Does Thinking Silently Help Students Learn?

Teachers often assume students 'think along' when other students answer. A new study shows that unless they're actively engaged, they might not be learning much.

A new meta-analysis by Yu and colleagues, published in Educational Psychology Review, has done something very interesting: they've quantified the learning benefits of what researchers call "covert retrieval", the academic term for mentally recalling information without producing an overt response.

Picture this: A teacher asks a question and calls on Sarah. While Sarah formulates her response, what are the other 29 students doing? Probably half-listening, maybe pretending to be focussed, but maybe not learning much. Now picture the alternative: the teacher poses the same question but before calling on anyone, says "ok I want everyone to think about this for 30 seconds and I’m going to ask someone at random." This small technique (a key component of ‘cold calling’) is one of the most effective tools in a teacher’s repertoire.

And this isn't just good classroom management, there's compelling evidence that simply having those 29 students think about the answer, without ever speaking it, could measurably boost their learning BUT the way in which we put this into practice matters and as ever, it’s easy to get it wrong.

The Study

The researchers analyzed 18 studies involving 2,560 participants to answer two crucial questions: Does thinking silently about answers actually help you learn? And is it as effective as writing or speaking those answers aloud?

The first answer is yes, with an important caveat. Covert retrieval (thinking about an answer) produces a small but significant learning benefit (effect size g=0.23). To contextualise this: if a student could normally recall 11 out of 22 items on a test, covert retrieval practice would help them remember one additional item, 12 out of 22. Now one extra correct answer might not sound revolutionary, but multiply that across hundreds of learning opportunities throughout a semester, and the cumulative effect becomes more substantial.

But there are important considerations: this benefit only emerges when students receive feedback about their thinking. Also I’d suggest, it’s because the internalised thinking comes into contact with some kind of external reality. And that's where techniques like cold calling appear to be so effective. Cold calling (randomly selecting students to answer questions) is consistent with many of the core principles of cognitive science and this study seems to provide further evidence to support it. The logic seems sound: if any student might be called upon, all students must stay engaged. It's essential for checking understanding across the whole class, not just the eager hand-wavers.

The Rub

But before we declare victory for quiet contemplation, the research delivers a somewhat sobering truth: overt retrieval consistently outperforms covert retrieval on its own. Speaking, writing, or typing answers provides a moderate but reliable advantage (g=0.17) over silent thinking. On this, the authors cite a very interesting term which I’ve not come across; "truncated search":

"For instance, the truncated search theory hypothesizes that people put less effort into retrieval when making delayed JOLs (or when performing other forms of covert retrieval) and truncate their retrieval attempts prematurely rather than making full-blown retrieval attempts to generate a complete answer (Tauber et al., 2015)."

In other words, when we think silently, we're lazy searchers. We stop digging through our memory once we hit something that feels "good enough"; a half-formed idea, a vague sense of familiarity, or a partial answer that satisfies our mental itch. But when we must speak or write, we can't get away with such cognitive shortcuts. We're forced to keep searching until we can produce something coherent and complete.

It's the difference between recognizing a song when you hear it versus being able to sing it from memory. The latter demands a more thorough excavation of what we actually know. This suggests that the very difficulty of overt production (the struggle to find the right words, to organise thoughts coherently) is precisely what makes it more effective for learning.

Not All Silent Thinking Is Equal

The research distinguished between three types of covert retrieval, and the differences matter:

1. Answer thinking (silently generating responses): Effective

2. Delayed judgments of learning (predicting future performance): Effective

3. Answer monitoring (judging others' responses): Ineffective

To me this again underlines why minimally guided approaches such as discovery learning are so ineffective: Learning requires active mental generation of responses, not passive observation of others' thinking. In discovery learning or project based activities for example, students are often expected to learn by watching peers explore, discuss, or stumble toward solutions, but this is essentially "answer monitoring" writ large.

Implications for teachers

Here's where it gets interesting for teachers and instructional designers. The power of covert retrieval isn't constant, it depends entirely on what happens next. When students engage in silent mental practice without receiving the correct answer afterward, the learning benefit nearly disappears (g=0.04). But when they get corrective feedback, ie when they discover whether their mental answer was right or wrong, the effect jumps dramatically to g=0.59. Perhaps most importantly for educators focused on lasting knowledge rather than performance theater, both covert and overt retrieval work best over time. The meta-analysis revealed that retrieval practice, whether silent or spoken, shows minimal benefits on immediate tests but substantial advantages after delays of a day or more.

I’ve long thought that ‘checking for understanding’ is perhaps the most powerful level of classroom instruction but this research exposes an uncomfortable truth about the technique and how it can lethally mutate: most of the time, we're not actually checking, we're sampling. And our sample size is woefully inadequate. (I think it was Dylan Wiliam who said that if you are measuring learning by taking answers from the 4 or 5 most confident students in the class, then you’re not measuring much at all.)

This aligns with what we know about memory consolidation and creates an interesting problem for teachers obsessed with immediate evidence of learning. The brain needs time to strengthen learning, and the initial difficulty of retrieval triggers processes that make memories more durable. It's "desirable difficulty", struggles that feel harder in the moment but pay dividends later.

The research also reveals something crucial about timing. Both covert and overt retrieval work best over time, showing minimal benefits on immediate tests but substantial advantages after delays of a day or more. This creates an interesting problem for teachers obsessed with immediate evidence of learning which is more often about performance not actual learning.

Key takeaways

So what does this mean for classrooms? One thing I really like about this paper is that the researchers are clearly thinking about classroom practice and how to apply this research. For example:

Three key shifts can transform how we use questioning. First, before calling on any student, have ALL students mentally prepare their answer, this isn't dead time, it's active learning time for everyone. Second, after revealing the correct answer, acknowledge the silent work with questions like "How many were thinking along similar lines?" which transforms what would otherwise be wasted mental effort into actual learning. Finally, don't expect immediate payoffs. The research shows that both covert and overt retrieval work best over time, so design your assessments to capture learning over days and weeks, not minutes.

Ultimately, this research contributes to a growing understanding that learning isn't passive absorption but active construction. Every time students generate answers, wither silently or aloud, they're building stronger, more retrievable knowledge structures.