Philosophy of Economics Crash Course 12 – “A Pure Tangle”


Dr. Alexander X. Douglas‘s biography states: “I am a lecturer in philosophy in the School of Philosophical, Anthropological, and Film Studies at the University of St. Andrews. I am a historian of philosophy, interested in the philosophy of the human sciences, particularly from the early modern period. I am interested in theories of human reasoning, desire, choice, and social interaction – particularly work that questions the foundations of formal theories in logic and economics from a humanistic perspective. I am particularly interested in the thought of Benedict de Spinoza, which continues to inspire alternatives to the dominant paradigm in economics and social science. My first book, Spinoza and Dutch Cartesianism, proposed a new interpretation of Spinoza, situating him in the context of debates within the Dutch Cartesian tradition, over the status of philosophy and its relation to theology. I am completing a book manuscript, which aims to introduce and develop Spinoza’s theory of beatitude. This is the culmination of Spinoza’s theory of desire, since it describes the condition of ultimate satisfaction. Although Spinoza saw the revelation of true beatitude as the ultimate goal towards which his philosophy reached, there are few interpretative works devoted primarily to this theme. Spinoza’s theory of beatitude is, in my view, the keystone that holds together diverse parts of his philosophy – his theory of desire and the emotions, his metaphysics of time, his theory of human sociability, and his philosophy of religion. These are often studied separately; my introduction to beatitude aims at helping readers understand Spinoza’s philosophy as a unified whole. I have also published a book examining the concept of debt from the perspective of language, history, and political economy. I’m interested in the philosophy of macroeconomics, which receives considerably less attention from philosophers than microeconomics. I am a member of the Centre for Ethics, Philosophy, and Public Affairs, the Executive Committee of the Aristotelian Society, the Management Committee of the British Society for the History of Philosophy, and a Research Scholar at the Global Institute for Sustainable Policy.”

In this series, we discuss the philosophy of economics. For this session, we come back after some time with session 11 on ‌fundamental‌ ‌premises‌, ‌utility-maximization‌ ‌automata, ‌a‌ choice, ‌Dr.‌ ‌Carolina‌ ‌Christina‌ ‌Alves‌, ‌human‌ ‌behaviour‌, ‌a metaphysical‌ ‌theory‌ ‌of‌ ‌fundamentally‌ “rational”‌ ‌human‌ ‌nature, ‌normative‌ ‌stance‌ ‌or‌ ‌ethic‌ ‌reflective‌ ‌of‌ ‌ ‌ideology, ‌political‌ ‌examples‌ ‌of‌ ‌Optimal‌ ‌Control‌ ‌Theory, ‌‌profit-motive‌ ‌examples‌ ‌of‌ ‌Optimal‌ ‌Control‌ ‌Theory, ‌understanding‌ ‌colonial‌ ‌narratives‌, and ‌the‌ ‌pretense‌ ‌of‌ ‌“control.”

Scott‌ ‌Douglas‌ ‌Jacobsen:‌ When looking at some of the philosophical systems sitting behind the economic theories, orthodox and heterodox, there’s, as you noted, a “very big step from ‘can be represented as’ to ‘is in fact.’” This seems as if a great point at which to begin to connect the philosophy of economics background to the heterodox economics expertise of Dr. Carolina Alves/Dr. Carolina Cristina Alves in “An Edge of Heterodox Economics 1 – Everything has a History.” Her series will sprinkle into this one, as yours will in hers.

With Dani Rodrik’s art/science or choosing models/building models split, you had an interesting non-throwaway phrase, “…‌I‌ ‌really‌ ‌haven’t‌ ‌seen‌ ‌the‌ ‌justification‌ ‌for‌ ‌taking‌ ‌that‌ ‌step,‌ ‌at‌ ‌least‌ ‌not‌ ‌in‌ ‌most‌ ‌cases.‌” On the opposite of “not in most cases,” there exist some cases. What are some of those cases? Those cases where the art of selection can be justified based on the models built as “science” (quoting Rodrik).

Dr. Alexander Douglas: Well, for example, the mathematical solution to noughts and crosses is quite simple, and adults play it reliably when they’re told the rules of the game and instructed to try to win. So here you have a mathematical model that reliably predicts and explains human action, in a very limited domain. A key point here is how much control needs to be exercised over the subjects for this explanation to work. The subjects need to follow the rules carefully, and they’re guided on what to do (try to win the game). Notice that they’re not setting their own agenda. Thus the model is no good for predicting how adults will behave when playing with young children whom they are trying to teach and encourage.

The economist and philosopher Don Ross has argued that the mathematical models used by economists should not be seen as explanations of human rationality. He thinks that human rationality is a crooked concept; there just isn’t one thing that it means to be rational independent of particular contexts and specific situations. Economic models are, rather, models that explain the workings of institutional mechanisms. The institutions make people behave in the algorithmic, maximizing way described by the models.

Ross is trying to defend economics, but he makes a very revealing admission. Economics, according to him, describes how people behave, not in general but within the institutions that make them behave in those ways. So he’s admitting that the theory works because reality is engineered to match it. Since this is something I’ve been arguing in the previous interviews, as part of a critique, I was surprised to find it being put forward as a defence of economics. Economics is often accused of being ideology rather than science. Ross thinks he is countering that accusation, but he seems to frame a new way of making it: if we build social institutions to make us behave in certain ways, and economists describe those, then economists are describing modes of control rather than patterns of behaviour. That sounds a bit like the role of the practical theologian or liturgist with respect to the church. It isn’t pure ideology, but it isn’t mere description either. 

This also relates to questions about decolonization in economics. If we start thinking of economists as sociologists or anthropologists with a specialization in certain cultural institutions of eighteenth-century European origin, we should rethink the role they have with respect to global policy.

Jacobsen: With these “human actions, choices, preferences,” and so on, ‘all having meanings.’ It raises some interesting questions about meaning as only a property in minds in relation to the world, not vice versa. If meaning arises in the context of any subject dealing with objects in the universe, then subjectivity imbues meaning, which isn’t seen as “relevant.” How do you build this aspect of subjective significance of things into the models? Is it even reasonably feasible with any precision?

Douglas: Yes, I was trying to avoid the very difficult question of what a meaning is. But the inference that meaning is irrelevant because it’s purely subjective works only if we assume that subjective factors aren’t themselves causes within the system. For example, whatever the shining of the stars might mean to us, the nuclear reactions that cause them to shine are one and the same. Here the meaning is irrelevant because it’s subjective. 

But with human action meaning is (I believe) among the causes. Let’s go back to the trading floor. When a trader buys some stocks in some manufacturing firm, we could describe her action as “investing in the production of peanut butter”. But this description gets the meaning pretty wrong. The trader might not even know what the stocks she’s buying are connected with, and she might be planning to sell them again in the next few hours. If she were investing in the production of peanut butter, we shouldn’t expect her to sell out very soon, but if she’s simply taking a temporary position or in the middle of a short-selling gambit then our expectations should be very different. Assigning a different meaning to one behaviour classifies it as a different action, and the predictive consequences are different.

Nor does it have to be the case that the meaning is represented by the actor for it to be causally relevant. Perhaps our trader isn’t even thinking about what she’s doing. Maybe she’s an old hand who has cultivated instincts and can run on autopilot most of the time. All the same, the institution in which she cultivated those instincts imbued her actions with the meanings. That’s why buying stocks on the trading floor is very different from, e.g., investing in a friend’s start-up company, even though, abstracting the actions away from their institutional context, they can fall under a single description (investing).

So we need to understand the meanings of actions, and there’s no science of this. We have to depend on our “commonsense” understanding, infused as it is with our moral instincts and cultural biases. We can’t depend on the scientific method to close these out, so the best we can do is keep the conversation open to a diversity of perspectives.

Jacobsen: How do you separate the “explanatory models” as “mathematical models” and the “descriptions under which the human actions fall,” while using this clear distinction to link them? In short, how can these subjective (and intersubjective) categories of meaning imbue the mathematical models with more robustness of aim?

Douglas: I’m not sure a mathematical model on its own can represent human actions at all. Human actions aren’t paths through some state space in which each dimension maps some salient variable. I struggle to communicate this, but take a simple example. Suppose we reduce a person’s driving behaviour to two variables: l, which is the number of times turning left and r, which is the number of times turning right. We can model driving as an optimization problem: maintain equality between l and r, or minimize |l-r|. More left turns will trigger more right turns, and vice-versa. I expect that model probably gets the quantities right over the long term. But of course it completely misses the point of what a driver is doing. Somebody who had only that model wouldn’t even understand what the point of driving was.

And I don’t think that simply adding more variables would get you closer to understanding what the driver is doing. A mathematical model just outputs a vector of quantities. These could be left turns, right turns, speed, distance, position, etc. But turning left to avoid an accident isn’t the same as turning left to test the steering wheel, or to correct for a previous mistake, or to follow the road, or to switch to a different road… Can you add more coordinates to the vector to track these differences? Of course, just as you can add more coordinates to track the colour of the car, the population of Paris, the number of craters on the moon… Which of these are salient and should go in the model? Well to know this you need to already understand driving, at some hermeneutic, non-mathematical level. When we’re looking at behaviour whose meaning we don’t already understand then we don’t know how to build the right mathematical model for it. And so mathematical models can’t explain behaviour. They can only regiment and formalize the understanding we already have.

Returning to Ross’s point: why then can algorithmic models, run on computer simulators, describe aggregate human behaviour within certain institutions? I say, because the institutions are themselves computers. At the limit you have a single piece of software implemented on two machines. One is the electronic computer running the economist’s model; the other is the computer running through the brains and institutions of human beings. The computer works by disciplining electricity to move in regular and predictable patterns through the circuits, rather than flowing more wildly as it does outside the machine. The institution does the same thing with human action; it regiments our thoughts to move in regular patterns like the current through a circuit board. A computational model can explain human action when human action is rendered computational. The mirroring can look like magic, but the conjuring trick is to cover up the institutional mechanism that makes it work.

Jacobsen: If the course of orthodox economic theorizing directs the “‌the‌ ‌dehumanizing‌ ‌language‌ ‌to‌ ‌the‌ ‌false‌ ‌mass‌ ‌psychology‌ ‌theory‌ ‌to‌ ‌‌ad‌ ‌hoc‌ ‌‌terminology‌ ‌to‌ ‌the‌ ‌complex‌ ‌mathematical‌ ‌models‌ ‌to‌ ‌the‌ ‌implied‌ ‌metaphysical‌ ‌theory” may not be a choice, is it, fundamentally, down to the consequences – economic cohort by economic cohort – of specific ‘sets of techniques’ where the advances happen by “pushing these techniques further”?

Douglas: Philosophers of science often talk about how the institution of science works by filtering out our natural human biases, blind-spots, etc. Scientific institutions pit humans into semi-competition against one another so that various idiosyncrasies and epistemic vices carry a cost and the elimination of less competitive theories drives convergence towards the truth. This works when there is a truth to converge towards. But with economics, I’ve suggested, reality is often engineered to match the theories rather than vice-versa. Then the scientific institutions of competition and filtration – peer-review for example – have a very different result. They work to force convergence onto a general plan for society – e.g. a model for how to build institutions – rather than onto some objective truth. I think the same is true of philosophy and other disciplines, so I’m not singling out economics for attack here.

Jacobsen: Dr. Alves argues Lionel Robbin’s An essay on the Nature and Significance of Economic Science (1932) became the point at which economics began formalization as a defined discipline, as old as some people’s grandparents. Economics, Dr. Alves, quotes, becomes “the science which studies human behaviour as a relationship between [given] ends and scarce means which have alternative uses.” A “science,” so a natural philosophy, given our conversations, this seems sincerely polyannaish, as per your example of the healthcare catastrophe happening in the UK (and elsewhere) with COVID-19.  Now, Dr. Alves notes the use of this term widely in economic discourse. What seem like the obvious consequences of asserting economics as a “science” on the state of economics over time – one person’s definition widely used?

Douglas: Carolina points out how Robbins’ definition works well with the ambition of Léon Walras to render economics as much like the “hard sciences” as possible. Physics in Walras’s time benefited greatly from models based on solving for equilibrium. You can use the same mathematics to explain human behaviour, if you reduce it to a problem of allocating means among ends. The solution to the model is a balance between competing demands, just as a physical model is solved as a balance between competing forces. Economists like Gary Becker made a big game of explaining unlikely behaviours as allocation problems and then creating sophisticated mathematical models to “solve” them.

Since Robbins there’s been a grand revolution in economics through the development of game theory. Economists can now discuss human institutions in a richer way, since they now model strategic interactions among agents rather than the “games against nature” that are allocation problems. But it seems no less fundamentalist to describe every human interaction as a strategic game than to describe every human activity as an allocation problem. So perhaps the modern-day version of that Robbins quotation is what’s found at the start of Ken Binmore’s Game Theory: A Very Short Introduction: “a game is being played whenever human beings interact”.

I don’t think that either of these reductions – of human activities to allocation problems and human interactions to game theory – can be justified on sociological or anthropological grounds. That’s to say, I see no reason to believe that most human activities and interactions are, in their ultimate meaning, allocation problems and strategic games. Why, then, are they all modelled as such? Because modelling them like that allows for fancy mathematics to be invoked. By contrast, representing all human activities as sacrificial ceremonies, as some twentieth-century anthropologists did, doesn’t allow for mathematization.

So I stress, there’s nothing inherently scientific about mathematizing something. Let this interview be modelled as an ordered pair of numbers (7,9). Let it be modelled as a vector of integers x,y Z2. Let it be a pair of elements, x, y, of a ring, R, defined as a countable set of elements {a, b, c, …} and an operation, ⊕, forming an abelian group. Have I explained anything you didn’t know before? Of course not – I’ve just bamboozled you with a lot of symbols and terminology. Buyer beware, with this sort of “science”. And buyer beware with fancy models of meaningful human activities as allocation problems with budget lines and indifference curves and local maxima, or games with dominant strategies and information sets and Nash equilibria.

Jacobsen: What are the benefits of having a career with the big journals using specific techniques? What are the benefits of doing things one’s own way as an amateur blogger?

Douglas: Of the first: research funding, academic positions, social status. Of the second… I’ll get back to you! 

Jacobsen: You note:

Economics‌ ‌goes‌ ‌off‌ ‌in‌ ‌so‌ ‌many‌ ‌different‌ ‌directions,‌ ‌even‌ ‌within‌ ‌the‌ ‌“orthodox”‌ ‌space.‌ ‌When‌ ‌you‌ ‌question‌ ‌economists‌ ‌about‌ ‌gaps‌ ‌in‌ ‌their‌ ‌theory,‌ ‌it‌ ‌feels‌ ‌a‌ ‌bit‌ ‌like‌ ‌being‌ ‌run‌ ‌around‌ ‌a‌ ‌bureaucracy… ‌With‌ ‌economics,‌ ‌it‌ ‌feels‌ ‌like‌ ‌the‌ ‌snapshots‌ ‌are‌ ‌all‌ ‌at‌ ‌different‌ ‌angles,‌ ‌and‌ ‌cut‌ ‌across‌ ‌each‌ ‌other‌ ‌in‌ ‌baffling‌ ‌ways.‌ ‌If‌ ‌you‌ ‌run‌ ‌them‌ ‌together,‌ ‌you‌ ‌get‌ ‌a‌ ‌pure‌ ‌tangle.‌ ‌

‌Dr. Alves stated:

Modern economics or what we call orthodox economics is about studying human interaction mainly through markets, where markets are theorized as being about the interaction between demand and supply, with equilibrium as a central concept and enduring reliance upon methods of mathematical modelling. This approach went to become ‘the mainstream economics,’ as it is the main and widely taught and researched approach.

Something “widely taught and researched” in “orthodox economics” that “goes off in so many different directions,” which produces “a pure tangle.” It sounds hopeless. If it can’t tell us “much about what we really want to know,” what do we really want to know if there you “don’t see any scientific approach to answering these questions emerging”?

Douglas: Well I probably should avoid the word “we” in that way; philosophers are always going on about “our” intuitions, “our” questions, and so on, and it betrays a lot of groupthink and ignorance of human diversity. But what I want to know is which institutions society should retain, which it should reform, and which it should replace. Now that I’ve read the Ross book, I think I have a clearer sense of why economic theories form such a tangle. They don’t describe human behaviour in general; they describe it within different institutional contexts. In the recent past, economists got in the habit of modelling everything as an abstract market in the sense Carolina means there: a mathematical optimization problem. Now they look at specific institutions (though these too are formulated as solutions to optimization problems – namely “games”). Institutions overlap in confusing ways.

But I’ve said that to mathematically model an institution or activity, you need to already understand it. How do we understand our institutions? I don’t know, and I don’t know that we do a good job of it. I just don’t think that social science as we have it helps us to gain understanding rather than to formalize understanding we already have. But the same faculty that allows us to understand our institutions, insofar as we do, is what we must rely on to think about how we might redesign our institutions. The insight of a novelist or an essayist might be more valuable here than all the mathematics in the world. De Tocqueville didn’t need equilibrium solutions to gain his insights into the ancien régime, nor Madame de Staël to understand the Napoleonic system. Ross writes at one point that while informal insight might have worked for people like Emile Durkheim or Max Weber, they don’t yield great results in the hands of “mere mortals”. But are we in any danger of running out of “immortals”? In any case, if we restrict ourselves to only looking at the institutions that we’ve learned how to reduce to equations, aren’t we going to miss out most of human life?

Jacobsen:‌ Insofar as algorithms “can be represented by mathematical equations,” and if you “‌take‌ ‌the‌ ‌meaning‌ ‌out‌ ‌of‌ ‌action‌ ‌and‌ ‌it‌ ‌becomes‌ ‌dead‌ ‌motion,” and if “meaning is everything in human life,” is economics, as a self-proclaimed “science,” a fruitless endeavour in generating theories or proper mappings of “meaning… in human life”?

Douglas:‌ That’s a good question, but I think the answer is no, because I don’t think that economists have really expelled meaning. They’ve just suppressed it. Since Milton Friedman’s essay on “the methodology of positive economics” in 1953, economists have philosophized as if all they’re trying to do (as “positive” economists) is track patterns so as to predict them. The realism of their assumptions is entirely irrelevant. In other words, they make it sound as if all they’re trying to do is find equations that output the data.

But their practice doesn’t match the theory. No economist explains stock prices by assuming that some omnipotent being determines their movements by tossing coins, though that would correctly “predict” the observed random-walk pattern. Nobody explains recessions as being caused by cosmic rays, though with the right assumptions in place one could easily generate the appropriate time-series data. If all economists were trying to do was predict data, why wouldn’t their theories consist of pure, uninterpreted equations? In purely mathematical terms, setting up a system of optimizing agents is a needless detour; you might as well just curve-fit a polynomial that directly outputs the data series you want.

The truth is that economists see the world working a certain way, and their models reflect this. They model society as a system of self-interested agents because, despite all their protestations, they’re telling a story about human nature and human society. And in doing so, they do ascribe meaning to actions and institutions: the meaning of the actions is self-interest and the meaning of the institutions is strategic balance in a power-struggle. Whatever economists might say, that will never just be a pure fiction used to generate an empirically robust mathematical model; it will always be a story economists tell us about ourselves, and we will always be entitled to ask whether it’s the right story.

Jacobsen:‌ ‌What are some other ‘strong doses of philosophical anthropology’?

Douglas: Since Ross’s book has been a theme for this interview, let me end with his idea about behavioural economics. This, he thinks, is a thoroughly misguided enterprise. It takes results of studying humans in experimental contexts, isolated from ordinary institutions, and tries to apply them to the behaviour of humans outside those isolated experimental contexts. For example, psychologists put people in a lab and find that they don’t “maximize” the way they’re supposed by economists to do. But, Ross argues, take them out of the lab and put them in a market setting – put them, say, on a trading desk or on the board of directors of a firm in which they’ve invested – and there’s no reason to expect that they’ll act the same way they did in the psychologist’s lab. Here the institutional setting primes and trains them to act as economic agents rather than subjects of a psychological experiment. “Maximisation”, in other words, is a social behaviour into which people are enculturated through capitalist institutions – a ritual they’re trained into.

Well, how is that for philosophical anthropology? Ross has probably been in some board meetings for American companies, so I’m sure he knows what he’s talking about. But this is in the background of economic theories that explain how we act as economic agents; “economic agents” means participants in the rituals and culture of certain familiar capitalist institutions. Western capitalist institutions, that is. Would Ross be as confident that economists’ models will hold up with respect to the behaviour of the directors of, say, an Indonesian state-run firm? I doubt that he should be. Ross advocates for the fusion of economics with sociology, but the examples of his chosen sociology come again from the study of familiar Western institutions, and are again heavily mathematized. Here the implicit philosophical anthropology is the assumption that human agency is, in general, amenable to mathematical treatment, and that behaviour within Western institutions reveals certain fundamental and universal principles.

Photo by Tiko Giorgadze on Unsplash

Image Credit: Alexander Douglas.

Leave a Reply