# Which Motion Features Induce the Percept - Unknown **Abstract:** — The perception of animacy is one of the fundamental social skills possessed by humans. The aim of this study is to determine whether such a skill can be accurately reproduced computationally, and if so, to describe the underlying factors which contribute to the animacy decision. We hypothesize a set of motion features which lead to such a distinction, construct a database of sample movements from both synthetic and natural stimuli, collect human judgments on the animacy of these stimuli, and evaluate the effectiveness of a computational system trained on this data to discriminate animacy. Index Terms— Animacy, Perceptual development, Causality, Naive physics I. I NTRODUCTION Objects around us may be classified as either having a controlling intelligence behind them or those dependent on purely physical circumstances. We can refer to these objects as animate or inanimate, although the distinction is not a clear one. What information do we need about an object, in order to classify it as animate? There are many artifacts which remind us of animacy, foremost of them being live biological motion. However, there are often moments where we classify abstract or non-living objects as animate, for example a complicated object trajectory generated by a mathematical equation, or a battery powered toy. The nuances that discriminate animate from inanimate motion have not been well described. In this work, we construct a system that performs this discrimination and we analyze the compone **Authors:** [[Viksit Gaur]]