[quoting Wild, 4 Apr 96] Certainly there is much confusion arising over the idea of mental models, simply because accounts differ from one theorist to another. But basing our understanding on Johnson-Laird's considerable and most recent work in this area, mental models are essentially cognitive tools that have predictive and other cognitive properties. A mental model is a mediating intervention between perception and action...
I agree about the confusion, and I would just like to make a small observation about the conceptualization in the last sentence. There seem to me to be quite close, though certainly not fully overlapping, similarities between Johnson-Laird's mental models, Rumelhart's "schemata," Minsky's "frames" (in Howes, 1991), Schank and Abelson's "scripts" (1977), and Mezirow's "meaning perspectives" (1991). They all deal with expectations (Martyn refers to their "predictive" properties) and seem to determine that a perception "is" (NOT "represents") such and such. In other words, I find it difficult to think usefully of an "interface" between a perception and a mental model of whatever if (as I understand to be the case) it is IN TERMS OF the mental model that the perception is what it is and not something else. I guess the test is that if some influence (learning) modifies your mental model, it shifts your perception in parallel. If "mental model" and "perception" always shift in a parallel and co-occurring way, where is the space for an interface, or intervention? This leaves us with a two-stage rather than a three-stage process. I acknowledge what this view owes to constructivism.