-229).

And that's how I may come to know that I entertained a certain thought yet, due to not being able to tell the content of that thought from some relevant-though-phenomenologically indiscernible alternatives (water vs. twater; that bowl vs. that other bowl), not know, in a perfectly intuitive sense, what thought I entertained.

The locus classicus of presentation of the idea of a thought which I am in a position to denote (as I can know it by description) yet am not in a position to think is, as we saw, Russell's 'Knowledge by Acquaintance and Knowledge by Description'. In a symposium with Michael Tye on 'Externalism and Memory', Jane Heal suggested that, even after being slow-switched to Twin Earth, a subject can always evoke by description her former thoughts about water (that is, about H2O), using the term 'water' as an abbreviation of, say, 'the stuff I swam in as a kid' (HEAL, 1988: 105). But is seems evident to me that, far from retrieving the lost content, such detour amounts to a tacit acknowledgment of its unavailability. As the example of the Chinese bowls shows, the contents of de re attitudes display a sort of constitutive vulnerability: should certain conditions, on whose satisfaction epistemic access to the object of the attitude depends, not be satisfied,29 29 And I stress, once again, that the satisfaction of such conditions may lie entirely beyond the subject's control (in the sense explained in fn. 5). such contents are inaccessible - quite literally unthinkable - for the subject.

VI

What should we conclude? If what mattered most for Russell was to avoid the idealist view according to which there is no cognition which is not conceptual (in other words, that there is no cognition outside particular acts of judgment), then his attempt to secure direct cognitive access to objects beyond present experience may well have left him with an unsolved problem.

And if we are to draw a couple of lessons from Russell's predicament, the first is, I suppose, that contemporary arguments aimed at establishing the incompatibility between externalism and first person authority (beginning with the so-called Memory Argument first presented in Boghossian, 1989) apply, by parity, to a wider range of views on mental content than is usually assumed.30 30 To be sure, although an object of Russellian acquaintance is relationally individuated, nothing like Kripke's historical chain of uses or Putnam's social division of linguistic labor or Burgean deference takes place here; yet the argument applies all the same. We put the finger here on the need to tell internalism from individualism as two independent strands in what is (still) called 'Cartesianism'. And then the second is that the difficulty raised by present-day critics of externalism about preservative memory (or, as Russell would have it, "memory of truths") is likely to extend to episodic memory as well.

So, on the cheeky assumption that my reconstruction may be of any interest beyond the precincts of Russellian scholarship, what it possibly teaches us is that it is incumbent on internalists as much as on externalists to explain how such possibilities of loss and modification of memory contents as were rehearsed in this paper are compatible with some form of Russell's Principle. It doesn't seem to me that this has so far been achieved.

Recebido em 22/07/2009.

Aprovado em 30/09/2009.

  • BOGHOSSIAN, Paul A. (1989). 'Content and Self-Knowledge', Philosophical Topics 17: 5-26.
  • BURGE, Tyler (1988) 'Individualism and Self-Knowledge', Journal of Philosophy 85: 649-663; reprinted in Peter Ludlow & Nora Martin (ed.), Externalism and Self-Knowledge (Stanford: CSLI Publications): 111-127.
  • DAVIDSON, Donald (1984) 'First Person Authority', Dialectica 38: 101-111, reprinted in Subjective, Intersubjective, Objective (Oxford: Oxford University Press, 2001): 3-14.
  • DAVIDSON, Donald (1987). 'Knowing One's Own Mind', Proceedings and Addresses of the American Philosophical Association, 441-58, reprinted in Subjective, Intersubjective, Objective (Oxford: Oxford University Press, 2001): 15-38.
  • DIAMOND, Cora (2000). 'Does Bismarck have a beetle in his box? The private language argument in the Tractatus', in A. Crary & R. Read (eds.), The New Wittgenstein (London & New York: Routledge): 262-292.
  • DONNELLAN, Keith (1977). 'The Contingent A Priori and Rigid Designators'. Midwest Studies in Philosophy 2: 12-27.
  • EVANS, Gareth (1982). The Varieties of Reference, ed. John McDowell (Oxford: Oxford University Press).
  • FALVEY, Kevin (2003). 'Memory and Knowledge of Content', in Susana Nuccetelli (ed.), New Essays on Semantic Externalism and Self-Knowledge (Cambridge, MA & London: The MIT Press): 219-240.
  • HEAL, Jane (1988) 'Externalism and Memory' Proceedings of the Aristotelian Society 88: 183-209.
  • HEIL, John (1988). 'Privileged Access', Mind 97: 238-251.
  • KRIPKE, Saul A. (1975) 'Outline of a Theory of Truth', The Journal of Philosophy 72: 690-716.
  • MARTIN, M. G. F. (2001) 'Out of the Past: Episodic Recall as Retained Acquaintance', in Christoph Hoerl & Teresa McCormack (eds.) Time and Memory (Oxford: Oxford University Press): 257-284.
  • MCDOWELL, John (1969-70). 'Identity Mistakes: Plato and the Logical Atomists', Proceedings of the Aristotelian Society 70: 181-195, reprinted in Meaning, Knowledge, and Reality (Oxford: Blackwell, 1998): 157-170.
  • MCDOWELL, John (1973). Plato: Theaetetus Translated with notes by John McDowell (Oxford: Clarendon Press).
  • MCDOWELL, John (1986). 'Singular Thought and the Extent of Inner Space', in Philip Pettit & John McDowell (eds.), Subject, Thought, and Context (Oxford: Clarendon Press, 1986): 137-168, reprinted in John McDowell, Meaning, Knowledge, and Reality (Oxford: Blackwell, 1998): 228-259.
  • NEALE, Stephen (1990). Descriptions (Cambridge, MA: The M.I.T. Press).
  • PEARS, David (1975). 'Russell's Theoris of Memory', in Questions on the Philosophy of Mind (London: Duckworth).
  • RUSSELL, Bertrand (1903). The Principles of Mathematics (London: George Allen & Unwin, 1964).
  • RUSSELL, Bertrand (1905). 'On Denoting', Mind 14: 398-401, reprinted in The Collected Papers of Bertrand Russell, vol. 4: Foundations of Logic 1903-05, ed. Alasdair Urquhart (London & New York: Routledge, 1994): 414-427.
  • RUSSELL, Bertrand (1904). 'On History', The Independent Review, reprinted in Philosophical Essays (London: Routledge, 2000): 60-69.
  • RUSSELL, Bertrand (1911). 'Knowledge by Acquaintance and Knowledge by Description', Proceedings of the Aristotelian Society 11: 108-128, reprinted in The Collected Papers of Bertrand Russell, volume 6: Logical and Philosophical Papers, 1909-13, ed. John C. Slater (London & New York: Routledge, 1992), 148-161.
  • RUSSELL, Bertrand (1912). The Problems of Philosophy (Oxford: Oxford University Press).
  • RUSSELL, Bertrand (1914). 'On the Nature of Acquaintance', The Monist 24: 1-16, 161-187, 435-453, reprinted in R. C. Marsh (ed.) Logic and Knowledge (London, Routledge, 1956): 127-174.
  • RUSSELL, Bertrand (1919-19). 'The Philosophy of Logical Atomism', The Monist 28-29, reprinted in R. C. Marsh (ed.) Logic and Knowledge (London, Routledge, 1956): 177-281.
  • RUSSELL, Bertrand (1984). Theory of Knowledge: the 1913 Manuscript, ed. Elizabeth Ramsden Eames & Kenneth Blackwell (London: Routledge).
  • SEVERO, Rogério Passos (2005). 'Three Remarks on the Interpretation of Kant on Incongruent Counterparts', Kantian Review 9: 30-57.
  • SHOEMAKER, Sidney (1988). 'On Knowing One's Own Mind', Philosophical Perspectives 4: 187-214.
  • STRAWSON, P. F. (1959). Individuals: an Essay in Descriptive Metaphysics (London: Routledge).
  • WHITEHEAD, Alfred North & Russell, Bertrand (1910). Principia Mathematica (Cambridge: Cambridge University Press, 1963), Vol. I.
  • Memory as acquaintance with the past: some Lessons from Russell, 1912-1914

    Paulo Faria
  • 1
    In fairness to Russell, let it be noted that he took Frege's theory and his own former view to be, in his own words, 'very nearly the same' (RUSSELL, 1905: 42). A few absentminded readers still believe that to be the case.
  • 2
    See, e.g. Pears 1975. And, to be sure, by the time of
    The Analysis of Mind (1921) a very different picture was already in place, but that will not be my concern here.
  • 3
    I follow Evans (1982, 1) and Neale (1990, 15) in using 'singular terms' interchangeably with 'referential expressions', thus excluding descriptions by stipulation. The stipulation, admittedly at variance with current usage in the philosophy of language, is meant to stress the
    logical gist of Russell's repudiation of 'the unity of the intuitive category of referential expressions' (Evans 1982: 42): namely, that descriptions are quantifiers rather than referential devices. To be sure, this holds only of descriptions in
    attributive position, and in sticking to the stipulation I don't mean to be taking a stand on the issue whether these are the only cases of logico-semantic (as opposed, say, to merely pragmatic) significance. So-called referential uses of definite descriptions, however they should be accounted for, fall beyond the scope of Russell's theory anyway, and are accordingly left out of the present discussion.
  • 4
    See Martin (2001) and his forthcoming book on perceptual appearances. Mark Johnston has been dealing with acquaintance under the head of 'the manifest', the topic of his forthcoming book to bear that title. Jim Pryor has lectured on his 'epistemic theory of acquaintance' at a number of places, yet no printed statement of the theory is, to the best of my knowledge, available for the time being.
  • 5
    The point here is that what makes it the case that an object-dependent thought is about
    a rather than about its phenomenologically indiscernible "twin"
    a* is a matter of the subject's being properly related, both causally and epistemically, to the object. Russell's notion of acquaintance was meant to encompass both the causal and the epistemic requirements, the latter taken in its strictest possible form as a requirement of
    discriminating knowledge; hence the reading of the so-called Russell's Principle (that a subject cannot entertain a thought about something unless he
    knows what it is he is thinking about) as requiring that 'the subject must have a capacity to distinguish the object of his judgment from
    all other objects' (EVANS, 1982: 89, my emphasis). This is arguably too strong a condition to be imposed by any plausible account of object-dependent thinking; moreover, it is certainly something content externalists should know better than grant to their critics. For all that, I open-eyedly proceed, throughout this paper, as if a relaxed version of Russell's Principle were
    not available; in particular, I say nothing about the advantages of substituting
    reliable discrimination for
    incorrigible discrimination (say, along the lines sketched long ago by Martin Davies and embraced since by Neale and others); or of taking the quantifier in "all other objects" to range over a set of contextually
    relevant alternatives, rather than over an unrestricted domain. I have allowed myself so to proceed for two reasons, of which the first is purely historical: seeing why Russell could not find a place in conceptual space for anything like what Burge and others call 'partial understanding' enables one to see the rationale for his verdict about the outcome of lacking discriminative knowledge (in the strongest possible sense) of an object: namely that no singular thought about that object is available to the subject (see the Bismarck example in the following paragraph, and then the brief account of McDowell's criticism of Russell in section IV below.) My second reason bears on the current status of the debate surrounding the so-called Memory Argument against content externalism (on which see section V below): arguably, much of what is at stake in assessing the force of Boghossian's attempted
    reductio of compatibilism hinges on the plausibility of building one or another form of a discrimination requirement into the account of what it takes to know the content of one's own mental states.
  • 6
    Russell 1911: 158 (1912: 57). A related possibility is discussed, in the framework of a 'direct reference' view of names, by Keith Donnellan: we may sometimes be in a position to know that a certain sentence expresses a truth without knowing the truth that it expresses (DONNELLAN, 1977). As for Russell, the problem for Donnellan is raised by cases where we lack the mode of access to the relevant object which alone would enable us to think what in each case we fail to think. More on that below.
  • 7
    See, for a particularly striking statement, Russell (1904).
  • 8
    To be sure, the main elements of the frame were an integral part of Russell's philosophy long before 'On Denoting'. The idea of a direct, immediate, non-conceptual mode of cognition, and of its priority over judgment (which reaches its most striking formulation in the Principle of Acquaintance discussed in Russell 1911) was part and parcel of the Moore-Russell anti-idealistic picture of judgment as grounded on receptivity.
  • 9
    The latter had been playing a crucial role in Russell's account of logical and mathematical knowledge since his 'revolt against idealism'. Compare this often-quoted statement from the Preface to
    The Principles of Mathematics: 'The discussion of indefinables - which forms the chief part of philosophical logic - is the endeavour to see clearly, and to make others see clearly, the entities concerned, in order that the mind may have that kind of acquaintance with them which it has with redness or the taste of pineapple' (RUSSELL, 1903: xv).
  • 10
    We have here in a nutshell the Russellian project (to be pursued further by Carnap, Goodman and others) of a 'logical construction of the world'. The project is described in Russell's Harvard Lectures of 1914
    Our Knowledge of the External World as a Field for Scientific Method in Philosophy and in the important paper, published that same year, 'The Relation of Sense-Data to Physics' (reprinted, along with 'Knowledge by Acquaintance and Knowledge by Description', in the collection
    Mysticism and Logic and Other Essays).
  • 11
    This step is not going to be particularly hard to take. A pretty good job has already been done in Martin 2001, to which the reader is referred.
  • 12
    The principle is explicitly appealed to by Plato at a later stage in the
    Thaetetus, as a step in an argument designed to show that the condition to be satisfied by a true judgment in order for it to qualify as knowledge could not possibly be the ability to discriminate the object of the judgment from other objects, for the plain reason that such ability is a requirement of any judgment (whether true or false), hence nothing that could possibly be
    added to a true judgment so as to turn it into knowledge. See
    Theaetetus 208e-209d, and McDowell's comment in McDowell 1973: 255-256.
  • 13
    To the extent that being able to mean a particular object is matter of being properly related to that object - e.g. of partaking in a unique and non-disrupted historical chain of uses of a designator for that object - a speaker's atempt to mean an object is vulneable to failure through ungroundedness: that is the sense in which meaning something is apt to be a 'risky' business. I am here extending to subsentential components the notion of groundedness as it features in the explanation of the semantical paradoxes (actually, ungroundedness of
    statements is often, if by no means exclusively, derived from ungroundedness of subsentential components, most notably singular terms). Anyway, here's Kripke: 'an adequate theory must allow our statements involving the notion of truth to be
    risky; they risk being paradoxical if the empirical facts are extremely (and unexpectedly) unfavorable. There can be no syntactic or semantic "sieve" that will winnow out the "bad" from the "good" ones' (1975: 692).
  • 14
    Lest my brutally sketched presentation be dismissed as a hyperbolic parody of McDowell's talk about a 'favor from the world', I hasten to note that (to give one single example) the rhetorically much austerer Neale shares essentially the same diagnosis: 'Bent on exorcising the possibility of someone entertaining an object-dependent proposition when, in reality, there is no object which the purported proposition is about, Russell maintained that only a sense-datum could be trusted to be the "subject" of such a proposition. It is by now pretty generally agreed that this position is untenable, and I shall not spend any time addressing any of the familiar objections to it' (NEALE, 1990: 18).
  • 15
    In what follows I describe, as best as I can, Russell's attempt to dissever
    acquaintance and
    presence to consciousness. It should go without saying that the attempt, even if successful, is predicated on a thoroughly individualistic picture of the mind, with the consequence that nothing beyond the subject's own past can possibly fall in the scope of what Russell is prepared to count as retained acquaintance. To that extent, the acquaintance theory of episodic memory (the only kind of memory Russell deals with) falls radically short of providing the mode of acquaintance with the past that McDowell has in view: a thought about, say, Aristotle will still be deemed a general thought, to be dealt with, on a par with thoughts about nonexistent objects, by the theory of descriptions. I thank Maite Ezcurdia for urging me to acknowledge the extent to which McDowell's proposal remains apposite even granted my plea on Russell's behalf.
  • 16
    That there is such a thing as direct awareness of what one has experienced Russell deems 'obvious'; in spite of which he goes on to sketch an argument to the effect that immediate knowledge by memory (acquaintance with the past) is the source of
    all knowledge concerning the past: 'without it, there could be no knowledge of the past by inference, since we should never know that there was anything past to be inferred' (1912: 49).
  • 17
    The same point recurs in the 'Definitions and Methodological Principles in Theory of Knowledge', published in
    The Monist in 1914 in sequence to the three papers 'On the Nature of Acquaintance'; taken together, these constitute the first four chapters of the 1913
    Theory of Knowledge. 'Being in the past' writes Russell, 'is not an intrinsic property of an object, but a relation to the subject; thus memory will have to be distinguished from sense and imagination as a different relation to objects, not as the same relation to different objects'. (RUSSELL, 1984: 56). Russell goes on to claim that 'memory is excluded if we say that the acquaintance we are concerned with must not be of an object
    given as past' (
    Ibid. My italics.)
  • 18
    The criticism of the 'veil of ideas', if taken seriously, should have erased from our accounts the picture of Russell as a "Cartesian" thinker, in the (historically inaccurate) sense in which that adjective has been most often used in the analytic tradition. For an enlightened version of the charge of "Cartesianism" see McDowell 1986: 236-247.
  • 19
    And, on Russell's construal, this amounts to being in a position to
    incorrigibly discriminate that object 'from all other things' (see note 6 above).
  • 20
    See the
    Critique of Pure Reason, A 273 / B 328. My original example was that of incongruous counterparts - say the left and the right hand as they feature in Kant's
    1770 Dissertation. To which it was rightly objected that different descriptions can be given of the right and the left hands by specifying a coordinate system with the right multiplicity. That much is granted, yet it misses Kant's point, as nothing in the resulting pair of descriptions would single out one of them as a description of the
    right rather than the
    left hand, or
    vice versa (See SEVERO, 2005 for a thorough and illuminating discussion of the issue). For all that, the example was certainly a distracting one, which I am glad to let go.
  • 21
    The scenario I am appealing to was introduced by Gareth Evans in
    The Varieties of Reference; his example was that of two indiscernible steel balls: see EVANS 1982: 90.
  • 22
    See DAVIDSON, 1984, 1987; BURGE, 1988; SHOEMAKER, 1988; HEIL, 1988.
  • 23
    That remarks contains (among other things) the key to unravel the vexing logico-philosophical problem known, after Wittgenstein, as 'Moore's Paradox'.
  • 24
    As Davidson writes: 'Neither speaker nor hearer knows in a special or mysterious way what the speaker's words mean; and both can be wrong. But there is a difference. The speaker, after bending whatever knowledge and craft he can to the task of saying what his words mean, cannot improve on the following sort of statement: 'My utterance of "Wagner died happy" is true if and only if Wagner died happy'. An interpreter has no reason to assume this will be
    his best way of stating the truth conditions of the speaker's utterance' (1984: 13).
  • 25
    That holds, to be sure, of
    Russellian acquaintance, and of the restrictions it imposes on episodic memory. I myself would think that to the extent that acquaintance
    is retained, one should be able to single out a recalled object by using a memory demonstrative ('That man I met yesterday'), and that insisting on the possibility of disruption of the pertinent anaphoric chain (the one linking the present memory demonstrative 'that man' to the past perceptual demonstrative 'this man') is tantamount to insisting, about the perceptual case, on the possibilty of hallucination. But then I am also prepared to endorse the 'partial understanding' account of Twin Earth style scenarios, and the liberalized notion of acquaintance recommended by McDowell: in other words, I embrace anti-individualism. For all that, here I speak as Russell's counsel: I'm pleading for a fair hearing.
  • 26
    This is variant of an example of Kevin Falvey's (see FALVEY, 2003: 223-229), itself a successor of Gareth Evans's thought experiment about the two steel balls (see above).
  • 27
    On what is involved in such illusion, see EVANS, 1982: 46 and MCDOWELL, 1986.
  • 28
    In urging this analysis, I am of course building an assumption of uniqueness into the description of the successive thought episodes elicited by the view of what was on display at the shop's window. I willingly grant that, should unicity be of no concern to the prospective buyer, the story might as well be read as involving a case of deferred ostension, where 'that Chinese bowl' would be short for something like 'a bowl of that kind'. Then it might help readers to imagine that what the subject was after was, say, a bowl alleged to have belonged to a famous person, or the one which once was owned by his grandmother and had since been sold - in brief, an object with a particular individuating history.
  • 29
    And I stress, once again, that the satisfaction of such conditions may lie entirely beyond the subject's control (in the sense explained in fn. 5).
  • 30
    To be sure, although an object of Russellian acquaintance is relationally individuated, nothing like Kripke's historical chain of uses or Putnam's social division of linguistic labor or Burgean deference takes place here; yet the argument applies all the same. We put the finger here on the need to tell
    internalism from
    individualism as two independent strands in what is (still) called 'Cartesianism'.
  • Publication Dates

    • Publication in this collection
      19 July 2010
    • Date of issue
      June 2010

    History

    • Received
      22 July 2009
    • Accepted
      30 Sept 2009
    Faculdade de Filosofia e Ciências Humanas da UFMG Av. Antônio Carlos, 6627 Campus Pampulha, CEP: 31270-301 Belo Horizonte MG - Brasil, Tel: (31) 3409-5025, Fax: (31) 3409-5041 - Belo Horizonte - MG - Brazil
    E-mail: kriterion@fafich.ufmg.br