Me and my robot
Professor Owen Holland is pleased about the robotic finger. It has been created by a team of engineers at the University of Southern California, and is equipped with a special new tactile sensor modeled on human fingertips. It can identify a huge range of natural materials by their texture alone, and opens the way for robots of the future to have a better sense of touch than people.
“Touch technology is currently one of the most active areas in robotics,” Holland tells me. “Human skin senses are not very good in themselves, but the brain manages to extract a huge amount of information from them. We haven’t seen the ideal touch technology yet in a usable form – some kind of artificial skin would be good – but we’ll certainly take advantage of it when it appears.”
When electronic skin does make its debut, Holland will be draping it on Eccerobot, the world’s first anthropomimetic (ie human-aping) robot. This gangly, bug-eyed, slightly vulnerable-looking creation resembles Johnny 5 reimagined by Gunther von Hagens, and represents the cutting edge of global neurorobotics.
“It’s about answering the question at the centre of neuroscience, psychology, religion, philosophy and much of human art and enquiry for the past 2,000 years: what is consciousness?”
What makes it different from your run-of-the-mill droid is that while many robots created to date have copied the human form, Eccerobot is the first to copy the inner mechanisms and structures of human bodies. This is a robot with polymorph bones modeled on the human skeleton and kiteline tendons and shock cord muscles which map human musculature. It has been created at great effort and expense – and with the use of Gray’s Anatomy as an anatomical guide – by a stellar team of academics and engineers headed by Professor Holland.
But why is it so important for Eccerobot’s internal workings to mirror those of human beings? “The key idea was that our human consciousness depends closely on our brain’s model of our body,” says Holland. “Since we didn’t know which features of the body model were important, we needed a robot with a body model as close to ours as we could manage.” Recreating the body’s inner structures and mechanisms means that Eccerobot can interact with its environment – and gain knowledge of it – in the same way as humans.
It is the human-like nature of this cognitive interaction with the world that Holland and his team are most interested in. Ultimately the creation of Eccerobot is not about hastening the arrival of C-3PO robot flunkies, it’s about answering the question at the centre of neuroscience, psychology, religion, philosophy and much of human art and enquiry for the past 2,000 years: what is consciousness?
The hunt for consciousness
It’s the handshake that really gets people. Wherever Eccerobot is placed on display at conferences and science fairs, he puts out his polymorphic paw and bystanders tentatively grasp it, before breaking into beaming smiles as he gives it a firm, polite shake.
But Eccerobot has much more to offer than just good manners. Holland is hoping that if he keeps adding functionality to his robot, he will eventually be able to create machine consciousness, capturing that most elusive property in silicone and bungee.
“The main advantage of engaging in the practice of attempting to create machine consciousness is that it forces us to be really objective about consciousness and its possible mechanisms,” he tells me. “There is no room for discussion about anything that is not relevant to the engineering aspects of designing and building a system. The Nobel prize winner Richard Feynman famously wrote: ‘What I cannot create, I do not understand,’ and I believe that applies very well to the problem of consciousness.”
Holland believes that the creation of machine consciousness “will utterly transform the debate, because human consciousness will then be correctly seen as just one evolved implementation of a potentially quite general phenomenon intrinsic to certain classes of physical systems. Creating machine consciousness would merely emphasise what we know already – we’re not unique or special, just very complex evolved systems with particular characteristics.”
If we had scientific proof that there was nothing intrinsically special about humans, nothing that truly sets us apart, wouldn’t that be disastrous for religion? While Holland would regard the death of religion as “a real benefit to mankind”, he doesn’t believe that Eccerobot will bring the world’s faiths crashing down. “Religions don’t take evidence into account,” he says. “Humans are social and imitative, and this makes them very vulnerable to highly structured religious or political or nationalistic belief systems supported by large numbers of other people. Better education and a ban on indoctrinating young children might have some chance of success, but I don’t think that demonstrating robotic consciousness would make any real difference.” What’s more, he thinks that Eccerobot might get sucked in by evangelists. “It’s certainly possible that a conscious robot might fall victim to religious memes,” he admits.
“Creating machine consciousness would merely emphasise what we know already – we’re not unique or special, just very complex evolved systems”
Holland believes that attempts by other disciplines to tackle the problem of consciousness cannot give us an answer on their own. “Most philosophy derived from merely thinking about consciousness isn’t worth tuppence because we now know from scientific investigations that conscious experience is a very unreliable guide to what is actually going on inside us and in the external world,” he says.
As for neuroscience, Holland doesn’t think that identifying the part of the brain that deals with consciousness – its neural correlate – will necessarily help us to understand the nature and causes of consciousness. “Why? Because I believe consciousness is something that will be explained at a level higher than the level of the substrate that supports it. If and when machine consciousness becomes a reality, it will almost certainly involve electronics – digital and analogue – but I can’t imagine anyone asking about the electronic correlates of consciousness,” he says.
“Putting it rather crudely, if the substrate is a computer, then we need to think in terms of what the interacting programs are doing, rather than which chips are active. Finding out which structures in the brain are sufficient to support consciousness will only help if we know what those structures are doing individually and collectively, and we’re still in the dark where that is concerned.”
One major advantage to tackling the question of consciousness in robots instead of humans is access. Holland can monitor “everything that’s going on inside the robot, and we can mess around with every part of its ‘brain’ just as we wish. Of course, if the robot does become conscious, this may become subject to ethical restrictions, and people are already discussing possible future ethical frameworks.” Success in creating an independent-minded conscious robot could mean that it became entitled to the same protections as humans, including the right not to have its frontal lobe tweaked by inquisitive scientists.
Eccerobot grows up
The next few years will see a host of new functionality added to Eccerobot’s list of accomplishments. Top of the list are the powers of memory and learning. Memory is key, says Holland: “Humans spend a lot of their time consciously recalling and thinking about their past experiences – this is called episodic memory – and people who lose the ability to recall their past are very seriously handicapped, so it’s clearly an important function. People also spend a lot of time thinking about the future by imagining what might happen and what they might do. It turns out that our brains use the same circuitry for both memory and imagination, so by developing the one we’ll have done most of the work for the other.” Learning, meanwhile, is essential: “We can’t specify in advance everything that the robot will encounter, and so it will have to learn about the details of the world in the same way as a child does.”
But Eccerobot will receive more than just software upgrades. The breakthroughs in robotic touch by the team at the University of Southern California and other research groups mean that before long Eccerobot should be able to start to learn the spatial relationships between the world it sees, and the things that are actually in the world. “Without getting these two things in register, it’s very difficult to interact successfully with the world,” Holland tells me. “If you put on special spectacles that move your visual world to the left, you’ll miss objects that you try to pick up, but after a little while you’ll adapt and everything will seem normal again. Adaptive processes like these are going on in humans all the time, but we don’t notice them because they’re slow and unconscious.”
If Holland and his team achieve their aims and create machine consciousness, it will be an epoch-making achievement, which will force academics, scientists, lawyers, technology companies and pretty much every thinking person on the planet to reset their views on what it means to be human and how we are going to live alongside robots in the future.
But it won’t be the end of the journey for Holland and Eccerobot. Driven by a powerful curiosity about the origins and nature of human experience, the professor’s next step would be to shift the focus of his research, in partnership with Eccerobot, to “the relationship between consciousness and mental illnesses such as schizophrenia, depression, and dementia”. Provided, of course, that Eccerobot doesn’t decide to go and “find himself” or get seduced into joining a cult before the experiments begin.
Slow Journalism in your inbox, plus infographics, offers and more: sign up for the free DG newsletter. Sign me up
Thanks for signing up.