Monday, January 25, 2010

Sandy Pentland: Honest Signals

Sandy Pentland's Honest Signals talks about subliminal cues we all emit and makes a case that they provide reliable clues about all kinds of social outcomes. It's a quick read (100 pages of text, 50 pages of Social Science appendices, 30 pages of notes, bibliography and index) and makes a reasonably compelling case that our unwitting signals accurately foretell outcomes in many situations. (negotiation, sales, poker and dating were all studied.) The cues themselves are described in the appendices, along with a discussion of the gropus who have learned to read them (salespeople, poker players). Pentland says it will soon be possible to produce portable devices to make the signals evident. Some of the experiments he reports on were done using prototype portable units, others were done in the lab with earlier versions.

Several important questions aren't covered in the book: Why do we signal this way but not notice it consciously?, How could we effectively learn to detect these signals?, Can people learn to change the signals they produce (either to conceal their motivations, or to steer situations to more desirable outcomes)?

Pentland does show that there's reasonable evidence that the signals occur early in a conversation, that they are reliable, and that most participants don't notice the signals, even though they usually realize how things went for them by the end of a meeting. He tries to argue that we can use the predictions made about job satisfaction and organizational stress to optimize the way work groups are managed, but there is little indication that the tools can be used for steering. When the metrics show that information isn't flowing freely across a geographical boundary within a group, there are obvious things to do to change that, but is there an obvious response if they indicate that one person has more influence than others? Maybe making diagnostic information about the form of the interactions visibly shared within a group would make it possible for someone to intervene, but we're a long way away at this point from knowing how to optimally intervene in group politics or communication.

No comments: