Prior Knowledge Determines New Learning¶

Claim · All claims
Evidence · 2 studies · 1 quant-synthesis r3, 1 review r3 · q4 · i3 large · n=8,776 effect sizes
Currency · newest study 2023 · 1 of 2 entries checked against their sources (automated) · not reviewed by a person

New learning is built on what the learner already knows: existing knowledge structures determine what can be understood, how new information is organized, and where misconceptions will interfere. This is a foundational premise of cognitive load theory and schema-based accounts of learning rather than a single experimental finding — its Evidence section therefore awaits a dedicated synthesis of the supporting literature.

Subclaims¶

q4 i3 A meta-analysis of 8,776 effect sizes finds a strong correlation (r+ = .534) between learners' prior knowledge and their later knowledge on the same domain, but almost no relationship between prior knowledge and normalized learning gains (r+ = -.059), with wide variability across studies. → Simonsmeier et al. 2022

q4 i? A meta-analysis distinguishes stability of individual differences (r+ = .53 over 7772 effect sizes) from the effect on gains (rNG+ = −.06 over 697 effect sizes), which varied from −.94 to .93. → Bittermann 2023

Evidence¶

Simonsmeier et al. 2022¶

Simonsmeier, B. A., Flaig, M., Deiglmayr, A., Schalk, L., & Schneider, M. (2022). Domain-specific prior knowledge and learning: A meta-analysis. Educational Psychologist, 57(1), 31–54. doi:10.1080/00461520.2021.1939700

q4 · meta-analysis · i3 · large stability correlation, r=.534; near-zero for normalized gains, r=-.059 · n=8,776 effect sizes · quant-synthesis · r3

A meta-analysis of 8,776 effect sizes examined the relationship between domain-specific prior knowledge (measured at Time 1) and knowledge outcomes measured later, across studies spanning many ages, domains and designs. Two distinct relationships were separated: the stability of individual differences (correlation between pretest and posttest knowledge scores) was high, r+ = .534 — learners who know more at the start tend to know more at the end, supporting the claim that new learning builds on and is strongly predicted by prior knowledge. But the predictive power of prior knowledge for normalized learning gains (how much a learner improves, controlling for starting point) was low and centered near zero, r+ = -.059, with a large 95% prediction interval of [-.688, .621] — meaning the direction and size of prior knowledge's effect on gains varies enormously across contexts, sometimes helping and sometimes hurting. The authors argue this variability falsifies both "knowledge is power" and "the effect of prior knowledge is negligible" as general statements, and calls for research on the conditions (moderators) under which prior knowledge helps, hurts, or does nothing — directly supporting this page's Discussion section on expertise reversal, moderators and boundary conditions rather than a single uniform effect.

Bittermann 2023¶

Bittermann, A., McNamara, D., Simonsmeier, B. A., & Schneider, M. (2023). The Landscape of Research on Prior Knowledge and Learning: a Bibliometric Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-023-09775-9

q4 · i? · review · r3

The article reports a recent meta-analysis (Simonsmeier, Flaig, Deiglmayr, et al., 2021) in which stability was high (r+ = .53) while the gain correlation was virtually zero (rNG+ = −.06), with effect sizes varying from −.94 to .93.

"This stability was high (r P + = .53) averaging over 7772 effect sizes. The correlation between prior knowledge and knowledge gains from pretest to posttest reflects the effect of prior knowledge on learning something new. This effect size was virtually zero (rNG + = −.06), averaging over 697 effect sizes."

Studies bearing on this claim¶

Does knowing more beforehand mean learning more from the instruction? This table gathers the studies behind this claim and the 15 narrower claims that bear on it: 15 distinct studies (5 causal, 4 associational, 2 review, 2 theoretical, 1 quant-synthesis, 1 design), q2–q4, 5 of 15 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

This claim is one of the most robust premises in educational psychology, but its strength is also its boundary condition: because prior knowledge acts as a filter and scaffold, learners with different starting points respond differently to the same instruction. The same instructional treatment is rarely optimal across the prior-knowledge distribution [~S].

Activation matters, not just possession. Having relevant knowledge is not sufficient; it must be retrieved and connected to the new material. Deliberately prompting recall of relevant prior knowledge at the start of instruction improves comprehension and integration [~M] — see Activation improves learning. and the strategy of activating prior knowledge. Advance organizers work by providing a bridge between existing knowledge and new content [~M] — see Advance organizers improve learning.

Misconceptions resist correction. When prior knowledge is inaccurate, it does not simply get overwritten; learners often assimilate new information into flawed schemas or revert to misconceptions after instruction. Productive conceptual change typically requires creating cognitive disequilibrium — surfacing the conflict between the learner's prediction and the evidence — rather than mere re-explanation [~M].

Expertise reversal. The same prior knowledge that enables learning changes what instruction is helpful. High-knowledge learners are burdened by scaffolds designed for novices, and low-knowledge learners are overwhelmed by instruction designed for experts [~S] — see expertise reversal effect. This makes prior-knowledge assessment a prerequisite for sequencing decisions such as fading worked examples.

Working-memory interaction. Prior knowledge reduces the working-memory load of new material by allowing chunking and automatisation; learners without relevant schemas must assemble everything from parts [~S] — see chunking reduces working memory load. and cognitive load theory.

Design implications. Because prior knowledge governs what instruction can accomplish, designers should (a) assess prior knowledge before selecting treatments, (b) activate relevant knowledge at the point of need rather than assuming it will be retrieved spontaneously, and (c) diagnose misconceptions early, since unaddressed inaccuracies distort how new content is encoded. Sequencing decisions — from advance organizers to fading of worked examples — should be driven by measured learner starting points, not by a one-size-fits-all progression.

Open questions. How durable misconception correction is across time, and how precisely prior knowledge must match new content (versus merely being analogically related), remain active areas. Analogies can bridge partial gaps [~M] — see analogical reasoning improves transfer.

Related Claims¶

• Activation improves learning. — possessing prior knowledge only helps when it is activated at the right moment
• Advance organizers improve learning. — a structured bridge from existing knowledge to new material
• Chunking reduces working memory load. — prior knowledge supplies the chunks that free working-memory capacity
• Analogical reasoning improves transfer. — analogies leverage existing knowledge to structure unfamiliar content
• Cognitive disequilibrium motivates conceptual change. — how inaccurate prior knowledge is confronted and revised
• Expertise reversal effect. — prior knowledge determines which scaffolds help and which hinder
• Pre-instruction LCTSR scores correlate with normalized learning gains, most strongly for theoretical content (TUG-K r=0.59) and more weakly for descriptive content (DIRECT r=0.50) — reports the opposite
• Positive teacher–student relationships are associated with greater student engagement and, more weakly, with achievement — related
• Synaptic connectionism is argued to be implausible for complex cognition because synaptic plasticity has been studied only as an analog of simpler learning forms — related
• The pretest-posttest correlation is a factor in determining a study's statistical power in education evaluations using state assessments. — related
• Deep thinking and making connections help students build strong memories and learn more deeply — related
• Learning is effective when it builds on children's existing knowledge in real-world contexts, and early fraction foundations matter because many older children struggle with rational numbers — a narrower finding that bears on this claim
• Reported prior knowledge of lesson content differed sharply between sessions (r=−.330) and was treated as a measure of content divergence, not a covariate — a narrower finding that bears on this claim
• Students who transferred prior-domain knowledge tended to offer correct explanations in the other domain — a narrower finding that bears on this claim
• Advance organizers facilitate conceptual retention and far transfer of unfamiliar technical material for learners lacking relevant prior knowledge — related
• mPFC activity during schema-related encoding predicts normalized second-year course grades, with no such relation for MTL activity — related
• Pretest performance and average math grade correlate with conceptual knowledge but not procedural knowledge posttest scores — a narrower finding that bears on this claim
• Prior knowledge strongly predicts post-test application and transfer scores in both topics — a narrower finding that bears on this claim
• The relationship between learners' prior knowledge and the optimal instructional sequence may not be straightforward — related
• Prior mathematical knowledge correlates very strongly with later mathematics achievement (r = .63), largely reflecting stable differences between learners — a narrower finding that bears on this claim
• When the γ-process (associative interaction of correct and incorrect knowledge) is considered, the normalized gain correlates positively with pretest score — reports the opposite
• RT propensity shows no reliable main effect on learning rate, while prior proficiency is positively associated with learning rate — related
• Hake's survey found interactive-engagement courses achieved average normalized gains about two standard deviations greater than traditional courses, with overall gain–pretest correlation of +0.02 — related
• Prior knowledge is not significantly related to test performance in this sample — reports the opposite
• Passage familiarity (prior related knowledge) correlates .52 with recall, and only familiarity, not imageability, affected one-week delayed recall — a narrower finding that bears on this claim
• Prior biology knowledge predicted new AI knowledge in Lessons 1 and 2 but not in Lessons 3 and 4 — a narrower finding that bears on this claim
• Subjects with a more developed schema for a body of knowledge show higher recall for materials related to that knowledge — a narrower finding that bears on this claim
• The positive relationship between kindergarten entrance age and achievement primarily reflects skill accumulation prior to kindergarten rather than heightened in-school learning ability — a narrower finding that bears on this claim
• Knowledge gains correlate positively with belief change, more strongly for higher relational-understanding levels — related
• Conceptual feedback significantly reduces the effect of prior knowledge on achievement — related
• Chunking via prior knowledge can massively increase the information processed in working memory — related