I agree with your rough summary. As to your question: for the process of differentiation, one has to look at the neural level – sharing some neural pathways conceptually does not mean that there cannot be a differentiation.

The important thing rather with CMT is that it really shows how we can _understand_: I am not so much concerned with high level metaphor blending as in your example, but how understanding bootstraps in the first place.

And from scratch, all we have to go with is our sensorimotor perceptions. So understanding can only build on propping metaphors on these semantic primitives. That is the crucial point why I find CMT compelling.

Differentiation etc is something for neuroscientists to work out

It seems that CMT makes the strong claim that with a metaphor like “good is up”, the constructions we invoke for “good” include some of the constructions we invoke for “up” – in other words, metaphor reflects some underlying unity of representation (because metaphor is the basic process by which we make new constructions). This is opposed to the idea that the concepts each have their own separate set of constructions that might merely be similar enough to make such a metaphor easy to comprehend.

If you agree with this rough summary, how does CMT explain how concepts become differentiated from one another? In other words, what determines the boundaries of the metaphor? For example, if “argument is war”, how does CMT explain that most of us, hopefully, don’t expect to literally destroy our opponents if we win an argument?

Terren