Biology and Philosophy 16: 103–108, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.

Discussion

Richard Levins as Philosophical Revolutionary WILLIAM WIMSATT Department of Philosophy University of Chicago Chicago, IL USA

I first knew Dick Levins as the author of a provocative article on model building in population biology (Levins 1966), which I read while a graduate student in philosophy of science at Pitt. Model-building in science has been a central concern of mine ever since. In those days, philosophers who talked about model-building tended to talk about correspondences between deductively organized theoretical elements and empirical statements in a thinly disguised application of the correspondence theory of truth. They didn’t have very much interesting to say about either models or theories which scientists would recognize. Even worse, some of them, (like Patrick Suppes or Donald Davidson), insisted loudly that scientists were just confused about models, and claimed that if all of their talk was properly rendered in model-theoretic terms, they would (or at least should) soon realize that! This ignored the fact that scientists claimed to be modelling phenomena, not theories, and what Suppes and Davidson had to say somehow just never seemed to connect with the concerns they had about models. Whether philosophers liked it or not, scientists were just talking about different entities, or talking about them differently, or both. (For a recent critique of the model-theoretic approach to model-building which I sympathize with, see Downs (1992).) Levins was a refreshing change. He talked with the generality of a philosopher about classes of methodological approaches, but what he said was quite different. For one thing, he talked about strategies of modelbuilding, which philosophers never discussed because that was from the (presumedly non-rational) context of discovery, rather than the (seemingly endlessly rationalize-able) context of justification. He took for granted the complexity which biological modellers dealt with, and proposed ways of conceptualizing and dealing with it, rather than quickly changing the subject to impossibly idealized discussions of what one could do in principle, thus

104 reinforcing lessons I had already begun to learn from Herbert Simon. He talked about tradeoffs (between the generality, realism and precision of models), and classified the different approaches of distinct ecological theorists in terms of these – with a recognition that each approach had its strengths and weaknesses, even though he preferred and argued for one of them. This kind of pragmatic pluralism was again not the kind of thing one found among philosophers, though it is emerging today as a common theme among naturalistically inclined philosophers (see Callebaut 1993). I first met Dick Levins in January of 1969, when Dick Lewontin (whom I hoped to post-doc with) sent me over to see him when I was visiting Chicago. I didn’t at first realize that this was the author of Levins (1966) (until I found it in the collection of reprints he gave me), but found someone who spun out models as easily as you could bring up a topic of conversation. You might raise a question about the diversity-stability hypothesis, and Dick would respond (almost conversationally) with “. . . well, consider an n-species Lotka-Volterra community in which the community matrix is asymmetric and the off-diagonal elements . . . ” and he was off on a tour in which, without a board, arrays of equations were specified, their qualitative dynamical behavior was discussed, and the lesson was summarized. This was astounding, and could be intimidating – until you learned that Dick would patiently explain any parts of it you hadn’t understood in as much detail as you would like – as soon as you could figure out where to start. Part of what was astounding was the organization of his constructions on the fly. One of my friends said that Levins was the only person he knew who spoke in paragraphs. One of his most provocative papers on complexity (Levins 1970) arose when British psychologist Richard Gregory came up to him with a tape recorder at a conference cocktail party and asked “What do you think about complexity?” – producing the paper as an essentially unedited transcript. But the net effect was, it all came out awfully fast. Another friend compared listening to a Levins lecture to trying to take a drink from a fire hose. (The students in our later joint courses who taped it definitely had an advantage!) But then (after the usual array of differential equations) Dick would sum it all up with a methodological aphorism, so that anyone could get it at some level or another – something like: “So our truth is the intersection of independent lies.” or “So in ecology, the general does not contain the particular.” (The first is his definition of a robust theorem – something which is deriveable in each of a variety of differently simplified or idealized models of the same phenomenon. The second is a denial that the details of an application of theory to a particular case (either parameter values or structural specifications) are to be found in the general theory – something which makes

105 construction of an application not a trivial matter, and contradicts at least the spirit of the hypothetico-deductive world view.) When I met Dick, he said he was delighted that I was coming to Chicago: Did I want to sit in on his seminar on “Complexity” in the fall? This was a marvelously paradoxical title for a biology course: not “complexity in LotkaVolterra communities”, or “Complexity in multi-locus population genetic models”, although we talked about those cases. Just “Complexity”, with no subject-matter specifications. It could have been a title for a philosophy course, if that hadn’t been beyond the normal philosophical canon. It was very broad-ranging, with students (and subjects) from across the university. A variety of biologists (of different stripes), philosophers, mathematicians, anthropologists, and numerous others together read works by authors from a dozen disciplines, from city planning to the qualitative theory of differential equations. About fifteen of us met in the evening at his home, with his wife, Rosario and their youngest son, Alexandro (‘Handro) usually in attendance, and Ricardo (‘Cardo) and Aurora (‘Lori) often passing through. Life didn’t stop for seminars, and it felt all the more interdisciplinary for that. (Life also included a full variety of political activities – this was the fall of 1969, at the peak of the Vietnam War, the trial of the Chicago seven, Dick had also been active in the Puerto Rican independence movement, and Rosario had a growing interest in feminism. The joke going around was that calling Dick always sounded like it was long distance because there were so many phone taps on his line. 1969 was like that.) Dick and Rosario have been Marxists since long before I’ve known them and urged that science and academic pursuits should always be seen in their social contexts. This was a refreshing wholeness in approach, and lead naturally to a criticism of the “value free” assumptions that arose from a positivist methodology. At this time William Shockley was circulating racist ideas and copies of Arthur Jensen’s slipshod nativist critique of headstart programs in the National Academy of Sciences. This lead to Dick Lewontin’s involvement on the other side, and a number of penetrating articles (see e.g., Lewontin (1974), collected in Levins and Lewontin (1985), and references to earlier articles therein) showing the methodological flaws of Jensen’s analyses, and some of the biases driving them. I was astounded at the shoddiness of Jensen’s arguments and what passed for science, and impressed with how a positivist methodology appeared to legitimate his conception of scientific method. I was also struck with how bad science was so readily and uncritically accepted in the service of worse social conclusions by people who should have known better on both scores, and vowed to made debunking such conclusions a part of my teaching. Levins regarded his Marxism as inseparable from his

106 scientific methodology (see Levins and Lewontin (1985), for a variety of essays illuminating this theme), and regarded his holism as flowing from his Marxism. I got the impression he felt that Marxism was essential to holism, but couldn’t quite understand how this made Marxism more than a personally relevant motivation for his views. I remember arguing with him: “Surely you don’t mean that you had to be a Marxist to be a holist – there are other independent arguments for holism!” It seemed to me that he was committing a historicist or “genetic” fallacy – of regarding an explanation of where an argument came from as relevant to its soundness or justification. According to any philosopher of an “analytic” persuasion, I would have just won the argument, but I’m not so sure. I’d then go on to illustrate my claim with a list of holistic thinkers in various areas of biology and neuroscience. I had about six or eight who (I supposed) weren’t and had never been Marxists. The only problem was that as I learned more of the history of these thinkers over the following 4–5 years, one after the other had to be dropped from the list of counterexamples. I still remember Karl Pribram saying at a conference in 1973: “Well, I see what you’re saying, but actually, I was influenced by Marxism in the late 30’s when I was developing my theoretical orientations, and Roy (E. Roy Johns, an articulate opponent of naive reductionism in neurophysiology) still is!” If I won that battle (and I’m not saying that I did), I surely lost the war. I’m now convinced that Marxism was essentially the only intellectually viable source of holistic ideas through the 1930’s and after. Whatever the in principle arguments say, Marxism may well have been a historically necessary condition for holism in that period. And I no longer regard that – or other historical detail – as philosophically irrelevant, as I once did. Indeed, there is a new interest among philosophers of biology and others in narrative or historical explanations and historicism which shows signs of their (Levins and Lewontin’s) influence and that of Steve Gould. One other influence Dick’s Marxism had was a commitment to following problems across disciplinary boundaries – wherever they lead. This is uncommon in science, and only slightly less so in philosophy, but I took to it naturally. If you want robust answers to real problems it is inevitable. Dick did many of his own models in the seminar – including a variety of cross-disciplinary applications. In one of my favorites, he observed that physicians tended to study the problem of diabetic control solely as a problem in physiology – even though a variety of reports at the time (1971 or 2) indicated that psychological processes had a causally significant role – because they didn’t know how to include the psychological variables. Dick found a way (a little later, when we taught together) through the development of his “loop-analysis” models. (How to solve the mind-body problem? Ignore it,

107 and stick all of the variables into the same model. It worked, and predicted (qualitatively but correctly) kinds of psycho-physical interactions that were later found.) Stuart Kauffman gave an early presentation of his Boolean automata models of gene-control networks in the seminar. I was fascinated by his studies of these networks, which reminded me of neural net models, or instances of networks of statements in the propositional calculus, but never developed in the same way. What about contradictions I asked – how do you avoid them? What’s so bad about contradictions, Dick said – they’re just oscillations in the state of the network! As a philosopher, I was scandalized, but as a modeller, I saw that contradictions didn’t need to be avoided – they could be modelled, indeed should be modelled, dynamically! Philosophers wanted a static (equilibrium) model of timeless truth-assignments, and thus never considered this option. My willingness to consider models of conceptual schemes which contain contradictions dates from this time, and I now think that – although we try to avoid contradictions, and usually succeed in doing so locally, we must assume that essentially all of our conceptual schemes fail to be globally consistent. As a philosopher, I still wanted to apply standard philosophical tools – in this case, analyses in terms of necessary and sufficient conditions. I gave a presentation on the definition of feedback (later published as Wimsatt 1971), showing that all of the extant attempts to analyze it, and all of the plausible alternatives I could think of, seemed beset with problems. I was frustrated. You would think that anything which was so useful, and which applied in so many clear cases had to have an analysis in terms of intuitive necessary and sufficient conditions which would clearly decide each case? But I had been looking for half a dozen years, and hadn’t found any such. This didn’t bother Levins at all: our concepts are heuristic tools devised for the straightforward cases we invented them for – why should you suppose that they would be able to solve all problems? A piece of language was to him like a model – it had its strengths and weaknesses and there were no magic bullets anywhere. Scientific methodology was heuristics all the way down. (Thus a closer look as his notion of a “sufficient parameter” (Levins 1966, 1968) shows it to be a heuristic analogue to the much overused concept of “supervenience” later invented by philosophers. I would argue that every purported case of “supervenience”, once one looks at the real science, is instead a case of “sufficient parameters” (see my 1981), but philosophers of mind still don’t read enough biology or philosophy of biology to know any better.) Those perspectives resonated naturally for me. Dick was a methodological revolutionary in a lot of ways, and philosophy of science – or philosophy of biology at least – has

108 moved a long way in his direction since (Callebaut 1993). Some other areas of philosophy of science could benefit from doing so. Dick could build models and formalize (or more accurately, manipulate, almost playfully, sets of differential equations) with the best of them – but with a difference. To philosophers in those days, formalistic methods were ways of generating secure foundations, crisp definitions, inexorable theorems. To Dick, they were just another technique to study whatever you wanted – no better and no worse, for different purposes, than a well-turned analogy, or a dollar’s worth of field equipment (a can of tuna fish, a box of toothpicks, and some tape for labelling them) with which he took 30 of us out one summer afternoon in 1977 to study behavior in ant colonies. That’s a pragmatic materialism for you! Dick – thanks for 30 years of inspiration!

References Callebaut, W. (ed.): 1993, Taking the Naturalistic Turn, or How Real Philosophy of Science is Done, The University of Chicago Press, Chicago. Downs, S.M.: 1992, ‘The Importance of Models in Theorizing: A Deflationary Semantic View’, in D. Hull, M. Forbes and K. Okruhlik (eds), PSA-1992, Vol. 1, The Philosophy of Science Association, East Lansing, pp. 142–153. Levins, R.: 1966, ‘The Strategy of Model Building in Population Biology’, American Scientist 54, 421–431. Levins, R.: 1968, Evolution in Changing Environments, Princeton University Press, Princeton. Levins, R.: 1970, ‘Complex Systems’, in C.H. Waddington (ed.), Towards a Theoretical Biology, Vol. 3, University of Edinburgh Press, Ediburgh, pp. 73–88. Levins, R. and Lewontin, R.C.: 1985, The Dialectical Biologist, Harvard University Press, Cambridge, MA. Wimsatt, W.C.: 1971, ‘Some Problems with the Concept of “Feedback” ’, in R.C. Buck and R.S. Cohen (eds), PSA-1970 (Boston Studies in the Philosophy of Science, Vol. 8), Reidel, Dordrecht, pp. 241–256. Wimsatt, W.C.: 1981,‘Robustness, Reliability and Overdetermination’, in R. Brewer and B. Collins (eds), Scientific Inquiry and the Social Sciences, Jossey-Bass, San Francisco, pp. 124–163. Wimsatt, W.C.: 1986, ‘Forms of Aggregativity’, in A. Donagan, N. Perovich and M. Wedin (eds), Human Nature and Natural Knowledge, Reidel, Dordrecht, pp. 259–293.