Is it still speech? Different processing strategies in learning to discriminate stimuli in the transition from speech to non-speech including feedback evaluation

Processing of speech was investigated by using stimuli gradually changing from speech (vowels) to non-speech (spectral rotated vowels). Stimuli were presented in descending levels of vocalization blends, from pure speech to non-speech, through step-wise combinations, resulting in ambiguous versions of the sounds. Participants performed a two-alternative forced choice task: categorization of sounds were made according to whether they contained more speech or non-speech. Performance feedback was presented visually on each trial. Reaction times (RT) after sound presentation, and functional magnetic resonance imaging (fMRI) data during auditory and visual processing, were analyzed. RT data suggested individual differences with a distinct group, good performers, functioning better in distinguishing stimuli with a higher degree of ambiguous blends compared to poor performers, who were not able to distinguish these stimuli correctly. fMRI data confirmed this finding. During auditory stimulation, good performers showed neural activation in the ventral auditory pathway, including the primary auditory cortex and the anterior superior temporal sulcus (responsible for speech processing). Poor performers, in contrast, showed neural activation in the dorsal auditory pathway, including the bilateral superior temporal gyrus. Group differences were also found for visual feedback processing. Differences observed between the groups were interpreted as reflecting different neural processing strategies.