Multimodal exploration of objects during toy play is important for a child’s development and is suggested to be abnormal in children with autism spectrum disorder (ASD) due to either atypical attention or atypical action. However, little is known about how children with ASD coordinate their visual attention and manual actions during toy play. The current study aims to understand if and in what ways children with ASD generate exploratory behaviors to toys in natural, unconstrained contexts by utilizing head-mounted eye tracking to quantify moment-by-moment attention. We found no differences in how 24- to 48-mo children with and without ASD distribute their visual attention, generate manual action, or coordinate their visual and manual behaviors during toy play with a parent. Our findings suggest an intact ability and willingness of children with ASD to explore toys and suggest that context is important when studying child behavior.
This study demonstrates evidence for a foundational process underlying active vision in older infants during object play. Using head-mounted eye-tracking and motion capture, looks to an object are shown to be tightly linked to and synchronous with a stilled head, regardless of the duration of gaze, for infants 12 to 24 months of age. Despite being a developmental period of rapid and marked changes in motor abilities, the dynamic coordination of head stabilization and sustained gaze to a visual target is developmentally invariant during the examined age range. The findings indicate that looking with an aligned head and eyes is a fundamental property of human vision and highlights the importance of studying looking behavior in freely moving perceivers in everyday contexts, opening new questions about the role of body movement in both typical and atypical development of visual attention.
Human infants interact with the environment through a growing and changing body and their manual actions provide new opportunities for exploration and learning. In the current study, a dynamical systems approach was used to quantify and characterize the early motor development of limb effectors during bouts of manual activity. Many contemporary theories of motor development emphasize sources of order in movement over developmental time. However, little is known about the dynamics of manual actions during the first two years of life, a period of development with dramatic anatomical changes resulting in new opportunities for action. Here, we introduce a novel analytical protocol for estimating properties of attractor regions using motion capture. We apply this new analysis to a longitudinal corpus of manual actions during sessions of toy play across the first two years of life. Our results suggest that the size of attractor regions for manual actions increases across development and that infants spend more time inside the attractor region of their movements during bouts of manual actions with objects. The sources of order in manual actions are discussed in terms of changing attractor dynamics across development.
Recent self-supervised learning models simulate the development of semantic object representations by training on visual experience similar to that of toddlers. However, these models ignore the foveated nature of human vision with high/low resolution in the center/periphery of the visual field. Here, we investigate the role of this varying resolution in the development of object representations. We leverage two datasets of egocentric videos that capture the visual experience of humans during interactions with objects. We apply models of human foveation and cortical magnification to modify these inputs, such that the visual content becomes less distinct towards the periphery. The resulting sequences are used to train two bio-inspired self-supervised learning models that implement a time-based learning objective. Our results show that modeling aspects of foveated vision improves the quality of the learned object representations in this setting. Our analysis suggests that this improvement comes from making objects appear bigger and inducing a better trade-off between central and peripheral visual information. Overall, this work takes a step towards making models of humans' learning of visual representations more realistic and performant.