The Proteus Effect in Gaming: Avatars, Behavior and Limits

An avatar can be an experimental condition

The Proteus effect asks whether the digital body a person inhabits can influence their behavior. For gaming, it is an interesting question about self-representation, not proof that a particular skin gives its owner a better personality or higher rank. Much of the evidence discussed here comes from structured virtual-environment experiments rather than ordinary competitive matches.

This article examines older research, not a newly announced finding. Separate the visible change, the behavior measured and the setting in which it happened. An experiment that changes an avatar and records a negotiation decision is answering a narrower question than whether choosing that avatar permanently changes a person.

What the original experiments measured

Yee and Bailenson’s 2007 paper reported two experiments. In the first, thirty-two Stanford undergraduates were assigned avatars with faces rated as more or less attractive in a pretest. The researchers measured interpersonal distance and self-disclosure during a virtual interaction. They reported closer approaches and greater disclosure in the more-attractive-avatar condition.

In a separate experiment with fifty undergraduates, relative avatar height was manipulated during a negotiation task. Taller-avatar participants made more self-favoring offers in the later split and were less likely to accept an unfair offer than shorter-avatar participants. These were specific behavioral measures, not a direct reading of anyone’s character. The attractiveness categories were the study’s ratings, not a ranking we assign to people or players.

A later replication preserved some findings, not every result

Huber and Gajos’s 2020 PLOS ONE study included a conceptual replication of the height-and-negotiation task with 1,334 unpaid online volunteers. It changed important details: the partner was programmed, and the partner avatar’s scale conveyed the relative-height manipulation. The study was not an exact rerun of the laboratory procedure.

The authors found more self-favoring offers in the tall condition, with smaller effects than in the original negotiation study. They did not find a significant height difference in accepting the unfair offer. Reporting both outcomes matters: supporting part of a prediction is different from reproducing every measure. The paper’s broader aim was to examine whether virtual-environment experiments could be conducted with unsupervised online samples.

Do not turn a measured action into a player label

A person choosing an armoured character might enjoy its appearance, its fictional role, its mechanics or a memory attached to it. Without additional evidence, the selection alone does not tell you which explanation applies. If a game gives that character different abilities, performance differences also need to be separated from appearance.

Our earlier article about gaming and personality asks whether observed play can support an inference about a person. The question here runs in another direction: researchers deliberately alter a representation and observe a subsequent response. Keeping those questions distinct prevents a finding about one experimental manipulation from becoming a shortcut for judging a teammate.

Ask what would count as a useful comparison

When a headline claims an avatar changes behavior, look for the assigned conditions, the measured outcome and the comparison between groups. Check whether the avatar also changes what the player can do. Ask whether the researchers studied immediate behavior, later behavior or both, and whether the participants resemble the people to whom the headline applies.

A second study is especially useful when its differences are visible. Changes in equipment, recruitment, tasks or how the manipulation works can help explain why results vary. They also limit how confidently two numbers can be treated as the same quantity. A replication adds information; it is not a reason to stop reading the method.

Use a small observation without calling it an experiment

As an editorial reflection exercise, notice how you describe a cosmetic-only avatar before a relaxed session. Afterwards, write down one decision you remember and what else might explain it: a different opponent, a new map, a teammate’s suggestion or simply your mood. This is a way to notice assumptions, not a validated test of the Proteus effect.

Enjoy a character because you like playing it. These studies do not require treating avatar choice as a tool for correcting yourself, and a personal observation does not establish causation. The more useful habit is to keep the representation, the rules and the evidence visible before drawing a conclusion about the person behind the screen.

Keep exploring

FOLLOW THE SOURCE

Sources & further reading

Have a correction? Tell our editorial team.

✦ NICE. YOU LEARNED SOMETHING.