RESEARCH DOSSIER
Auditory Perception and Psychoacoustics
Abstract. Auditory perception and psychoacoustics investigate how controlled acoustic stimulation yields measurable detection, discrimination, loudness, pitch, timbre, timing, and spatial judgments. This fundamental research monograph reconstructs historical developments in experimental hearing science, examines the physical and physiological constraints of auditory processing, and critically synthesizes psychophysical experiments, cochlear and neural observations, quantitative models, and international acoustical standards. Separate chapters evaluate auditory thresholds, spectral resolution, masking and recovery, cochlear compression, loudness integration, harmonic pitch, temporal modulation, binaural localization, and the interpretation of speech-cue transformations. The methodology distinguishes stimulus calibration, behavioral inference, computational model fit, and causal explanation, with explicit attention to population generalizability, experimental artifacts, replicability, and unresolved controversies. The concluding synthesis sets out testable research priorities and a functional basis for comparisons between biological hearing and engineered auditory systems in the Artificial Era.
Keywords: auditory perception; psychoacoustics; hearing; auditory psychophysics; auditory thresholds; cochlear tuning; auditory filters; masking; loudness; pitch; timbre; temporal modulation; binaural hearing; sound localization; auditory neuroscience; experimental methods; artificial auditory systems
Auditory Perception as a Quantitative Scientific Problem
Auditory perception is the measurable transformation of acoustic stimulation into judgments and experiences of presence, magnitude, pitch, timbre, duration, location, and relations among sounds. Psychoacoustics investigates these transformations through controlled variation of acoustic signals and quantitative observation of responses. Its object is a relation between a physical stimulus, the state of a listener, a perceptual task, and the resulting response distribution. A tone presented at a calibrated sound pressure level can be audible in quiet and undetectable in competing noise; the perceptual response depends on the spectral and temporal structure of both signals. The subject therefore requires coordinated acoustic specification and psychophysical inference. Oxenham (2018) provides an integrative analysis of how hearing converts sound into perceptually useful information, while Moore & Glasberg (1983) illustrates how measured performance can constrain models of frequency selectivity.
The principal explanatory variables are frequency, acoustic level, spectrum, periodicity, envelope, onset, duration, and spatial transfer at each ear. Principal outcomes include detection probability, difference thresholds, loudness matches, pitch judgments, similarity spaces, localization error, and response time. A scientific explanation must specify the task and link the outcome to a mechanism with independently testable consequences. A change in reported loudness may reflect cochlear excitation, auditory integration, stimulus context, learned response mapping, or several of these processes. It is methodologically misleading to infer a unique neural cause from an isolated rating. Psychophysical experiments strengthen inference by manipulating physical factors separately, characterizing observers, measuring uncertainty, and comparing competing models across several tasks rather than treating subjective report as a direct physical measurement.
This dossier concentrates on elementary and intermediate hearing capacities under defined acoustic conditions: audibility and masking, spectral and temporal resolution, loudness, pitch, timbre, binaural and spatial perception, discrimination, and the validity of quantitative models. The neighboring research object of auditory scene analysis investigates the formation of separable streams and objects from simultaneous or sequential sources; speech comprehension belongs more fully to the psychology of language. These phenomena remain connected through the same auditory representations, and their boundary is identified by the causal question of each experiment. The analysis synthesizes original laboratory studies, physiological findings, computational models, and acoustic standards. Evidence from music and speech is used where it directly tests auditory coding or perception rather than expanding the dossier into a general theory of linguistic or musical understanding.
Historical Formation of Psychoacoustics and Experimental Hearing Research
Scientific accounts of hearing developed through the intersection of musical acoustics, physiological investigations of the ear, and experimental psychology. Nineteenth-century psychophysics established that perceived intensity could be investigated through controlled comparisons rather than informal descriptions alone. Twentieth-century laboratories acquired electronically generated tones, calibrated transducers, precision level controls, and quantitative procedures for measuring thresholds and masking. These instruments made it possible to vary spectral components, duration, and level independently. A historical reconstruction should recognize that a theory's influence depends on the questions and measurements it organized, whereas its present explanatory accuracy must be judged by contemporary experiments. The continuing value of psychophysics comes from its capacity to turn a seemingly private perceptual judgment into a reproducible relation between stimulation, conditions, and observer performance.
The experiments of Fletcher & Munson (1933) were formative for loudness measurement and the empirical relationship between sound frequency, level, and judged loudness. Their procedures predate modern international equal-loudness standards, so the historical data and later standardized contour estimates should be distinguished. The present ISO (2023) standard describes normal equal-loudness-level contours under specified listening conditions and a defined population, rather than a universal function independent of age, hearing status, or acoustic environment. This progression from pioneering observation to formalized stimulus conditions demonstrates how standardization refines earlier psychophysical results. Equal-loudness contours are products of empirical measurement and modeling; a standard provides practical comparability within stated limits, not a substitute for studying people and contexts outside those limits.
Later advances connected perceptual tasks to models of cochlear filters, nonlinear transduction, temporal processing, and binaural cue extraction. The parameterizations of auditory filters in Moore & Glasberg (1983) and experimental comparisons of maskers in Glasberg et al. (1984) exemplify the movement from descriptive masking curves to explicit models of excitation. Kemp (1978) introduced evidence for acoustic emissions generated within the auditory system, opening a physiological route to investigating cochlear mechanics. Studies of temporal modulation Viemeister (1979) and harmonic pitch Shackleton & Carlyon (1994) further established auditory perception as a field in which distinct experimental traditions can constrain the same account of neural coding. The history is one of progressively sharper operational distinctions and increasingly direct comparisons among psychological, mechanical, and neural observations.
Acoustic Variables, Measurement Scales, and Stimulus Construction
An acoustic waveform specifies pressure variations over time at a chosen position. A pure sinusoidal tone has a defined frequency and amplitude, while natural sounds generally contain changing mixtures of spectral components and temporal envelopes. Frequency, measured in hertz, describes physical periodic variation; sound pressure level describes a logarithmic ratio to a reference pressure; and duration, rise time, modulation depth, and bandwidth quantify further stimulus characteristics. A decibel value is meaningful only when its reference and measurement context are identified. A difference of decibels in sound pressure level is a physical relation, whereas perceived loudness follows frequency- and context-dependent psychophysical functions. Equal-loudness comparisons Fletcher & Munson (1933) and current standardized contours ISO (2023) show why sound pressure and loudness require separate descriptions.
Stimulus generation introduces possible spectral artifacts, transducer distortion, clipping, onset clicks, channel imbalance, and acoustic leakage. Abrupt gating spreads energy across frequencies and can create audible transient cues that are absent in an intended steady-state comparison. Headphones can deliver different spectra at different placements on the ear, while loudspeakers interact with room reflections and listener location. Calibration must therefore describe the signal chain, output transducer, coupler or microphone, intended level reference, and verification procedure. In binaural experiments, synchronization and channel matching are necessary because unwanted timing or level differences may masquerade as localization cues. These are scientific controls: invalid acoustic input prevents the resulting psychometric function from identifying the psychological mechanism of interest.
Signal representations must also respect the relationship between a physical decomposition and a perceptual computation. A Fourier spectrum can reveal components that a listener does not resolve individually, while an amplitude envelope can dominate perception even when spectral fine structure changes. Speech experiments using transformed acoustic information Shannon et al. (1995) and auditory chimera stimuli Smith et al. (2002) illustrate that distinct aspects of the waveform contribute differently to perceptual tasks. Interpreting these manipulations requires noting which acoustic properties were preserved, exchanged, or degraded. Reproducible research should archive the generating code, calibration information, presentation level, duration, and a sample or exact specification of the signals. The observer's task, trial timing, and response categories are part of the stimulus–response model.
Cochlear Mechanics, Frequency Selectivity, and Active Processes
The cochlea decomposes incoming mechanical vibration across an organized spatial gradient, and hair-cell transduction couples motion to neural signals. Frequency selectivity in human hearing is commonly inferred indirectly through behavioral masking, auditory-nerve models, and objective physiological measurements. Auditory filters provide useful psychophysical descriptions of frequency-dependent sensitivity, yet their parameters depend on stimulus level, masker type, temporal arrangement, and inferential method. Moore & Glasberg (1983) formulated widely used filter-bandwidth and excitation-pattern approximations; Glasberg et al. (1984) compared estimates derived with different masker configurations. Such models are effective when they predict independent data, and their bandwidth estimates should not be treated as invariant dimensions of a physical filter detached from experimental conditions.
Human cochlear tuning changes with level because mechanical amplification and compression produce nonlinear relationships between sound input and basilar-membrane response. Oxenham & Shera (2003) compared forward and simultaneous masking approaches to estimate tuning at low levels, providing a case study in how an apparent physiological parameter depends on the behavioral procedure used to infer it. Shera et al. (2002) compared otoacoustic and behavioral estimates of human cochlear tuning, illustrating the evidentiary value of convergence across different measurement modalities. Both approaches require model assumptions: otoacoustic emissions provide information about cochlear processes, while psychophysical masking provides information about a behavioral decision influenced by peripheral and more central processing.
The discovery of stimulus-related emissions by Kemp (1978) changed the range of observations available to hearing science. An emission recorded in the ear canal supports claims about active cochlear mechanics under the measurement conditions, while its relationship to a particular individual's performance on a pitch, speech, or loudness task requires further evidence. The distinction matters especially for cases of hearing impairment or altered cochlear function, where physiological and perceptual measures can diverge. A comprehensive account connects mechanical selectivity, auditory-nerve activity, and behavior through hypotheses that specify each level. The mechanical model describes the transduction of energy; the perceptual model predicts how those signals support decisions and experiences. Their agreement is established through cross-level experiments rather than shared terminology.
Neural Coding and the Limits of a Single-Representation Theory
The auditory nerve provides both an across-channel pattern of activity and temporal information related to the acoustic waveform. Different stimuli can yield similar total excitation but differing patterns of phase locking, modulation, and onset synchrony; conversely, different sound mixtures can support similar behavioral judgments. Oxenham (2018) integrates evidence that pitch, timbre, and other percepts depend on how spectral and temporal information are processed. A neural response recorded during stimulation is a candidate representation of a stimulus feature, not automatically the mechanism that generates the listener's reported judgment. Strong explanatory evidence comes from changes in both signal coding and task performance under interventions predicted by a model.
Temporal modulation coding has been documented at multiple auditory stations, and neural sensitivity to envelope rates is not uniform across frequency, level, or processing stage. The synthesis by Joris et al. (2004) links physiological responses to amplitude-modulated signals with unresolved questions about how those responses support perception. Behavioral modulation thresholds Viemeister (1979) offer independent constraints on functional sensitivity, but a match between neural modulation tuning and psychophysical sensitivity does not uniquely establish a perceptual decoder. Comparisons require matched stimuli, listener or species assumptions, measurement of neural variability, and explicit mappings from population responses to behavior. Without these conditions, a correspondence may remain descriptive rather than causal.
A scientific model should distinguish the code available in a sensory population from the code actually used to solve a task. In harmonic-pitch experiments, which partials are resolved by peripheral filters changes the strategies available for estimating a common periodicity Shackleton & Carlyon (1994). Chimera experiments Smith et al. (2002) provide a different way to test reliance on envelope and fine-structure components. These designs reveal task dependence: information sufficient for intelligibility may be insufficient for another aspect of pitch or source quality. A single representation defined broadly as sound identity is therefore inadequate for precise prediction. Competing models should state which features are encoded, how estimates are combined, and the characteristic errors expected under constrained or ambiguous stimulation.
Psychophysical Tasks, Signal Detection, and Explanatory Identifiability
Psychophysical evidence is generated through an explicit relation between controlled stimulation and observable responses. Detection experiments ask whether a signal was present; discrimination experiments ask which interval differed; matching tasks adjust one stimulus until a specified dimension appears equal; magnitude estimation asks for graded responses. The outcome of each task includes perceptual sensitivity and the consequences of instructions, memory demands, response criteria, and trial design. A listener may detect a quiet tone reliably in a two-interval forced-choice task while giving fewer affirmative responses in a yes–no task because of different decision constraints. Experimental descriptions must identify the task and psychometric convention before reporting a threshold as an auditory property.
Adaptive tracks concentrate trials near a response region of interest, while fixed-level designs sample broader psychometric functions. Both require handling lapses, trial dependence, nonstationary attention, and finite-sample uncertainty. The experimenter also needs to guard against acoustic cues that make a nominal intensity discrimination into a spectral-discrimination task. The same concern applies to comparing simultaneous and forward maskers: differences can reflect the time course of excitation and recovery as well as task-specific response mechanisms Glasberg et al. (1984) Oxenham & Shera (2003). Repetition across listening sessions can reveal learning, fatigue, or calibration drift, each of which changes the interpretation of a group mean.
Explanatory identifiability is the problem of whether a pattern of behavioral results can distinguish among competing mechanisms. A broad auditory filter, a changed decision rule, and unreliable attention can each elevate a masked threshold, yet they predict different outcomes under additional manipulations. Multiple maskers, temporal conditions, response formats, and physiological comparisons can therefore be more informative than simply expanding the sample for one task. The most defensible evidence synthesis keeps physical calibration, behavioral estimation, and causal mechanism separately described. It specifies which findings have been reproduced across methods and which remain conditional on signal generation, listeners, and the model's assumptions.
Audibility, Pure-Tone Thresholds, and the Meaning of a Hearing Limit
An audibility threshold is an operational estimate of the stimulus level required to achieve a specified detection performance under defined conditions. Pure-tone thresholds vary with frequency, duration, noise floor, presentation method, and individual listener. Absolute thresholds describe performance in a specified background rather than a property of a tone in isolation. The common practice of plotting an audiogram is useful for characterizing sensitivity across frequencies, yet its values depend on the calibration standard, transducer, and testing protocol. The theoretical significance of these observations is that detection is a probabilistic judgment supported by sensory evidence. An apparent threshold shift can arise from changes in acoustic access, sensory encoding, internal variability, or decisions about uncertainty. This is why experimental hearing research reports both method and criterion.
Masking experiments show how a detectable tone can become difficult to hear when energy from another sound interacts with the representation used for detection. Threshold elevation depends on masker spectrum, signal frequency, duration, relative timing, and presentation level. Moore & Glasberg (1983) provided formulas that relate excitation patterns and auditory-filter bandwidths to phenomena such as spectral masking. The formulas summarize a rich body of auditory-psychophysical evidence, although the accuracy of any fixed function must be evaluated against the stimulus region and population of interest. A threshold measured under narrowband masking characterizes selective sensitivity differently from one measured under broadband noise, and changes in one condition do not automatically predict every other listening condition.
Threshold estimates must be accompanied by uncertainty and an account of listener variation. Hearing may change with exposure history, developmental stage, transient illness, equipment placement, fatigue, or instruction comprehension. Repeated measures can improve precision but may also introduce perceptual learning and strategic response changes. The researcher should report the masking noise spectrum, its level, signal-to-noise conditions, psychophysical rule, and treatment of incorrect and missing responses. The methodological principle is straightforward: an observed response rate is meaningful only relative to a reproducible experimental condition. A numerical threshold without the conditions under which it was derived supports little inference about hearing mechanisms.
Auditory Filters, Equivalent Rectangular Bandwidth, and Spectral Resolution
Spectral resolution is the capacity to distinguish components or changes in the frequency composition of sound. Auditory-filter models represent frequency-dependent sensitivity around a center frequency and are frequently characterized by equivalent rectangular bandwidth, a summary of filter area relative to peak transmission. Moore & Glasberg (1983) proposed influential formulae for filter bandwidth and excitation patterns, and Moore et al. (1990) examined auditory-filter shapes at lower center frequencies. These models allow quantitative predictions about masking and the resolution of harmonics. Their empirical success is conditional: estimated bandwidth is affected by the level and spectral design of the masker, the probe configuration, and the assumptions used to recover a filter shape from detection data.
The notched-noise technique compares tone-detection thresholds with noise energy omitted from a frequency region surrounding the signal. The pattern of threshold changes constrains the selectivity of the effective filter, but the technique requires assumptions concerning off-frequency listening, nonlinear compression, and level dependence. Glasberg et al. (1984) explicitly compared filter estimates using different maskers, illustrating why methodological equivalence must be established rather than presumed. Forward-masking measures Oxenham & Shera (2003) add temporal separation between masker and probe, which can reduce some complications of simultaneous interactions while introducing assumptions about recovery and the representation persisting after masker offset. No single protocol provides an assumption-free image of cochlear filtering. A contemporary experiment López-Ramos et al. (2024) compared 30-ms and 400-ms notched-noise forward maskers in seven normal-hearing participants. It found a duration-dependent difference in inferred filter bandwidth at 500 Hz, while the corresponding difference at 4 kHz was not statistically significant under the study conditions. The sample and methodological caveats counsel particular care in interpreting low-frequency estimates.
A model of spectral resolution must predict both elementary tone masking and more complex perceptual consequences. Closely spaced harmonics may be insufficiently resolved into distinct channels, changing the cues that support pitch; wider spectral spacing can make individual components available Shackleton & Carlyon (1994). The relationship between resolvability and perceptual organization is therefore mechanistically important. However, a broad filter inferred from one condition is not itself a complete theory of musical pitch, timbre, or speech recognition. Those outcomes require models for temporal structure, integration across channels, and decisions based on previous experience. Strong cross-validation uses different acoustic paradigms to estimate the same parameter, then tests whether that parameter predicts a new task without being refitted.
Simultaneous Masking, Spectral Spread, and Nonlinear Interaction
Simultaneous masking occurs when signal and masker overlap in time and the presence of the masker raises the level needed to detect the signal. The degree of masking depends on their spectral relationship and relative levels. A high-level masker often masks signals across a broader frequency region than a lower-level one, reflecting both the pattern of mechanical excitation and subsequent processing. Psychoacoustic experiments that vary masker center frequency, bandwidth, and intensity can constrain a filter model more tightly than experiments measuring only the detection of one fixed probe. Glasberg et al. (1984) compared filter shapes under alternative masking conditions, demonstrating the need to take the particular masker seriously in estimates of auditory selectivity.
Masking can also involve cues or nonlinear products generated by the auditory system or the playback apparatus. A listener can exploit a change in masker envelope, distortion product, or other incidental cue instead of detecting the intended target energy. Notched-noise and tone-on-tone paradigms therefore need careful control of acoustical and physiological side effects. Glasberg & Moore (1982) examined how auditory-filter shapes inferred from forward masking vary with stimulus level; Oxenham & Shera (2003) used complementary masking techniques to study low-level tuning. Their findings motivate models in which the effective bandwidth and excitation shape depend on level and procedure. Broad claims about critical bands should state whether they refer to a particular estimation method or an idealized computational approximation.
The perception of a sound mixture can be decomposed into detectable signal energy, discriminable features, and higher-order source organization. The masking threshold of a pure tone does not directly measure understanding of speech amid competing voices, even though frequency selectivity constrains the information available for such understanding. A useful scientific synthesis treats masking as a set of quantified interactions among signals and internal representations. This approach supports applications in audio coding and hearing-device evaluation, provided the algorithm is validated under the temporal and spectral conditions for which the model was derived. Predictions derived from stationary noise may require additional testing with fluctuating, modulated, or reverberant signals.
Non-Simultaneous Masking and the Time Course of Recovery
Forward masking measures the effect of a preceding masker on the detectability of a later signal. By varying the silent gap and relative levels, experiments characterize recovery from prior stimulation and the effective persistence of excitation or central processing consequences. Because masker and signal do not overlap, some simultaneous nonlinear interactions can be reduced, making the method valuable for estimating auditory tuning. Glasberg & Moore (1982) and Oxenham & Shera (2003) used forward-masking approaches to investigate level-dependent filtering and the sharpness of human cochlear tuning. The measured function nevertheless remains a behavioral threshold: it reflects whatever sensory, temporal, and decisional processes are sufficient to affect the listener's choice.
Backward masking refers to impairment in detecting or identifying a signal by a subsequent masker. Its interpretation has historically motivated questions about auditory persistence, temporal integration, and decision processes. A later stimulus can affect access to an earlier acoustic event even after the early event has ended, which means physical separation in time does not guarantee perceptual independence. Conditions must distinguish detection from identification, monaural from binaural presentation, and very short signal intervals from longer segments. Unlike a direct physiological recovery trace, a backward-masking psychometric result is conditioned by the selected task and available response cues. An informative design manipulates onset asynchrony and masker structure while evaluating whether a model predicts the direction and size of interference.
Temporal masking provides a bridge between peripheral nonlinear response and central integration. Comparing simultaneous and forward conditions can reveal whether a proposed filter estimate is sensitive to overlapping acoustics or to sustained changes in sensory gain. An explanation should also predict the effects of signal duration, stimulus level, and silent intervals, and be tested on signals beyond those used to fit it. Evidence from Glasberg et al. (1984) and Shera et al. (2002) supports the broader methodological practice of relating behavior to independent physiological observations. The remaining issue is model identification: multiple temporal mechanisms can yield similar detection curves, making systematic perturbation and convergent measurement essential.
Cochlear Compression, Adaptation, and Dynamic Hearing
Input level changes the gain and selectivity of the auditory periphery. Cochlear amplification supports sensitivity at low stimulus levels, while compressive behavior reduces the growth of response across some input ranges. Changes in the auditory excitation pattern alter both masking and perceived magnitude, although these relationships are not identical. Studies combining psychophysical estimates with otoacoustic observations Shera et al. (2002) help evaluate which mechanisms are compatible with the behavioral patterns. A model that assumes a linear and level-invariant auditory filter can approximate some experiments yet fail when compared with stimuli presented at another level. Level dependence is therefore a basic dimension of hearing, not a correction added after a perceptual theory is built.
Adaptation refers to history-dependent changes in neural or behavioral responsiveness during and after acoustic stimulation. Its time course depends on the signal and measurement scale: milliseconds of neural adaptation are not interchangeable with seconds of perceptual habituation or long-term changes associated with sound exposure. Modulation experiments Viemeister (1979) quantify sensitivity to changing envelopes, and physiological investigations Joris et al. (2004) identify neural processes that respond to temporal structure. The interaction of compression and adaptation is especially relevant to natural sounds whose levels fluctuate, such as speech or environmental noise. Models should specify whether a predicted perceptual effect derives from changed input coding, changed weighting across time, or decisions about a sound's persistence.
Dynamic hearing is assessed by alternating between steady and time-varying acoustic conditions. An experiment might match total stimulus energy while changing onset duration, envelope fluctuation, or temporal context; changes in judged loudness then constrain models of temporal integration. Buus et al. (1997) and Florentine et al. (1996) investigated relationships among loudness, duration, and level, illustrating that temporal history affects magnitude judgments. The observed patterns require attention to stimulus design and psychophysical scale. A model trained on stationary pure tones should not be presumed reliable for rapidly modulated real-world signals without independent tests of its state variables, response dynamics, and cross-listener behavior.
Calibration, Psychometric Uncertainty, and Reproducible Hearing Experiments
Acoustic calibration establishes that the actual stimulus delivered to a participant matches the quantities used in scientific analysis. The necessary record includes signal generation, reference level, hardware chain, sampling rate, transducer, coupling, calibration device, and the conditions at the ear. In loudspeaker studies, room acoustics and listener position must be documented; in headphone studies, fit, ear-canal interaction, and left–right matching can be decisive. An experiment with binaural timing contrasts requires precise channel synchronization; a near-threshold experiment requires a noise floor low enough to preserve the manipulation. Reporting a digital amplitude without the corresponding acoustic output is insufficient for interpreting audibility or loudness.
Statistical uncertainty involves both the finite number of trials and heterogeneity among listeners. Thresholds estimated from adaptive staircases or fitted psychometric functions should identify the targeted response probability, constraints on lapse rates, and interval estimation. Confidence intervals and individual-level results make clear whether a reported group trend is stable or driven by a small subset of listeners. More trials can improve precision within a session, whereas broader recruitment and independent replication improve evidence about generalization. Psychoacoustic research additionally faces procedural differences across laboratories: the same nominal signal can be experienced differently if presentation level or transducer transfer functions differ. Reproducible code and acoustic test files reduce these risks.
Standardized documents such as ISO (2023) and ISO (2017) make comparisons possible for defined procedures, but they should be invoked with their actual scope. An equal-loudness contour is not an individualized medical assessment, and a standardized loudness calculation is not a universally verified measure of every auditory experience. Methods sections should specify how standards were implemented and whether participants met their applicability assumptions. Good scientific reporting also preserves negative or conflicting results: departures from a model can reveal limitations in stimulus range, populations, or mechanisms. The defensible goal of calibration and replication is a transparent mapping from physical stimulus through procedure to result, allowing independent investigators to test the same explanation.
Loudness, Equal-Loudness Contours, and Judged Magnitude
Loudness is the perceived magnitude of sound, influenced by sound pressure level, frequency, bandwidth, temporal history, and the listener. Its operational study involves matching, comparative judgment, or explicit magnitude estimation. A listener can match the loudness of two tones with different frequencies even when their physical pressures differ considerably. The experiments of Fletcher & Munson (1933) helped establish this relationship as a quantitative research object. Their measurements support the general principle that a physical level scale cannot be interpreted as a direct scale of experienced magnitude. Contemporary accounts refine that principle through calibrated stimuli, improved statistical methods, and precise statements about the conditions under which a loudness judgment is obtained.
Equal-loudness contours connect levels at different pure-tone frequencies that yield the same judged loudness under specified presentation conditions. ISO (2023) supplies standardized normal equal-loudness-level contours within its stated listening and population conditions. The contours should not be read as universal functions for every person, headphone, room, or hearing configuration. Individual thresholds, ear-canal transfer functions, exposure history, and age can modify the relation between a physical stimulus and perceived magnitude. A rigorous experimental report therefore separates the standard reference used for acoustic calibration from empirical matches obtained from a particular sample. That separation is important when transferring laboratory findings to headphones, audiovisual environments, and hearing-device algorithms.
Loudness can also change when the sound duration or spectral composition changes while nominal peak level remains fixed. Multiple components can contribute to the total excitation underlying a magnitude judgment, and the effective integration across them depends on level and timing. Buus et al. (1997) investigated loudness, duration, and discrimination jointly, establishing how several psychophysical measures can constrain the same model. The scientifically relevant question is whether a proposed magnitude function predicts new stimuli and listeners, rather than only reproducing the data used to estimate its parameters. The broader implication is that loudness is a dynamic, context-sensitive perceptual construct defined through reproducible comparisons and explicit uncertainty.
Computational Loudness Models, Spectral Summation, and Time Dependence
Loudness models translate acoustic signals into predicted perceptual magnitude using stages that approximate peripheral filtering, excitation, and integration. Glasberg & Moore (2002) presented a model designed for time-varying sounds, extending evaluation beyond a single steady pure tone. The international ISO (2017) document specifies a standardized procedure for calculating loudness using the Zwicker method. A calculation standard and a psychological explanation play complementary roles: the former supports repeatable outputs for the specified signals and implementation, while the latter must account for why particular computational stages correspond to human performance. Validation requires comparing model outputs with independent psychophysical data across level, bandwidth, duration, and temporal change.
Spectral summation describes how energy distributed over frequencies can alter perceived loudness relative to concentrated energy. Temporal integration concerns the growth or decline of judged loudness as the duration or recent history of a sound changes. Florentine et al. (1996) and Buus et al. (1997) provide experimental tests of how these dimensions interact. The results are relevant to the choice between simple energy-summation rules and dynamic mechanisms that depend on compression and time constants. Researchers must distinguish physical integration of pressure or intensity from the psychological operations needed to make a loudness judgment. Equivalent total acoustic energy can coexist with different envelopes, peak levels, and perceived magnitudes.
More recent model development has incorporated binaural interactions and time-dependent processing. Moore et al. (2016) introduced binaural inhibition into a model of loudness for time-varying sounds, providing testable predictions about the combination of inputs from the ears. Such a model should be evaluated under monaural and binaural conditions using held-out stimuli, not treated as confirmed solely because its architecture resembles a plausible physiology. Its parameters may vary in their identifiability, and a close fit may conceal compensating assumptions. Comparative tests should ask which components explain a stable pattern of judgments and whether additional complexity improves prediction outside the development dataset.
Pitch of Pure Tones, Harmonic Relations, and the Missing Fundamental
Pitch is the perceptual dimension by which many sounds are ordered as relatively high or low and by which periodic signals acquire a perceived fundamental. For a pure tone, frequency is a principal determinant of pitch, although level and other conditions can influence judgments. Complex tones pose a richer problem because the perceived fundamental can correspond to a frequency with little or no acoustic energy at that frequency. The perceptual system can infer periodicity from the relationships among harmonics. McDermott & Oxenham (2008) reviews the interaction of pitch perception, music, and auditory coding, while Houtsma & Smurzynski (1990) provides controlled evidence about identification and discrimination of complex tones containing many harmonics.
A missing-fundamental stimulus is constructed from higher harmonics of an underlying periodic series while omitting the fundamental itself. Listeners can nonetheless report a pitch corresponding to the common spacing of the components. This is a constraint on theories equating pitch with the strongest spectral component or with a single peripheral channel. Competing accounts emphasize resolved harmonic patterns, temporal periodicity, or their interaction. Shackleton & Carlyon (1994) manipulated harmonic resolvability in pitch and frequency-modulation tasks, helping reveal which cues are available in particular stimulus conditions. The relevant empirical question is how the cue weights and error patterns change as frequencies, component numbers, and masker conditions are varied.
Pitch measurements require defining the judgment. Detecting a pitch change, assigning a musical interval, matching the pitch of two stimuli, and estimating the fundamental are different tasks with differing memory and decision requirements. Training and musical experience can influence performance without changing the physical construction of the tone. A complete model should explain perceived pitch, discrimination thresholds, octave ambiguities, and failures under degraded conditions. It should also specify the range of frequencies and harmonics for which the hypothesized code is effective. The most persuasive support comes from predicting multiple psychophysical outcomes with the same model parameters and from relating those predictions to independently measured auditory selectivity.
Resolved Harmonics, Temporal Periodicity, and Competing Pitch Models
Resolved harmonics activate partly distinguishable frequency-selective channels; unresolved harmonics overlap more strongly within peripheral filters and can generate envelope fluctuations related to their frequency spacing. The experiments in Shackleton & Carlyon (1994) directly compared perceptual consequences of these conditions. A place-based account can derive pitch information from the pattern of resolved excitation, while temporal accounts emphasize periodicity in neural activity. The strongest theories make explicit predictions for both classes of stimuli and for their combinations. A behaviorally successful pitch judgment does not imply that a listener used a single invariant cue, because the reliability of spectral and temporal information can change with frequency, level, and masking.
Neural phase locking supplies a physiological basis for representing timing in parts of the auditory system, yet its limits depend on species, neural population, and stimulus conditions. The overview by Oxenham (2018) emphasizes the need to connect information available in neural codes with actual pitch performance. Temporal-modulation physiology Joris et al. (2004) offers relevant mechanistic observations but does not settle the perceptual decoder on its own. Comparisons between neural measurements and behavioral pitch estimates must control acoustic cues, sampling, and model flexibility. Apparent success on a restricted family of harmonic complexes may arise from sensitivity to envelope regularity rather than a general solution to pitch perception.
Evidence across experiments supports flexible cue use and persistent theoretical questions about the conditions under which spectral and temporal representations are indispensable. Houtsma & Smurzynski (1990) tested pitch judgments with complex tones of different harmonic construction, while McDermott & Oxenham (2008) situates such results within broader music-related perception. Future model comparison can use mixtures with matched periodicity but altered spectral distribution, individually calibrated cochlear tuning, and cross-task predictions for discrimination and identification. The scientific standard is explanatory discrimination: a proposed code should be preferred when it predicts a pattern its competitors cannot reproduce under the same parameter constraints.
Timbre, Sound Quality, and Perceptual Dimensions
Timbre describes aspects of perceived sound quality through which sources can be distinguished even when pitch, loudness, and nominal duration are approximately matched. The explanatory variables include spectral envelope, harmonicity, onset dynamics, temporal modulation, and the evolution of frequency content. Timbre is therefore a multidimensional psychophysical construct: the judgment that two sounds differ in timbre can arise from different acoustic cues and different weighting strategies. Controlled synthesis can vary one acoustic dimension at a time, while similarity ratings and multidimensional scaling can reveal perceptual relations among a set of sounds. The resulting dimensions are model-dependent descriptions of observer judgments rather than automatically universal axes of auditory experience.
Harmonic organization contributes to both pitch and timbre, creating a useful experimental opportunity to distinguish related constructs. McDermott & Oxenham (2008) describes the importance of pitch and spectral relationships for musical sound, and Smith et al. (2002) demonstrates that exchanging temporal envelope and fine-structure properties can produce different perceptual outcomes. Such results show why a measure developed for speech intelligibility, musical pitch, or source recognition should not be treated as a general index of timbral quality. A timbre model must identify the task it predicts and the conditions under which changes in level, duration, or instrument identity can alter the reported quality.
Individual learning and listening context affect which timbral differences are informative. An audio engineer may detect a subtle spectral shift that is irrelevant to the performance of an untrained listener on an identification task. The difference can reflect experience, choice of response categories, and the sounds included in the test. Reliable evidence requires testing whether a recovered perceptual space predicts fresh comparisons, whether listeners agree about relative distances, and whether the acoustic features remain diagnostic outside the original stimulus set. Psychoacoustic sound-quality measurements have practical value for audio rendering and assistive devices, but claims about naturalness or pleasantness require their own validated constructs.
Frequency Discrimination, Intensity Discrimination, and Threshold Geometry
Discrimination measures the ability to judge differences between sounds, distinct from the ability to detect a sound in isolation. A frequency difference limen depends on baseline frequency, intensity, stimulus duration, temporal regularity, and the task. An intensity difference limen is influenced by level, spectral content, and whether the comparison is sequential or simultaneous. Threshold estimates can be expressed as absolute differences or relative proportions, but their interpretation must account for stimulus range and psychometric function. Houtsma & Smurzynski (1990) provides evidence that pitch identification and discrimination in harmonic complexes depend on their construction, while Shackleton & Carlyon (1994) links harmonic resolution to frequency-modulation discrimination.
Intensity discrimination illustrates the importance of separating sensitivity to a change in physical level from the magnitude of the percept produced. Loudness discrimination can vary with signal duration and baseline level, as investigated by Buus et al. (1997). A participant might correctly select the more intense interval while giving different verbal descriptions of how much louder it sounds. The two outcomes are constrained by partially shared sensory information but require different response mappings. The problem becomes harder with broadband or modulated sounds, where a nominal level change can also modify audibility of components or alter envelope cues. Experiments should control these alternative routes to successful discrimination.
A rigorous comparison reports the fitted psychometric function, threshold convention, confidence interval, and variability among observers. If listeners learn the task across sessions, a lower threshold may reflect improved use of available information without an identifiable peripheral change. Discrimination thresholds are consequently tools for characterizing performance and testing models of encoding and decision, rather than direct readings of an internal resolution parameter. A mechanistic account is strengthened by predicting how the function changes when frequency, level, uncertainty, and response instructions are manipulated independently. Replication with distinct stimuli is crucial for establishing that the inferred mechanism generalizes.
Temporal Resolution, Gaps, and Auditory Event Timing
Temporal resolution describes the capacity to detect and distinguish changes in acoustic signals over time. A gap-detection experiment inserts a silent interval into a sound and measures whether a listener detects the interruption. The resulting threshold depends on stimulus bandwidth, level, duration, the relationship between the pre-gap and post-gap segments, and the statistical properties of background noise. Detecting a gap in broadband noise does not provide an unqualified measure of the resolution available for all speech or music. The interpretation requires a model of auditory filtering, envelope sensitivity, persistence, and decision. Temporal processing therefore constitutes a family of experimentally distinguishable functions, each defined by the stimulus manipulation and behavioral readout.
Temporal integration may improve detection of an extended signal by combining information across time, while rapid changes in the signal can carry cues to events or linguistic contrasts. Viemeister (1979) used temporal modulation transfer functions to characterize sensitivity to periodic changes in sound amplitude. Rosen (1992) examined temporal information in speech at acoustic, auditory, and linguistic levels, showing why a laboratory measure of envelope processing should be connected to a stated task. Timing information occurs at several scales: fine temporal relationships support periodicity, syllabic-rate fluctuations contribute to speech structure, and longer intervals organize phrases and rhythmic patterns. Model parameters appropriate to one scale may not generalize to another.
A test of temporal resolution is most informative when alternative physical cues are constrained. Introducing a silent gap can also change total energy or produce onset and offset transients, allowing participants to identify a changed sound by a cue other than elapsed silence. Randomizing duration, equalizing energy when justified, and comparing spectral conditions help distinguish mechanisms. Repeated measurement across sessions should evaluate learning and fatigue as well as performance at a fixed test level. The key explanatory requirement is to connect the observed limit of detection to the acoustic dimension actually varied, preserving uncertainty about which stage of hearing governs the result.
Amplitude Modulation, Envelope Coding, and Modulation Transfer
Amplitude modulation produces systematic changes in signal envelope and can be manipulated through modulation rate, depth, carrier frequency, and waveform. A modulation threshold is the minimum depth required for an observer to discriminate a modulated from an unmodulated signal under specified conditions. Viemeister (1979) established a major experimental paradigm for temporal modulation transfer functions based on such thresholds. The resulting functions describe behavioral sensitivity and can be compared across levels, carriers, and listener groups. However, their shape emerges from acoustic filtering, neural encoding, and task demands; the transfer function should not be interpreted as a measurement of a single neural filter without independent evidence.
Physiological work reviewed by Joris et al. (2004) reveals diverse neural responses to amplitude modulation along auditory pathways. These findings provide mechanistic candidates for the perceptual sensitivity documented in laboratory psychophysics, while leaving open how populations are decoded during different judgments. Bacon & Viemeister (1985) studied temporal modulation transfer functions in normal-hearing and hearing-impaired listeners, illustrating both the utility of the paradigm and the risk of treating one average function as representative of all hearing conditions. Auditory loss may change audibility, dynamic range, frequency selectivity, and the weighting of temporal information in ways that complicate straightforward between-group comparison.
Envelope cues have particular significance for hearing speech under transformed acoustic conditions. Shannon et al. (1995) showed that speech recognition can remain possible when primarily temporal cues are preserved by an experimental signal-processing procedure. This result constrains theories of speech encoding but does not establish that fine structure is unnecessary for every perceptual task. The auditory chimera experiments of Smith et al. (2002) make the complementary point that different information components can be important for different judgments. Research on temporal modulation should consequently report which task is being measured: detecting modulation, identifying phonetic material, recognizing a talker, localizing a source, or judging quality. Generalization requires evidence connecting those outcomes.
Binaural Hearing and Interaural Timing and Level Differences
Binaural hearing combines acoustic information received at two ears. A sound arriving from one side can reach the nearer ear earlier and at a different level than the farther ear, creating interaural timing and level differences. Their availability depends on frequency, head geometry, distance, and environmental acoustics. Localization experiments estimate how an observer maps these cues onto judgments of azimuth, lateralization, or perceived position. The physical relationship between source location and ear signals can be calculated, but the perceptual use of each cue must be measured. A coherent model should predict errors and ambiguities under altered cues rather than assuming that every physical difference contributes equally to behavior.
Click-train experiments by Hafter & Dye (1983) investigated detection of interaural time differences as a function of temporal spacing and the number of clicks, demonstrating sensitivity to the structure of successive binaural cues. Closely related studies of interaural level differences showed that task performance also depends on temporal presentation and integration. Such paradigms make clear that binaural comparison is dynamically weighted, not reducible to an instantaneous measurement of a static delay or level. Explanations may involve temporal emphasis on particular events, adaptation, and the reliability of available cues, each requiring specific experimental predictions.
Binaural judgments are influenced by the difference between headphone-based lateralization and localization of an actual source in an environment. Earphones can impose interaural delays and level disparities without reproducing the full directional spectrum of the outer ear. Free-field stimuli include head movement, reflections, and distance-dependent changes. Brungart & Rabinowitz (1999) investigated the head-related transfer functions of nearby sources, supporting a richer stimulus description for spatial hearing than a single interaural timing value. Comparisons among laboratory techniques therefore require calibrating the full ear-input signals and defining the response coordinate system used to infer perceived location.
Localization in Rooms, Reflections, and Acoustic Environments
Real rooms generate direct sound and delayed reflections, producing signals that vary with source distance, surface properties, and listener position. Reflections can create misleading localization cues or, under some conditions, provide useful information about environment and distance. Hartmann (1983) investigated the localization of sound in rooms, contributing controlled evidence on spatial perception under reverberant conditions. Such studies require a careful description of source geometry and room properties because the signals at the two ears are determined by both emitter and environment. A claim that a certain cue dominates localization should be tied to the tested range of delays, levels, and acoustic configurations.
Reverberation is often summarized by a decay measure, but the perceptual consequences of a room are influenced by more than a single scalar. Early reflections, direct-to-reverberant ratio, source spectrum, and binaural coherence may each matter depending on the task. An observer localizing a brief click receives different information from one locating a continuous broadband sound. Measuring only average angular error can hide distinct kinds of mistakes, including front–back confusion, distance bias, and occasional large deviations. Experimental studies should therefore preserve response distributions and test specific predictions of cue-integration models across several stimulus classes rather than declaring one room effect universal.
Knowledge acquired from repeated exposure to a room can change listening strategies and expectations, introducing an interaction between acoustic structure and perceptual learning. This makes ecological validity important: novel stimuli in an anechoic chamber provide precise control, while familiar or reverberant contexts test whether inferred mechanisms transfer to ordinary situations. Brungart & Rabinowitz (1999) and Hartmann (1983) together motivate experiments that vary physical distance and room acoustics in a factorial design. The scientific result sought is a transferable explanation of localization errors and successes, with clear limits on what the available signal could support.
Speech Perception as a Test of Auditory Representations
Speech is a complex acoustic signal with structured temporal envelopes, rapid transitions, harmonics, and spectral features whose perceptual relevance depends on the linguistic task. Its value for psychoacoustics is to test which auditory representations carry usable information. Rosen (1992) analyzed temporal information at acoustic, auditory, and linguistic levels, a distinction that guards against equating perceptual audibility with recognition of phonetic content. The detection of a modulation pattern, the discrimination of two syllables, and the understanding of a sentence rely on different combinations of sensory and learned representations. This dossier uses speech findings where they test general auditory constraints and leaves language-specific learning and comprehension to the corresponding research field.
The experimental transformations in Shannon et al. (1995) supported speech recognition using primarily temporal cues, challenging accounts that require faithful preservation of every spectral detail for all intelligibility. The chimera study by Smith et al. (2002) further separated contributions of temporal envelope and fine structure to different perceptual outcomes. These demonstrations are especially informative because stimuli were designed to dissociate information sources that normally covary. They also establish limitations: performance depends on language materials, listening condition, familiarity, task, and the particular transformations applied. A strong inference specifies which information was sufficient for the demonstrated behavior and does not generalize sufficiency across all forms of sound perception.
Psychophysical thresholds can help predict problems of hearing speech, but an isolated frequency or modulation threshold rarely determines speech understanding on its own. Additional factors include linguistic expectation, masker fluctuation, spatial cues, and the listener's experience. A rigorous research program combines elementary acoustic tasks with independent speech outcomes and tests whether a model estimated from one set predicts the other. For artificial speech-processing systems, comparisons should likewise isolate acoustic recognition from lexical knowledge, language modeling, and evaluation-set familiarity. The common principle is to state which auditory capacity is being tested and which broader competencies can influence the measured score.
Individual Differences, Hearing Status, and Population Generalizability
Auditory thresholds and perceptual functions vary among listeners because peripheral sensitivity, sensory history, neural coding, attention, task experience, and physiological state differ. A study involving young adults with clinically normal hearing establishes an empirical relation for its tested sample and acoustic conditions, and cannot by itself determine the distribution expected in older listeners or those with altered auditory sensitivity. ISO (2023) explicitly fixes the population and conditions for its equal-loudness contours; these limits are an integral part of interpreting the standardized model. When research compares groups, it should describe how hearing status was assessed and whether presentation levels were chosen to equate audibility, physical intensity, or sensation level.
Work on modulation sensitivity illustrates how heterogeneity can modify the interpretation of a psychophysical result. Bacon & Viemeister (1985) examined temporal modulation transfer functions in normal-hearing and hearing-impaired listeners, offering a concrete experimental basis for questioning the assumption of a universal temporal filter. A difference in performance can reflect changes in access to the carrier, nonlinear response, decision variability, or actual modulation coding. Distinguishing these explanations requires matched tasks and designs that make alternative predictions. Group-average curves may conceal individual response patterns relevant to mechanism and practical application.
Development, training, and musical or linguistic experience can further modify auditory judgments. Pitch discrimination, sound-quality ratings, and spatial-cue adaptation are each sensitive to the task and history of the observer. McDermott & Oxenham (2008) situates pitch perception in the context of musical hearing, while Houtsma & Smurzynski (1990) provides laboratory paradigms for studying complex-tone judgments. The challenge is to test whether an effect of experience transfers across different sound materials and measurement formats. A well-supported population claim requires representative recruitment or explicit boundaries, consistent calibration, transparent exclusion criteria, and analysis of uncertainty at both individual and group levels.
Psychoacoustics in Audio Technology, Environmental Sound, and Hearing Devices
Psychoacoustic models guide sound reproduction, coding, and the assessment of acoustic environments. An auditory-filter approximation can help predict which spectral components are likely to be obscured by a masker, while loudness calculations can provide a standardized basis for comparing changing programs. Moore & Glasberg (1983) and ISO (2017) support different parts of this applied chain. An application must respect the conditions of model development, including level range, time dependence, listener properties, and presentation device. A numerical prediction optimized for laboratory headphones does not automatically establish perceived naturalness or safety in a real room.
Spatial reproduction introduces related questions about the mapping between measured ear signals and perceptual location. HRTF-based systems can synthesize interaural differences and directional spectral cues, but their accuracy depends on device response, individual anatomy, head tracking, and the room or virtual scene. Kistler & Wightman (1992) and Brungart & Rabinowitz (1999) establish methodological foundations for representing HRTFs and their dependence on geometry. A convincing validation program measures more than directional accuracy: externalization, distance, front–back confusions, consistency across head positions, and performance with novel sources should be considered where relevant. Acoustical specifications and perceptual quality measures are distinct, complementary aspects of engineering verification.
Assistive hearing systems must additionally accommodate differences in auditory sensitivity and the nonlinear growth of loudness. Psychoacoustic experiments clarify which signal characteristics listeners can detect or distinguish, while clinical benefit requires appropriately designed outcome studies with the intended users. Bacon & Viemeister (1985) illustrates why auditory performance in hearing impairment cannot be inferred from normal-hearing modulation functions alone. Designers should state whether an algorithm targets audibility, comfort, speech intelligibility, or spatial awareness; each outcome needs its own validation. The interdisciplinary contribution of psychoacoustics is an experimentally grounded link between device signals and selected perceptual consequences.
Computational and Artificial Auditory Systems as Comparative Research Objects
Computational auditory models can implement spectral filtering, temporal analysis, pitch estimation, binaural feature extraction, and sound recognition. Their performance becomes scientifically interpretable when the input signals and evaluation tasks are specified with the same care used in human experiments. A model may recover a tone's frequency with high numerical precision while failing to predict the pitch judgments of human listeners for unresolved harmonics. Conversely, a system can classify environmental sounds accurately using statistical features that differ from the ones human listeners rely on. The scientifically relevant comparison involves matched stimuli, constrained task outputs, and competing mechanistic predictions rather than a superficial resemblance in reported labels.
Human auditory experiments provide useful adversarial tests of machine representations. The harmonic manipulations in Shackleton & Carlyon (1994), envelope–fine-structure exchanges in Smith et al. (2002), and masking comparisons in Glasberg et al. (1984) can be translated into controlled model-evaluation stimuli. These tests allow investigators to distinguish reliance on spectral periodicity, temporal envelope, source statistics, or a dataset-specific shortcut. If a model succeeds only when a recognizable recording category is present, its performance does not establish general psychophysical sensitivity. Mechanistic interpretability requires tracing how an input transformation affects intermediate computation and output under alternative signal conditions.
In the Artificial Era, comparative psychoacoustics can study artificial hearing functions as engineered signal-processing and decision systems with observable capabilities. A rigorous program characterizes signal detection, discrimination, calibration stability, transfer across acoustical domains, and adaptation when physical conditions change. It can then compare these performance profiles with corresponding human psychometric functions without assuming that shared outputs imply shared neural mechanisms or experiences. This functional framework connects the psychology of human hearing with research on artificial perception and sensor fusion, enabling systematic investigation of cross-system similarities and differences at explicitly defined levels of analysis.
Review Method, Source Selection, and Appraisal of Evidence
This monograph is a critical narrative synthesis of experimental psychoacoustics, auditory physiology, quantitative modeling, and technical standards. Its questions were derived from the registered research object: audibility, discrimination, masking, loudness, pitch, timbre, timing, and spatial hearing. The source corpus was selected to connect landmark original investigations with later direct tests of the same constructs, including studies published in 2024 and 2025. Primary experimental articles are used to support statements about measured outcomes; reviews establish the scope of established research programs; international standards document specific conventions. This work does not employ a preregistered search protocol, exhaustive database screening, formal risk-of-bias scoring, or pooled effect estimation, and its findings are presented as a scientific monograph rather than a systematic review or meta-analysis.
Evidence was appraised in relation to its inferential task: acoustic calibration and experimental manipulation, number and characteristics of listeners, degree of stimulus control, accuracy of operational definitions, and ability of a result to distinguish competing mechanisms. For example, the eight-listener forward-masking investigation Oxenham & Shera (2003) provides direct evidence for the tested cochlear-tuning procedure, while the seven-listener duration manipulation López-Ramos et al. (2024) reveals a specific source of method-dependent variability. Such results provide mechanistic constraints without representing population-wide prevalence estimates. The binaural-model study Moore et al. (2016) is evidence for the predictive performance of a specified mathematical model, which must be evaluated separately from a claim about biological implementation. These distinctions govern the language of inference throughout the chapters.
Bibliographic identifiers were traced to publication records and official standards where available. Access to some historical experiments was limited to abstracts, publisher metadata, or accessible excerpts; claims from those studies were restricted to confirmed information about the study, procedure, or general finding. Across evidence types, numerical estimates were retained only when verifiable from an original abstract or article, and no pooled quantitative conclusion was inferred from incompatible paradigms. The principal limitations of this synthesis are selective coverage of a large literature, heterogeneous tasks, historic differences in instrumentation, and incomplete direct full-text access for some sources. Future systematic reviews could preregister narrower questions on a specific task, use standardized extraction and bias assessment, and quantify study-level heterogeneity.
Evidence Synthesis, Model Comparison, and Causal Attribution
Evidence in psychoacoustics arises from several sources with different inferential strengths: controlled perceptual decisions, objective acoustic measurement, cochlear emissions, neural recordings, and model predictions. A successful synthesis does not collapse these measurements into one undifferentiated account. Studies of behavioral filtering Moore & Glasberg (1983) Oxenham & Shera (2003) estimate effective frequency selectivity under particular procedures; otoacoustic work Kemp (1978) Shera et al. (2002) provides complementary information about cochlear function. Agreement across them strengthens a model, but each contains assumptions and potential measurement artifacts. The most informative result is one in which the same parameter predicts independent measurements without post hoc adjustment.
Several theoretical disputes reflect incomplete identifiability from currently available tasks. In pitch perception, place-based and temporal explanations can account for overlapping sets of behavioral findings Shackleton & Carlyon (1994) McDermott & Oxenham (2008). In loudness research, alternative schemes for spectral integration, dynamics, and binaural combination may fit subsets of the same judgments Glasberg & Moore (2002) Moore et al. (2016). In temporal hearing, neural modulation responses and behavioral sensitivity are related but require a decoder model for a causal link Joris et al. (2004) Viemeister (1979). These disagreements create a precise experimental agenda: construct signals for which theories make different quantitative predictions, measure multiple response types, and predefine criteria for comparing model performance.
Methodological quality is established by acoustic validity, informative stimulus manipulations, adequate sampling, transparent uncertainty, and independent tests. Original experiments remain indispensable because their physical conditions and analytical choices determine what was actually observed. Review articles and standardized methods organize this evidence but do not substitute for its task-specific assessment. Quantitative synthesis must avoid comparing thresholds obtained under incompatible definitions as if they were measurements on one common scale. A cumulative account of hearing progresses when estimates can be traced to calibration, a particular psychometric rule, and a clearly stated model of inference.
Open Scientific Problems and the Next Generation of Hearing Research
A central unresolved problem is how representations of fine temporal structure, envelope, and spectral place are weighted across different sounds and listening conditions. Evidence from speech transformations Shannon et al. (1995) Smith et al. (2002) and harmonic-pitch experiments Shackleton & Carlyon (1994) demonstrates that a cue sufficient for one task may be insufficient for another. The next experiments should quantify cue reliability and integration when spectral resolution, temporal regularity, and level are manipulated factorially. Competing accounts should predict changes in behavior with fixed model parameters across several tasks, allowing genuine discrimination rather than a series of independent fits.
Another challenge is establishing the relationship between cochlear nonlinearity, central dynamics, and natural listening. Estimates obtained from simultaneous masking, forward masking, and otoacoustic procedures differ partly because they interrogate different processes and conditions Oxenham & Shera (2003) Shera et al. (2002). Natural environments add head movement, fluctuating competing sources, uncertainty about source distance, and acoustic reflections, as shown by spatial-hearing research Hartmann (1983) Brungart & Rabinowitz (1999). Experimental progress will require designs that preserve physical precision while testing transfer to more realistic signal statistics. Advances in individualized calibration and data sharing can make these comparisons more reliable.
A third challenge concerns the diversity of listeners and the changing conditions of technological hearing. Existing loudness standards ISO (2023) and temporal-processing studies of hearing impairment Bacon & Viemeister (1985) illustrate why generalization requires explicit populations and outcomes. Computational systems introduce new comparative opportunities, yet their evaluation needs robust tests of hidden stimulus shortcuts and cross-domain transfer. Future evidence should integrate longitudinal listening measures, independent acoustic replications, causal manipulations where feasible, and uncertainty-aware models. The shared goal is to explain how measurable sound structures support perception under controlled and ecologically meaningful conditions, with every claimed mechanism tied to discriminating observations.
Conclusion
Auditory perception and psychoacoustics constitute an integrated experimental science of the relations among physical sound, auditory processing, and perceptual judgment. The historical progression from loudness comparisons Fletcher & Munson (1933) to quantitative auditory-filter models Moore & Glasberg (1983), physiological evidence of active cochlear processes Kemp (1978), and contemporary synthesis Oxenham (2018) has established a rich set of reproducible phenomena and computational tools. Detectability is shaped by spectral and temporal interactions; loudness varies with frequency, level, duration, and binaural context; pitch can arise from harmonic relations; and localization depends on interaural, spectral, and environmental cues. Each finding is defined by specific tasks, observers, and acoustical conditions.
The strongest explanatory position relates distinct forms of evidence without erasing their differences. Behavioral thresholds constrain but do not uniquely reveal cochlear filters. Neural modulation tuning provides physiological possibilities but does not directly identify the perceptual decision rule. Standardized equal-loudness contours and calculation methods ISO (2023) ISO (2017) increase comparability while retaining explicit limits of applicability. Original experiments on temporal cues Viemeister (1979) Shannon et al. (1995), harmonic pitch Shackleton & Carlyon (1994), and spatial transfer functions Kistler & Wightman (1992) demonstrate how carefully chosen stimulus manipulations can adjudicate competing explanations. Scientific progress relies on model discrimination and independent prediction as well as on increasingly precise measurement.
The research program that follows from this synthesis is a multilevel account of hearing in which acoustic calibration, sensory coding, integration over time and frequency, task-dependent decisions, and listener variability are jointly specified. It supports applied investigation of sound reproduction and hearing technologies, while providing a methodologically controlled basis for comparisons with artificial auditory systems. The enduring question is how reliable information about acoustic events becomes available for judgment and action across changing environments. A satisfactory answer will connect physical stimuli to perceptual functions through models that predict observable performance beyond the experiments from which those models were originally constructed.
References
- Bacon, S. P., & Viemeister, N. F. (1985). Temporal modulation transfer functions in normal-hearing and hearing-impaired listeners. Audiology, 24(2), 117–134. https://doi.org/10.3109/00206098509081545
- Brungart, D. S., & Rabinowitz, W. M. (1999). Auditory localization of nearby sources. Head-related transfer functions. Journal of the Acoustical Society of America, 106(3), 1465–1479. https://doi.org/10.1121/1.427180
- Buus, S., Florentine, M., & Poulsen, T. (1997). Temporal integration of loudness, loudness discrimination, and the form of the loudness function. Journal of the Acoustical Society of America, 101(2), 669–680. https://doi.org/10.1121/1.417959
- Fletcher, H., & Munson, W. A. (1933). Loudness, its definition, measurement and calculation. Journal of the Acoustical Society of America, 5(2), 82–108. https://doi.org/10.1121/1.1915637
- Florentine, M., Buus, S., & Poulsen, T. (1996). Temporal integration of loudness as a function of level. Journal of the Acoustical Society of America, 99(3), 1633–1644. https://doi.org/10.1121/1.415236
- Frankenbach, C., Herzog, C., Seitz, J., & Fels, J. (2025). Evaluating localization accuracy of dynamic binaural synthesis: Differences between individual, estimated, and generic HRTFs. Journal of the Audio Engineering Society, 73(1–2), 5–23. https://doi.org/10.17743/jaes.2022.0190
- Glasberg, B. R., & Moore, B. C. J. (1982). Auditory filter shapes in forward masking as a function of level. Journal of the Acoustical Society of America, 71(4), 946–949. https://doi.org/10.1121/1.387575
- Glasberg, B. R., & Moore, B. C. J. (2002). A model of loudness applicable to time-varying sounds. Journal of the Audio Engineering Society, 50(5), 331–342. https://aes.org/publications/elibrary/elibrary-page/?id=11081
- Glasberg, B. R., Moore, B. C. J., & Nimmo-Smith, I. (1984). Comparison of auditory filter shapes derived with three different maskers. Journal of the Acoustical Society of America, 75(2), 536–544. https://doi.org/10.1121/1.390487
- Hafter, E. R., & Dye, R. H., Jr. (1983). Detection of interaural differences of time in trains of high-frequency clicks as a function of interclick interval and number. Journal of the Acoustical Society of America, 73(2), 644–651. https://doi.org/10.1121/1.388956
- Hartmann, W. M. (1983). Localization of sound in rooms. Journal of the Acoustical Society of America, 74(5), 1380–1391. https://doi.org/10.1121/1.390163
- Houtsma, A. J. M., & Smurzynski, J. (1990). Pitch identification and discrimination for complex tones with many harmonics. Journal of the Acoustical Society of America, 87(1), 304–310. https://doi.org/10.1121/1.399297
- International Organization for Standardization. (2017). Acoustics—Methods for calculating loudness—Part 1: Zwicker method (ISO Standard No. 532-1:2017). ISO. https://www.iso.org/standard/63077.html
- International Organization for Standardization. (2023). Acoustics—Normal equal-loudness-level contours (ISO Standard No. 226:2023). ISO. https://www.iso.org/standard/83117.html
- Joris, P. X., Schreiner, C. E., & Rees, A. (2004). Neural processing of amplitude-modulated sounds. Physiological Reviews, 84(2), 541–577. https://doi.org/10.1152/physrev.00029.2003
- Kemp, D. T. (1978). Stimulated acoustic emissions from within the human auditory system. Journal of the Acoustical Society of America, 64(5), 1386–1391. https://doi.org/10.1121/1.382104
- Kistler, D. J., & Wightman, F. L. (1992). A model of head-related transfer functions based on principal components analysis and minimum-phase reconstruction. Journal of the Acoustical Society of America, 91(3), 1637–1647. https://doi.org/10.1121/1.402444
- López-Ramos, D., Eustaquio-Martín, A., López-Bascuas, L. E., & Lopez-Poveda, E. A. (2024). Effect of stimulus duration on estimates of human cochlear tuning. Hearing Research, 451, Article 109080. https://doi.org/10.1016/j.heares.2024.109080
- Marggraf-Turley, N., Shiell, M. M., Pontoppidan, N. H., Cappotto, D., & Picinali, L. (2025). Electroencephalographic decoding of sound location: Comparing free-field to headphone-based non-individual head-related transfer functions. Journal of the Acoustical Society of America, 158(2), 859–870. https://doi.org/10.1121/10.0038635
- McDermott, J. H., & Oxenham, A. J. (2008). Music perception, pitch, and the auditory system. Current Opinion in Neurobiology, 18(4), 452–463. https://doi.org/10.1016/j.conb.2008.09.005
- Moore, B. C. J., & Glasberg, B. R. (1983). Suggested formulae for calculating auditory-filter bandwidths and excitation patterns. Journal of the Acoustical Society of America, 74(3), 750–753. https://doi.org/10.1121/1.389861
- Moore, B. C. J., Glasberg, B. R., Varathanathan, A., & Schlittenlacher, J. (2016). A loudness model for time-varying sounds incorporating binaural inhibition. Trends in Hearing, 20, Article 2331216516682698. https://doi.org/10.1177/2331216516682698
- Moore, B. C. J., Peters, R. W., & Glasberg, B. R. (1990). Auditory filter shapes at low center frequencies. Journal of the Acoustical Society of America, 88(1), 132–140. https://doi.org/10.1121/1.399960
- Oxenham, A. J. (2018). How we hear: The perception and neural coding of sound. Annual Review of Psychology, 69, 27–50. https://doi.org/10.1146/annurev-psych-122216-011635
- Oxenham, A. J., & Shera, C. A. (2003). Estimates of human cochlear tuning at low levels using forward and simultaneous masking. Journal of the Association for Research in Otolaryngology, 4(4), 541–554. https://doi.org/10.1007/s10162-002-3058-y
- Rosen, S. (1992). Temporal information in speech: Acoustic, auditory and linguistic aspects. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 336(1278), 367–373. https://doi.org/10.1098/rstb.1992.0070
- Shackleton, T. M., & Carlyon, R. P. (1994). The role of resolved and unresolved harmonics in pitch perception and frequency modulation discrimination. Journal of the Acoustical Society of America, 95(6), 3529–3540. https://doi.org/10.1121/1.409970
- Shannon, R. V., Zeng, F.-G., Kamath, V., Wygonski, J., & Ekelid, M. (1995). Speech recognition with primarily temporal cues. Science, 270(5234), 303–304. https://doi.org/10.1126/science.270.5234.303
- Shera, C. A., Guinan, J. J., Jr., & Oxenham, A. J. (2002). Revised estimates of human cochlear tuning from otoacoustic and behavioral measurements. Proceedings of the National Academy of Sciences, 99(5), 3318–3323. https://doi.org/10.1073/pnas.032675099
- Smith, Z. M., Delgutte, B., & Oxenham, A. J. (2002). Chimaeric sounds reveal dichotomies in auditory perception. Nature, 416, 87–90. https://doi.org/10.1038/416087a
- Viemeister, N. F. (1979). Temporal modulation transfer functions based upon modulation thresholds. Journal of the Acoustical Society of America, 66(5), 1364–1380. https://doi.org/10.1121/1.383531