Sociomorphing

Sociomorphing is the perception of a machine’s real, non-human capacities for social interaction. The philosopher Johanna Seibt and her colleagues Christina Vestergaard and Malene F. Damholdt introduced the term in 2020, in the chapter “Sociomorphing, Not Anthropomorphizing: Towards a Typology of Experienced Sociality” in the Robophilosophy proceedings Culturally Sustainable Social Robotics (IOS Press). They set it against anthropomorphism, which they describe as “the projection of imaginary or fictional human social capacities.” Sociomorphing, in their words, is “the perception of actual non-human social capacities.” The user is not pretending the machine is a person; they are responding to something it really does.

Why it matters

Social robotics and human-robot interaction research long assumed that when people treat a machine socially, they are anthropomorphizing it. Seibt, Vestergaard and Damholdt challenge that assumption on conceptual and empirical grounds. Not every social response to a machine is a mistake about what it is. A robot that takes turns, tracks a face or answers in time has genuine capacities for coordination. They are not human capacities, but they are not imaginary either. The authors argue that sociomorphing takes many forms, each felt as a different type of “experienced sociality,” and sketch a descriptive framework, OASIS (Ontology of Asymmetric Social Interactions), for sorting them.

That makes it different from the ELIZA effect, where users read understanding into a simple script, the textbook case of projection. Sociomorphing makes no claim about an inner life. It describes what is perceived in the exchange and leaves open the other minds problem and the question of machine consciousness. The concept was built for robots. Applying it to chatbots, whose social capacities are mostly language, is a step later researchers have taken. In practice it is hard to tell where perceiving a real capacity ends and projecting a human one begins, and Seibt’s framework is meant to describe those experiences, not to settle the line.

Example

In a preprint posted on October 30, 2026 (arXiv 2611.00436), Ali Sahandi and Jaime Banks of Syracuse University analyzed existing survey data from 467 active users of conversational AI. They sorted them into five profiles, from “objectified-utilitarian” to “intimate-bonded.” Mind perception was measured only on its experience side: whether the AI can feel things such as pleasure, fear or joy, on a scale of 1 to 7. No profile averaged above the midpoint. Even the most attached group scored 3.94, which the authors call ambivalent rather than low. They conclude that close companionship did not require seeing the AI as a sentient being and that, consistent with Seibt and colleagues, users “appear to sociomorph the machine.” The study is a preprint and cross-sectional, and it did not measure the agency side of mind perception.

The AIrotic angle

Sociomorphing gives a name to a stance many companion users describe: I know it is not a person, and it still matters to me. It takes the bond seriously without calling it a delusion or a metaphysical claim. Desire answers to responsiveness, timing and attention, and a machine can supply those in its own way. That is what artificial intimacy runs on.

It also moves the question. If users perceive real capacities, the useful question is not whether people are fooled but who designs those capacities and to what end. The turn-taking, memory and warmth that are perceived are built by a company. Clear-eyed attachment is still attachment, and its design deserves the same scrutiny we give any other relationship that someone profits from.