Methods for estimating infant thresholds

Abstract

Detection thresholds for 1000-Hz, 16-ms tone bursts were estimated for 3- and 6-month-old infants and for young adults. The test procedure used was the observer-based psychoacoustic procedure. Thresholds were estimated using two different adaptive procedures and the method of constant stimuli. There was little difference among the average thresholds determined by any of these techniques. The least variable thresholds were obtained in the method of constant stimuli. In addition, 10 infants at each age completed two 30-trial blocks of trials in the method of constant stimuli; 22 adults completed 8 blocks of 30 trials. For 3- and 6-month-olds and for adults, there was no significant change in average threshold between blocks. Individual 6-month-olds' thresholds rarely changed by more than 5 dB between blocks, and the correlation between the thresholds obtained in the two blocks was significant. Individual 3-month-olds' thresholds, however, sometimes changed by as much as 10 dB between blocks, and the correlation between first and second block threshold was not significant. The effects of response bias on threshold were assessed by examining receiver operating characteristic plots of hit and false alarm rates at threshold and the correlation between false alarm rate and threshold. Although there was some variability in response bias, infant/observer teams tended to respond in an unbiased fashion. In one of the adaptive procedures only, false alarm rate was significantly correlated with threshold. In all procedures, the exclusion of infants with high false alarm rates changed the average thresholds obtained by less than 5 dB.(ABSTRACT TRUNCATED AT 250 WORDS)