My essay on the Ghost of Descartes was written by cannibalising a longer, not quite finished essay entitled “Cartesian vices, Copernican moments”. In writing something else, I find myself wanting to refer to another part of it, so I’m hastily topping and tailing the relevant section and posting it here. I’ll follow it up with another fragment on Michael Polanyi who was an all-round better neoliberal guy than Hayek for reasons I’ll elaborate when I publish it.
Hayek’s markets as embodied cognition
The ideas of embodied cognition help vivify economists’ appreciation of markets. The issues were brought to life most fully in the ‘socialist calculation debate’ of the 1920s. It was already well understood that administrative mechanisms could blunt market incentives for self-interest to drive economic improvement.1 But Ludwig von Mises, and then his fellow Austrian, Friedrich Hayek observed that even if the incentive problem could be overcome, in a centrally planned economy, decision-makers could not obtain the information they needed to make economically rational decisions.
How would the centre determine the value of different commodities if they weren’t traded in a market? It seems intuitively clear that it would be massively more complex to run such information from all corners of the market to some central brain to make production and allocation decisions. Yet, socialist economists, Lange and Lerner proposed just such a scheme. To simplify it somewhat, consumers would pay prices and producers would be paid to produce with the central planner sitting between them receiving and paying each appropriately. Where production surpluses and deficits appeared, the central controller would change its instructions, the prices it set and/or the payment producers received in a ‘trial and error’ process just as markets mutually adjust demand and supply within the market.
(As an aside on re-reading this, the ‘socialist’ side of this debate is as classic a case as one can imagine of an intellectual pathology in which the scientist imagines their own construct as part of the world itself, rather than a construct of it. The neoclassical construct Lange and Lerner are seeing in markets has been built from one particular intellectual mapping of what goes on in a market. The neoliberals are right that that’s for deeper reasons. Nassim Taleb often tackles something similar when he critiques neoclassical theorists of risk for imagining that the world is the casino of risk they imagine it is, when most of the risk people must manage in the world is a very different thing – uncertainty. There’s a similar split between Kahnemann’s work which is quick to assume that naïve decision making has ‘biases’ when compared with ‘optimal’ decision theory, when Gigerenza and the embodied cognition crowd more generally think that those heuristics we can discover are the building blocks of ecological rationality.)
Be that as it may, in distributing important decisions and actions throughout the organism and its environment, causal spread does not just disperse the cognitive load and increase the system’s responsiveness to knowledge that is dispersed throughout the economy.2 In just the way the fly manages to fly3 and we manage to catch a ball,4 it also lightens the cognitive load – often massively. In prosecuting his case, Hayek identifies “the most significant fact about this system” but it is not the capacity of markets to align incentives, to organically adjust supply to demand or to utilise distributed knowledge, though Hayek stressed the latter as fundamental. Rather, he identifies “the economy of knowledge with which [the market] operates, or how little the individual participants need to know in order to take the right action” (emphasis added).5
Quite apart from the incentive problems they glossed over, the socialist models also lacked this informational parsimony. With economists from Adam Smith to Friedrich Hayek assuring us of the miraculous properties of markets, as Tabarook puts it, the relatively new subdiscipline of mechanism design shows us also how difficult it is to replicate this feat of informational efficiency and incentive compatibility – or in the language of embodied cognition, causal spread. “Theorists may one day design a better market mechanism … but for now the gains will come from using our deeper understanding to gently improve something that’s already pretty marvelous”.6
Causal spread also often unifies different domains within the organism and its environment bringing the brain and other subsystems in the organism into alignment around a particular activity like catching a ball or evading a predator. This too can be illustrated by reference to the Austrians’ appreciation of the economic efficacy of markets. The Austrians’ claims in the socialist calculation debate showed how markets unified informational efficiency with incentive compatibility. Hayek went on to explain that markets reconcile two very different domains. They reconcile systematic knowledge of the kind one learns from books and at university with temporary, context specific and even tacit knowledge that the agent possessing it may be unable to articulate.
John Gray makes the point powerfully:
In social theory, Hayek’s devastating critique of Cartesian rationalism entails that, whatever else it might be, social order cannot be the product of a directing intelligence. It is not just that too many concrete details of social life would always escape such an intelligence, which could never, therefore, know enough. Nor (though we are nearer the nub of the matter here) is it that society is not a static object of knowledge which could survive unchanged the investigations of such an intelligence. … Such an impossibility of central social planning rests, firstly, on the primordially practical character of most of the knowledge on which social life depends. Such knowledge cannot be concentrated in a single brain, natural or mechanical, not because it is very complicated, but rather because it is embodied in habits and dispositions and governs our conduct via rules which are often inarticulable. But, secondly, the impossibility of total social planning arises from the fact that, since we are all of us governed by rules of which we have no knowledge, even the directing intelligence itself would be subject to such government. It is naive and almost incoherent to suppose that a society could lift itself up by its bootstraps and reconstruct itself, in part at least because the idea that any individual mind—or any collectivity of selected minds—could do that, is no less absurd.7
Hayek often quoted Alfred North Whitehead to generalise his point about the need for informational efficiency. Whitehead traduced the pompous maxim “repeated by … eminent people … making speeches” that we should always think what we are doing, or as it is more commonly expressed these days we should ‘question everything’. As Whitehead put it “the precise opposite is the case. Civilization advances by extending the number of important operations which we can perform without thinking about them”. Hayek expands on the social significance of this insight
We make constant use of formulas, symbols, and rules whose meaning we do not understand and through the use of which we avail ourselves of the assistance of knowledge which individually we do not possess. We have developed these practices and institutions by building upon habits and institutions which have proved successful in their own sphere and which have in turn become the foundation of the civilization we have built up.
The contrast between Asimo’s and passive dynamic robots’ navigation of the world provides a useful analogue of the difference between a centrally planned economy and a well functioning market. So too would be a ‘Cartesian’ implementation of catching a ball in which the central brain takes charge, processes signals from the visual system to predict the ball’s landing point whilst sending all the relevant individual instructions out to the scores of muscles that must coordinate to carry the catcher to the right spot and then softly close their hands around the falling ball in just the right way at just the right time.
The two robots also provide an analogue of the contrast between the Austrians’ focus on cognitive efficiency in an economy and the Rube Goldberg/Heath Robinson machine neoclassical economists have built to make good on their scientistic method. As in the case of the Austrians, the neoclassical understanding of a market disposes of the need for a Godlike central planner. But where, for Austrians, the whole point of their reflections and the resulting abstraction was to render an economy populated by human beings in all their fallibility and variety, neoclassical economics’ can only remove the godlike central planner by making everyone in the economy godlike, omniscient about the economic conditions, today and forever into the future. Hayek argued that the central economic problem was how “the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess” might most productively be brought into relation. It’s a powerful formulation. The neoclassical approach provides as breathtaking a begging of the questions it asks as could be imagined.
- Murray N. Rothbard, 1991. “The End of Socialism and the Calculation Debate Revisited”, The Review of Austrian Economics, Vol. 5, No. 2 (1991): 51-76. ↩
- In fact even in the Lange-Lerner story, the central planner does access knowledge distributed throughout the economy through trial-and-error adjustment of prices. But it does so in a cruder way than a market does. ↩
- You might think that when a fly flies, its primitive brain has decided to do so, though the word “decided” might be in scare quotes, meaning no more than that the fly’s brain controls this function of switching to flying. It “looks like” it. But you’d be wrong. As David Marr tells us, the fly’s body has a reflex system according to which, when their middle legs kick off from a surface the wings buzz and, as we’ve all experienced, they do so until the fly lands again. (The closely coupled wiring of these reflexes – by-passing the fly’s brain – is what makes flies’ reflexes so quick and thus so hard to swat!). Marr, D. 2010. (1982) Vision:A Computational Investigation into the Human Representation and Processing of Visual Information: San Francisco, Freeman.pp. 32-3. ↩
- If you ask someone how they run to the right spot to catch a ball in a park almost everyone will plump for the Cartesian explanation that their mind predicts where it will land. It might “look like” it to you, but it doesn’t if you look more attentively. The brain does play an important role, but it doesn’t tacitly or otherwise make calculations to predict the ball’s trajectory. Instead, our bodies and minds use their collective resources far more craftily, and with vastly greater cognitive economy. We move so that the image we have of the ball appears to us as a straight-line, constant speed trajectory against our visual field. ↩
- Hayek, F. (1945). The use of knowledge in society. American Economic Review, 35(4), 519–530. At https://www.econlib.org/library/Essays/hykKnw.html. As one of the architects of mechanism design and Nobel Laureate for his troubles Eric Maskin observes “Hayek had a remarkable intuitive understanding of some major propositions in mechanism design – and the assumptions they rest on – long before their precise formulation. Indeed, his understanding seems to have been a guiding influence in their formulation”. Note relaxing these assumptions does not suggest that other mechanisms are more informationally efficient than markets, but rather prevents any strong conclusions being drawn. Maskin unpacks Hayek’s case for the informational efficiency of markets into two claims which have subsequently been validated by mechanism design – though, it must be admitted, not without the customary list of unrealistic assumptions.
First, they are more parsimonious in what they require participants to know than administrative mechanisms. They must know only what they want and how much at going market prices – things which they will almost invariably know better than others (who do not know their preferences and, in markets do not need to know!). Second they have no incentive to mislead those with whom they trade about this information – as they would for instance if they were asked how much they wanted some good the full cost of which they did not have to pay. In the language of mechanism design, there is incentive compatibility.
Eric Maskin, 2015. “Friedrich von Hayek and mechanism design”, Review of Austrian Economics, 28:247–252. ↩
- Tabarrok, Alex, 2007. “What is Mechanism Design? Explaining the research that won the 2007 Nobel Prize in Economics”, October 16, ReasonOnline, at http://reason.com/news/show/122998.html, recovered at https://web.archive.org/web/20090903163108/http://reason.com/news/show/122998.html. ↩
- Gray goes on:
…It is the neglect of how all economic life depends on this practical knowledge which allowed the brilliant but, in this respect, fatally misguided Joseph Schumpeter (1883–1950) to put a whole generation of economists on the wrong track, when he stated in his Capitalism, Socialism and Democracy (1942) that the problem of calculation under socialism was essentially solved. It is the neglect of the same truth that Hayek expounded which explains the inevitable failure in Soviet-style economies of attempts to simulate market processes in computer modeling. All such efforts are bound to fail, if only because the practical knowledge of which Hayek speaks cannot be programmed into a mechanical device. They are bound to fail, also, because they neglect the knowledge-gathering role of market pricing.
Here we must recall that, according to Hayek, knowledge is dispersed throughout society and, further, it is embodied in habits and dispositions of countless men and women. The knowledge yielded by market pricing is knowledge which all men can use, but which none of them would possess in the absence of the market process; in a sense, the knowledge embodied or expressed in the market price is systemic or holistic knowledge, knowledge unknown and unknowable to any of the elements of the market system, but given to them all by the operation of the system itself. No sort of market simulation or shadow pricing can rival the operation of the market order itself in producing this knowledge, because only the actual operation of the market itself can draw on the fund of practical knowledge which market participants exploit in the their activities.
The “problem of the division of knowledge,” which Hayek describes as “the really central problem of economics as a social science,” is therefore not just a problem of specific data, articulable in explicit terms, being dispersed in millions of heads: it is the far more fundamental problem of the practical knowledge on which economic life depends being embodied in skills and habits, which change as society changes and which are rarely expressible in theoretical or technical terms. ↩