On pitch, the ear and the brain of the beholder. Focus on "neural coding of periodicity in marmoset auditory cortex.".

The pitch of a sound is perhaps one of its most salient properties. It is the pitch of sounds that allows us to distinguish and recognize melodies, and the salience of pitch and melody appears to be exploited by a strikingly wide variety of vertebrate life-forms, from song birds to “singing” hump-back whales all the way to human composers of advertising jingles. But while perceptually pitch has a great “immediacy,” physiologically pitch is a surprisingly complex phenomenon. It is sometimes said that the pitch of a sound is “related to its frequency content” but that relationship is anything but straightforward (Plack and Oxenham 2005). Many natural sounds contain large numbers of frequency components, and it is possible to produce an infinite variety of sounds with rather different frequency composition that nevertheless share the same pitch. Not all sounds evoke a clear pitch, but one thing that all sounds that do evoke a pitch percept have in common is that they are periodic, i.e., the waveform of the sound consists of a short “motif” that is repeated very rapidly over and over again, and the “speed” at which these repeats occur is the chief determinant of the perceived pitch value. Note that these repeated motifs are too brief to be perceived as distinct events. For humans to hear a clear musical pitch, their period must be roughly between 25 and 0.33 ms long (corresponding to pitches between ca. 40 and 3,000 Hz), and while the motifs needn’t be absolutely identical from one repeat to the next, they nevertheless have to be “similar.” If the pattern of repeated motifs becomes increasingly less regular, then the sound becomes increasingly noise-like, and the pitch becomes increasingly less salient until it becomes indistinguishable. Perceived pitch can therefore be thought of as a measure of the underlying “regularity” of a sound. The requirement that sounds with a clear pitch must be periodic constrains their frequency content (i.e., their Fourier spectrum cannot take just any shape). A sound wave can only be periodic if all the sine wave components that make up the periodic sound “conform to the common underlying rhythm.” In other words, only sine waves that can fit a whole number of cycles into one period of the overall sound can be part of the spectrum of a periodic sound. Consequently, the spectra of periodic sounds are always composed of the harmonics (integer multiples) of a given “fundamental” frequency. So-called “place theories of pitch” assume that these harmonically spaced maxima in the spectrum of periodic sounds produce distinct peaks of excitation along the basilar membrane in the inner ear, and that the position and spacing of these peaks is then interpreted by the brain and determines the perceived pitch. However, the fact that periodic sounds must be composed of harmonically related frequencies does not constrain them very tightly. For example, periodic sounds can vary greatly in how strongly the various harmonics are represented. In fact, while some periodic sounds that produce strong pitch percepts contain dozens or hundreds of harmonics within the audible frequency range, others contain only a very small number of harmonics, and pure tones, perhaps the “archetypal” periodic sounds, contain only a single one. Usually periodic sounds which have a lot of energy at higher harmonics sound “brighter” than those that do not, but while changing the relative amplitude of high and low harmonics can dramatically change the sound’s “timbre,” it usually does not change its pitch. In fact, in sounds with several prominent higher harmonics one can even reduce the amplitude of the lowest harmonic, the fundamental, to nothing, and while such “missing fundamental” stimuli will sound “less rich” than equivalent sounds that do carry significant sound energy at the fundamental, their pitch still remains the same. And one need not stop at removing only the fundamental; harmonic complexes missing not several of the lowest harmonics may still evoke a recognizable pitch at the missing fundamental. Place theories of pitch have a hard time explaining how the perceived pitch can remain stable in the face of such radical alterations of a sound’s spectrum. Alternative explanations for the basis of pitch known as “timing theories” are therefore gaining widespread acceptance. These propose that our brains derive pitch mostly from “time domain” cues that are extracted from the phase locked temporal firing patterns of auditory nerve fibers (Cariani and Delgutte 1996a,b). But whether pitch information enters the auditory system mostly through temporally or through place-coded information, much additional neural processing is clearly required in either case before a clear and stable pitch percept can emerge. The neural structures and mechanisms that underpin our pitch perception are still only poorly understood and are the subject of much active research. Indeed in two recent papers, Bendor and Wang have described discoveries made during recordings from the auditory cortex of the marmoset that shed new light on this issue. Their first paper (Bendor and Wang 2005) documented neurons that appeared tuned to the periodicity of harmonic tone complexes, even if these tone complexes were “missing fundamental” stimuli. The responses of these neurons are therefore apparently stable when the sound spectrum changes, as long as the periodicity of the sound remains constant, just as the perceived pitch remains constant under identical manipulations. Neurons exhibiting this stable periodicity tuning seemed to cluster anatomically in a location near the low frequency border of primary auditory cortex, an area that Bendor and Wang have therefore designated as a putative “pitch area.” In a second recent paper, Bendor and Wang (2010) go on to show that these putative “pitch neurons” not only respond to missing fundamental stimuli but appear tuned Address for reprint requests and other correspondence: J. Schnupp, Dept. of Physiology, Anatomy and Genetics, Sherrignton Building, University of Oxford, Parks Road, Oxford OX1 3PT, UK (E-mail: [email protected]). J Neurophysiol 103: 1708–1711, 2010. First published February 17, 2010; doi:10.1152/jn.00182.2010.