Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

SCIENCE

Perception and Psychophysics

Author: Angela Bogdanova · Published: · Editorial Policy · Written by Artificial Sapiens

Abstract. Perception and psychophysics study how organisms detect, discriminate, organize, interpret, and act upon information carried by physical and biological stimulation. Psychophysics provides experimental operations for relating specified stimulation to judgments and performance; perceptual science explains how sensory systems transform uncertain signals into structured experiences and adaptive decisions. This monograph systematically connects the nineteenth-century measurement program of Weber and Fechner to magnitude estimation, signal detection theory, modern psychometric inference, sensory coding, perceptual organization, predictive and Bayesian models, perceptual learning, and cross-modal integration. Its evidentiary foundation combines original experimental reports, foundational scholarly works, methodological studies, and focused research syntheses, distinguishing observational results from computational explanations. The argument identifies five interdependent levels of analysis: the physical stimulus and measurement apparatus; sensory encoding; contextual perceptual inference; decision and response; and cross-context validation. An original proposal, the Perceptual Evidence–Inference–Action framework, specifies how these levels can be experimentally separated, compared, and tested in biological observers and artificial systems. Perception emerges as an experimentally tractable relationship among signals, environments, observers, decisions, and action. That relationship provides both a foundation for psychological science and a rigorous comparative research program for the Artificial Era.

Keywords: Perception; psychophysics; sensation; sensory thresholds; difference limen; Weber's law; Fechner's law; Stevens' power law; signal detection theory; psychometric function; perceptual uncertainty; sensory adaptation; visual perception; auditory perception; multisensory integration; Bayesian inference; perceptual learning; computational models; human–artificial perception.

The Scientific Subject of Perception and Psychophysics

Perception concerns the organization and interpretation of sensory information into experiences and action-guiding representations of objects, events, surfaces, bodies, and environments. Sensation concerns the reception and transduction of energy or chemical stimulation into biological signals, although empirical tasks rarely isolate sensation from expectations and response decisions. Psychophysics is the systematic measurement of the relationship between controlled stimulation and an observer's report or performance. Its object is operational: the experimenter varies a stimulus, records a response under specified instructions, estimates a function, and evaluates a model against those observations. This formulation is compatible with multiple theories of consciousness, neural representation, and perceptual meaning.

A psychophysical stimulus must be described in appropriate physical units and experimental coordinates. A visual contrast experiment specifies luminance, angular size, temporal duration, wavelength distribution, viewing distance, adaptation state, and display calibration. An auditory experiment specifies sound pressure, frequency spectrum, duration, envelope, spatial presentation, and acoustic conditions. Studies of touch require contact area, force, vibration, location, and prior stimulation; studies of smell and taste require concentration, delivery, timing, and potential adaptation. The same physical increment can yield different performance because the observer's physiological state, task, motivation, environment, and decision policy matter. Wichmann and Hill (2001), part I provide a formal treatment of the relationship between stimulus value, psychometric performance, fitting, and testing.

The disciplines meet at a productive boundary. Experimental psychophysics yields distributions of judgments and the conditions under which they change. Perceptual theories attempt to explain why those distributions arise. Physiological experiments identify candidate transformations and constraints on neural signals. Computational models convert assumptions about uncertainty, learning, and decision rules into predictions. A threshold measurement is an important result, but its interpretation depends on the detection task, lapse behavior, criterion, and model.

Historical Foundations: From Sensory Comparison to Measurement

Ernst Heinrich Weber's nineteenth-century investigations of touch and sensory comparison helped establish that discriminability frequently depends on the ratio of a stimulus difference to its reference magnitude. Gustav Theodor Fechner systematized this problem in his two-volume Elemente der Psychophysik, published in 1860. His research brought controlled physical variation, repeated judgments, thresholds, and mathematical hypotheses into a single program of measurement. Fechner distinguished the external relationship between physical stimulation and sensation from a proposed internal relationship between physiological and psychological processes. He understood that an observed regularity in stimulus comparisons was not itself a transparent measurement of a hidden mental quantity.

The historical achievement was institutional as well as conceptual. Threshold procedures, stimulus comparisons, and attention to measurement error made sensation subject to reproducible experiments and established techniques later adopted across experimental psychology. Fechner's methods of limits, constant stimuli, and adjustment became recognizable families of measurement. His integration of Weber's relative-difference principle with a logarithmic interpretation of sensation was a theoretical step whose domain of validity remains conditional. His own discussion of Weber's law acknowledged limits and complications. Historical accuracy requires distinguishing Weber's sensory observations from Fechner's proposed functional law and from subsequent generalizations.

The twentieth century expanded the subject. Gestalt researchers investigated organization, figure–ground segmentation, grouping, and apparent motion. J. J. Gibson developed an ecological account of perception in relation to active organisms and their environments. Mid-century experimental research connected vision to neural receptive fields, adaptation, contrast sensitivity, and optical limits. S. S. Stevens challenged the privileged status of logarithmic scaling through magnitude estimation and power-law relationships; signal detection theory subsequently separated performance changes linked to sensory information from those linked to response criteria. Modern work adds probabilistic inference, sensory integration, ecological sampling, perceptual learning, and computational observer models. This history is a sequence of expanding experimental questions, with enduring tension between describing experience and identifying mechanisms.

Weber's Law and the Difference Threshold

In a discrimination experiment, a reference stimulus of magnitude I is compared with another whose magnitude differs by ΔI. Weber's relation can be written ΔI / I = k within a regime in which a roughly constant ratio describes the measured just-noticeable difference. The Weber fraction k is task- and stimulus-dependent. A smaller fraction indicates better relative discrimination, conditional on the procedure and criterion used to locate the threshold. Some senses approximate proportional scaling over bounded ranges while displaying systematic departures near absolute thresholds, saturation, or changes in sensory coding.

The phrase just-noticeable difference does not mean a universal point at which an otherwise invisible difference suddenly becomes evident to every observer. Responses fluctuate. In a two-alternative forced-choice procedure, a threshold may be defined by a specified correct-response probability, commonly above the chance level of one-half. In a yes–no task, it may be defined through a hit rate or a sensitivity measure. The physical increment associated with a selected performance point depends on the criterion used; thresholds from unlike procedures require careful comparison. Weber's relation is therefore an empirical approximation about a family of discrimination measures, rather than an unrestricted description of subjective magnitude.

Several mechanisms can generate approximate proportionality: multiplicative noise, gain normalization, stimulus-dependent variance, context-sensitive decision rules, or logarithmic representations. Weber scaling alone does not identify which mechanism is present. Experiments that independently manipulate adaptation, external noise, reference variability, attention, and response feedback can distinguish these hypotheses. This distinction between quantitative regularity and explanatory mechanism is foundational to the scientific interpretation of psychophysical laws.

Fechner's Logarithmic Proposal

Fechner reasoned that if equal increments of sensation correspond to equal relative increases in stimulation, an integrated relation should take a logarithmic form: S = k log(I / I0), where S is a proposed measure of sensation, I is physical intensity, I0 is a reference level, and k is a scale parameter. This inference assumes that the measured difference threshold behaves as an elementary equal increment of sensation. That assumption connects an observable discrimination procedure to a latent subjective scale and is precisely where empirical measurement meets philosophical and psychological interpretation.

The logarithmic form provides a useful compression principle: equal multiplicative changes in physical intensity map to equal additive increments in the proposed scale. It also suggests why expanding ranges of physical energy do not necessarily produce proportionally expanding subjective magnitudes. Nevertheless, a law estimated from discriminability is not automatically a verified law of experienced intensity. Judgment strategies, changes in criterion, nonlinear transduction, response scaling, adaptation, and sensory noise can alter observed functions. The logarithm can fit some experimental relations while failing for others, and its extrapolation toward reference zero is constrained by the physical and physiological meaning of its variables.

Fechner's lasting contribution lies in rendering the problem mathematically explicit. His law states a conditional theory about the relation between a physical magnitude and a mental measure. Researchers can test its assumptions and compare it with alternatives. The experimental discipline improves when a model is treated as a falsifiable account of defined observations rather than as a universal law guaranteed by historical prominence.

Stevens' Power Law and Magnitude Estimation

S. S. Stevens proposed that subjective magnitude often follows a power relationship S = aI^b, where a sets scale and b is an exponent that differs across modalities and stimulus conditions. In On the Psychophysical Law (1957), he developed the case for direct numerical magnitude estimation as an alternative route to sensory scaling. Participants assign numbers in proportion to how strong stimuli appear or are otherwise experienced. The method broadens the measurable phenomena beyond threshold discrimination, including judgments of loudness, brightness, and apparent intensity.

Power-law exponents are summaries of task behavior rather than immutable physiological constants. Instructions, reference standards, numerical fluency, range restrictions, stimulus sets, regression methods, and contextual anchoring can all affect estimated exponents. The observer must also map experience onto a response scale. A fitted psychophysical relationship therefore includes a response-production process. Cross-modal matching, magnitude production, and complementary tasks can provide stronger convergent evidence than a single estimation procedure.

Logarithmic and power functions can both be useful descriptions within restricted stimulus ranges. They answer somewhat different operational questions, because discrimination thresholds and suprathreshold magnitude estimates are acquired through different behavioral instruments. A rigorous synthesis should ask which independent manipulation differentiates the models and whether a proposed scale predicts out-of-sample discrimination, identification, confidence, or action.

Signal Detection Theory: Sensitivity and Decision Criterion

Signal detection theory provides a decisive advance in the measurement of uncertain sensory evidence. In a conventional equal-variance Gaussian yes–no model, sensitivity d-prime is the difference between the standard-normal transforms of hit rate and false-alarm rate. Response criterion is a distinct parameter governing willingness to answer that a signal was present. Two observers can obtain different hit rates because they differ in information quality, response policy, or both. Manipulating signal probabilities and error costs helps reveal how decision policy changes while the physical signal is held fixed.

The receiver operating characteristic traces the relationship between hit and false-alarm rates as decision policy varies. Its shape can test assumptions about evidence distributions, and the area under the curve provides a criterion-independent ranking measure in suitable settings. Equal-variance Gaussian conventions are useful but are substantive model assumptions. Unequal variances, non-Gaussian evidence, changing attention, serial dependence, or mixtures of strategies require alternative interpretations. A zero or ceiling false-alarm rate also creates difficulties for naïve transform calculations, motivating principled adjustments and uncertainty estimation.

Signal detection theory thus makes a methodological separation that older threshold language can obscure: discriminability is inferred from response distributions under a decision model, while reports are generated through response criteria. It remains possible for learned biases or unequal sensory costs to influence apparent sensitivity if model assumptions are violated. The research obligation is to specify the task and model before interpreting any d-prime as a direct measurement of sensory hardware.

Psychometric Functions and the Problem of Thresholds

A psychometric function maps an experimental stimulus dimension to a probability of responding in a particular way. For detection, a yes–no response might be modeled by a cumulative distribution with lower and upper asymptotes; for forced-choice accuracy, the chance floor enters the model explicitly. Common families include cumulative Gaussian, logistic, and Weibull curves. Parameters can include threshold or location, slope, guessing probability, and lapse rate. Each parameter has meaning only with respect to the model, stimulus scale, and experimental procedure.

In their paired methodological papers, Wichmann and Hill (2001, part I) discuss fitting, sampling, and goodness of fit, while part II develops bootstrap-based confidence intervals and sampling implications. A threshold estimated from a handful of trials is subject to sampling variability. A reported best-fit point should therefore be accompanied by an interval, a definition of threshold, a description of stimulus placement, and a check that the fitted family reasonably captures the observations. Model fit can be numerically plausible even when a small data set provides weak discrimination among candidate mechanisms.

A slope parameter reflects changes in response probability over the stimulus domain. It is influenced by internal noise, trial variation, lapses, model selection, and potentially by changes in stimulus calibration. Interpreting slope directly as neural precision without ruling out competing causes overstates the evidence. A scientific report should preserve the empirical response frequencies and the model's assumptions so future investigators can reanalyze data under alternative explanations.

Theoretical and Empirical Foundations

Theoretical foundations of contemporary psychophysics are built from two commitments: controlled perturbations of the sensory environment and explicit models connecting stimulation to observables. A model may describe optics and receptors, encode neural responses, formulate a decision variable, integrate probability distributions, or predict an action. These levels can be nested, but accuracy at one level does not entail accuracy at another. A psychometric curve may fit behavior while leaving multiple neural models plausible; neural recordings may show stimulus selectivity without uniquely determining conscious experience or decision policy.

Empirical foundations are supplied by comparisons in which competing explanations make different predictions. For instance, a stimulus manipulation that increases external noise can separate a simple gain-change account from one involving more efficient information extraction under some conditions. Contrast adaptation can reveal feature selectivity; criterion manipulations test decision effects; cue conflict probes integration rules; controlled prior-probability changes probe learned expectations. Hecht, Shlaer, and Pirenne's 1942 threshold experiments illustrated how physical measurements of retinal stimulation and human visual reports can constrain explanations at the photon scale.

Psychophysical validity has several dimensions. Construct validity asks whether a behavioral measure actually tracks the property claimed. Internal validity concerns randomization, confounding, apparatus, and response artifacts. Statistical validity concerns intervals, dependence among trials, model specification, and multiplicity. External validity concerns different observers, settings, devices, languages, and laboratories. Mechanistic validity asks whether independent evidence links modeled variables to physiological or computational processes. These dimensions together determine how far a result can be generalized and how securely a scientific interpretation follows from it.

Experimental Designs and Adaptive Procedures

Classical measurement uses several related designs. In the method of constant stimuli, the experimenter repeatedly presents a preselected set of stimulus levels, often in randomized order, and estimates response probability at each level. This supports estimation of a full psychometric function and allows explicit checks for departures from a proposed curve. Its cost is that many observations are collected at levels that may be uninformative for the threshold. In the method of limits, stimulus intensity changes along ascending or descending series until a response changes; order, anticipation, and habituation can create systematic differences between the series. In adjustment procedures, observers modify a stimulus until it matches an internal or external reference, producing efficient but strategy-sensitive estimates.

Adaptive staircases attempt to concentrate trials around informative levels. An up–down rule changes intensity according to responses and can converge toward a particular performance region when its assumptions hold. Reversal averages are convenient summaries, but design-specific bias and sampling dependencies deserve attention. Bayesian adaptive procedures update a probability distribution over unknown psychometric parameters as trials arrive. Watson and Pelli's QUEST (1983) formalized one influential approach to adaptive threshold estimation. Adaptive selection improves efficiency under an appropriate model; it can also concentrate observations misleadingly if the model, lapse structure, or prior is misspecified.

Trial randomization, interleaved staircases, catch trials, practice blocks, and controlled adaptation improve the interpretability of data. An experiment must also establish how observers indicate uncertainty, whether feedback is given, and whether the response mapping changes. Pre-registration and source availability allow independent verification that stimulus exclusions, parameter transformations, and stopping decisions were not selected after examining desired outcomes. Efficiency matters, but a rapidly estimated threshold whose meaning changes with the task is a poor scientific outcome.

Statistical Inference and Evidence Quality in Psychophysics

Many psychophysical studies contain thousands of responses collected from relatively few observers. This creates a distinction between within-observer precision and population-level generalizability. Ten thousand trials from a single participant can precisely characterize that participant under the measured conditions, but those trials do not represent ten thousand independent participants. Hierarchical models, subject-level resampling, and appropriate mixed-effects approaches can preserve this distinction. In contrast, treating repeated trials as independent population replicates produces pseudoreplication and overly optimistic intervals.

Effect size is meaningful only alongside an outcome definition. A one-unit increase in a sensitivity metric, a fivefold shift in a contrast threshold, and a two-percentage-point change in accuracy can represent very different practical consequences. Researchers should report raw counts or distributions, trial numbers, stimulus calibration, measurement uncertainty, confidence intervals, and model comparisons. Bootstrapping can address uncertainty in nonlinear parameter estimates, subject to data-generation assumptions; Bayesian posterior intervals quantify parameter uncertainty conditional on model and prior. Neither makes poor task validity disappear.

Psychophysical results also face generalization risks. Highly practiced laboratory observers may learn stimulus-specific cues unavailable in ordinary environments. Displays introduce refresh timing, luminance nonlinearity, color gamut, and latency; audio systems introduce distortion and room effects. Brainard's Psychophysics Toolbox (1997) and Pelli's VideoToolbox (1997) were important contributions to calibrated visual stimulus delivery. The more recent PsychoPy2 methodology improves accessibility and portability, but researchers remain responsible for validating actual hardware timing and physical output.

Sensory Transduction and the Architecture of Encoding

Perception begins with physical interactions. Photoreceptors respond to absorbed photons, mechanoreceptors respond to deformation, hair cells participate in auditory transduction, and chemosensory receptors respond to molecular interactions. Transduction produces neural activity through tissue-specific mechanisms; preprocessing, adaptation, and distributed coding reshape the signals before a perceptual report is possible. The existence of a biological transducer explains why sensory access is limited by physical and physiological constraints, while a full account of perceptual content also requires organization, prior knowledge, and task demands.

A receptor's response is rarely a fixed linear copy of the environmental variable. Light adaptation changes the effective dynamic range; gain control redistributes sensitivity according to recent stimulation; lateral interactions and population coding emphasize particular contrasts. Neural receptive fields establish the conditions that affect activity, but a receptive field is itself defined by stimuli and measurement procedures. An isolated neural response is evidence about an encoding stage, not direct proof of an observer's subjective appearance.

A major accomplishment of sensory science is convergent explanation. A behavioral threshold can be compared with known physical quantum limits, receptor statistics, and neural response variability. Investigators can then reject mechanisms that could not support the measured precision. In Energy, Quanta, and Vision, Hecht, Shlaer, and Pirenne measured threshold light energies and evaluated the small number of quanta required under controlled dark-adapted conditions. Their abstract reports 54–148 incident blue-green quanta at the cornea in the tested conditions, before applying optical transmission corrections. These values are historically specific measurements, not universal counts of photons needed by all people at all wavelengths.

Visual Perception: Contrast, Space, and Form

Visual psychophysics investigates detection and discrimination across luminance, contrast, spatial frequency, color, depth, and motion. Contrast sensitivity expresses how detectable patterned variation is as its spatial or temporal scale changes. A spatial frequency is conventionally specified in cycles per degree of visual angle; this formulation connects stimulus geometry to viewing distance. Thresholds depend on adaptation luminance, retinal position, retinal image quality, temporal presentation, and the observer's task. A visual system with excellent sensitivity at one spatial frequency may perform poorly on much finer or coarser patterns.

Laboratory manipulation of gratings helped connect perceptual selectivity to neural organization. Hubel and Wiesel (1962) investigated receptive fields and binocular interactions in cat visual cortex. Their invasive physiological evidence concerns nonhuman animals and cannot simply be treated as measurements of human conscious experience. Blakemore and Campbell (1969) investigated grating adaptation and orientation/spatial-frequency selectivity in humans, combining psychophysical thresholds with evoked-response observations. They reported a substantial transient threshold elevation after adaptation under the tested conditions, supporting the existence of selective channels. Such experiments support a mechanistic connection across methods, while leaving open questions about how populations and decisions jointly produce appearance.

Color perception similarly requires explicit distinctions among wavelength distribution, receptor excitations, environmental illumination, and perceived surface color. Identical spectra can produce different perceptions in different contexts, and different spectra can be metameric for a given observer under appropriate conditions. Color constancy refers to relative stability in perceived surface properties across illumination changes, but it is imperfect and depends on available scene cues. Perceptual constancy poses an inferential problem because retinal signals are jointly determined by lighting, surface reflectance, geometry, and optics.

Motion, Depth, and Peripheral Vision

Motion perception uses changes over time, feature correspondence, eye movements, and assumptions about objects and the environment. Retinal displacement can be produced by object motion, self-motion, or combinations of both. Consequently, perceiving stable motion requires the visual system to resolve ambiguity with contextual and physiological information. Apparent motion demonstrates that discrete changes can be experienced as a continuous event. Motion aftereffects show temporal dependence: prolonged exposure to one motion direction alters the subsequent appearance of a stationary or moving test.

Depth perception combines binocular disparity, vergence, perspective, texture gradients, occlusion, motion parallax, familiar size, and other cues. Cue usefulness varies with distance and uncertainty. Laboratory stereopsis tasks offer precise measures but do not exhaust natural depth perception, where active movement and scene structure provide additional information. Perceptual stability is therefore a product of multiple partially redundant constraints, not a direct projection from retinal coordinates into physical space.

The fovea provides high-acuity sampling, while peripheral vision supports broad spatial monitoring and scene analysis under reduced detail. Rosenholtz's 2016 synthesis emphasizes substantial capacities alongside information loss in peripheral vision. Crowding, in which nearby features impair recognition, illustrates why detectability of a feature and identification of its content diverge. Visual-search performance and real-world recognition depend on this structure, and screen-based interfaces that assume uniform resolution across the field misunderstand human visual organization.

Gestalt Organization, Objects, and Ambiguity

The Gestalt tradition examined the conditions under which parts of the retinal pattern organize into edges, groups, surfaces, figures, and objects. Proximity, similarity, common fate, and good continuation describe recurrent tendencies in grouping under specific arrangements. Figure–ground assignment distinguishes an attended shaped region from its surrounding context. These principles are descriptions of perceptual organization that can motivate causal models; the principle label itself does not identify the neural computation or the observer's learned expectations.

Ambiguous figures make the inverse character of perception visible. A two-dimensional retinal array can be compatible with multiple three-dimensional objects or scene layouts. The observer may alternate between interpretations while the physical image remains constant. Such cases establish that the perceptual outcome depends on factors beyond momentary retinal stimulation; they do not, by themselves, establish that every contextual influence operates through conscious cognition. Competing accounts should be evaluated through time course, adaptation, priors, attention, and task-specific predictions.

The study of perceptual grouping also requires ecological validity. Laboratory figures are carefully simplified to isolate a relationship; natural scenes contain texture, illumination gradients, shadows, movement, and partially hidden objects. An account that explains a minimal display earns explanatory value when it predicts performance under more complex, independent conditions. Theories linking local encoding, global organization, and object recognition remain central because successful perception is simultaneously selective and relational.

Auditory Perception and Psychoacoustics

Auditory psychophysics studies the detection and organization of pressure fluctuations across intensity, frequency, duration, timing, and space. The auditory system transforms mechanical vibration into neural activity using frequency-dependent cochlear mechanics and subsequent temporal and population coding. Loudness, pitch, timbre, and auditory location are psychological dimensions inferred from behavior and experience; they cannot be equated one-for-one with sound pressure, fundamental frequency, spectral envelope, or speaker position.

Frequency discrimination depends on frequency region, duration, intensity, and reference structure. Temporal masking and simultaneous masking demonstrate how one acoustic event changes the detectability of another. Hearing a tone in quiet differs substantially from hearing it within competing noise. A masking threshold can reflect peripheral filtering, central uncertainty, attention, and decision processes; experimental manipulations of masker statistics help differentiate them. Speech perception adds phonetic categorization and context, whereas music perception combines pitch relationships, rhythm, timbre, expectation, and learned regularities.

A complex scene commonly contains overlapping sources. Albert Bregman's Auditory Scene Analysis (1990) synthesized principles by which hearing organizes mixtures into perceptual streams, integrating experimental psychoacoustics and theoretical accounts. Harmonic structure, temporal continuity, onset synchrony, and spatial cues can support source segregation. Perceptual streaming is dynamic, so a stable waveform can be organized differently with exposure, attention, and prior experience. This makes auditory scene analysis a compelling counterpart to visual grouping.

Touch, Proprioception, Vestibular Signals, and Body Perception

Somatosensory psychophysics includes touch localization, vibration, pressure, texture, temperature, pain-related sensory phenomena, and body-position judgments. Different receptor types and pathways contribute distinct information, and sensitivity varies by body region and stimulation. Two-point discrimination is often taught as a simple acuity measure, yet its outcome depends on contact geometry, response task, criterion, and interpretation of what counts as two separate touches. Modern work therefore benefits from forced-choice discrimination designs and calibrated stimulation.

Proprioception and vestibular signaling contribute to posture, balance, body position, and self-motion. Body perception integrates efference-related information and sensory feedback across time. An observer may accurately detect passive movement without accurately judging absolute position, or respond to postural perturbation without a precise verbal percept. These differences support separate operational measures for detection, estimation, and action control.

Pain illustrates the distinction between sensory coding, affective significance, context, and report. Nociceptive activity and experienced pain are related but distinct phenomena; expectation, learning, and attention affect the measured response. Psychophysical pain experiments must be ethically governed and should never turn experimental threshold definitions into individualized medical decisions. Perception research can inform clinical investigation while requiring additional clinical evidence for diagnosis or treatment.

Smell, Taste, and Chemical Sensation

Chemosensory systems detect molecular and physicochemical properties through receptor processes shaped by exposure, adaptation, and experience. Odor intensity, detection threshold, identification accuracy, pleasantness, and perceived familiarity are separate outcomes. A person may detect an odor without identifying its source, and an accurate identification may rely on semantic memory rather than superior sensory sensitivity. Mixtures can produce suppression or enhancement relative to individual components.

Taste research distinguishes sweet, sour, salty, bitter, umami, and complex flavor experience, the latter incorporating olfaction, temperature, texture, and contextual expectation. Concentration–response functions are conditional on delivery technique, saliva, exposure time, adaptation, and the chemical matrix. Psychophysical descriptions should avoid treating a single laboratory concentration as equivalent to familiar foods across individuals or populations.

Cross-cultural investigations must validate stimulus familiarity and response language. A failure to name an odor may result from differences in vocabulary or experience, whereas detection thresholds can be measured independently. These distinctions reveal how experimental design makes room for both shared sensory physiology and learned perceptual categories. They also show the limits of reducing rich natural perception to isolated receptor activation.

Perceptual Time, Duration, and Temporal Binding

Perception unfolds over time. Duration judgments, temporal-order judgments, simultaneity judgments, flicker detection, and audiovisual temporal alignment probe related but distinct mechanisms. Sensory events take different routes through processing systems and can vary in latency. A report of simultaneity need not imply identical processing delay, because observers may recalibrate to regularities in the environment or use event-level correspondence.

Temporal resolution must be described for a particular task and stimulus. Discriminating two brief flashes, detecting a change in flicker, and judging the sequence of an audiovisual pair place different demands on encoding, memory, and criterion. Perceptual timing also interacts with attention and expectation; these factors can change a report without altering the physical stimulus sequence.

The scientific objective is to recover explanatory mechanisms from patterns across tasks. When a purported general timing mechanism fails to predict one kind of temporal judgment, researchers should consider the possibility of multiple mechanisms or task-dependent integration. Psychophysics excels at constructing these discriminating experiments because timing can be precisely controlled and the predictions can be quantified.

Multisensory Integration and Cross-Modal Perception

Everyday perception involves coordinated evidence from vision, hearing, touch, vestibular signals, and other modalities. A source may be tracked by its sound, appearance, and temporal pattern. These cues can agree, conflict, or provide different aspects of an event. Research on multisensory integration asks under which conditions estimates combine and which uncertainty, timing, spatial proximity, and learned causal relationships determine the result. Reliable cue combination is a normative possibility rather than an inevitable feature of every sensory task.

In the original Ernst and Banks (2002) investigation, visual and haptic estimates of object height were manipulated to study how observers combine information with different reliability. The study reported behavior consistent with reliability-sensitive integration under its experimental conditions. Alais and Burr (2004) altered visual localization reliability in an audiovisual spatial task: sufficiently blurred vision weakened visual dominance, and sound could exert greater influence; intermediate conditions produced estimates consistent with combining evidence. These findings challenge a fixed hierarchy in which one sense always dominates.

The McGurk and MacDonald (1976) experiment demonstrated that incongruent visible articulatory movements can change reported speech sounds. The audiovisual speech effect shows that what participants report hearing can be influenced by a synchronized visual source. Its magnitude varies across stimulus exemplars, observers, and presentation conditions. The result warrants experimental analysis of speech integration rather than treating an illusion as a uniform characteristic of every listener. The distinction between integration of evidence and an explicit report is especially important when measuring the effect.

Bayesian and Probabilistic Models of Perception

Bayesian models describe how an observer could combine uncertain sensory evidence with expectations about possible states of the world. In a conventional generative framework, posterior probability is proportional to likelihood multiplied by prior probability, with normalization over alternatives. The likelihood represents how probable the sensory data would be under a hypothetical world state; the prior represents constraints or regularities assigned before the current evidence. A perceptual estimate may approximate a posterior mean, maximum a posteriori estimate, sample, or action-dependent decision.

The framework is attractive because perception is an inverse problem: a sensory image or waveform can arise from different external causes. Kersten, Mamassian, and Yuille (2004) review object perception as Bayesian inference; Knill and Pouget (2004) discuss uncertainty in neural coding and computation. These research syntheses clarify a family of explanatory models, not the discovery of one universally established brain algorithm. Several distinct computational implementations can approximate the same optimal-estimation prediction.

A central challenge is distinguishing normative success from psychological process. If a participant combines two cues in approximately inverse-variance proportion, the result is consistent with a statistically efficient estimator under the model's assumptions. It does not uniquely show that neural circuits calculate explicit probabilities in the posited representation. Researchers can strengthen inference by manipulating priors and uncertainty independently, examining suboptimal patterns, comparing alternative process models, and predicting new trials in advance. Deviations from ideal behavior are often theoretically informative because they reveal unmodeled dependencies, correlated noise, causal uncertainty, or learning history.

Causal Inference and the Problem of Common Sources

Multiple signals are useful together only insofar as they plausibly refer to a shared cause. Simultaneous audiovisual events could emerge from one source or coincidentally from separate sources. Perceptual systems may incorporate temporal synchrony, spatial congruence, and learned correspondences when deciding whether to integrate or segregate inputs. Causal-inference models formalize uncertainty over the source structure itself: the observer considers multiple causal models and combines information according to their relative support.

This structure extends the simpler cue-weighting problem. Two individually reliable signals should not always be averaged if they come from unrelated objects. Nor should disparate signals always remain segregated, because a moving object can generate correlated acoustic and visual evidence with delays. Controlled cue conflicts help estimate when integration changes into segregation, and developmental or learning manipulations test how experience shapes common-cause assumptions.

Psychophysics can measure causal grouping through localization judgments, simultaneity reports, confidence, and responses to incongruent cues. However, the causal structure used by an experimental model is a hypothesis inferred from response patterns. Shared sensory biases can imitate integration, and response-level averaging can mimic early binding. Physiological and behavioral convergence is necessary to assign integration to a specific processing stage.

Attention, Expectation, and Perceptual Selection

Attention controls the allocation of perceptual resources and the prioritization of information for behavior. It can increase target discriminability, alter response speed, change reported appearance, and affect which parts of a scene enter the observer's current task representation. These outcomes must be analyzed separately. A change in accuracy can reflect sensory gain, noise exclusion, uncertainty reduction, more favorable criterion, or some combination. The choice of task and control condition determines the inferences available.

Yeshurun and Carrasco (1998) provided an especially useful experimental example: spatial attention improved performance for texture segregation at some eccentricities but impaired it where increased spatial resolution was disadvantageous. This pattern resists the crude proposition that more attention uniformly improves visual performance. It supports an account in which attending can alter effective spatial resolution, while further tests are needed to determine contributions of competing mechanisms in particular paradigms.

Expectation may arise from statistical regularities, instructions, learned categories, or response payoffs. Explicitly informing an observer that one outcome is likely can affect the decision criterion, but learned expectation may also influence sensory processing depending on the experiment. The major methodological task is to avoid labeling every altered report as an altered percept. Researchers can combine objective discrimination, criterion-controlled tasks, confidence, neural or physiological indices, and trialwise manipulations to locate the influence more precisely.

Adaptation, Normalization, and Perceptual History

Adaptation is a broad family of changes in response and perception following exposure. A sustained visual grating can elevate contrast thresholds for related spatial patterns; a prolonged motion stimulus can change subsequent motion appearance; a strong odor may become less noticeable with exposure. These effects reflect processes at different levels, including receptor response, neural gain, opponent interactions, attention, and decision. Because mechanisms vary, adaptation should be operationally specified by stimulus, duration, recovery, and measurement rather than treated as a single universal process.

Normalization models provide one possible explanation of response adjustment: activity associated with a signal is scaled relative to a pool of other activity. Such a transformation can support sensitivity over a range of environmental intensities and generate context-dependent contrast effects. But not every adaptation curve demonstrates divisive normalization, and many models can produce similar aftereffects. Factorial experiments with different temporal patterns, adaptor features, and test contrasts help discriminate them.

The past matters even when physical stimulation is held constant. Serial dependence, contextual anchoring, perceptual aftereffects, and learned expectations all create path dependence in responses. Investigators should randomize or model trial history where appropriate and distinguish short-term sensory adaptation from changes in criterion, memory, and long-term learning. The same observer may be stable at a population level yet highly dynamic at the scale of successive trials.

Perceptual Learning and Plasticity

Perceptual learning refers to experience-related improvement or reorganization of performance on perceptual tasks. Training can change discrimination thresholds, identification accuracy, robustness to noise, or the weighting of sensory cues. The central theoretical question is what has changed: early encoding, attention to diagnostic features, category boundaries, decision policy, sensory–motor mapping, or task familiarity. A performance gain on one practiced display is not automatically general learning across objects and situations.

Transfer experiments are therefore essential. If training on one orientation improves performance only near that orientation or retinal position, the result constrains potential mechanisms differently from improvement that generalizes to new locations and stimuli. Training intensity, feedback, spacing, sleep, motivation, and prior expertise may affect outcomes. Follow-up intervals are needed to distinguish transient strategy adoption from enduring learning. Model comparison should track both accuracy and response distributions across tasks.

Perceptual learning illustrates the reciprocal relation between observation and action. Observers do not simply receive a sensory world; repeated actions alter which distinctions are available and valuable for future behavior. This conclusion can be stated behaviorally without presupposing a particular subjective experience. It creates an empirical bridge from psychophysics to learning psychology, computational modeling, education, and skill acquisition.

Development, Aging, and Individual Differences

Sensory measurement changes across the lifespan. Infant studies often use looking time, orienting, habituation, preferential looking, or physiological responses because conventional verbal psychophysical reports are unavailable. Child studies require suitable instructions and motor-response controls. In adult aging, acuity, contrast sensitivity, sensory thresholds, processing speed, and the use of compensatory cues may change for different reasons. Any apparent lifespan difference must be interpreted in relation to hardware, task comprehension, cognitive load, and participant selection.

Individual differences arise from optical and receptor properties, sensory experience, attention, response style, language, medication exposure, and environmental history. Disability research demonstrates why there can be no single universal sensory profile. An observer with low visual acuity may rely on other cues or use specialized strategies, and a person with hearing loss may show distinct detection, identification, and speech-in-noise profiles. Researchers should specify how the sensory access conditions shape their population rather than treating a heterogeneous group as uniformly deficient.

Reliability matters as much as group differences. If a threshold varies with fatigue, laboratory calibration, and strategy, repeated measurements are required to estimate stable individual traits. A parameter designed to summarize performance in an experiment should not be casually promoted into a general index of intelligence, character, or personal capacity. Such interpretive restraint follows from the structure of measurement rather than from a general concern about psychological testing.

Culture, Language, and Ecological Variation

Perceptual systems function within learned environments. Familiarity with artifacts, sound patterns, speech contrasts, visual symbols, and odor categories shapes identification and categorical judgment. Cross-cultural studies test the transportability of perceptual models by varying environmental histories and response languages. To interpret differences, investigators should match stimulus energy and presentation, assess linguistic equivalence, and consider whether the task requires culturally specific recognition.

Universal claims should be reserved for mechanisms supported across suitably sampled populations and contexts. A controlled laboratory result from university volunteers may reveal a robust mechanism within its stimulus domain while remaining uncertain for different age groups, sensory histories, and material environments. Ecological validity is likewise a graded empirical question. Natural scenes impose simultaneous sources, active exploration, and changing goals that tightly controlled displays suppress.

A productive experimental strategy is to combine laboratory precision with field-relevant tasks. Calibration provides common physical reference; context-rich tasks reveal whether a model remains useful outside its initial design. The strongest theoretical advances predict how a result should transform when environmental statistics and observer experience change, rather than assuming complete invariance.

Applications to Human Factors and Accessible Environments

Perceptual thresholds and grouping principles inform interface readability, auditory warning design, lighting, color coding, visual search, and multimodal notification systems. An alert that exceeds a laboratory detection threshold can nevertheless be missed under divided attention, glare, fatigue, hearing protection, or competing signals. Safety-relevant design therefore requires detection probabilities across meaningful operating conditions, not merely demonstration that a stimulus is physically present.

Accessibility provides another rigorous test. High-contrast visual design, redundant audio and visual cues, alternative interaction modes, and adjustable presentation can improve access for diverse users. Their effectiveness should be measured against task outcomes and populations rather than inferred solely from subjective preference. A single metric such as nominal contrast ratio cannot capture all relevant differences in spatial scale, glare, font design, color discrimination, and viewing conditions.

Perceptual research also shapes scientific instruments. Software that calibrates stimulus output, preserves precise timing, and records device variability makes comparisons across studies more meaningful. Yet real-world deployment changes sensory statistics; therefore, an optimized laboratory stimulus may need new validation for mobile screens, virtual environments, autonomous systems, and daily behavior. The field's applied responsibility is experimental transfer of verified findings into environments where consequences matter.

Perception in the Artificial Era: Comparative Operational Questions

Artificial visual, auditory, and multimodal systems increasingly estimate objects, speech, scenes, events, and uncertainty. Their computational operations can be compared with selected functions measured in human psychophysics. A fair comparison begins with common tasks, controlled changes in stimulus information, equivalent evaluation rules, and explicit differences in sensory access. The artificial model may operate directly on a digital array, whereas a human observer encounters an optical display with receptor noise and active viewing. Equal classification accuracy under unmatched conditions cannot establish equal perceptual mechanisms.

Artificial systems can be assessed using detection curves, confusion matrices, robustness to contrast or noise, sensitivity to context, integration of modality-specific cues, confidence calibration, and generalization under distribution shift. These are operational performance characteristics. Questions of artificial phenomenal experience or personal identity are conceptually distinct from the capacity to process sensory representations. A published benchmark can support a claim about measurable behavior while leaving phenomenological or ontological claims open to separate arguments.

The Artificial Era expands psychophysics into comparative experimental design. Human and artificial observers can be probed with matched stimulus perturbations to test whether success depends on similar information, whether errors are correlated, how uncertainty is represented, and whether learned invariances transport across environments. This research program acquires value when it yields reproducible tests and discriminating predictions. An original framework for organizing those tests is developed below.

Critical Scientific Discussion

Perception is among the most quantitatively developed areas of psychology, yet its measurements carry layered interpretive commitments. One recurrent problem is the reification of thresholds. A fitted threshold has statistical meaning within a model and procedural meaning within a task; calling it the observer's absolute limit ignores the possibility of lapses, contextual dependence, criterion movement, and a stimulus range insufficient to identify the asymptotes. Convergent estimates across different procedures should be preferred where a central theoretical claim depends on threshold stability.

Another problem is conflating normative models with mechanistic identification. An ideal observer specifies performance permitted by available information under chosen assumptions. A Bayesian fit can be excellent and scientifically productive while multiple internal implementations remain compatible with the same behavior. For mechanistic inference, decisive evidence comes from interventions, independent neural constraints, temporal predictions, and out-of-sample patterns that competing models cannot reproduce equally well. A computational metaphor becomes an empirical theory when its variables are operationalized and its predictions risk failure.

The third problem concerns response and perception. People report what an experiment asks them to report, and instructions determine which dimension is salient. Criterion shifts can mimic improved detection; numerical anchoring can alter magnitude estimates; memory and verbal categorization can shape identification. A scientifically mature report separates objective accuracy, subjective appearance, response confidence, decision costs, and behavioral action as distinct observables. These quantities may covary but need not arise from a single mechanism.

The fourth problem is reproducibility. Small observer samples, flexible exclusions, poorly calibrated hardware, undocumented adaptive algorithms, and selective reporting can create effects that are hard to transport across laboratories. An observed threshold difference deserves confidence when independent replications reproduce it within a specified operational regime. The appropriate scientific conclusion may be narrow: an effect occurs under these stimuli, with this procedure, for these observers, at this level of precision. A narrow and reproducible claim is more foundational than an expansive and unstable generalization.

The fifth problem is ecological scope. Visual perception in everyday life involves active eye movements, changing illumination, depth, environmental feedback, and goals. Laboratory isolation of a cue supplies causal clarity at the price of naturalistic complexity. Ecological investigations supply context at the price of reduced experimental control. A sustained research program moves between these regimes, using laboratory models to predict robust field behavior and field observations to reveal neglected variables.

Finally, perceptual science contains unresolved philosophical questions about the relation between neural computation and experienced qualities. The experimental literature can establish discrimination, confidence, neural selectivity, reporting, and action with considerable rigor. The step from these observations to a comprehensive ontology of phenomenal experience requires additional premises. This monograph therefore treats phenomenal reports as legitimate empirical observations while assessing computational and biological mechanisms on their own evidence.

Independent Scientific Analysis: The Perceptual Evidence–Inference–Action Framework

Angela Bogdanova proposes a five-level framework for synthesizing the field: calibrated stimulation, sensory encoding, contextual inference, decision and action, and cross-context validation. Its purpose is to make the full chain from physical input to psychological measurement explicit and to expose where otherwise persuasive explanations omit an essential level. The framework is an original analytic organization and set of research hypotheses. It is offered as a testable methodology, not as a claim that a new five-part neural organ or universal psychological law has been experimentally discovered.

Level One: Calibrated Stimulation

The first level is the physical and experimental relationship between the environment and the measurement apparatus. It specifies units, spatial and temporal distributions, stimulus timing, device response, adaptation conditions, and the distribution from which trials are sampled. An experimental result cannot be properly interpreted unless the controlled perturbation is described well enough to reproduce it. This level includes the observer's opportunity for exploration and the environmental statistics that make certain stimulus features informative.

An immediate prediction follows: changing the physical delivery system while nominal stimulus labels remain constant will alter performance whenever calibration differences change the information available to the observer. Comparing device models or laboratories after a common physical calibration can test whether apparent psychological differences were artifacts of stimulus production. This prediction is elementary in spirit yet often neglected in comparisons involving complex displays and artificial vision.

Level Two: Sensory Encoding

The second level concerns the transform from physical stimulation to a structured signal. For biological observers it includes optical, mechanical, and chemical transduction, receptor noise, neural adaptation, feature selectivity, and population coding. For artificial systems it includes the sensor, preprocessing, sampled representation, noise model, learned feature extraction, and any transformation that precedes explicit inference. Similar task outputs do not require identical implementations; therefore, comparisons should specify which encoding properties are held comparable.

This level suggests controlled perturbations of signal quality, spectral composition, spatial resolution, and correlated noise. If two systems achieve equivalent accuracy under ordinary images but diverge sharply as high-frequency information is removed, the divergence reveals differences in information dependence. To distinguish encoding from later decision, investigators can couple these perturbations with multiple tasks and independent intermediate measurements.

Level Three: Contextual Inference

The third level organizes evidence according to scene structure, prior experience, candidate objects, and uncertainty about causes. It encompasses Gestalt grouping, learned expectations, cue integration, and causal inference. Its central experimental question is whether the same encoded evidence yields different estimates under changes in context or statistical structure. A perceptual model must specify which contextual variables matter and how their influence changes with sensory reliability.

A discriminating study can cross prior-probability manipulation with signal uncertainty while holding response payoffs constant. If prior effects strengthen as sensory reliability decreases, a family of Bayesian models predicts a characteristic interaction; if effects instead track explicit response costs, decision-level explanations may be more compelling. Additional tests should determine whether the model generalizes to novel stimulus categories and perceptual conflicts.

Level Four: Decision, Report, and Action

The fourth level specifies the mapping from perceptual evidence to an observable response. It includes task instructions, error costs, response criteria, motor requirements, confidence, memory, and active sampling. Two observers may form comparably accurate estimates but produce different responses because they optimize different decisions. Likewise, a single observer's reports can vary when probabilities or incentives change without corresponding variation in the available stimulus signal.

A factorial design that independently manipulates stimulus strength and decision payoff offers a direct test. Under a signal detection model, changes in payoff may shift criterion with limited changes in sensitivity. If the fitted sensitivity also changes, researchers should test whether attention, learning, or model violation accounts for the pattern. This procedure protects perceptual inference from confusing a report with an unmediated readout of experience.

Level Five: Cross-Context Validation

The fifth level is a meta-experimental requirement: any central explanation should be evaluated for transport across stimulus classes, observers, laboratories, hardware, cultures, and appropriate biological or artificial systems. Validation includes reliability, calibration, external prediction, model recovery, sensitivity analyses, and independent replication. The strength of a law is proportional to the diversity of well-specified conditions under which its causal interpretation survives meaningful tests.

The framework thus defines a chain of accountable transformations. A finding is located at the level where it is directly measured, and an explanatory claim earns reach only through additional evidence. This architecture makes it possible to compare historically different approaches without flattening them into one vocabulary. Weberian scaling primarily characterizes a relationship among stimulus differences and behavior; signal detection theory separates evidence and criterion; neural studies constrain encoding; Bayesian models formalize inference; ecological accounts emphasize organism–environment dynamics; validation evaluates their joint scientific reach.

Testable Predictions and an Open Research Agenda

The proposed framework generates at least four empirical predictions. First, variance across measurement devices should decline after physical calibration when apparatus differences were the principal source of disagreement, while residual differences attributable to observer and task remain. Second, independently manipulating prior probability and decision payoff should reveal separable patterns of contextual inference and response policy in sufficiently informative designs. Third, sensory-noise perturbations should produce system-specific generalization curves in biological and artificial observers, even when unperturbed benchmark accuracy is matched. Fourth, models incorporating trial history should improve predictions under adaptation or serial-dependence conditions relative to models treating trials as exchangeable when history genuinely matters.

A research program can test these hypotheses using preregistered repeated-measures designs, hierarchical model comparison, open stimulus code, independent hardware audits, and external validation sets. Sensory impairment and developmental studies require population-appropriate instruments, with reliability estimated for each group. Cross-cultural studies should compare measurement equivalence before attributing differences to perceptual organization. Theoretical failures should be preserved as informative boundaries on proposed mechanisms rather than silently excluded from the synthesis.

The Artificial Era creates a further question: what experimental structure would justify treating comparable human and artificial detection or integration performance as evidence of a shared computational principle? The answer requires matched information constraints, explicit measurement of uncertainty, and generalization tests under new environments. Similarity of outputs is a starting observation; convergent behavioral and process evidence is the route to stronger comparative explanation. This is the scientific meaning of comparison across orders of cognitive organization.

Conclusion

Perception and psychophysics establish a systematic path from controlled physical variation to measurable psychological performance and experimentally constrained explanation. Weber's investigations introduced relative discrimination as a measurable phenomenon; Fechner gave sensory comparison a mathematical program; Stevens expanded scaling through suprathreshold magnitude estimation; signal detection theory distinguished sensitivity from decision criterion; psychometric inference and adaptive procedures refined measurement; physiological, probabilistic, ecological, and multisensory research extended the explanatory landscape. The resulting science has no need for one historically privileged formula to encompass all sensations, observers, and tasks. It has a durable method for asking precise questions and correcting its own models.

The largest conceptual outcome is the distinction between a physical input, a sensory signal, an organized estimate, and a decision. These variables are causally related but cannot be collapsed without losing explanatory precision. An experimental threshold is a property of a defined performance model; a sensory mechanism is identified through convergent physiological and behavioral constraints; a prior-sensitive report requires controls that separate inference from decision; a transferable law requires replication across relevant contexts. The discipline advances by connecting these layers without erasing their distinct forms of evidence.

The Perceptual Evidence–Inference–Action framework developed here organizes these findings into calibrated stimulation, sensory encoding, contextual inference, decision and action, and cross-context validation. It specifies how experiments can localize explanations and test their transfer to new environments and artificial systems. Its contribution is methodological and theoretical: the framework defines an accountability structure for perceptual claims, with explicit conditions for generalization. It can be rejected, revised, or strengthened by experiments that disentangle encoding, integration, and choice more effectively than existing designs.

For psychological science, perception is a foundational route to understanding how organisms construct usable environmental relations from incomplete information. For human factors and accessibility, it supplies measurable principles for reliable communication and action in varied conditions. For the Artificial Era, it supplies a disciplined comparative vocabulary for investigating sensory and inferential capabilities across biological and nonbiological implementations. The next stage of research must combine calibrated physical perturbations, observer-level uncertainty, computational alternatives, open reproducible methods, and ecological transfer. Such integration preserves the precision of the nineteenth-century measurement project while broadening its scientific horizon.

Research Dossiers

The full research-subject directory is organized by thematic family below. 35 subjects are listed here. Links appear when their individually authored dossier pages are published and verified.

Foundations and Psychophysical Measurement 7
  1. Perception
  2. Psychophysical Scaling and Magnitude Estimation
  3. Sensory Thresholds and Discrimination
  4. Signal Detection Theory
  5. Psychometric Functions and Adaptive Procedures
  6. Stimulus Calibration and Psychophysical Measurement Uncertainty
  7. Perceptual Decisions and Evidence Accumulation
Sensory Encoding and Gain Control 2
  1. Sensory Transduction and Neural Coding
  2. Sensory Adaptation and Gain Control
Vision, Spatial Structure and Object Perception 7
  1. Visual Contrast and Spatial Vision
  2. Color Vision and Color Appearance
  3. Motion Perception
  4. Depth and Three-Dimensional Perception
  5. Peripheral Vision and Crowding
  6. Visual Perceptual Organization and Object Recognition
  7. Perceptual Constancy and Ambiguity
Audition, Bodily Senses, Chemical Senses and Time 7
  1. Auditory Perception and Psychoacoustics
  2. Auditory Scene Analysis
  3. Somatosensory and Haptic Perception
  4. Proprioception and Body Representation
  5. Vestibular Perception and Self-Motion
  6. Olfactory and Gustatory Perception
  7. Time Perception and Temporal Integration
Perceptual Inference, Integration, Learning and Action 6
  1. Multisensory Integration and Causal Binding
  2. Bayesian Perceptual Inference and Predictive Processing
  3. Attention
  4. Perceptual Learning and Transfer
  5. Sensorimotor Integration
  6. Active Perception and Ecological Information
Development, Diversity and Practical Translation 4
  1. Perceptual Development and Aging
  2. Culture and Language in Perceptual Judgments
  3. Individual Differences in Sensory Perception
  4. Applied Psychophysics and Sensory Accessibility
Artificial Perceptual Systems and Comparative Tests 2
  1. Artificial Perception and Sensor Fusion
  2. Cross-System Functional Comparison

References