Frege's Begriffsschrift: A Formal Language for Inference
Gottlob Frege's 1879 Begriffsschrift introduces a formal language that can express claims involving several objects and make the structure of an inference explicit.
PhilosophyAll chapters Reading tools Notes & comments
Making inference explicit › What “Every Student” Leaves OpenFull chapter · Sections in this chapter1 / 8 ↓
What “Every Student” Leaves Open
Imagine trying to express a simple thought. Every student admires some teacher. Anna may admire Dr. Chen, Ben may admire Dr. Okafor, and the statement can be true even if no teacher is admired by both. Compare it with the claim that there is a teacher whom every student admires. This also talks about all the students and the existence of a teacher. It would be false of Anna and Ben if the only teachers they admire are different people. Making that difference explicit is a useful way into Gottlob Frege's Begriffsschrift.
Traditional syllogistic could handle inferences that turn on the overlap between categories. If all oak trees are plants and all plants need water, then all oak trees need water. Its familiar forms can also accommodate “all students admire a teacher” if “admires a teacher” is taken as a single predicate. The trouble begins when we want the inference to depend on relationships within that predicate, such as whether different students admire the same teacher. The structure we need to see is inside a unit that the syllogism treats as given.
In the preface of his 1879 book, Frege explains his aim through the question of how much arithmetic could be derived by reasoning alone. He wanted a chain of inference with no gap through which an unacknowledged assumption might enter. Ordinary language proved an obstacle to that aim, since its sentences do not consistently show the inferential relationships he needed to record. The title describes the proposed remedy as a formula language of pure thought modeled on the formula language of arithmetic.
Frege likens his notation to a microscope alongside the eye. A scientific instrument can give a precise view of a narrow range of things. The eye remains better suited to a great many ordinary purposes. In the same way, the concept-script is worth judging by what it reveals about chains of reasoning. Asking it to preserve every distinction we make when we speak would give it a different job from the one Frege designed it to do.
A New Language for Thought
In §3, Frege compares two ways of saying that the Greeks defeated the Persians at Plataea. One grammatical construction makes the Greeks the subject. A passive construction makes the Persians the subject. In the respects that matter to the inferences he is recording, the two sentences can be used in the same way. This is part of what Frege calls their conceptual content. Word order and voice can guide a reader's attention while leaving that content intact.
In §§9–10, Frege explains how to analyze a content by holding one part fixed and allowing another to vary. His example is that hydrogen is lighter than carbon dioxide. We can hold fixed the comparison “is lighter than carbon dioxide” and ask which gases satisfy it. We can also hold fixed “is heavier than hydrogen” and vary the gas in that comparison. Either way, the starting claim is the same. The decomposition changes according to which term we treat as replaceable. Frege calls the remaining part a function and what occupies the replaceable place an argument.
Write the comparison in modern shorthand as L(hydrogen, carbon dioxide), where L(x, y) means “x is lighter than y.” Holding the second place fixed lets us compare something else against carbon dioxide. Holding the first place fixed lets us compare hydrogen against something else. The order of the places matters. L(y, x) makes a different claim from L(x, y). Frege explicitly discusses that distinction for two-argument functions in §10.
This way of taking a statement apart connects mathematical operations with logical analysis. Frege can isolate what stays the same when we replace a particular word, and he can repeat that operation for different parts of a statement. His later account of a concept as a function whose values are truth-values gives a sharper ontology to the notion. In 1879 his explanation starts from an expression and the different ways we can consider part of it replaceable. The flexibility of the analysis is already doing substantial work.
In §11, Frege introduces an explicit sign for generality. A letter appears inside a concavity in the horizontal line of the formula. Its reach is governed by the portion of the expression the sign applies to. §12 uses generality together with negation to express existence. We would now write “there exists an x that satisfies P” as ∃x P(x). Frege can express it by denying that every x fails to satisfy P. His primitive sign is for universality, and the two quantifier symbols used today were absent from his pages.
How the Argument Works
The first two sections distinguish the expression of a content from the act of judging it. The horizontal content stroke belongs with what can be judged. Adding the vertical judgment stroke indicates that the content is asserted. If a conditional asserts that one claim follows on a condition, it need not separately assert either component. A proof therefore needs a way to record which expressions figure as asserted lines and which occur inside the content of another line.
In §§5–7, Frege explains the conditional, inference, and negation. To see the inference pattern behind §6, imagine holding that if the tank leaks, pressure will drop. A second assertion says that the tank leaks. Taking these together entitles us to assert that the pressure will drop. The judgment about the pressure would have no such warrant from the conditional by itself. Frege gives the distinction a visible form by writing the two premises as judgments and specifying what can be detached from them. The relation in the conditional is a logical one. An engineering explanation of how such a leak drops pressure would supply a different kind of information.
§8 treats identity of content. Frege asks why it is worth having more than one sign for the same thing. His example uses a fixed point on a circle and another designation for a point obtained by rotating a line through it. At a particular orientation, the moving designation picks out the fixed point itself. Learning they coincide is a geometrical discovery. If we erased the second designation at the start, we would obscure the reasoning that establishes the coincidence. Frege's account in this section belongs to 1879. His later treatment of the significance of identity claims is a further development.
Part Two starts with an explicit defense of the deductive presentation in §13. Frege numbers judgments and gives derivations that can refer back to preceding ones. A reader can trace an asserted conclusion through stated dependencies, checking when a law is invoked or an expression substituted. Proving a particular mathematical result still calls for finding the right intermediate claims. The notation gives the reader a way to inspect the result of that work.
What the Early System Does for Arithmetic
In Part Three, beginning with §23, Frege tests his notation on general facts about series. The idea in §26 can be introduced with a finite toy example. Suppose a trail map allows you to go from A to B and from B to C. The first definition of a direct link need not say there is a route from A to C. To capture which places can follow A in a chain, one can consider properties that hold at every direct successor of A and pass from each place with that property to its direct successors. C shares every such property. An unrelated point D could be disqualified by a property that holds throughout the reachable trail and fails at D. In §26, Frege uses quantification over properties and an inheritance condition to define following in a series.
That is why simply identifying the concept-script with modern first-order predicate logic understates its resources. Its treatment of generality can range over functions or properties as well as individual objects. The theory of series shows how useful that wider reach can be. Frege can express a general form of reasoning about chains without introducing a new primitive for each kind of series or listing all the terms.
Frege develops his program in The Foundations of Arithmetic (1884) and the two volumes of Basic Laws of Arithmetic (1893 and 1903). In the later formal system he introduced a law governing the identity of the objects he associates with functions, called courses of values. It allows functions that agree on every argument to be assigned the same course of values. Russell's 1902 letter exposed a contradiction arising from the machinery of that later account. Frege's reply, translated in the van Heijenoort sourcebook, acknowledges the blow to his treatment of arithmetic. The 1879 text and the later text share an ambition and some of the apparatus. Keeping their axioms distinct is essential to saying what the paradox challenged.
What Predicate Logic Can Do
The distinction between two readings of the student sentence can be tested on a small roster. Say there are two students and two teachers. Each row of a table represents a student and each column represents a teacher. We mark a cell when that student admires that teacher. If there is a mark in each row, every student admires a teacher. If there is a column marked in every row, there is one teacher every student admires. With two marks along the diagonal and none elsewhere, the row test succeeds and the column test fails.
In current notation, restricting the variables to students and teachers respectively, the first claim takes the form ∀x ∃y A(x, y). The second takes the form ∃y ∀x A(x, y). A common teacher supplies a witness for the first claim, since each student admires that very teacher. In the other direction, the table tells us why the inference fails. The symbol order lets us inspect the mistake, and the roster supplies a concrete counterexample. Frege's §11 already allows the scope of one bound letter to include the scope of another.
Or consider the statement that all that glitters is not gold. This is famously ambiguous in English. Does it mean "not everything that glitters is gold" (some glittering things are not gold) or "everything that glitters is not gold" (no glittering thing is gold)? Predicate logic forces us to disambiguate by placing the negation at different points in the formula, before the quantifier or after it. This is not pedantry. It reflects a real distinction in meaning, one that natural language obscures.
Consider a real-valued function f of a real variable and a point a in its domain. Continuity at a means that for every positive tolerance ε, there is a positive δ such that every x in the domain with |x − a| < δ satisfies |f(x) − f(a)| < ε. The quantifiers occur in the order ∀ε ∃δ ∀x. Writing out the order shows where each choice is made.
The example of continuity also shows why bound letters need a range and scope that a reader can trace. Once a tolerance ε has been chosen for the values of the function, δ has to work for all the arguments x permitted by that choice. Letting δ be chosen after x would permit different neighborhoods for individual points and would miss the condition the definition is meant to impose. The quantifiers tell us how to test the definition, and they make clear which variable choices a valid argument can depend on. Its ε and δ notation is a modern illustration of the expressive resources at issue.
How Scope Changes an Inference
Take a slightly longer claim. Every student admires some teacher who has advised a doctoral candidate. In the simple table, a marked cell established just one relationship, the student's admiration for the teacher. With the additional condition, we would need to know that the teacher in that cell really did advise someone. It can be a different candidate for each teacher. The order in which the sentence introduces its people determines which of those choices has to be held fixed when we test it.
The scope of a negation matters in the same way. Denying the longer claim requires finding a student for whom no admired teacher meets the advising condition. That student might still admire a teacher, and there might be many advisers among the teachers admired by other students. It would take a much stronger claim to say that no student admires any adviser. What makes the difference visible is the position of the denial relative to the claims about each student and at least one teacher.
This gives a more specific sense to Frege's comparison with a microscope. We can inspect one part of a complicated claim while keeping track of just the connections needed to test an inference. The formal translation makes choices about the meaning of the original. A correct derivation from that translation still leaves us the question of whether we understood the original sentence correctly.
What the Notation Leaves Out
Frege shows that the same formal resources can strip away distinctions speakers care about. In §7, he observes that “and” and “but” can agree in their contribution to the content he wants to represent even while “but” signals a difference from what the listener would have expected. “The road was closed, but the shop was open” conveys some surprise or contrast. The two clauses can both be true without any such contrast. A record of the inference that the shop was open can set the contrast aside. An account of why someone chose the word “but” would need to recover it.
This choice also helps explain why learning to read a page of concept-script can take longer than learning the symbol for conjunction in a modern textbook. The page puts visible inferential features on separate parts of a diagram. Each branch has a role inside a larger judgment. Once that role is clear, the diagrams can show the difference between an asserted conditional and an assertion of one of its components. It is useful to begin with reasons for the notation and then read an example at a time. Substituting contemporary symbols can ease the work, provided the translation keeps track of assertion, scope, and the order of the argument places.
The project begins with a question about foundations and yields a method one can use before making up one's mind about that project. Take two sentences that sound alike. Ask what must be the case for each to be true, whether one actually follows from the other, and what a counterexample would look like. In the case of the students and teachers, a small table is enough to answer. In the case of a formal proof about arithmetic, the same concern with dependencies has to be sustained across many more lines.
Reading Frege
Read the 1879 German scan with the section numbers in view. The preface explains the intended use of the notation. §§2–3 distinguish judgment and content, §§5–8 introduce the main operations and identity, and §§9–12 set out functions and generality. §§23–26 reward a separate reading for the treatment of series. The folio numbers in the printed book differ from the PDF viewer's page count.
Join the discussionAsk a question or share what you learned.
Keep exploring
Reference
Citation details
Cite or share this article.

Comments
Add to the conversation
A question, a correction, or an example from your own work.
No comments yet. Be the first to share your thoughts!