To Örebro University

Listening to speech in noise is often reported to be effortful, especially for individuals with hearing impairment, and many studies have shown that the ability to recognize speech in noise is positively associated with working memory capacity. We reasoned that if working memory capacity could be increased by training this might improve the ability to recognize speech in noise and modulate the neural activation associated with it. Adults with normal (NH) and impaired hearing (HI) were randomized to five weeks of CogMed QM training followed by five weeks of no training, or vice versa, according to a crossover design. Auditory and cognitive abilities were tested on four occasions: pre-training, T1; after 5 weeks, T2; after 10 weeks, T3 and after a further six months, T4. During fMRI scanning at T1, T2 and T3, the participants listened to stereotyped matrix type sentences in pink noise and competing talker noise at individually adapted 50% and 90% intelligibility levels as well as in quiet. Behavioural results show that although HI had worse auditory abilities than NH, there was no significant difference in cognitive ability, with the exception of phonological processing, which tended to be slower (cf Classon et al. 2013). Performance on most of cognitive tasks improved across sessions, although this could not be specifically attributed to training. We found no consistent pattern of correlations between working memory and the ability to understand speech in noise either before or after training. fMRI results did not reveal any significant effect of training and furthermore there was no significant effect of hearing status. However, there was a significant between group difference in activation of the left temporal gyrus (-44 -23 10) for the contrast speech in pink noise (across intelligibility levels) vs clear speech. There was also an interaction (p < .001 uncorrected) between group and testing occasion in the right superior frontal gyrus (7 58 16) for the contrast speech in noise (across types and levels) vs clear speech. Further, activation in left superior temporal gyrus (-56 -20 -2) correlated more strongly with intelligibility in NH compared HI participants.

These results suggest that even when cognitive abilities are matched and intelligibility is individually adjusted, there are differences relating to hearing impairment in the neural mechanisms supporting speech in noise processing. The pattern of results suggests hearing-related differences in bottom-up processing mechanisms across time and hearing-related differences in top-down mechanisms that change over time. However, we found no evidence that working memory training improves speech recognition in noise.

References:Classon, E., Rudner, M. & Rönnberg, J. (2013). Working memory compensates for hearing relatedphonological processing deficit. Journal of Communication Disorders, 46, 17-29.doi:10.1016/j.jcomdis.2012.10.001CogMed QM, developed by CogMed Cognitive Medical Systems AB, Stockholm, Sweden, 2006