Auditory Frequency Analysis, Neural

Speech, music, and every other sound one hears can be described by the pattern of amplitude at different frequencies. This pattern, which is called the sound's spectrum, distinguishes one sound from others. In human speech, for example, the vowel in the word heat sounds different from the vowel in hat because amplitude peaks occur at different frequencies in the spectrum of the first vowel. Frequency analysis, described in this entry, refers to the ability to process different regions of the spectrum separately. This ability allows a person to discriminate two tones of different frequencies, but more importantly, it allows the locations of the amplitude peaks in the spectrum of any sound to be encoded and represented in the brain. Frequency analysis begins in the cochlea, where the spectrum of the acoustic signal is converted to a representation called a place code for frequency. The place code for frequency that is created in the cochlea is preserved as the signal is represented in the responses of neurons and processed in the brain. Throughout the auditory system, the representation of frequency by a neural place code is the predominant organizational principle.

Sound arriving at the eardrum is transmitted through the middle ear into the cochlea, where it causes a long, narrow structure called the basilar membrane to vibrate. Because of the way the stiffness of the basilar membrane varies along its length, it responds to a sine wave stimulus with a traveling wave whose maximum displacement occurs at a location that varies systematically with the frequency of the tone. Each location along the basilar membrane responds best to a restricted range of frequencies; that is, each location is “frequency-selective.” The frequency selectivity of any single location can be described by its tuning curve, which shows the pattern of thresholds as a function of stimulus frequency. Each threshold is defined as the level of a sine-wave stimulus that displaces the basilar membrane by a fixed amount. Tuning curves are V-shaped, and the frequency for which the threshold is lowest is called the characteristic frequency (CF). Thresholds increase rapidly for frequencies above CF, and increase more slowly for frequencies below CF. This pattern of thresholds means that each location responds to some frequencies but not others.

The mechanical frequency analysis accomplished by the basilar membrane is enhanced by the action of outer hair cells (OHCs). When stimulated by low-amplitude vibration of the basilar membrane, OHCs generate a mechanical response of their own. The effect of this “electromotile” [Page 158]response is called the cochlear amplifier. The feedback provided by the cochlear amplifier increases sensitivity to low-amplitude sounds near the CF and improves frequency selectivity.

The next step in frequency analysis occurs when the vibration of the basilar membrane is converted into neural activity. Motion of the basilar membrane stimulates inner hair cells (IHCs). IHCs in turn stimulate auditory nerve fibers (ANFs), the neurons that carry information about sound from the cochlea to the brain. Because each ANF receives input from a single IHC, it is also frequency-selective, with a tuning curve that is essentially the same as the tuning curve of the basilar membrane location that it innervates. A sine-wave stimulus will activate a subset of ANFs that has CFs that are close to the frequency of the tone. When the frequency of the tone changes, it activates a different subset of ANFs. This change in the place of activation provides the neural basis for frequency discrimination. Of course, most sounds are more complex than sine-wave tones. For a sound like one of the vowels mentioned previously, ANFs whose CFs are close to the frequencies of peaks in the vowel's spectrum will be activated most strongly. When the sound's spectrum changes, so that spectral peaks occur at different frequencies, the locations of peaks of neural activation will also change. This description leaves out many details, especially the contributions of cochlear nonlinearities, but in general, the pattern of activation as a function of place provides the basis for neural frequency analysis.

...