Why the best learners talk to themselves | Tania Lombrozo

Host Elise Hu introduces cognitive scientist Tania Lombrozo's argument that having instant access to the internet, AI and social media has not necessarily made learning easier. Instead, the sheer volume of available information can make learning harder by creating “information overwhelm.” Hu says people commonly respond by searching outside themselves for still more information, but that approach can intensify the problem rather than solve it. Lombrozo's proposed alternative is “learning by thinking”: begin with the knowledge already stored in your own memories and skills, and explain ideas to yourself before reaching for a search engine or chatbot. Hu compares this to software engineers' practice of “rubber ducking,” in which programmers talk through code line by line to a rubber duck that cannot answer back. The value is not that the duck provides information; explaining forces the programmer to discover what they understand and where their reasoning breaks down. Hu previews the “illusion of explanatory depth,” the experience of trying to explain something and realizing that one's understanding is much thinner than expected. She presents that moment of self-humbling as useful rather than discouraging: it gives learners a foundation, identifies gaps and helps them decide what information to seek next.

Lombrozo begins by asking the audience questions whose answers many people may not immediately retrieve: how many windows are on the front of their home, what the fifth letter of the alphabet is, and whether Z sits on the left or right side of a typical keyboard. She argues that people often already possess the information needed to answer. Visual memories can be used to count the windows, auditory memories can reconstruct that E is the fifth letter, and motor knowledge from typing can locate Z on the left side of the keyboard. These examples illustrate her central definition of learning by thinking: extracting knowledge from the mind by transforming and combining representations that are already there. She compares the process to cooking. If a recipe requires chocolate chips and salted butter but the kitchen contains only a chocolate bar and unsalted butter, the cook can transform the chocolate into pieces and combine butter with salt. Likewise, information stored as a series of visual memories can be transformed into a single answer, such as the number of windows. Lombrozo adds an important limitation: thinking alone cannot supply information that has never been encountered. If she has never visited someone's home, she cannot reason her way to the number of windows it has. However, asking the question before visiting tells her exactly what to look for. Learning by thinking therefore prepares the learner to acquire new information efficiently.

Lombrozo says there is no universal learning recipe, but a strategy that comes close is seeking explanations. Research beginning in the late 1980s examined why students learn different amounts despite having the same textbooks and passages. The studies found that successful learners often questioned themselves while studying and tried to provide their own answers. They acted as their own rubber ducks. Later research, according to Lombrozo, found that both children and adults generally learn more effectively and apply knowledge more broadly when they explain to themselves. She calls this the “self-explanation effect.” She illustrates the mechanism with a card task. A reference card has two ducks; one comparison card has two frogs and another has one duck and one fox. Adults usually pair the reference card with the two-frog card because they represent both cards as containing matching sets—two of one object—even though the objects differ. Five- and six-year-old children usually choose the duck-and-fox card because it shares an actual duck with the reference card. In one study, researchers showed children examples in which the adult matching pattern was correct, then asked half of them to explain why the examples belonged together. The children received no corrective feedback and no new information, yet those prompted to explain subsequently matched new cards like adults. Lombrozo concludes that explaining changed how they represented the information, enabling a different solution.

Lombrozo next explains why self-explanation is valuable even when it exposes ignorance. She tells the story of her friend Roxanna, whose young son asked how tornadoes start. Roxanna had taken a college class on climate and natural disasters and initially thought she could answer. But while explaining pressure differences, she stalled: she could not say what sets the process off or exactly how low pressure produces a funnel. The attempt left her realizing that she knew far less than she had assumed. Lombrozo identifies this experience as the “illusion of explanatory depth.” People often feel that they understand a process until they are asked to explain it step by step. Rather than treating the discovery as failure, Lombrozo says learners should use it to build an information “shopping list.” By trying to combine what they know into an explanation, they uncover a specific gap—in Roxanna's case, the relationship between low pressure and the formation of a tornado funnel. This process also determines how new information should be interpreted. Once learners know what they are missing, they can ask more precise questions and connect the answers to existing knowledge. Lombrozo's broader point is that learning begins not with indiscriminate information collection but with an internal attempt to explain, because the attempt reveals both the boundaries of current understanding and the information that would actually help.

Lombrozo applies the argument to an environment in which tools for retrieving and synthesizing information are becoming increasingly pervasive. She says people will be tempted to turn immediately to search engines, AI-generated responses and chatbots. She describes these as “incredible tools,” but warns that they should not replace learning by thinking. If people begin with externally generated answers, they may outsource the crucial work of transforming and combining what they already know. Without that internal work, learners can lose track of the boundary between knowledge held in their heads and information they can merely access. Lombrozo says this makes them vulnerable to “illusions of understanding”: they may recognize or retrieve an explanation without being able to reconstruct or use it independently. Self-explanation is her way to counter that risk. It forces learners to answer new questions from existing knowledge, identify where the answer breaks down and then seek information targeted to the gap. She does not claim that learning by thinking makes learning easy or eliminates the need for new information. Instead, it gives effort a direction. Her closing recommendation is direct: before looking outside your head, ask yourself questions and explain the subject in your own words. “Be your own rubber duck,” she says. The result is a path through information overwhelm toward better understanding and more effective use of whatever information comes next.