Greg Detre
Wednesday, 24 October, 2001
Dr John Iles, week 5
Konishi & Knudsen�s (1977) study with owls� sound localisation abilities was an attempt to answer the question, �Why do we have two ears?�, given that one is more or less sufficient for identifying sounds. Most hearing creatures can identify where sounds are coming from using auditory cues alone, but owls (especially barn owls) so excel at the task that they can hunt in the dark. The localisation of sounds in space is achieved in the brain by comparison of differences in the intensity and timing of sounds received in each ear.
For example, it is possible to determine a sound's source direction by comparing Interaural Intensity Differences (IID) if the structures separating the two ears shadow the sound sufficiently (or if the distance is large enough for there to be differential attenuation). This effect is wavelength dependent (high frequency sound with shorter wavelengths are attenuated more and are shadowed more easily by the head).
In addition, if the ears are far enough apart, then the time of arrival of the sound at the two ears will differ and Interaural Time Differences (ITD) can be used, taking advantage of both Ongoing Time Disparities, OTD, and sound Onset Disparities, OD. Single-unit recording of cells in the ventral cochlear nucleus (Oertel) highlight two types of cell that appear to play contrasting roles. When depolarised by steady current pulse, the stellate cell generates a spaced series of action potentials at regular intervals, called a chopper response. In contrast, the bushy cell generates only one or two spikes at the beginning of the pulse, signaling the onset and timing.
Onset disparities relate to the duration of the delay between the sound being first heard by both ears. It depends on the distance between the two ears, the speed of sound and the location of the sound source. If the sound is along the midline, front or back, the delay would be zero. At 90� to the right or left, the inter-aural delay would be up to 50�s � between these extremes, there is a spectrum of interaural time difference.
Ongoing time disparities compare the minute phase shifts in the waveform of an ongoing sound caused by the sound reaching one ear lagging behind the other slightly. For low frequency sounds (<1400Hz), a continuous tone can be localised by phase difference. At higher frequencies, where the wavelength of the sound is less than the distance between the two ears, the phase or time difference of a continuous tone becomes ambiguous, because the phases could align along multiple cycles. At these frequencies, the head acts as a sound shield, reflecting and absorbing the shorter wavelengths of sound to produce an interaural intensity difference. Moreover, as discussed below, the brain appears to use neurons in concert to phase-lock onto different cycles of high frequency sounds, preserving the information about high-frequency sounds obtained in the cochlear.
Following Konishi & Knudsen�s (1977) original study, a great deal of work has been done on sound localisation in owls, especially the barn owl. They identified an area in the midbrain, where ten thousand 'space-specific' neurons would fire only when sounds were presented in particular location. These cells are topographically arrayed, such that aggregates of space-specific neurons pinpointed precise vertical and horizontal coordinates of the speaker, firing only when a tone was played at that location.
Payne had shown that the owls make use of their superb scotopic vision to pounce on mice in very dim light. However, they are still able to catch the mouse even if the lights are turned off as the owl is leaving its perch, as long as the mouse makes some sort of noise. This ability is impaired if the owl's ears are plugged.
Konishi & Knudsen used microelectrode recording while a remote-controlled speaker was moved around the owl's head in a dark, sound proof, anechoic room, imitating sounds the owl would hear in the wild. The owl had been trained to turn its head toward the direction of a sound, while the horizontal and vertical movement of the head was monitored by the variation in current of copper coils mounted around the owl�s head (the search coil technique).
Owls' ears are not symmetrical - the left one points down and the right one points up. A partial block of the left ear causes the animal to miss the speaker location by orienting the head just a little to the right and up from the actual location, and vice versa for the right ear. Like mammals, owls employ both interaural intensity and time differences. Konishi & Knudsen found that altering the timing between two sounds caused the owl to move his head in the horizontal plane, and that altering the intensity between sounds in the two ears caused the owl to move its head in the vertical plane
The owl's midbrain contains a map of sound location. Neurons located near the midline of the nucleus are sensitive to sounds to the left of the animal's head and neurons located near the lateral edge of the nucleus sensitive to sounds in front of the animal's head. The neurons' RFs overlap, providing a continual representation.
Each cell in the auditory nerve projects to two distinct brain nuclei: the nucleus magnocellularis and the nucleus angularis.� Here the information about timing and intensity splits into two pathways: the timing pathway includes the nucleus magnocellularis and laminar nucleus, and the intensity pathway includes the nucleus angularis and lateral lemniscal nuclei.
Cells from the nucleus magnocellularis project to the laminar nucleus, a bilateral structure in the central auditory pathway in the bird�s medulla, where the inputs from the two ears are first combined in order for timing calculations to be performed. This, the timing pathway of the brain, is organised tonotopically into phase-locked isofrequency zones. Carr vindicated Jeffress� (1948) model, showing that the auditory system (in man and owl) contains �coincidence detectors�, neurons which fire only when inputs from both ears are closely synchronised. Various lengths of �delay lines�, fibres anatomically organised so that their signal takes an arbitrary extra period of time to reach their destination, allow an array of such coincidence detectors to neurally encode a time differential between a signal reaching the two ears.
In the intensity pathway of the brain, neurons in the principal relay nucleus of this pathway are excited by contralateral stimulation and inhibited by ipsilateral stimulation of the ear. When both ears are stimulated equally, excitation and inhibition balance, and there is little response. When sound amplitude is decreased in one ear, neurons in the intensity pathway on the same side respond, since ipsilateral inhibition has been decreased. Neurons are arranged according to the specific interaural intensity difference to which they are most responsive.
At higher levels, the timing and intensity pathways converge, where there are neurons broadly tuned in frequency but specific in spatial localisation of sound.
In the mammalian brain, axons from the cochlear nuclei project to several brain stem auditory nuclei. From the cochlear nucleus, axons project along three pathways: the dorsal acoustic stria; the intermediate acoustic stria; and the trapezoid body. The trapezoid body contains fibres that go to superior olivary nuclei on both sides of the brain stem. The medial superior olive is concerned with sound localisation on the basis of interaural time differences � this nucleus is composed of spindle-shaped neurons with one medial and one lateral dendrite, which receive input from the contralateral and ipsilateral cochlear nuclei, respectively. The binaural cells in the medial superior olive are very sensitive to phase differences between continuous tones presented to the two ears. The lateral superior olive is concerned with interaural differences in sound intensity
Most experiments on auditory localisation have been concerned with horizontal and vertical positions of sound sources, ignoring the third dimension - distance. There has been little work on this since von Bekesy, until Bronkhorst and Houtgast's study using virtual sound technology (the cues provided by the head, the ears and the room are measured, digitally synthesised and mixed with the acoustic characteristics of the presenting headphones, making them indistinguishable from real sounds presented within rooms by distant loudspeakers), and Graziano et al.�s work with frontal cortex neurons in monkeys. Bronkhurst and Houtgast simulated source distances of a metre or more, while the Graziano et al. study focused on the near field. They showed that small distance processing can rely on monaural spectral cues and interaural level differences, but it is not yet clear whether auditory neurons in non-echolocating mammals are sensitive to the other cues available for more distant sound sources.
Moore and King consider the idea that perception (in all three major sensory systems) divides neatly into what and where processing pathways. This thesis is probably clearest and most defensible for the visual system, where support comes from lesion studies in monkeys, in which impairment of spatial abilities or object identification can be separated, and anatomical studies which show that connections in the visual pathways are fairly strongly segregated. Of course, the auditory system too must analyse both identity and location of stimuli, but it is not clear to what extent these are functionally and anatomically isolated.
As discussed above, the initial signals from the two cochleas are conveyed to a complex network of pathways and nuclei in the brainstem and thalamus, where spectral and temporal information is extracted to determine the identity and location (requiring binaural comparison in the brainstem) of the sound source. Romanski et al examined the connectivity of higher auditory cortical areas in macaque monkeys, using a combination of anatomical tracer dyes with electrophysiological recordings. Their results support the ventral/dorsal temporal/parietal what/where processing dichotomy, contributing to functionally distinct regions of the frontal lobe. Further parallelism is evident in the primate auditory cortex, which has three similar primary or primary-like areas, each tonotopically organised and receiving activating inputs directly from the auditory thalamus. The projections from these to 7 or 8 proposed fields seem to provide anatomical support for the beginnings of ventral and dorsal cortical processing streams - they project to largely different portions of the frontal lobe. However, the middle belt area makes connections to the frontal lobe that overlap those of the two putative streams, indicating possible intermediate or additional auditory streams (also analogous to the additional functional streams or 'streams within streams' found in the visual system).
In addition to parallel pathways, the auditory system has an extensive set of feedback connections, e.g. layer V in the auditory cortex sends axons back to the medial geniculate nucleus and layer VI to the inferior colliculus. The inferior colliculus sends recurrent fibres to the cochlear nucleus. A cluster of cells near the superior olivary complex gives rise to the efferent olivochlear bundle, which terminates either on the hair cells of the cochlea directly or on the afferent fibres innervating them. These connections may be important for regulating attention to particular sounds by modulating the transduction mechanism in the organ of Corti.
Like owls, the mammalian brain incorporates both interaural timing and intensity differences to gauge where sounds are coming from. The auditory system incorporates a huge number of topographic representations of one kind or another, as well as spiking synchronicity, to calculate and encode sound localisation. In the spatial acoustic maps in the higher levels of the auditory cortex, extraneous factors like volume and sound identity have been removed.