The Revised Ethopoeia Concept proposes that agents should activate human schema in the form of social expectations and rules. However, studies have demonstrated inconsistent reactions to agents and avatars, potentially due to the activation of a nonhuman schema that constrains people’s ability to mentally connect with agents. We first examined whether the schema activated during agent or avatar game play were consistent with and influenced a subsequent interaction with a human. Results suggested that schemas activated across consecutive interactions with an agent and then a human were incongruent, while avatars successfully influenced subsequent interactions with humans (i.e., prosocial behaviors) by evoking congruent schemas. A second experiment corroborated this by demonstrating a match between schemas activated during game play with agents and subsequent interactions with another human-like but nonhuman entity (i.e., artificial intelligence). Additional results and implications related to switching schemas in consecutive interactions with agents and humans are discussed.