I am interested in how children explore the world and make decisions.
It’s about the Hunt, not the Treasure: Children are more explorative on explore-exploit tasksImagine that it’s your second day at a new job. You are standing at the coffee cart outside your office building, considering the unfamiliar menu. Yesterday you had a cappuccino and enjoyed it; today you must decide whether to get the cappuccino again like yesterday, or try the matcha green tea latte, which you might not like. This is known as an explore/exploit problem, because you must choose between exploiting a familiar option (the cappuccino) or exploiting a new one (the matcha latte). We used two different explore-exploit paradigms with children (see photos below). We found that children didn’t play in a way that maximized payout and explored more than would be optimal. However, children correctly identified the machines with the highest level of payout and overwhelmingly preferred this bandit. We found that adults played in a way that maximized payout when they were told to maximize payout or were given no explicit instructions. However, when adults were asked to maximize learning, they explore more -- like children.
Collaborators: Barbara Sarnecka, Mark Steyvers, & Celeste Kidd
Digital Measures of Risk Propensity and Executive Function for ChildrenLife is full of decisions about when to play it safe and when to take risks. These decisions can be as mundane as deciding whether to try a new cereal or as paramount as choosing between alternative treatments for a life-threatening disease. In adults, risk-taking behavior is associated with many undesirable behaviors such as substance abuse, gambling etc. However, little is known about the development of an individual’s propensity for risk. Specifically, there are no standardized laboratory procedures that assess young children’s risk-taking preferences. In this line of work, we created and validated a measure of risk propensity. The Child Risk Utility Measure (CRUM) is a touchscreen application which enables us to assess individual differences in young children’s risk propensity. The CRUM respects the cognitive limitations of three- to six-year-old children. In this task, children try to help Cookie Monster take cookies from a cookie jar without Oscar waking up. However, the chances of Oscar waking up increase along with the number of cookies on the plate. We used this, along with other child-friendly measures of executive function to investigate the relationship of risk propensity, working memory, and inhibitory control.
Collaborators: Susanne Jaeggi, Barbara Sarnecka, Ryan Stokes, & Percy Mistry
Shima: A Virtual Reality Measure of Risk PropensityThe greatest challenge of developing a measure of risk propensity is creating a game that feels real. In this line of work, we designed Shima: a virtual reality measure of risk propensity. The game Shima was developed as part of the 2017 Virtual Reality BrainJam Hackathon during the 2017 Games for Change Festival. In this game, you are a photographer on an island inhabited by new species of animals. Your goal is to get as close as you can to each animal and take their picture. The closer you get, the more points you get. But if you get too close, the animal gets scared and runs away. We are currently collecting data to validate this measure.
Developed with: Grace Lin, Roldan Melcon, Matt Cooper, Grace Lin, Helena Kent, Angel Lopez, Armando Somoza, Russell Cohen Hoffing, Aaron Seitz, and Susanne Jaeggi.
Toddlers Always Say the Last Word: Recency drives verbal question answering during early childhoodThe learning benefits of a greater working memory are well known. However, limited working memory could also offer previously unappreciated advantages. Specifically, working memory limitations could be useful for young language learners, allowing them to engage in some of their first verbal communications without yet entirely understanding the meaning of the words they are speaking. In some cases, such as in response to questions containing fixed choices, children’s ability to parrot back chunks of linguistic material could be mistaken for a deeper understanding of a words’ semantic content. We found that (1) toddlers demonstrate a robust verbal recency bias when responding to questions that present choices, (2) this bias exists outside of the laboratory setting, and (3) the bias is likely due to working memory constraints in young children, as evidenced by the fact that it can be created in older children by increasing the working memory demands of the task. These findings shed light on both how we evaluate young children's lexical knowledge and the link between response biases and working memory.
Collaborators: Celeste Kidd, Noah Goodman, Erika DeAngelis, & Mara Hyatt
Sumner, E., DeAngelis, E., Hyatt, M., Goodman, N., & Kidd, C. (2015) Toddlers Always Get the Last Word: Recency biases in early verbal behavior. Proceedings of the 37th Annual Meeting of the Cognitive Science Society. [PDF]
Sumner, E., Stokes, R., Mistry, PK., Jaeggi, S., & Sarnecka, BW. A freshly baked perspective on how we measure risk propensity. [In Prep, email for preprint]
Sumner, E., DeAngelis, E., Hyatt, M., Goodman, N., & Kidd, C. Toddlers Always Get the Last Word: Recency drives children's question answering. [Under review, email for preprint]
Sumner, E., Lomeli, A., Lee, M., & Sarnecka, BW. Too risky for you, but not for me: individual differences in preschooler's decision-making strategies [in prep]
Sumner, E., Harder, E., & Jaeggi, S. Delay Discounting: A measure of risk propensity or working memory? [in prep]
Sumner, E. (1999) My MRI. Published by Mass General Hospital for Children, over 10,000 copies distributed.
I grew up outside of Boston, MA. I completed my undergraduate education in Brain and Cognitive Sciences at the University of Rochester. I am a strong supporter of open science. Other than science, I enjoy running, hiking, baking, and folk music.