Tracking Action in Cross-Reality Mark Test Reveals Developing Body Representation Among Toddlers

Research on multisensory integration--particularly visuo-proprioceptive integration--has renewed interest in how body representation develops in early childhood. In the past decade, researchers have attempted to adapt body-illusion tasks such as the rubber hand illusion for young children, yet difficulties with verbal reports and inconsistent behavioral responses limit their usefulness for toddlers. At the same time, children naturally begin to recognize their mirror image as self during the early toddler period, suggesting that mirror-based localization may offer a developmentally appropriate window onto emerging body representation. However, the classic mark test is not suitable for repeated measurements. We developed Bodytoypo, a cross-reality body-part localization task that projects a life-size self-image onto a screen and overlays AR markers on multiple body locations using real-time skeleton detection. Toddlers touched the corresponding location on their actual body, allowing quantitative evaluation of localization accuracy. Thirty 2- and 3-year-olds participated. We recorded first-touch outcomes, touch error, reaction times, and hand trajectories. Conventional coding was combined with an automated statistical analysis to estimate factors predicting touch error. Results showed that Bodytoypo is feasible for repeated use with toddlers and provides reliable indices of body-part localization. Automated statistical analyses identified several reaching strategies that appeared to reflect differences in localization precision. These findings suggest age-related changes in how visual information and motor prediction are weighted during visuo-proprioceptive integration. Bodytoypo offers a promising behavioral index for clarifying how sensorimotor information (body schema) and spatial understanding of body parts (body topology) jointly contribute to the early development of body representation.