How would you like to explore?
This site shows how babies see, hear and feel the world at each age, and the research on computer models that learn from baby-like senses. Pick the view that fits you. You can switch at any time from the top of the page.
See the world the way a baby does
A baby's senses start out blurry and muffled and sharpen month by month. Pick an age, then look and listen.
Built from published research on groups of babies. It cannot tell you how a particular baby is doing, and it is not a medical test.
What does a world model learn from a baby's senses?
Each sense (vision, hearing, touch and body position) gets its own small predictive world model, fed input filtered to match an infant's at each age, and a fusion model combines them. In a small synthetic world, the quality and order of the input change what is learned and how fast.
Learning Like Infants: Input Constraints in Predictive World Models
How degraded, staged input and network growth change what predictive world models learn, with four propositions and a controlled simulation.
Simulating Infant Senses with Age-Tuned Networks
What infants sense at each age from birth to 24 months, how age-tuned networks simulate it, and the Baby Senses Lab.
Both papers are preprints in preparation; arXiv identifiers will be added when they are posted.