About

Elizabeth Bonawitz is the David J. Vitale associate professor of Learning Sciences at Harvard University. Her work focuses on the basic science theories of learning with the broader goal of informing educational practice. Her research bridges two research traditions: cognitive development and computational modeling. Specifically, Bonawitz’s empirical approach concentrates on the structure of children's early causal beliefs, how evidence and prior beliefs interact to affect children's learning, the developmental processes that influence children's belief revision and curiosity, and the role of social factors such as learning from others in guiding learning. Bonawitz received her Ph.D. from MIT in Brain and Cognitive Sciences in 2009, working with Dr. Laura Schulz. She completed a post-doctoral fellowship at the University of California, Berkeley with Thomas Griffiths and Alison Gopnik from 2009 to 2013. She has served as an Assistant and Associate Professor of Psychology at Rutgers University, Newark, from 2013 until 2020, when she moved to Harvard. Her research is funded by several NSF grants, the Caplan Foundation, and the Templeton Foundation, and her work has been published in top journals in psychology, cognitive science, and education. Bonawitz has also served as Associate Editor for the journal Open Mind and is on the governing board of the Cognitive Development Society and Children Helping Science. Her current sponsored projects include studies on cognitive mechanisms of guided instruction in early elementary years, focusing on how different teaching styles impact children's understanding of core STEM concepts, and research on causal reasoning about contagious illness in young children, aiming to develop tools for better science reasoning and decision-making. Additionally, she is involved in the Creating Impact Science Program, supporting high-risk, high-reward research projects in the science of learning with educational applications.

Research topics

• Computer Science
• Artificial Intelligence
• Developmental psychology
• Psychology
• Cognitive psychology
• Machine Learning
• Computer Security
• Data science
• Neuroscience
• Social psychology
• Cognitive science
• Epistemology
• Software engineering
• Business
• World Wide Web
• Risk analysis (engineering)

Selected publications

• Daily Dynamics of Children's Science Learning: Associations With Fluctuations in Attention and Family Stress

  Mind Brain and Education · 2026-02-01

  articleOpen access

  Abstract Though children's learning occurs through daily experiences, few studies have examined how variability in those experiences predict day‐to‐day learning. Here, we explored associations among learning, attention, and parental stress within and across children. Children aged 5–7 ( N = 103, M age = 6.3 years) participated in eight virtual science lessons over 2 weeks ( N = 813 lessons). Within‐child, greater attention during lessons predicted greater learning ( d within = .23). Within‐child, parent stress was not associated with attention or learning, but across children, greater parent stress was associated with lower attention ( d between = −.63). Children's attention ( d within = −.21) and science learning ( d within = −.24) declined across lessons. Declines in learning were stronger for children whose parents reported higher levels of daily stress ( d = −.27). These results suggest that differences in children's science learning may emerge from differences in attention in the moment and variability in learning trajectories over time.

  PublisherOA PDFDOI

• Parent-child interaction styles relate to preschooler’s causal play, learning, and generalization

  Early Childhood Research Quarterly · 2026-01-01

  articleOpen accessSenior author

  • This study investigated the effects of a brief instructional intervention on parent-child interactions and children’s learning. • Parents were assigned to one of three conditions: teaching (teach the child), helping (help the child), or no-involvement (watch the child). • Parents in the no-involvement condition asked fewer questions and provided less feedback and encouragement than those in the other two conditions. • Parents’ use of questioning and guidance predicted better learning outcomes for children not only during the play but also in the subsequent independent exploration. • The findings suggest that guidance (in particular in the form of pedagogical questions) supports “learning to learn”. Parental involvement plays an important role in children’s learning within everyday social contexts. This study investigated whether and how a brief instructional intervention affected parent-child interactions during play, and how parents’ and children’s behaviors impacted children’s learning during the play and subsequent independent exploration. Parents and their 3- to 6-year-old children ( N = 72) engaged with causal toys, with parents pseudo-randomly assigned to one of three conditions: teaching (teach the child), helping (help the child), or no-involvement (watch the child). During the Interactive Phase, parent-child dyads had four minutes to explore the toys and figure out the causal rule together. This was followed by the Independent Phase, in which children explored a different set of toys with a new causal rule on their own for three minutes. Children's learning outcomes were assessed after each phase. Results indicated that parents in the no-involvement condition asked fewer questions and provided less feedback and encouragement compared to those in the teaching and helping conditions. However, there were no significant differences in parental behaviors between the helping and teaching groups. In terms of learning outcomes, children in the teaching condition had a higher success rate than those in the no-involvement condition. Across all conditions, parents’ use of questioning (particularly pedagogical questions) and guidance in helping children generate evidence during the interaction predicted better learning outcomes for children. These improved learning outcomes were even observed following children’s independent exploration, suggesting that guidance (particularly in the form of pedagogical questions) supports “learning to learn”.

  PublisherDOI

• Neural responses to state curiosity in young children

  Developmental Cognitive Neuroscience · 2026-01-28

  articleOpen access

  Curiosity scaffolds children's exploration and learning. Yet, the neural mechanisms of curiosity-modulated learning in children remain unclear. Here, we designed an fMRI task to test how curiosity, as defined by children's self-reported excitement about learning information, modulates memory and neural activity in 5- to 8-year-olds (n = 60 with behavioral data, n = 51 with fMRI). We observed greater learning when children reported more curiosity. In whole-brain analyses, high-curiosity was associated with greater activation in inferior frontal gyrus, lateral occipital cortex, the thalamus, and the putamen. Curiosity did not modulate activation in preregistered regions of interest (dorsal attention network, hippocampus, nucleus accumbens) but did modulate activation in an exploratory region of interest, the amygdala. Multivariate searchlight decoding revealed local activity patterns that reliably distinguished reported curiosity levels in dorsolateral prefrontal cortex, fusiform gyrus, angular gyrus, precuneus, and cerebellum. Together, these findings are consistent with prior work on curiosity-related activation during information receipt in adults, suggesting that neural systems that support curiosity-driven learning are already engaged in early childhood.

  PublisherDOI

• The development of the “first thing that comes to mind”.

  Developmental Psychology · 2025-06-05

  articleSenior author

  = 189, U.S. English speakers, recruited online) produce both "first-to-mind" judgments and predictions about random samples. In Experiment 1, providing information about whether being longer or shorter made a fictional tool better or worse led adults to provide first-to-mind judgments that were biased toward the prescriptive ideal, but unbiased random sample predictions. However, 6-9-year-old children provided judgments that were biased by the prescriptive ideal in both cases. In Experiment 2, with 6-9-year-olds and adults, we manipulated whether the prescriptive information focused exclusively on positive (i.e., only "better") or negative (i.e., only "worse") properties. In the positive-focus condition, all age groups showed an effect of prescriptive ideal on first-to-mind judgments, but only 6-7-year-olds showed an effect of prescriptive ideal on random sample predictions. However, in the negative-focus condition, there was no effect of prescriptive information on either type of judgments for any age group, including adults. We discuss what changes in development in the ability to represent different kinds of information and apply the best kind of information to a specific task. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

  PublisherDOI

• Children’s sensitivity to automatic behavior relates to pedagogical reasoning and Theory of Mind

  2025-03-14

  preprintOpen accessSenior author

  Pedagogy is a powerful way to learn about the world, and young children are adept at both learning from teaching and teaching others themselves. Theoretical accounts of pedagogical reasoning suggest that an important aspect of being an effective teacher is considering what learners need to know, as misconceptions about learners’ beliefs, needs, or goals can result in less helpful teaching. One underexplored way in which teachers may fail to represent what learners know is by simply “going through the motions” of teaching, without actively engaging with the learner’s beliefs, needs, and goals at all. In the current paper, we replicate ongoing work that suggests children are sensitive to when others are relying on automatic scripts in the context of teaching. We then look at the potential link to two related measures. First, we hypothesize that sensitivity to a teacher’s perceived automaticity will be linked to classic measures of pedagogical sensitivity and learning – specifically, how children explore and learn about novel toys following pedagogical versus non-pedagogical demonstrations. Second, we hypothesize that the development of Theory of Mind (and age differences more broadly) relate to these pedagogical sensitivities. Our online adaptation of the novel toy exploration task did not invoke pedagogical reasoning as expected, and so we do not find robust links between these tasks. We do find that Theory of Mind predicts children’s ability to detect automaticity in teaching when controlling for age. This work thus highlights the connections between sensitivity to teaching and reasoning about others’ knowledge, with implications for the factors that support children’s ability to teach others.

  PublisherOA PDFDOI

• Parent-Child Interaction Styles Relate to Preschooler’s Causal Play, Learning, and Generalization

  2025-03-05

  preprintOpen accessSenior author

  Parental involvement plays an important role in children’s learning within everyday social contexts. This study investigated whether and how a brief instructional intervention affected parent-child interactions during play, and how parents’ and children’s behaviors impacted children’s learning during the play and subsequent independent exploration. Parents and their 3- to 6-year-old children (N = 72) engaged with causal toys, with parents assigned to one of three conditions: Teaching (teach the child), Helping (help the child), or No-involvement (watch the child). During the Interactive Phase, parent-child dyads had four minutes to explore the toys and figure out the causal rule together. This was followed by the Independent Phase, in which children explored a different set of toys with a new causal rule on their own for three minutes. Children's learning outcomes were assessed after each phase. Results indicated that parents in the No-involvement condition asked fewer questions and provided less feedback and encouragement compared to those in the Teaching and Helping conditions. However, there were no significant differences in parental behaviors between the Helping and Teaching groups. Across all conditions, parents’ use of questioning (particularly pedagogical questions) and guidance in helping children generate evidence during the interaction predicted better learning outcomes for children. These improved learning outcomes were even observed following children’s independent exploration, suggesting that guidance (in particular in the form of pedagogical questions) supports “learning to learn”.

  PublisherOA PDFDOI

• Children's sensitivity to automatic behavior relates to pedagogical reasoning and Theory of Mind

  Frontiers in Developmental Psychology · 2025-04-24 · 2 citations

  articleOpen accessSenior author

  Pedagogy is a powerful way to learn about the world, and young children are adept at both learning from teaching and teaching others themselves. Theoretical accounts of pedagogical reasoning suggest that an important aspect of being an effective teacher is considering what learners need to know, as misconceptions about learners' beliefs, needs, or goals can result in less helpful teaching. One underexplored way in which teachers may fail to represent what learners know is by simply “going through the motions” of teaching, without actively engaging with the learner's beliefs, needs, and goals at all. In the current paper, we replicate ongoing work that suggests children are sensitive to when others are relying on automatic scripts in the context of teaching. We then look at the potential link to two related measures. First, we hypothesize that sensitivity to a teacher's perceived automaticity will be linked to classic measures of pedagogical sensitivity and learning—specifically, how children explore and learn about novel toys following pedagogical vs. non-pedagogical demonstrations. Second, we hypothesize that the development of Theory of Mind (ToM) (and age differences more broadly) relate to these pedagogical sensitivities. Our online adaptation of the novel toy exploration task did not invoke pedagogical reasoning as expected, and so we do not find robust links between these tasks. We do find that ToM predicts children's ability to detect automaticity in teaching when controlling for age. This work thus highlights the connections between sensitivity to teaching and reasoning about others' knowledge, with implications for the factors that support children's ability to teach others.

  PublisherOA PDFDOI

• The Mind Lab: Thought Experiments as a Means to Teaching Science Effectively and Efficiently

  2025-08-14

  preprintOpen accessSenior author

  Western educational philosophy has long emphasized the importance of first-hand observation in learning. This view, however, overlooks another important form of learning, sometimes called learning by thinking. Here, we compared three conditions: instruction using thought experiments (Thought Experiments condition), instruction using real experiments (Real Experiments condition), and no instruction (Baseline condition). A total of 100 children (MAge = 84.79 months; 43 girls; predominantly White) were assigned to the conditions. All participants completed a pre-training physics assessment, received physics instruction (Thought and Real Experiments conditions only), and completed a post-training assessment. Results showed that only children in the Real and Thought Experiments conditions improved from pre- to post-training. Critically, there was no significant difference between the two instructional methods.

  PublisherDOI

• The Development of Sensitivity to Automatic Behavior

  2025-11-14

  articleOpen access

  People’s behavior can be roughly categorized into two modes: either reflective and thoughtful, or automatic and rote. Past work on Theory of Mind has focused on the first category. But do children notice when people are acting in an automatic way? This paper examined five- to ten-year-old children’s reasoning about others’ rote behavior, focusing on the consequences of this inference in teaching contexts (N = 660 across four studies, 327 girls). Children’s sensitivity to rote behavior increased with development, with consistent competence emerging around age 7. Rote behavior was also associated with worse teaching. These results indicate when and how reasoning about automatic behavior matters to children’s perception of others, and suggest novel extensions to models of Theory of Mind.

  PublisherDOI

• The Development of Sensitivity to Automatic Behavior

  2025-10-03

  articleOpen access

  People’s behavior can be roughly categorized into two modes: either reflective and thoughtful, or automatic and rote. Past work on Theory of Mind has focused on the first category. But do children notice when people are acting in an automatic way? This paper examined five- to ten-year-old children’s reasoning about others’ rote behavior, focusing on the consequences of this inference in teaching contexts (N = 660 across four studies, 327 girls). Children’s sensitivity to rote behavior increased with development, with consistent competence emerging around age 7. Rote behavior was also associated with worse teaching. These results indicate when and how reasoning about automatic behavior matters to children’s perception of others, and suggest novel extensions to models of Theory of Mind.

  PublisherDOI

Recent grants

• EAGER: MAKER: The origins of making: A Data Science Approach to Investigating Cognitive and Affective Basis of Learning through Constructing

  NSF · $298k · 2016–2019

• Choosing to learn: Investigating the factors that drive preschoolers' exploration

  NSF · $651k · 2016–2020

• A Computational Framework for Assessing the Integration of Category Knowledge in Episodic Memory in Early Childhood

  NSF · $165k · 2019–2021

Frequent coauthors

• Patrick Shafto

  57 shared

• Alison Gopnik

  University of California, Berkeley

  47 shared

• Laura Schulz

  36 shared

• Ilona Bass

  Harvard University Press

  34 shared

• Yue Yu

  Eastern Liaoning University

  20 shared

• Kimele Persaud

  Rutgers, The State University of New Jersey

  20 shared

• Stephanie Denison

  University of Waterloo

  19 shared

• Thomas L. Griffiths

  19 shared

Labs

• Computational Cognitive Development LabPI

Education

• PhD, Brain and Cognitive Sciences

  MIT

  2009

Awards & honors

• James McDonnell Foundation Understanding Human Cognition Sch…
• Jacobs Early Career Research Fellowship