History

Alfonso Corti, an Italian anatomist working in Würzburg, published the first detailed description of the sensory epithelium of the cochlea in 1851. Working with hardened preparations and early compound microscopes, he identified the rows of pillar cells, the arch they form (the tunnel of Corti), and the hair-bearing cells on either side. He did not know what the cells did — the connection between hair cells and hearing would take another half-century — but the anatomy was precise enough that the structure still carries his name. Corti left science shortly afterward and returned to manage his family's estate; the work that made him famous was essentially his only major publication.

Hermann von Helmholtz proposed in his 1863 *Die Lehre von den Tonempfindungen* (On the Sensations of Tone) that the cochlea performs frequency analysis by resonance: that structures of graded stiffness along the basilar membrane act as a bank of tuned resonators, each responding selectively to its matched frequency. The idea was motivated by his own experiments with tuning forks and acoustic resonators, and by the anatomical observation (then recent) that the basilar membrane is narrow and stiff at the base, wide and compliant at the apex. Helmholtz imagined the transverse fibers of the basilar membrane as independent strings, each tuned to a different pitch — a "piano inside the ear." The resonance theory was qualitatively correct: the cochlea is indeed a frequency analyser, and basilar-membrane stiffness does grade from base to apex. But the mechanism is not independent resonators. Békésy showed in the 1940s that the membrane supports a traveling wave whose envelope peaks at the frequency-matched place. The modern picture — active, nonlinear, and wave-based — descends from Helmholtz's insight but replaces his piano strings with coupled fluid-membrane dynamics.

Hermann von Helmholtz, in his 1867 *Handbuch der physiologischen Optik* (Treatise on Physiological Optics), argued that perception is not a passive registration of sensory data but an active process of *unconscious inference* — the brain automatically and involuntarily constructs hypotheses about the external world from incomplete and ambiguous sensory evidence. The idea was radical for its time: the dominant view held that perception was a direct readout of stimulus properties. Helmholtz's framework anticipated by more than a century the Bayesian and predictive-coding accounts of perception that now dominate computational neuroscience. The modern formulation — that the brain maintains a generative model of the world and updates it via prediction errors — is essentially Helmholtz's unconscious inference rewritten in the language of probability theory. The hearing book's treatment of Bayesian perception in this chapter is a direct descendant of Helmholtz's 1867 insight.

Lord Rayleigh (John William Strutt) proposed in 1907 that the auditory system uses two distinct cues to localise sound in the horizontal plane: interaural time differences (ITDs) for low frequencies, and interaural level differences (ILDs) for high frequencies. The crossover occurs near 1500 Hz, where the wavelength is roughly twice the head diameter. Below this, the head is too small to cast a significant acoustic shadow, but phase differences between the ears are unambiguous. Above it, the head shadows effectively, but the phase cue becomes ambiguous because the wavelength fits inside the interaural path more than once. Rayleigh demonstrated the theory with elegant psychophysical experiments using tuning forks and a rotating chair. The duplex theory remains the organising framework for spatial hearing, though modern work has added spectral cues (pinna filtering for elevation) and temporal-envelope ITDs at high frequencies that Rayleigh's original formulation did not anticipate.

Georg von Békésy, a Hungarian physicist working at the Budapest telephone exchange, began his cochlear experiments in 1928 with a practical question: what limits the frequency range of telephone communication? His approach was direct and physical — he built large-scale mechanical models of the cochlea, then moved to cadaveric human cochleas observed under stroboscopic illumination. By the late 1940s he had shown that sound entering the cochlea produces a traveling wave on the basilar membrane: a displacement pattern that propagates from base to apex, grows in amplitude as it approaches the place tuned to the stimulus frequency, peaks sharply there, and dies out beyond. The traveling wave replaced Helmholtz's resonance theory of independent fibers with a hydrodynamic picture: the membrane and the fluid are coupled, and the wave's behavior is set by the position-dependent impedance of the membrane. Békésy received the Nobel Prize in Physiology or Medicine in 1961 — the only physicist to win in that category. His measurements, made on cadaveric cochleas with passive mechanics, showed broad tuning; the sharp frequency selectivity of the living cochlea would require the discovery of the cochlear amplifier two decades later.

In 1930 Ernest Glen Wever and Charles Bray at Princeton placed an electrode on a cat's auditory nerve and connected it, via an amplifier, to a loudspeaker in another room. When they spoke into the cat's ear, their words were reproduced in the loudspeaker with startling fidelity. The electrical signal they recorded — which they initially interpreted as a neural response — turned out to be mostly the cochlear microphonic: an extracellular potential generated by hair-cell transduction currents that follows the acoustic waveform cycle by cycle. The Wever-Bray experiment demonstrated that the cochlea produces an electrical signal that mirrors the acoustic input with remarkable precision. Disentangling the cochlear microphonic (generated by outer hair cells) from the compound action potential (generated by the auditory nerve) took another decade of work by Hallowell Davis and others. The cochlear microphonic remains a clinical tool today — it is recorded during electrocochleography and cochlear-implant surgery.

In 1933 Harvey Fletcher and Wilden Munson at Bell Telephone Laboratories published the first systematic measurement of equal-loudness contours — curves in the frequency-intensity plane along which tones of different frequencies sound equally loud. The work required thousands of loudness-matching judgments from trained listeners and produced the now-familiar family of curves showing that human hearing is most sensitive near 3--4 kHz (the ear-canal resonance) and falls off steeply at low and very high frequencies. Fletcher and Munson's contours were adopted as an international standard (ISO 226) and are the basis of the A-weighting filter used in environmental noise measurement. The ear-canal and middle-ear transfer functions — the physics developed in this chapter — directly explain the shape of the contours: the 3 kHz sensitivity peak is the quarter-wave resonance of the ear canal, and the low-frequency rolloff reflects the stiffness-dominated impedance of the middle ear.

In 1948 Lloyd Jeffress proposed a neural circuit for measuring interaural time differences: an array of coincidence-detector neurons, each receiving input from both ears through axons of different lengths. A sound arriving at the left ear first would travel further along the left delay line before reaching the coincidence detector that compensates for exactly that ITD. The detector fires maximally when the two signals arrive simultaneously — converting a time difference into a place of maximum activity. The Jeffress model became the canonical picture of binaural processing for half a century. In birds, the nucleus laminaris implements something very close to it, with anatomically measurable delay lines. In mammals, the medial superior olive (MSO) performs ITD computation, though the mechanism appears to rely more on inhibitory timing than on pure axonal delay. The model's enduring value is conceptual: it shows how a temporal code can be converted to a place code using nothing more than conduction delays and coincidence detection.

In 1948 Thomas Gold, a physicist better known for his work in cosmology, published a remarkable theoretical argument: the viscous damping in the cochlea is far too strong for the basilar membrane to achieve the sharp frequency tuning that psychophysical experiments demand. He proposed that the cochlea must contain an active feedback mechanism — a biological amplifier that injects energy on each cycle to counteract the damping. The idea was largely ignored by the auditory community for three decades. Vindication came in stages. David Kemp's 1978 discovery of otoacoustic emissions — faint sounds emitted by the ear itself — proved that the cochlea was indeed an active device. In 1985 William Brownell demonstrated the mechanism directly: isolated outer hair cells change length when their membrane potential changes, on a cycle-by-cycle basis at audio frequencies. This electromotility, later traced to the motor protein prestin, is the engine of the cochlear amplifier that Gold had predicted purely from physical reasoning.

Albert Bregman's 1990 monograph *Auditory Scene Analysis* synthesised decades of psychophysical research into a unified framework for how the auditory system parses a complex acoustic mixture into separate perceptual objects — voices, instruments, environmental sounds. Bregman identified two classes of grouping process: *primitive* (bottom-up, driven by physical regularities like harmonicity, common onset, and frequency proximity) and *schema-based* (top-down, driven by learned templates and attention). The framework gave the field a vocabulary and an experimental program. The streaming phenomena explored in this lesson — the bistable ABA_ triplet, the role of frequency separation, the build-up of streaming over time — are all experiments motivated by Bregman's taxonomy. Modern computational models of auditory scene analysis, including Bayesian and deep-learning approaches, remain organised around the primitive/schema-based distinction he articulated.