The following is a transcript of a videotaped interview of
Dr. Robert Rosen
done in July, 1997, in Rochester, New York, U.S.A.
by Judith Rosen
(This is a corrected version [July 14, 2000] thanks to the help of Esther Wieringa)
JR: Hello, my name is Judith Rosen. This is an interview with Dr. Robert Rosen. You're looking at a painting that he did in oils of a famous lighthouse in Halifax, outside of Halifax, Nova Scotia.
JR: This is an interview for Belgian television for a set of meetings in August that are about anticipatory systems. You wrote a book with the title Anticipatory Systems in the late 70s, which came out in the early 80's sometime. Could you talk about what anticipatory systems are ?
RR: Well, “anticipation”, the way I use the term, is a style of control. And it's based, not as cybernetic systems are on a deviation from a desired behavior, anticipation is based on having a predictive model of the system you're trying to control and using the predicted behavior to generate the control which will modify the behavior in a desired way. The combination of the free system that you want to control and a control based on such a model, predictive model… its behavior is what I call an “anticipatory system”.
JR: Now, is anticipation an emergent property?
RR: In general, I would say it is an emergent property but that's another aspect of their behavior. When I was trying to develop the idea of an anticipatory system, I was interested more in how it behaved as such than in how it got to be an anticipatory system. But, in fact, anticipation does have emergent aspects. Both emergence and anticipation are different aspects of what I call “complexity” in a system.
JR; You're working on a book right now called “Complexity”.
JR: What is complexity?
RR: Well that's a little bit harder to describe. Complexity is really recognized by the failure of all our attempts to deal simply with these systems. Simplicity is easier to define. I define a system to be simple if it has certain properties and anything else is a system that isn't simple; I call “complex”. Simplicity is one of the things we inherited from physics; a philosophy of science: all systems can be broken up in a certain canonical set of ways and all systems are built up out of pieces that arise from such decompositions, again in a certain canonical set of ways. So, a system is simple if you can take it apart in a familiar fashion or put it together from pieces in a familiar fashion. That's what basically it means for a system to be simple. The whole idea behind physics was that all systems were simple. And that's the way science progresses, by finding the right pieces and the right ways of putting the pieces back together. The lesson I bring from biology is that most systems, MOST systems are not even simple. Most systems are more like organisms. There's no one fixed set of parts into which they can all be decomposed…
JR: This is one of the more controversial ramifications of your research, isn't it? That physics is a limited science…
RR: That's right. Especially physicists don't like that. But it's unfortunate that most systems are more like organisms than they are like machines– simple systems.
JR: And the way they are more like organisms is that they are “complex”…
RR: Exactly. They are complex, and as such they have properties with anticipation, they have properties from emergence and many other kinds of similar things arising from their very nature and they cannot be understood comprehensively the way a physicist likes to understand a rock or a grain of sand.
JR: Now, what is it about complexity that makes it different… a complex system different from a simple system?
RR: Well, as I say, complex systems cannot be all decomposed through the same algorithm, there's not one way of taking it apart or putting it back together. It's a much wider set of possibilities than you can get out of a simple system.
JR: Is that because there are relationships in a complex system that can't be measured by any current technology? And the relationships are part of what make the system what it is?
RR: Well, I think that's true. When we try to interact with an organism as a physical system, we find ourselves interacting with it in many more ways than we have instrumentalities. And we have our own physical aspect, which is much richer than we give it credit for. So as you would expect just from the intuitive use of the term complexity, you have many more possibilities, many more capabilities, and many more modes of interaction than just a limited set of canonical ones.
JR: And most experimental scientific techniques look only at the parts and not at the relationships between the parts…
RR: Exactly. And. also they look in limited sorts of ways even with those interactions. And ifs true it's interesting to see, you know, how far you can get with just one limited set of interactions. But that's not all there is to nature. That's like trying to write all music in diatonic scale– there's a great deal of music that cannot be expressed easily that way.
JR: Now, your book, Life- Itself. How do anticipatory systems relate to living systems?
RR: Well, living systems are anticipatory in general. The way they adapt, the way they learn to navigate in the world is by means of models. Actually, I didn't use the concept of anticipation much in that book. I was trying to really come to terms with some aspects of complexity… trying to explore some of their philosophical or epistemological consequences… why organisms are different from trying to do physics. I was really trying to explore the limitations of the machine [metaphor] rather than the full fabric of complexity in that book.
JR: Now, if life is an emergent property of a complex system, and all living systems are anticipatory, your book on complexity is going to explore how these things develop in a complex system?
RR: Well, again, complexity is more than life. Complexity is, I feel, a basic feature of the world, the material world, even the world of physics, the world of machines.
JR: From the atomic level on?
RR: Yeah. And I want to explore the… sort of, well, they're quasi-biological… they're not part of biology themselves, not originally biology…
JR: Will you talk about how complex systems develop? Or how they become more complex?
RR: I hope to, insofar as I understand it. But again, complex systems– their nature is that they avoid one mode of understanding of them. They are —-, they have more capabilities than any one formalism, say, can grasp. What you need to be able to do it is to put all of these formalisms together to indicate sort of the very rich mix of properties that these have, which you wouldn't expect from approaching them in a straightforward physical way. It's just Re-doing mathematics. Mathematics is richer than any one part of itself. Mathematics is richer than geometry, say. Mathematics is richer than algebra. And richer than any finite single combination of parts of mathematics. And so if you want to understand the full potentialities of mathematics as a habit of mind, say, you have to go more by example. You can't do it by construction, which is basically collapsing everything down into one formalism again.
JR: OK, we're back. Why don't we talk about some of the more controversial ideas or aspects that are ramifications of some of the theories you've been developing? What was the most surprising thing to fall out of these theories– anticipatory systems, complex systems?
RR: Well, again, this depends on what you consider familiar or unusual. What kinds of intuitions you have about the way things “should be”. The tendency over the last several hundred years, perhaps since Newton, is to try to capture all of the world, the external world, everything that science pertains to, in one principle-one way of grasping reality. And that leads directly to the concept we call the “machine”. So nature is a big machine, an organism is a machine, mechanism is the goal and the end of science, and mechanism itself can be embodied in one principle or one set of principles. They're the principles of Newton, the principles of Descartes, or they're principles of mathematics… There are many attractive features, which flow from the idea of the machine. One of them is the idea of objectivity. You want to explain nature in a way in which individual consciousness, or “will”, has no part. That's what it means to say that nature is “objective”. If you ask most people what they understand by objectivity, that's what they will tell you. Consciousness, or will, or volition, all of the things which are characteristically human, play no part. As I say, that has been attractive and that has set up the ideal. And that is partly why the Cartesian ideal of the machine was so nice; because it's inherently objective. If something can be done by a machine, then it clearly doesn't involve will, doesn't involve subjectivity or consciousness or anything like that. And that has animated most of epistemology for the last 300 years. Part of the attractiveness of mathematics was that it embodies this kind of objectivity, even though mathematics exists only in the mind. Well, anyway, complex systems are not like that. If you try to compress a complex system into that kind of mold, you'll miss it completely. And I was getting pushed more and more to the idea that organisms are not machine-like. And that to understand them, you had to get rid of these vestiges of the machine. You need consciousness. You had to run around into… circles of causality, circles of time; there was no way of straightening out causal loops… This was, in fact, the way I defined complexity in my last book on anticipatory systems, uh excuse me, on complex systems…
JR: Life. Itself, you mean?
RR: Life- Itself, yeah. That there were… that you could not straighten them out in the way you could do with automata or simple systems. And that you had to go, not to one set of tools from which everything could be constructed and in which you would never get into a causal loop…
JR: Well the whole idea of anticipation…
RR: Well, anticipation has these temporal loops in it.
RR: You have to have more than one time scale, more than one thing that you could call “real time” in an anticipatory system. In my first approaches, the anticipatory system was based on predictive models, as I've said before. Something in the system is running faster than real time in the system, or else you have no anticipation– the system is not anticipating what its own subsequent behavior is going to be. So, you have to have more than one inherent time scale. Then I realized that the time scale is related to a way of decomposing the system into units or parts and that implied the capability of reducing what you would call the same system into different parts that were not equivalent. Now, in mathematics itself there are many situations where a way of taking the system apart is inherently limited. You can show that it is inherently limited. Just to give you a simple example: Measure Theory is a way of talking about volume or area or lengths, and its very much tied to properties of intervals. Intervals have lengths associated with them. There are certain mathematical operations you can perform on sets with lengths to bring you new sets, which also have lengths. And when people were developing these ideas of measure theory, which technically are a little over a hundred years old now, they wanted to see whether you could extend the idea of lengths so that if you had a bunch of sets that you could measure, you could extend that notion of measurement to give any set a length. And it turns out that this is not possible. Most sets, in fact, don't have a length because most sets are not constructible from intervals, in an easy fashion. So there are these huge families of nonmeasurable sets. And if you have two different kinds of sets embodying two different kinds of measures, that's one of the prototypic examples I always think of about complexity. It's generic for sets not to be measurable. But our experience is such, we are so closely tied to intervals, that to us a set without a measure is a very unusual, pathological kind of object. But that's sort of what complexity is like. It's like having many different notions of length co-existing in the same universe. And you have to talk about all of them in order to deal with them. So, starting from the idea of anticipation– the idea that you have to have more than one scale of real time in a system, like two notions of measure in a family of subsets, you come to the idea that, uh, these various times are tied to modes of system decomposition and these lead fairly directly into the wider notion of complexity.
JR: Now, all anticipatory systems are living systems, is this true?
RR: Well, all that we know about. One of the origins of these ideas I think it might be good to talk about… uh, when I was a student, there was a vaguely defined area, which impinged on my own areas of interest, it was called “bionics”, which was the use of biological prototypes to design technology, OK? In fact, one of the approaches that I had to anticipatory systems was just in this bionic line, to use them because we need; I think, as part of our technology, modes of anticipation. We have to look ahead. Say, is the planet getting warmer or is it not? A lot of the problems we face are problems, which are only amenable to anticipatory control. And, in fact, that's what people are doing. What I claimed was, years ago, that what people fought about were models. They fought about what the models told them about the future and what they ought to do about it. They fought over what strategies we should use to try to improve the circumstances in the future. And that required going to more biological modes of control, anticipatory control, more complex system control, uh, than we were used to doing. This was part of what was called “bionics”, was to try to use particularly the behavior of organisms to give us guides as to the kind of technologies we needed to invoke to solve the problems which technologies themselves, among other things, were creating for us. In fact, the first chapter, if I remember, of my book on anticipatory systems was sort of the background of my formal work in this area. But it was meant as, at least in part, an example of bionics: the use of biological prototypes to mold our own technologies. And as such it revolved not so much around science as engineering, and the kind of engineering that it involved had to do with what appears in biology as FUNCTION. What a specific organ or piece of an organism does is carry out a function in the larger behavior of the system as a whole.
JR: Well, isn't the notion of function considered to be non-objective? It's one of the things that reductionists don't want to deal with.
RR: Very much so. But as I said, if you want to try to control a behavior, what you're trying to control is a function, or how a function is being expressed. One of the habits of thought that was being imported from physics was that the idea of function was not a scientific notion. If you wanted to make biology scientific, you had to extrude all notions of function. And in fact, you see, in many books on the philosophy of biology, there were attempts to just cast out the notion of function. The worst sin was to use a phrase like; “The function of the heart is to pump blood.” There were the most elaborate circumlocutions made to get away from using an expression of that form. It was felt to have no scientific content. Now, a pioneer in the use of the term of function in the biological sense was my teacher, Rashevsky. And, in fact, he separated biology into two aspects; one of which was the structural aspect, which he called “metric”, which had to do with more or less the physics of an organism– the sort of things which he, as a physicist himself, was used to dealing with. But then there was what he called the “functional aspect” or the “relational aspect”, which he felt was equally important. In fact, MORE important. But this was one of the reasons that Rashevsky was derided, was because he insisted on this notion of biological function and the way of incorporating it into science, relating it to structure.
JR: How was Anticipatory Systems ,the book, greeted when it first came out? The school of ideas about the anticipatory nature of complex systems…
RR: Well, there were many people who, should I say, the ideas of functionality and ideas of anticipation raised hackles because, again, it was easy for them to feel it was a step away from science as they understood it. They felt that science was a simplifying– dealing only with simple systems. And any attempt to complexity or any attempt to invoke a larger context- several larger contexts, in fact- was a step away from the science that they wanted, or they were seeking. Which was basically the science of simple systems.
JR: Can you remember some of the reactions that you encountered?
RR: Well, you also have to remember that as I got better at defending myself and putting forward my point of view, fewer and fewer people would talk to me.
JR: Directly, you mean?
RR: Yeah. Well, I became really too good at defending my point of view and people wouldn't take me on, it's as simple as that. They could spend a whole lifetime without confronting the ideas that I was putting forth and that was their choice. If they wanted to go ahead without talking about anticipation or they want to exclude certain inherent biological modes of behavior, that's their privilege. So it was both to their advantage and simplified their life to stay away from me. So I didn't get the full fruits… Every now and then someone would take me on in that sense, but..
JR: At what point were you accused of “trying to find answers to questions that nobody wanted asked”?
RR: Oh, that was very early in the game. Yeah. At that time, people were still talking to me, when they told me that my basic problem was that I insisted on tying to answer questions that nobody wanted to ask.
JR: Well, this book on complexity; what got you to decide that you were going to write a book about the whole idea of complexity?
RR: Well, I really think that it's a fundamentally new paradigm. Now, I don't usually regard myself as basically a critic. I just follow what look to me to be intuitively fruitful paths; open things up. But the notion of complexity is so fundamental and carries with it a fundamental change of so many deep habits that I think it's well worth trying to pursue it in detail.
JR: Also, the word “complexity” is being used differently by a lot of other people than the way you mean it, isn't that true?
RR: Yes. What they perceive as complexity, I perceive as… or I call “complication”. In other words, what they call complex is still simple in my terminology, it's just harder for us to deal with because of a rather superficial quantitative aspect.
JR: So, in other words, something could be very… a system could be very complicated and yet still be computable…
RR: Still simple, yeah. Suppose you look at the asteroid belt. You've got thousands, maybe millions of asteroids in a band between Mars and Jupiter. Well, what's the gravitational field of the asteroid belt? You can't compute it, but the reason you can't compute it is because of this quantitative aspect- there are many asteroids. If there were only a few asteroids, you could compute it, just as a technical problem if you should want to do so. You could compute, for instance, the gravitation field of the nine major planets because there are only nine planets. But if there are a few hundred thousand asteroids, the analogous problem becomes, as a technical matter, impossible to deal with. OK?
JR: And yet it's still a simple system.
RR: A perfectly simple system. In fact, that's why you could compute it if you wanted to. But it's not a complex system. You can't do the same thing for the particles, which comprise an organism, and solve anything. That is not doing biology.
JR: Because what makes an organism an organism is not because of the particles…
RR: Well, it's not because the particles are simple. It's not because they're just gravitational, OK? The particles have… I think there are a lot more properties in just matter, the things that physicists study, which show up in organic behavior but do not seem to show up at all in the traditional ways that physicists look at particles, and that's why its not so instructive to take an organism apart into its particles. That's the problem I had to deal with for years– sort of the tenets of naive reductionism… that all there is to know about an organism is the particles you get by decomposing it in various canonical ways and pulling particles out of it. Now everybody knows this is not really quite enough and people express it in various sort of naive ways, like “the whole is greater than the sum of its parts”.
JR: But that's true, isn't it? In a living system?
RR: It is true. It is a very simple and still a good way to characterize the difference between simplicity and complexity. The point is the two can co-exist but the problem is more: What makes certain subsystems behave like, you know, discrete parts, and behave as if they were simple? That's more of the mystery than to try to talk about emergence of new properties in the, sort of… canonical parts. I think the parts were always more complex than people think they are and the circumstances which reveal these anomalies, these complexities, are the interesting ones, these are the ones biology deals with. You have to understand that I consider biology by far the most interesting of the sciences, the most mysterious of the sciences.
JR: Well, your goal all along has been to understand why living things are alive.
RR: Yeah. What makes something live? That is really the great mystery in all of science, in all of nature, and in all of thought.
JR: So can you sum that up by saying that the reason they are alive is because they are complex enough to be alive?
RR: I feel that complexity is almost another way of saying these systems are alive.
JR: They achieve a certain level of complexity and life is an emergent property of that complexity?
RR: Yeah. I would… That's a fair statement.
JR: In your book, Life Itself, you talked about synthesis, but you stopped short of describing in detail what you meant. Is there a reason why?
RR: Well I worry about that. It has an unpleasant technological side to it… And many people over the last certainly 30 or 40 years have worried about biotechnologies. I don't think that many of the things they worried about– like genetic engineering, broadly– were really radical things, but they were real enough to panic a lot of people. I don't know if you remember.. Well, you wouldn't remember, but… 30 years ago, 40 years ago, there were panics about genetic engineering, or biotechnologies, the invocation of, uh, well, machine-like principles… if you want to change the nature of species. Actually this panic goes back even further, to H.G. Wells book, The Island of Doctor Moreau. The monkeying of.. they blame the biologists for the tinkering or tampering with biology, with the biological basis of what it means to be human. It provoked very strong reactions.
JR: And so what is your concern with publishing your ideas, your theories on synthesis of living systems?
RR: Well, I have a fundamental lack of trust in my species. Now, as long as biologists are approaching organisms from the point of simplicity, they can do relatively little harm. None of these concerns arise. But I think it starts to cut close to the bone if you begin to approach them from ideas of complexity. In a certain sense, I feel that as long as, you know, biology stays the way it is: “The main concern of biology is to try to turn organisms into machines” (which they aren't), uh, they can't do any harm. Ifs just not possible for them to do damage any more than it was for, say, Luther Burbank to create a moral or ethical conundrum in biology by hybridizing flowers. He was very limited, so limited, in the techniques he was using that he couldn't do any harm.
JR: But you feel that if somebody were to embrace the theories that you've developed about living systems and then use those to synthesize…
RR: Yeah, I really feel that's dangerous. That's far more dangerous in the long run than, for instance, the development of atomic fission, even though it could have destroyed the planet. It was a dangerous technology for us to get into, especially with the lack of sophistication we had around 1950… But if you did the same sort of thing in biology; use the narrowness of the models that were available in 1950, and have the capacity to really fundamentally change the nature and characteristics of organisms… uh, I feel that is… I wouldn't want to take the responsibility for that. Put it that way.
JR: You think there is a lack of wisdom inherent in our…
JR:… species and our society?
RR: A massive lack.
JR: So, you'll never publish the ideas that you've developed on…
RR: Probably not. I may leave them somewhere… You know, like Leonardo da Vinci, write it in code backward in the mirror… in some kind of cipher…
JR: So, to get into this area of science, would you make any statement about ethics required or any kind of warning to people that are interested in following these ideas that they should also consider the ramifications and responsibilities inherent in doing so?
RR: Well, I'm a sort of Millsian democrat. Just because I don't want to assume the responsibility for it doesn't mean, you know, that I think it's necessarily completely evil, or shouldn't be pursued by anybody. It s up to each individual to decide these matters for himself- or herself. But the point is that it does take a conscious decision to assume the responsibility. And even if I'm not willing to do it myself, I'm not willing to say that no one should ever… But its true that I have very little faith in my colleagues. And the way it stands right now, although the receptivity to some of my approaches is increasing enormously during my lifetime, there aren't many that can do it. I don't worry about my colleagues. I used to joke that, uh, most of my colleagues (there are a few exceptions, but…) most of my colleagues are only getting about 15% of what I was trying to say in my papers. Different colleagues are getting a different 15% but there's no one who could… You know, if I were to disappear, there's nobody who could appear to do what I do. So I take a very personal responsibility for these things, I mean not a diffuse thing, like, say, the development of atomic energy was. It involves certain habits of thought, certain capabilities that don't come along all the time, and I know very well…
JR: Well, the work has had a lot of ramifications.
RR: Oh, yeah.
JR: A lot of different areas of science are interested in these ideas or in applications of these ideas. You once said, for example, that, uh, systems theorists regard you as a systems theorist. And chemists regard you as a chemist. Mathematicians look at you as a mathematician…
JR: What other ramifications can you see coming along from some of these ideas?
RR: Well, as I said, earlier in the game this idea of bionics; the idea of biological prototypes for our own technology, and the use of technology to give insight into how to approach biological… say, biological systems in general… it shows how interrelated things are. That's one of the reasons I'm trying to think about writing a book about complexity in general, because it does stress the interrelationships between purely symbolic things, like mathematics in itself and language… OK?… and organisms… not only organisms but, oh, things like climate, and certain things in the physical science, which involve, you know, a lot of interaction, dynamic interaction,… control. See, what I tried to do in my own books was to suggest, in just a line or two, that a particular development that I'd just finished had other applications besides the ones that were discussed in detail. But I didn't want to take the time to explore these ideas myself. I did in some out of the way papers, here and there, mostly for meetings. But I think an area like complexity is a good way to stress the interrelationships across entirely different fields of application- different fields of science, different fields of engineering, and so on.
JR: Any thought on when you might get that book finished?
RR: Well… I can't predict. I can't “anticipate”.
JR: Well, I think that's a good place to end. I hope this is of some use to Belgian Television and we will see you in August, Belgium.
RR: Well, we'll see. I hope it's what they wanted.
Copyright J. Rosen 1997