Hearing

Abstract

Hearing is the sensory process by which the ear turns airborne pressure waves into the neural signals the brain interprets as sound — its pitch, loudness, and location. Cognitive psychology studies it because what a listener hears is not a faithful copy of the acoustic input but a reconstruction: the cochlea performs a frequency analysis, and central mechanisms infer pitch, segregate overlapping sources, and place them in space. This article traces the path from air to auditory nerve, the place and timing codes for frequency, the measurement of loudness and the audible range, and the auditory scene analysis that lets a listener follow one voice in a crowd. It also covers otoacoustic emissions, the faint sounds a working ear itself produces. Three interactive demonstrations let the reader drive cochlear tonotopy, an auditory filter, and the streaming of a tone sequence.

Keywords: hearing, pitch perception, cochlea, auditory filter, auditory scene analysis

Hearing is the transduction and perceptual analysis of sound: the auditory system converts the pressure fluctuations of an acoustic wave into a neural code and, from that code, recovers the properties a listener experiences — how high or low a sound is, how loud, where it comes from, and which of several overlapping sources it belongs to (Oxenham, 2018). It is a textbook case of perception as inference rather than recording: the ear does not relay the waveform intact but decomposes it into frequency components and hands the brain a representation from which pitch, timbre, and auditory objects must be reconstructed (Moore, 2013). Cognitive psychology's stake in hearing is exactly this reconstructive character, which is why the same physical sound can be heard in different ways depending on context, attention, and what else is present.

What Hearing Is

Hearing is one of the senses, and MeSH files it as a form of sensation within the broader domain of the vestibulocochlear physiological phenomena. As a perceptual system it has a clear job: to take the one-dimensional pressure-versus-time signal arriving at each eardrum and recover the structure of the acoustic world that produced it — the identities, pitches, and positions of the sound sources present (Oxenham, 2018). This is a hard inverse problem, because many different combinations of sources can produce the same waveform at the ear, and the auditory system solves it with a mixture of peripheral frequency analysis and central grouping rules.

The first and most consequential fact about hearing is that the ear is a frequency analyser. Rather than encoding the raw waveform, the cochlea decomposes an incoming sound into its frequency components and represents them along a spatial map (Moore, 2013). Almost everything downstream — the perception of pitch, the timbre that distinguishes a violin from a flute, the ability to pull one talker out of a background of others — is built on this initial spectral decomposition and on the two complementary codes, one based on place and one on timing, by which the frequency content of a sound is carried into the brain (Oxenham, 2012).

Types of Hearing

MeSH places hearing beneath sensation and, in the tree that runs through sensation (F02.830.816.263), records the narrower descriptors shown in Table 1. As with any MeSH placement, this is an indexing classification for organising the literature, not a theory of how hearing divides into natural kinds; the two subtypes below pick out particular routes and by-products of hearing rather than partitioning the whole faculty, and neither yet has its own article on this site. Bone conduction concerns how the stimulus reaches the cochlea, while spontaneous otoacoustic emissions are a sign that the cochlea is an active, energy-producing organ rather than a passive microphone.

From Air to Auditory Nerve

The journey of a sound begins with the outer ear funnelling pressure waves onto the eardrum, whose vibration the three middle-ear bones transmit to the fluid-filled cochlea. Inside the cochlea the sound sets up a travelling wave along the basilar membrane. Georg von Békésy's direct observations of the membrane, for which he received the 1961 Nobel Prize, showed that this wave peaks at a position that depends systematically on frequency — high frequencies near the stiff base, low frequencies near the flexible apex — so that the cochlea lays out sound as a spatial map of frequency (Békésy, 1960). This tonotopic arrangement is the anatomical basis of the place code discussed below.

The cochlea is not a passive frequency meter. The outer hair cells actively amplify the travelling wave, sharpening its peak and enormously increasing sensitivity, so that the ear behaves as a set of very finely tuned, nonlinear filters (Hudspeth, 1989). The clearest sign of this active process is that the ear emits sound of its own: David Kemp's discovery of otoacoustic emissions showed that a healthy cochlea, stimulated or even at rest, produces measurable acoustic energy in the ear canal — a by-product of the amplifier at work (Kemp, 1978). Because emissions depend on cochlear mechanics, they can be used to estimate the sharpness of human cochlear tuning non-invasively, and such estimates indicate tuning considerably sharper than older physiological measurements had suggested (Shera et al., 2002). The inner hair cells, by contrast, are the true sensory transducers, converting membrane motion into the trains of nerve impulses that leave the ear along the auditory nerve.

Coding Pitch: Place and Timing

How does the brain know the frequency of a sound? Two codes operate together. The place code reads frequency from where along the tonotopic map the response is greatest — the principle Hermann von Helmholtz anticipated when he proposed that the cochlea resonates, different places tuned to different frequencies, and that Békésy's travelling wave later grounded in mechanics (Békésy, 1960). The timing (temporal) code reads frequency from when auditory-nerve fibres fire: the fibres phase-lock to the waveform, firing at a consistent phase of each cycle, so the pattern of spike intervals carries the period of the sound. Because a single fibre cannot fire on every cycle of a high-frequency tone, Ernest Glen Wever and Charles Bray proposed the volley principle, in which groups of fibres take turns so that their combined firing still tracks the waveform (Wever & Bray, 1930).

The decisive evidence that pitch is not simply place is the missing fundamental: present a listener with only the upper harmonics of a complex tone — say 800, 1000, and 1200 Hz — and they hear a pitch at the absent 200 Hz fundamental, the common spacing of the components. This residue pitch, robust even when a low-frequency masker rules out any distortion product at the fundamental's place, shows that the auditory system infers pitch from the periodicity of the whole pattern rather than reading it off a single cochlear channel (Schouten et al., 1962). It is the phenomenon that forces any account of pitch to include a timing- or pattern-based stage beyond tonotopy.

Neither code is sufficient alone, and pitch perception draws on both (Oxenham, 2012). Phase-locking is strong at low frequencies but degrades above a few kilohertz, which is roughly where the musical sense of pitch also fades; the place code, conversely, is coarser at low frequencies where the travelling-wave peak is broad. The interplay explains why the frequency limits of melody, the accuracy of pitch judgements, and the perception of musical intervals behave as they do, and why consonance between two tones is closely tied to whether their components fall within the same cochlear filter (Plomp & Levelt, 1965). Pitch is therefore not a single quantity read off one channel but an inference the auditory system draws from converging place and timing evidence — and, as newer work shows, one that different listeners can weight differently (McPherson & McDermott, 2018).

Loudness and the Audible Range

Loudness is the perceptual correlate of sound intensity, but the mapping is neither simple nor uniform across frequency. Harvey Fletcher and Wilden Munson measured the equal-loudness contours — the sound levels at which tones of different frequencies are judged equally loud — and found the ear most sensitive in the range of a few kilohertz and progressively less sensitive toward the low and high extremes of the audible range (Fletcher & Munson, 1933). The human audible range spans roughly 20 Hz to 20 kHz in young ears, but sensitivity within it is shaped by the mechanics of the outer and middle ear and by the cochlear amplifier.

Frequency resolution is captured by the auditory filter. Fletcher framed the ear as a bank of overlapping bandpass filters and introduced the critical band: only noise falling within the band centred on a tone contributes to masking it (Fletcher, 1940). Eberhard Zwicker mapped these critical bands across the spectrum (Zwicker, 1961), and Brian Moore and Brian Glasberg refined the measurement into the equivalent rectangular bandwidth (ERB), derived from notched-noise masking, which gives the width of the auditory filter as a function of centre frequency (Glasberg & Moore, 1990). The auditory filter is why one sound masks another when they are close in frequency, why the ear resolves the lower harmonics of a complex tone but not the higher ones, and why frequency selectivity is a central variable in models of hearing.

Locating Sounds and Auditory Scene Analysis

Hearing must also answer where a sound is and what it belongs to. Localisation in the horizontal plane relies on two binaural cues: the interaural time difference (a sound reaches the near ear first) and the interaural level difference (the head shadows the far ear), with time differences dominating at low frequencies and level differences at high (Middlebrooks & Green, 1991). Elevation and front–back judgements depend instead on the spectral colouration the outer ear imposes. These cues let the auditory system assign a direction to each source.

Assigning sound to sources at all is the problem of auditory scene analysis, Albert Bregman's framework for how the auditory system parses a single mixed waveform into distinct perceptual streams using cues such as common onset, harmonic relationship, and continuity of frequency (Bregman, 1990). Streaming is what lets a listener follow one instrument in an ensemble or one talker at a party, and it is closely bound up with attention: selecting a source is an act of object-based auditory attention operating on the streams the system has formed (Shinn-Cunningham, 2008). The resulting perceptual entities — the auditory objects a listener hears and can attend to, identify, and locate — are increasingly treated as the central unit of hearing beyond the periphery (Bizley & Cohen, 2013). Recent computational models trained to perform real-world hearing tasks reproduce aspects of this behaviour and predict cortical responses, offering a new handle on how the brain builds auditory objects (Kell et al., 2018).

Worked Example

The width of the auditory filter — how finely the ear resolves frequency at a given point — is captured by the equivalent rectangular bandwidth. Glasberg and Moore's formula gives the ERB in hertz as a function of centre frequency F in kilohertz (Glasberg & Moore, 1990):

ERB (Hz) = 24.7 × (4.37 × F + 1), F in kHz.

At a centre frequency of 1 kHz, ERB = 24.7 × (4.37 × 1 + 1) = 24.7 × 5.37 = 132.6 Hz — the filter passes a band about 133 Hz wide. At 4 kHz, ERB = 24.7 × (4.37 × 4 + 1) = 24.7 × 18.48 = 456.5 Hz, more than three times wider. At 0.5 kHz, ERB = 24.7 × (4.37 × 0.5 + 1) = 24.7 × 3.185 = 78.7 Hz. So the absolute bandwidth grows with frequency, but the relative bandwidth narrows: at 1 kHz the filter is about 13% of the centre frequency wide, whereas at 4 kHz it is about 11%.

This has a direct perceptual consequence for resolving the harmonics of a complex tone. A 200 Hz voice has harmonics at 200, 400, 600, … Hz. Near the third harmonic (600 Hz) the ERB is 24.7 × (4.37 × 0.6 + 1) = 24.7 × 3.622 = 89.5 Hz — narrower than the 200 Hz spacing, so that harmonic falls in its own filter and is resolved. Near the tenth harmonic (2 kHz) the ERB is 24.7 × (4.37 × 2 + 1) = 24.7 × 9.74 = 240.6 Hz — wider than 200 Hz, so several harmonics share one filter and are unresolved. The transition from resolved to unresolved harmonics, which shapes how pitch and timbre are computed, follows directly from how the ERB grows with frequency. The first demonstration lets the reader watch the tonotopic place of a tone move along the cochlea; the second plots the auditory filter and its ERB live.

Discussion

Hearing earns its place in cognitive psychology because the sound a listener experiences is manifestly a construction, not a recording. The cochlea's frequency analysis, the twin place and timing codes, the auditory filter, and the grouping rules of scene analysis are all stages at which the raw waveform is transformed into something the brain can use — and at each stage the transformation can be probed, modelled, and, in the laboratory, pushed around. Pitch that does not correspond to any single frequency present, a talker heard clearly in a noisy room, two tones that fuse or stay separate depending only on their timing: these are the signatures of an inferential system, and they are why the psychophysics of hearing has been so productive a testing ground for theories of perception.

The account also has clear limits and open seams. The place and timing codes are complementary but their exact division of labour, especially for pitch above a few kilohertz, remains debated; the neural computations that turn cochlear output into auditory objects are only partly understood; and individual differences — in cochlear tuning, in how listeners weight pitch cues, in the effects of ageing and hearing loss — are larger than a single normative model admits. Age-related hearing loss in particular has consequences that reach beyond the ear into cognition and communication, a reminder that hearing is embedded in the wider cognitive system rather than sealed off from it (Peelle & Wingfield, 2016).

Current Directions

Contemporary research is pressing on three fronts. The first is the auditory cortex: new recording and analysis methods are revising the classic picture of cortical organisation, revealing hierarchies of processing and representations of complex sound features beyond simple tonotopy (King et al., 2018). Task-optimised deep neural networks have become a central tool here, reproducing human performance on real-world hearing tasks and predicting cortical responses well enough to suggest a functional hierarchy from primary to non-primary areas (Kell et al., 2018). A second front concerns individual variation: careful psychophysics shows that listeners differ systematically in whether they hear pitch by spectral or temporal cues, unsettling the assumption of a single universal mechanism (McPherson & McDermott, 2018). A third refines the periphery itself, using otoacoustic emissions and modelling to pin down human cochlear tuning and its relation to emission generation (Shera & Charaziak, 2019). Cutting across all three is the clinical and cognitive weight of hearing loss, whose links to communication difficulty and cognitive load keep the psychology of hearing tied to questions of healthy ageing (Peelle & Wingfield, 2016).

Common Misconceptions

The ear works like a microphone, faithfully recording sound.

The cochlea decomposes sound into frequency components and actively amplifies them; what the brain receives is a transformed, analysed signal from which perception is reconstructed, not a copy of the waveform (Moore, 2013; Hudspeth, 1989).

Frequency is coded only by place along the cochlea.

A timing code based on phase-locking, organised by the volley principle, carries frequency information alongside the place code, and pitch perception depends on both (Wever & Bray, 1930; Oxenham, 2012).

The ear only receives sound; it never produces it.

A healthy cochlea emits faint sounds of its own — otoacoustic emissions — a by-product of active amplification that can be measured in the ear canal and used to test hearing (Kemp, 1978).

Glossary

Auditory scene analysis.

The process by which the auditory system parses a single mixed waveform into distinct perceptual streams corresponding to separate sound sources.

Basilar membrane.

The membrane running the length of the cochlea whose stiffness gradient makes it respond to high frequencies at the base and low frequencies at the apex.

Bone conduction.

Transmission of sound to the inner ear through the bones of the skull, bypassing the outer and middle ear.

Cochlea.

The fluid-filled, spiral organ of the inner ear that transduces sound into neural signals and performs a running frequency analysis.

Critical band.

The band of frequencies around a tone within which other sounds interact with it, for example in masking; the perceptual bandwidth of the auditory filter.

Equal-loudness contour.

A curve of the sound levels at which tones of different frequencies are judged equally loud, showing the ear's uneven sensitivity across the audible range.

Equivalent rectangular bandwidth (ERB).

A measure of auditory-filter width as a function of centre frequency, derived from notched-noise masking; the ideal rectangular filter passing the same power.

Interaural time difference.

The small difference in a sound's arrival time at the two ears, a primary cue for horizontal localisation, dominant at low frequencies.

Missing fundamental.

The pitch heard at a complex tone's fundamental frequency even when that frequency is physically absent, evidence that pitch is inferred from the periodicity of the harmonic pattern rather than read off a single cochlear place; also called residue pitch.

Otoacoustic emission.

Sound generated by the cochlea and measurable in the ear canal, spontaneously or in response to stimulation, reflecting active outer-hair-cell amplification.

Phase-locking.

The tendency of auditory-nerve fibres to fire at a consistent phase of a sound's waveform, providing a temporal code for frequency.

Place code.

The representation of a sound's frequency by the location of maximal response along the tonotopic map of the cochlea.

Tonotopy.

The orderly spatial mapping of frequency onto position, established in the cochlea and preserved through much of the auditory pathway.

Travelling wave.

The wave of displacement that moves along the basilar membrane in response to sound, peaking at a frequency-dependent place.

Volley principle.

The idea that groups of auditory-nerve fibres fire in alternation so their combined activity tracks a waveform faster than any single fibre could.

Key Researchers

Albert S. Bregman. Cognitive psychologist at McGill University; he founded the study of auditory scene analysis, showing how the auditory system organises a mixture of sounds into distinct perceptual streams. Wikipedia

Georg von Békésy. Biophysicist at Harvard University; his direct observations of the cochlear travelling wave established the mechanical basis of place coding and earned the 1961 Nobel Prize in Physiology or Medicine. Wikipedia - Wikidata

Harvey Fletcher. Physicist at Bell Telephone Laboratories; he measured the equal-loudness contours and introduced the critical band, founding the psychoacoustics of loudness and frequency selectivity. Wikipedia - Wikidata

Hermann von Helmholtz. Physicist and physiologist at the University of Berlin; his resonance theory of hearing framed the cochlea as a frequency analyser, anticipating the place code. Wikipedia - Wikidata

Josh H. McDermott. Auditory neuroscientist at the Massachusetts Institute of Technology; he studies real-world hearing with computational models and cross-cultural experiments spanning pitch, texture, and scene analysis. ORCID - Faculty Page

Brian C. J. Moore. Auditory psychologist at the University of Cambridge; he formalised the ERB scale of auditory-filter width and authored the standard textbook on the psychology of hearing. ORCID - Faculty Page - Wikipedia

Andrew J. Oxenham. Auditory psychologist at the University of Minnesota; he studies pitch perception and cochlear frequency selectivity, integrating psychoacoustics with the physiology of the auditory periphery. ORCID - Faculty Page

Christopher A. Shera. Auditory biophysicist at the University of Southern California; he uses otoacoustic emissions to estimate human cochlear tuning and develops coherent-reflection theory of emission generation. ORCID - Faculty Page

Barbara G. Shinn-Cunningham. Auditory neuroscientist at Carnegie Mellon University; she studies the neural basis of auditory attention and how object-based selection lets listeners solve the cocktail-party problem. ORCID - Faculty Page

Ernest Glen Wever. Experimental psychologist at Princeton University; with Charles Bray he discovered the cochlear microphonic and advanced the volley theory of temporal frequency coding. Wikipedia - Wikidata

Frequently Asked Questions

What is hearing?
Hearing is the sensory process by which the ear converts airborne pressure waves into neural signals and the brain reconstructs from them the pitch, loudness, location, and identity of sounds (Oxenham, 2018).

How does the cochlea tell frequencies apart?
The cochlea acts as a frequency analyser: a travelling wave peaks at a place that depends on frequency, high near the base and low near the apex, laying sound out as a spatial map (Moore, 2013).

What is the difference between the place code and the timing code?
The place code reads frequency from where the cochlea responds most; the timing code reads it from the phase-locked firing of nerve fibres, organised by the volley principle. Pitch perception uses both (Wever & Bray, 1930; Oxenham, 2012).

What is the auditory filter or critical band?
It is the band of frequencies the ear treats together at a given point; its width, the ERB, sets how finely frequency is resolved and governs masking (Fletcher, 1940; Glasberg & Moore, 1990).

Why is the ear more sensitive to some frequencies than others?
Equal-loudness contours show the ear is most sensitive around a few kilohertz and less so at the extremes of the 20 Hz to 20 kHz range, reflecting the mechanics of the ear and the cochlear amplifier (Fletcher & Munson, 1933).

What are otoacoustic emissions?
They are faint sounds the cochlea itself produces, a by-product of active amplification by the outer hair cells, measurable in the ear canal and used to test hearing and estimate cochlear tuning (Kemp, 1978; Shera et al., 2002).

How do we tell where a sound is coming from?
Horizontal localisation uses the difference in a sound's arrival time and level at the two ears; elevation uses spectral cues from the outer ear (Middlebrooks & Green, 1991).

How can we follow one voice in a noisy room?
Auditory scene analysis groups the mixture into streams by cues such as common onset and harmonicity, and object-based attention selects the stream of interest (Bregman, 1990; Shinn-Cunningham, 2008).

References