Wednesday, 26 February 2014


I read the behavioural literature as a meta form of self-help. Here are some standard ways that people can screw things up; here are some heuristics they use that work on average but can yield failures when applied to the wrong domain; here are some strategies for applying the right heuristic at the right time and for avoiding applying the wrong one; here are some common spots where people need to be extra-vigilant to avoid making errors.

Gareth Morgan tweets a link to a write-up of the standard Wansink findings around food:
Sure, in field experiments, you can induce overeating by making people think that they've eaten less than they have (for example, by surreptitiously filling the bowl from below). But does that mean that they're irrational and always subject to error? Or might it mean that people eat until one of two conditions are met: satiation, or end of current portion? If the latter typically comes before the former, people stop eating at the end of the bowl. If the former tends to come before the latter, they'll leave some behind. What interest would a restaurant have in supplying you with more food than you'd really want when doing so might make you less likely to order dessert and will make you more likely to linger longer at the table?

The linked piece also takes a self-help approach to the findings: Try using smaller bowls or smaller plates; don't go for "value" deals if that isn't what you really want to eat.

And so Matt Nolan replied to Gareth:
Morgan replied,
This kind of line really bugs me; it reminds me of the kind of thing that non-economists will come out with when criticising economics. Imperfect information hardly seems to be what's driving food choices. And, perfect knowledge is hardly necessary to make precommitment viable. You just need to know that you often screw up particular kinds of choices.

Odysseus didn't need perfect information about just how lovely the Sirens' call was in order to have the sailors bind him to the mast; he just needed to know that the temptation had proved too tempting for many others. I've never played World of Warcraft, but that doesn't mean that I've erred in deciding never ever to start playing multiplayer online games. I definitely don't have perfect information about it - I've never played it! But I know that I'd find it hard to avoid spending too much time playing online games if I had the added pressure of friends wanting me to come help them on a raid. So I just don't play. Imperfect information has led me to consume what's likely too little gaming relative to an ideal: you don't need to assume perfect information to get precommitment.


  1. "I've never played World of Warcraft, but that doesn't mean that I've
    erred in deciding never ever to start playing multiplayer online games."

    That's why I didn't try MDMA when many others did. I had a fair idea of what it did (or thought I did) and it just seemed too darn attractive to stop once I'd started. Becker and Murphy would be proud of me.

    If you're interested in anomaly research as self-help, you might like Gerd Gigerenzer's work. I haven't read nearly all of his stuff, but his paper on how to teach people Bayesian reasoning so that they get it is a bit of a classic (short version: use natural frequencies, not probabilities).

  2. Oh, and as for the perfect information nonsense, you're right to be annoyed given that this problem was adressed quite satisfactory by Riker and Ordeshook in 1973, on pp. 21-23 of their Introduction to Positive Political Theory. Not exactly a bestseller, but I would expect someone who wields technical phrases like "perfect information" to have read some of the literature.

  3. Hello Eric! I found your blog when searching for professor jobs in New Zealand. Where should an American start looking if she'd like to teach university in NZ?