The economists teamed with psychologists at Stanford to turn an M.R.I. machine into a shopping mall. They gave each experimental subject $40 in cash and offered the chance to buy dozens of gadgets, appliances, books, DVDs and assorted tchotchkes. Lying inside the scanner, first you'd see a picture of a product. Next you'd see its price, which was about 75 percent below retail. Then you'd choose whether or not you'd like a chance to buy it. Afterward, the researchers randomly chose a couple of items from their mall, and if you had said yes to either one, you bought it; otherwise you went home with the cash.
The good news, for behavioral science, was that the researchers saw telltale patterns, which they report in the Jan. 4 issue of the journal Neuron. "We were frankly shocked at how clear the results were," said Brian Knutson, the Stanford psychologist who led the experiment. "It was amazing to be able to see brain activity seconds before a decision and predict whether the person would buy it or not."
The first was my nucleus accumbens, a region of the brain with dopamine receptors that are activated when you experience or anticipate something pleasant, like making money or drinking something tasty. In the experimental subjects at Stanford, this region was activated when they first saw pictures of things they wanted to buy. My nucleus accumbens just happened to respond more strongly than the typical subject's, so what else could I do? If it feels good, buy it.
The other culprit -- the main villain, really -- was my insula. This region of the brain is activated when you smell something bad, see a disgusting picture or anticipate a painful shock. It was typically activated in the brains of the other shoppers when they saw a price that seemed too high. I'd like to think of my insula as particularly stoic, the strong, silent type, but he's probably just an oblivious slob.
This data shouldn't be too surprising. There's an extended literature documenting the insula's aversion to losses (like paying for a good), and the nucleus accumbens (NAcc) has long been a card-carrying member of the dopamine-reward pathway. (Whenever we expect a reward, the NAcc floods our mid-brain with neurotransmitter.) This study also extends the dual process model, which is quickly becoming the unifying theory of neuroeconomics. According to this model, our behavior depends upon the interaction of two (or more) antagonistic decision-making systems. At any given moment, one part of our brain is busy evaluating costs and losses, while the other part of our brain is monitoring expected gains. Sometimes, these conflicting approaches can be divided into "rational" (mPFC, etc.) and emotional (amygdala, insula, NAcc, etc.) brain areas. But often, the tension is entirely emotional, as in the case of shopping. Our decisions often depend entirely upon our feelings. (There's a long history of neurological patients who have suffered injuries to their "emotional" brain systems, and are thus unable to make "rational" decisions. As David Hume put it, "reason is, and ought to be, the slave of the passions." He was right.)
Of course, this is yet more evidence that the classical assumptions of economics are entirely false. Instead of being wired to maximize income (we are supposed to be selfish and rational), our mind contains a collection of competing and contradictory parts that are suffused with emotion. Our decisions emerge from this mosaic of neural activity. To borrow a metaphor from Isaiah Berlin, if economists believe that we think like the hedgehog - we always fall back on the same strategy of selfish rationality - neuroeconomists have discovered that we actually think like the fox, and employ different strategies in different circumstances.
Personally, I was most surprised by what the Knutson/Loewenstein study didn't see. Unlike other decision-making studies, they didn't detect activation in the orbitofrontal cortex, the anterior cingulate cortex, or the parietal cortex. The authors excuse this anomaly by noting that their experiment didn't involve any learning, and that the ACC and other parts of the prefrontal cortex are most activated during decision-making tasks that force people to take new information into account. Shopping, apparently, has zero educational value. (But I was still surprised to not see the ACC, especially since it is often activated by "mental conflict," such as mediating between situations that generate both "loss" and "gain" signals.)
Those minor caveats aside, I hope this study is read by economists and policy makers eager to increase the anemic savings rate of Americans. As the authors note, credit cards take advantage of our brain, since they minimize the activation of the insula:
With respect to economic theory, the findings support the historical notion that individuals have immediate affective reactions to potential gain and loss, which serve as inputs into decisions about whether or not to purchase a product (Kuhnen and Knutson, 2005). This finding has implications for understanding behavioral anomalies, such as consumers' growing tendency to overspend and under-save when purchasing with credit cards rather than cash. Specifically, the abstract nature of credit coupled with deferred payment may ''anaesthetize'' consumers against the pain of paying (Prelec and Loewenstein, 1998). Neuroeconomic findings thus might eventually suggest methods of restructuring institutional incentives to facilitate increased saving.
Excellent post. I liked Tierney's suggestion for a credit card that showed the bank account statement. That might get the insula going!
"Of course, this is yet more evidence that the classical assumptions of economics are entirely false. Instead of being wired to maximize income (we are supposed to be selfish and rational), our mind contains a collection of competing and contradictory parts that are suffused with emotion."
This is a mischaracterization of the classical economic behavioral assumptions. The individual is assumed to maximize utility, not income. Money is often used as a metric of utility because people reliably have monotonically increasing utility functions for money due to fungibility (or in lay terms, people reliably want more than they currently have because it's easy to exchange for other stuff). The classical model of individual decision making is that there is a black box which determines the expected gain or loss of utility from an option and those valuations are acted on rationally.
The study actually butresses the classical model. It appears to validate the idea of utility as having some psychological grounding rather than simply being a convenience for modeling, which actually suprised me. The sequence of activity also corresponded in the areas that would be expected - the NAcc for anticipating benefits, insula for costs, and mPFC for evaluating the tradeoff - this actually matches quite nicely with how preference has been treated historically, with the NAcc and insula corresponding to the black box which generates preference function and the mPFC evaluating options in terms of these results. The separate mechanisms for evaluating loss and gain and that costs seem to be evaluated as direct losses rather than as a loss of other positive options is the most radical departure for classical models, and may provide a basis for explaining a variety of odd behaviors, such as endowment bias.
Hi Matt. Thanks for your great comment, as usual. And you're right: I should have referred to a maximization of utility, not income or wealth. (Although this study, like so many others, used money as a proxy for utility.)
But I'm going to have to disagree with you about whether or not this study buttresses the classical model. This study, like so many others, demonstrated that our brain contains distinct neural mechanisms for evaluating losses and gains,. This is not how rational agents are supposed to behave. In fact, this anatomical setup is probably responsible for all sorts of economic anomalies, the most notable of which is loss aversion. Foregone gains (opportunity costs) are supposed to be treated equivalently to out of pocket costs (losses), but they aren't, because they are detected by different brain areas. If Milton Friedman had designed our brain, he wouldn't have the insula and NAcc doing different things. Instead, there would be a common calculator inside our head that responded to both losses and gains simultaneously.
It's also worth noting that this process is largely driven by emotional signals, not rational or explicit utility calculations. And while I subscribe to the "affect-as-information" model, in which our emotions are signals of value (like a neural price tag), it's pretty clear that our emotions are triggered by all sorts of "irrational" cues, like a loss frame. So even if the mPFC is "evaluating" the tradoff and trying to maximize utility, it's clear that the signals it is using to make a decision are highly biased and distorted by their emotional sources. (In other words, the amygdala or insula or NAcc are only interested in specific types of information.)
That said, one of the lingering mysteries of neuroeconomics - at least as far as I'm concerned - is that the more reductionist you get, the more our cells seem to obey the assumptions of classical economics. Single cell recordings (see Glimcher's work, or the Padoa-Schioppa Nature paper) show that the firing rate of our cells in the PFC closely obey utility theory, in the sense that a single neural currency is used to evaluate various alternatives. The alternative with the highest rate of firing is what we select.
But I think from the perspective of systems neuroscience, the relationship of utility theory and neural anatomy is far more contradictory. Time and time again, we have seen that our utility calculations do not take place in any single brain region, and do not depend upon the sort of explicit analysis that rational agents are supposed to undertake. Instead, there are only contradictory emotions and conflicting gut instincts, which other parts of our brain (usually the mPFC or ACC) do their best to decipher.
I agree with Jonah on this one. I think its easy to imagine a brain imaging study that would have confirmed the economic model of decision making. There would be a single black box in the frontal cortex that would do a cost-benefit analysis, and then choose the maximizing alternative. this calculation would probably be accessible to conscious introspection. What they would not find is a distributed network of brain areas that respond to discrete parts of the problem and are defined by their generation of implicit, automatic cues. What these neuroscientists found might support prospect theory, but it doesn't support conventional utility theory.
I think we agree about the facts of the matter but see it as whether the glass is half empty or half full with regards to the usefulness of the classical way of looking at preference. By default I assume that economic models aren't necessarily accurate descriptions of the physical systems they represent - what may behave in aggregate in a particular way can be composed of unexpectedly varied components, especially in biological systems. A lot of concepts in economics are simply useful fictions for describing the behavior of large groups - I don't even entertain the idea that there is an actual value for "inflation" out there corresponds to the relationship between nominal and "real" rates of interest, but it's an useful way of looking at prices and yields reasonably accurate predictions.