Signal-Detection Theory Applied to Selective Listening

Abstract

Recently, 2 papers have appeared that have applied signal detection theory to selective listening. Both have certain defects, and the present paper is intended to clarify some of the issues raised by the earlier papers. Broadbent and Gregory used a tone detection task on one ear while presenting a memory load to the other and found that d'' but not B altered for the detection task. Treisman and Geffin recently reported on the "shadowing" situation (continuous repetition of an ongoing message) and again found evidence that d'' but not B was affected. Treisman interpreted her results to mean that selection of input, not of response, was operative, basing her conclusions on the way in which the subject responded to homophones and homonyms. Detection of signals was signaled by the subject tapping the microphone. Unfortunately, Treisman was forced to use pooled data for the estimation of d'' and B which involved some strong assumptions about the relative performance of individual subjects. The present experiment was designed to provide a relatively "pure" selective listening task for the applicability of signal detection theory to the process of selective attention.