Regarding the "primacy" of affect or cognition, I liked the dual interacting streams metaphor several posts back and the idea just put forward that we artificially carve up biological processes in order to comprehend them - leaving us confused.

Describing two dynamic biological processes as streams (of thought and affect) which interact by "setting the context or state" for the next moment in the other stream (and vice versa) is a mental model that seems more truthful than a static box-and-arrow model and that preserves the distinction necessary to do cognitive therapy. I wonder if there are some useful mathematical tools to help model these two dynamic and interactive processes.

The problem is (and this also speaks to the assertion that "affect is first") that perception has to be included and distinguished from cognition in the model. It seems to me that the model has to start with "pure perception" at some (early) point in development... and then we need some developmental framework to account for the increasing and yet variable influence of memory, automatic and conscious association upon perceptual events (as well as the affective influence). So I think the two-stream model works, but the cognitive stream needs some extra work. It seems to me that including and developing a distinction between "pure" and "influenced" perception would be helpful and important to the clinician. What we're after is a model that is pretty accurate and more importantly, that helps us do effective therapy.

Distinguishing between "automatic" and conscious is also important, but not sufficient - for there are automatic and relatively "influenced" or "distorted" appraisals (which are of course impacted by the conscious work of CT despite being automatic) and there are conscious and relatively veridical perceptual events...