« How would tighter regulation affect mortgage origination? | Main | Tax rebates don't always work »
In defense of (some) neuroeconomics
Andrew Samwick at Vox Baby reports on the neuroeconomics sessions at the AEAs. He is justifiably impressed with the work of David Laibson. Laibson has pioneered the theory and implications of hyperbolic discounting.
But a clever economist can always come up with a rational (time-consistent) model to explain what appears to be irrational hyperbolic discounting. Laibson, however, uses fMRI scans to show that different parts of the brain are activated when making decisions at different time-scales. As Andrew notes, the isolation of the different decisions to different parts of the brain gives Laibson's argument significant credibility against more standard neo-classical explanations for the same phenomena.
Posted by Alex Tabarrok on January 18, 2008 at 02:06 PM in Economics | Permalink
Comments
Does it? As I understand it, explanations from rationality are a species of functional explanation. Rationality is a function of the brain. Function, however, is distinct from mechanism. And fMRI scan will reveal something about a mechanism. But it will not reveal the function of that mechanism, or the absence of function.
Posted by: Constant at Jan 18, 2008 2:24:33 PM
Like you, I admire this work. But keep in mind we don't know exactly what the scanis telling us. A "different part" of the brain may light up, but that itself doesn't refute the view that there is a unified decision-maker. I reject the unified decision-maker hypothesis for other reasons, one of which is simple introspection.
Posted by: Tyler Cowen at Jan 18, 2008 2:55:34 PM
One has to be Bayesian about this - it's not a question of prove or refutation but of evidence. Suppose Z=f(x,y) and someone argues that x and y are chosen to maximize Z. If it turns out that x and y are chosen by different people that, all else being equal, ought to reduce the probability that Z is maximized.
Posted by: Alex Tabarrok at Jan 18, 2008 3:34:47 PM
I'm always skeptical of the applicability of studies like this to economics. People can learn to do all sorts of crazy things their nature and brain chemistry never intended. They can learn to sky dive, race cars at 300 mph, or do all sorts of other things that would light of the fear centers of normal people's brains like christmas trees. The division of labor obviously takes care of "deficiencies" in many people's brains, at least when there are incentives which reward rational behavior. My usual reaction to some study that says "most people make irrational decisions when deciding between X and Y" is: "So what? Most people can't figure out the best valve angle to use in their car's cylinder head, but that doesn't keep cylinder head design from improving".
I wonder to what extent the evolved systems in the economy have corrected for hyperbolic discounting? The average wage-worker gets his health care paid for, and probably a retirement fund as well. He gets paid a constant value every week, regardless of what the future holds, and is usually incapable of taking on more debt than he can pay back (since lenders try to avoid that sort of thing). He does have to plan for his children's future, but that is something which there seems to be strong incentives to do and so less hyperbolic error (not being a parent, I wouldn't know). Entrepreneurs and professional investors, on the other hand, trade short-term consumption for greater long-term gain. Are they evolved specialties which came to be because hyperbolic discounting prevents others from being as rational with long-term plans?
If there is still room for improvement to correct for hyperbolic discounting, will that lead to more of those goal-setting services with financial incentives? (i.e., "Give us $X. If you meet goal A, you get $X + $Y back, where Y is some fraction of X") I don't ask this pertaining to health care, since I am completely unfamiliar with many health care regulations, which for all I know prevent insurance companies from being able to offer similar incentives.
Posted by: Grant at Jan 18, 2008 4:48:43 PM
It is not new that different parts of the brain light up when people
are engaging in apparently different kinds of thinking. The Ur-paper
on neuroecon that may yet win a Nobel is the 2001 one in the PNAS by
McCabe, Dickhaut, Houser, and Smith (with Houser and McCabe apparently
being the only ones out of Vernon Smith's shop remaining at GMU, or so
I hear). The big finding? That when people are playing prisoner's
dilemma games, different parts of the brain light up if they are
contemplating cooperating or if they are contemplating "cheating,"
aka "being economically rational and going to the Nash equilibrium"
Amusingly, in fact tellingly, it is when one is thinking about cooperating
that the prefontal cortex lights up, the area long thought to be the main
seat of "rational intelligence." This reflects the complaint of the
mathematician who was playing with the economist Armen Alchian in the
first repeated PD game experiment at Rand back in the early 50s (Dresher
and Flood), who kept being annoyed by Alchian's repeated tendency to
"selfishly" cheat: "Why is this guy so stupid?"
that the
Posted by: Barkley Rosser at Jan 18, 2008 5:11:46 PM
One would imagine from most of the time-integrated history of evolution that the short term discount rate is highly survival based [no meat today - I die] and the the long term discount rate is barely, if at all, encoded in the brain [everyone will die in 10 years, anyway]. So the hyperbolic stuff makes a lot of sense.
Posted by: JPC at Jan 18, 2008 7:54:30 PM
Economists need to be convinced at very single step of the way that their model is wrong and that they ought to discard their assumptions and accept biological assumptions. What incredibly slow and stubborn learners they are. Faith dies hard.
Social science ought to be based on biological science and Darwinian evolution.
Unified decision maker: It is an obvious illusion.
Posted by: Randall Parker at Jan 18, 2008 9:30:33 PM
Grant,
Regarding "normal people": People who skydive get a huge reward out of that experience that most people do not get. Those people who skydive and do other extreme sports are the subject of biological research to identify which genetic alleles are responsible for their thrill-seeking behavior. Alleles in dopaminergic pathways are the biggest suspects in this research.
When you see that someone has taken of the practice of doing X or Y you should always ask whether the people who do that have some average cognitive difference based on genetic of epigenetic influences that give them more of a reward for that behavior or which otherwise compel them to do the behavior even though they might feel no reward.
Look at people with obsessive compulsive disorders. Do these do these obsessive compulsive behaviors because they've learned to overcome fears? In most cases no.
Posted by: Randall Parker at Jan 18, 2008 9:42:02 PM
Perhaps skydiving was a bad example (having never done it, I wouldn't know), but motor sports wasn't. Your average person may think that racing around a track at 150 mph is scary, extremely difficult, or maybe insane. But its actually a rather easy thing to learn (at least, to learn mildly well, no where near a pro's level), and becomes routine after a while. The professionals who participate in motor sports (or similar activities) don't get a "thrill" out of it (maybe they did their first time?). With the exception of a few hot-heads who hopefully do not remain in the sport for long, they are professionals. They do their job with surgical precision.
You mentioned rewards for that behavior. Where there are incentives for people to overcome pre-programmed behaviors, I think they will often do so.
I just don't see much of a difference here between the race car driver and the professional investor. Both attempt to do things that are utterly alien to the environment which humans evolved in, and both utilize whatever non-human (computer) tools are available to do so.
Those who plan poorly for the future will generally tend to spend their money rather than accumulate it. This seems like a good thing to me, because more of that money will end up in the hands of people who do not exhibit temporal errors in their decision-making: the professional investor. I would think that would lead to better investing and therefore capital distribution.
Although I suppose this is much more of a Neoclassical blog than an Austrian one, so the Neoclassical assumptions of rationality are probably more directly challenged?
Posted by: Grant at Jan 19, 2008 12:18:26 AM
It might interest some of you that there is now a paper forthcoming in the QJE by Andrew Caplin and Marc Dean which takes some neuroscientific insights to economics!
It´s about reinforcement learning and reward prediction errors.
The paper is called "Dopamine, Reward Prediction Error, and Economics" and can be found on Andrew Caplins Webpage. It´s a theory paper, by the way.
Posted by: Neuro at Jan 19, 2008 3:29:32 AM
But a clever economist can always come up with a rational (time-consistent) model to explain what appears to be irrational hyperbolic discounting.Seems to me that by the point neoclassical economists are resorting to this sort of objection, they've hit the point where they're explaining away the data that look like irrational hyperbolic discounting rather than explaining them. You can't just come up with an arbitrary mathematical model ex post: there has to be some theoretical justification for your model other than that it's consistent with your preexisting theory, unlike the simple, obvious explanation proposed by your opponents.
So I don't really disagree, but I think you've given neoclassical economics a little bit too much credit.
Posted by: Elliot Reed at Jan 19, 2008 11:11:54 AM
