How You Can Learn in a World of Information Overload | Tania Lombrozo
Access to information was once a major constraint on learning. Today, the harder problem is often deciding what deserves attention. Search engines, social platforms, online courses, and generative AI can produce answers faster than we can evaluate them. In her TEDxNewEngland talk, cognitive scientist Tania Lombrozo argues that better learning does not always begin with gathering more material. Sometimes it begins by working more deliberately with what is already in the mind.
Lombrozo calls this process “learning by thinking.” The idea is not that reflection can replace evidence or supply facts we have never encountered. Rather, thinking can retrieve hidden knowledge, reorganize familiar information, generate new conclusions, and reveal precisely where our understanding fails. That makes internal reflection a powerful first step—and a useful defense against information overload.
Why Explaining a Problem Can Help Solve It
Software engineers often describe a stubborn problem aloud to a rubber duck or another silent object. The duck contributes no technical expertise. Its value comes from forcing the programmer to state the situation clearly, reconstruct the relevant steps, and make implicit assumptions visible.
Lombrozo uses this practice to illustrate a broader principle. We commonly treat learning as information moving from the outside world into the mind: a teacher gives an explanation, a book provides facts, or a search engine returns an answer. Yet people can also learn by examining and manipulating information they already possess.
Consider a question about a familiar room or childhood home. You may never have memorized how many windows it contains. Even so, you might mentally walk through the space, inspect a visual memory, and calculate the answer. The number itself was not stored as a ready-made fact. It was produced by transforming one kind of knowledge—a remembered scene—into another form.
The same can happen with sounds, movements, spatial layouts, and partially understood concepts. A person may know more than they can immediately state because useful information is distributed across different kinds of memory. Thinking is one way to recover and recombine it.
Learning Requires Transformation, Not Just Retrieval
Lombrozo compares internal knowledge to ingredients in a kitchen. What you possess may not match what a recipe requests, but it can sometimes be converted or combined. A chocolate bar can become smaller pieces; salt and unsalted butter can be used where salted butter is needed.
Knowledge works similarly. A memory, rule, example, or physical skill may contain the raw material for an answer without containing the finished answer itself. Effective thinking involves changing the representation of that material: turning an image into a count, several examples into a pattern, or separate facts into a causal explanation.
This matters because a person can consume large quantities of information without performing much transformation at all. Reading another article or requesting another summary may increase what is available, yet availability is not the same as understanding. Learning deepens when we connect new material to prior knowledge, test its implications, and use it to answer questions that were not answered for us.
There are obvious limits. No amount of introspection will reveal a historical date, an unfamiliar scientific measurement, or the contents of a document we have never encountered. Learning by thinking cannot manufacture missing evidence. Its purpose is to determine what can be derived internally and what must be acquired elsewhere.
The Research Behind Self-Explanation
A particularly effective form of learning by thinking is self-explanation. Rather than merely rereading material, learners pause to ask questions such as:
Why does this step follow from the previous one?
How does this example fit the general principle?
What would happen if one condition changed?
Can I explain the idea without looking at the source?
Which part of my explanation remains vague?
Lombrozo points to research beginning in the late 1980s that examined why students exposed to similar lessons and resources could emerge with very different levels of understanding. One important difference was what successful learners did while studying. They generated explanations, questioned intermediate steps, and tried to account for the relationships in the material.
Later studies established what researchers call the self-explanation effect. Across age groups and learning contexts, people who explain ideas to themselves tend to develop stronger understanding and become better able to apply what they have learned beyond the original example.
The benefit does not come from producing polished prose. A hesitant explanation can be productive because the attempt itself requires retrieval and organization. It makes the learner decide which facts matter, how they relate, and whether they genuinely support the conclusion.
How Explanation Changes the Way We See a Problem
Lombrozo describes an experiment involving cards that could be matched in more than one way. Adults tended to notice an abstract relation—for example, pairing a card showing two identical animals with another card that also showed two matching animals. Young children were more likely to focus on a shared object, matching cards because both included the same kind of animal.
When five- and six-year-olds were asked to explain why certain cards belonged together, their choices shifted toward the more abstract pattern favored by adults. They were not given corrective feedback or additional facts. The request for an explanation changed how they organized the information already in front of them.
This finding captures something important about learning. Explanation does more than report a completed thought. It can reshape the thought itself. By looking for a general reason, learners may move beyond conspicuous surface features and notice a deeper structure.
That shift supports transfer. If someone understands only the visible details of one example, the lesson may remain tied to that case. If they identify the underlying relation, they are more likely to recognize it when the objects, language, or setting changes.
The Illusion of Explanatory Depth
People frequently feel that they understand familiar systems until they try to explain how those systems work. A toilet, zipper, bicycle, weather event, or economic mechanism may seem obvious from everyday experience. The confidence often collapses when someone attempts a detailed causal account.
Researchers refer to this mismatch as the illusion of explanatory depth. Familiarity creates a sense of comprehension that exceeds what a person can actually reconstruct. We confuse recognizing something, using it, or having easy access to information about it with understanding it.
Self-explanation punctures that illusion. Each incomplete transition becomes visible: “I know these two events are connected, but I cannot say how.” That moment is not a failure of learning. It is valuable diagnostic evidence.
Lombrozo offers the example of trying to explain the relationship between low atmospheric pressure and a tornado’s funnel. The learner may begin with several relevant ideas but discover that a causal link is missing. Instead of searching broadly for “everything about tornadoes,” they can formulate a narrower question about pressure, airflow, or funnel formation.
The attempted explanation therefore creates an information shopping list. It identifies which pieces are already present and which must still be found.
Turning Information Overload Into Targeted Inquiry
When people feel uncertain, the natural response is often to collect more: more tabs, more videos, more saved papers, and more AI-generated overviews. This can produce the sensation of progress while making the original question harder to see.
Learning by thinking suggests a different sequence:
State what you are trying to understand or accomplish.
Explain the topic using only what you currently know.
Mark assumptions, weak links, and unanswered questions.
Search specifically for the information needed to repair those gaps.
Rebuild the explanation with the new evidence included.
This sequence does not reduce the amount of knowledge in the world. It reduces the portion that is relevant to the learner’s immediate need. A vague desire to “learn about climate” might become a concrete question about how a particular feedback loop operates. A broad search for help with a software bug might become an investigation of one state transition or error condition.
Targeted inquiry also makes it easier to judge new information. When you know which gap a source is supposed to fill, you can ask whether it actually supplies the missing evidence. Without that purpose, almost any related content may appear useful.
What AI Should—and Should Not—Do for the Learner
Lombrozo’s argument becomes especially consequential as AI systems make synthesis effortless. A chatbot can retrieve definitions, propose explanations, compare theories, and present a coherent answer within seconds. These capabilities can support learning, but they can also externalize the mental operations through which understanding develops.
If a tool performs all the selecting, combining, and explaining, the user may receive a good answer without building a usable mental model. The result is another illusion: knowledge feels personally possessed because it is instantly accessible.
The solution is not to reject external tools. It is to use them after—or alongside—active thought. Before requesting an explanation, try to produce one. Ask the system to challenge a causal link, supply evidence for a specific uncertainty, or compare your model with an alternative. After reading the response, close it and reconstruct the argument in your own words.
This approach preserves a productive division of labor. External resources provide facts, perspectives, and evidence that memory cannot supply. The learner remains responsible for integration: deciding how the pieces fit, where confidence is justified, and which questions remain unresolved.
A Better Starting Point for Learning
Learning by thinking is not necessarily easier than looking up an answer. It requires effort, tolerates uncertainty, and sometimes exposes how little we understand. Those are features rather than defects. The effort is directed toward building connections and diagnosing needs, not merely accumulating material.
Lombrozo’s central recommendation is simple: begin within before reaching outward. Retrieve what you know, transform it, combine it, and attempt an explanation. Then let the failures of that explanation guide your search.
In a world of abundant information, strong learners are not simply those who can find the most content. They are those who can determine what they already know, recognize what is missing, and acquire the next piece for a clear reason.
