In the late 50�s and early 60�s I started out in theoretical physics at the University of Saskatchewan.� During my time there I was actively involved not just in social causes but also intellectual discussions.� Toward the end of my undergraduate studies I came to the view that the most interesting question was how physicists came to formulate new theories.� First question: How do we know that we were not overlooking a yet better theory.� In contrast, most of my classes focused on the verification (falsification) of existing (already formulated) theories.� As an illustration of the direction in which my interests were veering, early on (1961) I became fascinated by the apparent tension between the theory of evolution, leading to greater complexity, and the second law of thermodynamics, leading to a reduction in differentiation.
Taking my questions/speculation seriously, I next went to UBC to study Sociology and Anthropology, with a view of studying small religious groups with unsupportable and falsified theories, such as that the end of the world was at a given date � and then the date passed.� I also looked at small groups with peer pressure, etc.� During this research I became aware of the difficulty of formulating theories about the process of creating and validating theories.� The advantage of physics was that theories were formal and in mathematics, so that predictions could be �cranked out mechanically�.� I then looked at computer simulation models (in Fortran, in punched cards, in the mid 60�s), as well as at Information theory (Shannon, etc.).� This became part of the focus of my masters research in Sociology.� During that process I came to more of an appreciation of the role of language in theorizing, and decided to continue by investigating language comprehension through computer simulation.
In 1967 both Stanford and the University of Michigan decided to give me a chance to try my ideas as research for a Ph.D.� Michigan was more flexible in allowing me to combine Social Psychology with Computer Science, Linguistics, and some Philosophy, Math, etc. So to Michigan I went (with the support of a Canada Council Fellowship).� I was fortunate to have some excellent mentors and great faculty.� I could explore mathematical psychology and measurement theory, computational linguistics, modal logic and the foundation of computers, and many other topics.
My research focused on building a computational model of language comprehension using Lisp and Logic as the underpinnings of the semantic system.� (more)
This brought me to the question of validating a model of language comprehension.� Since computational models are very complex, it is not even clear how many degrees of freedom they have, and what measurement would suffice to falsify them.� I decided on an engineering approach:� provide support through use.
The initial approach was in education.� Starting at the University of Michigan, and later at Queen�s University in Kingston (1974 ..), I used a computational model of language comprehension to help in teaching university students to understand the scientific method.� At Michigan, I split a class into groups, where one group could formulate �scientific laws� into the language-based model.� Another group would then �run experiments� to try to uncover these laws.� At Queen�s I built an elaborate simulation, where the students were the researchers giving orders in simple English to the computer-based Research Assistant.� Based on the student�s design, the computer would �run the experiment� and report the results � based on a simulation.� The student would then order the computer to analyze the data with standard statistical techniques.� (Graduate students would provide the simulation models based on their research and on literature reviews.)
At the same time I was running some experiments were a student would predict how another student would answer some questions or deal with a social situation.� (A simplified version of this kind of approach briefly became a TV fad.)
The results of these investigations led me to question my understanding of the underlying semantic space.� Following the research of Chomsky, and the theory that there was a universal grammar for all humanity, there had been a quiet assumption that there would be a parallel universal semantic space.� (This assumption had been questioned by some research in Anthropology.)
On gradually leaving the university, I decided to investigate by doing knowledge engineering across most industrial sectors.� My tentative conclusion is that there are overlapping semantic spaces mostly following professional training and application areas (industrial sectors).� It became clear to me that schooling played a large role in establishing shared semantic spaces.