Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Selective neuronal tuning to spoken words in the dorsal auditory stream

Poster C8 in Poster Session C, Friday, October 7, 10:15 am - 12:00 pm EDT, Millennium Hall

Lillian Chang¹, Srikanth Damera¹, Xiong Jiang¹, Josef Rauschecker¹, Maximilian Riesenhuber¹; ¹Georgetown University Medical Center

Cortical models of auditory processing (Rauschecker & Scott, Nat Neurosci, 2009; Hickok & Poeppel, Nat Rev Neurosci, 2007) propose a dual-stream architecture consisting of anteroventral and dorsal streams, respectively. In the context of auditory speech processing, the ventral, “what” pathway is specialized for spoken word recognition, and recent evidence supports the existence of an auditory word form area (AWFA) in the anterior superior temporal cortex (aSTC) that contains neuronal populations tightly tuned to whole spoken words (DeWitt & Rauschecker, PNAS, 2012; Damera et al., Soc Neurobiol Lang, 2019). The AWFA parallels the well-established visual word form area (VWFA), an analogous region with lexical representations for written words in the ventral visual stream. In contrast, the dorsal, “how” stream is thought to subserve speech production through the mapping of speech sounds to their motor articulations (Rauschecker, Hearing Res, 2011; Hickok et al., Neuron, 2011; Chevillet et al., J Neurosci, 2013). Two key areas in the dorsal stream hierarchy for sensorimotor integration of auditory speech are the posterior superior temporal cortex (pSTC) and the premotor cortex (PMC; Wilson et al. Nat Neurosci, 2004). We previously reported effective connectivity from the pSTC to the PMC when listening to speech sounds, thought to reflect the mapping of these sounds to their motor articulations (Chevillet et al., J Neurosci, 2013). We here hypothesized a hierarchical representation for auditory words in the dorsal auditory stream, with a sublexical representation in the pSTC (paralleling our earlier finding of a sublexical representation for written words in the pSTC, Glezer et al., NeuroImage, 2016; Brain Lang, 2019) and a lexical representation in the PMC. To test our hypotheses, we first conducted localizer scans to identify the pSTC and PMC ROI individually in each of N=23 subjects. We then conducted two fMRI rapid adaptation (fMRI-RA) experiments with auditorily presented real words (RWs) and novel pseudowords (PWs; i.e., unfamiliar nonwords), respectively. In these experiments, two words of a single stimuli category (RW or PW) were presented in quick succession; they could be identical (SAME), differ by a single phoneme (1PH), or share no phonemes at all (DIFF). Supporting our hypotheses, we found patterns of release from adaptation in the PMC compatible with an auditory lexicon in the dorsal stream, with significant adaptation effects for RWs (SAME vs. DIFF: p=0.0013; SAME vs. 1PH: p<0.001; but no significant differences between 1PH and DIFF, p=0.547, indicating full release from adaptation already for two words differing by a single phoneme) but not for PWs, replicating our earlier results for the AWFA in the anteroventral stream. In contrast, there was a gradual release from adaptation in the pSTC compatible with a sublexical representation (SAME vs. 1PH: p<0.001; 1PH vs. DIFF: p=0.021; SAME vs. DIFF: p<0.001). Our findings therefore support the the proposed “inverse” model mapping speech sounds to their motor articulations in the dorsal auditory stream, starting with sublexical representation in the pSTC and culminating in an articulatory lexicon in the PMC.

Topic Areas: Speech Perception, Speech Motor Control