• If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • Finally, you can manage your Google Docs, uploads, and email attachments (plus Dropbox and Slack files) in one convenient place. Claim a free account, and in less than 2 minutes, Dokkio (from the makers of PBworks) can automatically organize your content for you.


Predictably Irrational

Page history last edited by PBworks 12 years, 4 months ago

Notes on Dan Ariely (2008) Predictably Irrational: The hidden forces that shape our decisions, Harper Collins. ISBN: 978-0-00-725-652-5


This is a gem of a book: short, engagingly-written and connected both to the science and the policy implications. Ariely is a seasoned bias researcher (he sees himself as a behavioural economist rather than a psychologist) and this book runs through many of the experiments he has been involved in, as well as related research. There is some, but not a huge amount, or overlap between the biases discussed by Ariely and by the other books I recommend on this site.


The main thrust of the book, as suggested by the title, is the economist's idea of rational choice does not describe how we actually behave, because we are irrational in predictable ways (or, as I prefer to call it, biased). This has a number of implications for how we can live better and happier lives: we cannot leave it to the free market to fix things; voluntarily restricting our choices can help us to achieve more; we need to self-police our selves to counter our tendency to rationalise immoral behaviour.


I recommend anyone interested in the topic of bias taking this on a long train ride, and taking a lot longer to think about the implications.




1: The Truth about Relativity

The decoy effect on preference: given a choice between A and B which are hard to compare, subjects are much more likely to choose option A when a third option, like A but visibly inferior, is on offer. The decoy is an "irrelevant alternative" in the language of economics (because it is never chosen itself) yet, contrary to economic theory, it has a major effect on the choice.

Examples: Economist online subscription versus print-and-online subscription; a photograph of an attractive student is preferred to a photograph of a similarly attractive student when we add a photograph of the same person photoshopped to look ugly.

This is an example of how contrast effects are ubiquitous in perception.


Social comparison is another example: the rich are made unhappy by envy of the super-rich. By restricting the social circles we move in, we can restrict the social comparisons we make and so avoid dissatisfaction.


2: The Fallacy of Supply and Demand

People decide whether something is "cheap" or "expensive" not on some absolute scale but by the anchoring-and-adjustment heuristic. What they anchor on is what they previously paid for that good. When the price decreases, they see it as cheap and (in the short run) buy more. These changes in buying behaviour can give the illusion of a stable demand curve which is derived from personal preferences (a la orthodox economic theory). The difference is that this comparison phenomenon is dependent on memory, so we should not expect these changes to survive in the long run, or to survive a loss of memory. This phenomenon is called arbitrary coherence, in that the initial anchored price can be arbitrary but people react to changes of price in a coherent way.


Tahitian black pearls were regarded as inferior until the seller took the deliberate step of selling them alongside diamonds and other gems with very high prices to match.


Some anchoring experiments are described: suggested payments for enduring an aversive noise affect the subjects' subsequent bids in an auction. When a second anchor is introduced which is the same for all subjects, the first anchor still dominates. Primacy turns out to be more important than recency in anchoring price judgements.


Self-herding: just as people's choices are shaped by what they see other people doing, their behaviour is also shaped by what they remember doing. So once people can be persuaded to pay a lot of money for a cup of coffee, they get used to it and think of that amount as the price of coffee, even if it's much higher than what they previously paid.


3: The Cost of Zero Cost

Zero cost has a special psychological appeal which is not accountable for in terms of elastic demand. Ariely's explanation is in terms of loss aversion: with "free" there's no implied loss (even though there may be a non-monetary loss in terms of missed opportunity).


Subjects offered a luxury chocolate at 27 cents or ordinary chocolate at 2 cents. Most choose the luxury option. Change prices to 26 cents and 1 cent (equivalent to the original choice plus a gift of one cent) and the pattern is the same. Change again to 25 cents and 0 cents respectively (equivalent to the original choice plus a gift of two cents) and the majority now choose the ordinary chocolate. Students were presented with these options at the till of a cafeteria, so finding change was unlikely to be a factor.


This effect also applies in non-monetary transactions (chocolate bars experiment).


Shoppers offered a free $10 Amazon voucher or a $20 for only $7 mostly choose the former option.


4: The Cost of Social Norms

Two contexts for doing something for other people: 1) social norm (you do them a favour and they will do an unspecified favour for you down the line, or pass on the favour to someone else); 2) market norms: doing something to get paid. E.g. relationship sex versus prostitution. (cf. intrinsic versus extrinsic motivation; motivation by incentives and motivation by self-image)


Experiments compare performance on a task when the subjects are doing it in return for money against doing it as an unpaid favour. Unpaid subjects work best on the task.

Subjects given a gift rather than money work as hard as unpaid subjects (a more expensive gift does not increase their work).

Subjects given a gift and told how much the gift cost show the same low motivation as when given the equivalent amount of money.

S's primed to think about money by a word puzzle were less willing to ask for help when doing an unrelated task, and also less likely to offer help to others.

Corporations are increasingly trying to employ social norms (making customers think of it as a friend or family), but doing it half-heartedly is worse than not doing it


5: The Influence of Arousal

Student's opinions of whether they would do various taboo or dangerous activities (unprotected sex, being spanked, homosexuality, bestiality, ignoring "no") is very different when asked the questions in a "cold" state or an aroused state. Poor self-knowledge: e.g. teens thinking they will be able to abstain from sex in the heat of the moment. (The experiment involved a Mac iBook covered in Saran wrap so that masturbating men could answer the survey!)


First-time mothers predicting they will not need pain relief when giving birth but asking for it when the time comes: example of how people in the "cold state" fail to take into account that their choices will be different in the "hot state" and so predict their own behaviour badly.


6: The Problem of Procrastination and Self-Control

Student self-control experiment with three groups. All the students take a 12-week course with three essays to be marked at the end of the 12 weeks.

  1. Total freedom of choice: No deadlines within the term
  2. Dictatorial regime: One essay due every four weeks with a penalty for essays handed in after the relevant deadline.
  3. Students set own deadlines at start of course. For each essay, they select a week by which they will have submitted it. They are committed to this, with penalties for late submission. (libertarian paternalism?)

In terms of final marks, dictatorical regime did best, total freedom worst.

Result contradicts economic principle that more choice makes us better off. In this case, students in week 1 fail to take into account their own tendency to procrastinate, which will bunch all their work into week 12 if they take the "rational" step of setting the deadlines as late as possible. Self-imposed deadlines (in economic terms, a self-imposed cost), make for better results than total freedom.


Looking at those students in the self-set deadline group, the ones that chose weeks 4, 8 and 12 (the same as in the dictatorial condition) did as well as the dictatorial group. The reason the self-set deadline group did not do as well as well as the dictatorial group is that some did not space their deadlines out.


One thing contributing to procrastination in looking after a car or a person (i.e. health) is that there are so many different tasks to be done that there is an overload of choice. Solution: bundle up check-ups so you don't make separate visits to get different parts of your car (or your body) checked.


"Self-control credit card": would impose penalties, chosen by the user, for exceeding certain categories of spending.


7: The High Price of Ownership

Endowment effect

Replicating a classic loss aversion experiment: subjects are allocated tickets to a sports game by lottery. Those who won tickets valued them at $2400. That was the lowest price, on average that they would accept to part with their tickets. Students who did not win tickets valued them at $175 (the maximum they were prepared to pay).


People putting a high price on things that they own is perhaps an example of failure to de-centre: they attach particular memories and significance to the thing, and fail to realise that other people won't see it the same way (e.g. selling an old van).


Endowment effect is exploited by companies with "free trial" offers or money-back guarantees.


8: Keeping Doors Open

Subjects play a computer game in which staying in each of three rooms earns them a different sum of money (amount varies to introduce some uncertainty). If a room is not visited for a number of turns, it is no longer available (decreasing options). Moving from one room to another takes up a move (which could be used to accumulate money). Subjects persist keep their options open, losing much money in the process. This is a case of people being made worse off by the pursuit of maximum freedom of choice.


Staying with any one option for the whole game would have made more money than keeping all options open. When the game was made easier and the subjects were given practice, they still persisted in this self-defeating behaviour. Analogy for career, child activities, hobbies?


Buridan's Ass shows that the costs of indecision can outweigh the costs of the options being chosen between.


9: The Effect of Expectations

Students in a blind taste test prefer beer with a few drops of balsamic vinegar (compared to the same beer without). Subjects who are told the difference beforehand like the vinegar-laced beer less. Subjects who are told the difference after sampling the two beers prefer the one with vinegar.


A similar experiment: subjects rate the taste of coffee more highly if a variety of unusual condiments (e.g. orange peel) are presented along with it. These even though none of the subjects used the condiments. [most taste is smell: could it have been that the condiments had strong smells which affected the drinker's sense of taste]


[cf Placebo effect! addressed more in next chapter]


Pepsi does better in blind tests than Coke. Coke does better when people know what they are drinking. Brand loyalty seems to actually affect the experience of drinking the cola.


Stereotype threat/stereotype enhancement: Asian-american women have both a positive and negative self-stereotype concerning their ability at maths. These can be activately selectively to boost or lower their maths exam performance.


10: The Power of Price

Placebo surgery for chest pain (performed in the 1950s by a Dr Leonard Cobb) was as effective as real heart surgery. A similar effect was found in 1993 with knee surgery for osteoarthritis.


Subjects given a fake painkiller (actually Vitamin C) report less pain when given a series of powerful electric shocks. Price moderates this effect. When the drug is described as costing two and a half dollars, almost all subjects report it having an effect, but when it is reported as 10 cents, only half do.


Students given energy drink and an anagram test. When given the drink at a "discount", performance is worse. Hype of the drink's alleged cognition-enhancing properties raises performance as well.


Moral and economic dilemmas raised by more expensive placebos having apparently more effect than discount ones.


11: The Context of Our Character, Part I

We treat dishonesty very differently depending on the context. Stealing something from a shop we treat as more morally serious than office crime such as falsifying business expenses, even though tax fraud and office crime cost many times more than street crime.


Student subjects paid ten cents for each correct answer in a multi-choice test. Given the task of transferring their answers to the scoring sheet (hence an opportunity to cheat). Different conditions give different opportunities to cheat without being detected. Outcome is the total number of cents taken by the group.


Cheating was present in all groups with an opportunity, but didn't differ much between them. Subjects do not take the opportunity to cheat on a grand scale.


Priming for honesty: In another experiment where they had opportunities to cheat undetected, subjects were either asked to recall the Ten Commandments or a list of ten books. The former group did not cheat at all, although most of them could hardly remember any of the commandments themselves. A same result happened when students were prompted to think of "the MIT honor code" which doesn't even exist.


The challenge is to make honesty and ethics salient at the crucial time when temptation is available.


12: The Context of Our Character, Part II

Leave cans of coke all around a dormitory: they are gone in two or three days. Leave plates of money lying around in the same places and they are still there three days later. It is somehow easier for people to steal goods than to steal money to the value of the goods.


Honesty experiments like in the previous chapter, where students were paid for correct answers to a quiz, and had the opportunity to lie about their score. This time they were paid in tokens by one experimenter, then went to another experimenter across the room (12 feet away) to exchange the tokens for cash. Subjects cheated for tokens to twice the extent that they did for cash. In the tokens condition, far more people cheated as far as possible (claiming the maximum amount of money).


Cash seems to be treated with a moral significance not shared by other things, even when they have the exact same value. Other examples: "stretched" insurance claims, "wardrobing" clothes bought from a store, questionable expense receipts. These things aren't like grabbing cash from a safe. They are one step removed from cash.


These experiments demonstrated the existence of ethical conscience, because nearly everybody had limits to how much they would cheat. It doesn't cover what more greedy, selfish people would do with the ability to cheat.


As cash declines and credit cards ascend, this propensity to cheat has increasingly worrying implications.


13: Beer and Free Lunches

Waiters offer four different beers to patrons in a bar. "Public condition": each patron in turn picks a beer by stating out loud. "Private condition": patrons tick a box on a form to select beer. Public condition results in more diverse choices and less customer satisfaction. Need for uniqueness in this context results in lower happiness. In Hong Kong, public condition resulted in more conformity (and again lower happiness) than private condition.


Orthodox economics implies rational agents and "no free lunch". Predictable irrationality implies there are free lunches, i.e. opportunities for better decisions.


Irrationality and illusions are an inevitable part of how humans work, but that does not mean we are helpless: we can take steps at a personal or policy level to compensate.

Comments (0)

You don't have permission to comment on this page.