Halliday & Matthiessen (1999: 566-7):
The everyday folk models are more likely to be embraced unconsciously by everybody in a culture, because they are everyday models, instantiated in casual conversation, and because they are construed as congruent in the cryptogrammar. The general model of the phenomena of our experience, including those of our own consciousness — seeing, thinking, wanting, and feeling — is of this highly generalised kind.
In contrast, scientific models are much more contextually constrained: they are developed, maintained, changed and transmitted within those situation types that we associate with scientific language. … these situation types are quite restrained relative to the context of culture as a whole … In this respect, scientific models are clearly sub-cultural models: contextually they are located somewhere between the potential and the instance.
… an inherent property of instantiation is variation; and scientific models (like other subcultural models) vary one in relation to another. Sometimes they are complementary, sometimes they conflict.