← all systems · see it in The System
Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it
In mathematics, a function from a set X to a set Y assigns to each element of X exactly one element of Y.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | 1673 (the term); precursors perhaps medieval (Oresme) |
| born precision | the term to the year; the concept contested (C-1014) |
| boundary | Inside: the practice of letting a quantity depend on another and writing that dependence down (formula, graph, table, rule, tone), evaluating it, composing and inverting it, learning it from sample points, and using it to model the world; and the structure that practice works on (a domain, a codomain and a set of pairs with one output per input, with its kinds and operations). Outside: the quantities in the world a function describes (a planet's position), the equations that ask for a function's zeros (algebra), the numbers it takes and gives (number), the proportions it can keep (ratio), the figures its graph draws (geometry). |
| nested level | A part inside mathematics (a concept that 'in most fields of mathematics' is 'the central objects of investigation'); runs on number (the sets most functions are defined on), on writing (notation and plots on a surface) and on people (taught, learned and extended by them); its structure sits under calculus, analysis, computing and the sciences' models. In set theory a function is a set of ordered pairs; set theory in turn measures size by bijections (src:zach-slc-2025). |
| rewalk on opus | True |
| rewalk why | K-12 wave 11 was walked on Sonnet during the 11-09 model trial; the blind review found Opus walks deeper (research/backchain/MODEL-TRIAL-W11.md). Re-walk on Opus under the current walkSpec. |
| scale | social |
Measures (23) published science
| characteristic clock: from the word 'function' (Leibniz, 1673) to the first formal definition (end of the 19th century) | about two centuries years (derived from two dates the sources give) |
| date: the term 'function' introduced (Leibniz's letter) | 1673 year |
| date: functional notation first used (Euler) | 1734 year |
| waiting time: a question on Takagi's function (infinite derivative) left unanswered | 77 years |
| date: Hill equation formulated (hemoglobin O2 binding) | 1910 year |
| gap: empirical fit (Hill) to biophysical justification (universal Hopfield barrier) | over a century years |
| minimal width for ReLU universal approximation | d_in + 1 (d_in = number of inputs); width d_in is not enough at any depth neurons per hidden layer |
| carry settling: serial (ripple) against Kogge-Stone prefix, for n places | Kogge-Stone depth ceil(log2 n) in prefix gates (2 ceil(log2 n) as a logic circuit); serial graph on 4 places: depth 3 in prefix gates, 6 as the ripple-carry logic circuit gate levels |
| characteristic clock: a carry out of one lookahead group | at most 5 gate delays |
| payment: minimum energy to erase one bit (Landauer limit) | E >= k_B T ln 2 (k_B Boltzmann's constant, T absolute temperature); erasing in finite time tau costs extra work falling in proportion to 1/tau energy per bit |
| sample: GeoGebra quasi-experiment, quadratic functions (Grade 9, Ethiopia) | 42 experimental, 45 control; p < 0.05 for conceptual understanding and problem solving students |
| sample: math anxiety and resting-state EEG integration | 60 tenth graders; r = -0.28 to -0.34 students; correlation |
| odds: next number after 9, 15, 21: 27 against 149.5 (marginal likelihood) | 27 is millions of times more probable ratio |
| setting: in-context learning of linear functions | 20-dimensional inputs; error comparable to the optimal least squares estimator dimensions |
| date: function adopted as a term in the Leibniz-Jean Bernoulli correspondence | 1694-1698 years |
| date: Dirichlet separates function from its analytical representation (an arbitrary correspondence) | 1837 year |
| date: Cantor proves no set maps onto its power set | 1892 year |
| date: the convention of graphing data (Playfair) | 1786 year |
| date: Kolmogorov and Arnold's superposition papers | 1956-1957 years |
| Kolmogorov-Arnold representation: outer functions needed for any continuous function of d variables (q = 0 to 2d) | 2d + 1 outer functions, each of a sum of d one-variable inner functions |
| sample: learning the function concept with static or dynamic visualizations (laboratory experiment) | 157 secondary-school students |
| sample: function estimation from scatterplots (hand-drawn functions) | 170 psychology undergraduates |
| sample: rule versus exemplar learning strategies across a task battery | 96 participants (online, data collected 2017) |
Open contentions (6) 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-917 Does doing algebra use the brain's language areas?
booked: The map follows the studies showing that in trained adults algebra uses visual and spatial areas rather than language areas, and keeps the view that language plays a part alongside.
the other reading: see C-917: mathematical syntax arose historically from language competence and is 'compiled' into visuo-spatial areas - C-1012 What is a mathematical function?
booked: The map uses the set-based definition and keeps the older and alternative definitions alongside.
the other reading: the older 'assignment' reading, which 'cannot be formalized' (Leibniz, Newton, Euler); a function as a rule applied to an argument (the lambda calculus: abstraction and application); a function as a primitive notion, not defined from sets (type theory); Ponte 1992: 'There is no single way of looking at functions': dependency among numerical variables (analysis), relationship (algebra), the algorithm (logic and computer science); in the lambda calculus 'not ... a relation but ... a computational rule'; a close relative is primitive in category theory - C-1013 How do people learn mathematical functions?
booked: The map keeps both accounts, learning rules and learning by similarity, along with a proposed way to unify them.
the other reading: rule-based accounts alone; similarity-based accounts alone; rule-based and exemplar-based learning as stable individual traits, independent of working memory (Herzog and Goldwater 2026); some people draw functions that track local changes, others the global trend (Little et al. 2025) - C-1014 When was the idea of a function invented?
booked: The map dates the term and concept to the 17th century, and keeps the view that it has ancient and medieval precursors alongside.
the other reading: an implicit concept 'with an ancient lineage' (Medvedev, as the history page reports it); precursors in medieval work such as Oresme's (Ponte); particular instances in ancient epochs: counting, the four arithmetic operations, Babylonian tables of reciprocals, squares and roots; Oresme's latitudes of forms (Ponte 1992); Ponte himself dates the concept as 'a clearly individualized concept' to the end of the 17th century - C-1477 Is our inborn sense of number logarithmic?
booked: The map keeps both readings of the evidence: an inborn curved number line that school straightens, or a straight line from the start.
the other reading: Cantlon et al. 2009: the data do not rule out linear encoding with scalar variability, the space mapping built by analogy (the authors answered in a 2009 response, not read)
Cross-scale patterns seen here (18) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it. Each links to every other system that shows it.
- borrowed-then-held (structural)
Borrowed phase: the function concept arose as the tool for the quantitative study of nature, and what counted as a function was set by physical problems and their regularity (falling bodies, planets, the vibrating string, heat flow). Held phase: the concept 'began evolving on its own, moving away fr - causal ceiling (structural)
Some functions cannot be computed by any Turing machine, and by the Church-Turing thesis 'the so-called halting problem not only cannot be solved by Turing machines, it cannot be effectively solved at all': a limit set by the structure of computation, not by any machine's speed or memory. - nested ladder (structural)
Functions form a ladder of added constraints: injective and surjective together give bijective, which is exactly the functions with an inverse; the functions most considered in applications have regularity, 'continuous, differentiable, and even analytic'. Below the function sits a wider rung: every - same role, different builds (structural)
Stating a function for a reader is done by a formula, a graph, a table, a rule in words or a tone: different builds for one job. - templating (structural)
The concept 'one output for each input' and its notation f(x) are reused unchanged in calculus, programming, the lambda calculus, neural networks and a cell's response curve. - The Payment to Entropy (structural)
A function that is not injective merges inputs that cannot be told apart afterwards (the square function sends -3 and 3 to one output), and Landauer's principle ties that logical merging to a physical toll: any logically irreversible step must be accompanied by an entropy increase in the surrounding - the whole sets the local reference (structural)
A multi-valued function is single-valued on each local branch, yet its value 'may "jump" if one follows a closed loop around a singularity' (the monodromy): each local reading is a function, but the readings do not close into one global function. - trade one property for another (structural)
Fewer variables bought, smoothness paid: Kolmogorov-Arnold writes any continuous function of d variables with one-variable pieces, but some differentiable functions of d variables are not superpositions of differentiable functions of fewer variables, so the inner pieces cannot all be smooth; Kurkova - Umwelt: each system carves its own slice of one shared medium (structural)
One set of pairs, four slices: the eye takes the graph (a curve, or Fermat's spiral if drawn in polar coordinates), the eye takes the formula (the rule), the eye takes a table (a sample or a finite whole), and the ear takes the values as the pitch of a tone. The structure is the Umgebung, the channe - namespace completeness (the map predicts its gaps) (partial)
Cantor's construction is a map that names its own gap: for any function g from a set to its subsets it builds the subset {x : x not in g(x)} and shows where it must be missing from g's range. Every list of subsets predicts the member it leaves out. Also the shelf of kinds (polynomial, rational, elem - outside-to-inside (partial)
Outside phase first: the linear mapping of number onto space is 'a cultural invention' (rulers, number lines, schooling); inside phase: educated adults hold it as their own linear mapping for small or symbolic numbers, while the Mundurucu, without formal education, map logarithmically at all ages. C - same shape, opposite sign (partial)
The pairs of a bijection read from output to input are its inverse: the same set, read the other way; a restriction to an injective piece gives the inverse trigonometric functions. - seed (partial)
One definition (one output for each input), once put in terms of set theory, 'greatly increased the possible applications of the concept': kinds, compositions, inverses, function spaces and the many fields that now use functions grew out of it. - symmetry breaking (name the cut) (partial)
The cut that makes a function out of a set: sets have no order ('In an unordered pair {x, y}, the order does not matter'), and the ordered pair (Wiener-Kuratowski) chooses a direction, input first, output second; the direction chosen sets what is built after it (graph, composition, inverse as the or - The carry (partial)
Two faces seen through function. The wait: each carry is a function of the carry below it, so a ripple adder is a chain of compositions and 'each stage must wait until the previous carry bit has been calculated'. The lookahead: composition is associative, and the operator on (generate, propagate) pa - who opens the door (partial)
A function crosses into a person only through a representation (it is 'not directly accessible apart from external representations'). For number on space, the receiver's own physics opens a door (an initial logarithmic intuition, universal), and a port is installed ahead of time by culture: the line - lightning (surface)
The 'monster-functions' (continuous with no derivative anywhere) came as a shock, 'moments of horror' that 'seemed to have no other purpose than to steal from them their confidence in their own reasoning'; after them the concept evolved on its own. A sudden change, but no stored difference released - partner-becomes-part (surface)
Set theory, a partner, became the function's body: 'functions can be identified with certain relations, i.e., with certain sets of tuples', while set theory takes its notion of size from bijections (a gain/gain exchange). No inheritance is involved, so surface.
Valence: what it trades, and with whom (18) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Vector space: sets whose elements can be added and scaled, as mathematics describes it, with bases and dimension, linear maps, inner products, and their use as the state spaces of physics and data via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Linear algebra (the practice of solving, transforming and taking apart linear systems with vectors and matrices, and the structures it works with) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Cell via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Neuron (nerve cell) via sys:aqueous-solution · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Human (Homo sapiens) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Large language model via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Number and counting (the number sense, the practice of counting, and the structure it works on) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Shapes (geometry): the practice and the structure via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Computer via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Algorithm (the practice of step-by-step procedures and the structure they compute) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Ratio and proportion: comparing two quantities by how many times one contains the other, the practice of keeping proportions, and the structure it works on via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Algebra (the practice of solving for unknowns and transforming expressions, and the structures it works with) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Trigonometry (the practice of tying angles to lengths, and the structure it works on) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Differential equations (the practice of writing and solving laws of change, and the equations it works on) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Set theory: the study of sets (collections taken as single objects), membership, Cantor's infinities and the axioms, and the practice of the people and machines who work with them via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Limit (mathematics): the value a sequence or function approaches, and the practice of approaching, bounding and proving it via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on) via sys:sensory-environment · Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it gain
Partners and parts published science
Built from / builds (7)
- Domain (the set of inputs)
- Codomain (the set the outputs are taken from)
- Graph of a function (the set of pairs)
- Composition (applying one function to the output of another)
- Functional notation (f(x), the arrow, the placeholder)
- Ordered pair (Wiener-Kuratowski)
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
Sits inside / holds (2)
- Human society (from local community up to societies)
- Mathematics (the field: the practice and the structures it studies)
expressed by (8)
- Generation/Synthesis
- Recombination/Composition
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Interoperability/Transduction (Bridging)
- Encapsulation & Modularity (Abstraction)
- Mapping/Modeling
- Uncertainty Reduction/Learning
floor filled by (25)
- Cell
- Neuron (nerve cell)
- Human (Homo sapiens)
- Science (as a human social and cultural system)
- Writing (the practice of marking language on a surface, and the script plus rules it works with)
- Human brain
- The world through the senses (sensory environment)
- Number and counting (the number sense, the practice of counting, and the structure it works on)
- Measurement (the practice of comparing a quantity with a unit, and the units and standards it uses)
- Computer
- Algorithm (the practice of step-by-step procedures and the structure they compute)
- Artificial neural networks (trained sequence-to-sequence models)
- Inverse function
- Lambda calculus
- Sonification
- GeoGebra (dynamic graphing software)
- Function learning
- Covariational reasoning
- Mathematical notation
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Graph of a function (the set of pairs)
- Composition (applying one function to the output of another)
- Function space
- Set theory: the study of sets (collections taken as single objects), membership, Cantor's infinities and the axioms, and the practice of the people and machines who work with them
- Differential equations (the practice of writing and solving laws of change, and the equations it works on)
Kinds (15)
- Injective function (one-to-one)
- Surjective function (onto)
- Bijection (one-to-one correspondence)
- Partial function
- Multi-valued function
- Multivariate function (several inputs)
- Real function (real-valued, real variable)
- Nowhere-differentiable ('strange') function
- Polynomial function
- Rational function
- Elementary function
- Vector-valued function (vector field)
- Function space
- Lambda term (a function as a rule)
- Function in programming (subroutine)
Notes published science
- Every continuous function of many variables, however tangled, can be built from functions of a single variable and plain addition, and the inner one-variable pieces are the same for every function; only the outer pieces change. Kolmogorov and Arnold proved it in 1956-57, against what Hilbert's 13th problem expected. Source: src:toscano-2025.
- Your first number line may be logarithmic: Mundurucu adults and children, with little formal schooling, place numbers on a line so the gaps shrink as numbers grow, and Western adults do the same when they can't count. A straight, even number line may be something school installs. Other researchers say the same data fit an even line. Source: src:dehaene-2008, src:cantlon-2009.
- Single brain cells do arithmetic on functions: a neuron's firing rate is a function of its input, and other inputs can slide that function sideways or stretch it, multiplying instead of adding. Source: src:silver-2010.
- Ask people for the next number after 9, 15, 21 and they say 27, not 149.5, though a rule can be found that gives either; the Bayesian arithmetic says 27 is millions of times more probable. Source: src:wilson-2015.
- A function that merges inputs costs heat: any logically irreversible step, such as erasing a bit or merging two computation paths, must be paid for with an entropy increase in the surroundings, and any deterministic computation can be redone reversibly if it keeps a copy of its input. Source: src:bennett-2003.
- Archibald Hill fitted a curve to haemoglobin data in 1910 with no justification beyond the fit; in 2024 the same Hill function turned out to be the universal limit on how sharp any equilibrium molecular switch can be, so a cell that does better must be spending energy. Source: src:martinez-corral-2024.
- A question about Takagi's nowhere-differentiable function, asked in a 1933 editorial note as 'interesting and probably not too difficult', took 77 years to answer correctly. Source: src:allaart-kawamura-2011.
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.
a function: each input assigned exactly one output, shown as a graph, formula, table, rule or tone (reader's view)
- L1: The dependence in the world: a quantity that varies with another (a planet's position with time; a year's motor vehicle deaths with the year).
- L2: Seeing or hearing a representation: a graph, a formula, a table, or a tone whose pitch follows the value. A plotted function is taken in by its local features (single points, small groups) and its global trend, and readers differ in which they follow.
- L3: Codes: formula, graph, table, rule in words and tone are different codes for the same set of pairs; the expert brain's network for mathematical statements spares the language areas. A function is reached only through its representations (tabular, graphical, algebraic, situational); translating between them is part of understanding it.
- L4: Transforming: evaluate at an input, compose, invert, restrict; formula and graph moved together in dynamic software. In the brain's own wiring, single neurons transform input into output by functions that modulatory inputs shift or scale (gain is the slope of a neuron's input-output relationship).
- L5: Learning and control: extrapolating from sample points, preferring the simplest rule that fits, reasoning about how two quantities change together. Two ways of thinking a function (covariation and correspondence); two stable learning strategies (rule learners extrapolate, exemplar learners stay within the range seen).
- L6: The structure: domain, codomain and a set of pairs with one output per input; kinds (injective, surjective, bijective, partial, multi-valued); composition and inverse; function spaces. Built from sets: an ordered pair (Wiener-Kuratowski), a function identified with its graph, partial functions with at most one output; bijections measure size (Cantor: no function from a set onto its power set); the computable functions are the same under every proposed model (Church-Turing), and some are not computable (halting); every continuous function of many variables is built from one-variable functions and addition (Kolmogorov-Arnold).
- L7: Use: models of dependence in science and engineering; calculus and analysis; programs; approximators; a cell's response curve. Modelling runs in cycles: from the situation to a mathematical description and back, checked against observed data and changed.
the boardroom (shared floors)
- L3: Codes: a function held as a graph, a formula, a table, a rule in words, a tone or a set of recorded pairs; in expert mathematicians the judgement of mathematical statements, algebra, analysis, topology or geometry alike, runs on a bilateral frontal, intraparietal and ventrolateral temporal network that spares the language areas. In mathematics education a function is held to be reachable only through its external representations; four typical forms (tabular, graphical, algebraic, situational), and translating between them is one aspect of understanding.
- L4: Transforming: a function evaluated at an input, composed with another, undone by its inverse where it has one, restricted to a part where it has one, and moved between its representations (graph and formula changing together). The brain's own neurons do this kind of work on signals: a rate-coded neuron's input-output relationship is shifted or scaled by modulatory inputs (gain is its slope), performing arithmetic on the function itself.
- L5: Learning and control: people extrapolate a function from a few sample points and prefer the simplest rule that fits; covariational reasoning (how a change in one quantity goes with a change in another) is effortful even for well-prepared students; anxiety goes with a less integrated network. Covariation (how one quantity changes as another does) and correspondence (pairwise assignment) are two ways of thinking a function. Learners differ stably: rule learners extrapolate, exemplar learners stay within the range they saw; in drawing a function behind noisy data some track local changes, others the global trend.
- L6: The structure: a domain, a codomain and a graph of pairs with one output per input; injective, surjective and bijective functions; composition (associative) and inverses; restriction and extension; partial and multi-valued functions; function spaces; and what a function is at bottom, which the sources do not settle. In set theory a function is identified with its graph, a set of Wiener-Kuratowski ordered pairs; bijections define when two sets are the same size, and no function maps a set onto its power set (Cantor 1892); the effectively computable functions are the same under every proposed model (Church-Turing), and the halting function is not among them; Kolmogorov and Arnold showed every continuous function of several variables is a superposition of one-variable functions and addition.
- L7: Functions put to use: dependence of one quantity on another modelled in science and engineering; functions as the objects of calculus and analysis; programs; neural networks as approximators of functions; a cell's input-output response. Function arose as the tool for the quantitative study of natural phenomena (Galileo, Kepler) and is used through modelling cycles that go from situation to description and back.
way in: Emotion: tension in front of mathematical problems
- L1: Math anxiety: 'feelings of tension, fear, and apprehension when individuals engage with mathematical problems or numerical information'; mathematics in general, not shown for functions in particular.
- L4: Network integration: mean eigenvector centrality of resting-state EEG was negatively related to math anxiety in three bands, r = -0.28 to -0.34 (60 tenth graders).
way in: Hands: dragging and typing in dynamic graphing software
- L1: Entering an equation or dragging a parameter in dynamic software; the graph and the algebraic form update together.
way in: Hearing: a function heard (pitch following the value; sonification)
- L1: A sonified function: an audio track that plays a continuous tone, its pitch following the dependent value as the input steps through its domain.
way in: Machine input: a program calls the function (argument in, value out)
- L1: A call: an argument handed to a callable unit with a well-formed interface; it returns a value.
- L2: Source code compiled to machine code that implements similar semantics: two different kinds of callable unit for the same function.
way in: Machine input: input-output pairs in a transformer's context
- L1: A prompt of in-context examples (input-output pairs from one function) and a new query input; the model gives the output with no change to its parameters.
way in: Measured pairs: input-output values recorded from the world
- L1: Paired measurements of a quantity against another (oxygen bound against oxygen pressure; a planet's position against time), the raw material a function is fitted to.
way in: Sight: graphs and plots of functions (curves, bars, other coordinate systems)
- L1: The picture of a function: points (x, f(x)) drawn in a coordinate system as a curve, or as bars for a finite domain; the same pairs drawn in polar coordinates give a different curve.
- L2: Graphical perception: the reader of a plot attends to local features (single points, pairs, small groups) or to global features (overall trends); some people draw the function behind noisy data by tracking local changes, others by following the global trend, and more points give smoother drawn functions.
way in: Sight: tables of values
- L1: A table of input-output values: complete for a finite domain (a multiplication table), a sample for a continuous one, with interpolation between rows.
way in: Sight: written formulas and notation (f(x), the arrow, indices)
- L1: A function written as a formula or in notation: f(x) = ..., the arrow x -> ..., an index f_x, a placeholder; the letter f, g or h names it.
way in: Thought: a function held and extended in the mind (a rule imagined, the next value guessed)
- L1: A function held as a rule in mind and run past the data: asked for the next number in 9, 15, 21, people are likely to guess 27 rather than 149.5. Where such a rule comes from is not ruled.
- L2: The number-to-space mapping a thought of quantity builds on: a universal intuition that starts logarithmic, with the linear number line arriving through formal education (Dehaene et al. 2008); contested: the same data fit linear coding with scalar variability (Cantlon et al. 2009).
way in: Words: a function stated in a sentence or a word problem (read or heard)
- L1: A rule in words: 'assigns to each element of X exactly one element of Y'; a dependence in a story (the position of a planet is a function of time).
way out: Applying: predicting and extrapolating with the function, and building it into models and programs
- L1: The world a model is used on: a planet's position at a later time.
- L7: A dependence modelled and used.
way out: Entering and dragging: an equation typed or a parameter moved, the graph redrawn at once
- L1: The screen redrawn: graph and algebraic form together.
- L4: The change of the function, as an equation edited or a parameter moved.
way out: Speaking: reading a function aloud ('maps to'; 'f of g')
- L1: The sentence spoken aloud.
- L6: The rule named in words.
way out: Writing: the rule as a formula, the pairs as a table, the curve as a plot
- L1: The marks on the surface, read back by the eye.
- L4: The plot: pairs placed as points in a coordinate system.
- L6: The rule as a formula that computes the value from any input.
Sources (37) published science
Wilson AG, Dann C, Lucas CG, Xing EP (2015) The human kernel. NIPS 2015; arXiv:1510.07389 (behind its links)
History of the function concept (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Hanin B, Sellke M (2017) Approximating continuous functions by ReLU nets of minimal width. arXiv:1710.11278 (behind its links)
Ponte JP (1992) The history of the concept of function and some educational implications. The Mathematics Educator 3(2) (behind its links)
Allaart PC, Kawamura K (2011) The Takagi function: a survey. arXiv:1110.1691 (behind its links)
Hill equation (biochemistry) (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Held S, Spirkl ST (2014) Fast prefix adders for non-uniform input arrival times. arXiv:1411.2917 (behind its links)
Garg S, Tsipras D, Liang P, Valiant G (2022) What can transformers learn in-context? A case study of simple function classes. NeurIPS 2022; arXiv:2208.01066 (v3) (behind its links)
Rolfes T, Roth J, Schnotz W (2020) Learning the concept of function with dynamic visualizations. Front Psychol 11:693. doi:10.3389/fpsyg.2020.00693 (PMC7212367, CC BY) (behind its links)
Calculus (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Function (computer programming) (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Silver RA (2010) Neuronal arithmetic. Nat Rev Neurosci 11:474-489. doi:10.1038/nrn2864 (PMC4750293, author manuscript, available for text mining) (behind its links)
Zimmerman C, Olsho A, Brahmia SW, Boudreaux A, Smith T, Eaton P (2020) Exploring student facility with 'goes like' reasoning in introductory physics. arXiv:2007.10409 (behind its links)
Vector space (Wikipedia, via Wikimedia Enterprise) (behind its links)
Linear algebra (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Trigonometry (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Narasimhan 2009. Laplace, Fourier, and stochastic diffusion. arXiv:0912.2798 (behind its links)
Amalric M, Dehaene S (2016) Origins of the brain networks for advanced mathematics in expert mathematicians. PNAS 113:4909-4917. doi:10.1073/pnas.1603205113 (behind its links)
Injective function (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Inverse function (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Function space (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)
Lambda calculus (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01) (behind its links)