Word Recognition in Multitalker Babble Measured with Two Psychophysical Methods

The purpose of this experiment was to determine the relationship between psychometric functions for words presented in multitalker babble using a descending presentation level protocol and a random presentation level protocol. Forty veterans (mean = 63.5 years) with mild-to-moderate sensorineural hearing losses were enrolled. Seventy of the Northwestern University Auditory Test No. 6 words spoken by the VA female speaker were presented at seven signal-to-babble ratios from 24 to 0 dB (10 words/step). Although the random procedure required 69 sec longer to administer than the descending protocol, there was no significant difference between the results obtained with the two psychophysical methods. There was almost no relation between the perceived ability of the listeners to understand speech in background noise and their measured ability to understand speech in multitalker babble. Likewise, there was a tenuous relation between pure-tone thresholds and performance on the words in babble and between recognition performance in quiet and performance on the words in babble.

Are Special Mechanisms Chosen in Short Experience-Based Identification of Degraded Geometric Forms?

Perceptual performance schema are consistently prepared by human beings. Minimal ‘experience’ with pertinent stimulus attributes seem necessary. We presume that with new stimuli, a ‘linkage’ is formed with inner representations on the basis of commonly shared stimulus components. To investigate this hypothesis, we used ‘confusion stimuli’, ie degradations of five-corner and six-corner stars. These were presented within each of four paradigms: upright (u), rotated (10°; r), foveal/local viewing (A) or eccentric/global viewing (B), that is uA, uB, rA, and rB paradigms. Random presentation of stimuli prototypes preceded the stimulus so that in comparing the presented stimulus to one and the other prototype, the subject faced a 2AFC task. Series (4 to 6) each of 32 randomised paradigms were run on each of six subjects. The subjects keyed 5 or 6 on identifying one or the other stimulus. On the assumption that exposure time and/or acuity grades impose stimulus intensities, psychometric curves were derived by plotting probability of correct choice against either parameter. The results indicate that in hybrid paradigms such as the present uA and uB, or rA and rB subjects may use shared component strategy: ‘u strategy’ with uA and uB, ‘r strategy’ with rA and rB, shared A and B in generating viewpoint invariance. Similar log-linear psychometric curves and slopes for pairs of paradigms were obtained, and ‘transfer’ of learning presumably also based on shared stimulus properties was observed.