Tuesday, April 2, 2013

Greedy Reductionism, Science and Economics

John E. King’s new book The Microfoundations Delusion: Metaphor and Dogma in the History of Macroeconomics (Cheltenham, 2012) charges neoclassical economics with the delusion that macroeconomic processes must just be explained in terms of microeconomics, and indeed that microeconomics is the proper foundation for all macroeconomic phenomena.

These ideas can be called “strong reductionism” and, curiously enough, the neoclassical delusion is a perfect example of what philosophers call “greedy reductionism.” The philosopher Daniel Dennett usefully calls “strong reductionism” the error of “greedy reductionism”: the belief that higher-level principles and traits of complex systems of all sciences (including social sciences) should be abandoned or simply explained in terms of lower-level terms (Dennett 1995: 80–81).

So what is the influence of this idea in the natural sciences and social sciences?

I. Reductionism in the Natural Sciences
King, citing the work of others, charges certain scientists with the “strong reductionism” fallacy, such as Richard Dawkins.

The objection is described by Lewontin, Rose, and Kamin in the following terms:
“Broadly, reductionists try to explain the properties of complex wholes—molecules, say, or societies—in terms of the units of which those molecules or societies are composed. They would argue, for example, that the properties of a protein molecule could be uniquely determined and predicted in terms of the properties of the electrons, protons, etc., of which its atoms are composed. And they would also argue that the properties of a human society are similarly no more than the sums of the individual behaviors and tendencies of the individual humans of which that society is composed. Societies are ‘aggressive’ because the individuals who compose them are ‘aggressive,’ for instance. In formal language, reductionism is the claim that the compositional units of a whole are ontologically prior to the whole that the units comprise. That is, the units and their properties exist before the whole, and there is a chain of causation that runs from the units to the whole.” (Lewontin, Rose, and Kamin 1984: 5).
But in my view the scientists or philosophers who are sometimes accused of being guilty of crude reductionism do not in fact hold that “strong reductionist” viewpoint.

Dawkins, for example, repudiates the type of “strong reductionism” or “precipice reductionism” that seeks to “explain complicated things directly in terms of the smallest parts, even, in some extreme versions of the myth, as the sum of the parts!” (Dawkins 1991: 13; emphasis in original). Instead, Dawkins endorses a “hierarchical reductionism” which recognises that the type of explanations suitable at a lower level are not the same as at higher levels (Dawkins 1991: 13; see also Dawkins 1986; Weinberg 1994).

Steven Pinker elaborates:
“Reductionism, like cholesterol, comes in good and bad forms. Bad reductionism—also called ‘greedy reductionism’ or ‘destructive reductionism’—consists of trying to explain a phenomenon in terms of its smallest or simplest constituents. Greedy reductionism is not a straw man. I know several scientists who believe (or at least say to granting agencies) that we will make breakthroughs in education, conflict resolution, and other social concerns by studying the biophysics of neural membranes or the molecular structure of the synapse. But greedy reductionism is far from the majority view, and it is easy to show why it is wrong. As the philosopher Hilary Putnam has pointed out, even the simple fact that a square peg won’t fit into a round hole cannot be explained in terms of molecules and atoms but only at a higher level of analysis involving rigidity (regardless of what makes the peg rigid) and geometry. And if anyone really thought that sociology or literature or history could be replaced by biology, why stop there? Biology could in turn be ground up into chemistry, and chemistry into physics, leaving one struggling to explain the causes of World War I in terms of electrons and quarks. Even if World War I consisted of nothing but a very, very large number of quarks in a very, very complicated pattern of motion, no insight is gained by describing it that way.

Good reductionism (also called hierarchical reductionism) consists not of replacing one field of knowledge with another but of connecting or unifying them. The building blocks used by one field are put under a microscope by another.” (Pinker 2003: 70).
But the point is that roughly hierarchical processes at each higher level might be understood at different levels of analysis and the lowest one is not necessarily of any use outside its restricted domain, and the higher-level analysis may well be the most important one.

In general, I would argue that, when properly limited to its correct place, “hierarchical reductionism” is no threat to science, the social sciences, and certainly not economics.

In a nice discussion in Debunking Economics (2011), Steve Keen notes how “strong reductionists” in the natural sciences thought that all the sciences could ultimately be reduced to physics (Keen 2011: 206), but this project has failed.

As Keen says,
“Strong reductionism implies that the behaviour of any complex system can be entirely understood by considering the behavior of its constituents, and then summing their effects: ‘the whole is the sum of its parts.’” (Keen 2011: 206).
But when complex systems have multiple nonlinear relations between variables the reality is that properties at the aggregate level can emerge which cannot be found at the lower level of individual components (Keen 2011: 207). The aggregate level phenomena are called “emergent properties” (Keen 2011: 207). Each higher-level science is not just an applied version of the science below it. The rough hierarchy of sciences exists, but
“‘at each higher stage entirely new laws, concepts and generalizations are necessary, requiring inspiration and creativity to just as great a degree as in the previous one. Psychology is not applied biology, nor is biology applied chemistry’” (Keen 2011: 208).
II. Strong Reductionism in Economics
In economics, the same “strong reductionist” method manifests itself in the desire to directly derive the properties of the macroeconomy from microeconomics (Keen 2011: 206). Under the neoclassical view, the whole economy or macroeconomy is just the sum of individual behaviour.

But this is just another instance of the “strong reductionist” fallacy. Macroeconomics is not applied microeconomics.

A case in point:
“despite its adherence to strong reductionism, neoclassical economics provides one of the best examples of emergent phenomena ever: the ‘Sonnenschein-Mantel-Debreu conditions’ … . This research proved that a market demand curve derived from the preferences of individual consumers who in isolation obeyed the Law of Demand – i.e., they had ‘downward-sloping demand curves’ – will not obey the Law of Demand: a market demand curve can have any shape at all.” (Keen 2011: 208).
While I will leave the question of how Austrian economics commits the “strong reductionist” fallacy for another post, one can certainly see it in the patron saint of modern Austrian economics, Ludwig von Mises:
“Society is concerted action, cooperation. Society is the outcome of conscious and purposeful behavior. … The total complex of the mutual relations created by such concerted actions is called society. … But society is nothing but the combination of individuals for cooperative effort. It exists nowhere else than in the actions of individual men. It is a delusion to search for it outside the actions of individuals. To speak of a society’s autonomous and independent existence, of its life, its soul, and its actions is a metaphor which can easily lead to crass errors.” (Mises 1998: 143).
In other words, Mises is just a crude “strong reductionist,” and for him society is just the “sum of its parts”: the “total complex of the mutual relations created by ... concerted actions is called society.” He has no understanding of emergent properties or downward causation.

Curiously, some few Austrians might have grasped the unsoundness of “strong reductionism,” despite (sometimes) protestations to the contrary. Hayek’s “spontaneous order” captures some elements of why greedy reductionism is wrong and how “emergent properties” or “emergent order” can occur in human society.

III. Conclusion
I would conclude that moderate reductionism or “hierarchical reductionism” is not a threat to good social science, nor to heterodox economics. It is “strong reductionism” that is the fallacy and bad science.

Certain scientists accused of “bad reductionism” (as far as I can see) are probably not guilty of it.

And finally the neoclassical attempt to derive macro from microfoundations is bad science.


Dawkins, Richard. 1986. “Sociobiology: The New Storm in a Teacup,” in S. Rose and L. Appignanesi (eds.), Science and Beyond. Blackwell, Oxford. 61–78.

Dawkins, Richard. 1991. The Blind Watchmaker. Penguin Books, London.

Dennett, Daniel C. 1995. Darwin’s Dangerous Idea: Evolution and the Meanings of Life. Penguin Books, London.

Keen, S. 2011. Debunking Economics: The Naked Emperor of the Social Sciences (rev edn.). Zed Books, New York and London.

King, J. E. 2012. The Microfoundations Delusion: Metaphor and Dogma in the History of Macroeconomics. Edward Elgar, Cheltenham.

Lewontin, Richard C., Rose, Steven and Leon J. Kamin. 1984. Not in Our Genes: Biology, Ideology, and Human Nature. Pantheon Books, New York.

Mises, L. 1998. Human Action: A Treatise on Economics. The Scholar's Edition. Mises Institute, Auburn, Ala.

Pinker, Steven 2003. The Blank Slate: The Modern Denial of Human Nature. Penguin Books, London.

Weinberg, Steven. 1994. Dreams of a Final Theory. Vintage Books, New York.


  1. Hierarchical reductionism is still reductionism. The Wiki page summarises it:


    "In his book The Blind Watchmaker, Dawkins introduced the term "hierarchical reductionism" to describe the view that complex systems can be described with a hierarchy of organizations, each of which is only described in terms of objects one level down in the hierarchy."

    In economics -- and probably in other sciences -- certain phenomena cannot be explained "in terms of objects one level down in the hierarchy". The paradox of thrift, for example, can only be explained at a truly macro-level -- it cannot be explained with reference to the "objects one level down in the hierarchy" which in this case would be individual savers. The same is that case of the Keynesian/Kaleckian theory of profits, the Keynesian theory of speculation (the beauty contest) and many other phenomena.

    Dawkins' arguments, frankly, are driven by his obsession with religion -- which in ublic comes across as vulgar preaching of doctrinaire atheism. His criticisms of religion -- which are shallow -- all rely on this idea that phenomena can be explained with reference to objects further down the so-called "hierarchy". This is what drives Dawkins at a fundamental level. It's quite obvious -- to me, at least.

    1. Yes, I concur that many of Dawkins' arguments "are driven by his obsession with religion -- which in public comes across as vulgar preaching of doctrinaire atheism".

      But I think that Wikipedia article does not accurately describe "hierarchical reductionism".

      Pinker is quite clear:

      "Good reductionism (also called hierarchical reductionism) consists not of replacing one field of knowledge with another but of connecting or unifying them. "

      I think that leaves room for "emergent processes" and all the insights King makes.

      Also, is no form of reductionism at all valid?

      For example, can we really deny that humans can be analysed in terms of biology -- organs, tissues, cells and biochemistry?

      Of course, that level of analysis is often useful but is to be restricted to the specific domain of biology (It does not tell you what you can learn from anthropology, for instance.)

      But connecting, say, psychology with biology in this way does not mean psychology is just derived biology.

      E.g., psychology studies phenomena not predicable from biological principles.

      Finally, when writing this post, I looked again at Dawkins's The Blind Watchmaker.

      There on p. 13 he disavows “strong reductionism”. This type of reductionist, he says,

      “tries to explain complicated things directly in terms of the smallest parts, even, in some extreme versions of the myth, as the sum of the parts!”

      You might complain that he is a hypocrite or inconsistent, but his alleged adherence to crude reductionism is not clear cut.

    2. I might just add that the conscious human mind looks like the ultimate example of an incredible "emergent property", maybe not predicable from individual behaviour of neurons.

      Yet surely you can "connect" -- or in a sense unify, as Pinker says -- psychology with biology by analysing the mind as a thing produced by some complex operation of the brain, namely, neurons, synapses, neurotransmitters and all of its chemistry.

      Is reductionism in this sense unsound even when one admits that psychology is not simply applied biology or applied neuroscience? And that the mind has properties not just the sum of the individual neurons?

    3. http://uath.org/download/literature/Richard.Dawkins.The.Blind.Watchmaker.pdf

      "For any given level of complex organization, satisfying explanations may normally be attained if we peel the hierarchy down one or two layers from our starting layer, but not more. The behaviour of a motor car is explained in terms of cylinders, carburettors and sparking plugs. It is true that each one of these components rests atop a pyramid of explanations at
      lower levels. But if you asked me how a motor car worked you would
      think me somewhat pompous if I answered in terms of Newton's laws
      and the laws of thermodynamics, and downright obscurantist if I answered in terms of fundamental particles. It is doubtless true that at bottom the behaviour of a motor car is to be explained in terms of interactions between fundamental particles... For those that like '-ism' sorts of names, the aptest name for my approach to understanding how things work is probably 'hierarchical reductionism'. If you read trendy intellectual magazines, you may have noticed that 'reductionism' is one of those things, like sin, that is only mentioned by people who are against it. To call oneself a reductionist will sound, in some circles, a bit like admitting to eating babies." (The Blind Watchmaker, pp12-13)

      Two notes:

      (1) Note how Dawkins ties being a reductionist to his ridiculous martyr-complex. This is the connection to his "theories" about religion. Dawkins loves playing the heretic and the victim and takes every opportunity to strike such a posture.

      (2) It is quite clear that Dawkins' "hierarchical reductionism" does indeed say that you can explain things by their parts. He just says that sometimes its useful to aggregate. This is still reductionism. I would say, as would Keynes, that some things are simply NOT reducible to their parts.

    4. "For example, can we really deny that humans can be analysed in terms of biology -- organs, tissues, cells and biochemistry?"

      Yes. I would absolutely deny that. You cannot explain consciousness and its derivatives from this level of analysis.

    5. Perhaps I am wrong in that Dawkins may neglect downwards causation and emergent properties.

      I would still say some moderate form of reductionism is defensible.

      I suppose you cannot explain the conscious human mind solely in terms of the individual behaviour of neurons or individual brain chemicals or an individual firing of a neuron.

      But if the brain seems to cause our mental life, then a moderate reductionism must be true in the sense that there are basic parts --- neurons, synapses, neurotransmitters and all of its chemistry -- underlying the mind, even if the mind itself is produced by very complex emergent properties and relations of these parts, not predicable from individual behaviour of neurons. E.g., the mind is not just the sum of the individual neurons, but how can one deny that parts do exist? (e.g., Broca's region is connected with language processing).

      The trouble is science does not fully understand the process at present, but surely there is an explanation.

      E.g., we can understand macroeconomic phenomena
      but the problem of consciousness is much more difficult.

    6. I don't think that it has to do with lack of knowledge. Its the logical sequence that is misunderstood by the likes of Dawkins and the other biologist types.

      We understand neurons etc. only through our consciousnesses. In a very real sense the neuron, as an ideational entity, is just the product of consciousness. Even the word "neuron" is a product of consciousness. To think that we can understand consciousness by studying its products seems to me grossly misguided -- and for the exact same reason that microfoundations and any form of reductionism is misguided: it misses the bigger picture. In the case of macro this is what I point out with the football stadium analogy; in the study of the mind it is the fact that the argument itself is being structured by the very thing it seeks to understand.

      With regards to consciousness all this has been pointed out since at least George Berkeley. And it is an argument that has NEVER been dealt with by the materialists. They just avoid it. I think materialism is a scam to be honest -- something like a scientistic religion (which is why some proponents are so obsessed with replacing religion). And its one, like neoclassical economics, that deals with its critics by ignoring them.

    7. By the way, I assume that we're all aware that Keynes was a subjective idealist and not a materialist. His theory of probability and his later work on uncertainty (and his critique of econometrics) is all in line with his a priori subjective idealism. This has all been dealt with at length in this excellent collection:


      Sample quote from Anna Carabelli's article "Keynes on Cause, Chance and Probability" (which is the best in the book):

      "The cognitive approach [i.e. subjective idealism] behind Keynes' view of cause implied a general shift from the realm of empirical uniformity, recurrence and regularity of events [i.e. the realm of materialism], to the thinking agents' reasons and grounds for believing." (pp155) -- All writing in square brackets is my own.

    8. Yes, the book looks interesting and I see it is edited by Tony Lawson too. I now have another book to read!

      You've given me a lot of food for thought..

  2. Also, let me just say I fear we may disagree on some issues here (like materialism), but I do get a lot out of your comments on my blog and appreciate them! they are challenging for me in a way that the annoying Austrians's comments here I aren't.

    I will read that Carabelli paper on idealism, but my first reaction would be: is idealism strictly logically necessary for the Post Keynesian view of economics? Sure, it seems compatible with it, but isn't materialism compatible too?

    E.g., Tony Lawson (who edited that book) seems to defend Post Keynesianism and materialism, but more than that:

    "A strata of reality can be said to be emergent, or as possessing emergent powers, if there is a sense in which it (1) has arisen out of a lower strata, being formed by principles operative at the lower level, and (2) remains dependent on the lower strata for its existence but (3) contains causal powers of its own which are both irreducible to those operating at the lower level and (perhaps) capable of acting back on the lower level. Thus organic material emerged from inorganic material. And, according to the conception I am defending, the social realm is emergent from human (inter-) action, though with properties irreducible to, yet capable of causally affecting, the latter.

    So interpreted, the theory of emergence commits us to a form of materialism which ultimately entails the unilateral ontological dependence of social upon biological upon physical forms coupled with the taxonomic and causal irreducibility of each to any other. Thus, although, for example, the geo-historical emergence of organic from inorganic matter and of human beings from hominids can be acknowledged, when we come to explain those physical and biological states which are due, in part, to intentional human agency it is necessary to reference properties, including powers, not designated by physical or biological science (again see Lawson, 1997)."

    Tony Lawson, "An Orientation for a green economics?," International Journal of Green Economics 1.3-4: 250-267 at pp. 257-258.

    1. Sure, Lawson can defend materialism if he feels like it. So can you. I think you're both wrong and I've already laid out some pretty clear arguments as to why. (Note that if the consciousness argument holds then ALL structures that rely on consciousness -- i.e. all human social structures -- cannot be understood at the level of materialist analysis). However, Keynes was not a materialist. In fact, he was quite consciously not a materialist. Kalecki was a materialist. Keynes was not.

      Are Keynes' theories about uncertainty compatible with materialism? I would say no, but that is a hard case to make. Are Keynes' theories of asset price speculation compatible with materialism? Categorically: no.

      The beauty contest dynamic is the subject of two or more consciousnesses trying to grasp at each other at a non-material level. It is an almost identical dynamic to the Hegelian master-slave dialectic [http://en.wikipedia.org/wiki/Master%E2%80%93slave_dialectic] -- by the way, I think I'm the first to point this out. This is a definitively non-materialist analysis.

      Does this mean you cannot squeeze it into the confines of materialism? Probably not. You could probably imagine that there is some bunch of neurons firing in each person's brain that is producing these effects (I believe these are called "mirror neurons" [http://en.wikipedia.org/wiki/Mirror_neuron]). Then you could say that you've "reduced" the activity. Well, I don't think this is what Keynes the idealist was talking about at all. And it runs into the problems I laid out with regard to attempts by materialists to understand consciousness. But you can make the leap if you really want. (Although I'm inclined to see this as materialists stealing from immaterialists and then repackaging their good ideas... but I'm not overly territorial in this regard...).

      By the way, I scribbled a few thoughts on the beauty contest here (not sure how relevant they are in this context, but hey...):


  3. So if there is a good and a bad reductionism, how do we distinguish between them in practice? And more to the point, why is neoclassical economics an example of bad reductionism? Because while it's certainly true that e.g. contemporary macroeconomists place a lot of weight on microfoundations, I'd argue that Keen's claims are strawmen. A general equilibrium model doesn't say that aggregate behavior is just a sum of individual behaviors - how could it be, when according to the model, individual agents take prices as given, yet those prices are determined by aggregate market-clearing conditions?

    So, for example, bad news about future growth could make individuals want to save more. But a GE model doesn't necessarily predict that aggregate savings will increase just as a sum of individual reactions would suggest - in an extreme case when the supply of saving assets is fixed, such model would predict that aggregate savings stay the same and only prices (interest rates) would change. Thus I'd argue that equating microfoundations with "greedy reductionism" is just false.

    1. "A general equilibrium model doesn't say that aggregate behavior is just a sum of individual behaviors - how could it be, when according to the model, individual agents take prices as given, yet those prices are determined by aggregate market-clearing conditions?"

      This is a very strange statement and I think that it may reflect a misunderstanding about the issues raised.

      Market-clearing prices are determined by supply and demand in neoclassical models. Supply is ultimately determined by peoples' decision to forgo leisure while demand is ultimately determined by individual tastes. The models try to incorporate these "microfoundations" and do not take account of the fact that they can lead to macro paradoxes (paradox of thrift, SMD conditions etc.).

      I'm afraid that I cannot really say any more. If you don't understand these issues then your best bet is to read up on them and then join the conversation.

    2. Actually, on reflection this has to do with exogenous imposed conditions on your GE models. You are imposing exogenous aspects from the outside and then claiming that this shows that its not all the work of the agents. This does not mean that the model is non-reductionist.

      For example, the saving example; it doesn't count. You're assuming a fixed supply of saving. I.e. you're imposing something on the actors exogenously. When I call a model reductionist its because it cannot endogenously generate macro-effects. Just imposing conditions exogenously and saying that the agents are constrained doesn't count at all.

      That would be like Lord Keynes saying to me that Dawkins is non-reductionist because he recognises that gravity constrains evolution and thus it is not "all the work of the individual genes".

    3. Yes, prices are given by supply and demand, but you can't say that price is a sum of properties of individuals, because individuals don't have any property called "price". So neoclassical economics is reductionist (like most of science), but it's not "crudely" reductionist in the same sense LK seems to have in mind when he cites Dawkins or Keen.

      Regarding SMD theorem, I hope you're aware that it's a _result_ derived from properties of general equilibrium models (and by general equilibrium theorists), not some external critique. In fact one could interpret SMD theorem as saying that since the structure of general equilibrium imposes little restrictions on aggregate excess demand function, GE models could in principle account for wide range of emergent behavior, i.e. in precisely the opposite way as you do.

      In fact one can obtain paradox of thrift (and other paradoxes) from microfounded models under certain conditions, for an example see "The Paradox of Toil" by Eggertson, or his later paper with Krugman.

    4. You can generate whatever on earth you want by imposing the right conditions on a nonsense model.

      The trick with Krugman is that he imposes a load of assumptions on his models about the state of the economy and then shows some effects. This is a very odd way of modelling. Modelling is supposed to explain the state of the economy, models are not supposed to be explained by it. That's completely bizarre.

      It's just another indication that neoclassical economics has gone completely into cloud cuckooland.

  4. I'll add to this by mentioning how this renders neoclassical economics a non-science. The trick is that the economist will look at the world around them and then start imposing them on the models. Eggertson looks around and sees ZIRP and output contraction and then imposes these on his models to generate his "paradox".

    Is Eggertson engaged in science? Is he EXPLAINING anything? Is he making predictions about the future? The answer is "no" to all three. This is more like fashion. Economists try to reflect their immediate reality and then show these results to one another. One could be quite rude about what is actually going on here. But I'll refrain. Let's just say that the word "ideology" would not go amiss.