Conservatism (belief revision)

In cognitive psychology and decision science, conservatism or conservatism bias is a bias in human information processing, which refers to the tendency to revise one's belief insufficiently when presented with new evidence. This bias describes human belief revision in which persons over-weigh the prior distribution (base rate) and under-weigh new sample evidence when compared to Bayesian belief-revision.

According to the theory, "opinion change is very orderly, and usually proportional to the numbers of Bayes' theorem – but it is insufficient in amount".[1] In other words, persons update their prior beliefs as new evidence becomes available, but they do so more slowly than they would if they used Bayes' theorem.

This bias was discussed by Ward Edwards in 1968,[1] who reported on experiments like the following one:

There are two bookbags, one containing 700 red and 300 blue chips, the other containing 300 red and 700 blue. Take one of the bags. Now, you sample, randomly, with replacement after each chip. In 12 samples, you get 8 reds and 4 blues. what is the probability that this is the predominantly red bag?

Most subjects chose an answer around .7. The correct answer according to Bayes' theorem is closer to .97. Edwards suggested that people updated beliefs conservatively, in accordance with Bayes' theorem more slowly. They updated from .5 incorrectly according to an observed bias in several experiments.[1]

In finance

In finance, evidence has been found that investors under-react to corporate events, consistent with conservatism. This includes announcements of earnings, changes in dividends, and stock splits.[2]

Possible explanations

The traditional explanation for this effect is that it is an extension of the anchoring bias, as studied by Tversky and Kahneman. The initial "anchor" is the .5 probability given when there are two choices without any other evidence, and people fail to adjust sufficiently far away. However, a recent study suggests that the belief revising conservatism can be explained by an information-theoretic generative mechanism and that assumes a noisy conversion of objective evidence (observation) into subjective estimates (judgment).[3] The study explains that the estimates of conditional probabilities are conservative because of noise in the retrieval of information from memory, whereas noise is defined as the mixing of evidence. Memories of high likelihoods are mixed with evidence of low likelihood and the resulting estimate is lower than it should be. The retrieval of memories of lower likelihoods are higher than they should be, and the result is conservatism (low is not low enough, and high is not high enough, the result is not extreme enough, which is conservative).

In an incentivized experimental study, it has been shown that the conservatism bias decreased in those with greater cognitive ability, though it did not disappear.[4]

See also

References

  1. 1 2 3 Edwards, Ward. "Conservatism in Human Information Processing (excerpted)". In Daniel Kahneman, Paul Slovic and Amos Tversky. (1982). Judgment under uncertainty: Heuristics and biases. New York: Cambridge University Press. ISBN 978-0521284141 Original work published 1968.
  2. Kadiyala, Padmaja; Rau, P. Raghavendra (2004). "Investor Reaction to Corporate Event Announcements: Under-reaction or Over-reaction?". The Journal of Business. 77 (4): 357–386. doi:10.1086/381273. JSTOR 10.1086/381273.. Earlier version at doi:10.2139/ssrn.249979
  3. Hilbert, Martin (2012) "Toward a synthesis of cognitive biases: How noisy information processing can bias human decision making". Psychological Bulletin, 138(2), 211–237; free access to the study here: http://www.martinhilbert.net/HilbertPsychBull.pdf
  4. Oechssler, Jörg; Roider, Andreas; Schmitz, Patrick W. (2009). "Cognitive abilities and behavioral biases". Journal of Economic Behavior & Organization. 72 (1): 147–152. doi:10.1016/j.jebo.2009.04.018.
This article is issued from Wikipedia - version of the 9/18/2016. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.