It’s been a rough week for the General Services Administration, whose chief resigned on Monday after revelations that the agency had blown nearly a million dollars on a lavish conference in Vegas—complete with a comedian, a clown, a $2,000 loft suite party, and a mind reader. That last expense seems a little silly: After all, everyone knows that mind reading isn’t real. Or is it?

A 2004 paper explains that, in very simple ways, “we are all mind readers,” just “not the magical sort.” We make reasonably accurate inferences every day about the mindsets of people around us—our coworkers, our friends and loved ones, even total strangers. But the study asks exactly how we make those inferences, so it presents an experiment wherein subjects were asked to fill out a form indicating their preferences for one of two jokes, one of two paintings, and one of two verdicts (innocent or guilty) about an ambiguous court case. They were then introduced (via a computer program) to their mind reading “target”: A man named “Michael” who, they were informed, either agreed with their choices or disagreed with them. After scoring the level of similarity they felt towards “Michael,” the subjects then watched his behavior in a video and were asked to make judgments about his mindset and character. The result: “When participants found they had such specific cues in common with targets (e.g., a shared preference for a painting by Klee over one by Kandinsky), they were considerably more likely to project their own desires and attitudes onto targets,” and “when participants found they did not share such attributes with targets, they were more likely to rely on stereotypes.” One wonders if the GSA’s Vegas mind reader was a spendthrift or a miser—after all, it could have determined which method he employed while helping his customers get fired.