Noninvariance of the ear advantage in dichotic listening

Listeners received 50 dichotic trials of both natural and synthetic CV syllables, and marked two responses for each of 30 syllables pairs per trial. The paired responses were distributed as a bivariate normal; left and right-ear scores had equal variance (σ=8%) and were uncorrelated (r=−0.02 for synthetic; +0.15 for natural). Percent ear advantage also was distributed normally with a standard deviation about equal to the mean (synthetic, X̄=12.9%, σ=11%; natural, X̄=9%, σ=10%). Thus, listeners with a mean REA evidenced many trials with either no ear advantage or an LEA (29% natural; 19% synthetic). Variability in left and right-ear scores is assumed to arise from random, uncorrelated, time-varying “noise” in pathways from each ear. The ear advantage, left or right, is modeled as a function of the S/N ratio in the respective pathways. The ear advantage is not invariant and, by this account, may not always be indicative of unilateral hemispheric specialization for speech processing. [Supported by PHS Grant No. NS-12125.]