Millikan’s thesis that verbal understanding is an extended form of perception is meant as an alternative to the Gricean thesis that verbal understanding is an exercise in mindreading. The further question arises to what extent Millikan’s teleosemantic account of the proliferation of public language conventions is consistent with the Gricean separability thesis, i.e., the distinction between verbal understanding and either acceptance (belief) or compliance. I will first argue that there is a restricted sense in which Millikan’s teleosemantics seems to be consistent with the separability thesis. But I will further argue that in a broader sense Millikan’s rejection of the mindreading thesis undermines the separability thesis.
On Millikan’s teleosemantic account, for a speaker’s descriptive utterance of an indicative sentence to meet the requirement of cooperation (and mutual interest) between the sender (or producer) and the receiver (or consumer), its direct proper function must be to cause the addressee to form a (true) belief. For a speaker’s prescriptive utterance of an imperative sentence to meet the requirement of cooperation, its direct proper function must be to cause the addressee to act in compliance with the content of the speaker’s request.
In the terminology of the relevance-based framework, a speaker who utters a descriptive utterance makes manifest to her addressee her communicative intention to make manifest her informative intention to make some fact manifest to him. The addressee may fulfil the speaker’s communicative intention by recognizing her informative intention and yet fail to fulfil her informative intention by resisting endorsing the relevant belief. A speaker who utters a prescriptive utterance makes manifest to her addressee her communicative intention to make manifest her informative intention to make manifest to him the desirability of turning some possible state of affairs into a fact by his own action. The addressee may fulfil the speaker’s communicative intention by recognizing her informative intention and yet fail to fulfil the speaker’s informative intention by resisting endorsing the intention to act in accordance with the speaker’s request.
Origgi & Sperber (2000, pp. 160–161), who subscribe to the Gricean thesis of the separability between verbal understanding and acceptance or compliance, have argued that the direct proper function of either a descriptive utterance or a prescriptive utterance could not be to reliably elicit the addressee’s response “at the level of belief or desire formation” (i.e., “the cognitive outputs of comprehension”), but instead “at an intermediate level in the process of comprehension”. Millikan might reply that according to her teleosemantic framework, an utterance may have a direct proper function and yet remain unfulfilled. If so, then the fact that an addressee may fulfil the speaker’s communicative intention (by recognizing her informative intention) and yet fail to fulfil the speaker’s informative intention seems entirely compatible with the teleosemantic framework.
However, to the extent that Millikan explicitly rejects the mindreading thesis, which is presupposed by the separability thesis, it is unlikely that she would find the separability thesis itself acceptable. On the relevance-based approach, it is a sufficient condition for securing what Austin (1975) called the “uptake” (or success) of a communicative act (or speech act) that the speaker causes the addressee to fulfil the speaker’s communicative intention by recognizing her informative intention. It is not necessary that the addressee further fulfil the speaker’s informative intention. Successful communication does not require the addressee to accept either a new belief or a new desire, in accordance with the speaker’s informative intention. But on Millikan’s teleosemantic framework, failure of the addressee to comply with the speaker’s goal of causing the addressee to accept either a new belief or a new desire looks like a failure of the addressee to cooperate with the speaker’s conventional action, and therefore like a breakdown of the speaker’s communicative action. In fact, Millikan (2000, 2004, 2005) has offered two broad grounds for rejecting the mindreading thesis, both of which make it unlikely that she would support the separability thesis; the second of which is based on developmental evidence. I start with the non-developmental argument.
First, Millikan (2004) rejects the mindreading thesis as part of her criticism of the reasoning that leads to the separability thesis: she rejects the joint assumptions that human predictions of others’ behavior are based on mindreading and that cooperation in human verbal communication is vulnerable to the risks of deception. On the one hand, she argues that “most aspects of social living involve cooperation in ways that benefit to everyone […] for the most part, social cooperation benefits both or all parties. There is nothing mysterious about its evolution in this respect” (Millikan 2004, pp. 21–22). In a nutshell, Millikan argues that the urge to explain how the benefits of human communication are not offset by the risks of deception is misplaced on the grounds that the interests of speakers and hearers are sufficiently similar, if not identical.[10]
On the other hand, she argues that we use belief-desire psychology, not for prediction, but “for explanation after the fact” (Millikan 2004, p. 22). This is consonant with her (1984, pp. 67–69) earlier claim that while human adults have the ability to reflect on a speaker’s communicative intention if the automatic flow of conversation is interrupted for one reason or another, normal verbal understanding does not require representing the speaker’s communicative intention. Instead, normal verbal understanding should be construed as a conventional transfer of information whereby the speaker translates her belief into an utterance, whose meaning is in turn translated back by the addressee into a newly acquired belief.
Thus, Millikan rejects two of the major assumptions on which the separability thesis rests. She underestimates the gap between the interests of speakers and hearers in human communication and she minimizes the role of belief–desire psychology in the prediction of others’ behavior. Interestingly, her rejection of both assumptions rests in turn on her own competing account of communicative acts. As she puts it, “a surprise of this analysis of the conventional nature of the information-transferring function of the indicative is that believing what you hear said in the indicative turns out to be a conventional act, something one does in accordance with convention” (Millikan 2005, p. 46).[11] First of all, Millikan (2004, p. 23) argues that humans expect others to behave in conformity with social conventions, not on the basis of others’ beliefs and desires. Second, she further speculates that the conventional behaviors that are caused by a disposition to social conformity may derive from natural selection the memetic function of serving a coordinating function (ibid.).
Clearly, being disposed to social conformity and expecting others to be similarly disposed may help solve coordination problems (as shown by driving on one side of the road). However, being disposed towards social conformity is not sufficient to comply with social conventions. Compliance requires learning, i.e., the acquisition of relevant true beliefs about the contents of social conventions. Thus, the basic challenge for Millikan’s claim that humans expect others to behave, not so much in accordance with the contents of their beliefs and desires, but in conformity with social conventions, is to offer an account of how humans come to learn and thereby know what it takes to act in conformity with social conventions.
This issue has been highlighted by the exchange between Tomasello’s (2006) comments on Millikan’s (2005) book and Millikan’s (2006) response, which focuses on Millikan’s (1998) thesis that many conventions, whose function it is to solve coordination problems, are reproduced by what she calls counterpart reproduction (or nuts and bolt reproduction). Typical coordination problems involve at least two partners, who share a common purpose that can be achieved only if each partner plays its assigned role, where both partners can be required to perform either the same act or two distinct complementary acts. In counterpart reproduction, when the respective roles of each partner require them to perform two different complementary acts, one typically adjusts her behavior to the other’s and vice-versa. Counterpart reproduction is exemplified by, e.g., handshake reproduction, the reproduction of the respective postures assumed by men and women in traditional dancing, the reproduction of social distances appropriate for conversation, or the reproduction of the use of chopsticks for eating. Similarly, Millikan (2005, 2006) argues that counterpart reproduction also underlies the continued reproduction of conventional public-language signs.
Millikan (2005) further mentions open, partially or completely blind, conventional leader–follower co-ordinations involved in joint actions based on shared goals, whereby one agent (the leader) introduces a component of a pattern whose completion requires her partner (the follower) to perform a complementary component (ibid., pp. 12–14). One example of open conventional leader–follower coordination is the pattern whereby one agent selects her seat at an arbitrary table in a restaurant and her partner follows suit and selects his accordingly. One example of a partially blind conventional leader–follower coordination is the couch-moving pattern whereby the leader affords the follower anticipatory cues of her next move by ostensibly exaggerating her own movements, where the follower’s familiarity with the pattern enables him to recognize the leader’s ostensive cues and thereby to reproduce the complementary portion of the joint action. Another of Millikan’s examples of a partially blind conventional leader–follower coordination is the US mailbox-flag convention, whereby the leader puts up a flag after she has placed mail in the mailbox and the postman picks up the mail after perceiving the flag.
Much comparative work by Tomasello and colleagues (reported by Tomasello et al. 2005 and summarized by Tomasello 2006, 2008) shows that while most communicative gestures in chimpanzees are learnt by ontogenetic ritualization, most communicative behaviors in human infants are acquired by imitative learning. As Tomasello (2006) argues, Millikan’s own requirement that the reproduction of a conventional pattern depends on “the weight of precedent”, not on its perceived intrinsic superior ability to produce a desired result, seems better fulfilled by a process of imitative learning than by a process of trial and error whereby one individual adjusts her behavior to another’s. There seems to be nothing arbitrary (as there should if it were conventional) about an individual’s adjusting her behavior to another’s. While Tomasello (2006) does not deny that counterpart reproduction plays a significant role in cultural transmission, he disputes the claim that the output of counterpart reproduction qualifies as conventional.
Part of the gap between Millikan and Tomasello lies in what they take to be the proper unit for the analysis of the mechanism underlying the continued reproduction of conventional patterns involved in solving problems of coordination. While Tomasello focuses on the learning capacities of single individual minds, Millikan focuses on what can be achieved by the reciprocal adjustments of pairs of cooperative partners. For example, when Millikan (2005, 2006) argues that counterpart reproduction underlies the continued proliferation of the custom of using chopsticks for eating in some cultures, she construes the convention of using chopsticks as a solution to the problem of coordination between pairs of partners, some of whom buy chopsticks and use them for eating and others who manufacture chopsticks. The latter would not manufacture chopsticks unless the former bought them and used them for eating. Conversely, the former would not buy them and use them for eating unless the latter manufactured them.
But of course, as Millikan is aware, this leaves open the question of how young children learn to use chopsticks for eating. As Millikan (2006, pp. 45–46) rightly observes, young human children understand their native language long before they can speak it. Nor can they learn to understand by imitating mature speakers: as she puts it, “they don’t watch how other people understand and then copy”. She further argues that young children would never understand their native tongue unless “their teachers” spoke to them, but “their teachers” would never speak to young children unless “they had had some reasonably successful experience” with previous listeners. This makes the continued reproduction of conventional public-language signs fit the pattern of counterpart reproduction. But still the question arises: how do young children learn to produce words of their native tongue? Vocal imitative learning may well play an important role (cf. Hauser et al. 2002). In a nutshell, according to Millikan the function of conventions is to solve coordination problems. She offers an elegant account of the proliferation of conventions based on counterpart reproduction. Her account must make room for the role of imitative learning in the way young human children learn either to use chopsticks for eating or to produce (and not just understand) words of their native tongues. As I shall argue in section 6.2, evidence shows that imitative learning in young children rests on their ability to construe the model’s demonstration as an ostensive communicative action. If so, then Millikan’s view that counterpart reproduction underlies the proliferation of conventions must make room for the role of children’s ability to recognize the model’s communicative intention.