Learning is not fully flexible until it can survive a changed surface
Transfer of learning, knowledge transfer, problem solving, transferable skills and applying knowledge all point to one difficult question: can a learner use what was learned when the next problem does not look like the lesson?
The National Academies describes transfer to new problems and settings as an important index of adaptive, flexible learning, while also documenting how difficult transfer can be. Expertise is deeply domain-specific: knowing how to reason well in one field does not automatically create a general-purpose reasoning engine for every field.
Return to How Learning Works →
1 · Why transfer fails
Learning binds to cues. The exercise heading names the method. The teacher’s diagram supplies the representation. The worked example supplies the first step. The classroom supplies familiar language. Remove those cues and the learner may possess relevant knowledge without recognising that it applies.
2 · Near transfer and far transfer
Near transfer changes relatively little: different numbers, a new sentence or another example of the same structure. Farther transfer changes representation, context or domain. The farther the change, the more difficult it is to recognise the invariant relation.
3 · Initial learning sets the ceiling
Transfer cannot rescue knowledge that was never sufficiently learned. A learner needs a stable enough model to carry. This is why transfer practice belongs after meaningful initial learning, though small transfer probes can also diagnose what the learner actually understood.
4 · Abstraction grows from comparison
Several concrete examples can reveal a common relation when they are aligned deliberately. Ask what stays the same while the surface changes. Abstraction is not the absence of examples; it is the structure extracted across examples.
5 · Variation must preserve something
Random variety creates noise. Educational variation changes a feature while preserving a target relation. Change the numbers, then the wording, then the representation. Each move asks whether the learner knows what is essential.
6 · Interleaving trains recognition
When several problem types are mixed, the learner must identify the structure before executing a method. This is a bridge from procedural fluency to transfer because future problems rarely announce their category.
7 · Explanation makes the transferable relation explicit
“Why does this method work?” and “What feature tells you to use it?” force the learner to articulate a relation that can later serve as a retrieval cue. Explanations should be checked against evidence; fluent stories can still be wrong.
8 · Counterexamples teach boundaries
A rule transfers intelligently only when its conditions travel with it. Faith’s arithmetic-average shortcut becomes useful when she understands when equal weighting is justified and when it is not.
9 · Retrieval supports transfer when retrieval cues vary
If knowledge is always retrieved from the same prompt, access can remain narrow. Ask for the same concept through a definition, diagram, explanation, prediction and problem. Multiple routes make the knowledge more usable.
10 · Transfer is domain-sensitive
The National Academies notes that problem-solving expertise depends strongly on domain knowledge. Teaching “critical thinking” without substantive knowledge risks producing slogans rather than capability. General strategies should be embedded in real disciplines.
11 · Alicia: evidence across subjects
Alicia learns to ask what a detail proves in English. Transfer is tested when she must judge evidence in Science or a source-based humanities task. The general relation—claim, evidence, warrant—travels only if she understands how each discipline defines acceptable evidence.
12 · Beatrice: fractions across changing wholes
Beatrice’s fraction knowledge transfers when she identifies the relevant whole in discounts, remaining quantities, probability and ratios rather than relying on one familiar diagram.
13 · Ciara: checking across representations
Ciara’s “answer job” check transfers when she uses it in tables, graphs, word problems and practical Science questions. The surface changes; the control principle remains.
14 · Denise: logic across language
Denise first represents “unless” through two cases. Transfer occurs when she can recognise the same conditional structure in new sentences without drawing the cases every time.
15 · Emily: algebra beyond the worksheet heading
Emily’s algebra transfers when she selects the method in a mixed problem, constructs the equation from a context and preserves the conditions of operations without the example beside her.
16 · Faith: patterns with conditions
Faith’s pattern recognition becomes transferable expertise when she carries the boundary conditions along with the shortcut. Fast recognition plus verification is stronger than fast recognition alone.
17 · Transfer and examinations
Examinations are transfer environments. Questions combine familiar knowledge with changed wording, representation and time constraints. Practice that remains identical to teaching examples can create local fluency without examination flexibility.
18 · Transfer and professional learning
A procedure learned in training must survive the workplace. A policy must survive a novel case. A communication principle must survive a difficult conversation. Simulation, case variation and feedback help bridge the training environment and the performance environment.
19 · Transfer and AI
AI can generate varied examples quickly, but variation is only useful if the learner still identifies the invariant. Ask AI for contrast cases, then require the learner to explain what changed and why the same principle does or does not apply.
20 · A transfer ladder
Same structure, different numbers. Same structure, different wording. Same relation, different representation. Mixed categories. Delayed return. New context. Combined concepts. Real performance. Move upward when evidence shows the lower rung is stable enough.
21 · What transfer does not mean
Transfer does not mean one skill magically applies everywhere. It does not mean every lesson needs a wildly novel problem. It does not mean novices should be thrown into far-transfer tasks without sufficient knowledge. It means instruction deliberately prepares knowledge to be recognised and used beyond the exact conditions of acquisition.
22 · The return
The strongest evidence of learning is not that the learner can reproduce yesterday’s example. It is that the learner can meet tomorrow’s problem, recognise enough of the underlying structure, adapt what was learned and know when the old method no longer fits.
Research route
- National Academies — Learning and Transfer
- National Academies — Expertise and Transferable Competencies
- National Academies — Transfer of Knowledge
How Learning Transfer Works → · Science of Learning → · How Learning Works →
