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

Introduction

The similarity of Wittgenstein’s and Kant’s projects is a much-discussed topic. Hacker (1972) is an example. He assimilates Kant’s critique of reason with Wittgenstein’s critique of language, exploring the limits of reason and of language respectively (1972, 30-32). He calls the private language argument (PLA) “an endeavour to extend and elaborate the Kantian dictum that intuitions without concepts are blind.” (1972, 216.)

Williams’ (1974) assessment of Wittgenstein’s idealism fits well in this topic. He agrees with Hacker, that the PLA is related to a long-term project of exorcising solipsism (cfr. Hacker 1972 in: Williams 1974, 79). He opposes, however, Hacker’s connexion of solipsism and idealism (Hacker 1972, 214 in: Williams 1974, 80). Williams’ argument, to which Malcolm (1982) has objected, chiefly aims to suggest that the move from the tractarian solipsistic ‘I’ to the later Wittgenstein’s ‘we’ “was not unequivocally accompanied by an abandonment of the concerns of transcendental idealism” [TI] (79).

In this paper I will argue that Williams’ interpretation ought not succumb to Malcolm’s objection (section 1). A comparison of Kant’s and Wittgenstein’s approaches to causality, however, will exhibit, besides some prima facie similarities, an important difference between both purported variants of TI. Due to this difference Williams´ claim turns out to be untenable (section 2). From this assessment, I will draw some conclusions (section 3).

1. Wittgenstein and Idealism

Williams qualifies the development from the tractarian to the later Wittgenstein as a move from ‘I’ to ‘we’, from solipsism to a pluralised idealism (Williams 1974, 93). The reason for this move is that solipsism involves phenomenalism (cf. TLP 5.6).

According to Williams, a phenomenalist maintains his empirical realist stance by claiming that material objects exist unperceived. Accordingly, on his assumption that observers are material objects, his translation of the claim that “Even if there were not any observers, certain material objects would exist” would be of the form:

(i) If P were not the case, then, if P were the case, then Q

In view of this unsatisfactory result Williams concludes that a phenomenalist cannot deny the mind-dependence of material objects. In this form, it amounts to empirical idealism (81 cf. TLP 5.631).

A phenomenalist might avoid this result by eliminating his assumption about the existence of observers (81):

(ii) If P were not the case, then Q

He, then, cannot maintain that the given sense data are (related to) observations. He, therefore, cannot account for these items unless they are supposed to be “in some sense” mental. Since he then cannot truly assert “any form of mind-dependence of the world”, the fact that these items are mental “show[ s ] that the world is mental”. Therefore, while we cannot say “(except empirically and falsely) that the world is the world of experience (…)”, “the world of experience conditions everything we say”. Williams calls this phenomenalist position a form of TI (82; orig. emph. throughout).

To avoid 0 the early Wittgenstein adopts 0 by postulating that the ‘I’ does not belong to the world, but is its boundary (TLP 5.632, 5.641). This postulate, and its corollary,

(iii) “The limits of my language mean the limits of my world.” (TLP 5.6)

point, according to Williams, taking into account that these limits “could not be staked out from both sides”, to transcendental solipsism (1974, 78, 82).

To avoid (ii), the later Wittgenstein’s arguments intend to “remove the need even to try to point (…) in a solipsistic direction” (79). Thus, by adopting 0,

(iv) “The limits of our language mean the limits of our world.” (82.)

the PLA argues against an empiricist theory of meaning which sets out with first-person immediacies. The move from ‘I‘ to ‘we’, therefore, is a move from the phenomenalist form of TI—which already has left behind empirical idealism (cf. 82-3) — to a non-phenomenalist form of pluralised TI.

Though Williams stresses that 0 should neither be interpreted as a blank tautology, nor as an empirical claim (83) he signals a persistent uncertainty in the interpretation of ‘we’ (90), due to its pervasive vagueness (79). He attributes this uncertainty to the relativistic elements in its use (90).

It is here that Malcolm’s objection enters. Malcolm agrees with Williams’ assessment of the tractarian ‘no-self solipsism’ (1982, 250-1, fn. 2). He recalls that the ‘I’ is a necessary condition (presupposition) of the world. Not being in the world, it is not itself a possible object of experience (250 cf. NB 79). As to the later Wittgenstein, however, though he concedes that ‘we’ has a shifting reference (252), he denies that its use is vague and indefinite (251); it always refers to some actual human group or society, in contrast with another or imagined one (cf. e.g. Z 380; OC 608-9 in Malcolm 1982, 254).
This objection does not convince. First, the allegedly steady reference of ‘we’ induces him to interpret

(v) “The possibility of a language game is conditioned by certain facts” (OC 617)

as an empirical proposition that he considers to be “surely contrary to idealism” (Malcolm 1982, 266). However, although Williams’ non-empirical interpretation of 0 is vulnerable to Moyal-Sharrock’s criticism, that he makes an unwarranted leap from the denial of a rational connection of language to reality to the assertion of its autonomy (2004, 221 n. 17), the non-autonomy of language, which is presupposed in (v), does not entail that (iv) should be read empirically. We must, therefore, not endorse Bolton’s view, that the “move from ‘I‘ to ‘we’ is the move from being transcendental, to being in the world” with which idealism is finished. (1982, 284). Wittgenstein’s assertion that, as he is not doing natural science, his interest does not fall back upon the possible causes of the formation of concepts (PI 230) provides textual support for a transcendental interpretation of 0 in accordance with Williams’ assessment of 0.

Second, Malcolm concedes that Wittgenstein’s treatment of knowledge claims, which conventionally are of the form:

(vi) Kp ? p

is prone to idealism. For, since the language game defines p to be the implication of the correct use of ‘Kp’, “what is the case, is determined by human language” (1982, 263). As Williams stresses, however, the point of language games is not that they establish truth conditions of empirical propositions. Wittgenstein’s constructivism (1974 89) rather aims to establish assertion-conditions (94). Hence, Malcolm’s distinction between knowledge claims and other assertions is unwarranted.

Third, Malcolm mitigates Williams’ worries about the impossibility of an evaluative comparison of the adequacy of language games by distinguishing between a cross-game understanding at the level of description, which he considers unproblematic, and an understanding at “the deeper level of instinctive action and reaction” which is problematic indeed (1982, 260). This distinction, however, disregards the impact of the ‘meaning-is-use’ notion and falls short as a heuristic device.

Williams’ worries, which are induced by his seeking clues for allowing Wittgenstein to leave solipsism behind are, therefore, not completely without reason. Accordingly, I subscribe to Bolton’s (1982, 279 fn. 14) suggestion that Malcolm does not bring out the full strength of Williams’ interpretation because he does not see the reasoning behind it. Hence, I conclude that Williams’ interpretation ought not succumb to Malcolm´s criticism.

Causality in Kant and Wittgenstein

Kant´s argument for his claim, that

(i) all alterations take place in conformity with the law of the connection of cause and effect (B232)

operates with the basic assumption of the ideality of time (B52), i.e. its unobservability (B233; cf. Paton 1970, 253). He argues that since mere sense and intuition cannot connect successive perceptions, their connection is produced by a synthetic faculty of the imagination, which determines inner sense in respect of time-relation (B233). This determination warrants that the order of the subjective succession of perceptions [representations] parallels the order of the objective succession of perceived events. This, the imagination achieves by subjecting the representations to a rule which compels us to observe them in that order (B242). Following this rule, we ascribe to an event a determinate position in time by presupposing that every event is determined by an antecedent (B240) upon which it follows invariably (B243). Paton considers this argument as the most fundamental application of the transcendental deduction (1970, 222).

In contrast, the later Wittgenstein admits of nonnecessary causality (Z 608) and denies a causal nexus. He claims that “it is high time that our concept of causality is upset” (Z 610). He attributes the “urge (…) to see everything through the scheme of cause and effect” (CE 393) to our unconsciously following a ‘natural law’ in our reasoning (OC 135).

This contrast leaves room, however, for advancing some prima facie similarities that might point in the direction of TI in the later Wittgenstein. First, Kant´s argument aims, by proving a correspondence between the successive order of occurrences of sense-data and a succession of events in an objective time-order (cf. B235), like Wittgenstein’s move from ‘I’ to ´we´, to refute phenomenalism. Second, Kant stipulates that his solution contradicts inductivism, the received view of establishing a causal nexus (B 241). This resembles Wittgenstein’s assertion that we do not need a law of induction to justify our predictions (OC 287). Accordingly, both philosophers emphasise that causality is not an empirical matter, not grounded (OC 131), but rather presupposed in experience. It is part of our ‘world-picture’ (134-5; cf. 167), “fused into the foundations of our language” (558), or presupposed in the concept of an object (B240) respectively. Third, Kant qualifies his law of empirical representation of events as a necessary law of our sensibility and a formal condition of all perceptions (B244). This resembles Wittgenstein’s calling “all those a priori certain principles [Sätze], like the principle of sufficient reason, of the continuity in nature, a priori insights concerning the possible laws [Sätze] of natural science.” (NB 42.)

These similarities notwithstanding, an important difference between Wittgenstein’s indeterminism (cf. Scheer 2004) and Kant´s account of causality consists in the nature of the necessity of the regularities within the world we perceive, and hence in the difference between Kant’s notion of form, and Wittgenstein’s life-form (cf. Bolton 1982, 273-4, who associates the former with TI, the latter with activity, practice (cf. OC 95)). In Kant, this necessity, and hence our certainty about them (cf. B239) is entailed by the assumption of the ideality of time as an a priori form of our sensibility. This assumption, or its corollary, esse est concipi, is, as Harrison (1982, 215) contends, “precisely the assumption of idealism” or, more precisely, of TI as an epistemological position. Thus, causality is a necessary a priori principle. Contrariwise, (v) implies, that Wittgenstein’s certainty “which underlies all questions and all thinking” (OC 504) is contingent a priori. This explains the relativistic elements in the use of ‘we’. Certainty does not entail knowledge in the sense of (vi). In accordance with the non-empirical provision of 0, it merely produces good grounds for an outlook (OC 608), though a “new way of looking at things cannot give a lie to scientific thinking” (OC 292-3). This, I suggest, is meant by “(t)he subsequent physical moments of things are already in the world.” (CE 421-444).

This difference shows that our unconsciously following a ‘natural law’ in our reasoning, which ‘compels’ us to think that objects do not (dis)appear without a cause (OC 135) does not point in the direction of TI, like Williams assumes the allusion to ‘our nature’ in the citation on top of this paper does.

Concluding Remarks

This result leaves us to question whether Williams’ assessment of Wittgenstein’s move from 0 to 0 is correct. Since 0’s solipsism entails that empirical idealism is indiscernible from realism (cf. TLP 5.64) and that scepticism is not a sensible position (6.51), it is unlikely that this move was inspired by worries about its consequence, empirical idealism. Following Hintikka´s (1966, 160) assessment of the early Wittgenstein´s solipsism in 0, as asserting the impossibility of getting “beyond the boundaries of myself”, the purported pluralised idealism of 0 rather seems to be an elaboration of the idea that solipsism, without being false, lacks linguistic expressibility (TLP 5.62; cf. PI secs. 24, 402-3). Thus, 0 is continuous with :

(i) “No part of experience is a priori” (TLP 5.634),

and elaborates this idea in the direction of a what Haller (1981 40) calls ‘praxeological foundationalism’.

In contrast, Kant saw TI, and hence, the necessary apriority of time, as the only possible solution (A378) to answer the global scepticism that he took the empirical idealism of his peers to entail. Accordingly, the contingent a priority of causal statements in the later Wittgenstein fits in a strategy to deny the use Kantian semantics makes of a priori concepts and to base a conception of the a priori on a development of semantics.

In this conception pure intuition, or, rather, private ostensive definitions as the replacement of their philosophical equivalent, possible experience, play no role (cf. Coffa 1991, 22). Their role is taken over by notions like ‘language game’, ‘forms of life’ that produce ‘grammatical propositions’ of neither deniable nor assertable certainty.