Saturday, December 29, 2007

consumer decision making

I was just reading an article on the Journal of Consumer Affairs about consumer decision making in a very emotionally charged, confusing, and time pressured context (funeral services). This is a tough combination of a challenging environment, novice decision makers, and aggressive sales practices.

Not a situation I would wish on anyone, but a great scenario for research because it is an serious edge case - bringing out the extremes of the human decision making process. The researchers found that consumers of these services generally used many of the cognitive load-reducing decision heuristics that have been called biases specifically because of these results. I don't blame any consumer facing the death of a loved one for trying to reduce the attention dedicated to picking a coffin, plot, embalming services et&c. But the problem is that they

1. evaluated very few alternatives
2. trusted the reliability of the information they were given (by salespeople)
3. did not search for additional information

The authors suggest government regulation. My default opinion about regulation is that we should give consumers all the information they need, but not to constrain their choice. But in this case, perhaps stronger regulation is needed because it is hard to require consumers in this context to be responsible information acquirers and processors.

Saturday, December 01, 2007

chicken sexing

OK, this is my favorite example of fractionation. Thanks to Jared Spool who talks about it in his webcast The Dawning of the Age of Experience. Basically, chicken farms profit by identifying the gender of their baby chicks as early as possible. Unfortunately, there are no physical differences between the male and female babies for several months. With no expertise, that gives them a 50/50 chance of guessing correctly. There are experts who can get up to 80/20, but they can't articulate how they know. They have fractionated their male/female schema so that they can recognize the differences, but there are no connections to any labels. They "just know."

So the training process simply repeats the learning process. A new hire categorizes the baby chicks by gender (at an accuracy of just 50/50 because they don't have any clue), and the expert hits them on the shoulder when they are wrong (feedback). After a few weeks (and hundreds of chicks), the new hires can get up to 70/30 accuracy rates, even though they also can't say quite how they know.

It is a great example of how much cognition goes on in areas of the brain that are simply not accessible to conscious examination.