The interpretation of Wittgenstein on rule-following is by no means straightforward. It will probably be most helpful if I present Norman Malcolm's version since this offers the
or 'a' - standard case against private rule-following. The basic idea is that rule-following is inherently public since without public scrutiny, correction and control there is no way in which one can tell (certainly the putative private rule-follower can't) whether s/he is really following a particular rule or merely thinks they are :
The standard account - Norman Malcolm
In 'Wittgenstein on Language and Rules', Norman Malcolm argues
that following a rule must be, and according to Wittgenstein is, a social
practice. This social practice requires 'a human community in which
there is agreement as to whether doing such-and-such is or is not
following a particular rule' (p. 5). Malcolm contends, against the
individualist position of Gordon Baker and P. M. S. Hacker, that a
person's following a rule in complete separation from other human
beings for his whole life is impossible. Malcolm also contends that this
was Wittgenstein's own view. ... My main concern is Malcolm's argument for the
view that rule-following requires a social community.
Let us begin with Malcolm's interpretive contention. He holds that
Wittgenstein's remarks on solitary language users concern only individuals who were trained in a linguistic community. Malcolm denies
that Wittgenstein ever suggested the possibility of a 'forever solitary
person with a language', i.e., someone who follows linguistic rules but
has not been trained to do so by the members of a linguistic community.
Malcolm contends that Wittgenstein's solitary rule-followers all share a
key feature of Defoe's Robinson Crusoe: they have acquired language
via training in a social context. On this reading, Wittgenstein holds that
all language must be socially established, but grants that Crusoe can
speak a socially established language to himself.
Malcolm believes PI 202 supports his contention that social training
is essential to rule-following and language. The presumed support
comes from 'the necessary distinction' between one's following a rule
and one's thinking one is following a rule. Malcolm explains:
If you try to imagine someone who had never participated in human
society, inventing a rule for himself and undertaking to follow it, you
will realize that there would be no foothold there for that necessary
distinction. . .. There must be a use of a sign that is independent of
what an individual speaker does with it, in order for the latter's use of
the sign to be correct or incorrect' (p. 28).
Social training is necessary, on Malcolm's account, because it provides
the independent basis for correctness and incorrectness of language
use. More specifically, 'what a rule requires and what following it is,
presupposes the background of a social setting in which there is quiet
agreement as to what "going on in the same way" is; [and] this is an
agreement in acting, not in opinions' (p. 21). In sum, then, social
training is necessary because it provides the background of 'agreement
in acting', which provides the independent basis for the 'necessary
distinction' between correctness and incorrectness of language use.
We now have Malcolm's answer to my crucial question raised above.
Problem with the standard account
But his answer faces a troublesome question: Why should we think that
'agreement in acting' with respect to the actions of a group of people can
provide for the distinction between correctness and incorrectness,
whereas 'agreement in acting' with respect to various actions of an
individual cannot? This is the crucial question that Malcolm must
answer, but does not answer.
My own view is that I can see how it is impossible for a private rule-follower to know whether they are really following a rule or merely (mistakenly) thinking they are. But I see no incorrigibility in a public dimension unless one takes an extreme conventionalist position that the rule and following it just are what other language users decide or judge that it is, so that (from this perspective) they cannot be mistaken.
Norman Malcolm, 'Wittgenstein on Language and Rules', Philosophy 64 (1989), 5-28.
Gordon Baker & P.M.S. Hacker, 2 Wittgenstein: Rules, Grammar and Necessity (Oxford: Blackwell, 1986), chapter 4.
Paul K. Moser, 'Malcolm on Wittgenstein on Rules', Philosophy, Vol. 66, No. 255 (Jan., 1991), pp. 101-105.