Research in child development has shown that embodied experience handling physical objects contributes to many cognitive abilities, including visual learning. One characteristic of such experience is that the learner sees the same object from several different viewpoints. In this paper, we study how learning signals that equate different viewpoints---e.g., assigning similar representations to different views of a single object---can support robust visual learning. We use the Toybox dataset, which contains egocentric videos of humans manipulating different objects, and conduct experiments using a computer vision framework for self-supervised contrastive learning. We find that representations learned by equating different physical viewpoints of an object benefit downstream image classification accuracy. Further experiments show that this performance improvement is robust to variations in the gaps between viewpoints, and that the benefits transfer to several different image classification tasks.

Geometric analogy problems remain an intriguing part of intelligence scales, which is closely correlated to many cognitive studies, such as perception, conception, memory, abstract and inductive reasoning. The problems not only target the most fundamental element --- analogy-making --- in human cognition, but also require integration of multiple components and stages: looking at the test booklet, thinking for a minute or two, and deciding the answer. Great efforts and achievements have been made to explain different individual aspects of this process. In this paper, we take a more holistic approach from the perspective of problem-solving, by modeling the entire process, from the moment the visual stimuli are received to the moment an answer is decided. Therefore, we explore how the final solution can be built upon visual inputs and necessary components that lie between the perceptual input and conceptual output. Particularly, we designed a novel similarity metric and a correspondence-finding method based on mapping and optimization. With these two basic blocks, we implemented a computational model, and report our initial results on a classical problem set.

Studying the mechanisms and trajectories of child development continues to be of critical importance, especially in thecontext of fast-changing digital environments, ubiquitous screens, and ever-increasing permeations of technology intoour lives. In this work, we study historical changes in the visual complexity of information presented in different mediacategories, including an eighty-year history of Disney movies and a forty-six-year history of NBC news programs andtelevision commercials. Our analyses include metrics of static visual variance in single frames as well as several metricsof visual change over time. By performing similar analyses on a dataset of egocentric videos, we compare trends in digitalmedia with data that more closely resemble real-life visual experiences. Understanding the visual characteristics of themedia we consume is an important step towards further investigating the effects that these characteristics might have onour perception, attention, and learning, especially in young children.

Understanding the computational mechanisms enabling visuospatial reasoning is important for studying human intelli-gence as well as for exploring the possibility of introducing human-like reasoning into artificial intelligence systems. Inour work, we investigate how a collection of primitive image processing operations can be combined into different co-herent strategies for solving a range of visuospatial reasoning tasks. We evaluate our approach on 20 subtests from theLeiter International Performance Scale-Revised (Leiter-R). Through our computational experiments, we show that withonly four primitive operations similarity, containment, rotation, and scaling we can form strategies that solve, to differentdegrees of success, at least portions of 17 of the 20 subtests. These results lay foundations for our future work to study howintelligent agents can learn and generalize strategies from simple task definitions in order to perform complex visuospatialreasoning tasks.