Modeling Infant Attachment & Exploration Behaviors Using Reinforcement Learning

How does caregiver-infant relationship impact infant locomotion exploration? Prior behavioral experiments have examined infant locomotion on its own. However, in real life, infants rarely navigate novel environments alone; therefore, infant exploration behaviors cannot be fully understood without accounting for the presence of caregivers in the environment. Prior research on Attachment Theory (AT) has highlighted the significant impact caregiver-infant relationship has on infant exploration behaviors. AT claims that different caregiver-infant relationships lead to distinct patterns of exploration in infants. Specifically, while infants with secure relationships with caregivers explore the most and the most freely, infants with insecure attachment relationships demonstrate constrained exploration. Prior computational work has used Reinforcement Learning (RL) to model infant exploration alone. More recent study has looked at infant learning through interactions with their caregiver in the environment, yet without examining how different types of infant-caregiver relationships could generate various kinds of exploration and play behaviors. The present study uses RL to train an infant agent in an environment containing toys and one of three different caregivers: loving and responsive, punishing and responsive, and loving and sometimes responsive. AT has characterized how these kinds of caregivers lead to secure and insecure caregiver-infant relationships, or attachment styles. After training, the infant agents display emergent exploration behaviors in alignment with all three expected attachment styles, but do not exhibit the kind of exploration characteristic of real infants. This work suggests that: 1) infant attachment styles can be computationally acquired from previous interaction history with the caregiver while maximizing rewards from the environment; 2) real world infants may explore the environment with additional intrinsic reward signals beyond attachment; and 3) caregiver-infant relationship modulates infant locomotion exploration and learning.

Publication Details

Journal
Stanford Digital Repository
Published
2024-08-21
DOI
https://doi.org/10.25740/wf578dv1281
Primary Topic
Child and Animal Learning Development
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Modeling Infant Attachment & Exploration Behaviors Using Reinforcement Learning

Stanford Digital Repository
Child and Animal Learning Development
article

Modeling Infant Attachment & Exploration Behaviors Using Reinforcement Learning

article en

Abstract

How does caregiver-infant relationship impact infant locomotion exploration? Prior behavioral experiments have examined infant locomotion on its own. However, in real life, infants rarely navigate novel environments alone; therefore, infant exploration behaviors cannot be fully understood without accounting for the presence of caregivers in the environment. Prior research on Attachment Theory (AT) has highlighted the significant impact caregiver-infant relationship has on infant exploration behaviors. AT claims that different caregiver-infant relationships lead to distinct patterns of exploration in infants. Specifically, while infants with secure relationships with caregivers explore the most and the most freely, infants with insecure attachment relationships demonstrate constrained exploration. Prior computational work has used Reinforcement Learning (RL) to model infant exploration alone. More recent study has looked at infant learning through interactions with their caregiver in the environment, yet without examining how different types of infant-caregiver relationships could generate various kinds of exploration and play behaviors. The present study uses RL to train an infant agent in an environment containing toys and one of three different caregivers: loving and responsive, punishing and responsive, and loving and sometimes responsive. AT has characterized how these kinds of caregivers lead to secure and insecure caregiver-infant relationships, or attachment styles. After training, the infant agents display emergent exploration behaviors in alignment with all three expected attachment styles, but do not exhibit the kind of exploration characteristic of real infants. This work suggests that: 1) infant attachment styles can be computationally acquired from previous interaction history with the caregiver while maximizing rewards from the environment; 2) real world infants may explore the environment with additional intrinsic reward signals beyond attachment; and 3) caregiver-infant relationship modulates infant locomotion exploration and learning.

Stanford Digital Repository
Openalex Percentile: Top 55%
Child and Animal Learning Development
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Modeling Infant Attachment & Exploration Behaviors Using Reinforcement Learning · Stanford Digital Repository (2024) | TGRS Research Map | TGRS