See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262798060

History of Probability

Chapter · January 2011

CITATIONS READS
6 9,758

1 author:

Jordi Vallverdu
Autonomous University of Barcelona
197 PUBLICATIONS   1,017 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

COST Action 16207 European network for Problematic Internet Usage View project

Foundations of Cognition View project

All content following this page was uploaded by Jordi Vallverdu on 03 June 2014.

The user has requested enhancement of the downloaded file.

 P Probability, History of

Kolmogorov AN () Foundations of the theory of probabil-  possibilities. But the first occurrence of combinatorics
ity. Chelsey, New York (Original work: Grundbegriffe der per se arose from Chinese interest in future prediction
Wahrscheinlichkeits Rechnung, , Berlin: Springer-Verlag)
through the  hexagrams of the I Ching (previously eight
Rudas T (ed) () Handbook of probability: theory and applica-
tions. Sage, Thousand Oaks
trigrams derived from four binary combinations of two
elemental forces, yin and yang).
In  Luca Paccioli defined the basic principles of
algebra and multiplication tables up to  ×  in his book
Summa de arithmetica, geometria, proportioni e propor-
Probability, History of tionalita. He posed the first serious statistical problem of
two men playing a game called “balla,” which is to end
Jordi Vallverdú when one of them has won six rounds. However, when
Universitat Autònoma de Barcelona, Barcelona, they stop playing A has only won five rounds and B three.
Spain How should they divide the wager? It would be another
 years before this problem was solved.
In  Girolamo Cardano wrote the books Ars magna
Five thousand years ago dice were invented in India (David (the great art) and Liber de ludo aleae (the book on games
). This fact implies that their users had at least a com- of chance). This was the first attempt to use mathemat-
mon sense approach to the idea of probability. Those dice ics to describe statistics and probability, and accurately
were not the contemporary cubical standard dice, but fruit described the probabilities of throwing various numbers
stones or animal bones (Dandoy ). They must surely with dice. Galileo expanded on this by calculating prob-
have been used for fun and gambling as well as for fortune- abilities using two dice. At the same time the quantifica-
telling practices. The worries about the future and the tion of all aspects of daily life (art, music, time, space)
absurd idea that the world was causally guided by super- between the years  and  made possible the numer-
natural forces led those people to a belief in the explanatory ical analysis of nature and, consequently, the discovery of
power of rolling dice. the distribution of events and their rules (Crosby ).
In fact, cosmogonical answers were the first attempt to It was finally Blaise Pascal who refined the theories
explain in a causal way the existence of things and beings. of statistics and, with Pierre de Fermat, solved the “balla”
The Greek creation myth involved a game of dice between problem of Paccioli (Devlin ). All these paved the way
Zeus, Poseidon, and Hades. Also in the classic Hindu book for modern statistics, which essentially began with the use
Mahabharata (section “Sabha-parva”), we can find the use of actuarial tables to determine insurance for merchant
of dice for gambling. But in both cases there is no theory ships (Hacking , ). Pascal was also the first to apply
regarding probability in dice, just their use “for fun.” probability studies to the theory of decision (see his Pen-
Later, and beyond myths, Aristotle was the strongest sées, ), curiously, in the field of religious decisions. It
defender of the causal and empirical approach to reality is in this historical moment that the Latin term “proba-
(Physics, II, –) although he considered the possibility of bilis” acquires its actual meaning, evolving from “worthy
chance, especially the problem of the game of dice (On of approbation” to “numerical assessment of likelihood on
Heavens, II, a) and probabilities implied in it. These a determined scale” (Moussy ).
ideas had nothing to do with those about atomistic chance In , Antoine Arnauld and Pierre Nicole published
by Leucippus and Democritus nor Lucrecius’ controver- the influential La logique ou l’art de penser, where we can find
sial clinamen’s theory. Hald () affirms the existence of statistical probabilities. Games and their statistical roots
probabilistic rather than mathematical thought in Classical worried people like Cardano, Pascal, Fermat, and Huygens
Antiquity; we can accept that some authors (like Aristotle) (Weatherford ), although all of them were immersed
were worried about the idea of chance (as well as about the in a strict mechanistic paradigm. Huygens is considered
primordial emptiness and other types of conceptual cul- the first scientist interested in scientific probability, and in
de-sac), but they made no formal analysis of it. Later, we  he published De ratiotiniis in aleae ludo. In  Pierre
can find traces of interest in the moral aspects of gambling Raymond de Montmort published his Essay d’Analyse sur
with dice in Talmudic (Babylonian Talmud, Book : Tract les Jeux de Hazard, probably the first comprehensive text on
Sanhedrin, chapter III, Mishnas I to III) and Rabbinical probability theory. It was the next step after Pascal’s work on
texts, and we know that in , Bishop Wibolf of Cambrai combinatoricsanditsapplicationtothesolutionofproblems
calculated  diverse ways of playing with three dice. De on games of chance. Later, De Moivre wrote the influential
Vetula, a Latin poem from the thirteenth century, tells us of Demensurasortis(),andyearslater,Laplacepublished
 Probability, History of P 

hisPhilosophicalEssayAboutProbability.Inthes,Daniel because it is erroneously supposed to do no harm (. . .) The

Bernoulli (Jacob Bernoulli’s nephew) developed the idea principle “same cause, same effect,” which philosophers
of utility as the mathematical combination of the quantity imagine to be vital to science, is therefore utterly otiose”
and perception of risk. Gottfried Leibniz at the beginning (Suppes , p. ). Nevertheless, scientists like Einstein
of the eighteenth century argued in several of his writings were reluctant to accept the loss of determinism in favor of
against the idea of chance, defending deterministic theories. a purely random Universe; Einstein’s words “God does not
According to him, chance was not part of the true nature play dice” are the example of the difficulty of considering
of reality but the result of our incomplete knowledge. In the whole world as a world of probabilities, with no inner
this sense, probability is the estimation of facts that could intentionality, nor moral direction. On the other hand, sci-
be completely known and predicted, not the basic nature entists like Monod (Chance and Necessity, ) accepted
of things. Even morality was guided by natural laws, as this situation. In both cases, there is a deep consideration
ImmanuelKantarguedinhisFoundationsoftheMetaphysics of the role of probability and chance in the construction of
of Morals (). the philosophical and scientific meaning about reality.
In  an influential paper written by the Reverend In the s there arose from theworks of Fisher ()
Thomas Bayes was published posthumously. Richard Price, and Neyman and Pearson () the classic statistical
who was a friend of his, worked on the results of his paradigm: frequentism. They use the relative frequency
efforts to find the solution to the problem of computing concept, that is, you must perform one experiment many
a distribution for the parameter of a binomial distribu- times and measure the proportion where you get a posi-
tion: An Essay towards solving a Problem in the Doctrine of tive result. This proportion, if you perform the experiment
Chances. Proposition  in the essay represented the main enough times, is the probability. If Neyman and Pearson
result of Bayes. Degrees of belief are therein considered wrote their first joint paper and presented their approach
as a basis for statistical practice. In a nutshell, Bayes pro- as one among alternatives, Fisher, with his null hypoth-
posed a theorem in which “probability” is defined as an esis testing, gave a different message: his statistics was
index of subjective confidence, at the same time taking the formal solution of the problem of inductive inference
into account the relationships that exist within an array of (Gigerenzer , p. ).
simple and conditional probabilities. Bayes’ theorem is a From then on, these two main schools, Bayesian and
tool for assessing how probable evidence can make a given Frequentist, were fighting each other to demonstrate that
hypothesis (Swinburne ). So, we can revise predic- theirs was the superior and only valid approach (Vallverdú
tions in the light of relevant evidence and make a Bayesian ).
inference, based on the assignment of some a priori dis- Finally, with the advent of the information era and all
tribution of a parameter under investigation (Stigler ). the (super)computer scientific simulations, Bayesianism
The classical formula of Bayes’ rule is: has again achieved a higher status inside the community of
P
experts on probability. Bayesian inference also allows intel-
P(B∣A)P(A)
P(A∣B) = , ligent and real-time monitoring of computational clusters,
P(B)
and its application in belief networks has proved to be
where our posterior belief P(A∣B) is calculated by multi- a good technique for diagnosis, forecasting, and decision
plying our prior belief P(A) by the likelihood P(B∣A) that analysis tasks. This fact has contributed to the increas-
B will occur if A is true. This classical version of Bayesian- ing application of parallel techniques for large Bayesian
ism had a long history, beginning with Bayes and con- networks in expert systems (automated causal discovery,
tinuing through Laplace to Jeffreys, Keynes, and Carnap AI…) (Korb and Nicholson ).
in the twentieth century. Later, in the s, a new type
of Bayesianism appeared, the “subjective Bayesianism” of Acknowledgments
Ramsey and De Finetti (Ramsey ; de Finetti ; This research was supported by the project “El diseño del
Savage ). espacio en entornos de cognición distribuida: plantillas y
At the end of the nineteenth century, a lot of things affordances,” MCI [FFI-/FISO].
were changing in the scientific and philosophical arena.
The end of the idea of “causality” and the conflicts about About the Author
observation lay at the heart of the debate. Gödel attacked Jordi Vallverdú is a lecturer professor of philosophy
Hilbert’s axiomatic approach to mathematics and Bertrand in science and computing at Universitat Autonoma de
Russell, as clever as ever, told us: “The law of causality (. . .) Barcelona. He holds a Ph.D. in philosophy of science
is a relic of a bygone age, surviving, like the monarchy, only (UAB) and a master’s degree in history of sciences (UAB).
 P Probit Analysis

After a short research stay as a fellowship researcher at Hacking I () The taming of chance. Cambridge University Press,
Glaxo-Wellcome Institute for the History of Medicine, Cambridge
Hald A () A history of probability and statistics and their
London (), and a research assistant of Dr. Jasanoff at
applications before . Wiley, New York
J.F.K. School of Government, Harvard University (), Korb KB, Nicholson AE () Bayesian artificial intelligence. CRC
he worked in computing epistemology issues and bioethic Press, Boca Raton
and synthetic emotions. He is listed as an EU Bioso- Moussy C () Probare, probatio, probabilis’ dans de vocabulaire
ciety Research Expert and is a member of the E-CAP’ de la démonstration. Pallas :–
Neymanv J, Pearson ES () On the use and interpretation certain
Steering (http://www.ia-cap.org/administration.php). He
test criteria for purposes of statistical inferences. Biometrika
leads a research group, SETE (Synthetic Emotions in (A):–, –
Technological Environments), which has published about Ramsey FP () Truth and probability (). In: Braithwaite
computational models of synthetic emotions and their RB (ed) The foundations of mathematics and other logical
implementation into social robotic systems. He is Editor- essays, Ch. VII. Kegan Paul, Trench, Trubner/Harcourt, Brace,
London/New York, pp –
in-Chief of the International Journal of Synthetic Emotions
Savage LJ () The foundations of statistics. Wiley, New York
(IJSE) and has edited (and written as an included author) Stigler SM () The history of statistics: the measurement of
the following books: Handbook of research on synthetic uncertainty before . Harvard University Press, Cambridge
emotions and sociable robotics: new applications in affec- Suppes P () A probabilistic theory of causality. North-Holland,
tive computing and artificial intelligence (with D. Casacu- Helsinki
Swinburne R (ed) () Bayes’s theorem. In: Proceedings of the
berta, Information Science Publishing, ) and Thinking
British academy, vol . Oxford University Press, London
machines and the philosophy of computer science: concepts Vallverdú J () The false Dilemma: Bayesian vs. Frequen-
and principles (Ed., ). tist. E – LOGOS Electron J Philos –. http://e- logos.vse.cz/
index.php?target=indexyear
Cross References Weatherford R () Philosophical foundations of probability the-
ory. Routledge & Kegan Paul, London
Actuarial Methods
Bayes’ Theorem
Bayesian Analysis or Evidence Based Statistics?
Bayesian Versus Frequentist Statistical Reasoning
Bayesian vs. Classical Point Estimation: A Comparative
Overview Probit Analysis
Foundations of Probability
Philosophy of Probability Tiberiu Postelnicu
Probability Theory: An Outline Professor Emeritus
Statistics and Gambling Romanian Academy, Bucharest, Romania
Statistics, History of

References and Further Reading Introduction

Crosby AW () The measure of reality: quantification in Western The idea of probit analysis was originally published in
Europe –. Cambridge University Press, Cambridge Science by Chester Ittner Bliss (–) in . He was
Dandoy JR () Astragali through time. In: Maltby M (ed) Inte-
primarily concerned with finding an effective pesticide to
grating zooarchaeology. Oxbow Books, Oxford, pp –
David FN () Games, gods and gambling, a history of probability
control insects that fed on grape leaves. By plotting the
and statistical ideas. Dover, Mineola response of the insects to various concentrations of pesti-
De Finetti B () La Prevision: Ses Lois Logiques, Ses Sources cides, he could visually see that each pesticide affected the
Subjectives. Annals de l’Institute Henri Poincaré :– insects at different concentrations, but he did not have a
Devlin K () The unfinished game: Pascal, Fermat, and the
statistical method to compare this difference. The most log-
seventeenth-century letter that made the world modern. Basic
Books, New York
ical approach would be to fit a regression of the response
Fisher RA () On the mathematical foundations of theoretical versus the concentration or dose and compare between the
statistics. Philos trans Roy Soc London Ser A :– different pesticides. The relationship of response to dose
Gigerenzer G et al () The Empire of chance. How probability was sigmoid in nature and at that time regression was
changed science and everyday life. Cambridge University Press,
only used on linear data. Therefore, Bliss developed the
Cambridge
Hacking I () The emergence of probability: a philosophical
idea of transforming the sigmoid dose–response curve to a
study of early ideas about probability, induction and statistical straight line. When biological responses are plotted against
inference. Cambridge University Press, Cambridge their causal stimuli (or their logarithms) they often form a

View publication stats