Abstract
Having curiosity by nature, humans start to explore surrounding environment at very early age. Such exploration starts by interacting with objects in immediate reach, like toys. Human infants have been observed to make generalisation about their actions on objects and related outcomes by learning from interactions with them. In psychology, it has been found that these generalisations are strongly influenced by objects’ visual properties. In this study we present simulation of generalising object properties based on interaction and visual features. The results show that object shape is more reliable than other object visual feature for generalising outcome.