The biophysical limits of cognitive computation

In this diavlog with Glenn Loury the behavioral economist Sendhil Mullainathan recounts the results of an experiment.

- If given the option of paying $100 for an item vs. $80 for an item, but in the second case having to go across town for the item, respondents choose $80 and going across town

- If given the option of paying $1000 for an item vs. $980 for an item, but in the second case having to go across town for the item, respondents choose $1000 and not going across town

This the result of a heuristic bias whereby we seem to perform comparisons as percentages, and not the absolute value of currency. You save $20 in both cases, but the proportion is much smaller in the big ticket item. That's not too interesting, what is interesting is this: Mullainathan claims that poor respondents actually choose to go across town in both cases because to them the $20 is very tangible. In other words, in this case the poor behave as the rational actors which neoclassical economics is predicated upon, because they operate in a world of such cash scarcity that they perform numerical calculations habitually to estimate trade offs. Much of the rest of the discussion involves these sorts of insights (if you've ever been on vacation and start out with a fixed amount of money you're going to expend, I suspect you see how this works as you progress through the vacation and have less and less money you can spend).

Mullainathan as a behavioral economist emphasizes the importance of psychology in understanding human action, but listening to him I felt that the real truth here was more fundamental and neurological. In How We Decide Jonah Lehrer reports plenty of data which show that the utilization of the prefrontal cortex to engage in rational action which overrules impulse is strongly conditioned on how much stress we are under and energy we have on hand. In my discussion with Jake Young last year, Jake, a neuroscientist with a deep interest in economics. pointed out how that the fixation on relative calculations may simply be a function of the fact that general relative rules and heuristics are computationally less expensive than encoding specific absolute values.

It is an irony here that one of the problems with modern economic science is that the scholarship of scarcity does not account for the scarcity of resources on a deep neurological level. Rather, the rational actor's prefrontal cortex has an infinite amount of cognitive juice at its disposal, and choice itself is not taxing. But in reality, a utility function in regards to choice always has to take into account that the act of choice itself is subject to the rules of scarcity. The brain uses 25-30% of our calories, far more than in a typical organism.


More like this

Don't ever buy them. Ever. Each year, millions of people gladly pay an additional 10 to 50 percent of a product's original price to extend a warranty. These snap purchases help fuel a booming, $15 billion-a-year business and feed a lucrative profit stream for retailers that sell the warranties and…
Temporal discounting is our tendency to want things now rather than later. In order to encourage us to save money, banks have to offer us a reward in the form of an interest rate. In order to delay gratification, we have to be convinced that the reward in the future is going to be sufficiently…
I read Scibling Jonah Lehrer's How We Decide some time ago, but Moveable Type ate my half-finished review, and it's taken me until now to get back to it. You may have seen quite a few reviews elsewhere by now - Adam Kepecs reviewed it for Nature back in April, and to make a long story short, I…
I think one of the most important tests of behavioral economics will arrive in the next few years, as we attempt to persuade consumers to improve energy efficiency in the home. Just imagine if, instead of installing granite on every kitchen countertop, we'd instead spent that money on better window…

George Stigler emphasized cognitive costs even without understanding their neurological origin. Hayek was also interested in this, and spent his later career trying to unify economics/biology.

I'd be very interested in seeing what impact this has long-term on people who change income bracket. It would be interesting to see if well-off people from poor backgrounds are better at making these decisions or whether they adjust to reduce the cognitive cost issue.

This also interacts in an interesting way with claims that the poor are poor due to their own behavior. Yet, they are making a more careful monetary decision here. I don't know how to reconcile this sort of thing with the fact that the people who buy lottery tickets are almost overwhelmingly from the lower income brackets.


1) i've read it takes 4 months for people to get used to a higher level of income after getting a new job or a raise. in other words, they simply ramp up their baseline consumption which they don't perceive as discretionary. not quite you're talking about, but that's all i know.

2) how much does a lottery ticket cost anyway? googling doesn't yield anything super-quick because every state has a different lottery.

This the result of a heuristic bias whereby we seem to perform comparisons as percentages, and not the absolute value of currency.

That is possible, but I think your comment about the response of the poor might provide a solution: it may have less to do with an incorrect perceptions of percentages than it does built-in assumptions about the respondent's net worth at the time they could afford to make such a purchase. It would be interesting to scale up the amounts to $1,000,000 and $999,980 and see if even the poor don't respond differently at that level.

In conjunction with my previous comment, a very cynical interpretation could be that the poor are more likely to be impressed by a longer string of digits in a figure. The middle-class buyer may be operating more on the assumption of purchasing power while the poorer respondent is thinking both figures are impressive and the $100 and $1000 have a lot of zeros. This could be controlled by picking figures with the same number of digits.

Razib, I've seen the 4 month claim before but don't know anything else. About that.

Regarding lotteries, I don't know although it seems like the prices vary between about 50% to about $3 with most in the middle range. (Incidentally, I suspect that the lottery issue may be getting connected to the confiding issue that humans are just really bad at probability. Whereas the situation involving 980 v. 1000 doesn't involve probability at all).

Maybe people figured that the $1000 item was something as large as, say, a TV set, and it would be more costly or difficult to transport it from across town.

fyi, he specified a computer for the $1000 one. don't remember what it was for the cheap thing, but it might have been clothing.

As someone who has spent a couple of years calculating down to the last franc (1 Euro ~ 6.5 Franc), I find the results entirely plausible.

When you give out $20, some people will think "well, that's 2 CDs, or a couple of books." Others will think "well, that's all my food for the next week and beyond, and I might even buy a few nice meat bits with that". Estimating money in units of meals is something I have done a lot - not sure if there's any research on that.

It's quite understandable that better-off subjects will not allocate as much resources to calculate everything down to the last dollar. Not performing these calculations for every single purchase is perhaps the most tangible benefit of a stable income. Scarcity management is enforced by "cognitive laziness", and how much mental effort you can be bothered to muster depends on the relative value of the stakes - relative to your own situation, that is.

most readers of this weblog have been students with little money after tuition, fees & lodging, so i think most have some idea of always translating money into units of food mentally :-) or, more specifically, translating the money you have into calories per meal you buy/make....

I suspect it has at least something to do with the feeling of what constitutes 'a good deal' rather than about the value of the money itself: a larger percentage discount feels like 'beating the house' in some fashion, whereas obtaining a piddling percentage discount doesn't offer any emotional reward. OK, it's not 'rational' - but it's not nonsensical or stupid, either: our consumption behaviour is inevitably motivated at least in part by the pursuit of emotional rewards, and there's nothing odd about acting in accordance with that.

Another point: arguably, crossing town for a $20 discount is the 'irrational' act for almost anyone not on a low income, regardless of the percentages involved. What's $20, really? I spend amounts like that on unnecessary consumables all the time without batting an eyelid just because those things give me some limited emotional reward (choosing a 25-dollar bottle of wine, say, over a 5-dollar bottle of Chateau Rough from the discount bin). Going to any great effort for such a saving on its own just isn't worthwhile to me (and I'm not exactly in a high income bracket). The emotional reward of a high-percentage but low-value saving might just be worth it - but the benefit would be the emotional satisfaction, not the financial saving. (And it cuts both ways: I resent paying, say, 20 per cent more for a meal in one restaurant than the same meal would cost in a similar restaurant down the road, but it's because I feel cheated not because that buck or two really has any meaningful value to me.)

I've noticed that in my own comparative shopping behaviour I respond to *amounts saved* rather than to percentages only when that amount actually becomes 'meaningful' - when it starts to reach levels where I actually would think twice before spending so much. $20 bucks is below that threshold for me, though, and I expect for most people of my earning power or above.

Beyond poverty, there's a certain kind of person who fights for the best price and for the bargain on everything. It reminds me of something I saw the other day -- a rare book for $5,000 plus $7 shipping. The kind of person I'm thinking of, after bargaining down the $5,000, will ask for free shipping at the last moment.

I sold Christmas wreaths for an educational charity once, and some of the richest customers were the most demanding. Basically we were selling $10 wreaths for $20, but they had costed out the $10 extra in terms of tax writeoffs and good publicity (people seeing their names as a donor) and if they could get $10 publicity for $9 they would.

This kind of person is usually ethnically described, but it's situational -- underfinanced entrepreneurs fighting for tiny margins. After generations, it becomes a family trait even after the family has money. (For example, people in entertainment, sports, media, politics, etc. are often big tippers. Rich people from the business world usually aren't).

Ethnic groups accused of this are famously Jews, but I've also seen it said about Chinese outside China, Scotsmen, and New England Yankees in the US. It's a function of business specialization. (There are also big-money people who fight for every nickel, for example the ones who rescue mismanaged companies -- penny-pinching is one of the main things they bring to the company).

In my brother's coffee shop, there are some people who reach into the tip jar for change so they don't have to break a bill -- but the tip jar is there for you to throw your change into after you break a bill. And he says that the people who do that are always very well off and very well dressed.

By John Emerson (not verified) on 24 Mar 2010 #permalink

I'm thinking that the people choosing to drive across town for $20 in savings are the ones making the irrational decisions in each case. Transportation costs will eat into the $20. Also, one has to figure out what one's time is worth. That part is tricky, but generally the higher up the income ladder the more valuable the time. (Time might be scarce for other reasons, too.) If income is high enough it might be more effective to hire someone to do the shopping in any case. Then one must to figure out if it's worth the cost of the trip for the employee to drive across town.

Lehner also explains very well how the brain records patterns of punishments and rewards and then makes those available to the decision making process in the form of somatic markers (Damasio et al).

In that respect the brain is like a democracy - it may make mistakes (sub-optimal choices) yet those same "errors" provide the means for adjusting to a better strategy in the future - even including metabolic costs into the equation.

IMO it's not a question of percentages vs. absolutes as much an example of the brain using the method that has delivered the most reliable results in the past - yet always being sensitive to changes in the patterns of its environment and the relative success of its decisions in response to it - so it can adjust to new conditions.

It's like any control problem; the error in the outcome provides the signal that allows the system to reduce it.

By Ray in Seattle (not verified) on 24 Mar 2010 #permalink

"most readers of this weblog have been ..."

I can attest to the fact that some of us still are/still do.

Not sure I need to remind anyone but when you're in this situation, it also becomes a habit to calculate the costs of other goods within that specific framework - it's not just money you evaluate this way, it's other goods too (ie. 'one cheap book = four days worth of food').

I've noticed this type of behavior in myself. When my resources are relatively high, I become more concerned about relative value; when my resources are low (college), I tend to concentrate more on absolute value.

There used to be a name for the kind of behavior I described: "bourgeois". Traditionally the bourgeoisie were penny pinchers grubbing for the dollar, while the aristocracy was extravagent. The peasantry was as frugal as the bourgeois, but usually peasant communities are bound into gift networks and communal obligations which institutionalize a kind of generosity.

By John Emerson (not verified) on 24 Mar 2010 #permalink

This has absolutely nothing to do with the "biophysical limits of cognitive computation".

People spend $100 frequently and $1000 a lot less frequently. Which means that the proposition indirectly implied expended effort against saving $20 many times or against saving $20 once. Far from being limited, most arrive to the correct conclusion that in the long run saving 20% makes more sense than saving 1%.

Razib (#10), so true. I remember living (rather unhealthily) for about a month on a diet primarily of vitamin pills, ramen noodles, beans and rice and just praying I was getting enough protein. Glad those days are gone.

Taking off on what someone said above, perhaps when people are spending large amounts they feel their time is worth more, but when they're spending smaller amounts they feel more average.

A couple of times I've bought books I knew I owned because I didn't want to spend an hour or so digging it out of whichever box it was. But for < $10.

By John Emerson (not verified) on 24 Mar 2010 #permalink

Speaking of bargaining. From reading stories related by retail workers, it seems the more common method is to not go across town, but demand the $100 (or $1000) item be pricematched down to the $80 (or $980) that the store across town is selling it for.

Also I nearly typed (or $800). Damn.

By Katherine (not verified) on 24 Mar 2010 #permalink