Friday, November 25, 2011

Brain Science and Economics

It's becoming more possible every day to see what happens in the human brain. One tool is functional magnetic resonance imaging, or fMRI, which measures which parts of the brain are especially active when facing certain kinds of choices. There is "transcranial magnetic stimulation" and "transcranial
direct current stimulation," where a researcher can increase or decrease neural activity in a specific region of the brain, and observe the effect on choices. There is electroencephalography or EEG data, where a bunch of wires and stickers record electrical activity on the scalp, which is picking up neural activity in the brain.

What can economics learn from this evidence? The Fall 2011 issue of my own Journal of Economic Perspectives offers a couple of articles on this theme.  Two economists, Ernst Fehr and Antonio Rangel, offer their sense of what has been learned in "Neuroeconomic Foundations of Economic Choice—Recent Advances."  Two brain scientists, Marieke van Rooij and Guy Van Orden and then offer a counterpoint in "It’s about Space, It’s about Time, Neuroeconomics and the Brain Sublime."

Fehr and Rangel note several times that economic applications of brain science to how people make choices are really just getting . However, they also believe that these studies are coming together around a five-part model of how the brain makes choices:

1. The brain computes a decision value signal for each option at the time of choice.
2. The brain computes an experienced utility signal at the time of consumption.
3. Choices are made by comparing decision values using a “drift-diffusion model.”
4. Decision values are computed by integrating information about the attributes associated with each option and their attractiveness.
5. The computation and comparison of decision values is modulated by attention.

While the evidence on some of these statements is stronger than for others, and while complex choices like whether to save for the future are going to harder to understand than a choice about which music you would prefer to listen to while having your fMRI scan, they make a solid case for what has been learned.

Van Rooij and Van Orden are more skeptical about what has been learned. They point out that looking at parts of the brain with fMRI scans and trying to identify them as the nexus of cooperation or calculation or risk-taking or other behaviors has been only a modest success so far. Different studies often seem to locate decisions in different parts of the brain, or in different constellations of brain activity.

They point out that looking at data from fMRI scans is actually a matter of looking at about 130,000 "voxels," which are sort of like pixels in a video screen, except that they estimate whether a certain point in the brain is more or less active. In doing such comparisons, the obvious question is how many of these "voxels" must be different to show that a certain area of the brain has really acted? Indeed, what actually happens in these studies is that the fMRI brain scans of people who undertake these activities under different conditions have their voxels averaged together, and then these averaged-together images are morphed onto a common brain format for comparison. The results of such studies are, understandably, not always robust.

They argue further that it may be mistaken to think of the brain as a set of spatial locations where different decisions happen. Instead, they argue that the brain may be characterized by movements over time, where activities start in certain parts of the brain but then propagate--or not--into other parts of the brain. They tend to favor EEG data over fMRI scans, because the EEG data can capture lots of spots of brain activity second-by-second over time. They argue that the brain over time follows patterns that can be discerned using analysis of fractals.

Both sets of authors are careful to emphasize that the field is very young, and both are optimistic about its possibilities. Fehr and Rangel write: "Neuroeconomics is a nascent field. Much of the basic work remains to be done, and many of the details of the computational models of choice described
here are likely to change and evolve over time. However, we hope that this description of the current frontier of neuroeconomics convinces economists that a great deal has already been learned about how the brains make choices, and that these findings already provide insights that are useful in advancing our understanding of economic behavior in many domains." Similarly, van Rooij and Van Orden write: "[B]ringing together reliable economic paradigms with reliable brain science holds a possibility for taking this new science beyond anyone’s wildest dreams."