Halliday & Matthiessen (1999: 602-3):
In the last chapter we reviewed the "scientific" model of the mind that informs cognitive science, looking at it from our point of view of construing experience through meaning. We showed that 'mind' is a construct of the ideation base, owing much to the commonsense picture of the world that is embodied in everyday grammar; but problematic because it draws on this account one-sidedly.
The scientific model takes off from the grammar of mental processes (seeing, feeling, thinking), but ignores verbal processes (saying) — although the two are both processes of consciousness, are closely related grammatically, and share the critical feature of being able to create meaning by projection.
It takes off from the ideational metafunction, but ignores the interpersonal — although our folk perception of consciousness derives from both.
Our sense of ourselves as conscious beings comes as much from the fact that we talk as from the fact that we think and feel; and owes as much to the nature of meaning as social action as it does to the nature of meaning as individual reflection.