Re-assessing the Evidence for MR Abilities in Children Using Computational Models

Area of Science:

• Cognitive Development
• Developmental Psychology
• Computational Neuroscience

Background:

• Evidence for adult mental rotation (MR) abilities is robust, but research in children relies on limited, adapted paradigms.
• Recent computational models of object recognition development offer a new lens to examine MR in early childhood.

Purpose of the Study:

• To re-evaluate the evidence for mental rotation (MR) abilities in children using computational models.
• To determine if non-MR object recognition strategies can explain children's performance on established MR tasks.

Main Methods:

• Utilized computational models simulating infant object representation acquisition through embodied interaction.
• Assessed two object recognition strategies against results from three classical MR tasks in children aged 6 months to 5 years.
• Modeled children's behavior using pixel-wise comparison and expectation-based rotation transformation models.

Main Results:

• A simple pixel-wise comparison strategy adequately modeled behavior in a common MR task for young children.
• Computational models demonstrated that MR may not be necessary for explaining results in children under 5 years.
• The proposed recognition strategies failed to explain performance when MR tasks were adapted to be more adult-like.

Conclusions:

• The findings challenge the assumption of early developing genuine mental rotation (MR) abilities in children.
• A simple recognition strategy, not MR, may underlie performance in classic MR tasks for young children.
• Further research is needed to understand the developmental trajectory of true mental rotation capabilities.