Dear list, does anyone know of any work that has formally applied signal detection theory or a related framework to continuous confidence ratings? By continuous, I mean a rating scale that is not discretized. For example, say we
have a recognition memory task in which subjects are presented with old or new stimuli. The discretized version might have response options “definitely old”, “maybe old”, “maybe new” “definitely new”, whereas the continuous version might require subjects to
use a slider going from 0 to 1. By formally, I mean that the model was actually fitted to the continuous ratings and not to the data after artificially discretizing them, for example into quartiles.
--
Samuel R. Mathias, Ph.D.
Associate Research Scientist (ARS)
Neurocognition, Neurocomputation and Neurogenetics (n3) Division
Yale University School of Medicine
40 Temple Street, Room 694
New Haven CT 06510
http://www.srmathias.com