Languages differ greatly in how they express causal events. In languages like Japanese, the subjects of causative sentences, or "causers," are generally animate and intentional, whereas in other languages like English, causers range widely from animate beings to inanimate objects (e.g. Wolff, Jeon & Li, 2009). This paper explores "when" children learn to represent cause in their native tongue and "how" this learning occurs over the course of development. English- and Japanese-speaking preschoolers watched animations that were caused by (i) humans acting intentionally, (ii) humans acting accidentally, (iii) objects that generate energy (e.g. a machine), and (iv) objects that do not generate energy (e.g. a tool). Children were then asked to choose a good description of the event between two options. At age three, English- and Japanese-speaking children performed the task in similar ways, attending only to the intention of causal agents; however, at age four, speakers of the two languages diverged. English speakers were more likely to accept energy-generating objects such as machines as the subject of a lexical causative sentence than Japanese speakers.