The molecularistic option is normally considered the stance where you have some basic beliefs (or some basic semantic properties) which are definitory of the meaning of words and sentences. This option falls under the criticism of Fodor and Lepore who claim that any attempt to build up a molecularist stance (we share only some beliefs) drives you directly either into holism or into the acceptance of the analytic/synthetic distinction (either meaning is given by all beliefs or it is given by analytic definitions). However Perry 1994 and others claim that there is a weak form of molecularism which escapes this criticism. We have therefore to distinguish a strong and a weak molecularism, where the distinction is given by the scope of the quantifier (I use here Marconi 1997):
- the strong molecularism, which is supposed to lead to holism or to A/S distinction, might be expressed as such:
p q (q≠p & Nec (p is shared q is shared)
that is, if two people share a belief p, there is some other belief q which must also be shared. Fodor and Lepore claim that, unless you give a class of privileged sentences, you have no idea of how to choose the sentences q to be shared. But to isolate a class of privileged sentences is to come back to the idea of a set of beliefs analytically connected with p, therefore to come back to the analytic/synthetic distinction, which is to be rejected.
What is wrong, however, with analytic/synthetic distinction? Quine himself was certainly aware of the apparent utility of a weak distinction between analytic and synthetic, or theoretic and observation sentences. He claims that there is no "principled" distinction, universally valid; however, for practical purpose a distinction of the kind is needed. We might therefore accept a relatively innocuous distinction of the kind linguistic/factual (practically most of the knowledge representation systems accept one, since the distinction between assertional and descriptional in the KL-ONE semantic networks). Certainly strong molecularism (which seem the position still hold by Dummett 1991) is a stable position and coherent with the traditional view of communication. However, for the sake of the argument, we grant a suspicion on the appeal to any kind of distinction between analytic and synthetic sentences. Is it possible to build a molecularist position completely free from this danger?
- the weak molecularism, which is supposed to avoid both holism and the A/S distinction, might be expressed as such:
p Nec (p is shared q (q≠p & q is shared)
that is, necessarily, if you share p, there are other beliefs that are also shared. However there is no privileged class of beliefs; there must be other shared beliefs if p is shared, but there is no reason why these beliefs should belong to a class of analytic or privileged sentences. It is easy to see how the weak molecularist option is different both from the strong molecularist one and from the holistic one. Certainly you do not need to understand all the language in order to understand a sentence; on the other hand this option maintains the classical view that communication is based on sharing some common content (which content, is not relevant at all; if you say to a hearer that your ring is gold, maybe you share just the wrong belief that gold is yellow, or the belief that gold is a metal; who cares? ). However the position is highly unstable. To assert that you need to have some shared beliefs q if you hold a belief p cannot avoid the question of which kind of beliefs we expect to be shared. If two people may share few beliefs, certainly they are expected to hold some beliefs which are relevant to the matter. Maybe they do not share all the relevant beliefs, but we, speakers of a language, must distinguish what is relevant and what is not. Sharing beliefs is on one hand an individual matter, on the other hand a social matter; there are beliefs shared in the society, even if not shared by all the individuals. Idealized individual competence is neither a universal encyclopedia nor a set of casual beliefs. If communication is, on this view, a sharing of contents, we cannot avoid the normative problem of which contents are the best to be shared to improve successful communication.