Brains and Machines

New paper: Being Analog

2011-02-22T22:48:07+00:00

I attended the 3rd International Workshop on Optical SuperComputing in Bertinoro, Italy, back in November. It don't get out so much these days, so I was pretty shocked when a total of 10 people showed up (or was it 11, you'd think I'd remember!) It's a shame, because there is some interesting stuff going on in this field...

In any case, despite a couple talks delivered remotely, the program turned out to be pretty thin (perhaps not surprising given the attendance). So I offered to give a presentation about my stuff, even though it's not strictly about optical. The talk, called Being Analog, was well received, so they asked me to write it up as a paper for Springer's Lecture Notes on Computer Science (which had a deal to do the proceedings). Uploaded it today: so here it is, hot off the presses, if you're interested.

Photo: The view out of my window from Bertinoro Castle.

A silicon vocal tract

2010-04-15T16:46:39+01:00

Researchers at MIT have developed the first integrated-circuit vocal tract, which could eventually make it's way into high-end PDAs. It's biologically inspired and combined with a bionic-ear processor in a feedback loop. This means it can not only be used for producing speech but also recognizing it: the vocal tract can help to model what the ear thinks it is hearing to verify whether it's likely to have it right or not.

Essentially, the system is a biological model of our own method of producing speech that has been implemented in silicon. It is not just good at synthesizing speech but (in conjunction with the ear/feedback) but in interpreting it because it can literally figure out what muscles etc. would have to be used to produce a particular sound and so can 'reverse engineer' what sound was actually intended. F and S may sound similar, but the way the sounds are produced in terms of muscles are quite different. So if you're able to get into the physiology from small differences in what you hear then you can make much better guesses at what's being said.

The chips used (one for the vocal tract and one for the ear) are both based on human have been implemented in custom analog circuitry. Digital computers can be used, but the computational complexity of the problem means that the analog solutions are drastically smaller, faster, and less power-hungry.

There are a number of really interesting commercial applications. The most obvious is robust speech/speaker/language recognition in noisy environments, but they are also building a glove that can drive the chip and a brain machine interface that can be implanted in the brain for speech impaired subjects, and are building muscle interfaces that would allow silent phone calls (you talk silently on the train, the system figures out what sounds you are trying to make and makes them for you down the phone).

Cloning developmental robots

2010-03-09T16:36:52+00:00

I was at a conference about humanoid robotics, and particularly the iCub, yesterday and Mark Lee from the University of Aberystwyth was talking about the difficulties of 'raising' truly developmental robots: robots that learn about their bodies and environments through experience the way we do. This got me thinking.

Being an analog girl at heart, and given that robots have a lot of essentially analog components (even if they are driven with digital controllers), I'd always assumed that truly intelligent humanoids would have to be raised developmentally. Each individual would have to learn about it's unique set of motors and sensors and processors and what they could do and how they could interact with the world before they would be able go out and do things. Now I wonder if it has to be as drastic as that.

My thinking now is that the question turns on just how different each robot will be to its 'siblings': robots turned out in the same batch and that are (at least intended to be) identical. There are bound to be subtle differences because of manufacturing tolerances, but perhaps these don't matter so much for digitally-driven machines. After all, a robot has to be able to cope if some of its components fail or change their performance (for instance) due to changes in temperature. So perhaps we could clone the 'brains' of one robot and successfully transplant it into another, which would just wake up feeling a bit out of sorts and have to re-optimize.

If that's true, it brings up a lot of interesting questions. Is the best idea is to focus energy on the development of a single machine to clone, or to raise a whole bunch of robots to a certain point and then, in a natural-selection-type way, clone the best brains and ditch the rest? Perhaps the latter would be the best way to learn the best teaching methods for robots? And at what point in their development should they be cloned? Presumably you don't want to have to send a 'baby' robot to each new workplace so that it can learn about it's new environment and tasks at the same time as it's learning how to see and control its actuators from scratch. On the other hand you want it to have plenty of scope for adapting to its new environment...

I wonder.

Photo: Yan Wu working with the Imperial College London iCub.

A silicon spinal cord?

2009-08-27T15:43:43+01:00

There has been a huge amount of progress in the area of using brain probes to read our intentions, and then relay those to a limb that would otherwise be paralyzed (or, for that matter, to a prosthetic limb). I've written about related subjects in the past, both for EE Times and the IEE (now IET), but was really impressed to see that there has now been a full demonstration of the technology (albeit, in a monkey) and that there are new clinical trials underway to show how implanted brain probes can help real human patients. There's also an important trend towards wireless implants. If you're interested in this, you might want to read my new piece in New Scientist magazine on the subject.

Some cool research I didn't have space to cover in this feature is described in a new paper by Jose Carmena and Karunesh Ganguly about how our brains can learn to cope with this kind of implant. Until recently, most experiments with brain probes required regular, and often laborious, manual ‘tweaking’ of chip parameters. But it turns out that the brain can adapt to both the neural probe system (the bit that is connected in the brain and is interpreting the brain signals) and the behavior of whatever it is controlling. This is true whether the output is connected to something virtual, robotic, prosthetic, or part of the user’s own damaged body.

What is critical is that the mapping between the brain signals and actuator movement stays constant over time: i.e. that you don’t change the way the electronics work or which muscles they are controlling. In this case, the brain will create a stable memory of how best to create neural signals to drive the probe and what’s connected to it, potentially allowing a patient to build up a sophisticated set of motor skills.

Of course, it does make sense that this would be true, given what we know about plasticity in the brain, but now we know for sure. This is yet another reason to suggest that if we keep plugging away with these technologies we will eventually reap real benefits for real people.

No free will for Christof

2009-06-09T19:46:16+01:00

Christof Koch and I see eye to eye on a few things. For one, we're both Mac-heads. (Although I have no Apple logos anywhere on my body). For another, he used to be a neuromorphic engineer: building analog neurons out of silicon. A couple of nights ago I came across a podcast talk by Koch and found we had something else in common... neither of us see any evidence for the existence of free will.

Since I'm a physicist by training (he's a biophysicist and neuroscientist, now focusing on consciousness), much of his argument was familiar to me. First, obviously, classical mechanics is deterministic: cause and effect. No room for free will there... Second, even though some classical systems are mathematically complex (or chaotic)—which means that you cannot predict their behavior far into the future—they are nevertheless still deterministic. You might not be able to figure out what's going to happen next, but it's still going to happen.

Then we get to the ugly subject of quantum mechanics. Although we don't know what's going to happen next—whether the photon will transmit or reflect—we know what the probabilities are. So those probabilites cannot be messed with. The only place where there is wiggle room for something spooky to happen is in how the actual result emerges from the probability distribution of possible results: in the jargon, the collapse of the wavefunction. However, Koch points out, there is absolutely no evidence that anything supernatural is happening here... He also torpedoes the claim that some kind of quantum computing is going on in the brain: the timescales, distances, and temperatures at which neural processes operate are such that quantum processes average out and are not significant.

Next in his talk he makes a little diversion into the subject of downward causation. This is the idea that events don't have to start from particle physics (or string theory, or whatever) and work up. They can go from big subjects like economics or politics down. You lose your job because of economics, not physics, goes the argument. As long as everything operates within the bounds of physics, it's claimed, there's no problem. But how does economics make the particles that were stationary get up (literally en masse) and walk out the door? I've never understood this idea since philosopher and computer scientist Aaron Sloman introduced it to me eight years ago or so. Christof could not make sense of it either.

So back to real science. It turns out that, even if you thought you could get a little spookiness out of quantum mechanics, the brain seems not to be set up for free will either. According to work by Benjamin Libet, we may think that we are making conscious decisions, but it turns out that our unconscious brains make the decision first, then lets our conscious brains know. This can be seen through a surge of brain activity that occurs before we are aware of having made a decision. So any sense that we are consciously weighing up our options before deciding what to do seems to be an illusion. (There's some interesting evidence on this related to hypnosis too).

So, Koch argues, even if there is some remote chance that quantum mechanical outcomes can be modulated by something supernatural, it seems like we make decisions unconsciously—without conventional deliberation—anyway.

It was a great talk, and one that I have probably not done justice to here. I recommend you listen to it yourself if you're at all interested.

Two reservations though. I wish it hadn't been presented in a school of divinity, where I strongly suspect that many in the audience will ignore all the caveats and say the soul resides in quantum behavior. I also wish it hadn't been funded by the Templeton Foundation, a group with the goal of blurring the boundaries between science and religion.