In general, a metonymic model has the following characteristics: - There is a “target” concept A to be understood for some purpose in some context. - There is a conceptual structure containing both A and another concept B. - B is either part of A or closely associated with it in that conceptual structure. Typically, a choice of B will uniquely determine A, within that conceptual structure. - Compared to A, B is either easier to understand, easier to remember, easier to recognize, or more immediately useful for the given purpose in the given context.
- Lakoff (1987), Pag. 84-85

We have seen the following kinds of metonymic models: social stereotypes, typical examples, ideal cases, paragons, generators, sub-models, and salient examples. They have a cognitive status, that is, they are used in reasoning. And they all yield prototype effects of some sort.
- Lakoff (1987), Pag. 90

Metonymic models are models of one or more of the above types, together with a function from one element of the model to another. Thus, in a model that represents a part-whole structure, there may be a function from a part to the whole that enables the part to stand for the whole.
- Lakoff (1987), Pag. 114

Metonymic models do not mirror nature. If metonymic models are real-if they are used to make judgments and draw inferences, and if they lead to prototype effects-then they constitute counterevidence to objectivist cognition. They constitute a kind of conceptual resource that is not objectivist.
- Lakoff (1987), Pag. 204