On beyond an algorithmic universe
Very rich conversations this week with Stuart Kauffman, a theoretical biologist, Alan Baker, a philosopher, and Scott Gilbert, a developmental biologist, first over dinner and then during a panel discussion with additional input from Mark Kuperberg, an economist, and Billie Grassie, founder of the Metanexus Institute on Religion and Science. Delighted if any of them wanted to weigh in with their own thoughts in the on-line forum below (along with anyone else interested in the subjects discussed).
Kauffman and I, unbeknownst to either of us, have been exploring related but distinct paths to a similar place, a conception of the universe which is consistent with but not constrained by physical law, a universe whose evolution is non-algorithmic and so open to the appearance of phenomena that could not in principle be predicted from a complete description of physical laws and starting conditions, a universe within which humans are causal agents in that sense, able to contribute to the appearance of things that change the nature of the universe (cf Kauffman's Beyond Reductionism and my From Complexity to Emergence and Beyond (.doc file)).
The conversations were intense, often technical, and in part reviewed diverse materials several of us had been over in the past (cf Emergent Systems working group), but the gist of them primarily concerned a relatively simple and straightforward issue: if we knew everything there was to know about the material universe (physical, biological, and human) at any given point in time, could we then predict all future events? Are "unpredicted outcomes" and "unconceived alternatives" simply reflections of human ignorance or do they reflect openings to potentials inherent in the material world but in principle unknowable until they are realized? Are humans (and perhaps other organisms) capable of recognizing and making use of those potentials?
No one, I think, went away persuaded that anyone else had a definitive answer to those questions, but I certainly thought there was significant progress in sharpening them and clarifying their connections to a host of related issues in physics, in biology, and in human culture. And I heard myself saying some things that may turn out to open a new line of approach to the question, at least for me.
Discussions of this kind, about the nature of the material universe, almost invariably start with physics, and presume that it is in the laws of physics that an answer to the question of whether we live in an algorithmic universe or not must be found. Kauffman has, I think appropriately, suggested that there are things going on in the biological world, in evolution and in development, that might provide a better perspective from which to answer the question. There are things happening in both developmental and evolutionary processes that would seem not, even in principle, to have been predictable in advance. My inclination is to go one step further, to suggest that the brain might provide a still better perspective to explore the issue of what kind of universe we live in. It is not only that the brain seems to do things that could not be predicted in advance but that the brain has the capacity to decide whether or not to do things that could not be predicted in advance, and so can play a significant role in deciding whether we and others things live in an algorithmic universe or not.
The starting point for this line of approach is a recognition that the universe we live in is, in a very important way, a construction of our brains. Our brains receive inputs (from our genes, from making observations, from listening to other people), create explanations ("stories") to account for those inputs, and test and revise those stories based on additional inputs. The stories are always subject to revision and, at least as importantly, are also always underdetermined. There are an infinite numbers of ways to account for any finite set of inputs, and so there are always unconceived alternatives, ways to account for the observations that haven't occurred to us (maybe masses move toward each other not because of a force between them called "gravity" but rather because masses produce curvatures of space that influence their future trajectories). The upshot is that the universe we live in is only one of an infinite number of possible universes, one we have brought into being by a choice (conscious or unconsconscious) among possibilities. It is not, of course, a fully unconstrained choice. We have to account for the observations, but there remain an infinite number of possibilities from which a choice is made. And those choices in turn alter the infinite array of possibilities from which future choices are made (let's stop worrying about explaining forces and try and understand curvatures of space instead). In short, if we take seriously what we currently know of our brains, we live not in "the universe" but in a constructed universe that is continually being reshaped by, among other things, our undetermined choices of how to conceive it.
It is, in this sense, not only the universe we live in but "the universe" itself that is a construction. Most people understand by "the universe" something that exists independently of our actions and thoughts, something that we don't fully understand but is there and gives rise to unexpected things, including unpredicted consequences of our own thoughts and actions. And most people presume that if we are smart and attentive and persistant, we will eventually work out the laws of that thing out there and there will be no longer unexpected things or unpredicted consequences or unconceived alternatives. And that's certainly a legitimate construction based on observations (scientific and otherwise). But its not the only one. There are an infinite number of other ways of thinking of "the universe" that also fit the observations. Among them is the idea that "the universe" is itself not only out there but also in here, and that it too is underdetermined, evolving, even in our absence, in somewhat unpredictable ways. The "algorithmic" universe is a human construction and, like all human constructions, should be understood not as a necessary baseline but rather as one possibility among many. An "indeterminate" universe, one with rules but also elements of randomness, fits the observations as well as an "algorithmic" one does (cf Alternate perspectives on randomness and its significance).
What follows from all this is itself an instance of underdetermination. And so we can make a choice, unconsciously or consciously, between conceiving the universe as algorithmic or indeterminate (alternatively, we can create still another way of thinking about it). There are real advantages to thinking of the universe as algorithmic, amongst which is that it gives us a clear task ("reduce uncertainty"), a way to measure success ("get better at predicting"), and a vision of a promised land in which uncertainty has been done away with and we can rest contentedly having achieved our goal. On the other hand, an indeterminate universe offers us the perhaps no less appealing alternative prospect of eternal participation in a work in progress, one in which there will always be surprise and uncertainty, the latter giving us the room to be not only discoverers but also creators (cf Evolution/science: inverting the relationship between randomness and meaning).
science itself will teach man... that he himself is something of the nature of a piano-key or the stop of an organ... so that everything he does is not done by his willing it, but is done of itself, by the laws of nature....even if this were proved to him by natural science and mathematics, even then he would not become reasonable, but would purposely do something perverse out of simple ingratitude, simply to gain his point.... the whole work of man really seems to consist in nothing but proving to himself every minute that he is a man and not a piano-key! .... Fyodor Dotoevsky, Notes From Underground, 1864
my first act of free will shall be to believe in free will ... William James, Letters, 1870
For me, as I suspect for Dostoevsky and James, the choice between algorithmic and indeterminate universes boils down to a personal and pragmatic one. In an algorithmic universe, all that will be is inherent in what is and has been. In an indeterminate universe, there is not only the possibility of alternative futures but a role for me and other people in expanding the array of possible futures by our own thoughts and actions. I like the idea that people might matter in that way and like still more the idea that the task of humans (myself included) is not to figure out what we are ignorant of but instead to create what has yet to be tried out by an evolving universe of which we are a part.
An indeterminate universe can be thought of as one exploring the space of possible forms of existence, a space which was expanded by the appearance of living things (as Kauffman argues), and further expanded by the evolution of the brain, with its ability to conceive things other than what exists at any given time ("counterfactuals"). While the construction of the algorithmic universe was quite effective at generating new possibilities for an extended period I think it is, at this point, becoming a constraint. When our constructions constrain our abilities to conceive new possibilities more than they expand them, the time has come to move on to new constructions. The indeterminate universe, with its emphasis on the generation of new constructions, seems to me to be an appropriate replacement. Whatever the potentials of the "physical" universe, the one that existed prior to our existence, I'd like to think we have increased them.
Many thanks to my colleagues in the Evolving Systems project for conversations that contributed importantly to this essay. For some practical implications of these ideas in academic and educational contexts see On beyond a critical stance and Evolution of science education as story telling and story revising. Reactions and/for additional thoughts are welcome in the on-line forum below.
- Universal grammar or similar problem solving processes? (27 Feb 2010)