A parietal–temporal sensory–motor integration area for the human vocal tract: Evidence from an fMRI study of skilled musicians

Abstract

Several sensory–motor integration regions have been identified in parietal cortex, which appear to be organized around motor-effectors (e.g., eyes, hands). We investigated whether a sensory–motor integration area might exist for the human vocal tract. Speech requires extensive sensory–motor integration, as does other abilities such as vocal musical skills. Recent work found that a posterior superior temporal-parietal region, area Spt, has both sensory (auditory) and motor response properties (for both speech and tonal stimuli). Brain activity of skilled pianists was measured with fMRI while they listened to a novel melody and either covertly hummed the melody (vocal tract effectors) or covertly played the melody on a piano (manual effectors). Activity in area Spt was significantly higher for the covert hum versus covert play condition. A region in the anterior IPS (aIPS) showed the reverse pattern, suggesting its involvement in sensory-manual transformations. This finding suggests that area Spt is a sensory–motor integration area for vocal tract gestures.

Introduction

There has been a great deal of work over the last decade aimed at understanding the neural circuits supporting sensory–motor interaction. Most of this research has centered on sensory–motor integration within the context of the visual system, and much has been learned. For example, regions in the posterior parietal cortex (PPC) in both human and non-human primates have been identified as critical components of visuomotor integration circuits (Andersen, 1997; Milner & Goodale, 1995). These regions appear to be organized primarily around motor-effector systems, such as ocular versus hand/limb action systems (Colby & Goldberg, 1999; Culham & Kanwisher, 2001; Grefkes, Ritzl, Zilles, & Fink, 2004; Kertzman, Schwarz, Zeffiro, & Hallett, 1997) and they may be computing coordinate transformations, mapping between sensory representations and motor commands during movement planning, online updating, and control of movement (Andersen, 1997, Castiello, 2005; Colby & Goldberg, 1999).

Much less research has been conducted on sensory–motor integration within the context of the auditory system. This is likely because most investigators who are interested in sensory–motor interactions have focused on eye or limb movements where visual information plays a dominant role in movement planning and online guidance. However, there is a domain of action where the auditory system plays a critical role, namely in the movements of the vocal tract for speech. For example, the developmental task of learning to articulate the sound patterns of one's language is cued externally by auditory input primarily (somatosensory feedback plays an additional important role, see Tremblay, Shiller, & Ostry, 2003). As adults, we can, simply by listening, learn to pronounce new words, or pick up regional accents, sometimes unconsciously. Experimental work has also shown that delayed or otherwise altered auditory speech feedback affects speech articulation (Houde & Jordan, 1998; Yates, 1963), and it is well known that late onset deafness results in articulatory decline (Waldstein, 1989). Finally, neuropsychological research has shown that damage to left hemisphere auditory regions leads to deficits in speech production (Damasio & Damasio, 1980). All of this shows clearly that the auditory system plays an important role in speech production and therefore, there must be a neural mechanism for interfacing auditory and motor representations of speech (Doupe & Kuhl, 1999). See Hickok and Poeppel (2007) for a recent review.

Until recently, little was known about the neural circuit(s) supporting auditory–motor integration. But recent fMRI experiments have made some progress in this respect. The design of these studies relied on the observation from the visual domain, that many neurons in PPC sensory–motor integration areas have both sensory and motor response properties (Murata, Gallese, Kaseda, & Sakata, 1996). Thus, an area supporting auditory–motor integration for speech should respond both during perception and production of speech (covert production is used in prior studies, including the present study, to avoid overt auditory feedback (Buchsbaum, Hickok, & Humphries, 2001; Hickok, Buchsbaum, Humphries, & Muftuler, 2003). Using this approach, a fronto-parietal–temporal network of auditory + motor responsive regions was identified in human cortex. Included in this network is a region in the left posterior Sylvian fissure at the parietal–temporal boundary, area Spt.¹ Area Spt appears to be functionally and anatomically connected with a frontal area known to be important for speech (area 44) (Buchsbaum et al., 2001, Buchsbaum et al., 2005a; Catani, Jones, & Ffytche, 2005; Galaburda & Sanides, 1980; Hickok & Poeppel, 2004) and the area Spt region, when disrupted via lesion or electrical stimulation, results in speech production deficits (Anderson et al., 1999; Damasio & Damasio, 1980). For these reasons, and also because of its anatomical location, area Spt has been hypothesized to be a sensory–motor integration area (Hickok et al., 2003) analogous in function to the sensory–motor integration areas previously identified in the PPC (Andersen, 1997; Colby & Goldberg, 1999).

Additional work showed that area Spt is not speech-specific, but responds equally well to the perception and production (covert humming) of tonal melodies (Hickok et al., 2003). Area Spt also responds during the temporary maintenance of a list of words in short-term memory independently of whether the items to be maintained were presented auditorily or visually (Buchsbaum et al., 2005b). This latter finding parallels claims that visual–motor integration areas have working memory-related properties (Murata et al., 1996), as well as the observation that sensory input from multiple modalities can drive neurons in PPC sensory–motor fields (Cohen, Batista, & Andersen, 2002; Mullette-Gillman, Cohen, & Groh, 2005). In sum, the response properties of area Spt – particularly that it shows both sensory and motor responses – are consistent with the hypothesis that it is a sensory–motor integration region similar to those found in the primate intraparietal sulcus (Buchsbaum et al., 2005b).

While area Spt has been shown to respond quite well in tasks involving a variety of sensory inputs (speech, tones, written words), its response to varying motor output conditions has not yet been investigated. While Baumann et al. (2007) report that the planum temporale is active in musicians generating hand/arm movements associated with playing a piece of music (the piece was “played” on a board), suggesting that Spt is active not only during orofacial-associated output tasks, but also manual tasks, without an orofacial task to use as a contrast it is impossible to gauge whether that region may be relatively selective for one or another output modality. If area Spt is organized in a manner analogous to PPC sensory–motor regions – which are organized around motor-effector systems – it should be fairly tightly linked to the vocal tract-related motor system. This, in turn, predicts that manipulation of the output modality should affect activity levels in area Spt even if the sensory stimulation were held constant. We set out to test this prediction in the present experiment.

One population in which auditory inputs can be mapped efficiently to two different motor-effector systems is skilled musicians. In at least some musicians, an aurally presented novel melody can be reproduced either by covertly humming the melody (vocal track effectors) or by covertly playing the melody (manual effectors). Based on our hypothesis that area Spt is a sensory–motor integration area involving vocal track-related motor systems, we predicted that area Spt would activate more strongly when musicians (pianists) were asked to covertly hum a heard melody than when they were asked to covertly play a heard melody. As a secondary hypothesis, we postulated that the use of manual effectors in covertly playing the melody will activate PPC, as this region has been implicated in several manual tasks including visually guided reaching, apraxic assessments, and piano-playing (Burnod et al., 1999, Haslinger et al., 2005; Makuuchi, Kaminaga, & Sugishita, 2005; Meister et al., 2004).