Social Science

It’s not exactly news that whether you’re asking someone out, pitching a business plan, or running for office, your audience responds not just to what you say but also to the way you say it. Speak with confidence and people will think you know what you’re talking about. Act thrilled to be chatting with someone and you often meet enthusiasm in return. “All these folk theories about how people work are to a fair degree true,” says Alex (Sandy) Pentland, MIT’s Toshiba Professor of Media Arts and Sciences and director of the Media Lab’s Human Dynamics Research Group. In his new book, Honest Signals, he explores just how these nonverbal cues work and how understanding them could make life better.

Using data collected by wearable sensors that he and his students designed, Pentland has measured subtle behaviors that are difficult to suppress or fake–signals like unconsciously mirroring someone’s gestures and varying (or maintaining) the volume and emphasis of your speech. These signals, he writes, operate independently of language to “change other people’s impressions of your attention, trust, interest, and focus,” even if they’re too slight to notice at a conscious level. They make up a type of communication that evolved to help ancient humans coördinate group activity, he says, and ignoring their influence can be a big mistake.

Pentland found that out firsthand when he served on a board of directors with a group of unusually smart, powerful people. “We made terrible decisions,” he says. “People would say things with passion and conviction, and other people would just say, ‘Yes, it must be true!’ There was too much charisma in the room.” If only there were a “warning light,” he thought, that could flag the social signals likely to inspire such unthinking assent. His “sociometers” now pick up these and other signals that define the roles ­people play in social groups. These roles influence human interactions profoundly, both for better and for worse–so profoundly that the outcome of an encounter can often be predicted just by analyzing them.