Humanlikeness as design, anthropomorphism as inference: a conceptual framework for human–robot interaction

Humanlike robots are increasingly deployed in public and private settings, yet research and design efforts are hindered by persistent conceptual conflation between robots' humanlikeness and individuals' anthropomorphism, as well as by a hyperfocus on appearance-based definitions. This paper distinguishes humanlikeness as a multidimensional design property-the extent to which the design of an agent intentionally mimics human characteristics, traits, processes, or behaviors-from anthropomorphism as a psychological attribution process in which people ascribe human properties to non-human agents. We argue that treating these constructs as analytically distinct, while recognizing their reciprocal and dynamic interaction over time, enables more precise theorizing, clearer operationalization, and more actionable design guidance. To support cumulative interdisciplinary work, we situate both constructs within an integrated framework that organizes adjacent constructs (e.g., humanness, machinelikeness, animism, personification, and dehumanization) and provide curated measurement resources to facilitate principled study design and evaluation.