Talk of the Nation, National Public Radio: “Decoding the Science of Decision Making”
July 24, 2009 –
PAUL RAEBURN, host:
This is SCIENCE FRIDAY from NPR News. I’m Paul Raeburn, sitting in for Ira Flatow.
From the time your alarm clock rang today, you’ve probably made thousands of decisions. The first was whether to hit the snooze button. Some of the decisions are big and deliberate. Do I actually go to work today or do I stay home and listen to SCIENCE FRIDAY? Others are miniscule and mostly unconscious, putting the toothpaste down on the right side of sink instead to the left or deciding to quickly and unconsciously scratch that mosquito bite on your elbow.
In the last few decades, psychologists, neuroscientists, even economists have been trying to pull back the curtain on our brains to see what’s going on when we make these decisions, large or small, figuring out for example why someone would decide on Coke instead of Pepsi, could mean millions of dollars for someone in the soda business. If you’re a lawyer, it could be a question of life or death for your client. How does a juror decide whether to convict?
For the first part of this hour, we’ll talk with three researchers studying decision making. If you want to get a piece of the conversation, give us a call. Our number is 1-800-989-8255, that’s 1-800-989-talk. If you’re on Twitter, you can twit us your questions by writing the @ sign followed by Scifri. And if you want more information about what we’re talking about this hour, go to our Web site at, www.sciencefriday.com, where you’ll find links to our topic.
Now I’d like to introduce my quests. Michael Frank is assistant professor of cognitive and linguistic science, psychology and psychiatry. He is director of the Laboratory for Neural Computation and Cognition at the Brown University Institute for Brain Science. He joins us today from Providence, Rhode Island. Thanks for being with us, Dr. Frank.
Dr. MICHAEL J. FRANK (Assistant Professor, Cognitive & Linguistic Sciences; Director, Laboratory for Neural Computation and Cognition, Brown University Institute for Brain Science): I’m very happy to be here.
RAEBURN: Jennifer S. Lerner is a professor at the Harvard Kennedy School, director of the Decision Science Laboratory at Harvard University. Thanks for talking to us today, Dr. Lerner.
Dr. JENNIFER S. LERNER (Professor, Harvard Kennedy School; Director, Decision Science Laboratory, Harvard University): I’m pleased to be here, thanks.
RAEBURN: And Colin Camerer is the Robert Kirby Professor of Behavioral Economics at the California Institute of Technology in Pasadena. Thanks to you too for joining us.
Professor COLIN CAMERER (Behavioral Economics, California Institute of Technology): My pleasure.
RAEBURN: Again, our number is 1-800-989-8255. Give us a call. Colin Camerer, what are you economists doing messing around with the psychologists and neuroscientists?
Prof. CAMERER: Well, I would say trying to reform economics in the sense that the standard economics theory that was developed mathematically starting around 1950s, and which continues to be taught and subject of every exam and also guides a lot of policies, is that people know what they want; they optimize by choosing the right – spend the right amount of time and money on things, even when decisions are very complicated – hundreds of mutual fund choices, distant – there are words are distant like at age 25, you’d have to choose a package that will give you a term and funds when you’re in your 70s, and there are very complicated probabilities to calculate it or in some cases, what we call ambiguity, we really know very little about the probability.
So, the economic model has been a very mathematical stylized, kind of Martha Stewart meets Mr. Spock on emotional calculating caricature. And it may apply in some cases but often not. It’s not a great model of human nature to build the entire social science on. So, what we’ve been trying to do, is to really import psychology and the increasing (unintelligible) looking directly at brain function when people are making somewhat complicated economic choices.
RAEBURN: Now, people talk about – what you’re talking about I think is behavioral economics or neuroeconomics, what do these terms mean?
Prof. CAMERER: So, behavioral economics takes the idea that people have some normal limits – not pathological limits, although pathology is useful to understand general mechanisms – normal limits on how much they can compute and figure out on greed and on willpower, and try to put those in mathematical terms that kind of soften this sharp caricature of perfect rationality that’s been the standard theory. Behavioral economics basically doesn’t say very much about detailed brain mechanisms, neuroeconomics does.
RAEBURN: Can you do something with the psychology, with mathematics or do you have to abandon the mathematics and use a different approach altogether?
Prof. CAMERER: No, in fact, I think one of the – one of the biggest innovations is that, first the mathematics just becomes harder. But that’s actually great because there are a lot of mathematical economists who’ve figured out all the simple stuff and are actually enjoying the challenge nowadays, or trying to mathematize something like envy or social influence.
RAEBURN: So, harder becomes harder just because these things are so difficult to quantify.
Prof. CAMERER: Yes, exactly. And also there’s a lot of different possible avenues, something like the things we’re talking about, willpower, addiction, exploration. Michal will talk about, often there are, you know, there are a lot of dimensions and you have to kind of comment to a particular approach. So, typically, for example, nowadays economists are very interested in modeling self control, and there’s not a single mathematical model, there are several. So then the ball goes in the court of empiricist to say, which are these seems to be on the right track?
RAEBURN: Michael Frank, it was your paper in Nature and Neuroscience that prompted this conversation today. Tell us a little bit about that. What did you find?
Dr. FRANK: Well, people have an automatic tendency, based on strong drives at the motivational system, to choose actions that have been associated with good outcomes in the past and to avoid those that had negative outcomes. That’s of course, and adaptive thing to do. But, sometimes we don’t know, really, what the best outcome is from our actions until we try other actions, essentially until we explore. And then we have to override this basic reinforcement system in order to gain more information and to explore when we’re more uncertain. So, we found was that people who’d a particular gene variation – their DNA – were more likely to export choices that had uncertain outcomes than those who had the other form of the gene.
RAEBURN: So, if choice that has an uncertain outcome, what would be an example of that?
Dr. FRANK: So, one example would be if you’re just used to going to the same restaurant again and again, and you kind of enjoy it, but you have lots of other options out there and you’re not sure whether you’re actually going to like the other option or you might not like the other option. So, the more you’re uncertain about that, the less you know about it or the more variability that you have in your experience is where the particular choice that you might make, the more uncertain you are both that and then in order to gain more information to find out whether it’s actually better than what you’ve been doing before. That’s the condition where you might want to explore.
RAEBURN: So, when I’m deciding whether to go to my favorite restaurant or to try this new restaurant, you’re telling me that seems like a fairly trivial thing that wouldn’t require a whole lot of computing power up in my head. But you’re telling me that it does require enough to at least be encoded partly in our genes.
Dr. FRANK: Right. I mean essentially you have to compute a lot of the things, you’re not actually computing, explicitly in your head. You’re not sort of measuring in your head exactly how much other uncertainty you have about going to a particular restaurant or anything. But, an example that I like to give sometimes, is if you’re faced with choosing – like you had in your introduction, salmon versus steak – you might have chosen both of those in the past many times and what you go on, which one you chose on a given day, might depend on your sort of integrated experience of all the times you’ve had steak or all the times you’d salmon in the past, together with a bunch of other contextual factors, of course.
But it’s not like you’re remembering each individual instance which you had salmon or which you had steak. Instead your brain essentially is remembering that in one sort of integrated value. But the computations necessary to compute that integrated value is a function of your past experience, of course, require complex brain mechanisms which themselves are influenced by the gene. And similarly the computations of how uncertain you are about that value are also influenced by brain mechanisms.
RAEBURN: I’m little afraid of what we think about this too much. I’m not going to have any idea what I have for dinner tonight by the time we’re done talking about it.
(Soundbite of laughter)
RAEBURN: Dr. Lerner, you’ve talked about emotions and their role in decision making, just to make this thing even more complicated sounding that it already is, go ahead and tell us what role emotions play?
Dr. LERNER: Sure. Well, it’s useful to first think a semantic distinction between two different kinds of emotion that can enter into judgment and decision making. The first is what we call, integral emotion and these are feelings that one may have about the judgment or decision at hand. So, for example, if I’m about to make a decision about whether to invest in a particular stock or change the allocation of my retirement savings, that sort of thing, I may have feelings about that decision and feelings about – and I can anticipate feelings like how I would feel if the market dropped further or if it started to go up, I can have an excitement or worries.
Those are all integral to the judgment or decision at hand, and from a -traditionally a standard economic model can actually do – incorporate those feelings relatively easily.
There’s another kind of feeling, which we call incidental emotion, and these are feeling that one just happens to be having at the time of a judgment or decision and that have no normative relevance to the judgment or decision at hand.
So say, you’ve just had a – you’ve just driven into work and someone cut you off in traffic, which actually happens quite often here in Cambridge, and you nearly had a fender bender, and you’re all mad, and then you go in and you’re making this decision about re-allocating your retirement savings. Your anger over having this unpleasant altercation and the bad driver who ended up maybe treating you badly, your anger about that should not influence your decisions that you’re about to make, and in fact they should just be incidental but not influential.
But what we find over and over again is that once the emotion systems are activated, for example, the anger system in your body, it creates a whole bunch of changes, and the brain is just not wired up to say, oh, that anger is from the traffic and should not be influencing my stock allocations, but instead it just carries over, and very often it carries over when we’re not even aware of it.
So those are the two kinds of emotion in simple terms: integral and incidental. Both have strong influences on judgment and decision-making. One of them is a kind normatively defensible influence because it is relevant to the judgment or choice at hand, and the other is not, but both occur with quite – with intensity and commonly.
RAEBURN: Are you or others trying to quantify that mathematically as well, as we discussed with some of the other aspects of this?
Dr. LERNER: Absolutely. We do experiments in our laboratory, and we – everything, all of our results boil down to a mathematical model, where we can actually quantify the degree of the emotional influence on a variety of judgment and decision-making outcomes.
RAEBURN: And so this is where the Mr. Spock image comes up again. The mathematics of emotions sounds inherently contradictory.
Dr. LERNER: Well, it isn’t really. Actually, emotion for much of the 20th century, there’s an interesting history and philosophy of science story here, was considered unscientific and not worthy of study, and…
RAEBURN: Let me stop you right now. We’ll hear more about that when we come back. We’ll be right back after this short break.
(Soundbite of music)
RAEBURN: From NPR News, this is SCIENCE FRIDAY. I’m Paul Raeburn. We’re talking this hour about the science of decision-making. My guests are Michael J. Frank, assistant professor of cognitive and linguistic science, psychology and psychiatry, and director of the Laboratory for Neural Computation and Cognition at Brown University; Jennifer S. Lerner, professor at the Harvard Kennedy School, director of the Decision Science Laboratory there; and Colin Camerer, the Robert Kirby Professor of Behavioral Economics at Cal Tech.
Jennifer Lerner, you were telling us about the emotions and mathematics. I think you also did an experiment about looking at anger in decision-making. Tell us about that.
Dr. LERNER: Okay. Want to hear about the experiment first or about the issue of whether emotion can be studied scientifically?
RAEBURN: Well, finish the discussion about emotion and scientific study, and then we’ll talk about the experiment.
Dr. LERNER: Okay. So B.F. Skinner, one of the founders of classic behaviorism in psychology right here at Harvard, said this famous quote. In the middle of the last century he said, Emotion is one of the fictional causes to which we attribute behavior. And essentially he thought it did not – emotion didn’t need to be studied at all. We just thought that emotion was influencing our behavior, but it wasn’t really. And his view really did dominate much of the last century’s thinking.
RAEBURN: That was one of the things that made him such a controversial figure, wasn’t it?
Dr. LERNER: It was, and emotion was really not studied very much in the 20th century, and it was – in fact, there’s – one of the first books that Charles Darwin ever wrote is called “The Expression of Emotion in Man and Animals,” but for the last century it went out of print, and you couldn’t even purchase that book. That is how out of fashion or out of scientific fashion the study of emotion was. But toward the end of the last century there were a number of methodological advances.
Brain imagining is one of them, also a lot of advances in studying the autonomic nervous system, and advances – some of them have become quite popular – in coding fine muscle movements in the face and being able to quantify different emotion expressions and how those expressions can actually predict underlying biological changes. And so we now internationally, across laboratories, have a number of very good tools for studying emotion, and in part through working with our colleagues in economics, improved tools for studying judgment and decision-making.
So there’s now this really thrilling field that’s burgeoning right at the intersection of psychology, economics and neuroscience and making a number of discoveries and also in particular with an exciting focus on how emotion influences these outcomes, again combining a psychological, an economic, and a neuroscience perspective.
RAEBURN: So tell us about the anger experiment. The anger seems to be – here’s my unscientific view. Anger is responsible for a lot of bad decisions by a lot of people in a lot of circumstances.
(Soundbite of laughter)
Dr. LERNER: Yes. We have been doing research on anger for about 15 years in terms of how it influences judgment and decision-making. So we have a lot of experiments, and the typical paradigm that we’ll use is one where we’re interested in incidental anger.
So as I described a few minutes ago, we’re interested in the times when anger from one situation carries over and influences decisions in another situation, even when people, for the most part, are not aware of it.
This is a very old problem. It was actually Aristotle who first said, Anybody can be angry, but to – and he said, that’s easy, but to be angry with the right person at the right time and in the right way, that’s not easy.
(Soundbite of laughter)
Dr. LERNER: And so that is what we find in our empirical data in the lab.
RAEBURN: Now we know Aristotle was right, after all this time.
Dr. LERNER: Aristotle was right. He’s actually – “Nicomachean Ethics.” I highly recommend the book. It has a lot of predictions that are bearing out in our highly controlled laboratory experiments.
RAEBURN: I want to take some calls, but just in a nutshell, what did you find in this experiment, that the anger does carry over to other situations.
Dr. LERNER: Anger does carry over, and I’ll give you one kind of non-intuitive result. Most people think that when you’re in a negative mood, like anger, you’ll have a negative outlook. You’ll be pessimistic. It turns out the opposite.
Anger makes you optimistic and makes you perceive less risk than if you were in a neutral state, and if makes you take more risks. So for example, you’re more likely to choose a gamble over a sure thing when you’re angry.
Anger does a lot of other things, as well. It makes you think more heuristically rather than systematically. It automatically activates relative left frontal hemisphere, which is associated with approach. So when you’re mad, it predisposes you toward believing things are going to work out your way, believing that you have some sense of control. It gives you a sense of certainty, makes you take more risks, perceive less risk, think less deeply, a whole series of choices, and we have, if anyone wants to read, a summary of a lot of these experiments in a paper that is easily accessible online called “Portrait of the Angry Decision-maker.”
RAEBURN: Okay, we’d like to take a question from Jonathan Wesley(ph) via twitter, who wants to know, Michael Frank, if big and small decisions are processed differently in the brain.
Dr. FRANK: Well, I guess it would depend on what big and small mean. If you mean small like some of the examples you gave before, in terms of just simple motor decisions that you’re not aware of, like where you put your soap and things like that, versus big decisions like which career you should take, then I was saw absolutely they’re processed in different ways in different parts of the brain.
Of course, they share some overlap in terms of how value might be assigned to some of these decisions. So where you put your soap originally, when you’re learning to move as a child and through development, you’re trying to figure out what is the best way to move and with the most efficiency and the least cost, the least effort, and some movements work better than others, and so that system slowly gets ingrained until it becomes unconscious.
Similarly, when you’re actually making deliberative decisions, it’s possible that some of the same brain areas that are involved in selecting between simple motor programs are also involved in selecting between sort of more complex cognitive decisions, although they’re not exactly in the same parts of the brain. They’re sort of overlapping circuits, we think.
RAEBURN: Colin Camerer, can you tell us a little bit about what light a functional MRI sheds on some of these questions?
Dr. CAMERER: Well, MRIs has a few advantages. It has severe limits, which people sometimes talk about and sometimes gloss over. It doesn’t have really great temporal and spatial resolution compared to some other techniques, like measuring directly the firing rate of a particular neuron, but at least it gives us a starting point, and it’s often a very useful compliment to other types of tools to try to understand geographical specialization in the brain.
And importantly, that doesn’t mean that we’re going to find, say, an envy region of the brain. More likely what we’ll find is, and what we do find, is sort of neural circuitry, several different regions that are causally influencing one another, maybe in a detectable time path, and that can then be the object of a computational model.
And in a few cases, in what’s sometimes called decision neuroscience, we’ve – there’s very good convergent evidence that in valuing a wide range of objects from very simple things like a drop of juice if you’re thirsty to more complex things like which charity to give money to, that there’s activity in fairly specialized areas in what’s called the medial orbital frontal cortex, that’s in the middle, above the eye sockets.
That’s the orbital part, like, right between your eyebrows, and also in regions called the ventral striatum and caudate(ph), deep in the kind of oldest part of the brain. And a little bit is also known about how people with damage in those regions sort of behave abnormally in terms of simple decision-making.
So there’s some converging evidence on lots of different objects that need to have values that are traded off to make complex in decisions where in the brain that’s occurring, and once we have some clues about where things are happening, then we can start to do other kinds of really interesting causal experiments, which you couldn’t do.
So in our lab and many others, people use something called TMS, transcranial magnetic stimulation. What that is is a magnetic coil that’s held on the outside of the brain and can stimulate an area on the kind of surface of the brain about the size of a quarter.
So if you know, for example, that a particular region is involved in sort of assigning a monetary value to food when you’re hungry, and we disrupt that activity in that area, does that mean that monetary values become more erratic? Does that mean monetary values go down?
So we can actually influence brain activity that way, and then there are lots of other tools for doing this, like electrical stimulation, pharmacological manipulations that adjust the amount of serotonin or dopamine or something in the brain.
So understanding the geography of the brain is really just partly a stepping stone to then be able to really confirm things with lesion patients and these causal influences.
RAEBURN: Jennifer Lerner, we have another question from Twitter from 1213, who asks whether decision-making and things we do based on habit come out of the same parts of the brain or the same system, or whether those are completely different things.
Dr. LERNER: That’s a fascinating question. We don’t have a definitive answer to that yet, but we are starting to make important inroads on that, in part because, for example, in the research that we’re doing on incidental emotion, a lot of times, they carry over a curse without people realizing it.
And so, for example, if they’re in a sad state, we observe that they spent more to buy something than if they were in a neutral state. And we made them sad by having them watch a short little film clip of about four minutes. And they’ve been randomly assigned to watch that clip or randomly assigned to watch a neutral clip.
And then afterwards, we give them a task that involves buying and selling goods, involving real commodities, real money. And afterwards, we’ll ask them, did what you saw in that video clip in any way influence the prices that you set? And people will tell us, no. And they actually are offended. And they say, you know, how could you suggest that I would? That would be ridiculous if it did.
And so, there’s a question between these sorts of non-conscious processes, where emotions are influencing how much we buy and – the price that which we buy and sell. But the process is opaque to intersection. So, the subjects themselves don’t see that it’s going on. And then getting back to your question, how similar different is that kind of non-conscious process from what we would call a habit?
So we are investigating this now on the behavioral level and increasingly on the neuroscience level. And these are some of the fascinating questions about how the mind works that the field is now taking up.
RAEBURN: Michael Frank, can the genetics work that you do be correlated with the geographic work that comes out of the fMRI scans?
Dr. FRANK: Absolutely. And there’s a whole line of research that’s attempting to do so and we’re following up on our original research and trying to do so as well.
What you can do is you can genotype people and put them in the scanner and look at whether activations in particular parts of the brain that are, in this case, involve in decision-making or reward processing, whether the activations that you see very according to the genotype.
So in some cases, you do see that where you’ll put someone in the scanner and then you’ll deliver rewards to them in a form of either financial rewards that are gambling and making money or you can actually deliver juice directly to their mouth if they’re thirsty and you actually can see that the reward area -sorry – the reward areas of the brain can sort of activate, they light up. And to the extent that they have particular genotypes that might actually modulate their sensitivity to reward, which could also predict their sensitivity to reward in other aspects of their life.
RAEBURN: Now, what other types of decisions can you study at the genetic level? These genes involved Dopamine, is that correct?
Dr. FRANK: That’s right. So you can study – and what I mentioned before is there’s the gene that predicts your tendency to sort of override your more basic reinforcement system in order to explore. But, of course, you can study the more basic process itself, and we and others have done that as well, so that your sensitivity to rewards versus your sensitivity to negative outcomes, like losses or punishment, and those depend on separate goal mechanisms within the brain. And there’s also genetics that are associated with us to aspects of learning.
There’s also other ways to sort of carve up the space of sensitivity to reward and decision-making, for example, temporal context. So, if went to a restaurant and you had uh, salmon once, and it was terrible, do you remember that very last experience and decide to change your whole strategy for restaurants, or you do sort of take a longer-term view? And that sort of poses a computational tradeoff in terms of the brain systems that are involved in learning from or responding to very recent history versus integrating sort of over the course of your lifetime what seems to be true on average.
RAEBURN: Let me take a moment to remind everyone, I’m Paul Raeburn. This is SCIENCE FRIDAY from NPR News.
So, Colin Camerer, tell us a little bit about the particular work that you do, some of your recent research and how that bears on some of these questions.
Prof. CAMERER: Okay. So a recent study we did, that came out in Science this May, had to do with self-control. And in that study, as I mentioned, it’s something which was talked about a lot. It seems to be a very important in the economy as – and we went down and noticed the savings rate in the U.S. in modern times is quite low, sometimes negative. And credit card debt is enormous, something in the scale of several thousand dollars per household. And in addition, where there’s had been a tremendous worldwide rise in obesity, people just eating too much and not exercising enough. And that has an enormous cause in terms of later health care.
So self-control is a big stakes thing that were interesting philosophically and from a critical point of view. So our paradigm is very simple. People see actual foods: yogurt, Snickers bars, Cheetos, and rate them on how healthy do you think they are and how tasty do you think they are. And then we asked them whether they want to choose this food or kind of a neutral food that’s in between on health and taste.
And these are people who say they’re dieting and struggling a little bit to control their diet. And what we find is that when there’s activity in the dorsolateral prefrontal cortex – I apologize for the jargon, but these do have names because we talk about in that way. Dorsal means sort of upward in the brain, lateral is on the side. It’s a spot maybe two or three inches above your eyebrow.
And when there’s stronger activity in that region, that’s predictive of when people are able to exhibit self-control in a sense of say no to things that aren’t healthy but are really tasty.
And as I mentioned, once we know a little bit about the neurocircuitry, the dorsolateral PFC and how it’s kind of modulating the weight that’s placed on health, which is also being expressed in another part of the brain and medial LC. Once we start to know something about the circuitry – and there are other regions involved, too. I’m just highlighting a couple. We can start to ask about genetic influences, particular neurotransmitters, we can do TMS and things like that. So we, you know, interpreted in an overly glib way. It’s something like an attempt to understand the locus of will power in the brain.
And related to what Jennifer works on, another thing which people can do outside of fMRI is – we know that this dorsolateral prefrontal region is also activated in working memory tasks, like if you’re trying to pay attention while driving or remember somebody’s phone number.
So what happens if you overload the DLPFC with an additional kind of cognitive distractor task, does that mean that people then have trouble resisting tempting foods? The prediction would seem to be yes.
So once we know something about these areas and what other things those areas are involved in, we can – it generates a big range of other cognitive experiments you can do. So there really is a kind of – fMRI is sort of the glamorous tool because pictures of the brain is so beautiful, but it also works best when it tells you other kinds of experiments that hundreds of people can’t do if you don’t have brain imaging.
RAEBURN: Well, I think that’s about all the time we have. I’d like to thank my guest for being with us. Michael J. Frank, assistant professor at the Laboratory for Neural Computation and Cognition at Brown University, Jennifer Lerner, professor at the Harvard Kennedy School, director of the Decision Science Laboratory there, and Colin Camerer, Robert Kirby Professor of Behavioral Economics at Caltech.
When we come back, a look at the new science-fiction movie “Moon.” Stay with us. We’ll be right back after this short break.
This is SCIENCE FRIDAY from NPR News.