NEUROSC M101C Lecture Notes - Lecture 15: Appendicitis, Computer Monitor, Bayes Estimator
Document Summary
Sdt is a general framework for understanding binary decisions under uncertainty. It is fundamental to understanding perception in general. Originally developed in the 60s by green and swets, the ideas came from how engineers could analyze the performance of electronic detection devices. Mathematically sdt is a variant of bayesian decision theory. The very basic idea: in psychology experiments we often ask people to detect a stimulus, and we want to know how good they are at doing this. One intuitive approach is to measure how many trials they get correct. But accuracy rate (% correct) is determined by two factors: capacity (a. k. a. sensitivity) and bias (a. k. a criterion). Capacity is the ability to process the information. Bias refers to the responding strategy: whether one is liberal or conservative in giving a positive answer, regardless of whatever capacity one has. The two things are different, and sdt is about formally sorting them out.
