← all systems · see it in The System
Probability: weighing chances, in brains and in people's mathematics, and the structure it works on
Probability concerns events and numerical descriptions of how likely they are to occur. ... Kolmogorov combined the notion of sample space, introduced by Richard von Mises, and measure theory and presented his axiom system for probability theory in 1933. ... There is strong behavioral and physiological evidence that the brain both represents probability distributions and performs probabilistic inference.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | not dated for the reading in brains; the calculus in 1654 (Pascal-Fermat letters) |
| born precision | undated for the sense; year for the calculus |
| boundary | Inside: the reading of chance in brains (relative-frequency inference, probability-coded neural activity, reward-probability signals, probability weighting), the practice of probability (odds, expectation, the calculus of chances from Cardano, Pascal and Fermat to Kolmogorov's axioms; its interpretations) and the structure it names (a probability measure on a sample space). Outside: the uncertain events themselves (dice, draws, rewards, quantum measurements), the people and animals who weigh them, and the statistics, games, insurance, markets and sciences that use probability. |
| nested level | Read in vertebrate brains (rung 8 up), practised and written by people at rung 11, naming a structure booked at rung 0 (roles); used in physics' own description of atomic-scale events (quantum mechanics, statistical mechanics). |
| scale | organism |
Measures (14) published science
| characteristic clock: riffle shuffles of a 52-card deck until the order is near random (total variation distance by shuffle 1-12: 1.00, 1.00, 1.00, 1.00, .924, .614, .334, .167, .085, .043, .021, .010) | distance stays 1.00 through 4 shuffles, then .924 (5), .614 (6), .334 (7), .167 (8): a sharp cutoff |
| shuffles needed as the deck grows (total variation) | 3/2 log2(n) + c shuffles |
| age at which infants infer population from sample | 8 months |
| kea tested | 6 birds |
| kea training to criterion (17/20 correct) | 120 (s = 61.97) trials, mean |
| dopamine neurons averaged in the reward-probability sequence | 26 neurons |
| house advantage in premodern Chinese games of chance | 5-30 % |
| gambling house commission on winnings | 5-10 % |
| share of wagers returned as prizes (lotteries) | 65-80 % |
| historical articles alluding to the house's probabilistic advantage | 7 of 94 (less than 8%) articles |
| base rate of red on an American roulette wheel | about .47 probability |
| streak lengths tested | 2-7 identical outcomes |
| moral certainty (classical reading) | 9999/10,000 probability |
| dates of the calculus: Pacioli's rule, Pascal-Fermat, Huygens, Markov chains, Kolmogorov's axioms | 1494, 1654, 1657, 1906, 1933 year |
Open contentions (4) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: see C-413 - C-940 Where does the gambler's fallacy come from?
booked: The map follows the 2023 account that it comes from experience with certain sequences and from how the 'law of small numbers' is taught, and keeps the older explanation alongside.
the other reading: the fallacy as a misreading of the law of large numbers ('There is no principle that a small number of observations will coincide with the expected value', encyclopedia); other accounts named in Rao & Hastie's review, not read individually - C-964 Does the brain work with probabilities?
booked: The map keeps both: the brain representing probabilities and making inferences with them, and the view that it relies on a bag of shortcuts instead.
the other reading: a set of heuristics or 'bag of tricks' instead of a probabilistic approach, raised by Pouget et al. as an objection; behaviour is often suboptimal, 'one of the most common criticisms of the probabilistic approach' (Pouget et al. 2013); probability weighting: neural responses nonlinear in probability (Hsu et al. 2009) - C-966 What does probability really mean?
booked: The map doesn't pick an interpretation and lists the main ones side by side: long-run frequency, physical tendency, and degree of belief.
the other reading: frequentist: relative frequency in the long run (Venn, Reichenbach, von Mises); propensity: a tendency even in a single trial (Popper and others); Bayesian/subjective: degree of belief (de Finetti, Savage); priors can differ; logical (Keynes, Carnap) and classical (Laplace); Kolmogorov's own: the formalism tied to experience by Cournot's principle (Shafer & Vovk 2006)
Cross-scale patterns seen here (6) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Probability: weighing chances, in brains and in people's mathematics, and the structure it works on. Each links to every other system that shows it.
- same role, different builds (structural)
Statistical inference built twice: in great apes and in a parrot whose brain has 'a highly different structure from primates'. - The Payment to Entropy (structural)
Form: signal (information). Along a Markov chain X → Y → Z no processing of Y, 'deterministic or random', can increase what it carries about X: I(X;Y) ≥ I(X;Z). Each hand-off can only keep or lose information about the source, so the loss compounds along a chain. No per-hand-off efficiency is given - the whole sets the local reference (structural)
The whole sample space is assigned one, and every event's probability is read against that whole; a Bayesian's posterior is read against the prior it started from, so agents with 'sufficiently different priors can lead to different conclusions, regardless of how much information the agents share' (l - templating (partial)
Odds as a template copied by selective imitation: house advantages converged on a band (mostly 5-30%) across games in premodern China, 'design without a designer'. - The carry (partial)
Face: the wait. A deck is not random until enough riffle shuffles have passed: the distance to random stays at 1.00 through 4 shuffles, then drops sharply (.924, .614, .334, .167 at 5-8) - the cutoff. Settling time grows with the deck as 3/2 log2(n). Booked as the wait only: no place overflows here. - Umwelt: each system carves its own slice of one shared medium (partial)
The same jar of mixed tokens is read differently: kea, great apes and infants take relative frequency, capuchins absolute number; people also read stated odds no animal can. Each reader's slice of one shared population.
Valence: what it trades, and with whom (16) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Quantum mechanics via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Entropy via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on even
- Statistical mechanics via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Human (Homo sapiens) via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Human brain via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Fraction (parts of a whole): the ratio sense, the practice of fractions, and the rational numbers they name via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Market and trade via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on mixed
- Kea (Nestor notabilis) via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Game of chance via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Insurance via sys:mental-representation · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on even
- Experiment, measurement and uncertainty (the physics lab as a practice: designing, measuring, estimating error) via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on even
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on) via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Risk management via sys:mental-representation · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on even
- Combinatorics: the practice of counting, arranging and choosing finite structures (permutations, combinations, partitions, designs, graphs), the structures it studies, and the people and machines who work with them via sys:sensory-environment · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
- Graph theory: the practice of studying vertices joined by edges (problems, proofs, algorithms, its people and books) and the graphs it works on via sys:written-record · Probability: weighing chances, in brains and in people's mathematics, and the structure it works on gain
Partners and parts published science
Built from / builds (8)
- Entropy
- Striatum
- Sample space
- Event
- Probability measure
- Midbrain dopamine neurons
- Probabilistic population code
- The base rate against the streak for the next prediction
Sits inside / holds (3)
- Human (Homo sapiens)
- Science (as a human social and cultural system)
- Mathematics (the field: the practice and the structures it studies)
expressed by (11)
- Generation/Synthesis
- Templating / Standardization
- Encapsulation & Modularity (Abstraction)
- Sensing/Scanning
- Mapping/Modeling
- Uncertainty Reduction/Learning
- Risk-managed Exploration
- Multi-Scale Data Fusion / Triangulation
- Surprise / Change-Point Detection
- Social Learning / Public Information Use
- Quality Control & Error Correction
floor filled by (22)
- Radioactive decay
- Quantum mechanics
- Entropy
- Statistical mechanics
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on
- Human (Homo sapiens)
- Writing (the practice of marking language on a surface, and the script plus rules it works with)
- Human brain
- Human society (from local community up to societies)
- The world through the senses (sensory environment)
- Fraction (parts of a whole): the ratio sense, the practice of fractions, and the rational numbers they name
- Market and trade
- Kea (Nestor notabilis)
- Striatum
- Midbrain dopamine neurons
- Game of chance
- Probability theory
- Insurance
- Physics (the field)
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on)
- Risk management
- Numeracy
Kinds (7)
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on
- Mathematics (the field: the practice and the structures it studies)
- Frequentist probability
- Propensity probability
- Bayesian (evidential) probability
- Quantum probability (Born rule)
- Stochastic process
Notes published science
- Gambling houses in old China set their odds so the house always won by a moderate margin, mostly 5-30%, yet almost nobody at the time understood why: only 7 of 94 historical articles even hinted at the house's mathematical edge. The odds seem to have been tuned the way living things are, by copying what worked (src:hong-2024).
- A New Zealand parrot, the kea, does what only great apes and human babies had been shown to do: it reads the proportions in a jar of tokens, then adjusts for a barrier and for a person who picks favourites. Statistical thinking may have evolved twice on Earth, in brains built very differently (src:bastos-taylor-2020).
- A deck of 52 cards stays fully ordered by the mathematics' measure through four riffle shuffles, then suddenly falls apart toward random between five and eight: randomness arrives at a cutoff, not gradually (src:assaf-2011, src:diaconis-1996).
- An electric eel that leaps out of the water and touches a person builds a full circuit through the eel, the arm and the water. Measured element by element, a small eel pushed 40-50 mA through a man's arm, far above what fires pain nerves (src:catania-2017-eel-leap).
- Fourier got his integral from a probability problem: between 1807 and 1811 Laplace's solution for sums of random variables showed him he could solve heat flow with integrals as well as series, and his own 1807 memoir then vanished after his death and was found more than fifty years later (src:narasimhan-2009).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
probability: uncertain events weighed by chance (reader's view)
- L1: Uncertain events in the world: a coin or die thrown, a sample drawn from a jar, a reward that may or may not come, a measurement outcome
- L2: The senses take samples and cues in: draws seen, a bar seen and felt, rewards tasted
- L3: Relative frequency, not absolute number: the proportion of kinds in a population set against a sample
- L4: Probability coded in neural activity: distributions in population codes, dopamine neurons following reward probability, the striatum weighting probabilities (the port)
- L5: Inference and updating: from sample to population and back, cues combined, physical and social information integrated, base rates updated by streaks
- L6: The calculus of chances and its structure: odds (Cardano), expected value (Pascal and Fermat 1654), a probability measure on a sample space (Kolmogorov 1933); its interpretations contested
- L7: Probability put to use on the world: bets and choices, odds set by gambling houses, insurance and market pricing, reliability design, quantum predictions
the boardroom (shared floors)
- L4: Probability in neural activity: population codes for distributions, dopamine prediction errors scaled by reward probability, striatal probability weighting
- L5: Inference: samples to populations and back, cues combined across senses and time, physical and social information integrated, base rates updated
- L6: The calculus of chances: odds, expectation, the axioms; a probability measure on a sample space, interpreted as frequency, propensity or degree of belief (contested)
- L7: Probability put to use: choices and bets, odds and prices set, insurance, reliability, physics' predictions; tied to the world by treating a very probable event as practically certain (Cournot's principle)
way in: Emotion: risk felt, rare and near-sure outcomes weighted out of proportion (inner channel)
- L1: Risk felt as bad: the possibility of not getting the reward expected
- L4: Striatal activity during valuation of gambles, nonlinear in probability (the port)
way in: Hearing and reading: stated odds, prize multiples and base rates
- L1: Stated odds: 1 of 36 figures, 30 times the stake
- L5: Beliefs about base rates set by what people are told
way in: Outcomes: rewards that come or don't (juice, money), at set probabilities
- L1: Risky rewards: binary gambles, a large or a small reward at P = 0.5
- L4: Midbrain dopamine neurons: responses follow reward probability as prediction errors (the port)
way in: Sight: samples drawn from mixed populations (jars of tokens, balls in a box)
- L1: A population of mixed objects in a transparent jar; a sampler drawing from it
- L3: Relative frequency read, not absolute number (capuchins use absolute number instead)
way in: Sight: sequences of binary events (streaks) from a device, an actor or a market
- L1: Sequences of eight binary events from a random device, a goal-directed actor or a financial market
- L3: Streaks of two to seven identical outcomes at the end of a sequence
way in: Thought: intuitions about chances (inner channel)
- L1: An intuition arrives: a total of 6 with two dice is likelier than 2; 1 in 30 beats 1 in 36
- L5: Intuitions about uncertainty are early: 12-month-olds compare the magnitude of uncertainty (as reported)
way in: Touch with sight: a bar's width felt and seen (cue combination)
- L1: A bar whose width is both felt and seen
- L4: Probabilistic population codes: products of distributions done by multiplying neural activity (the port)
way out: Choosing and betting (a hand picked, a gamble taken, a stake placed)
- L1: The hand chosen; the token or reward taken
- L5: The choice follows the inferred sample (the hand that sampled the favourable jar)
- L7: A choice under risk: a binary gamble taken
way out: Setting odds and prices (games, lotteries, insurance premiums)
- L1: Money moved between players and the house
- L5: Odds and fees chosen: 5-10% commission on winnings; 65-80% of wagers returned as prizes
- L7: Gambling houses under pressure to stay profitable set odds
way out: Writing the calculus (letters, books, axioms)
- L1: Marks kept outside the body: the doctrine of chances in letters and books
- L6: The axioms written out (Grundbegriffe, 1933)
- L7: Deciding to settle a question in writing (the problem of points put to Pascal)