Dependence of detection threshold estimates in rabbit on method of constant stimulus parameters

Abstract

The goal of this study was to estimate behavioral detection thresholds in quiet and noise for subsequent physiological studies of responses to near-threshold stimuli. The difficulty in measuring detection thresholds is maintaining stimulus control of behavior over schedule control. Tracking paradigms are efficient, but the large proportion of trials below threshold weakens stimulus control. The Method of Constant Stimuli controls the proportion of trials below, near, and above the SPL for a given percent correct. Here, percentages of trials at different SPLs and zero intensity catch trials were varied to determine proportions that yielded satisfactory stimulus control by a 500-Hz tone in quiet. Trials were initiated by nose-poke observing responses; a nose-poke reporting response within a 3-s window after tone onset resulted in food reinforcement. Reporting responses in the absence of tones resulted in timeouts. A metric for stimulus control was the ratio of the number of correct detections to the number of responses in the absence of the tone. This ratio was highest when approximately 20% of trials were near the 50%-correct SPL and 80% of trials were 20 dB higher. The ratio was lowest when significant proportions of catch and/or below-threshold trials were included. [Work supported by NIDCD DC-001641.]