Toward the end of my undergrad years in math. and physics, most of my fellow students were focused on the complex experimentation required to verify/falsify theories. I became more fascinated by the problem of where theories came from, how we could make better once, and how we would know that an even better theory might not be right around the corner, but hadn't been invented yet. I read about and discussed this question in philosophy of science, but I became convinced that it should be investigated via science.
My first idea was that the creation of scientific theories is a sociological problem. I went to UBC to investigate counterfactual religious theories about the world, in the idea that that might provide some insight. I ended up with the question of finding the information content in theories, using logic, information theory, and computer modelling, which I presented in Ottawa and submitted as an M.A. thesis.
I then went to U. of Michigan in the Doctoral Program of Social Psychology, to develop a computer simultation of language comprehension and usage, with the hope that a working simulation based on logic would take me further. The idea of a universal semantics, somewhat like Chomsky's universal grammar simply did not pan out.
I then went on to work with simplified subsets of language to re-engineer tasks and move along computer-assisted workplace automation. I became convinved that the search for a universal semantics was simply the wrong question.
I became convinved that a focus on tasks and action was approach. I also became convinced that one needs to look at evolution to see how the complexity of skills is developed and maintained. After years of building computer-based models for many tasks, I am convinved that information processing is central to working with complex skills, and that language was a central element in such information processing.