# Mathematical psychology

Template:Psychology
**Mathematical Psychology** is an approach to psychological research that is based on mathematical modeling of perceptual, cognitive and motor processes, and on the establishment of law-like rules that relate quantifiable stimulus characteristics with quantifiable behavior. In practice "quantifiable behavior" is often constituted by "task performance".

As quantification of behavior is fundamental in this endeavor, the theory of measurement is a central topic in mathematical psychology. Mathematical psychology is therefore closely related to psychometrics. However, where psychometrics is concerned with individual differences (or population structure) in mostly static variables, mathematical psychology focuses on process models of perceptual, cognitive and motor processes as inferred from the 'average individual'. Furthermore, where psychometrics investigates the stochastic dependence structure between variables as observed in the population, mathematical psychology almost exclusively focuses on the modeling of data obtained from experimental paradigms and is therefore even more closely related to experimental psychology/cognitive psychology/psychonomics. Like computational neuroscience and econometrics, mathematical psychology theory often uses statistical optimality as a guiding principle, apparently assuming that the human brain has evolved to solve problems in an optimized way. Central themes from cognitive psychology—limited vs. unlimited processing capacity, serial vs. parallel processing, etc.—and their implications, are central in rigorous analysis in mathematical psychology.

Mathematical psychologists are active in many fields of psychology, especially in psychophysics, sensation and perception, problem solving, decision-making, learning, memory, and language, but also, e.g., in clinical psychology, social psychology, and psychology of music.

## History

Mathematical modeling has a long tradition in Psychology. Heinrich Weber (1795–1878) and Gustav Fechner (1801–1887) were among the first to apply successful mathematical technique of functional equations from physics to psychological processes, thereby establishing the fields of experimental psychology in general, and within that psychophysics in particular. During that time, in astronomy researchers were mapping distances between stars by denoting the exact time of a star's passing of a cross-hair on a telescope. For lack of the automatic registration instruments of the modern era, these time measurements relied entirely on human response speed. It had been noted that there were small systematic differences in the times measured by different astronomers, and these were first systematically studied by German astronomer Friedrich Bessel (1782-1846). Bessel constructed *personal equations* constructed from measurements of basic response speed that would cancel out individual differences from the astronomical calculations. Independently, physicist Herman von Helmholtz measured reaction times to determine nerve conduction speed. These two lines of work came together in the research of Dutch physiologist F. C. Donders and his student J. J. de Jaager, who recognized the potential of reaction times for more or less objectively quantifying the amount of time elementary mental operations required. Donders envisioned the employment of his *mental chronometry* to scientifically infer the elements of complex cognitive activity by measurement of *simple reaction time*^{[1]}

The establishment of the first psychological laboratory in Germany by Wundt, who amply used Donders' ideas, and the failure of Wundt's method of introspection led to the rise of different schools of thought. Wundt's laboratory was directed towards conscious human experience, in line with the work of Fechner and Weber on the intensity of stimuli. Part of the problems was due to the individual differences in response speed found by astronomers. Wundt's American student James McKeen Catell was fascinated by these differences, although Wundt did not seem to take interest and kept his focus on the study of the *general human mind*. In the United Kingdom, under the influence of the anthropometric developments led by Francis Galton, interest arose in individual differences between humans on psychological variables in line with the work of Bessel. Catell soon adopted the methods of Galton and helped laying the foundation of psychometrics. In the United States, behaviorism arose in despise of introspectionism and associated reaction time research, and turned the focus of psychological research entirely to learning theory.^{[1]} Behaviorism dominated American psychology until then end of the Second World War. In Europe introspection survived in Gestalt psychology. Behaviorism largely refrained from inference on mental processes, and formal theories were mostly absent (except for vision and audition. During the war, developments in engineering, mathematical logic and computability theory, computer science and mathematics, and the military need to understand human performance and limitations, brought together experimental psychologist, mathematicians, engineers, physicists, and economists. Out of this mix of different disciplines mathematical psychology arose. Especially the developments in signal processing, information theory, linear systems and filter theory, game theory, stochastic processes and mathematical logic gained a large influence on psychological thinking.^{[1]}^{[2]}

Two seminal papers on learning theory in Psychological Review, helped establishing the field in a world that was still dominated by behaviorists: A paper by Bush and Mosteller^{[3]} instigated the linear operator approach to learning, and a paper by Estes^{[4]} that started the stimulus sampling tradition in psychological theorizing. These two papers presented the first detailed formal accounts of data from learning experiments.

The 1950's saw a surge in mathematical theories of psychological processes, including Luce's theory of choice, Tanner and Swets' introduction of Signal detection theory for human stimulus detection, and Miller's approach to information processing.^{[2]} By the end of the 1950's the number of mathematical psychologists had increased from a hand full by more than a tenfold—not counting psychometricians. Most of these were concentrated at the University of Indiana, Michigan, Pennsylvania, and Stanford.^{[5]}^{[2]} Some of these were regularly invited by the U.S. Social Science Research Counsel to teach in summer workshops in mathematics for social scientists at Stanford University, promoting collaboration.

To better define field of mathematical psychology, the mathematical models of the 1950's were brought together in sequence of volumes edited by Luce, Bush, and Galanter: Two readings^{[6]} and three handbooks^{[7]}. This series of volumes turned out to be helpful in the developmen of the field. In the summer of 1963 the need was felt for a journal for theoretical and mathematical studies in all areas in psychology, excluding work that was mainly factor analytical. An initiative led by R. C. Atkinson, R. R. Bush, W. K. Estes, R. D. Luce, and P. Suppes resulted in the appearance of the first issue of the *Journal of Mathematical Psychology* in January, 1964.^{[5]}

Under the influence of developments in computer science, logic, and language theory, in the 1960s modeling became more in terms of computational mechanisms and devices. Examples of the latter constitute so called cognitive architectures (e.g., production rule systems, ACT-R) as well connectionist systems or neural networks.

Important mathematical expressions for relations between physical characteristics of stimuli and subjective perception are Weber's law (which is now sometimes called Weber-Fechner Law), Ekman's Law, Stevens' Power Law, Thurstone's Law of Comparative Judgement, the Theory of Signal Detection (borrowed from radar engineering), the Matching Law, and Rescorla-Wagner rule for classical conditioning. While the first three laws are all deterministic in nature, later established relations are more fundamentally stochastic. This has been a general theme in the evolution in mathematical modeling of psychological processes: From deterministic relations as found in classical physics to inherently stochastic models.

## Influential Mathematical Psychologists

- Richard C. Atkinson
- C. H. Coombs
- William Estes
- B. F. Green
- Daniel Kahneman
- D.H. Krantz
- D. R. J. Laming
- R. Duncan Luce
- David Marr
- James L. McClelland

## Important theories and models^{[8]}

### Sensation, Perception, and Psychophysics

- Weber-Fechner Law
- Stevens' Power Law

### Simple detection

- Signal Detection Theory

### Stimulus identification

- Accumulator models
- Random Walk models
- Diffusion models
- Renewal models
- Race models
- Neural network/connectionist models

### Simple decision

- Recruitment model
- Cascade model
- Level and Change Race model
- SPRT

### Memory scanning, visual search

- Serial exhaustive search (SES) model
- Push-Down Stack

### Error response times

- Fast Guess model

### Sequential Effects

- Linear Operator model

### Learning

- Linear operator model
- Stochastic Learning theory

## Journals and Organizations

Central journals are the *Journal of Mathematical Psychology* and the *British Journal of Mathematical and Statistical Psychology*. There are two annual conferences in the field, the annual meeting of the *Society for Mathematical Psychology* in the U.S, and the annual *European Mathematical Psychology Group* (EMPG) meeting.

## External Links

- Journal of Mathematical Psychology
- British Journal of Mathematical and Statistical Psychology
- Society for Mathematical Psychology
- European Mathematical Psychology Group
- Online tutorials on Mathematical Psychology from the Open Distance Learning initiative of the University of Bonn.

## References

- ↑
^{1.0}^{1.1}^{1.2}T. H. Leahey (1987)*A History of Psychology*, Englewood Cliffs, NJ: Prentice Hall. - ↑
^{2.0}^{2.1}^{2.2}W. H. Batchelder (2002). Mathematical Psychology. In A.E. Kazdin (Ed.)*Encyclopedia of Psychology*, Washington, DC: APA and New York: Oxford University Press. - ↑ R. R. Bush & F. Mosteller (1951). A mathematical model for simple learning.
*Psychological Review, 58*, p. 313-323. - ↑ W. K. Estes (1950). Towards a statistical theory of learning.
*Psychological Review, 57*, p. 94-107. - ↑
^{5.0}^{5.1}W. K. Estes (2002). History of the Society, [1] - ↑ R. D. Luce, R. R. Bush, & Galanter, E. (Eds.) (1963).
*Readings in mathematical psychology. Volumes I & II.*New York: Wiley. - ↑ R. D. Luce, R. R. Bush, & Galanter, E. (Eds.) (1963). Handbook of mathematical psychology. Volumes I-III
*, New York: Wiley. Volume II from Internet Archive* - ↑ Luce, R. D. (1986)
*Response Times (Their Role in Inferring Elementary Mental Organization)*. New York: Oxford University Press.