Integrated versus independent processing of auditory features in speech sounds

Two MMN experiments investigate integrated versus independent processing of complex auditory information in linguistic sound. Our study asks if electrocortical techniques can be used to find feature additivity in the perception of linguistic sound and tests the efficacy of a roving-standard design, which allows for the direct comparison of multiple deviant contexts within a single session. We hypothesise that neurophysiologically evoked responses to multiple distinctive cues are stronger than responses to single cues. Two magnetoencephalographic (MEG) protocols are deployed, which are identical except for the order of presentation of standard and deviant stimuli. We observed larger evoked-response fields to deviants that differed in two phonetic features relative to deviants that differed in a single feature from the standard. Our results are consistent with the independent processing of acoustic cues and indicate that this novel methodology may be useful in testing questions of feature additivity.