From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

Introduction

Ameriks (1994, 331-2) points out that in Kitcher (1990) and Powell (1990) ‘the manifold ambiguity’ of the ‘strong apperception thesis’ (SAT), ‘plays a central role’. SAT holds that empirical consciousness (E) requires transcendental apperception (T) (334). According to Strawson (1966, 39, 247 ff.), this ambiguity pertains to Kant´s use of personal pronouns and possessives in such a way that:

(W)e have to ask what we human beings, Kant’s readers, can unambiguously understand by “us” and “we” and “our” when these expressions are so easily and loftily used to convey the doctrines of transcendental idealism.

Kitcher and Powell attempt to avoid this ambiguity focusing on the ‘functional’ character of apperception. Ameriks is not satisfied with these attempts. In contrast, he stipulates that:

(…) Kant himself is clear enough that he does, after all, want to say something about what the “nature” of the stuff of the self is, namely that (…) it does not constitute itself through reflection or synthesis, but rather, like everything else, it first of all exists as the non-material Ding an sich (…). (1994, 345.)

In this paper I will endorse Ameriks’ criticism, though I will object to his faithfulness to Kant. Instead, I will, with reference to the ‘private language argument’ elaborate Powell’s phenomenological approach to apperception.

This is done in section 3. Section 1. contains a description of SAT. In section 2. I outline and criticise the proposals of Kitcher and Powell.

1. Kant on the self

Kant’s concern is with the possibility of experience. Kant defines experience as cognition by means of connected perceptions. For perceptions to be cognitively relevant, their connection cannot be merely associative, like in Hume. A perception in turn, is defined as an empirical consciousness (E) of a connection (apprehension) of the manifold in an empirical intuition. In §26 of the transcendental deduction of the categories Kant explains how this synthesis of the apprehension is accomplished according to the same rules, viz. the categories, which connect these perceptions in transcendental apperception (T) in order for them to become experience.

The argument for this explanation is that time and space both are the forms of intuition as well as intuitions themselves. Therefore, the apprehension of items in a spatio-temporal frame is a priori accomplished in the same way as, and governed by, the synthesis of transcendental apperception. SAT claims that (E) is necessarily accompanied by (T) in the way that is described above. Thus, the relation between these two types of conscious states is not merely logical, but rather a ‘real potential relation’ (Ameriks 1994, 336; cf. B132). Consequently, the self-conscious state “I think.” (T) which effectively accompanies (E) is a consciousness of an identical subject of this act of synthesis in (E) (B132, B133, B135).

What has puzzled analytic philosophers in this ´real potential relation´ is ´the anomalous referential structure of self-consciousness´ that it entails. Since the subject of (E) is identical with the subject of (T) the ‘I’ cannot unambiguously be considered as an object of consciousness. (T) therefore, cannot unambiguously be considered as reflective consciousness. The conscious state (T) encompasses “only the idea of a subject of consciousness that is known only indirectly through the contents of consciousness (…).” (Sturma 1995, 201, 203.)

2. Empiricist reactions to Kant on the self

Kitcher (1990, 140) develops, in opposition with philosophers that advocate a conceptual analysis of Kant´s transcendental arguments (26), a transcendental psychological interpretation of SAT, in particular of §26. She regards this ´epistemic analysis´ of cognitive tasks as importantly contributory to cognitive science (205 ff.). It gives it direction (Kitcher 2000, 61).

Kitcher (1990), observes that Kant need not assume that reflective awareness is an “inseparable component of what it is to perceive, remember etc.” (107, in Ameriks 1994 339). Though one can, for the reason mentioned in sect. 1., agree with this observation, Kitcher fails to explain how the relation between (T) and (E) can unambiguously be interpreted alternatively. As a consequence, it remains unclear which conception of philosophy entitles Kitcher to call this appropriation of SAT in contemporary theory of mind philosophical, notwithstanding its “striking affinities with empirical psychology” (Kitcher 1990, 26 in: Ameriks 1994, 339).

In Kitcher’s interpretation, (T) is deprived of its subjective quality, since it becomes “no more than contentually interconnected systems of cognitive states” (1990, 122, in Ameriks 1994, 343.) Hence, it does not account for the transcendental claim that the relation between (T) and (E) is a real potential relation. It therefore does not satisfy as an interpretation of Kant.

In Powell’s interpretation of (T) ´I´ is itself not a referring expression and therefore might be taken to function as merely a logical operator, a quoting device. However, Powell (1990) is faithful to SAT’s involving a ‘real potential relation’ between (T) and (E). For, since the transcendental subject in (T) represents perceptions as elements of a representational system (232), this interpretation presupposes that the systematic connection of these perceptions in the subject is not merely empirical, but also transcendental. At the same time, it presupposes that it is an empirical subject having these perceptions that necessarily are to be considered as elements of a perceptual system. The expression ´I´, therefore, “(…) can only be used in such a way that its use carries an implicit acknowledgement of, and reference to that being who is the experiencing subject.” (234) Thus, Powell’s interpretation at first sight seems not vulnerable to Strawson’s objection of ambiguity since it does not commit him to the view that the transcendental subject is numerically identical with the subject of empirical apperception.

However, in his interpretation, (T) is of the form: ‘I think that-p’, in which p is an element of a system of representations that are the content of (E). This interpretation of (T) commits him to posit a content of (T), being a ‘meta-representing’ of (E). Thus, the relation between (T) and (E) is projected on a model in which (T) is a second-order mental state that is transitively conscious of (E). It is (T) then, that, by being reflectively conscious of (E) as a system of representations, confers intransitive consciousness on (E), the mental state which is its object. In this model, the intransitive consciousness of (E) is taken as a relational property that is dependent on (T), rather than as such. (Cf. Zahavi 2002, 15 ff.)

In order for an element of (E) to be conscious in (T), the transcendental subject must recognise it as something that it can self-ascribe. This now it can only do, if it has a prior acquaintance with itself. This requires that it is intransitively conscious of itself. Hence, (T) “must await either a third-order mental state that can confer intransitive consciousness on it, in which case we are (…) confronted with a vicious infinite regress, or it must be admitted that (…) [in (T), the I ] is itself already in possession of phenomenal consciousness from the very start, and that (…) would involve us in a circular explanation, presupposing that which was meant to be explained (…)”. (Zahavi 2002, 16; orig. emph.) In the former case, Powell’s (T) would, in contrast with Kant’s claim (B132) not be original. In the latter case, his introduction of the notion of ‘meta-representing’ would not have the illuminative value he claims it has.

Although Powell, regards SAT, in opposition with Strawson’s qualification of transcendental psychology, a ´very sophisticated phenomenology´ (1990, 58), his reconstruction is nevertheless functional. Since his interpretation fails to answer the question “What is (E) like?” with an explication of its intrinsic qualities it underscores Nagel´s (1974, 437) claim that functional characterisations of mental states or events do not exhaust their analysis.

For these reasons, one could endorse Ameriks’s criticism of these functional reconstructions of SAT. However, the elucidatory value of his own recourse to the Ding an sich is limited. For this merely amounts to the adoption of transcendental idealism, which doctrine, that is the source of SAT’s ambiguity, is presupposed in SAT. Moreover, one could contest that it makes sense to argue, as he does (1994, 333), that all uses of ‘I’ in (T) and (E) are co-referential in the sense that they refer to a numerical identical I whereas at the same time, the referent of the expression ‘I’ is urged to be considered as the noumenal self.

3. A Wittgensteinian perspective

SAT’s ambiguity can be attributed to its belonging to an epistemology as a ‘theory of getting to know’ (cf. Ryle 1949, 275, 299), which, though it is claimed to be philosophical (AIX), is marked by the blend of philosophical and psychological connotations of its concepts. The functional approaches share with Kant a predilection for physical language. Given this ambiguity, Ryle´s (1949, 298) question, as to the proper subject matter of the theory of knowledge is justified.

Fortunately, Ryle´s dismissal of epistemology as a theory of ‘getting to know’ does not force one to rush to the one alternative he offers: a conceptual analysis of scientific theories. Zahavi (2002, 17) presents a phenomenological analysis of self-awareness that prevents the answer to the question “What is (E) like?” to fall in an infinite regress, or to be viciously circular. Unlike Powell’s, this analysis does not take the intransitive conscious state (E) as a relational property, but accounts for its first-personal givenness. It involves a “distinction between intentionality, which is characterised by a difference between the subject and the object of experience, and self-consciousness, which implies some form of identity.” (16; orig. emph.) Thus it can answer the requirement that the identity of the I in (T) with the I in (E) “must be that of belonging to the same subject (…)”, whereas it excludes that they are taken as a numerical identity (16). This provision is important, for without it, this identity, that is a necessary condition for the possibility of my calling (E) mine, would leave no room for the difference between these two types of conscious states. This difference precisely consists in the one being transitive, whereas the other is intransitive.

That the identity of the I in (T) and in (E), though it is ‘some form of identity’ cannot be conceived as a numerical identity, is a consequence of what Sturma calls the ´constructive displacement´ of the I (B404). This, he states, does not depend on any vicious circularity but is a consequence of irreflexive relations inherent in the givenness of the empirical context. It is due to the fact—for which, as we have seen in sect. 1., Kant gives an argument in §26 of his deduction of the categories—that “consciousness of spatiotemporal objects is structured by a categorial synthesis according to the conditions of possible self-consciousness”. Hence apperceptive self-referential mental acts cannot be equated with knowledge of a ´self´ (1995, 201 ff.; cf. B404) as an object (B422). The identity of the representation ´I´ does not concern anything of the subject whereby it is given as an object (B408). Kant claims (B157) that in (T), I am conscious of myself not as I appear to myself (cf. B429), nor as I am in myself, but only that I am. This feature of (T) can be accommodated in a phenomenological analysis insofar as it presupposes the existence of pre-linguistic and non-conceptual forms of self-awareness. Reflective self-consciousness is “derivative (…) and always presupposes the existence of a prior unthematic, non-objectifying, prereflective self-awareness as its condition of possibility.” (Zahavi 2002, 17, 18.)

Sturma connects this feature with the experiential first-person perspective of the ‘private language argument’. (1995, 204). This argument is concerned with the analysis of a linguistic representation of phenomenal occurrences. As it demonstrates e.g. in §§269, 278 the redundancy of the ´transcendental´ subject, as well as of the metaphysical private object (cf. §§272, 293 and 299) it allows for an unambiguous use of the first-person personal and possessive pronouns by eliminating these metaphysical entities.

Moreover, the private language argument fits well in the phenomenological project insofar as it does not take intransitive consciousness as a relational property. This has the following advantages. First, it does not take linguistic behaviour as reflective on something else but opens the way for its full-fledged analysis. Second, interpreted as an argument for the impossibility of a phenomenal language, as Hintikka (1986, 241) does, it allows for pre-linguistic forms of self-awareness. Third, since this impossibility must not be attributed to some ineffable private property of phenomenal events but can rather be interpreted as a characteristic of the grammar that is decisive for what it makes sense to say (cf. Alva Noë 1994, 20) it does not entail an ontological commitment to some private object. This is expressed by Wittgenstein’s calling sensation “not a something, but not a nothing either!” (§304) Though Wittgenstein states this insight in a rather enigmatic way, his criticism of attempts to use the objectifying grammar of a physical language to represent phenomenal events is on a par with Nagel’s assessment that “physicalism is a position we cannot understand because we do not at present have any conception of how it might be true.” (1974, 447.)