We Don't Need Human Touch, We Need Better Engineering

Almost every mammal participates in some sort of grooming- or touch-based bonding. Even the semi-solitary orangutan, the great ape that finds group interactions as appealing as palm oil deforestation, spends its juvenile years traipsing through the forest hand-in-hand with a friend, a behavior scientists sensibly call “buddy travel.” Touch is central to the mammalian understanding of the world and a critical part of how we interface with things and people who matter to us. The question has long been whether or not its reciprocation can be engineered. Increasingly, we reach out to the technologies central to our lives; will they ever reach back? Will that touch calm us, comfort us, or leave us cold?

To understand how to engineer touch, it’s important to understand that our brains treat it as a reward. In a review of primate social behavior, Robin Dunbar, the British evolutionary psychologist of Dunbar’s social brain hypothesis fame, lays out an argument that interpersonal touch triggers the release of two types of neuropeptides: endorphins and oxytocin. Oxytocin has, famously and not quite accurately, been described as the “love hormone.” More accurately, as the American Psychological Association notes, oxytocin makes us more tuned for social contact. Endorphins, also hormones, are the brain’s morphine analogues — products of pain control circuits as well as drinking and sex. Remember that saying, “Hugs not drugs”? Well, turns out that hugs are, physiologically speaking, pretty similar to drugs — albeit not very good ones.

Positive hormonal responses could be why touch decreases stress and anxiety. Channeling ‘chi’ might not do much for acute wounds, but literature reviews of caregivers touching premature infants have demonstrated that touch contributes to increased weight gain in babies. Conversely, in experiments in the 1940s by sociologist Rene Spitz, children in an understaffed orphanage — one nurse per seven babies — fared poorly, from a developmental standpoint, compared with kids whose mothers were incarcerated but could still interact with their children. Touch is, in short, medically as well as socially important. It is also hard to prescribe because it currently requires a second participant.

Could we fill an orphanage, or a nursing home, with caregiver robots, to avoid the scenario that Spitz studied? If touch spurs on these hormones, can robots get us there, too? The robot Pepper, which made its debut this summer, is meant to elicit emotional attachments (though, its creators caution, not coital ones) and it has succeeded at that. Pepper’s appeal in Japan is both strong and unsurprising. The country is far from technophobic and has an aging population apparently open to android care. But it’s still important to remember that Pepper elicited a response because of what Pepper looked and sounded like. Turn Pepper off and you’ve got an appliance.

In order to make touch matter, researchers will need to make robots like Pepper that people want to touch and make touching them rewarding.

From a technical standpoint, groups like University of California, Los Angeles’ Biomechatronics Lab are pushing the boundaries of artificial hands. The UCLA lab is really researching human prosthetics, but that doesn’t mean the researchers are ignoring other implications. A current line of inquiry is artificial haptic intelligence, a system that programs robots to improve the way they handle things as they interact with the physical world.

Once robot hands work like human hands, which is to say adaptively rather than with cold deliberation, there’s little reason to believe we wouldn’t respond in the same way to being touched by an AI as by an I. Even if there is an emotional component to being touched by humans that contributes to our positive psychological and psychologic reaction, there’s a fair bit of evidence that we can feel affection toward lifelike machines. Soldiers, for instance, have developed the equivalent of bot-crushes on robots that defused bombs, and the Tamagotchi, that codependent bit of software that we were all so fond of, spawned genuine feelings of paternal and maternal bonding.

If trends in research hold, there’s plenty of reasons to think the line between simulated and actual touch will blur significantly over time. We are — as of now — unaware of a psychological or physiological hangup that could make it impossible for robot touch to release endorphins, and synthetic touch has very real economic value for medical, demographic, and, yes, sexual reasons. That points to a future where robots and humans “buddy travel,” but that future remains some years off.