I just read Cosma Shalizi's Must Macroeconomic Theories Have Microfoundations? Like most good scientists, he is obviously partial to reductionism: the idea that you understand a complex thing in terms of its constituent parts. Big things are made of little things, and those little things are simpler and more basic than the big. So he says: "Obviously, macroeconomic phenomena are the aggregated (or, if you like, the emergent) consequences of microeconomic interactions. What else could they be? Analogously, the macroscopic physical properties of condensed matter all ultimately emerge from molecular interactions."
Up until a couple of years ago I would have agreed with Cosma, but now I don't. I am not sure I can articulate why, but let's have a go at it anyway.
Let's get a couple of things out of the way. First, I've not gone all mystical. I'm as strict a materialist as anyone. I think I am anyway. But materialism is not the same as reductionism.
And second, I have no problem with reductionism in the physical sciences. I'll happily put my hand on my heart and recite after Dirac that, with the discovery of quantum mechanics "the underlying physical laws necessary for the mathematical theory of … the whole of chemistry are thus completely known, and the difficulty is only that the exact application of these laws leads to equations much too complicated to be soluble.”
But when it comes to social and even biological sciences things are different, and they are different because of evolution and because of people. So I can't agree with psychologists who think the key to understanding human behaviour lies in the neurochemistry of our brains – because where else could it be? – or with reductionist economists believe that "in the end" the behaviour of society can be explained in terms of the behaviour of the human individuals that comprise it, because what else is society, fundamentally, but a collection of individuals?
Are evolutionary explanations reductionist? Cosma says yes, but Jon Elster (in Explaining Social Behavior), says no. They provide explanations, but whereas reductionist explanations are also causal ("we explain an event by citing its cause. Causes precede their effects in time." [p271]), evolutionary explanations work in terms of consequences. Sure, mutation happens at the microscopic level, but selection is determined by how the mutated organism fits into the bigger world.
How to explain the stripes of a zebra? They help their bearer to survive by making it difficult for a predator to pick him or her out from a herd of similarly striped animals. As Richard Dawkins explains beautifully in The Extended Phenotype, we can explain the dam of a beaver as a genetic adaptation to its surroundings. So if a reductionist would say "Obviously, animal behaviour (macroscopic) is the consequence of its genetic and cellular makeup (microscopic). What else could it be?" you could say "it's the consequence of its environment too." No forests and lakes, no beaver dams.
Now forests are composed of trees are composed of cells are composed of molecules, so there is a microscopic aspect to the environment too, of course. That's materialism. But which came first, the dam-building genetic mutation or the logs to build with? To privilege one over the other is to miss the whole melody of the explanation. The closer we look, the more we want to know about the genes, but the more we want to know about the environment too: its history, its fluctuations, its other inhabitants. Evolutionary explanations lead outwards as well as inwards, and to me that makes them not reductionist.
There's more to social science explanations than evolution.
Many non-scientists have a reaction against reductionist explanations of human behaviour, as if they somehow diminish us, reducing us to nothing more than a pile of cells. As an ex-scientist myself I have had little sympathy with this attitude, but recently I've found myself thinking that there is something to it. Not the diminishment part–there's nothing wrong with being a pile of cells–but that it does miss an equally essential part of the puzzle. It may be old hat to you, but for me it's been novel.
Earlier this year, photons from a star that exploded when the universe was young finally reached NASA's orbiting "Swift" telescope after an uninterrupted journey of 13.1 billion years, according to the New Scientist. The exploding star was the most distant and hence most ancient object that humans have so far detected.
When the photons collided with the telescope, Swift's signal processing equipment transmitted a record of the event to earth, and "within an hour, astronomers began training ground-based telescopes on the same patch of sky to study the burst's infrared afterglow". Soon after, Neil Gehrels of NASA's Goddard Space Flight Center was giving a press conference, and a few days later Rachel Courtland of the New Scientist spent her day writing up an account of the event, dubbed GRB 090423 (Gamma Ray Burst on April 23, 2009).
To say this another way, an explosion that took place 13 billion years ago produced, through a cascade of intermediaries, a change in the movement of certain particles belonging to the 21st century bodies and clothes of astronomers and journalists. To understand, even in principle, the motion of Mr. Gehrel's particles in April 2009 we must take account of photons from an explosion that took place near the beginning of time.
But that's not all.
It was only because 21st century astronomers and cosmologists understand the laws of physics that the photon impact was understood to be important and the press conference was held. If 21st-century mathematics and physics (the subjects themselves) did not "exist" then the motion of Mr. Gehrel's particles would have been different. 21st century mathematics is an abstract thing: it is not made of particles. If the photons had come from a nearer source the results would have been quite different because the implications of those particles' frequencies at that time and angle would not have been interesting enough for a press conference.
The same goes for the Ms. Courtland's particles. Without the language we use, without a commercial media, an audience for the story, and countless other institutions of society, the neurons in Ms. Courtland's brain would have been firing in quite a different set of patterns on April 27. The motion of the particles that comprise Ms. Courtland and Mr. Gehrel was the result of physics and also the result of culture.
The reductionist program seeks to explain personality in terms of neurons, and neurons in terms of molecules, but here the tables are turned. To explain the motion of these atoms, we need to involve the neurons of the Courtland and Gehrel brains, and to explain the activity of those neurons we need to include language, society, and an ancient event from the beginning of time. And there is nothing to say that one of these sequences of causation is more basic or more fundamental than the other.
Yes, we can say that "in the end" our behaviour is explained by our cells and our cells are explained by our atoms. But that is one perspective among many, and not a privileged one. "In the end", the motions of atoms are determined by the properties of our culture and the "existence" of non-material structures such as mathematics, physics (the body of knowledge), and language.
Let me get a bit more romantic. The theoretical physicists, in their search for a theory of everything, claim to be making progress. But every step they take moves us further and further from any notion of a deterministic universe. How? Again, nothing profound or mysterious. If we demand of a deterministic worldview a weak notion of convergence – that the closer we get to the theory, the closer we get to an understanding, even "in principle", of the problems it purports to explain. Maybe fundamental physics fails this test.
How much of the universe do we need to include in our description to "explain" a particular event? Hundreds of years ago one could have said that if Ms. Courtland's ancestor had spent her day one way rather than another, we would only have to look in the circumstances of her immediate environment to understand why. But as we humans have discovered more about the universe and our place in it, so our understanding couples us to events that are more and more distant in space and in time, and we need to include more of the universe in any explanation of events. To explain Galileo's actions we needed to include the motion of the solar system. To explain Edmund Halley's actions we need to understand the motion of a distant comet. And so on. And the more of physics we understand, the more of the universe we need to explain the movements of our particles. There are events that took place even earlier than GRB 090423 that have not yet affected the motion of particles on earth, but which will do in the future as we develop ways to detect them.
There is no convergence here, and any argument that "in the end" we are just made up of nuclei and electrons has nothing more fundamental to it that the reverse claim, that the motions of nuclei and electrons are "in the end" determined by the remainder of the universe, by the properties of abstract cultural constructs such as language and media.
You can make the same reversal for claims that society must be understood in terms of individuals (or macro-economics in terms of micro). There's probably nothing new here for many in the social sciences and humanities, but it's new for me.
It is tempting to think that society is composed of individuals because we feel ourselves to be such an individual, with a separate and independent existence. We have stood, each of us, outside on a frost-covered night and looked up at the universe and felt the sense of insignificance in the face of that emptiness. But along with that rush of insight we feel an affirmation of our existence. However small we are, however indifferent the universe may be, we exist. Even if we agree with Samuel Beckett that "they give birth astride of a grave, the light gleams an instant, then it's night once more", at least it seems, at such moments, that our light does gleam for that instant.
A reductionist worldview gives credence to the idea of an isolated, self-contained individual because it sees individuals as the sum of their smaller parts. But this sense of individual, isolated, self-contained wholeness is illusion.
What is a person without language? If we had no language we would not be "individuals" in the same sense we feel on those nights. A two-year old who stares at the universe feels no awe because they have little sense of self and little understanding of what they are seeing, and our own individuality, our own consciousness, only emerges from that understanding and the culture that produces it. Daniel Dennett quotes Helen Keller on her life before the learned to communicate: "Before my teacher came to me, I did not know that I am. I lived in a world that was a no-world. I cannot hope to describe adequately that unconscious, yet conscious time of nothingness… Since I had no power of thought, I did not compare one mental state with another". If consciousness is, as Dennett claims, the "centre of narrative gravity" – the insubstantial eye in the middle of our storm of experiences and stories, then without language – without culture – there can be no such thing as an individual person. Society may be made of individuals, but individuals are equally made of society.
Adopting an evolutionary worldview, as opposed to a reductionist one, makes us as individuals simultaneously more and less significant. More significant because our cultural constructs become shapers of the universe no less than the laws of physics that we seek to uncover; less significant because our very individuality is an illusion, contingent on the culture from which we spring.
Cosma Shalizi has updated his post with a response to this one (here), and some interesting links. Unfortunately today has been too busy for me to re-respond.
So, would you agree with the statement: “If we can’t predict tomorrow’s weather, how can we predict climate change?”
Perhaps I simply don’t understand what you’re trying to say, and I’m sure I’m a little lost on your opinion on the evolutionary worldview (My guess: equally bad). However, the idea that a small complex system we can’t understand combines to form a large complex system we can’t understand… well.. isn’t that a little reductionist too?
On physics, the most promising theories of everything turn out to be emphatically deterministic and physical (and in only four dimensions). They are very much still on the fringe of a Kuhnian transition because of the enormous funding overhead they push against. A strong assertion on my part as I am not a physicist, but I have gone through the math and it seems compelling and extraordinarily exciting.
On complex non deterministic systems (humans, e.g.) arising from the primordial soup, I think the very best ideas I am familiar with are from Doug Hofstadter of “Gödel, Escher and Bach” fame in a recent book “I Am A Strange Loop.” His discussion of a concept called shared symbol networks seems to explain much about social systems to me.
Looks like Doug has a new book due March this year…”The Essence of Thought.” I expect it to be worth a read.
As a committed humanist, I’m pleased to read this post – especially the suggestion of language as fundamental to social being. I generally point to Gadamer’s Truth and Method for a somewhat round-about but nonetheless deeply compelling account of the social essence of language as the basis for individual being. But for a much more accessible and immediately enjoyable ride, check out this Radiolab episode on words, which includes (among other provocative forays into language) an account of a man who did not know that things have names until he was 27 years old – and can now not express what life was like previously:
I find it amusing that Shalizi resorts to petty rhetoric devices rather than sound argument in saying:
“Obviously, macroeconomic phenomena are the aggregated (or, if you like, the emergent) consequences of microeconomic interactions.”
First, there is nothing obvious about anything unless you arrogantly assume that people share your epistemic paradigm. Second, are those macroeconomic phenomena merely aggregated or are they emergent? You can’t have it both ways and if they are indeed emergent, then looking at the parts alone won’t suffice to understand the whole, because according to strong emergence the whole is greater than the sum of its parts.
So while I like that you are turning against reductionism, I do find your argument concerning language less than convincing. Wittgenstein had no clue about more recent advances in cognitive science that render the whole “the limits of my language are the limits of my world” argument quite pointless.
The RadioLab episode is fascinating – I had not heard of that story before. And I will look up the book you mention. Thanks.
I don’t think Cosma S. is being arrogant – see the disclaimers at the beginning of his post – but really, his post is just an excuse to trot out some things I have had lying around for a while. I’d be interested in pointers to these “recent advances in cognitive science”.
Shalizi may be partial to reductionism, but much less so than has been dominant among macroeconomics since, say, 1980. The Lucas critique/New Classical Macro really put that approach into high gear, and, as Shalizi put it, “pull[ed] down the theory.” His points 1 & 2 may have given away more than you are willing to concede, but I think that’s because he was more interested in points 3 through 7. So I think you are misreading him in calling him partial to reductionism, and your response to Jakob (currently the most recent comment) suggests that he is more straw man for you than anything else.
Urgh … trying to take this all in, I think you’re struggling with the idea that in politics, certain frameworks (individualism) are very bad ways of thinking about society, and putting that into quasi-scientific words, then concluding that since the politics is bad, science must be bad too. Because if the science is good, then the politics must be good, which is bad.
I hope the problems here are clear. Because that way lies madness, or at least post-modernism.
There’s an echo here of evolution vs. Social Darwinism (natural selection vs the rich are more fit than the poor), and hence one of the nominally religious objections to evolution being that it encourages Social Darwin.
Also, you seem somewhere to have assumed that “reductionism” denies feedback and interaction, which is an understandable mistake. There can be systems where e.g. atoms change depend on interactions with other atoms – that’s the atomic bomb.
Tom: I think that there is an argument against reductionism grounded in game theory. The rule, not the exception, in game theory is multiple equilibria. So given tastes and technology don’t determine the outcome. If you then think of social factors as equilibrium selection devices – things like norms, eg – then it seems to me reductionism doesn’t go through. Schelling says somewhere that solving a coordination game involves jointly creating a tradition: we need the We before we can determine what all the I’s will do!
Kevin – I had not thought about multiple equilibria in game theory that way. There is an essay by Roger Myerson along the lines you describe, about Schelling’s focal point: I wonder if that’s what you are thinking of? It’s here: http://home.uchicago.edu/~rmyerson/research/justice.pdf.
Seth – I’m trying to avoid the individualist politics/individualist science slippery slope. Yes, I have a bit more appreciation for some sociologists than before, but I think this is a purely scientific issue.
Let me try to clarify some things.
– A naive reductionism would say that we explain things in terms of their constituents, and solid state physics falls into that realm.
– It’s still reductionist to explain open systems in terms of their internal workings + their interaction with their environment (Seth and Sunny, I think). Usually this can be done through some idea of a flux through the boundary between the system and the environment.
– Speech, written communication, bodies of knowledge, and technology make the boundary between humans and their environment very difficult to define, to the point where it is not clear that there is something that is a human without its environment. Not only are we influenced by our surroundings, but via culture we are influenced by most of the known universe. This is not novel, of course. But it does make explanations of human behaviour (and the behaviour of our muscles, neurons, cells) non-reductionist. I don’t know if it matters to anyone else, but it does seem to matter to me.
Yes, he is more straw man than anything, but I’m sure he’ll survive that. And he does write this on reductionism: http://cscs.umich.edu/~crshalizi/notebooks/reductionism.html
Ya gotta be careful about putting periods right next to URLs, w/o any intervening spaces. It makes them hard to click on, w/o getting an error message (url not found, because the period is interpreted as part of the link).
ummm … how about statistical mechanics? We can make predictions about the behaviour of the population and its response to forces, but within the population itself individual particles may be doing all sorts of things.
And how come the countries with the best most advanced economies are those where the government has intervened to create a protected secure environment for industry with lots of state-sponsored activity? each company under the national umbrella is different and has its own niche, but together they form a distinctive collective.
And of course as always the truth is in the birds. Take ducks. They fly, the males are brightly coloured and the females are dull (mostly, but not shelducks). Except on islands with no ground predators when they don’t fly and the males are dull. So if a new species of duck is discovered where there are ground predators, I will confidently predict that the female will be brown, and the male brightly coloured, but I will have no idea what the male will look like.
Being a philosopher, all I can say, is: Of course. Isn’t it obvious? 🙂
One reason why macro is essentially different from micro is this: Micro can be modeled in terms of markets populated by individuals, and it is possible to conceive of a very simple model with a single “representative agent.” I don’t think that any economist actually thinks that this model is anything but a simple heuristic, and it allows us to think about the principles of markets in a simple way, boiling it down, so to speak.
However, macro does not deal with individuals and markets in the same way. The basic macro equation is Y=C+I+G+(X-M), that is, the sectoral relation linking households, firms, government, and the ROW. The operatve word here is “relation.” A society is not an aggregate of individuals but a system of elements (individuals), subgroups (families), groups (institutions), which form subsystems embedded in a larger system (nation).
“It’s the relationships, stupid.”
According to the sectoral balance approach, the sectoral balances sum to zero. Therefore, the government sector, the private domestic sector, and the external sector cannot all be in surplus or deficit simultaneously. That is to say, all sectors cannot act in tandem. Their relationship does not support it.
Those who try to reduce macro to micro often find themselves involved in fallacies of composition, like the paradox of thrift, where pursuit of individual self-interest in economic (market-based) situations in a way that is beneficial for the individual turns out to be deleterious for the economy considered from the vantage of the society as a whole. What may be effective and efficient for individuals as such may not be so for the larger group, owing to relationships linking individuals.
In addition, modeling social sciences on the physical or life sciences is a fool’s errand because of uncertainty. Human consciousness is not capable of standing outside of itself other than in very simple thought experiments (conceptual models). Modeling individuals as analogous to atoms or cells simply does not work in complex models owing to uncertainty. Humans are unpredictable as individuals, and the conditions they are presented with are uncertain. Reductionists can dream that they will be proved correct in the future, but it’s not in the cards at present. We can’t even write the equations, much less solve them.
and whilst I’m on about birds, songbirds such as finches when raised in captivity have weak poorly developed songs. In the wild they learn their songs from other birds. So birdsong is really a property of a population, not just an individual bird.