Infants' self-generated visual statistics support object and category learning

点击翻译按钮获取中文摘要

1 Project Summary 2 3 Human visual object recognition is remarkable in its ability to recognize individual objects in challenging 4 circumstances and to rapidly recognize even novel instances of tens of thousands of everyday categories. 5 Although a great deal is known about these processes at maturity, very little is known about their development 6 especially with respect to common everyday objects and the experiences that support robust object recognition 7 and categorization. This gap is critical because object recognition and categorization support early word 8 learning, physical problem solving, and the later learning of orthographies and mathematical symbols. This 9 research projects focuses on visual object learning in 1 year old toddlers, a developmental period that at the 10 front end of marked advances in visual object recognition and a period in which children with multiple risk factors 11 begin to fall behind the normative developmental trajectory. The approach focuses on the properties of real- 12 world visual experiences that support learning to recognize individual objects in challenging visual contexts and 13 generalizing that learning to same category members. The method uses head-mounted eye-trackers to capture 14 field-of-view images from 100 infants 17 to 22 months of age as they spontaneously interact and play with 15 objects; the supplemental projects adds 40 toddlers to the sample who have small productive vocabularies for 16 their age. These “Late talkers” are at risk for future diagnosis of Developmental Language Delay and also show 17 disruptions in the development of visual object recognition. Through active interactions with objects infants 18 generates their own packets of visual data for learning. Multiple visual properties relevant to object perception 19 will be algorithmically measured and quantified. Toddlers’ recognition of the actively-engaged object and a novel 20 object from the same category will be measured in challenging benchmark contexts including clutter, occlusion, 21 and different views. Category generalization will be measured in a name generalization task. Advanced statistics 22 and machine learning will determine the visual properties of self-generated experiences that support infants 23 object recognition and categorization. The research will provide the first characterization of the natural visual 24 statistics of toddlers’ active interactions with objects and potentially transformative evidence that the 25 developmental foundation for human prowess in visual object categorization lies not in experiences with many 26 different instances of a single category, the standard assumption, but in active visual experiences with individual 27 objects. 28