May 2009

Barbara J. King


Ancient Primate Signaling in a Brave New World

Say you’re on a first date. Your eyes glaze a bit as the guy you’re with drones on about his ex-girlfriend’s mean-as-a-snake personality, about his garage’s sporty contents, about the dead-boring novel he’s reading (alas, one of your favorites). He’s cute, and there was a hint of chemistry between you over the antipasto, but as the entrée portion of the evening wears on, you see he’s way too self-absorbed. This dance you’ve danced before, and to no good end; preferable to dancing it again would be a solo evening with a book, Ben, and Jerry.

All at once, though, something shifts; you notice a new patterning to your date’s word flow and a stilling of his body. He looks right at you and asks to hear more about your love of all things Italy. Better yet, he listens actively as your answer, with an openness that’s winning you over. He smiles, and you smile, and all at once there’s hope for the dessert course.

What just happened? Cute Guy was jerk-o-metered. That is, he had been wired for your date -- wired to a device that sent feedback on his social-signaling behavior, and that helped him make a mid-course correction (one that netted you some overdue attention).

As Alexander Pentland writes in Honest Signals, the jerk-o-meter is real, just one example of a class of devices called sociometers. Pentland, who heads the Human Dynamics Lab at MIT, thinks that sociometers will revolutionize our understanding of human communication. In the book, he considers how our “ancient signaling systems” affect our love lives and work lives, indeed any part of our lives that depend on communication in pairs, groups, or large organizations.

Honest signals are hard-to-fake “ancient primate signaling mechanisms" that “form an unconscious channel of communication between people.” Four key aspects of these mechanisms are influence (one person may cause another’s pattern of speaking to shift in line with theirs); mimicry (the unconscious copying we sometimes do of each other’s movements or inflections); activity (the slight increase in movement that accompanies a feeling of interest); and consistency (the smooth patterning of speech and gesture that indicates strong focus; its opposite is an uneven patterning that indicates an openness to influence).

Pentland’s sociometers record not the words people utter but data on how they communicate: face-to-face time (through infrared sensors); physical proximity to others; body movement; and speech features all get measured. Aiming a hard slap at academic psychology, Pentland comments, “this is not evidence based on U.S. college kids filling out questionnaires.”

Identifiable in the data are clusters of honest signals that translate to certain social activities: exploring, active listening, teaming and leading. (It was active listening towards which the sociometer nudged Cute Guy in our fictitious opening vignette.)

Some of the results are only logical: Pentland found that the more people influence the speaking style of their conversational partners, the more they are central to their social network. Other results, though, surprise. Honest signals “predict behaviors with very good accuracy, even without adjusting for personal characteristics or the history of previous interactions.”

This conclusion, I have to say, runs counter to all that anthropology has taught me about patterns of human meaning-making. Yet it emerges from the monitoring done by Pentland and colleagues: sociometer data predicted which speed-dating partners would trade contact information at 71% accuracy and rated customer-service calls for a large retail chain as successful or unsuccessful at 87% accuracy.

Would this predictability pan out in contexts not involving strangers or (also much-studied) work colleagues? That remains to be seen. Nonetheless, the next question is obvious: Do we humans have as much conscious control over our moment-to-moment lives as we like to think? Pentland offers an answer with his characteristic confidence: “The idea that our conscious, individual thinking is the key determining factor of our behavior may come to be seen as foolish a vanity as our earlier [historical] idea that we were the center of the universe.”

In questioning Western science’s traditional model, in which thinking and decision-making are seen to be carried out inside each individual head, Pentland is right on. I long ago converted to D-cog (distributed cognition), a robust alternative model explained in Ed Hutchin’s Cognition in the Wild. Through anthropological analysis of decision-making on a naval ship, Hutchins brilliantly brings home the notion that a system can itself have cognitive properties. It’s bang-on target to conclude, then, as Pentland does, that “human intelligence is in both the individual and the social network.”

Honest Signals has a lot to offer. It’s evolutionary, with apes and human ancestors popping up throughout. It’s systems-oriented, driven by the recognition that “an effective group can potentially be smarter than any of its individual members.” It offers data to support its theory.

Yet questions nagged at me, even as I read. Among them:

-- In the universe of snazzy tech devices, the sociometer is groundbreaking. Don’t Pentland’s claims for it, though, ignore over a century of careful observation by anthropologists, sociologists and psychologists? Consider this: “We have shown that people’s behavior is much more a function of their social network than anyone has previously imagined.” Whoa! Pentland isn’t obligated to dig deep and write, in a trade book, about social theorists like Radcliffe-Brown and Durkheim who conveyed that insight a century or so ago. Nor is he obligated to cite anthropologists like Paul Byers on nonverbal signaling or Hutchins on d-cog. But is it too much to ask for a hat tip, in the text, to a century of social science whose business it is to know that “the immersion of self in the surrounding social network is the typical human condition”?  

-- For Pentland, honest signaling comes from a completely different channel than does linguistic communication, because honest signals are “inherently two-way” whereas “conscious language” is “strictly one-way.” But the work of dynamic-systems thinkers like linguist Charles Goodwin and psychologist Alan Fogel shows that speakers reshape their in-progress sentences as they take in the verbal and non-verbal responses of their conversational partners. Language, too, is contingently constructed: what are the implications of ignoring that fact?

-- Does Pentland truly wish to split apart relationships and emotion? It seems so, when he writes that he’s concerned with “social relations rather than speaker emotion. That is, honest signals communicate about the relationship between people…” Show me a human relationship that occurs in the absence of emotion!  

-- And what about ethical considerations, like right to privacy? At one point, Pentland offers a concern about “George Orwell’s vision of an all-controlling state.” Alright, but let’s have more: To what ethical principles should scientists be bound? What is the role of informed consent in these data-collection projects? Indeed, what does that concept mean in a power hierarchy? If a boss thinks it’s valuable to have MIT wiring up its workers, does a worker really get to opt out -- without consequences?

I’m the biggest fan around of ancient primate signaling, and hey, who doesn’t want to help Cute Guy make it to the dessert course? Still, I remain under the influence of those questions about this brave new wave of human-communication studies.

-- Barbara J. King wraps up another year at William & Mary with a co-regulated, contingently constructed high-five gesture with her students.