Hearing Research 150 (2000) 225^244 www.elsevier.com/locate/heares Frequency-dependent responses exhibited by multiple regions in human auditory cortex1 Thomas M. Talavage a;b;*, Patrick J. Ledden b, Randall R. Benson b;2, Bruce R. Rosen b, Jennifer R. Melcher a;c;d; a Speech and Hearing Sciences Program, MIT-Harvard Division of Health Sciences and Technology, Cambridge, MA, USA b MGH-NMR Center, Department of Radiology, Massachusetts General Hospital, Building 149, 13th Street (2301), Charlestown, Boston, MA 02129, USA c Department of Otology and Laryngology, Harvard Medical School, Boston, MA, USA d Department of Otolaryngology, Eaton-Peabody Laboratory, Massachusetts Eye and Ear In¢rmary, Boston, MA 02114, USA Received 3 July 1999; accepted 29 August 2000 Abstract Recordings in experimental animals have detailed the tonotopic organization of auditory cortex, including the presence of multiple tonotopic maps. In contrast, relatively little is known about tonotopy within human auditory cortex, for which even the number and location of tonotopic maps remains unclear. The present study begins to develop a more complete picture of cortical tonotopic organization in humans using functional magnetic resonance imaging, a technique that enables the non-invasive localization of neural activity in the brain. Subjects were imaged while listening to lower- (below 660 Hz) and higher- (above 2490 Hz) frequency stimuli presented alternately and at moderate intensity. Multiple regions on the superior temporal lobe exhibited responses that depended upon stimulus spectral content. Eight of these `frequency-dependent response regions' (FDRRs) were identified repeatedly across subjects. Four of the FDRRs exhibited a greater response to higher frequencies, and four exhibited a greater response to lower frequencies. Based upon the location of the eight FDRRs, a correspondence is proposed between FDRRs and anatomically defined cortical areas on the human superior temporal lobe. Our findings suggest that a larger number of tonotopically organized areas exist (i.e., four or more) in the human auditory cortex than was previously recognized. ß 2000 Elsevier Science B.V. All rights reserved. Key words: Auditory cortex; Functional magnetic resonance imaging; Tonotopy; Human 1. Introduction strated in structures throughout the auditory pathway, including auditory cortex (e.g., Rose et al., 1959; Wool- Tonotopy, an ordered mapping of neuronal fre- sey, 1971; Guinan et al., 1972; Merzenich and Reid, quency sensitivity to spatial location, has been demon- 1974). These demonstrations have typically involved re- cording from single or multiple units in non-human species and relating the acoustic frequency yielding the lowest response threshold (best frequency, BF) to * Corresponding author. Present address: School of Electrical and unit position (e.g., Hind, 1960; Merzenich and Brugge, Computer Engineering, Purdue University, West Lafayette, IN 47907, USA. Tel.: +1 (765) 494 5475; Fax: +1 (765) 494 6440; 1973; Merzenich et al., 1976; Reale and Imig, 1980; E-mail: [email protected] McMullen and Glaser, 1982; Sally and Kelly, 1988; Morel et al., 1993). Although neuronal responses in 1 Portions of this work were presented at the annual meeting of the International Society for Magnetic Resonance in Medicine (1996) and auditory cortex can depend on stimulus frequency in the annual meeting of the Association for Research in Otolaryngology complex ways, a single BF can usually be assigned to (1997). neurons in primary and certain non-primary areas (e.g., 2 Present address: University of Connecticut Health Center, Department of Neurology, 263 Farmington Ave., Farmington, CT Hind, 1960; Sutter and Schreiner, 1991; Rauschecker et 06030-1845, USA. al., 1995). Within cortical areas that are tonotopically 0378-5955 / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved. PII: S0378-5955(00)00203-3 HEARES 3571 1-11-00 Cyaan Magenta Geel Zwart 226 T.M. Talavage et al. / Hearing Research 150 (2000) 225^244 organized, BF varies fairly systematically from low to tinen et al., 1993; Cansino et al., 1994; Huotilainen et high along the cortical surface, but is relatively constant al., 1995; Verkindt et al., 1995; Roberts and Poeppel, across cortical depth (e.g., Merzenich et al., 1975; Imig 1996; Diesch and Luce, 1997; Lu«tkenho«ner and et al., 1977; Rauschecker et al., 1995). The overall pic- Steinstra«ter, 1998; Mu«hlnickel et al., 1998). Studies tak- ture of cortical frequency organization from the animal ing this approach typically used moderate-intensity, literature is one of multiple tonotopic maps distributed narrow-band stimuli based on the hypothesis that (a) over the cortical surface. neurons responding to these stimuli would have a nar- In contrast with the extensive body of data docu- row range of BFs and therefore occupy a limited extent menting the tonotopic organization of the auditory of any underlying tonotopic map, and (b) the location pathway in animals, far less is known about this funda- of the responding neurons would move systematically mental organizing principle in humans. This is true even from one end of the map to the other with systematic for auditory cortical areas residing on the superior tem- increases or decreases in acoustic stimulus frequency. It poral lobe, perhaps the most studied part of the human has been reported that the position of brain activity central auditory system. Based on physiological record- generating several evoked response components ings, lesion studies and functional imaging, it is known changes systematically with frequency (Romani et al., that widespread areas on the human superior temporal 1982a,b; Elberling et al., 1982; Pelizzone et al., 1985; lobe respond to sound and play a critical role in the Pantev et al., 1988, 1990, 1991, 1994, 1995, 1996; Ya- perception of acoustic stimuli (e.g., Celesia, 1976; Tra- mamoto et al., 1988, 1992; Tiitinen et al., 1993; Can- mo et al., 1990; Zatorre et al., 1992; Binder et al., sino et al., 1994; Huotilainen et al., 1995; Diesch and 1994). The primary auditory cortex, which lies deep Luce, 1997; Lu«tkenho«ner and Steinstra«ter, 1998; Mu«hl- within the Sylvian ¢ssure on the medial two-thirds of nickel et al., 1998). In some studies examining more Heschl's gyrus, is an area exhibiting short-latency re- than one response component in the same individuals, sponses to transient acoustic stimuli (a de¢ning feature the generators of the various components were localized of primary cortex) as well as cytoarchitectonic and im- to di¡erent parts of the superior temporal lobe and munostaining properties typical of primary sensory cor- each showed a systematic relationship between position tical areas (e.g., Celesia, 1976; Galaburda and Sanides, and stimulus frequency (Pantev et al., 1994, 1995, 1996; 1980; Lie¨geois-Chauvel et al., 1991; Rademacher et al., Diesch and Luce, 1997; Lu«tkenho«ner and Steinstra«ter, 1993; Rivier and Clarke, 1997). The surrounding, 1998). These ¢ndings indicate that human auditory cor- acoustically responsive areas of the superior temporal tex includes more than one tonotopically organized lobe are anatomically and physiologically di¡erentiable area. from the primary area, and from each other (e.g., Gal- Several studies have examined human cortical fre- aburda and Sanides, 1980; Lie¨geois-Chauvel et al., quency organization using positron emission tomogra- 1994; Rivier and Clarke, 1997; Howard et al., 2000). phy (PET) or functional magnetic resonance imaging If, as in many animal species, the various di¡erentiable (fMRI), techniques that provide spatial maps of brain areas contain one or more frequency-to-place mappings activation (Lauter et al., 1985; Wessinger et al., 1997; (e.g., Imig and Reale, 1980; Morel et al., 1993; Kosaki Bilecen et al., 1998; Lockwood et al., 1999; Yang et al., et al., 1997; Rauschecker et al., 1997), one would expect 2000). These maps show changes in blood £ow (PET) there to be multiple tonotopic representations of the or blood oxygenation (fMRI) that re£ect changes in audible frequency range on the surface of the human neural activity (e.g., in response to a sensory stimulus; superior temporal lobe. Fox and Raichle, 1986; Fox et al., 1988; Bandettini et Evidence that parts of human auditory cortex are al., 1992; Kwong et al., 1992; Ogawa et al., 1992). The tonotopically organized has been provided by previous PET and fMRI studies examining cortical frequency studies using a variety of techniques. Some of this evi- organization employed essentially the same strategy as dence comes from single unit recordings from auditory the magnetic and electric recording studies: subjects cortex in humans (Howard et al., 1996). However, the were stimulated with band-limited sound with the idea majority derives from studies using non-invasive meth- that the resulting activity would occupy di¡erent parts ods and provides somewhat di¡erent information about of auditory cortex. Each of the studies examined the cortical activity. The most extensively used approach response to two or three stimulus frequencies and re- has involved recording sound-evoked magnetic or elec- ported either displacements in the volume of activation tric responses over the surface of the head and localiz- for di¡erent frequencies, or separable sites of maximal ing the brain activity generating these responses (Ro- activation ^ ¢ndings consistent with an underlying to- mani et al., 1982a,b; Arlinger et al., 1982; Elberling et notopic organization. al., 1982; Pelizzone et al., 1985; Pantev et al., 1988, While it is clear that human auditory cortex is tono- 1990, 1991, 1994, 1995, 1996; Yamamoto et al., 1988, topically organized, the number of tonotopically organ- 1992; Bertrand et al., 1991; Jacobson et al., 1992; Tii- ized areas, the spatial arrangement of these areas, and HEARES 3571 1-11-00 Cyaan Magenta Geel Zwart T.M. Talavage et al. / Hearing Research 150 (2000) 225^244 227 their relationship to the cortical anatomy remains 2. Materials and methods largely unresolved. Although magnetic and electric re- cordings have indicated more than one tonotopically 2.1.