In the November 5 issue of The New York Review of Books, physician/writer Jerome Groopman references a conference he led at Massachusetts General Hospital for interns and residents. The subject: The causes of misdiagnosis.
Groopman drew his audience's attention to a seminal paper by Amos Tversky and Daniel Kahneman (Science, September 27, 1974) on the cognitive pitfalls of human thinking. According to Groopman, the authors studied three kinds of bias:
1) "Anchoring," where a person overvalues the first data he or she encounters;
2) "Availability," where recent or dramatic cases come to mind and so skew one's thinking;
3) "Attribution," where stereotypes prejudice thinking so conclusions arise not from data but from preconceptions.
I took a look at the Tversky and Kahneman paper (available from the Science online archive by subscription only). It is long and technical, but Groopman's summary is fair enough. The good doctor is primarily concerned with how these biases affect medical diagnosis, but of course they also shape our judgments in matters of science, politics, religion, and general life choices.
The whole point of scientific methodology is to minimize these biases. Quantitative data and analysis, control groups, double-blind experiments, peer review, impersonal communication, and so on are all designed to dilute -- and ideally remove -- inevitable individual biases and preconceptions.
Perhaps nowhere are the effects of anchoring, availability and attribution more manifest than in matters of religion. The overwhelming majority of people commit themselves to the familial or cultural religion into which they were born. Even if upon reaching maturity we subject our beliefs to thoughtful analysis, anchoring can still have a hold on our allegiance.
None of us are free of these biases, no matter how hard we try, but being aware of them might at least evoke a degree of agnostic humility.