A blog over at Discover Magazine pointed me toward the latest episode of Terminator: The Sarah Connor Chronicles, which alludes to the importance of embodiment in the development of Artificial Intelligence. During the episode, which was first aired on December 15th, an Artificial Intelligence researcher poignantly notes that “…tactile experience is integral to A.I. development.”

Indeed! Classically, researchers in A.I. have attempted to build the mind first– a disembodied approach which implies that real time interaction with the world is not a necessary condition for intelligence. Rather, it turns out that the most successful A.I. programs have flipped this approach upside down. By starting with the body and a perceptual system instead, intelligence develops as that system learns to cope with its environment.

It’s good to see popular science fiction taking note of the latest developments in embodied cognition, and hopefully we’ll see more of this in the future with the series. It may have just won at least one newly intrigued viewer.

Rodney Brooks

Rodney Brooks

Rodney Brooks is a Professor of Robotics at the Massachusetts Institute of Technology, and he also the directs the MIT Computer Science and Artificial Intelligence Laboratory. Unquestionably, Brooks is the figurehead and principle leader of the embodied robotics movement.

Since his work was first published in 1986, Brooks has brought forth a new era in artificial intelligence. Instead of focusing on symbolic processing, which classical artificial intelligence was modeled on, he prioritized robotic architectures which were biologically-inspired. That meant focusing on sensorimotor and perceptual abilities– the capabilities an intelligent agent needs to interact successfully with the real world.

Brooks was also the first to point out that programming and embodying sensorimotor intelligence was far more challenging than programming basic symbolic reasoning skills. Thus, he argued that complex intelligence must ultimately be built out of those sensorimotor capabilities rather than from symbolic reasoning systems.

His classic article on this subject is titled Elephants Don’t Play Chess. His canonical books include the insightful Flesh and Machines, as well as Cambrian Intelligence. He is also a great popularizer of the subject, and has been featured in motion picture documentaries such as in Fast, Cheap and Out of Control.

Some of his new work is in developing low cost robots that will empower workers and evolve the world’s labor markets.

Although it’s a segment within a discussion about political framing, in this video clip George Lakoff discusses how embodiment comes to frame our ideas and perception through conceptual metaphor.

Within, he discusses how every word, in every language, is defined relative to a frame.

He also theorizes about the embodied source of certain well studied conceptual metaphors: “More is UP” and “Affection is Warmth”, the latter of which is also discussed in a recent study which we reported on here.

In conjunction with the International Joint Conference on Artificial Intelligence, a robotics exhibition is being held, called the Mobile Manipulation Challenge, from July 13-16 in 2009 in Pasadena, CA.

They need contributions, and are looking for demonstrations of physically embodied robots performing mobile manipulation tasks. Areas of interest include: Point-and-fetching, assembling structures, and searching for hidden objects.

Notifications of acceptance will be sent out March 20th of 2009. Go here for more information.

Donna Haraway

Donna Haraway

Donna Haraway is currently a professor of the History of Consciousness and Feminist Studies at the University of California at Santa Cruz. She began her career studying Zoology and Philosophy, and eventually earned her Ph.D. in Biology from Yale in 1972.

Haraway’s most central contribution to the study of embodiment comes at an intersection between her diverse scholarship in the history of philosophy, the science of biology and feminist epistemology. A critic of the traditional notion of objectivity as ‘a view from nowhere’, Haraway instead proposes that objectivity must be situated knowledge. In her own words, from Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective:

I am arguing for the view from a body, always a complex, contradictory, structuring, and structured body, versus the view from above, from nowhere, from simplicity.

Utilizing the metaphor of knowledge as vision, she argues that a nuanced understanding of vision and perception reveals that an object of sight (and likewise, an object of knowledge) cannot be conceptually or empirically removed from an embodied, structured and filtered context– from a situated point of view.

Her contributions to epistemology are paramount, but she is also acclaimed for her deconstructions of the masculinized metaphors and narratives which direct the science of primatology. She is also a major critic of essentialism, and is famous for her “Cyborg Manifesto”, wherein she utilizes the metaphor of the cyborg to fog traditional demarcations, dichotomies, conceptual divisions and social constructions regarding gender and race.

As a proponent of the power of technology to liberate, she is a constant contributor to cyberculture.

In this recent NY Times article, we get asked: “What happened to all of those early promises of having cogent robots, fully or partially integrated into our society, helping us out with all of our daily tasks?” Where are our robot maids, like in The Jetsons? Robots to dramatically and obnoxiously warn us of impending dangers, such as in Lost in Space? Robot pets? It’s already just about 8 years after Stanley Kubrick’s ominous prediction of 2001, so where are they?

‘Artificial intelligence’ has become a radical misnomer with the focus more on the ‘artificial’ rather than the ‘intelligence’. Cutting edge roboticists at the MIT robotics laboratory have an astute answer. Quite simply, early models of artificial intelligence took it for granted what seem to us to be simple bodily tasks, motor functions and perceptive abilities. It turns out that those are actually the most difficult kinds of abilities to program.

If cognition is fundamentally embodied, then it’s no surprise that intelligence hasn’t emerged in robots. Before you can have real, versatile intelligence, you have to master simple motor tasks in non-structured environments. It turns out our ability to do things like reach and grab for objects or walk through a changing environment has more to do with higher cognition than anything else. And so far, we can’t even put together a robot with the same motor capabilities of a newborn; or a cockroach, for that matter.

On the bright side, there’s also no foreseeable danger of assassin terminators taking over the world, either.



The RobotCub Project is fascinating, ongoing research which is studying cognition through the construction of a robot, called the iCub, which is a humanoid with roughly the appearance and size of a 3.5 year old child– which is currently about the same age as the project. This is the kind of research that is revolutionizing our knowledge of embodied, enactive cognition.

Even better, the project is entirely open source and public. At their website at robotcub.org, you can find source files, publications, pictures, videos and updates on the project. You can view the iCub software files. They’re always open to new international collaborators and partners.

The project also runs a yearly “summer school” where students have the chance to experiment with the iCub, and which helps to encourage embodied artificial cognitive systems research.