Short-term categorical/noncategorical processing in the delayed-discrimination of monaural and dichotic synthetic vowels and nonspeech tones

Auditory stimuli that can be distinguished by phonetic labels (e.g., the between-category vowels /i/ and /I/) are typically discriminated categorically and so prevail in memory beyond the short term. Auditory stimuli that cannot be distinguished by category labels (e.g., within-category variants of the vowel /I/ and tones whose frequency separations may be neither broadly salient nor musically scalable) can only be discriminated noncategorically, so last only momentarily. These between/within-category stimuli were employed to investigate phonetic/nonphonetic vowel memory, to compare the nonphonetic memory of vowels and tones, and to tease apart by discrimination accuracy and response time measures the potentially separate, potentially overlapping, sensory memory mechanisms accessible via left and right ears.^ Eight subjects hearing pairs of 70-msec monaural/dichotic steady-state vowels and tones separated by a 250, 500, 1250 or 2000-msec interval with or without another, intervening (distractor) vowel or tone (S1-Sd-S2), responded different if S1-S2 sounded different, same if same.^ As testing with trial-by-trial feedback extended over months, between-category vowel functions emerged essentially flat over the two-second interval sampled, within-category d\u27 declining, RT increasing.^ Doubling memory load did not halve accuracy nor greatly increase RT on simpler conditions (indicating sensory memory to be intrinsically passive, automatic, parallel). But simultaneous masking (largely by ipsilateral tone) and distractor suffix masking (primarily ipsilateral) on more complex conditions increased perceptual difficulty and handling time considerably. With these masking effects subtracted out, dichotic memory doubling proved less robust than monaural: one ear\u27s sensory memory is relatively better than two.^ Though the known factors contributing to ear advantage (namely, each ear\u27s privileged anatomical access to the contralateral hemisphere, and hemispheric specialization) could not be dissociated by the present design, a modest right-ear advantage (small discrimination accuracy and RT differences), not significant for vowels summed significantly with a left-ear advantage for tones (significant accuracy and nearly significant RT differences). Taken together with the results of ipsilateral/contralateral suffix masking, single/dual-memory loading, and crossed/uncrossed-ear comparisons, these results describe two passive, automatic, parallel, virtually independent vowel and tone sensory memory processes.