Redacción HC
03/09/2025
Children’s drawings are more than whimsical expressions—they are windows into the evolving architecture of the mind. A groundbreaking study published in Nature Communications (2024) by Bria Long and colleagues investigates how children’s ability to both produce and recognize line drawings matures in parallel during early childhood. By analyzing over 37,000 sketches from more than 8,000 children aged 2 to 10, the researchers reveal that even the messiest scribbles hold structured information about the world.
Although children start drawing and recognizing visual forms early in life, much remained unknown about how these abilities co-develop across many categories of objects. The central question was whether improvements in drawing and recognition reflect refinements in conceptual representations—knowledge about what makes an object identifiable—rather than merely better motor control. The team sought to measure, at scale, how “diagnostic” information—features that clearly identify an object’s category—emerges in children’s art as they age.
The research team set up a digital drawing kiosk in a children’s museum, where participants produced sketches of 48 different object categories. In total, they collected 37,770 drawings from children aged 2–10. To measure motor skills separately, they included tracing tasks. Recognition ability was tested by showing children peers’ drawings and asking them to guess the category.
For analysis, the researchers used deep learning models—specifically, latent features from the VGG-19 neural network and the CLIP vision-language model—to quantify how much diagnostic information each sketch contained. They also employed crowdsourced part-labeling, identifying which object components (e.g., ears, whiskers for a cat) were included. This multi-pronged approach allowed them to control for motor skill differences and assess whether improvements were truly conceptual.
The study found a steady increase in recognizability from age 2 to age 10. Importantly, this trend persisted even after adjusting for motor skill performance, underscoring that conceptual growth—not just better hand control—drives improvement.
A striking result was that even “unrecognizable” sketches to the human eye carried meaningful cues. Neural network analyses revealed that these early scribbles often still encoded broad properties like animacy (living vs. non-living) or relative size. This suggests that young children’s early attempts already reflect partial conceptual knowledge.
Crowdsourced part-labeling revealed that older children consistently included more diagnostic parts and correct spatial arrangements, which strongly correlated with better recognition by peers. The use of two independent AI models (VGG and CLIP) confirmed the robustness of these findings, despite differences in interpretability and sensitivity.
From an educational standpoint, this research highlights drawing as a low-cost, non-invasive diagnostic tool for assessing children’s conceptual development. Teachers and curriculum designers might integrate structured drawing activities into classrooms not only to foster creativity but to gauge and support visual literacy.
In clinical psychology, similar methods could be adapted for early detection of atypical development, such as in autism spectrum conditions or language disorders—though the authors caution that targeted studies are necessary before clinical application.
The findings also open avenues for experimental interventions. Could deliberate drawing practice sharpen children’s perceptual discrimination? There is some evidence in adults, but this remains to be tested in younger populations. Additionally, replicating such large-scale studies in culturally diverse settings could reveal how environment and exposure shape the development of visual concepts.
By combining large-scale data collection with advanced computational analysis, this study provides compelling evidence that the ability to depict and recognize objects develops hand in hand during childhood. Even the roughest early sketches hold the seeds of structured conceptual knowledge. For educators, psychologists, and parents, the message is clear: when a child draws, they are not just making marks—they are revealing how they see the world.
Topics of interest
EducationReference: Long B, Fan JE, Huey H, Chai Z, Frank MC. Parallel developmental changes in children’s production and recognition of line drawings of visual concepts. Nature Communications [Internet]. 2024;15(1191). Available on: https://doi.org/10.1038/s41467-023-44529-9
![]()