Abstract revolutions or: if something is new, the old thing must be bad

There is a part of this small, otherwise enjoyable article about the advent of neuroeconomics that bothers me. The premise of the piece is that neuroeconomics is “seeking a physical basis for [economic theory] inside the brain”. This is a field that is certainly sexy and possibly exciting; a while back I argued that it is a field that in some sense rediscovers the absolute primacy of “preferences” as the keystone of economic theory. I said:

The potentially exciting thing about neuroeconomics is that, even allowing for inexactness, it might tell us more about the actual hedonic motivators of people. Ambitious, yes, but not unimaginable. Of course, to an economist who wasn’t under the mistaken impression that simplified preferences are supposed to be realistic, it might just amount to saying “your simplification is a simplification”, which is slightly less exciting news. Or not news at all.

OK. In today’s article, we learn that 

modern economic and financial theory is based on the assumption that people are rational, and thus that they systematically maximize their own happiness, or as economists call it, their “utility.”

Since it’s hard to figure out what is going on in people’s heads, the argument continues, we employed an idea called revealed preference, the reconstruction of unobserved objectives from observed choice. Neuroeconomics, the claim runs, may one day be able to identify brain structures that are associated with various components of choice, and so neatly sidestep the problem of unobservability.

But this is much too much:

While Glimcher and his colleagues have uncovered tantalizing evidence, they have yet to find most of the fundamental brain structures. Maybe that is because such structures simply do not exist, and the whole utility-maximization theory is wrong, or at least in need of fundamental revision.

Utility-maximization theory is wrong. It is wrong by construction, because it is a model, and models are wrong by construction. Why must we go through this? It might be easier to sell me an iPod if you first convince me that CDs are useless, but that’s marketing. Why do we need to market neuroeconomics this way? That tiny little word “wrong” up there is a sin, because it belies the very essence of modeling as a means to make sense of things.

Is it perhaps that what we should understand by the quote is that looking into the brain will tell us that people are not, in some sense, maximizing their hedonic pleasure by the choices they make? This amounts to both an attempt to open the black box called “preferences” and to pin down the (biological?) process by which decisions are actually made. If this is the sense in which utility-maximization will be proved “wrong”, then in the first place it is not clear to me that neuroeconomics can accomplish such a thing. Leaving aside the tricky questions of intent, free will and consciousness, can there be anything inside the black box but another, and another? The leap from the correlates of physical choices in brain activity to the content or existence of a utility function is huge.
But more importantly, even taking literally the notion that we will be able somehow to trap preferences or process in a cage, it is surely impossible that any breadth or depth of evidence on what these preferences are could preclude the need for us to model. What if the thankless treadmill of refining imperfect models of the world could at last be switched off, that there is an apple of knowledge that will free us from the need for models ever again? This is a seductive idea, but it cannot be. To do away with models would be to be as complex as reality, and that is a fight that reality will win every time. 

Experimental philosophy, experimental economics

Interesting article at Prospect Magazine called Philosophy’s Great Experiment, about the rise of ‘experimental philosophy’. Doubly interesting to me, because it could equally well be talking about experimental economics, albeit a few years too late. Or about “neuroeconomics“; the philosophers in the article are using fMRI machines to look for patterns of neuronal activity when subjects are presented with philosophical problems”, just like the researcher who does the same for resource allocation – economics – problems. But here’s the rub:

Some philosophers quietly dismiss the movement as a cynical step by researchers to appear cutting edge and to tap into scientists’ funding.

Indeed, it’s easy to feel this way about the kind of experiments in which economists step on psychologists’ toes. The drive toward empiricism in philosophy that the article talks about seems to be symptomatic of social sciences’ and humanities’ desire to be taken “seriously” as science.
And as we know, that means we need something falsifiable or verifiable. “There is no article in Prospect Magazine called Philosophy’s Great Experiment” is falsifiable, because I can find such an article and falsify the statement. “There is at least one article in Prospect Magazine called Philosophy’s Great Experiment” is verifiable, because I can find such an article and verify the statement. 
In experimental economics, often it seems (at least to this observer) that we’re replicating, or at least mirroring, psychology experiments. Unfortunately, the economics experiment is much less likely to be “scientific”, not because of the method or the issue at hand, but because of the specific question. This is precisely what Lawrence Boland discusses in the paper “On the futility of criticizing the neoclassical maximization hypothesis” (pdf), which I read as a welcome withering put-down to all of those who claim to “disprove rationality“, etc etc. He says:
Properly stated, the neoclassical premise is: ‘For all decision makers there is something they maximize’… The person who assumed the premise is true can respond: ‘You claim you have found a consumer who is not a maximizer but how do you know there is not something which he is maximizing?’

For experimental economists and experimental philosophers alike, the challenge is to pose a scientific question; without that, no method will save us. “Are people ethical?”, for example, is equally a dead end as “are people rational?”. 

Looking at brains

A related question to what people want is what makes people happy. The answers aren’t necessarily the same, but in thinking about one, especially as a modeler, it helps to think about the other.

I saw Andrew Caplin talk a couple of months ago about this paper, in the growing neuroeconomics field. My layperson’s understanding of neuroeconomics is that it looks at the brain to try and figure out what’s going on while we make choices. It seems to be concerned at least as much with “what makes people happy” (or at least what makes their brain light up on the machines) as it is with what people actually choose, a point that separates it somewhat from other economic fields that take preferences as given and observe only choices. The argument, presumably, is that knowing more about what motivates people might shed more light on their choices.

Caplin described a paper titled, excellently, “Responses of monkey dopamine neurons to reward and conditioned stimuli during successive steps of learning a delayed response task” (link) which described a Pavlovian experiment on monkeys. First the monkeys were given food, which set off a dopamine response in their brains, supposedly a sign of pleasure. Then the researchers started to ring a bell before giving the food, and the dopamine response shifted from the food to the bell. Finally, they rang the bell and withheld the food, and the dopamine measure showed pleasure at the bell and disappointment at the time when the food was expected.

The upshot is that this physical measure of pleasure in the brain seems to show that news was just as important as the tangible outcome: the monkeys don’t just care about the food. People, too, clearly care about more than just money, goods or tangible outcomes: perhaps I am happy to see a good weather forecast for tomorrow even though I didn’t actually get anything.

It’s related to the great loss and rediscovery of preferences in economics. At some point (to grossly oversimplify the history) the profession settled on making abstract simplifications of people’s motivations. Then a while went by and someone said “people don’t conform to your economic theory, they must be irrational!”, when of course the more obvious explanation is that people’s motivations are more complex than the simplification. If people care only about money, then according to some theory I should see this. If I see that instead, I have not disproved the theory independently from the assumption on preferences.

The potentially exciting thing about neuroeconomics is that, even allowing for inexactness, it might tell us more about the actual hedonic motivators of people. Ambitious, yes, but not unimaginable. Of course, to an economist who wasn’t under the mistaken impression that simplified preferences are supposed to be realistic, it might just amount to saying “your simplification is a simplification”, which is slightly less exciting news. Or not news at all.

In fact, one interpretation of the monkey result might be that new information – or expectations – matters to people’s decisions. Macroeconomists – Nobel memorial winners, Phelps and Friedman – incorporated expectations into their models forty years ago. If anything, what we’re seeing is individual-level (or at least individual monkey-level) evidence that, yes, expectations are important. That might be another way of saying that evidence from experimental/behavioral/neuroeconomics like this is fun, but it doesn’t materially change what economists are up to: if more complexity makes our modeling more accurate, we often end up doing it.