This is the research companion to Curiosity as Brain State. Read the main post first.
The main post made one claim: curiosity is a brain state, not a personality trait, and the state is engineerable. That claim rests on more than the two studies the post cites. Behind the single fMRI scan sits a body of work asking the prior questions. What is curiosity, mechanically? Why does the brain treat not-knowing as a reward worth paying for? Can adults reliably switch the state on in a child, and does it survive contact with a real classroom?
These five papers answer those questions in order. One reframes curiosity as a deprivation state triggered by a perceived gap in what you know. One pulls the gap into reward circuitry and shows the brain pays real value for information. One builds the tool to measure curiosity-promoting teaching and finds it almost absent from ordinary lessons. One trains five-to-seven-year-olds to ask questions and watches what shifts. One pools forty-one trials to ask the blunt question -- can curiosity be raised on purpose? Read together, they move the main claim from striking finding to mechanism with a method.
The Gap That Makes You Itch
Loewenstein, G. (1994). The psychology of curiosity: A review and reinterpretation. Psychological Bulletin, 116(1), 75-98.
10.1037/0033-2909.116.1.75
What they found: This is the theory paper the rest of the field builds on. Loewenstein reviews two prior waves of curiosity research -- the 1960s drive-and-arousal accounts and the 1970s-1980s scale-building effort -- and argues both failed. The scale work, in particular, produced inconsistent correlations and overlap with neighboring traits like sensation seeking and need for cognition. In their place he offers a reinterpretation: curiosity is a cognitively induced deprivation state that arises when a person notices a gap between what they know and what they want to know. The larger and more salient the gap, within a tractable range, the more intense the curiosity. The framing explains why curiosity is topic-specific rather than general, why people voluntarily court it when they expect the gap to close, why intensity tracks proximity to the answer, and why satisfaction often disappoints. One counterintuitive prediction: curiosity rises with expertise, because the more you know in a domain, the sharper you feel what is still missing.
Why this matters for you: This is the engineering spec for the curiosity state. If curiosity is a felt gap, then the lever is the gap, not the child. You do not produce curiosity by telling a kid to be curious. You produce it by building enough base knowledge that a gap becomes visible, then leaving the gap open long enough to be felt. Loewenstein is explicit that simply encouraging students to ask questions, without first building the scaffold that makes gaps salient, is not enough. The parent move is to half-tell -- enough to make the missing piece itch, not so much that nothing is missing.
What it doesn’t answer: This is a theoretical reinterpretation, not an experiment. Loewenstein offers no new data and frames the gap account as a hypothesis-generator, not a proven model. The model assumes a learner who can represent what they do not know against a reference point, and its reach to very young children, or to children under stress or threat, is not worked out in the paper. It also predates the brain imaging that later tested it. Whether perceived gap can be measured cleanly in real time, and how stress changes the threshold at which a gap reads as curiosity rather than threat, are left open.
The Brain Pays for Information
Kidd, C., & Hayden, B. Y. (2015). The psychology and neuroscience of curiosity. Neuron, 88(3), 449-460.
10.1016/j.neuron.2015.09.010
What they found: Kidd and Hayden synthesize the psychology, computation, and neuroscience of curiosity, organized around the function, evolution, mechanism, and development of information-seeking. The central move is to treat curiosity as a drive for information that carries measurable value. In one monkey paradigm, the value of advance information about an upcoming reward was estimated at roughly 25% of the value of the reward itself, and that value scaled with stakes. Curiosity in trivia tasks showed a U-shape against confidence -- lowest when you have no clue or are certain, highest at intermediate confidence -- matching the information-gap account. Midbrain dopamine neurons and lateral habenula neurons signal information prediction errors, treating information like reward. And in infants, look-away probability is U-shaped against stimulus complexity -- attention peaks at intermediate information content, the pattern they call the Goldilocks effect, holding across visual and auditory displays and within individual infants.
Why this matters for you: This is the bridge from the gap to the reward system. The 25% figure makes the abstract concrete: the brain will spend to find out, the way it spends for food or money. And the U-shapes, from infants to adults, give you the calibration rule the main post implies but does not state. Too easy is boring; too hard is noise. The sweet spot is intermediate -- a gap big enough to feel, small enough to look closable. The Goldilocks finding means this is not something you teach a child to do. The system arrives pre-tuned to seek intermediate uncertainty. Your job is to keep supplying it.
What it doesn’t answer: This is a perspective article, not a primary study, so its conclusions inherit the limits of the work it cites. There is no single participant sample. The cross-species framing, from worms to humans, is suggestive rather than settled -- curiosity may not be one unitary thing across taxa, ages, and contexts, and the authors deliberately resist a firm taxonomy. Neural activations tied to curiosity may partly reflect salience or arousal rather than curiosity specifically. The paper itself notes that the strongest causal traction comes from studies that manipulate curiosity directly, not from the correlational imaging that dominates the field.
Counting Curiosity in the Classroom -- and Finding Almost None
Jirout, J. J., Zumbrunn, S., Evans, N. S., & Vitiello, V. E. (2022). Development and testing of the Curiosity in Classrooms Framework and Coding Protocol. Frontiers in Psychology, 13, 875161.
10.3389/fpsyg.2022.875161
What they found: Jirout and colleagues built an observational protocol to code curiosity-promoting and curiosity-suppressing instructional moves from classroom video, then tested it on 35 video-recorded fifth-grade math lessons from a national teaching database. The framework names eight specific practices. The feasibility test found these practices strikingly rare. Practices that promote feelings of curiosity averaged about 2 instances per lesson; practices that promote curious behaviors averaged under one. Modeling comfort with uncertainty was observed in only 9% of teachers. Opportunities to explore and figure things out appeared in only 6%. Most striking, prompting students to generate their own questions was observed zero times across all 35 lessons. Of the segments with usable audio, over half had zero student questions, and many of the questions that did occur were permission or clarification requests rather than curiosity-driven. Curiosity suppression was even rarer, but so was almost everything that would switch the state on.
Why this matters for you: This is the answer to a question the main post raises but cannot prove on its own -- why curiosity seems to fade through school. It is not that teachers actively crush it. It is that the moves that produce the state are simply absent. Zero question-prompting across 35 lessons is not a low number; it is a floor. For a parent, the practical read is that the curiosity state is not the default condition of formal learning -- it has to be deliberately installed, and most settings do not install it. The home becomes the place where wondering aloud and following questions can actually happen, because the dominant instructional environment leaves that work undone.
What it doesn’t answer: This is a tool-development paper, not an efficacy trial. The feasibility test is descriptive -- 35 lessons, a single grade, a single subject, a single topic, fifth-grade math on adding and subtracting fractions. The results should not be generalized to other grades, subjects, or to science, art, and play-based settings where uncertainty may be more naturally embedded. Coding was by consensus rather than independent inter-rater reliability. There was no direct measure of student curiosity, and the analysis linking teacher language to student questioning was exploratory, underpowered, and not statistically significant. The paper does not test whether these practices actually raise student curiosity or learning.
Teaching the Question, Not the Answer
Park, A. T., Colantonio, J., Delgado Reyes, L., Sharp, S. D. S., Koepp, A. E., Bonawitz, E., & Mackey, A. P. (2026). Question asking practice fosters aspects of curiosity in science content in young children. npj Science of Learning, 11, 2.
10.1038/s41539-025-00384-5
What they found: A preregistered randomized trial with 103 children aged 5 to 7. Across eight one-on-one science lessons over two weeks, children were randomly assigned either to practice asking questions or to practice listening carefully. The primary result: children in the question-asking condition showed significantly higher willingness-to-pay for new science information than the listening group (question-asking median 7 versus listening median 5; p = .002, surviving correction for multiple comparisons; Wilcoxon r = 0.23). This was the one primary outcome that held up after correction. The effect was larger for children who started with less science knowledge -- the kids who came in knowing less showed the biggest boost in valuing new information. There was no overall effect on science learning itself, though question-asking appeared to partly equalize outcomes across vocabulary levels. Within the question-asking group, relevant questions rose from about 5.5 to 7.4 per lesson from first to last.
Why this matters for you: This is the closest thing to a clean test of a single parent-usable move. Asking questions is a trainable behavior, and training it shifts how much a child values finding things out -- not just what they happen to know. The mechanism the authors propose maps directly onto Loewenstein: asking a question forces a child to survey what they know and notice what they do not, which sharpens the felt gap. And the equity angle matters at home as much as in school -- the children who benefited most were the ones who started behind. Modeling and inviting questions is not a nicety. It is the lever that most moves the kids with the least to start.
What it doesn’t answer: The effect was real but small -- a Wilcoxon r of 0.23 on one of five outcomes after a two-week intervention is a signal, not a transformation. The sample was higher-SES, predominantly White, recruited through Facebook for a virtual science camp, which likely preselected for science interest and limits how far the result travels. Delivery was one-on-one by trained researchers and remote during the pandemic; scaling to a classroom or a kitchen table introduces noise the study did not face. Several outcomes were null, including prompted question generation and persistence, and the listening comparison may not be a neutral baseline -- it could have suppressed curiosity rather than the question condition raising it. Whether this works below age 5 or in non-science domains is untested.
Can You Raise Curiosity on Purpose? The Pooled Verdict
Schutte, N. S., & Malouff, J. M. (2023). A meta-analytic investigation of the impact of curiosity-enhancing interventions. Current Psychology, 42, 20374-20384.
10.1007/s12144-022-03107-w
What they found: A random-effects meta-analysis of 41 randomized controlled trials, with a combined 4,496 participants, asking whether curiosity can be systematically increased. The pooled effect was moderate -- Hedges’ g = 0.57 (95% CI 0.44 to 0.70). The funnel plot showed small-study bias, and after a trim-and-fill adjustment that removed five studies the effect fell to g = 0.49, still roughly moderate. The effect was stable across measurement mode: self-report measures (g = 0.54) and behavioral measures (g = 0.54) produced nearly identical results, which argues against the worry that the whole thing is a reporting artifact. Among intervention types, those built on mystery (g = 0.60) and game play (g = 0.54) performed well; mindfulness-based and autonomy-based designs produced weaker, non-significant subgroup estimates, though those cells were small. The samples skewed adult -- 36 adult samples against only 4 child samples.
Why this matters for you: This is the durability check on the main post’s central claim. It is one thing to show curiosity is a state in a single scan, or that one classroom trial moved it. It is another to find that across dozens of independent trials, structured attempts to raise curiosity reliably work. The mechanism that emerges -- mystery and games outperforming everything -- is the gap account again, made operational. Mystery is a staged information gap. A game is a puzzle wrapped in stakes. The parent translation is not abstract: build a little mystery, withhold a little, turn the question into a game. The effect is moderate and real, not magic. But the direction is settled -- curiosity responds to design.
What it doesn’t answer: The most important caveat for parents is the thinness of the child evidence. Only 4 of the 41 trials used child samples, and that pooled child estimate (g = 0.33) had a confidence interval crossing zero -- meaning the strong overall result rests heavily on adults. The mystery and game-play findings, while promising, come from an overall moderator test that was not statistically significant, so the ranking among intervention types should be held loosely. Small-study bias was detected and adjusted for. Most primary studies came from education, workplace, or lab settings with relatively typical populations; effects under high stress, scarcity, or trauma are not directly tested. And the meta-analysis measures whether interventions raise curiosity, not whether that curiosity then improves learning, creativity, or wellbeing -- that downstream link is assumed, not demonstrated here.
Coming Up
Next up: The Engineering Failure. We take the most predictable developmental conflict in any house with a teenager -- the war over sleep -- and reframe it as a structural problem, not a discipline problem. The adolescent body reschedules itself, the school day does not, and the gap between them was engineered by adults. We will lay out what the biology actually says and why willpower was never the variable.
Every Functional Child Development post comes with a Dive Deeper companion. Subscribe for the full picture.