NN0508 NV.indd

For a lot of processing in the brain, two pathways are clearly better than one. In the visual system, two anatomically and functionally distinct pathways are thought to be specialized for processing information about where an object is versus what an object is. In a classic behavioral experiment1, dorsal lesions in posterior parietal cortex of monkeys resulted in worse performance only when the monkeys were required to make a spatial judgment (choosing which one of two locations was closest to a ‘landmark’ cylindrical object). Lesions to ventral temporal or parietal lobes reduced performance when animals were required to pick the object that did not match an initial ‘sample’ object, but performance was unaffected in the spatial task. This double dissociation was taken as strong evidence in favor of the ‘what, where’ hypothesis. An alternative hypothesis is that these separate pathways mediate perception (what) and action (how), respectively2. Nevertheless, obtaining distinct outcomes by lesioning different areas implies some kind of functional segregation. An article in this issue by Lomber and Malhotra3 now provides convincing evidence that two different regions of the auditory cortex are necessary for carrying out two different tasks. Given that parallel functional systems seem to be a general feature of brain organization, wouldn’t we expect the auditory system to be organized in the same way? This is by no means a foregone conclusion: unlike in the visual system, where information is transmitted relatively directly to the cortex, much auditory processing, such as that for localization, takes place in the brainstem. Although there is clear evidence for separate dorsal and ventral anatomical pathways from auditory cortex to frontal regions in primates4, the evidence for functional differences is less convincing. Physiological recordings in animals4 and human neuroimaging5 are broadly consistent with the two-pathway hypothesis. For example, neurons in the posterior auditory field of cats are more sensitive to the spatial attributes of sounds6 than neurons in other auditory cortical regions. Neurons in the anterior auditory field are more sensitive to spectral modulations that are probably important for identifying cat vocalizations7, hinting at a ‘what’ pathway. Neurons in some of the putative ‘where’ (dorsolateral) parts of prefrontal cortex are most responsive to sound during localization tasks8. On the other hand, neurons in putative ‘what’ (ventrolateral) parts of prefrontal cortex and in other parts of the putative ‘where’ pathway (parietal cortex) code both spatial and nonspatial attributes9,10. These data suggest that there is more of a bias to regional coding of spatial or object-related properties of sound than a clear segregation. Nor is what/where the only possible functional distinction; another hypothesis is that the dorsal pathway is specialized for spectral dynamics11. In the new report3, posterior and anterior regions of the auditory cortex were separately and reversibly deactivated in the same cat while the cat was awake and listening and responding to sounds. Each cat was trained to perform three different tasks. In one task, it had to detect a short noise burst. In a second task, it was required to indicate where the sound came from (localize the sound in space) by approaching the appropriate loudspeaker. In the third task, the cat had to signal a change in a ‘morse-code’ type sequence of short and long noise bursts (discriminate between different sounds). The main finding (illustrated in Fig. 1) was that deactivation of the posterior auditory field on both sides of the brain impaired the cats’ ability to localize sounds, whereas detection and discrimination were not affected (a ‘where’ impairment). In contrast, when the anterior auditory field was deactivated, the cats lost their ability to distinguish between different sounds (a ‘what’ impairment), whereas detection and localization were unaffected. This double dissociation shows, for a specific pair of tasks, a clear functional segregation in the auditory cortical fields. It also supports the hypothesis that these regions are preferentially involved in spatial coding and object identification, respectively. An important feature of the Lomber and Malhotra study3 is that, unlike surgical or chemical lesions, deactivation by cooling is rapid and reversible in a few minutes. Once the cooling device is switched off, blood circulation rapidly returns the tissue to normal working temperature and functionality. This rapid reversibility makes the experiment very well controlled: it is easily repeatable The authors are in the MRC Institute of Hearing Research, University Park, Nottingham, NG7 2RD, UK. e-mail: [email protected] The need for a cool head: reversible inactivation reveals functional segregation in auditory cortex