cannot be sure that children's pretense is in fact novel, or whether they simply copy or follows others' instructions. Some experimental work has attempted to capture children's novel pretense (Nielsen & Christie, 2008; Rakoczy, et al., 2004). However, we argue that what looked like novel pretense in these studies could be explained by deferred imitation. This is the first experiment to show preschoolers create their own novel object substitutions, without relying on deferred imitation. Generating Object Substitutions Pretend play differs from functional play as the actions performed during pretend play are technically incorrect (e.g., drinking from empty cup, talking to banana; Hoicka & Gattis, 2008; Hoicka, Jutsum, & Gattis, 2008; Hoicka & Martin, 2016). One form of pretend play, object substitution, requires temporarily suppressing the typical action for the object while performing an action that is typical for another object (e.g., pretending banana is phone; Tomasello, Striano, & Rochat, 1999). Naturalistic research suggests children perform object substitution during free play from 2 years (e.g., Belsky & Most, 1981; McCune-Nicolich, 1981). However, these studies do not provide information on the content of their play. Additionally, it is difficult to determine whether children's object substitutions are generated by children themselves, or whether they are copied from others (immediately after observation, or using deferred 1 imitation; see Hoicka & Akhtar, 2012). 2 Experimental research suggests 2-to 3-year-olds perform object substitutions (e.g.,