Self-Explanation Helps Learners Reframe Problems and Find Knowledge Gaps
Cognitive scientist Tania Lombrozo argues that when information is abundant, learning depends not just on finding answers but on working out what we know and what we still need to learn. In her TED talk, she makes the case for learning by thinking—especially by explaining ideas to yourself—as a way to expose gaps in understanding, ask better questions and make new information more useful.

More information is not the same as more learning
Software engineers sometimes explain a problem in their code to a rubber duck. The duck does not answer, and the programmer receives no new information. Yet working through the problem aloud can reveal what is wrong. Cognitive scientist Tania Lombrozo calls this learning by thinking: using what is already in memory and skill to produce an answer that was not immediately available.
The distinction matters as access to information expands. Lombrozo argues that the instinct to meet every question by looking outward—to the internet, social media or AI—can worsen information overwhelm. More information does not necessarily make learning easier. The challenge is not only finding material. It is knowing what to look for, and how to make sense of what we find. Starting with what one already knows can help with both.
Lombrozo says our minds contain extensive stores of knowledge in memories and skills, even when we do not readily recognize that knowledge as something we can use. She illustrates this with three questions: How many windows are along the front of your home? What is the fifth letter of the alphabet? Is Z on the left or right side of a typical keyboard? Many people may not have those answers ready as facts. They can nevertheless work them out from visual memory, the remembered sound of the alphabet, or the motor knowledge involved in typing.
In each case, the answer comes from transforming a different kind of knowledge into a form that answers the question. The windows may be represented as a collection of visual memories rather than a count. To get the number, one has to inspect and combine those memories. The result is new knowledge, though no new information has arrived from outside. Lombrozo’s point is that knowing something is not limited to having a ready-made answer stored in the exact form a question demands.
She compares the process to cooking. A recipe calls for chocolate chips and salted butter; a cook might make the first from a chocolate bar and the second by combining unsalted butter with salt. Learning by thinking likewise transforms and combines what is available to produce what is needed. The ingredients are already there, but they have to be worked with.
There is a limit: someone who has never visited your home cannot reason their way to its window count. They need new information. But thinking about the question before visiting can still prepare them to learn. It makes the missing fact explicit and gives them a target: when they arrive, they know what to look for. In Lombrozo’s cooking analogy, it is sensible to check the ingredients already in the kitchen before making a shopping list.
Learning by thinking is therefore a starting point, not a substitute for information. It cannot provide facts a person has never encountered. It can help them identify what they already know, what they still need to find out, and how new information might fit with what they know already.
Explaining changes how learners represent a problem
There is no single learning strategy that guarantees the best result in every situation, Lombrozo says. But seeking explanations comes close. Students with access to the same material can learn very different amounts. Research beginning in the late 1980s found that successful students often asked themselves questions as they studied and tried to answer them. They acted as their own rubber ducks.
Subsequent studies, Lombrozo says, have found that children and adults who explain to themselves tend to learn better and apply what they learn more broadly than those who do not. Researchers call this the self-explanation effect. Lombrozo’s own work illustrates how explanation can alter the way a learner represents information—not just help them recall it.
Imagine a reference card showing two ducks. Which of two new cards belongs with it: one showing two frogs, or one showing a duck and a fox? Adults typically choose the frogs. They can represent the cards not only as two objects of a particular kind, but as pairs of matching objects: duck with duck, frog with frog. On that representation, two frogs match two ducks in structure.
Five- and six-year-olds typically choose the duck-and-fox card instead. They represent the reference card as containing a duck and a duck, so another card with a duck on it seems more similar than one with no ducks. The displayed example makes the alternatives concrete: the reference card has two yellow ducks; the other cards show two frogs, two ducks, or a duck and a fox.
Even when children see examples that demonstrate the adult pattern—matching the same with the same—they may continue to choose according to the objects themselves. In one study, Lombrozo and her colleagues prompted half the children to explain why examples of two-frog cards went with two-duck cards. The researchers gave no feedback and supplied no new information. The children who explained subsequently matched new sets of cards according to the adult pattern. Lombrozo’s interpretation is that explaining led them to represent the task differently, enabling them to solve it in a new way.
By explaining, they changed the way they represented the information to solve the task.
The example makes explanation more than a way of rehearsing an answer. The children had already encountered examples of the pattern, but the act of explaining was associated with a change in how they approached new cards. In Lombrozo’s account, learning by thinking can reorganize what a learner has in mind, allowing the same information to support a different answer.
By explaining, they changed the way they represented the information to solve the task.
An attempted explanation reveals what is missing
Explaining can also expose the limits of what a person knows. Lombrozo recounts a story from her friend Roxana, whose young son asked how tornadoes start. Roxana had studied climate and natural disasters in college and believed she could explain the process. She began describing differences in pressure, then found she could not explain what set the process off or exactly how low pressure produced a funnel.
The moment is familiar, Lombrozo says: trying to explain something can reveal that one understands less than one thought. Psychologists call this the illusion of explanatory depth. A person may feel they understand a familiar phenomenon until asked to account for how it works in detail. The attempt to connect the pieces makes the missing links harder to ignore.
For Roxana, the gap was not simply “tornadoes” as a broad topic. She could identify the specific part she could not explain: how low pressure related to the funnel. That gave her a more useful next step than a general search for information. The attempt to explain had not supplied the missing account, but it had helped locate what was missing.
Lombrozo argues that this is one reason to begin learning with the information already at hand. Trying to make an explanation from what one knows can show where the account breaks down. The gap then points to a more specific question to pursue, and to the information that could help answer it. Once that information is found, the learner has some existing ideas to connect it to.
This is part of the practical value of thinking before searching. A learner who begins by trying to explain does not have to pretend that the explanation is complete. The uncertainty itself can be useful: it can distinguish what seems understood from the particular link or fact that still needs work. Thinking does not eliminate the need to seek information; it can make the search more directed.
Use outside information without outsourcing understanding
Lombrozo expects information overwhelm to intensify as tools for retrieving and synthesizing information become more pervasive. AI-generated search responses and chatbot explanations can be useful, she says, but they should not replace learning by thinking. Her distinction is not between using these tools and refusing them. It is between using external information and outsourcing the work of making sense of what one knows.
If people turn first to external answers, they may bypass the work of transforming and combining what they already know—the work that helps them formulate the right questions. They may also become less able to distinguish what they understand from what they can merely access. Lombrozo describes the result as an illusion of understanding: people can lose track of the boundary between knowledge in their own heads and information available outside them.
Seeking an explanation before relying on a ready-made one can help preserve that boundary. Asking oneself questions and trying to answer them can reveal what is already understood and where an explanation stops. That can guide a search toward a specific gap, rather than an open-ended request for more information. It can also help a learner connect new information to ideas and memories already in place.
Lombrozo’s point is not that learning by thinking makes learning easy, or that it removes the need for new information. In many cases, a learner cannot reach the answer without facts from outside their own experience. But the effort of explaining can prepare them to find those facts and to use them: it clarifies what they are missing and gives them a starting structure for understanding what they learn next.
Before seeking the answer outside your head, Lombrozo advises, ask yourself a question and try to explain. “Be your own rubber duck.”


