Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Implicit Learning and the Formation of Underspecified Mental Representations: an MMN Study
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
Mariya Kharaman1, Carsten Eulitz1; 1University of Konstanz
The human perceptual system can adapt to acoustic variability in incoming signals without storing every possible variation. In speech perception, phonological features that do not reliably signal phonemic categories are proposed to be underspecified in the mental lexicon, which increases processing efficiency. Our study investigates whether the formation of featurally underspecified category representations can also be generalized to the non-speech domain, which would suggest that underspecification of unreliable features is a general principle of cognitive systems. We compare the processing of acoustically complex non-speech stimuli before and after category learning using the Mismatch Negativity (MMN). Sixty-four different auditory stimuli are used in a computer game to signal four different categories to the participants. Acoustic parameters, such as the dynamic pitch contour and the frequency range of a superimposed high-frequency sine waves are defining the categories. After the initial MMN measurement, there was an implicit learning phase. Participants played a video game shooting bad aliens and capturing good ones coming from the left or the right side. Auditory stimuli (16 per category) indicated the alien type and the side of the screen they will pop up. The frequency range of a superimposed high-frequency sine waves reliably signalled that good aliens would pop up in the upper part of the screen and the bad ones in the lower part of the screen. The type of the dynamic pitch contour signalled only for the bad aliens the side of the screen they will pop up. For the good aliens the type of the dynamic pitch contour was an unreliable acoustic feature. We expect that the sound category representations for the good aliens will consequently be underspecified for the dynamic pitch contour. We predict a different pattern of MMN effects in the post-learning MMN measurement compared to the initial one with more asymmetric MMN effects when the featurally underspecified sound category serves as the deviant. The data collection is still ongoing and will be finished by July. The first preliminary data analyses show a change of pattern of MMN effects between pre- and post-learning measurements in every participant. Further data sets are required to assess the change in the pattern in more detail and to determine whether it is consistent with our predictions.
Topic Areas: Speech Perception, Phonology