Item

Abstract

Previous work indicated that task-dependent modulation of the medial geniculate body (MGB) is behaviorally relevant for speech recognition [von Kriegstein et al., 2008]. The MGB is commonly divided into lemniscal and non-lemniscal parts [Bartlett 2013]. To-date it is unclear which part is modulated by a speech task. Here we used 7 T fMRI to test the task dependent modulation of MGB to eventually address this question. Twenty-eight healthy participants (19 female) listened to vowel-consonant-vowel syllables spoken by male or female voices and pressed a button if the syllable changed (speech task). In the control task listeners attended to speaker voice and pressed a button if the speaker changed (speaker task). To independently localize the MGB a sparse sampling, fast event related design was used with natural sound stimuli. This paradigm together with computational modeling has enabled the localization of the tonotopically organized non-lemniscal ventral subdivision of the MGB (MGBv) [Moerel et al. 2015]. Data were analyzed using SPM12 with Frequentist and Bayesian estimation. A significant result was obtained for the left MGB explained by the behavioral percent correct speech score using small volume correction (p=0.035). For the Bayesian inference a log-odds of 2.020 was found in the left MGB for the same contrast, which translates to a probability of responses of 0.883. This research shows that the task based modulation of the MGB, especially with respect to fast varying stimuli can be seen using ultra high field fMRI at 7 T. The results reproduce previous ones found with a lower magnetic field, and strengthen the hypothesis of cortical feedback loops to Thalamic auditory centers involved in speech processing. In a next step we will investigate whether the task-dependent modulation is located in the MGBv.