How We Hear
Listening is essential for understanding human speech as well as for emotional communication and appreciation of music. It is fundamental to clinical empathy and psychodynamic psychotherapy. Animals listen to locate and identify predators, prey, and kin. Much neuroscientific research to date has focused on sound localization, which is of less clinical interest than more complex functions like understanding speech. Nevertheless, the emerging picture of early auditory processing is a fascinating example of the integrated systems of complex high-speed multilevel connections evolution has created. And it is very different from the sound-processing software used in hearing aids and speech recognition software.
Most of this discussion is based on Donata Oertel’s and Allison J. Doupe’s chapter on the auditory central nervous system in the textbook, Principles of Neural Science.
The brain uses the difference in time of arrival and sound intensity between the two ears to locate sound sources In addition, the asymmetries of the external ears introduce distortions, as does interference by our heads and shoulders; these provide more spatial information, which is coordinated at the level of the midbrain with information from the visual system.
The cochlea mechanically transforms sound waves in the air into waves in the inner ear’s fluids. The hair cells then transform this information into electrochemical impulses in the cochlear nerve. Since nerve cells produce all-or-nothing discharges, information about the quality and location of sounds is conveyed by which and how many neurons fire and by the frequency of discharges.
The neural path to the auditory cortex is more complex than in the visual system, which has one principal waystation, the lateral geniculate nucleus of the thalamus, between the retina and the primary visual cortex. Cochlear nerve cells project to the cochlear ganglia in the medulla, whose cells project via several parallel pathways to the superior olivary nuclei and the nuclei of the lateral lemniscus in the pons, as well as to the inferior colliculus in the midbrain. Inputs from all these brainstem levels come together in the inferior colliculus. That nucleus projects to the superior colliculus and to the medial geniculate nucleus of the thalamus, which projects to the primary auditory cortex in the medial temporal lobe. Several specialized types of neurons and glial cells are involved in processing information at these levels. Much of the input on both sides of the brain comes from both ears.
Individual hair cells respond to specific frequencies. In addition, the early stages analyze the temporal structure of sounds. How the system recognizes sound intensity is less clear; it appears to involve nonlinear response characteristics of the cochlear nerve as well as higher levels. Pleasant-sounding music produces regular impulses in the auditory nerve, while dissonant chords whose frequencies are close together produce impulses which interfere with one another.
Two pathways connect the inferior colliculus to the auditory cortex via the medial geniculate nucleus of the thalamus. The auditory cortex comprises several areas in the medial temporal lobe. The primary auditory cortex is organized tonotopically like the lower auditory processing areas. Interestingly, the direction of tonotopic organization in adjacent cortical fields reverses at the boundaries between the fields. The primary auditory cortical fields respond to pure tones, while the areas around them prefer complex sounds.
There is extensive projection from the cortex back to the medial geniculate body of the thalamus (ten times as many fibers project downward as upward!) Cortical neurons also project to midbrain auditory areas and even to the cochlear nucleus. Activation of these downward-projecting fibers improves auditory processing at the lower levels, so part of their function may be to fine-tune the system and improve its ability to discriminate sounds.
Unlike in the visual system, the auditory system from the cochlear nucleus all the way up to the primary auditory cortex is laid out tonotopically, that is, according to sound frequency rather than spatial location. It can detect time differences as short as ten microseconds. Like in the visual system, different aspects of auditory information are processed in discrete circuits, which converge at higher cortical levels into complex representations of sound.
While the primary auditory cortex’s is organized tonotopically in two dimensions, the columnar organization of cells in the layers of the cortex appears to facilitate sound localization. Cells are excited by input from both ears, with stronger input from the opposite side. Alternating columns receive inhibitory input from the ipsilateral ear. It is thus a three-dimensional array, with frequency in one plane and information about localization perpendicular to it. Some cortical neurons respond to broad and some to narrow ranges of frequencies.
In general, caudal areas of the auditory cortex, which are concerned with localization and movement of sound sources, project on to other dorsal and caudal areas of the temporal lobe, while rostral areas, which are more concerned with tone and pattern of sounds, project to rostral and ventral areas. These in turn project to different areas of the frontal lobe. This is similar to the visual system, where a dorsal stream is concerned with object localization and a ventral stream with object identification. However, some areas appear to participate in both pathways.
We are familiar with the critical period in development for acquiring language: it is very difficult to learn to speak a foreign language without an accent after puberty. Animal experiments have found such critical periods for development of the sound localization circuits in the midbrain. This is very early in the processing hierarchy—presumably far below any conscious awareness. Nevertheless, the auditory cortex has some plasticity, even in adults.
Learning ability also increases with motivation: adult owls allowed to hunt for live mice adapt more readily to prismatic lenses, which distort the localization of prey, than adults who are not allowed to hunt, but less readily than juveniles.
Oertel and Doupe note that both subcortical and cortical pathways respond to acoustic stimuli, with subcortical pathways from the medical geniculate nucleus to the amygdala mediating rapid responses to fearful stimuli, while responses requiring discrimination of more complex stimuli operate via slower pathways involving the cortex.
Hearing and imitating the speech of others is essential to learning to speak. Such vocal learning has been found in cetaceans, bats, songbirds, parrots, and hummingbirds, but interestingly, not in other mammals, including the great apes. In songbirds, hearing a bird’s own vocalization is essential for mastering the song of a tutor bird. Vocal learning is subject to the critical learning periods I mentioned above, and is influenced by the motivational state of the animal and by hormones such as testosterone.
The auditory system is a remarkable example of the kind of flexible, efficient, high-speed processing systems evolution has created. It is quite different from human-designed auditory processing systems: hearing aids operate by breaking down sound waves into mathematical wave functions which when added together approximate the sound being duplicated (these are called Fourier transforms,) and then amplifying some frequencies and suppressing others to improve hearing. Speech recognition software starts with Fourier transforms and uses various combinations of phonetic dictionaries, feature analyses, language models, and learning programs (neural networks) to transcribe and to a limited extent understand and generate speech.
Human listening relies on this intricate, strange, multiple-level arrangement which is consistent with the ad-hoc nature of its evolutionary development. While this has many advantages over other systems, it also has limitations. The critical period for learning languages is one. Experimentalists have found auditory misperceptions similar to visual illusions. It will be interesting to think about how the auditory system’s strengths and limitations play out in clinical work.
Donata Oertel and Allison J. Doupe, The Auditory Central Nervous System, in Principles of Neural Science, Fifth Edition, Eric R. Kandel, James H. Schwartz, Thomas M. Jessell, Steven A. Siegelbaum, and A.J. Hudspeth, editors, 2013. pp 682-711.