Brains and Machines

RSS Feed Homepage

Articles: 1, 2

On Intelligence by Jeff Hawkins
2007-05-03 20:28:00
I've finally read On Intelligence   by Jeff Hawkins, first published in 2004. It's deeply flawed but also required reading, pulling together a plausible theory of how the neocortex (and therefore intelligence) works. Let me know what you think of my review* and, as always, let me know what you think I should be reading. I started Ray Kurzweil's The Singularity is Near today. Note: Book reviews are always held on the Books on Brains and Machines blog rather than this one so that people can subscribe to them separately.

An Oxford Charade?
2007-04-25 17:28:00
Saw an interesting talk yesterday. The Visual Geometry Group from Oxford University claimed to have developed a technique for automatically tagging video content (both arbitrary objects and faces) so that it was easily searchable in real time. The faces could even be labelled as particular actors or characters using online resources. The goal was to have a video version of Google where you can search by names, faces, logos, object images, anything. Although searching is not an area I particularly follow, the work seemed to have important implications for machine vision and I was amazed at how effective it was as described in the presentation. The system works by breaking images down into small features (words) that together (in hundreds or thousands) combine to make objects and scenes that work regardless of the size of the object in the frame, it's position, or even its orientation. This is done off-line and is currently quite slow, though they're hoping to speed up to real time...
More About: Arad , Char , Hara

Hail to the neuromorphic engineers
2007-04-17 01:43:00
Mark Tilden changed my life. In about 1998 I started to become interested in analog computing for intelligence and came across a paper called Living Machines Mark wrote with Brosl Hasslacher a few years earlier. In it they talked about analog electronic creatures that were were very different to any other robots I had seen before. The 'nervous networks' that drove them were made of very few transistors, capacitors, and resistors—dozens rather than hundreds—and yet they, together, performed a rich, natural, and robust set of behaviors. The sun-seeking robots were even being used for interesting applications like satellite guidance and mine-clearing. It was a great story and it helped me understand what was important about intelligence in a way I hadn't before. What was more important, however, was that researching this piece made me more sensitive to other analog stories and introduced me to the community of neuromorphic engineers: people who try to build in silicon circuits...
More About: Engine , Engineers , Engineer , Romo , Morph

Making the best of what you've got
2007-04-12 16:51:00
When I discuss analog computation with most people, their instant reaction is to explain to me that there is no such thing as analog. For instance, one of the comments on an earlier post I wrote on analog vision pointed out that using a film-based camera rather than a digital camera doesn't give you infinite resolution, just a higher, finite resolution. In this case the the limit is grains of silver on the film rather than the number of pixels on a camera array. This is true, but it also misses the point. One of the main differences between physical computation (the slightly extreme form of analog I discussed in my PhD thesis) and digital computation is that it is good at extracting meaning out of tiny fluctuations. Digital computing, for very good reasons, is designed so that small 'glitches' that could lead a calculation astray are completely ignored. This is great when doing maths, but not when processing signals. Let me give a concrete example from a piece of research done ...
More About: What , Hat , Best , King , Best of

Books on Brains and Machines
2007-04-06 19:07:00
A reader suggested that it would be useful to build up a resource of reviews of books related to the stuff that we discuss here at Brains and Mac hines. Rather than keep those reviews here, it seems to make sense to give them a blog of their own. This will allow for easier categorization both by subject and by rating. However, I'll post a notice of the reviews (which I expect to post monthly or so) here as well, so you don't need to subscribe to both feeds. The first book I've reviewed is something I read when I was doing my research for the Wired article: Seeing Voices by Oliver Sacks. The book looks at deaf history and culture, and talks about the link between language and intellectual development and the differences between spoken and visual languages. I hope you find the review useful. Also, do feel free to get in touch with suggestions for other books that I should be reading and/or reviewing... Just keep them to the general area of brains and machines and no pseudoscience ...
More About: Books , Machine , Machines , Book

Senses: What are they good for?
2007-04-03 16:30:00
When I first heard the story about the Feelspace or North belt, one of the things that excited me the most was that it demonstrated how we learn from all the stimuli we are regularly in contact with, even those that don't seem very special. We learn that does and doesn't 'feel right' from the swish of our winter coats to the sound of our shoes on the ground to the vibration of our cars. We don't just do passive sensing, but active: we interact with the world in routine ways and can tell something about our environment by the way it reacts to us. Another thing that I thought was very interesting was that it showed how we extend ourselves with technology already. One of the things that occurred to me in the lecture where I first heard Peter König talk about this work was that taxi drivers might use their car as a kind of big North belt, and improve their sense of direction (in the car) by associating vibrations, the feeling of turns etc. with their position. So, for instance, i...
More About: Sense , What , Hat , They , Good

Loser: IEEE Spectrum on tongue display
2007-03-28 16:30:00
One of my colleagues told me at lunch that he'd seen an IEEE Spectrum article saying that the Wicab tongue display was one of the 'loser' technologies of 2007, so I thought I'd take a quick break from the idea of intuitive displays to deal with this. I was very disappointed with the article, in that none of the 'experts' quoted seemed to have any specific knowledge of the device, how it was perceived by users, what it felt like to use etc. Also, it was not noted that these people were, essentially, competitors. (I have no axe to grind except that I think it was bad journalism). I have no idea whether the project will succeed or not (something that will probably have more to do with business model and marketing than technological excellence), but it's very early days and if the company can, as they say they intend, turn the device into something that can be worn as a plate in the mouth, I think it at least has a chance. I won't reiterate why, as it's in my previous blog on ...
More About: Play , Display , Loser , Tongue

Physical intuition
2007-03-23 19:31:00
After the last couple of posts I got some interesting comments about familiarity being important to intuition. Roger Attrill pointed out that Adobe users would find Photoshop intuitive (while others wouldn't) and Bob Salmon pointed out that musicians might find software using some kind of musical-score based interface natural to use that the rest of us wouldn't. For the kind of sensory interfaces I'm interested in, however, I think you'd want something that is cross-cultural, something that relies on being human, maybe, or on a naive understanding of the physics of the world, but not otherwise requiring training. What I'm getting at might be considered a wider definition of a term I just learned for the first time last year: symbology. When I visited Tom Schell's lab, I talked to him and to Todd Macuda about the various ways that, mainly through visual display, they try to make jobs easier for pilots. One of the examples they showed me, for instance, was having a 'road' di...
More About: Physical , Intuit , Tuition , Intuition

Getting complicated
2007-03-20 19:21:00
During my recent research into the world of sensory interfaces, I had at least a dozen discussions, maybe more, about what the word intuitive means. I can't remotely claim to have cracked this: on the contrary, it seems to me that people are still at the stage of figuring out the right questions to ask. It's true that human-machine interfaces have been well studied and there's vast literature on the subject. However, though sensory interfaces that allow you to interact with the real world (rather than a computer world) will certainly have many things in common with virtual displays, I think there will be some differences too. Anyway, the first part of my definition is that the more intuitive something is, the less training it should require to be understood. Obvious, yes, but sometimes stating the obvious can be useful... So, if we are looking for an interaction that requires as little training as possible, what does that mean? From talking to two people in particular—Paul Ba...
More About: Cat , Comp , Cate , Complicated

Intuition versus attention
2007-03-16 20:28:00
One of the aspirations of those trying to feed information into the brain through the senses is how to make the process intuitive. Roughly we know what this means: we want to have the information be self-explanatory, to not require any further thought, to immediately and naturally provoke the behavior that we intended. But what exactly does this mean in practice? How does it break down? During my research for myWired feature on feeding information through the senses (April 2007 issue), this is one of the questions I really wanted to get to the heart of. Among the people I discussed the issue with was Charles Spence, an experimental psychologist from the University of Oxford here in the UK. Spence is particularly interested in creating devices that will make cars safer by warning drivers about hazards, waking them up, etc. In particular one of the things they are looking at is whether they can build some kind of vibrating device into the seat belt that can give us a "tap on the...
More About: Versus , Intuit , Tuition , Attention , Intuition

Seeing with your brain
2007-03-13 18:08:00
Biological brains can do something that we're still working on building into machines: rewire themselves to take advantage of the best information available. This is a well-known phenomenon, but it's not widely understood just how radically different the rewiring or optimized wiring can be from what is considered 'normal'. Probably the best known example of this kind of neural plasticity is that, if a mammal is born with an eye that is defective or missing, the brain will only develop connections to the good eye. This makes sense because the other one isn't delivering any useful information. The original work showing this was done by Hubel and Wiesel back in the '60s. More recently, Mark Bear and colleagues at the Howard Hughes Medical Institute (MIT) have showed that, in this kind of experiment, the normal eye actually improves in this case: it's not just as good as it would have been without the other eye being defective, it's better. Not only that, but in experiments wh...
More About: With , Rain , Your , Seeing , Brain

Hearing with your chest
2007-03-09 19:13:00
As well as explaining evolution (see my last post), analog feedback may explain the ability of deaf people to hear through their bodies as the percussionist Evelyn Glennie learned to do. According to her biography,Evelyn spent a lot of time when she was young (with the help of Ron Forbes her percussion teacher at school) refining her ability to detect vibrations. She would stand with her hands against the classroom wall while Ron played notes on the timpani (timpani produce a lot of vibrations). Eventually Evelyn managed to distinguish the rough pitch of notes by associating where on her body she felt the sound with the sense of perfect pitch she had before losing her hearing. The low sounds she feels mainly in her legs and feet and high sounds might be particular places on her face, neck and chest.This is pretty amazing. What her brain managed to do was to take physical structures in her body that were not 'designed' or evolved to be sensors, but use them that way anyway. This i...
More About: With , Hear , Your , Ring , Chest

Where a little means a lot
2007-03-07 18:49:00
One of the things that has driven my work over the last decade has been an interest in analog systems (which I alluded to in an earlier post) that perform what I call physical computation. What that means is that they are not ‘programmed’ except in the sense that, like all objects, they are forced to obey the laws of physics. When physical objects become sufficiently complicated, then they start to behave in interesting ways while still just doing what comes naturally. That’s interesting because biological organisms fall into this class of physical or analog machines. Not everone understands this. Many people mistakenly assume that because information is transmitted through the brain by uniform pulses of electricity, that this means the brain is like a digital computer. Not so. In fact they brain is a much more complex machine, where the time between pulses (which can vary continuously and so is analog) carries the information, not the pulses themselves. But, from my point...
More About: Here , Little , Where , Litt

I want my ESP
2007-03-03 01:53:00
Stefan Marti is a research scientist with Samsung in San Jose. When he was a student, he had the experience of walking around with a broken vibrating pager that gave him an unexpected extra sense: for certain kinds of electromagnetic fields. From the buzzing in his pocket, he knew when people were making popcorn in the microwave, when there was a wireless router nearby and, most interestingly, when a phone—whether his or someone else’s—was about to ring. The latter turned out to be a nice feature. Sensing an incoming call, he learned not to start a new sentence, and this experience ended up inspiring his doctoral work. His research involved making phones smart enough to figure out whether or not they should ring at all, and to figure out the appropriate action to take (passing on messages etc.) depending on the circumstances. Now he’s working on long-term projects for Samsung that will involve making phones interact with people in more-intelligent/less-disruptive ways. For...
More About: Want

A new kind of military conflict
2007-02-28 01:50:00
We are notoriously bad at simultaneously doing different tasks with the same bit of brain. For instance, in his book Seeing Voices, Oliver Sacks points out that it’s known to be extremely difficult to sign in American Sign Language while speaking in English, which has a completely different syntax and grammar. Likewise, it is thought to be impossible for someone to write Chinese while speaking English. So, if the tongue and the skin are providing imagery to the brain (as discussed in the last post), it may be important that it’s complimentary to that coming from the eyes: enriching the visual stream rather than feeding in some completely new kind of information. However, this may not be so easy, as the US military are finding out. A new program called the Multi-Spectral Adaptive Networked Tactical Imaging System (MANTIS), is being developed by Raytheon, Sarnoff, and Rockwell Collins Optoelectronics. The project involves providing different kinds of images to each eye: this coul...
More About: Military , Conflict

Choosing what to see
2007-02-25 01:48:00
In my previous posts I discussed using tactile displays (the tongue display and spatial-orientation vest) in order to sense the world in new ways. In both cases my immediate reaction was to close my eyes, which may seem foolish: I was cutting out visual information that might have been useful. However, there was method in my madness. Vision dominates the other senses for spatial tasks and the information provided by the eyes can be misleading. At the Royal Air Force base in Henlow, for instance, pilots take courses on why you shouldn’t trust what you see through your night vision goggles NVGs) too much. To bring the message home, the pilots are shown a film of an accident caused by one pilot misunderstanding what his NVGs were telling him: that the other plane is feet away, not hundreds of feet. This problem is anything but theoretical. Tactile displays like the one being tested at the Operator Performance Laboratory are intended to help pilots to get around problems with inform...
More About: What , Hat , Sing