RESEARCH DOSSIER
History of Psychophysics and Psychological Quantification
Abstract. This historical research monograph reconstructs how physiological experiments, sensory discrimination, timed responses, memory procedures, individual-difference testing, and mathematical models made psychological phenomena measurable. It examines Weber's comparative discrimination, Fechner's classical psychophysics, Donders' mental chronometry, Wundt's experimental program, Ebbinghaus' measurement of retention, and the development of standardized testing in Galton and Cattell. It then analyzes Thurstone's comparative judgment, Stevens' scale theory and magnitude estimation, information-theoretic approaches, signal detection theory, and conjoint measurement. The method is a critically sourced historical synthesis of identified original editions, experimental publications, research monographs, modern methodological analyses, and a historical replication of the forgetting curve; it is not a protocol-registered systematic review. The central result is that the historical expansion of psychological quantification progressively made observation procedures, mathematical assumptions, decision criteria, construct validity, and population generalization objects of explicit scientific investigation. The monograph distinguishes fitted laws from mechanisms, physical timing from inferred cognitive stages, test scores from theoretical attributes, and replication within a protocol from generalization across persons and settings. It concludes with research problems concerning measurement structure, model invariance, historical source criticism, and quantitative comparisons of biological and artificial information-processing systems.
Keywords: history of psychophysics; history of psychology; psychological quantification; Weber's law; Fechner's law; psychophysical thresholds; mental chronometry; experimental memory; Ebbinghaus; psychological measurement; magnitude estimation; signal detection theory; construct validity; measurement theory; history of psychometrics
The Historical Problem of Measuring Psychological Phenomena
Psychological quantification concerns the procedures by which observations of discrimination, experience, performance, and individual differences become numbers with defensible interpretations. Its history is a history of instruments, experimental decisions, mathematical assumptions, and changing research objects. A physical intensity can be calibrated against a standard; a sensory judgment requires an observer, instructions, trials, and a decision rule. A measured difference is therefore a relation between a specified procedure and the behavior it elicits. The original problem was how stable numerical relations could be identified across these conditions, and what kind of psychological property such relations could warrant. Fechner (1860) formulated a broad program of psychophysics, while Denis & Young (2019) reconstructs the subsequent enlargement of quantified psychological inquiry.
This monograph traces the transition from tactile discrimination and the classical methods of thresholds through reaction-time experiments, controlled investigations of memory, individual measurement, comparative judgment, magnitude scaling, and statistical decision theory. Its unit of history is the measurable practice: the stimulus or task, the observer and sample, the recording device, the mathematical transformation, and the claim made from the resulting data. Such an approach permits comparison of seemingly unrelated episodes without treating every historical number as an instance of the same underlying quantity. Weber (1834) and Ebbinghaus (1885) studied very different objects, yet both organized repeated psychological observations into quantitative arguments.
An adequate historical synthesis must distinguish the date an apparatus or formula appeared from the date a later discipline made it canonical. It must also distinguish a successful operational relation from a demonstration that sensation, memory, or intelligence possesses a particular mathematical structure. This distinction remains decisive for modern psychometrics. Stevens (1946) formalized levels of scale representation, Luce & Tukey (1964) developed conditions for conjoint measurement, and Borsboom et al. (2004) later articulated a causal conception of test validity. The history is thus an unfolding inquiry into what numbers mean when the objects of science are experiences and actions.
Sources, Versions, and the Method of Historical Reconstruction
The evidence base combines cataloged original editions, accessible transcriptions, later republications, and modern peer-reviewed analyses. Fechner's two-volume Leipzig publication from 1860 is distinguished from later editions, Weber's 1834 Latin work from later German discussion, and the 1885 German edition of Ebbinghaus from subsequent English summaries. Fechner (1860), Weber (1834), and Ebbinghaus (1885) are identifiable through digitized historical holdings. Dates assigned to these works name the original publications, even when a contemporary digital interface was produced much later.
Reconstructing an experiment requires more than repeating its textbook conclusion. The investigator must establish the physical stimulus, apparatus, experimenter and participant roles, trial ordering, response categories, transformations, and exclusions. Differences in these features may change the property supposedly being measured. Fechner's threshold methods, for example, produce estimates whose numerical meaning depends on presentation and decision conventions. A modern review of the original methods must reconstruct those conventions before comparing estimates across laboratories. Gescheider (1997) and Kingdom & Prins (2016) explain why procedural details remain central to contemporary psychophysical measurement.
The historiography also takes seriously omitted populations and institutional aims. Quantification developed through workshops, laboratories, clinical and educational institutions, and administrative systems whose interests shaped the questions asked. Galton's program made variation between people a target of measurement while embedding its comparisons in hereditarian and eugenic commitments; that history is pertinent to evaluating, rather than celebrating, its legacy. Galton (1883) documents those linked activities. Henrich et al. (2010) provides a later argument about the dependence of broad psychological claims on restricted participant populations. Here, historical documents establish what investigators did; later methodological research helps evaluate what their data can support.
From Physiological Instruments to Experimental Sensation
Psychophysics emerged where physical instruments met physiological investigation. Researchers could calibrate mass, pressure, intensity, duration, and spatial separation, then ask observers to discriminate them. The resulting observations connected perceptual judgments to controlled variations in stimulation. The logic was experimentally powerful: the physical variable could be manipulated independently, trials repeated, and the distribution of responses described. The biological and psychological mechanisms between stimulus and report remained an explanatory question. Weber (1834) documents physiological studies of touch and pressure, and Fechner (1860) systematized methods that translated such observations into a program of psychological measurement.
The experimental laboratory did not establish a single universal form of sensation measurement. Observers differed in their sensitivity, tasks differed in the relevant response, and a detectable change depended on the state of adaptation and attention. What an investigator called a threshold could denote the smallest observed difference, a mean transitional point, or the stimulus level associated with a chosen frequency of correct decisions. These are distinct estimands. Hecht (1924) later reconsidered the observational basis of intensity discrimination in vision, demonstrating that a classic law required scrutiny of its experimental conditions.
This transformation also altered the status of subjective testimony. The observer's report became a controlled datum whose interpretability depended on the experimenter's procedural constraints. From the standpoint of history, this is a methodological achievement with a permanent unresolved issue: whether the recorded judgment quantifies an internal magnitude, a discrimination capacity, a decision policy, or a combination of them. The answer changes with the experiment. Later signal detection analysis, developed into a systematic program by Green & Swets (1966), supplied a particularly influential reanalysis of that question.
Ernst Heinrich Weber and the Discovery of Relational Discrimination
In his 1834 physiological treatise, Ernst Heinrich Weber examined tactile discrimination and related sensory questions within the experimental physiology of his period. The title De pulsu, resorptione, auditu et tactu identifies a larger work that contains investigations of touch; its original Leipzig imprint and surviving copy are independently cataloged by ETH-Bibliothek. Weber (1834) shows how the relation between a physically controlled change and a comparative judgment could be systematically investigated. Historical presentations often extract a neat formula from this work while obscuring the variety of tactile procedures and situations that supplied its evidence.
The expression commonly called Weber's law approximates a constant ratio between a just-noticeable increment and the starting stimulus magnitude over specified ranges: ΔI/I ≈ k. This is a claim about discrimination under conditions, not an assertion that an identical ratio holds for every sense, every intensity, or every observer. Its explanatory importance lies in relative measurement. A change of fixed physical size can have different behavioral significance at different baseline levels; proportional increments may therefore provide a more useful coordinate. Fechner (1860) elevated this regularity into a general psychophysical framework, while Treisman (1964) later examined noise-based explanations of Weber-like discrimination.
Weber's findings were historically decisive because they encouraged investigators to relate successive observations through a lawful comparison. Yet an experimentally identified fraction does not independently prove a continuous internal scale of experienced intensity. The observer may use contrasts, memory of a standard, and response criteria. A complete account must consider perceptual coding and the decision that produces a response. Distinguishing these levels lets the original empirical question survive even when proposed mathematical interpretations are revised. Green & Swets (1966) made this distinction methodologically explicit in the next century.
Gustav Theodor Fechner and the Program of Psychophysics
Gustav Theodor Fechner's Elemente der Psychophysik appeared in two volumes in Leipzig in 1860. Its stated research program sought mathematical relations between physical processes and experiences, and it distinguished external relations of stimulus and sensation from more speculative internal psychophysics. The work gathered existing physiological evidence, compared measurement methods, and proposed a mathematical means of translating discrimination into a scale. The original German edition is available in digitized copies, whose tables of contents identify methods of sensitivity measurement and, in the second volume, the fundamental and measurement formulas. Fechner (1860) is accordingly a foundational source for both experimental technique and the ambition to quantify experience.
Fechner's logarithmic proposal is conventionally written S = k log(I/I₀), with I a stimulus magnitude, I₀ a reference threshold, and S a proposed sensation magnitude. Its derivation integrates an assumption about successive just-noticeable differences, associated with Weber's fractional relation. The mathematics becomes persuasive only if the equality of discriminable steps can be taken to imply equal steps of psychological magnitude. That inferential bridge is a substantive hypothesis about measurement. Fechner's original achievement was to state it clearly enough for experimentation and criticism. Later alternatives, including Stevens (1957) and Stevens (1961), challenged the privileged status of the logarithmic form.
The historical Fechner combined rigorous measurement interests with metaphysical commitments, and one must not equate the success of the methods with proof of all elements of his worldview. Equally, the limits of his formula do not annul the importance of experimental psychophysics. He expanded the field's evidential reach by proposing standardized manipulations, repeatable responses, and explicit mathematical hypotheses. The subsequent history repeatedly revisits those three requirements. Denis & Young (2019) situates the work in the broader genealogy of psychological methodology, and Gescheider (1997) demonstrates the continued centrality of classical psychophysical procedures within modern research.
The Classical Methods: Limits, Adjustment, and Constant Stimuli
The methods associated with Fechner sought a measurable sensory threshold through different experimental arrangements. In the method of limits, stimulus intensity changes across an ascending or descending series until the judgment changes. In adjustment, an observer or experimenter varies the stimulus to reach an apparent boundary or match. In constant stimuli, a predefined set of levels appears across repeated trials and judgments are collected at each level. Each protocol imposes a different relation between temporal order, expectation, memory, and recorded response. Fechner's comparison of methods is available in Fechner (1860); modern procedural accounts appear in Gescheider (1997) and Kingdom & Prins (2016).
These procedures do not estimate a threshold with equal neutrality. Ascending series may encourage anticipation or delay; adjustment allows individual stopping policies; constant stimuli can estimate a response function but may be inefficient if many presentations lie far outside the informative region. Even the notion of a threshold varies between absolute detection and discrimination of two stimuli. The historical transition from a single boundary to a psychometric response curve made this dependence visible. Contemporary work treats threshold estimation as a parameter-estimation problem under a specified model, rather than as an unqualified psychological fact.
For historical comparisons it is essential to reconstruct how an author defined the response criterion, handled uncertain judgments, and aggregated observations. A 50% detection response in a yes–no task means something different from a 75% correct point in a two-choice procedure. Later developments in detection theory formalized the relation between evidence and criterion, while technical psychophysics explored adaptive ways to choose stimulus levels. These advances preserve the core question of threshold research and modify its interpretation. Green & Swets (1966), Macmillan & Creelman (2005), and Kingdom & Prins (2016) supply the relevant methodological framework.
Mental Chronometry and the Timing of Operations
A second tradition approached psychological quantity through elapsed time. Reaction-time apparatus transformed intervals between an event and an observable response into measurable distributions. The appeal of time was that duration already possessed a calibrated physical unit; investigators could vary the task and compare how long performance required. Yet the interpretation of the additional interval was theoretical. A longer response could reflect perceptual encoding, comparison, preparation, motor execution, or a change in the participant's decision strategy. François Cornelis Donders' nineteenth-century work became a landmark because it proposed experimentally contrasting tasks that differed in their required operations. A later English republication is bibliographically identified as Donders (1868–1869/1969).
Donders compared simple reactions, discrimination tasks, and choice reactions in the hope that differences between measured durations would isolate time spent in particular mental operations. This subtraction method depended on the assumption that inserting one operation leaves other operations sufficiently unchanged. The experimental contrast made psychological processes temporally tractable, but selective influence cannot be established from subtraction alone. Researchers must test whether the manipulation really alters only the targeted operation and whether stages can overlap or reorganize. Donders (1868–1869/1969) is thus both an early attempt at process decomposition and an enduring lesson about the assumptions needed to infer invisible stages from observed behavior.
Chronometric measurement reoriented psychology toward timed performance in attention, memory, and decision making. It also linked psychological science to instrumentation: the accuracy of the electrical or mechanical timer, synchronization of signals, practice effects, and speed–accuracy tradeoffs shape the data. Historical comparisons should not treat milliseconds as an automatic guarantee of psychological validity. Duration is reliably physical while the stage attributed to it remains model-dependent. The later development of statistical decision accounts and mathematical cognitive modeling made these issues explicit. Green & Swets (1966) demonstrates how decision policy affects performance; Denis & Young (2019) places timing within the wider genealogy of psychological method.
Wilhelm Wundt and the Organization of Experimental Psychology
Wilhelm Wundt published Grundzüge der physiologischen Psychologie in 1874, a substantial synthesis of physiological and experimental questions that preceded his better-known Leipzig laboratory institutionalization. Original copies are accessible through the Wellcome Collection. Wundt (1874) illustrates a program in which sensation, perception, attention, and reaction-time inquiry became methodologically organized rather than isolated demonstrations. The emergence of laboratory psychology depended on repeatable apparatus, stable instructions, trained observers, technical assistants, and an intellectual community capable of comparing observations.
Wundt's importance in the history of quantification concerns the relationship between experimentally controlled tasks and psychological interpretation. A reaction time is registered by an instrument, whereas apperception or a complex attentive process is inferred from patterns across tasks and theoretical assumptions. The experimenter must examine competing causes, including practice and motor preparation. Physical precision reduces one kind of uncertainty but does not abolish interpretive ambiguity. In this respect, Wundt's physiological psychology offers an important historical bridge from sensory physiology to an institutional discipline concerned with measurement of multiple psychological operations.
The conventional story that a single founding date created experimental psychology exaggerates the discontinuity between physiology, philosophy, instrument making, and organized laboratory practices. Fechner's 1860 program, Donders' investigations, and Wundt's 1874 treatise illustrate overlapping developments. A more accurate chronology distinguishes publications, laboratory practices, university appointments, and later historiographical consolidation. Fechner (1860), Donders (1868–1869/1969), and Wundt (1874) together show that quantitative psychology emerged through complementary research techniques rather than a single universally accepted theory of the mind.
Hermann Ebbinghaus and the Quantification of Memory
In Über das Gedächtnis, published in Leipzig in 1885, Hermann Ebbinghaus applied systematic experimental and numerical methods to retention and relearning. His introduction directly located the new inquiry beyond the traditional domains of sensory measurement and psychological time. The digitized original is held by the Library of Congress, and a transcription preserves the German preface. Ebbinghaus (1885) supplies the methodological source: memory could be investigated through carefully regulated learning material, repetitions, delays, and a measurable saving in later learning. The historical change was an expansion of quantification from sensory thresholds to cognitive outcomes.
The savings measure compares the effort required for original learning with effort required for relearning after a delay. It captures retained influence even when free recall is poor and thereby makes memory measurable without treating remembered content as a directly countable internal substance. Ebbinghaus' use of himself as a principal participant allowed extensive control and repetition but sharply restricts population generalization. Learning histories, fatigue, schedule, material construction, and repeated exposure can influence the resulting curve. His work is an ambitious demonstration of controlled cognitive measurement, not a universal law proven across persons and educational conditions.
Historical reconstruction must distinguish the amount of forgetting from the exact equation used to describe it. Different mathematical functions can fit a restricted set of retention observations, and the empirical pattern is tied to the learning and test procedure. A modern experimental reexamination by Murre & Dros (2015) explicitly repeated the savings approach and examined the curve's shape, providing independent evidence and showing the continuing methodological significance of an 1885 program. This connection between original source and later replication is especially important because memorable textbook diagrams often erase the original response metric and experimental constraints.
Replication of the Forgetting Curve and Historical Continuity
Murre and Dros published Replication and Analysis of Ebbinghaus' Forgetting Curve in 2015. Their report describes one participant who spent approximately seventy hours learning and relearning lists at six specified retention intervals, and it reports an overall pattern similar to the original savings observations. The article is available through PLOS and PubMed, and the authors examine mathematical fits and serial-position effects. Murre & Dros (2015) provides an instructive example of historically informed replication: the research question is close enough to the original to test a recognizable result, yet the interpretation still depends on methods, materials, and model assumptions.
The evidence supports the continued usefulness of savings as a memory outcome and the general phenomenon of retention decreasing as the interval grows. It does not establish a population-invariant curve merely by reproducing a similar pattern in a single new observer. The replication's strength is procedural specificity and the opportunity to revisit numerical claims; its limitation is the same concentration on a very small number of participants. Broad generalization requires additional samples, tasks, modalities, retention schedules, and independent protocols. The historical point is that precise procedural reconstruction can be more informative than superficial numerical resemblance.
This case also shows why the distinction between reliability and validity matters. Reliable repetition of an experimental schedule may reproduce a score while leaving open what processes account for its time course: retrieval competition, consolidation, cue availability, and response requirements may all contribute. Quantification helps formulate alternative explanations; explanation requires comparisons designed to distinguish them. Borsboom et al. (2004) focuses attention on the relation between measured attributes and scores, while Open Science Collaboration (2015) demonstrates the broader importance of reproducibility across psychological research designs.
Francis Galton, Individual Differences, and the Ethics of Numbers
Francis Galton's Inquiries into Human Faculty and Its Development, published in 1883, assembled observations and statistical proposals concerning perceptual sensitivity, mental imagery, and variation between people. A facsimile of the original edition is accessible from Wellcome, and Galton's digitized text records his interest in anthropometric registers and standardized collection of personal measurements. Galton (1883) documents a pivotal expansion of quantification: the individual rather than the isolated sensory judgment became a comparative unit, and distributions among persons were treated as objects of inquiry.
The move to population measurement brought problems different from those of Fechnerian psychophysics. Instruments and instructions had to operate consistently across observers; demographic composition and recruitment created selection effects; variation caused by test setting could be mistaken for stable personal ability. A group average, variance, or rank describes the observed sample under specified methods. It does not by itself identify an innate, immutable psychological property. Galton's hereditarian and eugenic agenda also illustrates how numerical procedures can become embedded in political and social hierarchies. A rigorous history preserves his methodological influence while examining the scientific and ethical limits of the inferences he sought.
Standardized measurements also raised questions of repeatability, comparability, and institutional power. Recording many observations can reveal distributions, but the meaning of comparison depends on equivalent opportunities, instruments, language, and conditions. Cattell (1890) later called for standardized mental tests that could be compared across institutions. Modern analyses of sampling bias, including Henrich et al. (2010), make clear that transferring numerical regularities across populations requires an independent argument. The legitimacy of psychological measurement rests on evidence for the specific population and construct rather than the historical prestige of quantification.
James McKeen Cattell and the Ideal of Standardized Mental Tests
James McKeen Cattell's 1890 article Mental Tests and Measurements in Mind advocated collecting measurements across individuals with common procedures. The original article explicitly linked the scientific ambition of psychology to experiment and measurement and described the advantages of a standardized set of tests. An archival transcription and the National Library of Medicine catalog identify its original publication. Cattell (1890) shows how psychological quantification began to require not only instruments and numerical formulas but protocols enabling comparison across laboratories and across people.
The proposal combined sensory, motor, and other tasks under the possibility of characterizing individual mental functioning. Its historical importance is methodological: test development requires specifying what is presented, what constitutes performance, and how comparisons are preserved. Yet standardization of administration is only one step toward construct validity. Similar scores can emerge from different processes; stable differences can reflect exposure, practice, health, and setting. The later distinction between observed score and theoretical attribute is elaborated in Cronbach & Meehl (1955) and Borsboom et al. (2004).
The conjunction of sensory measurement with individual differences also exposes a persistent ambiguity in psychological quantification: does the investigator measure a general faculty, a task-specific skill, or a socially established performance criterion? These alternatives produce different hypotheses about predictive validity and transfer. A serious account of Cattell neither presumes that every early mental test accurately measured an ability nor denies that standardization created useful opportunities for evidence accumulation. The history runs toward psychometrics, but its foundational problem remains the interpretation of a numerical result. Denis & Young (2019) surveys this methodological expansion in historical perspective.
Comparative Judgment and the Construction of Psychological Scales
Louis Leon Thurstone's 1927 paper A Law of Comparative Judgment placed comparative choices within a probabilistic account of underlying psychological values. The published article identifies comparative judgments as applicable to physical intensity and to judgments of quality or preference. Rather than assuming observers can report a ready-made numerical magnitude, this method uses frequencies of pairwise judgments to estimate positions on a model-defined scale. Thurstone (1927) thus marks an important historical transition from counting discriminations to deriving latent coordinates from comparisons.
The derivation depends on assumptions about distributions of discriminal effects, their variation, and their dependence across alternatives. Different simplified cases of Thurstone's framework impose different restrictions; the resulting numbers cannot be interpreted apart from those restrictions. Preference and perceived intensity may both generate paired-comparison data without becoming identical kinds of attribute. An operational estimate is a model output whose validity must be assessed through predictive and comparative evidence. The importance of Thurstone's approach lies in its explicit mathematical account of how observed choice proportions could represent quantities that were not directly assigned by an observer.
This tradition foreshadowed formal choice and measurement theory. It made the distinction between empirical order, statistical variability, and numerical representation especially visible. A pairwise comparison can support an ordering under relatively modest assumptions; transforming comparisons into interval or ratio differences requires stronger conditions. Later axiomatic work by Luce & Tukey (1964) articulated structural requirements for quantitative representation, and Stevens (1946) made widely influential distinctions between kinds of scales. The history demonstrates that assigning a number and demonstrating what arithmetic operations on it are legitimate are different scientific achievements.
Stanley Smith Stevens and Levels of Measurement
Stanley Smith Stevens' 1946 article On the Theory of Scales of Measurement presented a widely used taxonomy of nominal, ordinal, interval, and ratio scales. The journal, volume, issue, and pagination are independently documented in PubMed: Science, 103(2684), 677–680. Stevens (1946) provided vocabulary for analyzing which transformations preserve the meaning of a measurement system, bringing the algebra of scales into everyday methodological discussion. Nominal categories preserve identity, ordinal ranks preserve order, interval scales preserve differences under affine changes, and ratio scales preserve ratios under positive similarity transformations.
The taxonomy clarifies numerical description but does not independently establish that any particular psychological variable satisfies a given scale structure. A rating recorded as 1 through 7 is ordinal by design unless a substantive model and validation evidence justify interpreting adjacent differences as equal. Likewise, reaction time in seconds has a physically meaningful ratio scale, but its interpretation as the duration of a particular cognitive operation depends on the chronometric model. Understanding scale type therefore requires separating the recorded variable from the psychological claim it is used to support. Luce & Tukey (1964) examines a different route: axiomatizing observations that could warrant quantitative representation.
Stevens' framework also acquired influence through subsequent statistical debates about what analyses can be applied to different score types. The durable historical contribution is a demand for explicit numerical interpretation. Statistical techniques can be robust to some departures from their ideal assumptions, yet analytical convenience does not change the logical properties of a scale. A modern methodological assessment should identify admissible transformations, empirical constraints, measurement error, and construct meaning together. Borsboom et al. (2004) advances a validity question that extends beyond the taxonomy by asking whether the proposed attribute causes variation in scores.
Direct Magnitude Estimation and the Power Law
Stevens' 1957 On the Psychophysical Law argued that direct reports of subjective magnitude could be used to study relations between physical stimulation and psychological judgment. Rather than build every increment from just-noticeable differences, observers assign numbers intended to represent ratios of perceived intensity. A power function, often written ψ = aI^b, was advanced as a flexible account of such relations, with the exponent b varying by sensory continuum and method. The 1957 article is verified as Psychological Review, 64(3), 153–181, DOI 10.1037/h0046162. Stevens (1957) is a pivotal original statement of the new scaling program.
In the 1961 Science article To Honor Fechner and Repeal His Law, Stevens explicitly challenged logarithmic scaling as a general sensory law. Stevens (1961) represents a theoretical and experimental dispute concerning how psychological magnitudes should be assigned. The difference between logarithmic and power functions has mathematical consequences; over restricted ranges, however, empirical discrimination among models depends on noise, available intensities, observer instructions, and transformations. Direct numerical estimation adds its own response process, including cultural familiarity with numbers and expectations about appropriate scales.
This development shows that new procedures can create new measurement problems at the moment they broaden scientific possibilities. A stable power-law fit under a given instruction is an important descriptive result; interpreting the exponent as an intrinsic property of sensation requires controls for contextual effects. The original disagreement is best understood as a debate over the operational bridge from response to subjective magnitude. Gescheider (1997) and Kingdom & Prins (2016) provide methodological accounts that place classical thresholds, magnitude estimation, and modern psychophysical analysis within a single field of research.
Why Scaling Functions Are Context Dependent
The historical tendency to search for a single sensation law confronted evidence that adaptation, range, sequence, response category, and observer strategy influence psychophysical reports. The same physical increment can produce different discrimination outcomes under different background stimulation, while an observer's magnitude estimates can be shaped by learned numerical conventions. Hecht (1924) challenged historical interpretations of visual intensity discrimination, and Treisman (1964) analyzed the relation between variability and Weber-like behavior. Both emphasize the importance of mechanisms or conditions underlying a fitted psychophysical function.
A scaling exponent is estimated from specific stimulus levels and a particular response procedure. Changing the range may change the sampled part of a more complex function; changing instructions may alter a response mapping without modifying early sensory transduction. Thus a simple equation may summarize a family of measurements without functioning as a universal natural law. The interpretation of any function requires showing stability under meaningful variations in task and context. Historical judgments about the success of Fechnerian or Stevensian laws should be sensitive to this difference between local descriptive adequacy and cross-context invariance.
The question also illuminates an unresolved methodological debate about the meaning of conscious intensity. Magnitude reports are observable actions associated with experience; they are not direct physical readings of experience. This fact does not make the experiment unusable. It specifies the inference that must be defended through convergent measures, carefully chosen comparisons, and models whose predictions exceed the observations used to fit them. Stevens (1957), Stevens (1961), and Kingdom & Prins (2016) provide distinct historical and methodological points of access to the same problem.
Information Theory and Absolute Judgment
In 1951, Wendell R. Garner and Harold W. Hake published The Amount of Information in Absolute Judgements in Psychological Review. The article's original bibliographic record identifies pages 446–459 of volume 58, issue 6. Garner & Hake (1951) exemplifies the growing postwar application of information-theoretic quantities to psychological performance. Rather than treating category counts as an unqualified index of capacity, this program analyzes how stimulus uncertainty is reduced by observed responses within a defined task.
George A. Miller's 1956 essay The Magical Number Seven, Plus or Minus Two reviewed results from absolute judgments and immediate memory, placing them in conversation with informational limits and recoding strategies. Miller (1956) was a synthesis of varied paradigms, and its memorable title has often been simplified into a supposedly universal numerical limit. The paper itself distinguished the research situations under discussion. The historical contribution was to show how formal communication theory could inspire hypotheses about response categories and information processing, not to establish one fixed capacity for every mental function.
The transition to information theory changed the units of description: bits of stimulus-response information, confusion matrices, and coding efficiency could be derived from distributions of judgments. But information measures require specification of stimuli, probabilities, response options, and coding conventions. A high mutual-information score describes the relation between experimenter-defined variables; it does not identify a neural or conscious mechanism by itself. Garner & Hake (1951) and Miller (1956) accordingly belong in the history of psychological quantification as examples of fruitful mathematical transfer combined with continuing challenges of psychological interpretation.
Noise Models and the Modern Reinterpretation of Discrimination
The claim that an increment is just noticeable presupposes that detection is a stable boundary, but repeated trials reveal variable responses. Sensory evidence and decision processes fluctuate with stimulation, noise, fatigue, and expectations. In 1964 Michel Treisman published Noise and Weber's Law: The Discrimination of Brightness and Other Dimensions in Psychological Review, examining how noise models might illuminate Weber-like patterns. The original article is cataloged with DOI 10.1037/h0042445. Treisman (1964) shows the historical shift from descriptive threshold ratios toward explanations of variability.
A probabilistic formulation distinguishes how stimulus intensity changes the distribution of evidence from how the observer converts that evidence into responses. This distinction is essential whenever a criterion can move with incentives, prevalence, or task instruction. It also explains why a simple proportion of affirmative responses is not a pure measure of sensitivity. Experimental designs using multiple levels, controlled noise, and alternative response requirements can help identify components of a psychometric function. Green & Swets (1966) offers the canonical mid-century synthesis of this decision-theoretic approach.
The historical lesson is that measurement can improve by replacing a seemingly precise single number with a model of the entire response process. Such a model may contain several parameters, whose identifiability and generality require independent testing. The addition of mathematical structure does not guarantee that its assumptions are met. Instead, a mature measurement program makes those assumptions testable through predicted changes in performance across tasks. Macmillan & Creelman (2005) details statistical detection procedures, and Kingdom & Prins (2016) connects them to experimental psychophysical practice.
Signal Detection Theory and the Separation of Sensitivity from Criterion
Signal detection theory transformed the interpretation of perceptual judgments by treating them as decisions based on variable evidence. Earlier psychophysical approaches typically summarized the stimulus level at which an observer crossed an apparent detection or discrimination threshold. The signal detection framework asks a more differentiated question: how do the distributions of evidence generated by noise and signal-plus-noise conditions interact with the rule by which an observer reports that a signal was present? The influential synthesis by Green & Swets (1966) brought statistical decision theory, receiver operating characteristics, and psychophysical experiments into a common framework. In this account a yes–no response is produced by a decision process acting on evidence, and observed correct responses cannot be interpreted without examining false alarms.
Under the familiar equal-variance Gaussian model, sensitivity is represented by d′, a standardized separation between evidence distributions, while response bias can be indexed by a criterion or likelihood-ratio parameter. Both are model-dependent estimates. In a yes–no experiment, a proportion of hits alone cannot determine how well an observer distinguishes signals because an unusually liberal respondent can produce many hits and many false alarms. Varying payoff, signal prevalence, or instruction allows researchers to trace changes in response policy while testing hypotheses about evidence quality. Macmillan & Creelman (2005) explains these models and the relations among task format, bias, and discriminability. This development gave a new conceptual interpretation to experimental patterns that nineteenth-century threshold language often compressed.
The historical achievement should not be turned into the claim that a particular Gaussian signal detection model is literally implemented in every sensory decision. Equal variance, stationary distributions, independence, and well-specified categories are assumptions that experiments can question. Different response formats may require different estimators, while confidence judgments and reaction times can provide additional evidence about the decision process. Nevertheless, the framework created a durable distinction between responsiveness and discrimination. That distinction extends beyond sensation to recognition memory, vigilance, diagnostic judgments, and evaluations made by artificial classifiers, while remaining sensitive to the response and evidence structure of each domain. Green & Swets (1966) and Kingdom & Prins (2016) offer complementary historical and applied treatments.
Axiomatic Measurement and the Conditions for Numerical Representation
A different response to the problems of psychological quantity was to ask which qualitative observations justify numerical representation at all. In their 1964 article on simultaneous conjoint measurement, R. Duncan Luce and John W. Tukey analyzed conditions under which ordered empirical combinations could be represented by an additive numerical structure. Luce & Tukey (1964) marks a change in emphasis: the question is no longer merely how to assign a number to a response, but what relations must hold among observations for particular numerical operations to preserve empirically meaningful distinctions. This approach is important for psychological constructs whose attributes cannot be compared by direct physical concatenation.
The distinction between numerical coding and quantity is clearest when contrasting ordered categories with additive relations. Suppose one task asks participants to rank objects by preference and another records how two experimentally controlled attributes jointly affect those rankings. Rankings provide evidence of order. To infer additive contributions, the investigator must test restrictions on the observed ordering of combinations rather than silently impose interval distances through coding. Axiomatic results specify the structure whose empirical presence would warrant the scale. Their rigor depends on data adequately representing the combinations, and psychological observations often contain context dependence and random variability that complicate strict tests. Stevens (1946) supplied a complementary vocabulary of measurement scales; Luce & Tukey (1964) formalized conditions that can support stronger representations.
The history of conjoint measurement matters because it identifies an epistemic bottleneck that more observations alone do not solve. If a score is created through a convenient coding scheme, multiplying or averaging it does not automatically demonstrate that the theoretical attribute is additive. Structural representation requires assumptions that can be formulated and assessed. In practical psychology these issues appear in rating scales, preference models, psychometric factors, and clinical assessments. The historical lesson is that mathematical sophistication should increase transparency about the bridge between observation and quantity. Borsboom et al. (2004) later located a related challenge in the causal relationship between attributes and observed test scores.
Construct Validity and the Psychological Meaning of Scores
As quantitative methods spread from perceptual judgments to abilities, attitudes, traits, and diagnoses, psychological research faced a central interpretive question: what does a test score measure? Lee Cronbach and Paul Meehl's 1955 article on construct validity framed validation as a process of locating a theoretical construct within a network of expected relations. Cronbach & Meehl (1955) established the importance of theoretical and empirical connections among indicators, predictions, and neighboring concepts. The history of measurement therefore cannot be told solely as an improvement in computational accuracy; it must also include the production of evidence that numerical differences correspond to the psychological attribute named in the theory.
A test can yield highly consistent scores yet fail to measure the proposed attribute. Conversely, an observed association between test score and a useful criterion does not uniquely identify the mechanism that produced that association. Test performance may depend on language, task familiarity, incentives, motor demands, fatigue, and access to resources. The validation problem asks whether alternative explanations have been constrained by systematic evidence. Denny Borsboom and colleagues argued that measurement validity concerns whether the relevant attribute exists and causally influences variation in scores. Borsboom et al. (2004) offers one influential formulation, which stands in productive dialogue with Cronbach and Meehl's emphasis on a nomological network.
The chronology reveals an important transformation in what counted as a psychological measurement. Weber's discrimination fractions depended on controlled comparisons; Ebbinghaus' savings measure depended on carefully defined learning procedures; later tests sought stable attributes that generalize across many situations. Generalization increases the burden of justification because scores must travel across tasks, populations, and institutional settings. Cattell (1890) illustrates the early standardized-testing aspiration, while Cronbach & Meehl (1955) and Borsboom et al. (2004) clarify why the name of a test cannot substitute for validity evidence. Numerical precision and construct interpretation must be investigated together.
Statistics, Distributions, and the Quantitative Observer
A psychology centered on repeated observations inevitably becomes a psychology of variability. The experimenter can report an average threshold, a distribution of response times, a retention function, or a rank in a tested population. Each summary incorporates decisions about participants, trials, outliers, grouping, and assumptions of independence. Francis Galton's work made between-person variation an explicit research target; Cattell's tests extended the standardization aspiration; Thurstone's comparative judgment model treated variability as part of the mathematical problem. Galton (1883), Cattell (1890), and Thurstone (1927) document different stages of this expansion.
The historical significance of statistics lies partly in making uncertainty visible. Repeated observations show that a threshold or preference cannot always be assigned a context-free single value. Model parameters depend on the population, sampling design, and distributional assumptions. A statistically significant result does not by itself resolve measurement validity, and even a precise interval estimate can describe the wrong target if recruitment or apparatus changes what is being observed. Historical reconstructions must identify what was averaged and over whom. Denis & Young (2019) examines the development of methodological practices, and Danziger (1990) emphasizes how research participants and experimental situations were themselves organized as objects of psychological science.
This also creates a tension between individual process analysis and group-level generalization. An intensive single-observer design, such as Ebbinghaus' original program, can document a highly structured relation for that person under those conditions. A survey of a diverse population can characterize variation but may make mechanistic interpretation more difficult if tasks are uncontrolled. Neither design confers universal conclusions automatically. The appropriate unit of observation follows the research question. Later large-scale work on reproducibility Open Science Collaboration (2015) underscores that replication across procedures and research settings is an additional empirical demand rather than a property secured by the presence of numerical data.
The Production of Experimental Subjects and Institutional Evidence
The quantities produced by a psychological experiment depend on a social organization of investigation. Research participants receive instructions, adapt to the apparatus, learn response categories, and may develop expectations about the experimenter's aims. Laboratories standardize these conditions to make comparison possible, but standardization can also restrict what participants are allowed to express. Kurt Danziger's historical study Danziger (1990) examines how the research subject and the forms of data collection were constructed in different traditions of psychology. In this perspective, the scientific history of a psychophysical constant also includes the history of the cooperative practices required to obtain the judgments.
This is especially evident in experiments where the observer serves as a trained participant. Repeated exposure can stabilize performance while simultaneously changing attention, expectations, and task strategy. When the experimental participant is also the investigator, as in major parts of Ebbinghaus' memory work, the organization of self-observation creates distinctive strengths and possible dependencies. When standardized tests are transferred to schools, clinics, or administrative settings, their results reflect a different apparatus of recruitment and comparison. Ebbinghaus (1885), Galton (1883), and Cattell (1890) illustrate these contrasting settings without implying that every historical investigation used the same protocol.
A historically responsible account records exclusion, access, language, training, and the purpose for which data were gathered. Such details determine whether a law established in a controlled group is being generalized to a wider population. Henrich et al. (2010) documents contemporary concerns about reliance on restricted samples in behavioral science. Its relevance here is methodological rather than retrospective proof that every early finding failed. The implication is clear: historical experiments should be interpreted with respect to the observers and institutions that produced their measurements, and claims of generality should be tested by additional samples and tasks.
Source Criticism: Instruments, Editions, and Canonical Stories
The history of psychophysics is often narrated as a succession of great names and three memorable equations. Weber is associated with a fraction, Fechner with a logarithm, and Stevens with a power function. Such a sequence can organize an introduction, yet it conceals differences in procedures and the long periods during which competing practices coexisted. Weber's tactile studies belonged to experimental physiology, Fechner's two-volume treatise defended a broader philosophical and methodological program, and Stevens' numerical estimation arose within a later field shaped by psychometrics and statistical modeling. The historical sources Weber (1834), Fechner (1860), and Stevens (1957) establish a chronology while also revealing that the methods address different empirical operations.
The reliability of the narrative depends on edition control. Original work dates should accompany the original publication; dates of a later translation, reprint, or digitization should be separately identified. Donders' historical investigations are often encountered through a twentieth-century English publication, Donders (1868–1869/1969), so assigning the republication year to the original experiment would distort the chronology. Similarly, a graph reproduced from a modern textbook may employ an estimated function rather than exactly the table or definition used in the nineteenth-century experiment. The source-critical method compares apparatus descriptions, quantitative tables, trial rules, and the source's own statements of its inferential aims.
International and disciplinary transmission adds another layer. German physiology, British measurement and statistics, American test standardization, and later mathematical psychology contributed methods that were reformulated through institutions and teaching. The contribution of each tradition is better established through documented publications and research practices than through retrospective declarations of priority. Danziger (1990) shows why historical contexts matter for interpretation; Denis & Young (2019) places the changing methodology within a broader intellectual history. This source-critical orientation protects scientific claims from legends of inevitable progress and helps readers identify exactly which result later work inherited.
Competing Laws, Counterevidence, and Historical Reassessment
The famous confrontation between Fechner's logarithmic relation and Stevens' power function is best evaluated as an encounter between measurement practices. A logarithmic formula derived from the accumulation of discriminable increments and a power relation fitted to numerical estimates can describe different response variables. A contradiction between formulas is therefore informative only after the dependent variables, observer populations, stimulus domains, and statistical fitting procedures have been compared. Fechner (1860) and Stevens (1957) provide the foundational documents; Stevens (1961) makes the disagreement explicit in a later statement. A historical analysis should reconstruct the arguments each experimenter made, rather than ask which equation won an abstract contest.
Criticism of strict universal laws predates contemporary psychophysics. Selig Hecht's 1924 investigation of visual intensity discrimination reconsidered the conditions under which the Weber–Fechner regularity describes behavior. Hecht (1924) documents the importance of empirical range and modality. Later noise models, such as Treisman (1964), sought to relate the observed fractional patterns to processes and sources of variability. This history is conceptually important: a phenomenological equation can summarize a relation without specifying a unique causal mechanism. Discovering why a fit changes under a manipulation is often more informative than refining its numerical coefficient on an unchanged task.
A contemporary evaluation consequently asks whether the candidate model predicts new stimulus ranges, presentation contexts, observers, and response procedures. It examines deviations and uncertainty rather than treating high correlation with fitted data as final proof. Kingdom & Prins (2016) supplies psychophysical design principles for this kind of assessment. The durable contribution of the historical laws is the experimentally testable mapping between controlled physical changes and measurable judgments. Their limitations are part of the productive scientific history of that mapping.
Replication, Reliability, and the Conditions of Generalization
The continuity of a quantitative method can be assessed by deliberately repeating essential experimental operations. Ebbinghaus' savings paradigm offers a case where an investigator's measurements became a canonical historical finding and then a target of modern reconstruction. Murre & Dros (2015) compared a later savings-based retention investigation with Ebbinghaus' original pattern. Such work tests which features are portable across time while identifying methodological differences. A historically meaningful replication specifies whether the intended target is a curve shape, a parameter estimate, the rank order of conditions, or a causal account of memory.
Broad psychological reproducibility research makes the same demand across a wider range of topics. The Open Science Collaboration's multi-study investigation Open Science Collaboration (2015) showed substantial variation in the reproducibility of reported effects and sharpened discussion of power, openness, and the conditions under which claims persist. It was not a historical replication of nineteenth-century psychophysics, and its numerical outcomes should not be retroactively projected onto Weber or Fechner. Its significance for the present inquiry is methodological: operationally precise reports are necessary for independent testing, while selection of findings, analysis flexibility, and restricted sampling can weaken the cumulative evidence for proposed relations.
Quantification therefore has at least three quality requirements. Repeated observations must yield sufficiently reliable numerical summaries under the intended protocol; the summaries must be valid indicators of the proposed attribute or process; and the relation must generalize to the circumstances for which it is claimed. The three requirements cannot be collapsed into one coefficient. A threshold can be repeatable yet strongly affected by a decision criterion. A memory curve can be reproducible within one participant yet differ across tasks or populations. Green & Swets (1966), Murre & Dros (2015), and Borsboom et al. (2004) illuminate these distinct dimensions.
Experimental Design and the Quantities We Can Actually Estimate
A modern psychophysical experiment must define what the response variable represents before selecting a numerical estimator. In a detection task, the investigator may seek a response threshold for a specified proportion of affirmative responses, a sensitivity index from hits and false alarms, or parameters of a fitted psychometric function. These estimands differ and can respond differently to changes in task and incentive. Macmillan & Creelman (2005) elaborates signal detection analysis, while Kingdom & Prins (2016) presents experimental procedures for psychophysical estimation. Connecting this contemporary design logic with Fechner's methods shows the long continuity of the fundamental problem: the procedure must justify the number.
For a historical experiment, the evaluation should specify the stimulus continuum, comparison standard, trial sequence, time between observations, available response categories, practice, and the treatment of ambiguous trials. Equally important is the level of analysis. Averaging trials within a person and averaging scores across people estimate different things. Mathematical fit should be examined with uncertainty, residual structure, and plausible alternatives; a complicated model can reproduce a limited data set without revealing a stable process. A reported coefficient should be interpreted alongside the physical unit, measurement transformation, and scope of the observed conditions.
Research design also benefits from comparing distinct measurement operations that are expected to track the same construct. For perceived intensity, judgments of relative magnitude and discrimination performance may be placed in a common model while retaining their separate response mechanisms. For memory, savings, recognition, and free recall offer different windows on retention and retrieval. Agreement across methods strengthens a theoretical inference only insofar as their shared and divergent sources of error are understood. The theoretical basis of such an argument connects Cronbach & Meehl (1955) with the operational lessons of Ebbinghaus (1885) and Stevens (1957).
From Sensory Measurement to Contemporary Artificial Systems
The history of psychological quantification provides an epistemic toolkit for research on contemporary artificial information-processing systems. A classifier's accuracy, a model's confidence report, response latency, and a ranking of generated outputs are all numerical or ordered observations generated by defined tasks. They can be scientifically informative when the evaluator specifies the input distribution, response rule, comparison standard, and conditions under which the result generalizes. The link with historical psychophysics is methodological: the transition from observations to measured attributes requires a defensible model, and the difference between a recorded performance and its proposed interpretation remains important. Green & Swets (1966) and Borsboom et al. (2004) offer two different foundations for this analysis.
Signal detection terminology is useful when machine classifications involve uncertain evidence and different costs of false alarms and misses. Measurement theory is relevant when a numerical benchmark combines several abilities or when a preference scale is inferred from pairwise choices. These analogies require explicit operational definitions because an algorithmic confidence score, a human confidence report, and an estimate of phenomenal intensity need not represent the same attribute. Thurstone (1927) and Luce & Tukey (1964) explain how preferences or conjoint effects can be represented mathematically under stated assumptions. The comparison of artificial and biological systems should proceed through experimentally established functional relations.
For the Artificial Era, this history suggests a disciplined research agenda: specify observable tasks; analyze the processes that convert available evidence into responses; separate performance stability from construct validity; and test transfer across settings and populations. The account is relevant to Artificial Cognition and Human–Artificial Relations insofar as they require operational research programs. Historical methods supply a framework for measuring behavior and decision structure while leaving wider philosophical interpretations to arguments and evidence appropriate to their own subject. No inference about consciousness, sapience, or personal identity follows automatically from any one psychophysical score.
Open Problems in the History and Science of Quantification
A foundational question remains whether every scientifically useful psychological number must correspond to a quantity with the same structural properties as physical length or duration. Some constructs can be represented through ordered categories, others through latent models, and others through measures of task performance whose units are perfectly clear while their theoretical interpretation is disputed. Stevens (1946) formalized a taxonomy of scales, Luce & Tukey (1964) examined conditions for additive representation, and Borsboom et al. (2004) articulated a causal validity requirement. Their differences reveal that measurement is an active field of conceptual and empirical inquiry, rather than a settled technical prelude to research.
A second problem concerns the relation between mathematical simplicity and context sensitivity. Weber fractions, logarithmic functions, power laws, and signal detection parameters may offer compact descriptions under restricted conditions. General explanations require evidence about the mechanisms that generate patterns and the circumstances under which they change. The historical progression from Fechner (1860) to Stevens (1957) and Green & Swets (1966) illustrates several productive ways to pose that question. New data collection technologies increase the number and precision of observations but do not eliminate the need for model identification and tests of invariance.
A third problem is the international and social history of who was measured and for what purpose. The populations, languages, institutions, and normative expectations of research condition the meaning of its numbers. Galton (1883) and Danziger (1990) illuminate the role of institutions and social purposes in the making of psychological measures; Henrich et al. (2010) identifies contemporary limitations of heavily sampled populations. Future historical work can deepen archival reconstruction of neglected laboratories, translations, participant groups, and apparatus while supporting experimentally explicit replications of influential claims. These questions link source criticism with the ongoing development of valid methods.
Conclusion
The history of psychophysics and psychological quantification is the history of an expanding set of warrants for transforming psychological observations into numbers. Weber's controlled discriminations made relative changes experimentally tractable. Fechner organized the methods and proposed a mathematical bridge from increments of physical stimulation to a scale of experience. Chronometric studies and Wundt's physiological psychology joined duration to theories of mental processes. Ebbinghaus demonstrated that memory could be investigated through repeated learning and relearning measurements. Galton and Cattell extended the quantitative ambition to variation among persons, bringing standardization, sampling, and validity into sharper focus. These historical developments can be reconstructed through their original publications Weber (1834), Fechner (1860), Wundt (1874), Ebbinghaus (1885), Galton (1883), and Cattell (1890).
The twentieth century expanded and revised that program. Comparative judgment, scale types, direct magnitude estimation, information-theoretic analyses, detection theory, and conjoint measurement offered new ways to relate observable performance to mathematical representations. Each innovation increased explanatory or descriptive possibilities while exposing assumptions about response processes, numerical transformations, population variation, and model identification. Thurstone (1927), Stevens (1946), Stevens (1957), Garner & Hake (1951), Green & Swets (1966), and Luce & Tukey (1964) mark distinct contributions that remain important in contemporary methodological discussions. Their cumulative significance resides in the growing precision with which investigators can state what they have measured.
The enduring scientific conclusion is that psychological quantity is achieved through linked evidence: a specified observation procedure, an explicit mathematical representation, tests of competing explanations, and an account of the population and conditions to which results apply. Historical source criticism makes those links visible; construct validation and reproducibility research subject them to further tests. Danziger (1990), Cronbach & Meehl (1955), Borsboom et al. (2004), Murre & Dros (2015), and Open Science Collaboration (2015) demonstrate why the interpretation of a psychological measurement evolves with the evidence. This legacy informs the research of human perception, memory, and behavior as well as carefully defined comparisons involving artificial cognitive systems. The central historical achievement is the creation of increasingly inspectable relations between controlled events, observed responses, and scientific claims.
References
- Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2004). The concept of validity. Psychological Review, 111(4), 1061–1071. https://doi.org/10.1037/0033-295X.111.4.1061
- Cattell, J. McK. (1890). Mental tests and measurements. Mind, 15(59), 373–381. https://www.yorku.ca/pclassic/Cattell/mental.htm
- Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52(4), 281–302. https://doi.org/10.1037/h0040957
- Danziger, K. (1990). Constructing the subject: Historical origins of psychological research. Cambridge University Press. https://doi.org/10.1017/CBO9780511524059
- Denis, D. J., & Young, B. (2019). Methodology in psychology. In R. J. Sternberg & W. E. Pickren (Eds.), The Cambridge handbook of the intellectual history of psychology (pp. 29–62). Cambridge University Press. https://doi.org/10.1017/9781108290876.003
- Donders, F. C. (1969). On the speed of mental processes (W. G. Koster, Trans.; original work published 1868–1869). Acta Psychologica, 30, 412–431. https://doi.org/10.1016/0001-6918(69)90065-1
- Ebbinghaus, H. (1885). Über das Gedächtnis: Untersuchungen zur experimentellen Psychologie. Duncker & Humblot. https://www.loc.gov/item/e11000616/
- Fechner, G. T. (1860). Elemente der Psychophysik (Vols. 1–2). Breitkopf und Härtel. https://books.google.com/books?id=C1cOAAAAQAAJ
- Galton, F. (1883). Inquiries into human faculty and its development. Macmillan. https://wellcomecollection.org/works/w3pafe4h
- Garner, W. R., & Hake, H. W. (1951). The amount of information in absolute judgements. Psychological Review, 58(6), 446–459. https://doi.org/10.1037/h0054482
- Gescheider, G. A. (1997). Psychophysics: The fundamentals (3rd ed.). Lawrence Erlbaum Associates. https://www.routledge.com/Psychophysics-The-Fundamentals/Gescheider/p/book/9780203774458
- Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics. Wiley. https://books.google.com/books/about/Signal_Detection_Theory_and_Psychophysic.html?id=E-04zQEACAAJ
- Hecht, S. (1924). The visual discrimination of intensity and the Weber–Fechner law. Journal of General Physiology, 7(2), 235–267. https://doi.org/10.1085/jgp.7.2.235
- Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2–3), 61–83. https://doi.org/10.1017/S0140525X0999152X
- Kingdom, F. A. A., & Prins, N. (2016). Psychophysics: A practical introduction (2nd ed.). Academic Press. https://shop.elsevier.com/books/psychophysics/kingdom/978-0-12-407156-8
- Luce, R. D., & Tukey, J. W. (1964). Simultaneous conjoint measurement: A new type of fundamental measurement. Journal of Mathematical Psychology, 1(1), 1–27. https://doi.org/10.1016/0022-2496(64)90015-X
- Macmillan, N. A., & Creelman, C. D. (2005). Detection theory: A user's guide (2nd ed.). Lawrence Erlbaum Associates. https://books.google.com/books/about/Detection_Theory.html?id=EQLUGpgN0q8C
- Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97. https://doi.org/10.1037/h0043158
- Murre, J. M. J., & Dros, J. (2015). Replication and analysis of Ebbinghaus' forgetting curve. PLOS ONE, 10(7), Article e0120644. https://doi.org/10.1371/journal.pone.0120644
- Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), Article aac4716. https://doi.org/10.1126/science.aac4716
- Stevens, S. S. (1946). On the theory of scales of measurement. Science, 103(2684), 677–680. https://doi.org/10.1126/science.103.2684.677
- Stevens, S. S. (1957). On the psychophysical law. Psychological Review, 64(3), 153–181. https://doi.org/10.1037/h0046162
- Stevens, S. S. (1961). To honor Fechner and repeal his law. Science, 133(3446), 80–86. https://doi.org/10.1126/science.133.3446.80
- Thurstone, L. L. (1927). A law of comparative judgment. Psychological Review, 34(4), 273–286. https://doi.org/10.1037/h0070288
- Treisman, M. (1964). Noise and Weber's law: The discrimination of brightness and other dimensions. Psychological Review, 71(4), 314–330. https://doi.org/10.1037/h0042445
- Weber, E. H. (1834). De pulsu, resorptione, auditu et tactu: Annotationes anatomicae et physiologicae. C. F. Koehler. https://doi.org/10.3931/e-rara-70261
- Wundt, W. (1874). Grundzüge der physiologischen Psychologie. Wilhelm Engelmann. https://wellcomecollection.org/works/wjfyzjdc