Halliday & Matthiessen (1999: 599-600):
At the same time the 'scientific' models of the mind fail to extend consciousness in the way it is extended by the grammar of English. There are, in fact, two complementary perspectives embodied in the semantic and grammatical systems of English; and together they point towards an alternative interpretation both of 'information' as constructed in cognitive science and of the individualised 'mind' that is its object of study.
(i) Ideational: Sensers and Sayers. The ideational resources of language are primarily a theory of experience, so they are reflected fairly directly in consciously designed theories such as those of cognitive science. If we stay within the ideational metafunction, where mental processes are construed, we also find other processes that are complementary to these: those of saying (verbal processes) and those of symbolising (a type of relational process).
(ii) Interpersonal: interactants. If we move outside the ideational metafunction to the interpersonal, the resource through which we interact with other people, we find that here we are acting out our conscious selves — "modelling" consciousness not by construing it but by enacting it. Since this kind of meaning is non-referential it is not taken account of in scientific theories at all.