How Much Does Prior Knowledge Really Matter for Learning?

New study challenges the "Knowledge-is-Power" hypothesis

So I just read a new paper that has troubled me. It troubles me because it questions something so fundamental about learning as to make me question everything I thought I knew, so I want to take a really close look.

“The most important single factor influencing learning is what the learner already knows. Ascertain this and teach him accordingly.”
— David P. Ausubel, Educational Psychology: A Cognitive View (1968)

One of the foundational assumptions in the science of learning is that prior knowledge facilitates new learning. It's an idea so widely accepted that it's often considered self-evident: the more you know, the more you can know. However a new experimental study by Zachary Buchin and Neil Mulligan has called all that into question.

The study, set out to test the “knowledge is power” hypothesis using a rigorous experimental design. Below is the full experimental procedure, which unfolded over 5–6 days in three phases: training, learning, and testing. In the training phase (3 days), participants studied three topics within one domain (e.g., sensation and perception), leaving one topic untrained to serve as a test of new learning. In the learning phase (2 hours), they read short texts on all four topics in each of two domains (trained and untrained), then engaged in either retrieval practice(answering and reviewing feedback on short-answer questions) or restudy (reading key ideas) for each topic, repeated twice. Finally, in the testing phase (2 days later), participants completed short-answer tests on all topics from both domains.

The researchers asked: Would students perform better on the new topic in the domain they had trained in (i.e., where they had high prior knowledge)? The answer? No!

Across every measure, final test scores, learning gains, performance on retrieval practice, there was no benefit of having more domain knowledge. (it feels weird even typic that sentence) Even more curiously, students felt that learning the new topic in the familiar domain was easier. It required less mental effort.

(Buchin & Mulligan, 2024, p. 93)

This is a surprising and frankly unsettling finding. If Ausubel’s above axiom is true (that prior knowledge is the single most important factor in learning) then we would expect at least some measurable advantage when students approach new material in a domain they’ve just trained in. Instead, the study found that learners performed no better on the untrained topic in the trained domain than on completely new topics in an unfamiliar domain. Even when the learning method was active and effective, like retrieval practice, the supposed advantage of prior knowledge vanished.

So…does knowledge not matter anymore?

Not quite. The study is well designed. It overcomes many of the flaws that plague previous research like relying on correlational data or failing to clearly define what "new learning" actually is. But its results need to be interpreted with caution. Here’s why.

The most fundamental issue for me is that participants received just 3 days of training. This isn't really "prior knowledge" in any meaningful educational sense. It's more like recent exposure. Real domain expertise develops over months and years, creating rich, interconnected knowledge networks. The study's manipulation may have created surface-level familiarity rather than the deep, structured knowledge that characterises true expertise.