Critical Thinking Can Transfer Only When Its Structure Is Made Visible

Critical thinking arrives with a suitcase full of very good verbs. Identify. Evaluate. Infer. Explain. Regulate.

The Delphi report gives it those six core skills, and the list has the pleasant authority of an airport departures board.

Then it makes a larger promise. Learn the procedure here, think better there.

Move from a laboratory report to a headline, from a history document to an epidemiological claim, from a suspicious graph to whatever strange creature the internet has produced before lunch.

Learning research enters with a less romantic question. What, exactly, moved?

Thorndike and Woodworth make the old faculty psychology sound like a gym membership for the soul.

Train attention with Latin or geometry, it said, and the strengthened faculty will stride confidently into every later problem.

Their theory of identical elements offered the cooler answer. Practice transferred when the new task shared concrete elements with the old one. Otherwise the trained faculty had apparently misplaced its shoes.

Judd’s underwater-dart experiment complicates the funeral.

Boys who learned the optical principle of light refraction did no better at the original target depth than the group who had not.

Change the depth, however, and the group with the principle adapted far better.

Judd’s generalization theory says that learning can travel when the learner has abstracted what is doing the work.

The two positions are not enemies so much as the field’s two weather reports.

Some learning stays local because its useful features stay local. Some moves because the learner has found a relation that survives a change of scenery.

Critical thinking lives precisely in that awkward interval.

A claim is not an answer. Evidence is not decoration. A warrant is the bridge between them.

Toulmin’s model calls the conclusion the claim, the offered support the grounds, and the mostly unspoken bridge the warrant. The last of these is where arguments hide their small animals.

Consider two sentences. All birds have wings. All valid premises support this conclusion.

The nouns have changed from feathers to logic, and the second sentence is not a good argument merely because it wears a tie.

But you can look in both cases for the same relation. What is asserted, what is said to support it, and whether the support actually reaches the claim.

Transfer begins when a relation, not just a phrase, becomes visible.

So the routine looks like the whole answer.

State the claim precisely, inspect the source, grade the evidence, apply base-rate discipline, consider alternative explanations. A general critical thinker will emerge, wielding an acronym like a compact umbrella against the weather.

The routines are real.

CRAAP asks for currency, relevance, authority, accuracy, purpose. SIFT says stop, investigate the source, find better coverage, trace claims to their original context.

Its crucial habit is lateral reading. Professional fact-checkers leave the page. Students, and even historians, tend to stay and admire its credentials.

But a good routine does not carry its own world inside it.

Credibility, expertise, and conflict of interest can be checked across subjects. Deciding whether a medical study’s design warrants a causal claim, or whether a historical source has omitted the relevant archive, requires knowledge the checklist cannot mail in later.

Willingham’s formulation is blunt. Critical thinking is knowledge-dependent.

Without the domain’s plausible alternatives, the alternative-explanations step becomes a polite empty chair.

This is not an insult to abstraction. It is an account of its price.

Near transfer crosses contexts that share most surface and structural features. Far transfer crosses substantially different ones.

Barnett and Ceci’s taxonomy separates distance in domain, physical setting, time, social context, and format, because calling everything far had permitted a century of research to put different journeys in the same hat.

Judd’s refraction principle is real transfer. It is not an all-access pass from dart throwing to foreign policy.

Research programmes that claim generally transferable thinking, including executive-function training and brain training, tend under rigorous designs to show gains on structurally similar tasks and attenuation on only abstractly related ones.

The live question is not whether far transfer exists. It is which cues, knowledge, and representations let it occur.

Bjork’s desirable difficulties make the same point with a small sadism clause.

Spacing, practice testing, and interleaving can worsen performance during learning while improving later retention or transfer.

The meta-analytic record is unusually good here. Practice testing beats restudying by a solid margin. Spaced practice beats massed practice in nearly every comparison pooled.

The effects are sturdy. The desired difficulty is not a universal property of suffering.

Rohrer and Taylor found interleaved geometry practice far ahead on a test a week later, despite worse accuracy during practice itself.

Yet interleaving can weaken or reverse for learners with no prior familiarity. The challenge-point framework makes the unhelpful but correct observation that a useful difficulty depends on the particular learner’s current skill.

A procedure can expose a structure only when there is enough structure there to see.

Learning styles do not rescue the general promise.

The VARK model offers categories. The meshing hypothesis makes the stronger causal claim that matching teaching to a preferred category improves outcomes.

The review that looked for adequate crossover studies found almost none, and no meshing benefit in the few using the necessary design.

A preference for diagrams is not a theory of how a warrant reappears in a different subject, however much stationery may be involved.

The measurement problem is therefore not a footnote with a tiny clipboard.

The Watson-Glaser appraisal offers five timed subtests: inference, recognition of assumptions, deduction, interpretation, and evaluation of arguments. The CCTST is a multiple-choice test built from the Delphi skills, whose internal consistency sits below the conventional benchmark for individual decisions.

Neither number dissolves the construct-validity question.

McPeck’s critique names it. A science item and a history item do not merely test a content-free operation in different costumes. They bring different knowledge and different standards of evidence into the room.

The domain-specificity claim has conceptual, epistemological, and empirical versions, and the evidence about how much transfer occurs is thin in all three.

A score predicting careful reasoning across all domains is therefore speculation, not an achievement certificate waiting for a lanyard.

What survives is neither a magic lever for general intelligence, nor a verdict that critical thinking was only pretending to be a superpower.

It is that both traditions ask the mind to notice what remains when the props change.

Reasoning does not travel because a checklist has been memorized. It travels, when it does, because a relation has become legible.

Grounds, claim, bridge. Evidence, hypothesis, alternative. A target under altered water, and the refraction that altered it.

The subjects change costumes. The structure waits backstage.