516 Cortical processing of complex sounds Josef P Rauschecker Work on the functional organization of auditory cortex in nonhuman primates has recently gained increasing attention. Neurophysiological studies using complex stimuli, combined with anatomical tract tracing, reveal a hierarchy of cortical processing comparable to other sensory systems. On the basis of these findings from animal studies, together with the advent of modern neuroimaging methods used in human cortex, the field of auditory neuroscience could soon arrive at a detailed understanding of the cortical representation of complex sounds, including speech. Addresses Georgetown Institute for Cognitive and Computational Sciences, Georgetown University Medical Center, 3970 Reservoir Road NW, Washington DC 20007-2197, USA; e-mails: [email protected] and [email protected] Current Opinion in Neurobiology 1998, 8:516-521 http://biomednet.com/elecref/0959438800800516 0 Current Biology Publications ISSN 0959-4388 Abbreviations Al primary audltory cortex AL anterolateral area CL caudolateral area CM caudomedial area FM frequency modulation, frequency-modulated fMRl functional magnetic resonance Imaging ML mlddle lateral area PET positron emission tomography R rostra1 area STG supenor temporal gyrus Introduction Chmples natural sounds arrive at the ear as a mixture of multiple frequencies. As has been known for some time, the cochlea of the inner ear efficiently breaks down thcsc complex, time-varying frequency spectra into narron’ bands [1,2]. Keural processing in the ascending auditory pathways, amplified by positive feedback from corticofugal projections, achieves increasingly finer tuning of single neurons for a best frequency [3*,4*]. Preservation of neighborhood relationships between best frequencies Icuds to the lvell-known tonotopic representation in primary auditory cortex, analogous to the topographic maps found in other sensory systems, such as retinotopy in the visual cortex and somatotopy in the somatosensory cortex. R’ork on the more peripheral stations of the auditor) system using pure-tone stimuli, however elegant, does not explain how information about the tonal composition of complex sounds is re-assembled, how this information gives rise to perceptual and cognitive performance, and how memory traces for music or voices that can be recognized among thousands of others are formed. The obvious place in the brain in which these higher functions arc performed is the cerebral cortex. If the ultimate goal is to understand the neural basis of human auditory cognition, including spoken language, then it may be advantageous to study animal models Lvith brains most similar (i.e. closest in evolution) to those of humans. Studies of nonhuman primates have led to a remarkable amount of knowledge about the cortical functional or- ganization underlying visual perception. In addition to a detailed understanding of the primary visual cortex (Vl) [S]. these studies have revealed the cxistcnce of multiple representations, seemingly specialized for the processing of particular aspects of the visual rvorld [6,7]. One influential suggestion [8] has been that the cortical visual pathuays are organized into t\\‘o processing streams: a ventral stream, which leads into the inferior temporal cortex, for visual pattern recognition or object identification; and a dorsal stream, which leads into the parietal lobe, for visual motion and spatial analysis. In this review, I will discuss recent results of single-unit recordings from primary and nonprimary auditory cortex in nonhuman primates, as well as anatomical tracer studies in the same animals. ‘I’he responses of neurons to \wious types of complex sounds, including species-specific vocal- izations, \viII receive particular attention, and the neural mechanisms of how selectivity for such sounds is attained wiII be discussed in relation to other species. Finally, I will present recent results of functional imaging studies from auditory cortex in humans and compare them \vith the findings from nonhuman primates. All these results taken together provide initial evidence for the existence of a dorsal stream for the processing of auditory spatial information and a ventral stream for the processing of auditory patterns, including communication sounds and speech. Multiple areas in monkey auditory cortex Perceptually, the auditory system has to deal with the same basic problems as the visual system: that is, identify patterns or objects and determine the spatial location of a stimulus. Both functions are achieved by integrating auditory information across its thvo major dimensions, frequency and time. By comparison with the visual system, much less is known about the functional organization of higher auditory pathways, even though a consider- able amount of anatomical and gross electrophysiological information was collected early on [9,10]. The studies by Pandya and colleagues (see e.g. [lo]) divided the auditory cortex (like other sensory cortices) into core and belt areas on the basis of cytoarchitectonics and connectivity. The first microelectrode mapping study of rhesus monkey auditory cortex was published a quarterof a century ago by hlerzenich and Brugge [ 111. The): described several tonotopic areas on the supratemporal plane. Some of these areas were later characterized with modern histochemical techniques [12-l-l]. The existence of 2-R core areas and several belt and parabelt areas has now also been confirmed on the basis of cortico-cortical connectivity [ lS**]. hly collcagues and I [lf,“] have recently investigated the connections from the thalamus to the tonotopic areas on the supratcmporal plant using 3 combination of lesion and anatomical tracing techniques. We