eduKateSG · Mathematical Logic · Worked learning guide
Find the reasoning step you need
Understand why a universal rule can hold when its category is empty, and why that result cannot establish that anyone qualified, participated or succeeded.
1. Begin with what would make a universal false
3. Compare all, none and some in a complete model
5. Add the existence condition that the intended claim needs
7. Distinguish an empty extension from an unknown extension
9. Practise with empty and populated categories
Full chapter index · Practice and checking · Mathematical Logic subject guide
“All submitted reports were checked” does not, by itself, say that any reports were submitted. A rule can have no counterexample because its relevant category contains no objects. Classical logic records this as vacuous truth.
The gold standard is to explain what such a statement establishes and what remains unknown. This guide works through empty categories, restricted quantifiers, existential claims and misleading summaries using original finite examples.
The setting is ordinary classical first-order logic with a nonempty overall domain. A predicate may nevertheless apply to no object. Empty categories inside a domain and an empty domain are different ideas. Use Mathematical Logic for the wider account of interpretations.
Chapter index
Chapters 1–4
Chapters 5–8
Chapters 9–10
The formula ∀x(A(x) → B(x)) says every A-object is a B-object. To make it false, there must be an object for which A is true and B is false. Such an object is a counterexample.
Suppose the domain is three labelled boxes, and none is marked Fragile. The claim “Every Fragile box has a protective insert” has no counterexample among those boxes. In the classical interpretation it is true, regardless of which boxes actually contain inserts.
The formula does not assert that a Fragile box exists. Its job is to constrain any object that belongs to the Fragile category. If no object belongs, the condition is never violated.
A direct truth-table explanation
A material conditional A → B is false only when A is true and B is false. For every object in this example, A is false. Therefore its conditional is true at each object, and the universal sentence is true.
This truth condition does not claim that fragility causes inserts, that the insert policy is useful, or that someone performed a successful protective action. Those are additional claims about explanation, purpose or events.
A learner who finds the result surprising should begin with the counterexample test. Ask them to identify a Fragile box without an insert. If no Fragile boxes exist, that task cannot be completed.
The difficulty often comes from ordinary conversation, where a speaker may imply that the category is populated. Formal truth conditions and conversational expectations can differ. The correct response is to keep both visible rather than changing the formal rule to match an unstated implication.
A first-order interpretation supplies a domain and meanings for its predicates. In the standard convention used here, the domain is nonempty. A unary predicate’s extension is the subset of domain objects to which it applies.
Let D = {a,b,c}. Let A have extension ∅, and let B have extension {a}. The domain contains three objects, the A-category contains none and the B-category contains one.
Then ∀x(A(x) → B(x)) is true. The existential ∃x A(x) is false. The universal ∀x B(x) is false because b and c do not have B. These three evaluations are consistent.
The empty category has not removed all objects from the interpretation. Variables still range over a, b and c, and names can still denote those objects.
Why the distinction matters
Some logical systems permit empty domains or use different conventions for names and existence. Results involving those systems require their own definitions. This guide does not silently move between conventions.
In a restricted universal such as “all registered participants…,” the domain may contain many people while the registered-participant predicate has an empty extension. No unusual logic is needed to represent that situation.
A useful drawing is one rectangle containing the domain objects, with no objects circled as A. The rectangle is not empty merely because the A-circle contains nothing.
When checking an inference, record separately: the domain is nonempty, the A-category is empty, and the B-category has the specified members. This prevents the common mistake of assuming that a nonempty overall domain guarantees members of every named category.
Use the same domain D = {a,b,c}, with A empty and B true only of a. Evaluate four standard categorical statements.
| Statement | Formula | Truth in this model |
|---|---|---|
| All A are B | ∀x(A(x) → B(x)) | True |
| No A are B | ∀x(A(x) → ¬B(x)) | True |
| Some A are B | ∃x(A(x) ∧ B(x)) | False |
| Some A are not B | ∃x(A(x) ∧ ¬B(x)) | False |
“All A are B” and “No A are B” can both be true when A is empty. They do not form a contradictory pair in this formalisation.
The contradictory of “All A are B” is “Some A are not B.” Exactly one of these is true in every classical interpretation. The contradictory of “No A are B” is “Some A are B.”
Do not invent existential import
From ∀x(A(x) → B(x)), one cannot infer ∃x(A(x) ∧ B(x)) without an additional existence premise. The empty-A model makes the premise true and conclusion false.
Adding ∃x A(x) repairs that inference. Choose an A-object from the existence premise, apply the universal rule, and obtain an object with both A and B.
The phrase existential import refers to an existence commitment associated with a statement. In the modern formalisation used here, the restricted universal alone lacks that commitment. Other historical or natural-language treatments may use different assumptions.
A strong explanation avoids saying simply that “all and none mean the same thing.” They do not. They agree in this particular empty-category case and differ in many populated models. Agreement in one interpretation does not establish logical equivalence.
A classroom system checks this rule: every submitted assignment has feedback. At the end of an interval, no assignments have been submitted. The universal rule has no violating assignment.
That is a different result from “feedback was provided on at least one assignment.” The latter requires a submitted assignment with feedback. It is also different from “every enrolled learner received feedback,” whose category is learners rather than assignments.
A summary such as “100% feedback coverage” may be ambiguous or undefined when the denominator is zero. The logical universal can be true while an ordinary numerical percentage has no defined value under the usual ratio calculation.
Keep the two calculations separate
If submitted_count = 0 and feedback_count = 0, the ratio feedback_count/submitted_count is 0/0, which is undefined. Do not derive 100% from that arithmetic. A reporting system might adopt an explicit convention, but the convention is additional to the ratio.
A logical check can instead report “No submitted assignment lacked feedback; no assignments were submitted in this interval.” This states the universal result together with the missing participation fact.
If one assignment is submitted without feedback, the universal becomes false. If one assignment is submitted and has feedback, the universal and the positive existence claim are both true.
The example illustrates a general reporting discipline. An absence of violations and the presence of successful activity answer different questions. Both may matter, but one should not be allowed to stand silently for the other.
For a practical record, also verify that the category is correctly populated. A missing submission caused by a data failure is not evidence that no submission occurred. Vacuous truth applies to an interpretation; the interpretation’s adequacy remains an empirical question.
Suppose the intended statement is “At least one registered participant exists, and every registered participant completed the induction.” Let R(x) mean registered participant and C(x) mean completed induction.
Write ∃x R(x) ∧ ∀x(R(x) → C(x)). The first conjunct supplies existence. The second supplies universal coverage. Together they exclude the empty-participant case.
If the intended claim is “Some registered participant completed induction,” write ∃x(R(x) ∧ C(x)). This is weaker than requiring everyone to complete it, but it still supplies a witness.
If the intended claim concerns every person in a roster, choose a predicate for that roster. Changing R from registered people to all eligible people changes the evaluation. A rule can cover all registrants while leaving eligible nonregistrants untouched.
A three-part specification
Separate population, condition and existence requirement. For example: the population is registered participants in this session; the condition is completed induction before the start; the existence requirement is at least one participant.
The date or session qualification can be included in the predicate meaning or represented with additional relation arguments. Either approach should be made explicit enough that a reader knows which records belong.
Do not add existence merely to force every universal sentence to feel more natural. Some statements are intended as rules that remain applicable even when no current object qualifies. “Every future submission must include a signature” need not assert that a submission will occur.
The repair follows the intended promise. Rules, reports of activity and guarantees of participation perform different jobs. Logical precision helps keep those jobs distinct, rather than imposing one template on all of them.
The negation of ∀x(A(x) → B(x)) is ∃x(A(x) ∧ ¬B(x)). In words: at least one A-object fails B. This is “not all A are B,” rather than “no A are B.”
If A is empty, “all A are B” is true and “not all A are B” is false. The negation correctly requires a counterexample that does not exist.
The statement “no A are B” is ∀x(A(x) → ¬B(x)), equivalent to ¬∃x(A(x) ∧ B(x)). It allows the A-category to be empty. It also allows a populated A-category whose members all fail B.
Test a populated mixed case
Let A = {a,b}, with B true of a but false of b. Then all A are B is false because b is a counterexample. No A are B is false because a is a counterexample. Some A are B and some A are not B are both true.
This model shows why “all” and “none” are not contradictory. They can both be false as well as both true in the empty-A case.
A reliable negation method uses one step at a time: negate the universal to obtain an existential, then negate the conditional, producing A ∧ ¬B. The result identifies exactly what would refute the original.
For ordinary prose, preserve restrictions. “Not every report submitted this week was checked” needs a report submitted this week that was not checked. An unchecked report from another week is not the required counterexample.
The transferable habit is to ask for a witness to the failure. Negative wording by itself does not guarantee that the logical negation has been expressed.
A category known to contain no objects is different from a category whose membership has not been established. A database query returning no rows may reflect an empty category, an incorrect filter, missing permissions, delayed data or an incomplete data source.
Classical model evaluation assumes that predicate extensions have been specified. If the interpretation states A = ∅, the universal evaluation is settled. If the record only says “no A entries were retrieved,” the empirical conclusion A = ∅ needs support.
Similarly, a blank cell is not automatically a false entry. If a system uses three statuses such as true, false and unknown, its reporting semantics must explain how unknowns affect a universal check. This guide’s classical examples use fully specified true-or-false predicates.
An original audit example
An induction roster has ten registered participants, but a faulty date filter returns zero. A report then announces no missing inductions. The logical calculation over the returned set may be internally correct while the selected set is wrong.
Repair the population query before reinterpreting the universal. Confirm the session identifier, time interval and membership rule. Then evaluate the condition for the actual relevant objects.
This distinction protects both logic and data work. A valid inference from an inaccurate premise can still produce an inaccurate account of the world. Calling the result vacuous does not resolve whether the premise about emptiness is true.
A good final statement makes the evidence boundary visible: “The supplied model has no A-objects” is a formal description. “There were no A-objects in the real process” is an empirical claim. They coincide only when the model is adequately grounded.
| Claim made | What the premise supports | Possible repair |
|---|---|---|
| All submissions passed, so someone passed | Universal coverage only | Add and verify at least one submission |
| No failures, so 100% succeeded | No failure witness | State the denominator and counting rule |
| The query is empty, so the category is empty | No retrieved rows | Check coverage, filters and missing data |
| All A are B, so all objects are B | A-restricted coverage | Preserve the category restriction |
| All A are B, so some A are B | No existence commitment | Supply an A-object |
A correction should preserve the valid universal result while removing an unsupported promotion. If the category is genuinely empty, say so. There is no need to label the formal rule false merely because an enthusiastic summary overstated it.
If existence was intended, add a separate existence test. This makes the desired behaviour reviewable: the rule-coverage check can pass while the participation check fails.
For learning, use three models: empty category, populated category with no violations, and populated category with a violation. Ask the student which statements change truth value across them.
A concise teaching script
First ask what counts as an A-object. Next ask what a violation would look like. Then ask whether any A-object exists. Finally separate the answer to the violation question from the answer to the existence question.
This sequence keeps the reasoning concrete. It also helps a parent or teacher respond to a plausible wrong answer. A learner who says “true means someone succeeded” needs an existence distinction; a learner who says “false because there are no examples” needs the universal counterexample test.
The goal is not to win an argument about unusual wording. It is to make the promise precise enough that the conclusion does not outrun it.
Exercise 1: no red tokens
The domain contains four tokens. None is red; two are square. Evaluate “all red tokens are square,” “no red tokens are square” and “some red tokens are square.”
The first two are true. The third is false. Neither universal statement asserts that a red token exists.
Exercise 2: add one red token
Suppose token a becomes red and is not square. “All red tokens are square” becomes false. “No red tokens are square” stays true. “Some red tokens are not square” is now true.
Exercise 3: repair an inference
Premise: every marked box was inspected. Conclusion: some marked box was inspected. Supply a missing premise that makes the inference valid.
Add: at least one marked box exists. A witness from that premise receives inspection through the universal rule.
Exercise 4: distinguish a percentage
No forms were submitted and no submitted forms were rejected. Is the ordinary acceptance ratio automatically 100%? No. Its denominator is zero. The statement that no submitted form was rejected can nevertheless be true in the model.
Transfer check
Choose a claim beginning “all,” “every” or “none.” State its category and identify its contradictory. Construct an empty-category model, then add one qualifying object that violates the condition.
Explain which change makes the universal false and why. Then decide whether the original practical task also needs a positive existence or participation condition.
A complete answer should keep the arithmetic question separate if a percentage appears. Logical universal truth, a count of zero and a ratio with denominator zero are related but different objects.
Frequently asked questions
Does vacuous truth mean a statement is meaningless? No. It has definite truth conditions but is satisfied without relevant category members.
Can all A be B and no A be B both hold? Yes, if A is empty under these classical formulas. If at least one A exists, they cannot both hold.
Does a nonempty domain guarantee a nonempty category? No. A predicate may apply to no domain object.
Should a rule always assert existence? No. Rules and reports of successful activity can have different purposes. Add existence when the intended claim needs it.
Connected reading
Return to Mathematical Logic for truth in interpretations, and Logic for counterexamples and supported conclusions. The education transfer guide explains why a definition should travel with its conditions and checks.
Sources and scope
Restricted quantifiers and existential commitments were checked against forall x: Calgary, Sentences with One Quantifier. Domain and predicate interpretation follow the Open Logic Project syntax and semantics reference. Accessed 22 October 2026. Token, submission and roster examples are original teaching models. The reporting discussion separates logical statements from numerical ratios; it does not claim that every software system uses the same convention.
Contents · Previous chapter · Continue to Mathematical Logic
Connected Mathematical Logic guides
Quantifier order and dependent choices