Another person whose intellectual habits influenced me at Columbia and who was of quite a different sort than any of the others I have mentioned was Paul Oskar Kristeller. He was interested in the fine details of historical scholarship, famous of course for his work in Renaissance philosophy, but I was more influenced by the seminar on Kant that he and Randall gave. Kristeller’s meticulous insistence on a textual basis for every interpretation suggested in the seminar, and his decisive way of handling the text, were a model of scholarship that I admired and learned from, even though Kristeller always modestly insisted that he was in no sense a Kantian specialist.

I received my PhD from Columbia in June of 1950 but my education scarcely stopped there. Although I began teaching upon arrival at Stanford in the fall of 1950, where I have been ever since, almost as soon as I could I also began to think about research in the philosophy of science. My problem was that I did not really know much about how to do any serious research and writing, since my graduate education did not involve the personal supervision of research of the kind so familiar to graduate students today. I was neither forced nor encouraged to produce early in my graduate career a publishable paper nor to because acquainted with the "how-to-do-it" aspects of research.

I was, however, full of energy and brimming over with ideas. I thrashed around for a few months but fortunately I soon became acquainted with J. C. C. McKinsey, a logician who had recently joined the Department of Philosophy at Stanford. McKinsey served as my postdoctoral tutor. It was from him that I learned the set-theoretical methods that have been my stock in trade for much of my career. It was not, however, just set-theoretical methods as such that MeKinsey taught me but also a passion for clarity that was unparalleled and had no precedent in my own prior education. I remember well his careful red-penciling of the initial draft I gave him of my first article on the theory of measurement (195la). McKinsey was just completing a book on the theory of games and consequently also had some of the interests that were of importance to me as my own interests in the social sciences began to blossom.

But there were other people, at Stanford and Berkeley who continued my education in those early years at Stanford. With McKinsey’s encouragement I attended Alfred Tarski’s seminar in Berkeley. McKinsey always claimed that he had learned everything he knew from Tarski. This was not true, but after attending Tarski’s seminar I understood why he liked to say this. Tarski was a ruthless taskmaster. I think he probably got the most out of his students of anyone I have known. His seminar provided perhaps the best example I know of vicarious learning. A student who made a poorly prepared seminar presentation was so ruthlessly and mercilessly questioned that the other students did not need any hints about the state of preparation they should achieve before making their own presentations. Tarski, as one of the great examples of the Polish school of logic, was unwilling to go forward on a single point unless everything covered thus far was completely clear—in particular, unless it was apparent to him that the set-theoretical framework within which the discourse was operating could be made totally explicit. It was from McKinsey and Tarski that I learned about the axiomatic method and what it means to give a set-theoretical analysis of a subject.

McKinsey died in 1953, and our collaborative work on the axiomatic foundations of empirical sciences was brought to a sudden stop. Another learning experience that influenced much of my later work was the summer research position I had early in the fifties for a couple of years with David Blackwell and M. A. Girshick while they were writing their influential book Theory of Games and Statistical Decisions (1954). As I remark later, I did not learn as much from them as I should have, but what I did absorb played an important role in a number of my early papers.