Skip to content

Cognitive vs. behavioral in psychology, economics, and political science

I’ve been coming across these issues from several different directions lately, and I wanted to get the basic idea down without killing myself in the writing of it. So consider this a sketchy first draft.

The starting point is “behavioral economics,” also known as the “heuristics and biases” subfield of cognitive psychology. It’s associated with various studies of cognitive illusions, settings where people systematically mispredict uncertain events or make decisions. Within psychology, this work is generally accepted but with some controversy which could be summed up in the phrase, “Kahneman versus Gigerenzer,” but it’s my impression that in recent years there’s been a bit of a convergence: for Kahneman the glass is half-empty and for Gigerenzer the glass is half-full, but whether you’re talking about “heuristics and biases” or “fast and frugal decision making,” there’s been a focus on understanding how our brains use contextual cues to decide how to solve a problem.

In economics, this work is more disputed because it seems to be in head-on conflict with models of utility-maximizing rationality from the 1930s-50s associated with the theories of Neumann and others on economic decision making. While some economists have embraced so-called “behavioral” ideas to explain imperfect markets, other economists are (a) skeptical about the relevance to real-world high-stakes behavior of laboratory findings on cognitive illusions and (b) wary of the political implications of social engineers who want to use cognitive biases to “nudge” people toward behavior they otherwise wouldn’t have done.

Within economics, I’d say that the behavioral/classical debate roughly follows left/right lines: on the left are the behaviorists who say that individuals and firms are irrational and thus we should not trust the judgment of the markets, instead we should regulate and protect people from their irrationality. On the right are the classicists who hold that people are rational when it comes to real economic decisions and thus any interference in the market, whether from governments or labor unions, will tend to make things worse.

The conservative position has some difficulties when dealing with customs and culture and roles and various non-governmental constraints on economic behavior: from one sort of conservative perspective these are unnecessary restrictions on the economy, silly traditions that rule-breaking entrepreneurs will shatter; from another conservative perspective, these traditions represent collective wisdom and we should be wary of reformers who try to start anew without recognizing that traditions are traditions for a good reason. But for now I will set all this aside and focus on the question of behavioral economics.

Step aside from economics for a moment, though, and things look a little different. Instead of thinking of “heuristics and biases” or “behavioral economics” in opposition to simplistic models of rationality (I’ve said it before and I’ll say it again; I see no reason why a long-discredited psychology model from the 1930s and 1940s should be taken as any sort of starting point for understanding human decision making; utility theory is, at best, one framework for such modeling), and put this work in a more general context of disparagement of human decision making.

To put it another way, think about “behavioral economics” not so much as “economics” but as “behavioral.” From a psychology point of view, behaviorism is a nearly century-old theory that was in many ways superseded by cognitive psychology. And, in many ways, “behavioral economics” is a sort of counter-revolution: it’s full of tropes under which people are doing things for irrational reasons, in which actions speak louder than words etc.

The full story here is complicated but one reason I think these ideas are popular in neoclassical economics is that they are, in some sense, anti-democratic. If people’s votes are determined based on the time of the menstrual cycle or on the outcomes of college football games, then elections are pretty silly, no? Which is an implicit argument in favor of lower taxes and more power for business, as compared to government (or, for that matter, unions).

This becomes particularly clear when we look at work along these lines in political science. If, for example, subliminal smiley faces have big effects on political attitudes, then this should cause us to think twice about how seriously to take such attitudes, no? Or if men’s views on economic redistribution are in large part determined by physical strength, or if women’s vote preferences are in large part determined by what time of the month it is, or if both sexes’ choice to associate with co-partisans is in large part determined by how they smell, then this calls into question a traditional civics-class view of the will of the people.

Luckily (or, perhaps, depending on your view, unluckily), the evidence for the empirical claims in the above paragraphs ranges from weak to nonexistent.

But my point is that there is a wave of research, coming from different directions, but all basically saying that our political attitudes are shallow and easily manipulated and thus, implicitly, not to be trusted. I don’t find this evidence convincing and, beyond this, I’m troubled by the eagerness some people seem to show to grab on to such claims, with their ultimately anti-democratic implications.

Let’s be clear here, though: I do have a dog in this fight, as the saying goes. In 1993, Gary King and I published an influential paper claiming that wide swings in the polls, swings that had often been taken as evidence of the capriciousness of voters or of their easily-manipulated nature, could be reinterpreted as evidence in favor of voters moving to their “enlightened preferences.” And then, more recently, David Rothschild, Sharad Goel, Doug Rivers and I updated this argument by providing evidence that some poll swings can be mostly explained by differential nonresponse without any large attitude changes. I’ve published work (with Aaron Edlin and Noah Kaplan) arguing why voting can be rational. And I’ve worked with Jeff Lax and Justin Phillips on their series of papers on the responsiveness of state legislators to state-level opinion. In my research I’ve been strongly committed, in many different ways, to the model in which voter preferences and attitudes should be taken seriously. So it would be fair enough to read my resistance to voters-are-influenced-by-irrelevant-stimuli arguments in that context. I’m providing you with my perspective, but I recognize that other perspectives are out there.

What I’m getting at is that I see a common thread in a lot of the counterintuitive, tabloid, Psychological-Science-type work out there, and that thread is a dismissal of human rationality and even human agency in the political (and, to some extent, the economic) arena. Here I’m speaking of “rationality” not in the limited sense of utility maximization but in the more general sense of thoughtful, purposeful decision making.

In the “Psychological Science” world, voters’ attitudes are determined by upper-body strength and the time of the month, their attitudes on important issues are influenced by meaningless subliminal stimuli, and their elections turn on the outcomes of late-October football games, and they flub any decisions involving uncertainty. Throw the words “Florida” and “bingo” at them and they walk slower, without even realizing why, they’re influenced by stereotype threat even without realizing it, and even their choice of clothing is not under their conscious control. Put it all together and you get a pre-cognitive conception of the citizen: not a man or woman who weighs the evidence, forms political views, and makes economic and political decisions, but a creature who is continually pushed to and fro by influences of which he or she is not even aware, an unstable product of hormones and the manipulators of political and social marketers, a sort of particle in the water being jostled by invisible Brownian forces.

Let me repeat that the evidence for many of these claims is weak, indeed I have the feeling that a lot of people want to believe in these things so they grab on to whatever “p less than .05” comparisons they find, and take them as representative of the general population, as scientific truth. On the other hand, I perhaps am coming from the opposite direction.

What I’m getting at is that there’s a political theme here, and also a scientific theme: I see a lot (although not all!) of this “behavioral” work as being behaviorist in the sense of being faithful to a pre-cognitive, and pre-modern conception of psychology.

The cognitive-psychology perspective, as I see it, is that we are thinking beings, and to the extent that we are influenced in irrational ways (whether by hormones, or subliminal marketing, or whatever), we mediate these influences through our thought processes. One reason I found the work of Cengiz Erisen so interesting (even while I disagreed with Larry Bartels’s more dramatic claims for the importance of that work) is that Erisen was not just treating his subliminal stimulus as a black box but rather was investigating how our conscious reasoning process might mediate the effects of a non-rational stimulus. In that particular case, the stimulus had no consistent effect on attitudes but I like the general approach of the study.


  1. Kyle C says:

    “to the extent that we are influenced in irrational ways (whether by hormones, or subliminal marketing, or whatever), we mediate these influences through our thought processes”

    I submit that this is why experiments on classroom convenience samples (including Kahneman’s famous endowment effect experiment with the mugs) should be treated as suggestive and exploratory at best, in terms of their generalizability to adult populations. In real life we know that college freshmen and sophomores are only beginning to be able to mediate their irrational urges through thought processes.

  2. JH says:

    “Wary of the political implications of social engineers who want to use cognitive biases to “nudge” people toward behavior they otherwise wouldn’t have done.”

    This seems to capture Gilles Saint Paul’s perspective, who argues that we shouldn’t confuse utility theory with the precept that individuals are unitary, rational, and responsible for their decisions.

  3. Dale Lehman says:

    I’ve been interested in these contrasting views for awhile – my background is as an economist. I think progress in reconciling these views is thwarted by a desire to have normative theories (which I do not share). Thus, it is important to believe people are rational utility maximizers if you want to conclude that free markets are best: or you believe in irrationality if you want to conclude that the government must intervene to prevent great harms. Personally, both rational and irrational behavior must both be true – to varying extents and under varying circumstances. In economics, I think you usually get further in explaining people’s actual behavior if you stick to the rationality paradigm – the clear examples of irrational behavior are typically more the exception than the rule.

    However, I think you go too far if you believe that underlying beliefs win out over short-term evidence of irrationality in experimental conditions. For example, I’m sure we can demonstrate that people’s views will change if we present an irrelevant image to which they respond strongly before we elicit their views. I would like to believe that given time and repeated behaviors, such short term influences don’t last – ultimately, the person’s “real” preferences will be felt. But I guess I come down on the Kahneman side of the psychology here – system 1 is much stronger than that. Rational thought can, but only with great effort, overcome the workings of system 2.

    If the empirical reality is that both sides of the debate have merit and neither can be ignored, then the only thing damaged would be desires to make normative statements, such as markets know best, or gov’t protection is required. If we can live with the grayness that both rationality and irrationality coexist – and some times under some circumstances one or the other may be stronger – then we might make more progress towards understanding human behavior (whether it be in markets, politics, consumer purchasing behavior, etc.). We would have to give up the (incorrect, in my view) illusion of certainty required for clear policy recommendations.

    As I write this, it sounds a lot like the debate over hypothesis testing, confidence intervals, etc. It is the desire to be decisive (as a result of a statistical study) that causes the problem – the evidence is always gray, and either gets more or less gray as a result of any particular study. Sounds almost Bayesian (to someone that was not trained that way).

    • Martha says:

      +1 to Dale’s last two paragraphs.

    • Darrin says:

      I think Dale has some excellent points. One of them (implicitly?) is that this post invests in a very black-and-white view of much of this “debate.” Gelman repeatedly says that the evidence for the “Psychological Science” claims is weak to nonexistent, phrasing that implies only the possibility of existence or nonexistence of the effects–not allowing the possibility of a range of effect sizes (IMO, the evidence is that many of the effects he dismisses are there, but so weak they might not have much effect in real-world political decision making). More troubling, but on the same lines, is his use of the name of the APS flagship journal as a clear outgroup label (with all the implications of homogeneity in outgroups, etc.).

      Finally, Gelman seems to have a fundamental misunderstanding of how “discredited” behaviorism is (or isn’t). It’s not just a theory from the 30s and 40s (with implications that what is old is bad); it’s a currently-viable meta-theory, descended from the original theory, with effects that are among the most consistent and replicable in all of the behavioral sciences. The “hard” Skinnerian version of behaviorism (i.e., “There’s nothing inside the black box worth worrying about”) was debunked very early on and repeatedly thereafter, but the principles of classical and operant conditioning are so robust and replicable that they might be called scientific laws rather than theories. There is a lively set of subdomains of research and practical application enjoying essentially uninterrupted scientific success for nearly three-quarters of a century, and many (most? all?) fields in psychology simply incorporate established behavioral principles as a strong thread in research and application. To ignore all that in order to have a an extra bullet point in one’s dismissal of the “other side” of the behavioral economics “debate” is somewhat silly.

      • Andrew says:


        As you say, Psychological Science is the flagship journal of its society and, as such, represents the in-group, not the out-group. That’s what’s so disturbing. When joke research is published in joke journals, it’s just a joke. When joke research is published in serious journals, that’s the problem.

      • Andrew says:


        To elaborate slightly, I agree that the topics that these researchers are studying are potentially interesting. The problem is that their measurements are too noisy for them to learn anything from their empirical studies. If they really want to learn about these effects empirically, they’re gonna have to up their game and put more effort into careful data collection. Otherwise they’re just coming up with a series of noisy numbers, it’s really like the classic problem of reading tea leaves, where every pattern tells a story.

  4. Popeye says:

    What I’m getting at is that I see a common thread in a lot of the counterintuitive, tabloid, Psychological-Science-type work out there, and that thread is a dismissal of human rationality and even human agency in the political (and, to some extent, the economic) arena.

    I always felt the common thread was reductionism, which feels “scientific” and which may be appealing if you aren’t in the group being “reduced” (or threatening/anti-humanistic if you are in that group). Human beings are complicated social/biological/chemical/psychological/rational/material creatures, and there’s something nice about the idea that science can cut through all the mess and come up with the clear underlying causes of behavior and events. Human beings also deceive themselves, so there’s something compelling/disturbing about the idea that people choose their spouses based on the first letters of names, the time of the month, the color of their shirt, how they smell, and whether they were implicitly primed for something while running across a shaky bridge.

    Alternately you can try to explain racism or unemployment or gender stereotypes as arising from natural self-interested economic rationality; the common thread isn’t respect or disrespect for human rationality, but rather the cargo cult science of “explaining” the world by reducing phenomena to the next level down.

  5. Alex says:

    I won’t claim to know a whole lot about behavioral economics or voting, but I am a cognitive psychologist. A colleague of mine did some interesting work on voting, particularly with people with amnesia. His thesis is here for those interested: . The short version is that people who have no memory of candidate’s positions still vote pretty similarly to normal people.

    I’m of two minds as to what this suggests about voting. On the one hand, it seems rational to vote for the person you like after having heard his positions, even if you can’t remember what they were. The fact that you like the person can be taken as evidence that you agree with him, so that would still be a rational basis for choosing a candidate. On the other hand, if you can’t even remember that you’ve seen that person before (let alone what he said), and given other work on how people’s preferences are affected by all sorts of non-relevant information (attractiveness, height, gender, etc.), it strikes me as disheartening that people with amnesia vote pretty much the same way everyone else does.

  6. Z says:

    I agree that there’s a similar flavor to the fallible humans lines of research in economics and political science, but I think they are subtly different. Classical economics requires that actors be *rational*. Classical political science just requires that voters have relatively stable preferences that are robust to minor fluctuations in their external or internal environments. People’s moral values need not be rational and definitely play into their political preferences, and I think most people agree that this is as democracy should be. So I think it’s OK to be generally approving of Kahnemann type behavioral economics and generally disapproving of voters don’t have real preferences type political science.

    Also, I agree that the voters-are-stupid research is politically undemocratic, but not necessarily conservative. I think that type of research is very appealing to lots of liberals when conservatives are in power as a way of explaining away their electoral success. Didn’t you spend a lot of time debunking “What’s the Matter With Kansas?” because it took up the voters-are-stupid mantle from a liberal perspective?

  7. ConvexPhil says:

    I’m not sure it’s best to think of cognitive psychology as having superseded behavioral; the two heavily coexist today, with still-active research traditions surrounding operant/classical conditioning models like Herrnstein’s Matching Law and the Rescorla-Wagner model or mixed techniques like cognitive-behavioral therapies. If I remember correctly, David Hothersall (author of “History of Psychology,” the text my undergrad history-of-psych class used: argues that Kuhnian revolution doesn’t really fit the history of psychology very well, with old traditions more often coming to coexist with new ones than outright replacing the old, unlike the physical sciences.

    Behavioral econ has a lot stuffed into it other than the Kahneman-and-Tversky heuristics and biases stuff, too. Research on social preferences, Shiller’s behavioral finance models with entirely random agents, work on constructed preferences, and the various stuff subclassified under neuroeconomics all seem very different from anything on framing effects or cognitive biases, for example. It is interesting, though, that many popular behavioral models—like Camerer et al’s experienced-weighted attraction—look and feel a lot more like behaviorist psych than cognitive psych, with relatively sparse models of cognitive processes but a lot of modeling of units-of-reinforcement/”utils.”

    I totally agree about the need to replace utility theory, though. We’ve got descriptively better models that still use representing functions, so it’s not like the theory has to be *totally* reworked. Models like prospect theory or rank-dependent utility still by-and-large take a representing function for preferences as their foundation; they don’t respect all the same nice properties as utility functions, but the rework they call for shouldn’t be as radical as, say, rejecting transitivity of preferences, or something equally destructive.

    • ConvexPhil says:

      Also, have you read any of Nobel laureate Elinor Ostrom’s institutionalist work on the tragedy of the commons? Your comments on conservative reactions to culture reminded me of it; she makes some interesting mixed theory-empirical arguments that cultural traditions often substitute effectively resolve what are traditionally thought of as coordination failures to be solved by government policy.

  8. Paul says:

    I was always uncomfortable being classified as a “behavioral” in Finance, mostly because I found preferences such as “loss aversion” an unsatisfactory or perhaps simplistic explanation of observed behavior. But there is certainly also a political component to it: true, especially in a very right-wing field such as Finance, the behavioral literature is considered more to left, at the same time I never liked the “look at how stupid those investors behave”-aspect of Behavioral Finance. It always seemed classist to me.
    I guess I always liked Popper in his discussion of the “The Rationality Principle” (1967) best:
    On the one hand “Rationality as a personal attitude is the the attitude of readiness to correct one’s beliefs” (something I believe most people are able to do). On the other, there is the “rationality principle”, i.e. a “minimum principle […] which animates all, or almost all, our explanatory situational models”.

  9. ZC says:

    >Within economics, I’d say that the behavioral/classical debate roughly follows left/right lines: on the left are the behaviorists who say that individuals and firms are irrational and thus we should not trust the judgment of the markets, instead we should regulate and protect people from their irrationality.

    I’m sorry, Andrew, but this is very wrong, at least in my experience as an economist surrounded by economists of both these stripes.

    • Dale Lehman says:

      From my experience of an economist surrounded by economists, I do not agree (with zc). I think it is a fair characterization in general.

      • Jonathan (another one) says:

        As another economist surrounded by economists, I agree with Dale. It’s even more pronounced in health care markets, in which I think the standard left opinion is that no individual consumer is even minimally competent to assess the adequacy or price/quality tradeoff of care received, while the standard right opinion is that virtually all cognitive errors in health care consumption are due to insulation from price/quality tradeoffs. (Both views are strawmen, I admit, but are defended regularly, and lead to very different fixes to health care market issues.)

  10. GabbyD says:

    “If people’s votes are determined based on the time of the menstrual cycle or on the outcomes of college football games, then elections are pretty silly, no? Which is an implicit argument in favor of lower taxes and more power for business, as compared to government (or, for that matter, unions).”

    why is that? there is nothing in these studies that say people in business (consumers or producers) are more rational than people in govt (voters or other actors).

    meaning if its silly in politics, its silly in consumer behavior as well.

    • Andrew says:


      The argument is that, when acting in the economy, people are more rational because they have “skin in the game”: bad decisions can cost them money so they are motivated to be rational, whereas in politics the probability of their vote making a difference is so low that the connection to personal outcomes is much less direct. So, yes, I do think that studies about the effects of shark attacks and menstrual cycles on voting do, to some extent, discredit the political sphere relative to the economic sphere.

      • GabbyD says:

        well, if “skin in the game” is the metric, then many consumer decisions also fit that metric. there are lots of papers that say people are irrational for instance in intertemporal decisions; their skin in the game, so to speak, at any given moment of time is small. They find this in savings randomized experiments too.

        • I read some paper about hyperbolic discounting as a natural consequence of some kind of an uncertain time horizon (ie. you don’t know how long you’re going to live, or more generally when or whether you will actually receive future payouts). Just mentioning it because that view, that “inconsistencies” in time preference are actually a consequence of uncertainty was kind of an interesting idea to me.

  11. Manoel Galdino says:

    Great post.

    Could anyone please point me to some literature review paper on the themes mentioned by Andrew? I read Kanheman’s book “Thinking Fast and Slow” and there are a lot of refereces there, but maybe there are a few “lit review” papers that will make it easier for me to imerse on the field?

    My background is in political science, so I’m more familiar with the literature on this area.

    Thanks in advance.

  12. Ryan says:

    Is there any research on the impact monthly cycles/football scores/etc have on voter turnout instead of the impact on voter “attitudes”? I could more easily accept that these unusual influences impact voter turnout than voter attitudes.

  13. Jose M. Vidal-Sanz says:

    Kanheman points out towards the existence of two mental systems, the racional and the intuitive-fast decision making approach. One might think of this as the fundament for applied economics: classical economics (rational) plus some noise (fast decisions), end then we would end up in econometrics structural modeling. So, economics is not out of fundament due to behavioral studies. On the contrary, behavioral studies offer new opportunities to quantitative economists to re formulate the specification of classical models, for example the existence of anchor points and asymmetries between loss and win outcomes may be used to improve the rational expectations model to build a better utility model. Analysis of intuitive decisions may also reduce the degree of errors in current models including heuristic rules. In this sense, economics can benefit from the fusion of knowledge from both streams of research.

  14. David Condon says:

    That’s not really behaviorist philosophy. A good description of behaviorism can be found in Chapter 1 of BF Skinner’s About Behaviorism:

    “The position can be stated as follows: what is felt or introspectively observed is not some nonphysical world of consciousness, mind, or mental life but the observer’s own body. This does not mean, as I shall show later, that introspection is a kind of physiological research, nor does it mean (and this is the heart of the argument) that what are felt or introspectively observed are the causes of behavior. An organism behaves as it does because of its current structure, but most of this is out of reach of introspection. At the moment we must content ourselves, as the methodological behaviorist insists, with a person’s genetic and environmental histories. What are introspectively observed are certain collateral products of those histories.”

    As ConvexPhil stated, it’s mostly more of a spectrum between cognitivist and behaviorist thinking than strictly one or the other. The debate is over how to deal with the human mind. Cognitivists take an inside-out approach; they try to understand the brain, and then show how this affects decisions. Behaviorists work outside-in; they try to understand the environment then show how this affects decisions. It’s very different from the debates in economics. Cognitive biases and processes are an outgrowth of cognitivism. It’s a bit of a misnomer in that neoclassical economics is mostly behaviorist to begin with and behavioral economics adds in a lot of cognitivist elements.

  15. jonathan says:

    What is good advertising? To change the old line a bit: is it that half the money spent on advertising is wasted or is it the other half of the money spent on advertising that is wasted? I bring this up because ads are ways of bringing attention to a product, service or company with the aim of encouraging buying that stuff, that service from them. I don’t think there’s good evidence that advertising can make people choose what they don’t want. Some advertising appears to nudge. The advertising we remember is either clever on its own terms – but of dubious actual value – or speaks to consumer needs in an appealing or fresh-seeming way. In other words, despite all the money, we don’t have a solid handle on either side of the rationality question: the impact of ads on buying decisions and the reason advertisers (irrationally?) put out this stuff.

    I think the topic, therefore, is less political overall though it is political in economic modeling. It isn’t political because it addresses deep uncertainty in our shared and individual realities and it is political because we need some way to shape our imperfect worldviews and thus our imperfect models. In that regard, I had some amazing conversations about “ricardian equivalence” – which I don’t capitalize because it wasn’t David Ricardo’s work; I had it cited to me often as though it’s a law and I would respond something like “it’s a first order assumption in a model”. And then I’d try to explain that’s all it is, just an imperfect assumption in a particular kind of imperfect model that may be useful in some cases but which you’re stupidly generalizing like it’s an invariant law of nature. I put those words at the end because that is part of the issue: we want to generalize, are made to generalize whether wrong or right, for good or bad, for stupid or genius, but not much generalization is actually invariant, meaning of course true in any frame or perspective and not just yours. I know I’ve mentioned before that if you spend time in NH, you think the world is all white people, that everyone has a weapon (and a truck) and that everyone has to scramble for money but that there’s enough money to go around … and that’s how you generalize, as though your frame is true in all frames (whether accelerated by city life or not!).

    Regarding irrationality research, etc., I find the results often interesting but the tendency to generalize beyond the clear meaning of the work annoying.

  16. David says:

    It seems to me this post is much more a critique of pop psychology (and especially social psychology) than of behavioral economics. The terms are confounded in the popular press, but there is a massive difference between a Psych-Science paper and a paper in QJE. That may also explain the emphasis on lab experiments, rather than the numerous field experiments building on findings from the lab.

    Maybe the most successful “product” to come out of behavioral economics is automatic enrollment in 401(k) plans, in some instances with automatically increasing contributions. It seems to me that retirement savings are a pretty high-stakes environment and someone who evaluates job offers from firms should incorporate benefits like matching contributions to retirement plans. Yet, a change of default (i.e. whether one is by default contributing or not contributing to the retirement plan) can drive participation rates from 40% to over 80%. Similarly, the default contribution rate significantly shifts how much people save. (Madrian and Shea’s 2001 QJE paper started it and there have been numerous replications and little tweaks here and there. It’s as robust a finding as we’re going to get.)

    John List and colleagues now run a school that takes findings from behavioral economics and decision research and applies them to education. Based on the first working papers coming out of this, it seems like a pretty successful endeavour. For example, they have a financial incentive study that eliminates the racial achievement gap — the treatment works for minority students (which make up a substantial part of the students in the low-income district), but not for more affluent whites. Similarly, they have bonuses for teachers that work better when paid upfront on the condition that they have to be repaid if students don’t meet certain goals (putting into practice loss aversion). Is education high stakes?

    “While some economists have embraced so-called “behavioral” ideas to explain imperfect markets, other economists are (a) skeptical about the relevance to real-world high-stakes behavior of laboratory findings on cognitive illusions and (b) wary of the political implications of social engineers who want to use cognitive biases to “nudge” people toward behavior they otherwise wouldn’t have done.”

    (a) That’s an empirical question. Lab studies have been replicated in developing countries with adults and high stakes (a month’s wage) and at most, performance in cognitive tasks has been shown to decrease with extremely high stakes (a choking effect). That is on top of the aforementioned field studies. Similarly, many studies are now run online with participants who are, at least on observables, fairly representative of the US population. It’s worth noting that no classical economic model suggests “degrees of rationality,” so there’s no theoretical reason to think that a model of time discounting looks different for students or adults. They may well have different discount rates, but that doesn’t resolve any of the original paradoxes.

    (b) That may well be true, but is immaterial to the validity of a model.

  17. Øystein says:

    Interesting argument, but I do not think economics supports it:

    1. In economics the behavioral literature is primarily about decisions in the economic sphere, not the political, so I do not see it contributing to “discredit the political sphere relative to the economic”.

    2. “one reason I think these ideas are popular in neoclassical economics is that they are, in some sense, anti-democratic”. But they are not popular in neoclassical economics – as you wrote earlier, “Within economics, I’d say that the behavioral/classical debate roughly follows left/right lines”.

  18. marcel says:

    “so they grab on to whatever “p less than .05″ comparisons they find”

    I prefer this myself.

  19. marcel says:

    (Oh, and don’t forget to let your cursor hover: by “this” I meant the subtext.)

  20. […] 5) Cognitive vs. behavioral in psychology, economics, and political science […]

  21. Alan Zaslavsky says:

    I remember a time in the early ’90s when behavioral econ was beginning to gain a foothold at my university. It seemed for a moment like this had some radical potential as a theory critical of the dominant “rational expectations” paradigm in economics. Over time, I have moved toward a narrower view of its value, although I still believe that behavioral econ provides material for a critique of the normative claims of neoclassical econ based on rational expectations, which is not a minor potential contribution. The fundamental limitation of behavioral econ may be that it is largely atomistic, concerned with experimenting on behavior of individuals regarded as units (whether in the lab or in the field). This is a face of the reductionism that “Popeye” commented on. Within that framework it is hard to look at questions of social change or transformation.

    Ultimately the value of intellectual work depends as much or more on the questions that are addressed as on the tools that are used to address it. If women’s voting patterns change over the menstrual cycle, what bearing does this have on the functioning of or prospects for democracy? Or for another example mentioned in an earlier comment, advertising might be regarded as a large-scale application of behavioral economics, but what value does it contribute? This is not to say that there is no socially valuable application of market research and behavioral ‘nudging’, but I would like to know whether behavioral econ has addressed how systemic change can take place. Can rational expectations and behavioral econ duke it out whether the masters of Wall Street were subject to irrational exuberance fostered by testosterone levels elevated by rising real estate markets, or simply responding to incentives to manipulate markets and get richer quicker?

  22. Erin Jonaitis says:

    I think you might be missing a contributing factor. Like all academic disciplines, I think, “psychology” is both a set of ideas for how to think about the world, and a community of people who have interests they’re trying to advance. One of psychology’s challenges — and as a social science, political science may face this too — is that a lot of findings can be classified coarsely into two boxes: “counterintuitive” or “your grandparents knew it.” I suspect it is harder to get money to study the stuff in the second box, in part because it tends to invite derisive snorts from non-scientists, including most of Congress, who ultimately signs the checks. This makes me wonder if there is some quasi-evolutionary pressure that favors lighter scrutiny on stuff in the first box. And I suspect findings of the form “entity X influences behavior Y without subjective awareness or intent” are more likely to belong in the first box.

    I’d be interested to hear your thoughts on this counterintuitive-vs-grandparents issue, as you’re someone who has benefited from NSF social science funding during a period (maybe multiple periods) when that specific budget line has gotten a lot of evil eyes from the right wing. Has that kind of thing come up in your own reviews? on panels you’ve served on? Am I delusional to think this is a thing?

  23. Thank you for these reflections. I accept that the evidence is weak overall and I too am unsure that we’ll get there with an approach that uses ‘p less than 0.05’ comparisons.

    One of the things I’m curious about is whether or not we are or are not rational – what is our goal? If our objective (which I hope would be shared) is to develop the intuition of human beings to tame the complexity around us, then doesn’t that mean ‘learning’? And if our emphasis is then on learning, we might be more oriented to develop systems around us that help us to learn instead of developing systems around us that thwart or stop us from thinking.

    • Martha says:

      “If our objective (which I hope would be shared) is to develop the intuition of human beings to tame the complexity around us”

      I’d say that one (presumably not the only) goal is to develop human beings’ willingness and ability to see, accept, and take into account the complexity around us — not to “tame” it.

  24. zbicyclist says:

    Andrew, I see this blog post is currently a featured article on “The Browser” as of Feb 1, 2015.

    The Browser is a subscription site that posts links to a small number of articles each day.

  25. Urstoff says:

    The political implications of behavioral economics (and even Psychological Science “psychology”) are whatever you want them to be. Are people irrational? Then they need a government to make them more rational. Are people irrational? Then the government, made of people, will itself be irrational and the damage that it can do must be limited. Are people subject to stereotype threat and unconcious biases? Then discrimination is probably pervasive and need the government to intervene. Ad infinitum.

  26. knowsie Farquar says:

    To my mind people are generally attempt a rational process of discerning the truth or otherwise of arguments using, primarily, ones own memory of previous arguments and personal histories. Given the known vulnerabilities of human memory, even if the processing of this information is conducted according to purely rational methods, the outcome of the decision can only be as good as the validity of the memories and is therefore not certain. Didn’t Stiglitz et al show this in the work that won them the Nobel prize? Perhaps I’m wrong, my memory could have played a trick on me. Either way, even if we are well evolved decision making machines that can effectively mediate the conflicting non-rational influences that seek to swing us one way or the other, one thing is for sure: the advertising industry is unlikely to start selling us toothpaste by telling us that basically they are all the same product so we might as well make the rational decision and by the cheapest one.

  27. […] Gelman seizes on what bothers me about so much current political science, particularly studies that focus on […]

  28. Jonathan says:

    If you want me to believe in political rationality provide a rational explanation for global warming denial.

  29. James says:

    This is a fascinating post. I tend to agree with Dale Lehman’s comments above in that it seems when strongly competing theories are able to compile loads of evidence in support of their position that both must have some validity, but that reality is more complicated than either side of the argument currently acknowledges. When thinking about this topic, I often recall a lecture by John Capaldi during a Psychology of Learning course in which he made a comment about how a stimulus-organism-response theoretical framework can be mapped onto the far more complex human brain system. His position was that thoughts be treated as both stimuli and responses. The implication of this is why the application of both classical and instrumental conditioning protocols in human systems — particularly by those who do not understand these limitations — is far less effective than often thought. It is because external stimuli can become internal stimuli for the desired outcome (another thought or behavior) or it can become a stimuli to an undesired outcome. When one places this in the complex environment of the real world and is applied to humans possessing differing temperaments, beliefs, knowledge, memories, learning histories, etc. — how does one control the stimuli and how does one even know what the outcome is and whether the observed behavior is a direct result of a supposed stimulus? For that matter, how would one even do so in an extremely constrained environment?

    Given this level of complexity and the amount of uncertainty in human cognition and behavior, it seems that the most likely outcome of such an application is Seligman’s model of ‘learned helplessness’.

    As an aside, I believe Capaldi’s comment has stuck with me so clearly and for so long because it so thoroughly challenged my ideas about the human mind at the time.

  30. […] Cognitive vs. behavioral in psychology, economics, and political science. […]

  31. matus says:


    you should read
    The role of causality in judgment under uncertainty. Krynski, Tevye R.; Tenenbaum, Joshua B., Journal of Experimental Psychology: General, Vol 136(3), Aug 2007, 430-450.

    From abstract:

    “Leading accounts of judgment under uncertainty evaluate performance within purely statistical frameworks, holding people to the standards of classical Bayesian (A. Tversky & D. Kahneman, 1974) or frequentist (G. Gigerenzer & U. Hoffrage, 1995) norms. The authors argue that these frameworks have limited ability to explain the success and flexibility of people’s real-world judgments and propose an alternative normative framework based on Bayesian inferences over causal models. Deviations from traditional norms of judgment, such as base-rate neglect, may then be explained in terms of a mismatch between the statistics given to people and the causal models they intuitively construct to support probabilistic reasoning.”

  32. Paul Herr says:

    I’m glad that I stumbled across this post by Andrew Gelman. It is timely because I just finished writing a paper for the Incentive Research Foundation on intrinsic, intangible, non-monetary rewards (rewards emanating from Kahneman’s intuitive “fast thinking system”). I reviewed the latest research in behavioral economics (BE), neuroeconomics (NE) and neuroscience (NS) before writing this paper so it is pretty up-to-date.

    Andrew is being rather glib in his post and he speaks disparagingly about Kahneman’s fast, intuitive (emotional) system and the “irrational” biases it creates. Andrew is the champion of cool, rational, intentional decision making which Kahneman calls the “slow thinking system.” According to Andrew, BE, and the research supporting it, is essentially a scam supported by dubious statistics.

    He doesn’t mention, however, that Kahneman’s won a Nobel Prize for his analysis of the fast-thinking system he is backed up by the latest research in neuroeconomics and neuroscience. The NE and NS research concludes that human emotions have been vastly underappreciated, and underestimated in all aspects of human behavior, including economic decision making.

    Simply stated, the fast, emotional system is actually very smart in most situations and tells us what we need to do to survive. The slow-thinking system, you and me, are in charge of satisfying these emotional needs. If we ignore our emotional feelings, by the way, we are dead–extinct. Lobotomy patients had their emotional feelings turned off, and it did NOT turn them into super efficient MR. Spocks. It make them lazy and inert and ruined their ability to make quality decisions.

    Hunger is part of the intuitive, fast-thinking system that we are all familiar with, so I will use it to illustrate some basic points. When we are hungry, the emotional mind’s message is, “it’s time to eat, dummy.” The rational mind can either take this hint, and find a path to food, or resist, say, to complete an important report. The pain signal then ramps up and says, “Ok, now it is REALLY time to eat.” The rational mind can be stubborn and resist the pull of emotional “gravity” through sheer willpower, but it will take more and more mental effort to do so. Eventually, the rational mind will do as it is told and find a path to food.

    The rational mind, in other words, is the emotional mind’s errand boy, and emotions are simply “proxies” for our core survival needs. In other words, emotions rule in business and in life, which is what BE experiments corroborate. Far from being irrelevant, emotional feelings lie at the core of customer and employee satisfaction, the core of the brand impression, the core of economic utility, and the core of every single financial transaction we make.

    The true economy is based not on monetary transactions, but on the emotional transactions percolating underneath. For example, if you give me a product or service that makes me feel better, more successful, or reduces my pain or effort, I will pay you for it. You can then take my money and buy something that makes you feel better, and that satisfies your emotional needs. This may be shocking, Andrew, but rewarding feelings emanating from the fast-thinking emotional system provide the true currency upon which the capitalist system is built. This doesn’t sound “irrational” to me.

    The time for bashing human emotions, and Kahneman’s fast-thinking intuitive system, is over. Andrew is on the wrong side of the scientific fence on this issue and should reevaluate his position. If anyone would like to see what the future might look like, I’d be happy to send you my paper when it’s finalized in a few weeks (just send an email requesting a copy to Paul at

    • Andrew says:


      You write that I speak disparagingly of Kahneman’s system, but here’s what I actually wrote above: “it’s my impression that in recent years there’s been a bit of a convergence: for Kahneman the glass is half-empty and for Gigerenzer the glass is half-full, but whether you’re talking about ‘heuristics and biases’ or ‘fast and frugal decision making,’ there’s been a focus on understanding how our brains use contextual cues to decide how to solve a problem.” I don’t think this is disparaging at all!

      I do disagree with Kahneman that “You have no choice but to accept that the major conclusions of these [social priming] studies are true.” But that doesn’t mean I don’t respect Kahneman’s work, it just means that I think Kahneman made a mistake in that case, that he was subject to the same sorts of cognitive biases that affect us all.

      So, in short: yes on the idea that we reason in different ways in different settings, yes on the idea that fast and frugal heuristics can be effective, no on many specific highly publicized claims such as social priming, power pose, ovulation and voting, fat arms and political attitudes, etc etc.

  33. Paul Herr says:

    Hi Andrew,

    Sorry that I over-reacted, but I am very sensitive to arguments were rational = good, and emotional = bad/primitive. Our motivational system evolved along side the rational one. They are both amazingly sophisticated and we need both to survive.

    The emotional system is severely under-appreciated because it is built to be stealthy and to guide our behavior without being the focus of attention. I often use a slide showing an actor hanging from clear, mono-filament strings. The caption reads, “We are puppets dancing on emotional strings, but we don’t know we are puppets, and we cant see the strings.” This slide, I believe, illustrates paradox of emotional regulation and why it takes clever experiments by behavioral economists, and others, to ferret out its secrets.

    Warm Regards,


    P.S. There was an error in the email address I gave with my last comment. For a copy of my research paper on intrinsic motivation, email me at

Leave a Reply