THRESHOLD THEORIES OF SIGNAL DETECTION

2-state low-threshold theory (2-LT), combined with a probabilistic response model, gives a plausible account of those features of confidence-rating detection data that have been cited as evidence against it. Nevertheless, there are data that conclusively reject 2-LT. A successful threshold theory must have both a low and a high threshold (and in consequence, must postulate at least 3 sensory states involved in detection). 3-state low- and high-threshold theory (3-LHT) is compatible with available detection data and is quantitatively quite similar to normal distribution signal detection theory with signal-plus-noise variance exceeding noise variance. The most promising way to reject 3-LHT, and all other threshold theories as well, would be to show that the posterior probability of the signal, given the response, can approach zero for moderate signal strengths. (26 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1969 American Psychological Association.