How is the hypothesis space represented? Evidence from young children's active search and predictions in a...

Area of Science:

• Cognitive Development
• Learning and Memory
• Computational Psychology

Background:

• Navigating uncertainty requires learning cue-outcome relationships.
• Active learning allows direct hypothesis testing.
• Understanding children's hypothesis space representations is key to active learning.

Purpose of the Study:

• To investigate how 5- to 7-year-olds actively learn cue-outcome relationships.
• To determine how children represent the hypothesis space during active learning.
• To compare computational models of hypothesis space representation in children.

Main Methods:

• Children selected monster pairs to observe their relative speed, learning cue-outcome (color/shape-speed) relationships.
• Two computational models were compared: cue-abstraction and permutation-based.
• Studies involved scaffolding and memory aids to assess model fit with children's behavior.

Main Results:

• Study 1: 5- and 6-year-olds used cue-abstraction representations with scaffolding.
• Study 2: Younger children favored permutation-based models; 7-year-olds used cue-abstraction with memory aids.
• Results indicate a developmental shift in hypothesis space representation.

Conclusions:

• The hypothesis space guides active search and learning.
• Active learning and hypothesis generation may trigger different representations.
• Children show a developmental progression in representing the hypothesis space.