How to Check Whether a Verbal Conclusion Must Follow

How do I use a counterexample to test a must-follow answer?

Keep the passage’s relevant statements true, then try to describe one coherent case in which the proposed conclusion is false. If the case fits without changing the wording or scope, the conclusion does not have to follow. If no such case comes to mind, the search has not proved the answer. The open textbook chapter “How to Think About Deductive Logic” defines validity by this guarantee: an argument is invalid when its premises can be true while its conclusion is false. University of Massachusetts lecture notes state the limit plainly: a counterexample can show invalidity, but failing to find one cannot establish validity.

For example, suppose a passage says every scheduled delivery has a tracking code, and one parcel has a tracking code. The proposed conclusion is that the parcel is scheduled. Imagine it is unscheduled but still has a code. Both passage statements remain true, while the conclusion is false. That is a counterexample, so the conclusion is not guaranteed. The check asks what the passage forces, not what seems likely. One faithful countercase defeats a must-follow answer. Finding none is only an unfinished search; a separate step would be needed to establish that the conclusion follows.

Sources: How to Think About Deductive Logic; University of Massachusetts Lecture Notes Unit 1

First decide whether the question asks for necessity or likelihood

Before looking for a counterexample, identify the standard the question asks you to apply. A counterexample is decisive when the task asks what must follow, must be true, or is necessarily implied. In that setting, one possible case in which the passage’s claims remain true but the proposed conclusion is false shows that the conclusion is not guaranteed. If the question instead asks what is likely, suggested, or best supported, a possible exception may weaken the conclusion without deciding whether it is still the most reasonable one. The University of Massachusetts lecture notes distinguish deductive arguments, which claim necessity, from inductive arguments, which claim a conclusion is probable given the premises. Entailment means that the stated premises guarantee a conclusion: in every possibility consistent with those premises, the conclusion is true. Likelihood is weaker. Evidence can make a conclusion more plausible than its alternatives while leaving open cases in which it is false. This is why “could this be false?” answers a must-follow question, but does not by itself answer a probability question. A possible exception defeats certainty; to assess likelihood, compare support for competing possibilities, using only evidence the task allows. The University of Minnesota chapter on deductive logic likewise defines validity by whether the premises can be true while the conclusion is false. Consider an invented example. A passage says some applicants who completed a skills course passed a screening, and everyone who passed received an interview. It follows that at least some course completers received interviews. But the passage does not establish that most course completers received interviews. A coherent possibility is that many other course completers did not pass. This leaves the statements given intact while making “most course completers received interviews” false. If asked what must follow, reject that stronger answer; if asked what the passage suggests is likely, the countercase is relevant but the wording and any further evidence determine the judgment. Do not add real-world assumptions about how screening usually works unless the item allows them. Read the exact instruction before choosing a method. Official EPSO testing guidance describes its own reasoning assessments and supplies sample materials; it does not establish a universal rule for every verbal reasoning test. Some formats ask for strict entailment, while others ask candidates to evaluate information or choose a supported conclusion. Look for the operative words, then apply the matching standard: for necessity, search for one passage-faithful exception; for likelihood, weigh the stated evidence without treating any imaginable exception as an automatic refutation. If the wording does not clearly establish the standard, record that uncertainty rather than silently turning a probability judgment into a proof task.

Sources: University of Massachusetts Lecture Notes Unit 1; How to Think About Deductive Logic; EPSO testing | EU Careers

What exactly must stay true in a counterexample?

A premise is a statement the passage supplies; scope is the set of cases it covers. A counterexample must keep those premises true while making the proposed conclusion false. Preserve its subject, quantifier, condition, time, and comparison group. If a scenario works only after one of those changes, it is not a counterexample to the argument as written. The Delftse Foundations of Computation chapter defines invalidity by just this possibility: at least one case in which all premises are true and the conclusion is false. The University of Massachusetts Lecture Notes Unit 1 likewise says validity concerns the connection between premises and conclusion, not their actual truth.

Consider an original illustration. A passage says: “If a staff member completes the safety course, the staff member may enter the archive.” It adds: “Rina completed the course.” The proposed conclusion is: “Rina may enter the archive.” To challenge the conclusion, you cannot keep both passage statements true while denying permission: that would violate the stated conditional. But change the proposed conclusion to “Rina is the only staff member who may enter,” and a countercase is easy: another staff member may also have completed the course. The passage says what follows for course completers; it says nothing about exclusivity. This illustrates why a counterexample must target the exact conclusion, not a stronger claim . The University of Massachusetts notes explain that a conditional’s if-part does not assert that the condition is true by itself; a separate premise is needed. That separate statement about Rina lets the result follow.

Three tempting countercases fail for the same reason: they alter a premise rather than preserve it. First, if the passage says “every registered applicant received a confirmation,” a case where only some registered applicants received one changes “every” into “some.” It cannot test the original claim. Second, if a conclusion concerns applicants in the evening group, citing a person from the morning group switches the reference class. That person is outside the stated comparison group. Third, if the passage says a badge is sufficient for entry, proposing that a badge is the only route to entry silently adds “only.” Another route challenges exclusivity only if exclusivity was asserted. Check each candidate against the wording: who is covered, how many, under what condition, and during which period?

Ordinary-world possibility and passage-allowed possibility are not always the same. Unless the passage includes that rule, importing it does not show that its conclusion fails; it adds background material. Conversely, a countercase need not describe what normally happens. It needs to be coherent under the claims the item gives. Formal logic makes the same discipline explicit by holding the inference’s logical form fixed while considering cases that could make the premises true and conclusion false. The Delft chapter uses truth-value assignments and says English arguments must first be represented by their actual propositions. For a test item, the point is to avoid a similar story whose condition, group, or relation differs. A faithful countercase changes only whether the proposed conclusion holds, while leaving every premise intact; if it requires rewriting the passage, it has not yet answered the test question.

Sources: How to Think About Deductive Logic; University of Massachusetts Lecture Notes Unit 1; Deduction

How can I build and check a passage-faithful countercase?

A useful countercase must keep the passage's claims true while making the proposed conclusion false. Separate what the text says from what the answer adds, then check the imagined case against each sentence. “How to Think About Deductive Logic,” in Open Textbook Library’s Good Reasoning, defines a counterexample as one with true premises and a false conclusion. Delftse Foundations of Computation presents the same test as a truth assignment. In verbal reasoning, preserve the printed statements’ meaning and scope while imagining a case they permit. First, isolate the conclusion in plain language. Write the exact claim, not a broad impression. Second, list its relevant premises, including who or what they cover, the condition, and limits such as “all,” “some,” or “only.” Third, ask what the premises leave open. Fourth, describe the smallest situation that leaves every premise intact but makes the conclusion false. Reread the passage sentence by sentence. If your situation contradicts a premise, it is not a counterexample. Consider an original illustrative rule: “If a staff member completes the safety briefing, the staff member may enter the archive.” The proposed conclusion is, “If a staff member may enter the archive, that person completed the safety briefing.” Suppose Jo may enter because Jo is the archivist, but has not completed the briefing. The rule remains true: anyone who completed it may enter. It does not say briefing is the only route to entry. Jo’s case makes the conclusion false without violating the rule. The conclusion affirms the result of the condition and assumes the condition must have caused it; the passage never supplied that restriction. Now compare a near-miss. Suppose the conclusion instead says, “If a staff member completed the safety briefing, that person may enter the archive.” Jo entering without completing it does not falsify this claim. It shows briefing may not be necessary for entry, while the conclusion says it is sufficient. To challenge the latter, you need someone who completed the briefing but may not enter, while the rule remains true. A countercase must falsify the actual conclusion, not a related claim. Keep the case as small as the wording allows. If the passage says “every trainee who passes the briefing may enter,” a visitor who never took it cannot refute the rule; visitors are outside its group. Changing “may enter” to “must enter,” or adding “only after,” changes the claim. Track the same person, condition, time, and meaning. If a sentence has multiple reasonable readings, state which one you test; otherwise your countercase may target a paraphrase instead of the passage. A passage-bound countercase is semantic: it asks whether this wording allows a possibility with true premises and a false conclusion. A formal demonstration can replace subject matter with symbols while preserving argument structure. The Good Reasoning chapter recommends extracting form to expose validity; Delftse Foundations shows how a truth assignment can refute a form. For a test item, keep the passage’s relationships visible; symbols help only if they preserve them. If a conclusion has several parts, negate only what it asserts. “All scheduled sessions are recorded” is false if one scheduled session is not recorded; imagining that none are recorded is a stronger claim. Aim for a precise failure case.

Sources: How to Think About Deductive Logic; Deduction

Why does one counterexample settle the issue, but no counterexample does not?

A counterexample has decisive force because “must follow” is a universal claim: it says that every situation allowed by the passage makes the conclusion true. One coherent situation in which the passage’s claims remain true and the conclusion is false is enough to disprove that guarantee. The University of Massachusetts lecture notes define deductive validity in these terms: an argument is invalid if its form permits true premises and a false conclusion. The Delftse Foundations of Computation chapter makes the same asymmetry concrete in propositional logic: a single truth assignment with all premises true and the conclusion false refutes validity. Think of the premises as defining a set of permitted possibilities. A must-follow conclusion succeeds only if it holds in every member of that set. A counterexample points to one permitted possibility outside the conclusion. It does not need to be common, likely, or true of the actual world; it needs to be coherent under the stated claims. Conversely, checking several familiar possibilities cannot show that every other permitted possibility also supports the conclusion. The burden differs: one surviving exception defeats “all,” while establishing “all” requires accounting for every relevant case. That asymmetry makes the method a useful rejection check, but not a complete proof procedure. Suppose an item says, “Every approved request receives a reference number. This request has a reference number. Therefore, it was approved.” A possible request could receive a reference number before approval, so the first two statements can hold while the conclusion is false. That case works only if the passage does not say that reference numbers are issued exclusively to approved requests. If it does, the imagined case changes a premise and is not a counterexample. A plausible story is not enough; it must preserve the passage’s exact conditions and scope. There is also a positive route. A reader may derive the conclusion step by step from the premises or recognize an inference pattern that guarantees it. A proof rules out counterexamples by showing why the premises cannot all be true while the conclusion is false. The Delft chapter contrasts refutation by one counterexample with deduction as a proof method. On a timed multiple-choice item, a full formal proof may be unnecessary, but the distinction matters: failing to invent an exception is not the same as demonstrating that none is possible. Use a modest stop rule. First reread the quantifiers, conditions, and referents; then try one alternative that makes the conclusion false without changing those statements. If it survives, reject “must follow.” If each candidate violates a premise, do not count it. If no candidate works and you cannot derive the conclusion, record the result accurately: “not disproved by this search.” That leaves the answer unresolved rather than quietly converting a limited search into certainty.

Sources: University of Massachusetts Lecture Notes Unit 1; How to Think About Deductive Logic; Deduction

Why can a plausible answer survive a quick read?

A conclusion can feel likely or familiar before you check whether the passage guarantees it. A likely conclusion may still have an exception; a necessary one cannot. Time pressure and task instructions can affect how people distinguish valid from invalid inferences. A counterexample check asks: can the passage still be true when this answer is false?

In “In search of counter-examples: deductive rationality in human reasoning,” two experiments examined conditional inferences. In the first, 56 participants judged inferences under a timing constraint and 44 judged them without one; the timed group showed a smaller effect of logical validity on judgments. In the second, 36 participants were told to endorse only necessary inferences, while 33 received a general instruction to make logical inferences. The necessity-focused instruction produced a larger validity effect. These results concern the specific conditional tasks and instructions studied. They do not show that every test-taker accepts plausible but invalid answers. They do suggest why wording matters: “sounds logical” leaves room for likelihood, while “must follow” sets a stricter standard.

A familiar cause and effect can make this distinction easy to miss. Imagine a passage says that when a building’s alarm sensor detects smoke, the alarm sounds. The proposed conclusion is that the sensor detected smoke because the alarm is sounding. Smoke may be a familiar reason for an alarm, but the passage does not say it is the only trigger. A counterexample check asks whether another cause could sound it without smoke detection. This illustrates the logic operation; it is not evidence about how often people make this error.

“Everyday conditional reasoning: a working memory-dependent tradeoff between counterexample and likelihood use” examined participants’ verbal reports while they reasoned about everyday conditionals. Participants classified as having lower working-memory capacity more often reported using probabilistic information; those with higher capacity more often reported using counterexample information. This links reported strategy with working-memory grouping in that task. It does not establish that capacity causes an individual’s answer on a verbal test or justify diagnosing someone from a missed item. The narrower lesson is that judging what seems probable and checking whether an exception exists are different routes to an answer.

“Counterexamples in sentential reasoning” reported three experiments on how participants refuted invalid inferences involving sentential connectives. In the first, participants explaining their judgments used counterexamples more often than any other strategy. In the second, they were more likely to use counterexamples against invalid conclusions consistent with the premises than against conclusions inconsistent with them. Strategy use, then, depended in part on how the conclusion related to the premises. This does not establish a universal success rate or cover every verbal item. It does help explain why a conclusion that fits the passage can be harder to challenge: fit may feel like support, although support is not guarantee.

Together, these studies support a modest inference: under some experimental conditions, timing, instructions, reported strategy, and task structure related to how people evaluated inferences or used counterexamples. They do not show that a particular reader will be misled, that the routine guarantees correct answers, or that it improves an intelligence score. The individual item’s logic must settle its answer.

When an answer feels obvious, inspect its force. Is the conclusion “some,” “usually,” or “must”? Look for a possibility that preserves the passage but makes the conclusion false. If a coherent case survives, the answer is not guaranteed. If none comes to mind, the search alone is not proof; check that you have not assumed a cause is exclusive or a likely event certain. The pause separates plausibility from the claim the test asks you to evaluate.

Sources: In Search of Counter-Examples: Deductive Rationality in Human Reasoning; Everyday Conditional Reasoning: A Working Memory-Dependent Tradeoff Between Counterexample and Likelihood Use; Counterexamples in Sentential Reasoning

What does research on counterexample search actually add?

Research supports counterexample search in some tasks, and a classroom study suggests it can be taught for a specific mathematical argument. Neither finding shows that it guarantees accuracy across verbal tests, improves standardized scores, or raises general intelligence. A counterexample can disprove necessity; whether people find one depends on task and conditions. In “The Search for Counterexamples in Human Reasoning,” a 1999 conference study, participants evaluated eight inferences involving non-standard quantifiers. The university research record reports that participants constructed a specific instance of the premises on 80% of trials, and every participant produced at least one counterexample. This shows spontaneous search in that task. It is not a success rate for verbal-test questions: the study used a specialized set of deductions, and the record does not establish representative sampling or standardized-test performance. The 80% describes trials where a strategy appeared, not the share of conclusions correctly rejected. Other experiments show why strategy use should not be treated as automatic. “In Search of Counter-Examples: Deductive Rationality in Human Reasoning” tested judgments of conditional inferences. Under time limits, the effect of logical validity was smaller among 56 participants than among 44 without the timing constraint. In a second experiment, emphasizing that only conclusions following necessarily should be endorsed produced a larger validity effect among 36 participants than the generic instruction to make “logical” inferences did among 33. These results concern those conditional tasks. They support a practical point: explicit instructions about necessity and enough time can affect how carefully an inference is evaluated. They do not show that timing has the same effect on every test or that teaching this technique raises test scores. “Counterexamples in Sentential Reasoning” reports three experiments on how people refuted invalid inferences involving sentential connectives. In the first, participants required to justify their evaluations used counterexamples more often than any other strategy. In the second, they were more likely to use counterexamples for invalid conclusions consistent with the premises than for conclusions inconsistent with them. The third found no reliable difference between participants who wrote justifications and those who did not. Counterexample use varied with task structure and response conditions. They caution against promising that a reader can always retrieve a countercase, even when one exists. PubMed provides an abstract, so conclusions should remain at that level. A classroom case offers another kind of evidence. “Eliminating Counterexamples: An Intervention for Improving Adolescents’ Contrapositive Reasoning” studied 15 students in one U.S. eighth-grade mathematics class. The intervention had students describe possible counterexamples to conditional claims and their contrapositives, compare those descriptions, and use them to construct, critique, or validate arguments. The article reports improved performance on targeted direct and indirect argument items after instruction, though many students did not explicitly use the framework in their written answers. The authors note key limits: one class in a special learning environment, the same assessment before and after instruction, and little or no improvement among students with lower prior mathematics achievement. This case study supports the possibility of teaching contrapositive reasoning in that setting. It does not demonstrate transfer to adult verbal tests, general reasoning, or IQ. The experiments examine judgments on particular inferences under particular prompts or constraints; the classroom study examines learning a defined mathematical practice. None tests this routine in every verbal-reasoning format. Its purpose follows from deductive logic: one passage-faithful case with true premises and a false conclusion shows that a conclusion is not guaranteed. Research adds a caution about execution, not a reason to abandon the check. State the task’s standard, preserve the wording, and treat a surviving counterexample as decisive for that item. Treat the absence of one as unresolved unless a separate valid derivation establishes necessity.

Sources: The Search for Counterexamples in Human Reasoning; In Search of Counter-Examples: Deductive Rationality in Human Reasoning; Counterexamples in Sentential Reasoning; Eliminating Counterexamples: An Intervention for Improving Adolescents’ Contrapositive Reasoning

What if the passage itself is ambiguous?

Ambiguity is the main reason a counterexample can look decisive and still miss the question. Before rejecting a conclusion, write down what the relevant sentence most defensibly says, then check whether the proposed case keeps that reading true while making the conclusion false. If two readings remain plausible, report the ambiguity instead of claiming a clean refutation. Consider an invented example: “Every guide who reviews a file sends it to the editor before publication. Mira reviewed the file, so she sent it to the editor before publication.” On one reading, “it” means the file Mira reviewed. The conclusion follows from the rule and the fact that Mira reviewed that file. But if the passage instead said “Every guide who reviews a file sends it to the editor before publication. Mira reviewed a file, so she sent it to the editor before publication,” a reader might ask whether the rule applies to every file she reviewed, or whether the conclusion refers to a particular file not identified in the passage. That wording leaves the referent less clear. A case in which Mira sent one file but not another could challenge a broad conclusion, yet not one about the specific file covered by the rule. The countercase’s force depends on which object the passage connects. The same care applies to category boundaries and comparison groups. “Some trained guides review confidential files” does not say that all guides do, or identify which guides belong to the trained group. A countercase that changes “some” to “all” has changed the passage, not tested its conclusion. If the conclusion drops that limit or applies the comparison to a different group, test the actual wording rather than a broader paraphrase. These are scope errors: the countercase changes who, what, when, or under what condition a premise covers. Words such as “only” and “unless” deserve special attention because their direction matters. “Only reviewed files are sent to the editor” says that being reviewed is required for sending; it does not say every reviewed file is sent. A countercase with an unreviewed file that was sent would contradict that sentence, while a case with a reviewed file that was not sent would not. If a conclusion reverses that relationship, ask which condition is required. The point is to preserve the relation expressed in the item, not to import a preferred interpretation. The chapter “How to Think About Deductive Logic” in the Open Textbook Library’s Good Reasoning explains that a counterexample can fail when the argument’s form has been extracted incorrectly. A vivid scenario may seem to preserve the passage while quietly changing a quantifier, condition, pronoun, or comparison group. University of Massachusetts lecture notes on validity define a counterexample by keeping the premises true and the conclusion false. Together, these sources support a disciplined check, not a license to rewrite a sentence until it becomes easy to defeat. Separate two outcomes. If a case contradicts a premise under that reading, it is not a counterexample to that reading. If a case preserves the premises under one reasonable reading but not another, it identifies an interpretive fork. The item’s instructions and full wording should decide whether context resolves that fork. Do not add outside facts to repair an underspecified passage, and do not assume the test intends the broadest possible reading. When an answer key is available, a useful explanation should point to the word or relationship that fixes the intended scope. If no wording resolves the issue, the precise conclusion is that the passage permits more than one reading, so the counterexample alone cannot establish what the test intended.

Sources: How to Think About Deductive Logic; University of Massachusetts Lecture Notes Unit 1

What should I do after the check?

If a passage-faithful counterexample keeps the stated claims true while making the proposed conclusion false, reject the claim that it must follow. If your candidate cases fail, reread the wording once: check the quantifier, condition, and who or what each sentence covers. Distinguish derivation from failure to disprove.The University of Massachusetts lecture notes explain the asymmetry: a counterexample can show invalidity, but failing to find one does not prove validity. Without a valid derivation, leave it unproved.

For a brief practice check, consider this invented passage: every room with a locked cabinet requires a key from the office; Room C has a locked cabinet. Does the passage guarantee the key is currently in Room C? Imagine the requirement holds, but no one has collected the key. If the passage does not rule out that case, the conclusion is not guaranteed. This evaluates the inference, not the person: one item records performance under its conditions, not identity, worth, diagnosis, or fixed potential.

For low-stakes practice, Test IQ Free offers a free 50-question reasoning test and worked answers. Its estimated IQ range is provisional, based on an unvalidated conversion with no validated adult reference population; it is not an established population comparison. The optional $9 report adds explanation, not validity. Use them to review reasoning operations, not to infer that practice raises general intelligence or make consequential decisions.

Sources: Test IQ Free reasoning test; How to Think About Deductive Logic

Questions readers ask