Response: Sound analysis in auditory cortex – from temporal decomposition to perception

We are grateful for Robert Zatorre’s insightful endorse-ment [1] of our view regarding the significance oftemporal decomposition of sensory signals in theauditory system [2]. Here, we would like to emphasizesome of the points raised in Zatorre’s excellent reviewthat appear to be particularly specific for audition, ascompared with vision. In the auditory domain, allsensory input is received in series, and it is the timedimension that is crucial for the construction ofauditory scenes. Consider that a painting by VanGogh or the face of a familiar person can be recognizedwithin a few dozens of milliseconds, whereas a melodyof Mozart or the voice of a loved one requires severalseconds of listening before it can be recognized. A briefsnapshot of a complex sound is often nothing but anincoherent burst of noise, whereas a visual snapshot ofthe same duration can easily produce a coherentpercept of useful information.As outlined by Zatorre, auditory neuroscience hashistorically taken a back seat to vision. However,recent lines of research have identified analogousvisual and auditory functional streams in the proces-sing of ‘what’ and ‘where’ information [3,4],andintherecognition of faces [5] and voices [6]. Researchers arealso starting to delineate the polymodal interactionbetween these two systems as they relate to theperceptionofspaceandtime.Tonamejusttwoexcitingtemporal examples, recent work has examined the casein which time is a common dimension between auditionand vision, being represented in processing of visualand auditory motion in the monkey and the humanbrain [7], and also how one sensory input (e.g. acoustic)can influence perceptual quality of another input (e.g.visual) in terms of cross-modal binding [8]. An excitingspatial example lies in unraveling the mechanisms bywhich auditory and visual space maps converge, sothat visually acquired space maps can be recruited andused by the auditory system when visual information isnot available (e.g. in darkness), both in animals[9] andhumans [10]. However, we would like to emphasizethat, although there are certainly analogies betweenhearing and seeing, which will be reflected in theirrespective neural substrates, these analogies are notlikely to be absolute. In ecological terms, audition andvision subserve different specific operations that haveevolved to solve different evolutionary problems. Forexample, navigational abilities such as traversingterrain, catching prey, guiding self-motion and perceiv-ing looming objects are informed by both vision andaudition. A specific advantage of the auditory system isthat it can provide spatial information about soundsources that are occluded, out of the line of sight or inconditions such as darkness or fog that make viewingimpossible. These abilities are particularly importantfor survival considering that an organism should beable to detect moving objects rapidly – particularlyobjects that are approaching. Although extensiveresearch has been conducted on the neural principlesunderpinning visual looming [11], little is known aboutlooming and the auditory system. Nevertheless, thisgap is beginning to be filled by an emerging body ofresearch examining information from auditory cortexunit recordings [12], psychophysical evaluations inanimals [13] and humans [14], and functional brainimaging [15].The ability of the auditory system to processtemporal information is at least an order of magnitudegreater than that of other sensory systems. At the levelof the cochlea, phase-locked neural activity can reachfrequency levels of up to several kilohertz; however,such phase locking is not observed at higher stages ofthe ascending auditory pathway. Thus, the temporalpattern of sound is recoded and represented usingother, as yet incompletely understood, principles. Inthe ascending auditory pathway of the human brain,for instance, recent evidence suggests that one of theseprinciples might be the conversion of temporal regu-larities into local neural activity levels [16]; however,the signal reconstruction principles remain to bedelineated as they have been in animal thalamocorticalcircuits [17] and synaptic cortical networks [18].Onahigher level of processing, one of the key aspects ofhow the brain constructs coherent concepts of sequen-tially received sensory information lies in the under-standing of auditory memory mechanisms, and itremains to be seen what role sensory memory tracesplay in assembling temporal information into anauditory experience. Functional magnetic-resonanceimaging (fMRI) provides not only excellent spatial