.png){kind=icon}
.png){kind=icon}
Return
Action chunking as conditional policy compression
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cognition.2025.106201.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Action chunking emerges as a trade-off between reward and cognitive effort. • People use temporal structure to reduce policy complexity and respond faster. • Higher working memory load leads to more chunking to conserve cognitive resources. • Chunking frees up memory for other, unchunked information.
Keywords:
action chunking
cognitive effort
temporal structure
working memory
policy complexity