David Le Page
Takeuchi was a tiny robot built from two cockroach legs and a silicon chip. An innocuous cyborg, he toddled for up to an hour at a time across flat surfaces at the University of Tokyo, where researchers built him in 1996.
The little machine had a conventional silicon chip micro-controller, an artificial body and was impelled by electrical pulses stimulating the cockroach legs.
But last year scientists probing methods for wiring up beings and machines moved their methods up the evolutionary ladder.
A lamprey recently contributed under general anaesthetic the most crucial parts of its anatomy to science.
The robot uses just a few neurons from what passes for a brain stem in the sea lamprey Petromyzon marinus to receive sensory information from electronic light sensors and translate it into “motor” nerve impulses.
Those impulses drive not cockroach legs, but a small, two-wheeled robot. The nerve cells are kept alive in a constantly oxygenated solution of salt water.
Ferdinando Mussa-Ivaldi of Northwestern University in Chicago and his co-researchers at the universities of Illinois and Genoa call the result, which shows distinct reactions to light stimuli, an “artificial animal”.
The current fascination of computer scientists and roboticists with biology stems in part from the failures of traditional artificial intelligence (AI) research. The 1956 prediction of pioneer AI researchers Allan Newell and Herbert Simon that intelligent machines would exist by 1970 has proved to be far off the mark. More recently noted roboticist and futurist Has Moravec has predicted that personal computers with processing power approximating that of the human brain will exist by 2030.
Of course, it’s not much use having the processing power if you don’t have the software to run it. Numerous attempts by AI researchers to somehow distil from scratch the “principles” that might combine to make a mind failed to achieve the desired results. Of late, the pioneers of god-level engineering have turned to another approach rather than trying to build minds, they are trying to create the conditions in which minds may create themselves.
Increasingly, they are looking to biology for inspiration in this work, designing robots that will design themselves and machines that will submit to Darwinian selection processes. The construction of minds and the bodies that will house them Takeuchi and the lamprey-mobile are great works proceeding apace towards the goal of artificial life, not to mention the blurring of the definition of life itself.
Miguel Nicolelis of Duke University in North Carolina hooked up a monkey’s brain to a robotic arm. In the age of the Internet, though, merely hooking it up to a machine five feet away from the monkey itself might have seemed somewhat unimaginative, so Nicolelis eventually hooked up the monkey to a machine 950km away. The monkey had learned tasks such as reaching for food placed at different locations on a tray.
The research is something of a breakthrough in that it represents the first time researchers have managed to monitor and process simultaneously impulses from so many nerve cells to produce complex real-time motions, with obvious possible applications for quadriplegics.
Quadriplegics, though, are more likely to be excited by the prospect of being wired up to their own limbs again, than by the possibility of controlling robotic arms. In the absence of the ability to repair the biological “circuitry” that runs through the nervous system, such apparently ghoulish technologies represent great hope for people whose lives have been shattered along with their spines.
It is not just machines for motion that can now be electronically wired into creatures of flesh and bone. Thousands of deaf people already benefit from cochlear implants that doesn’t just amplify sound they actually create signals fed straight into the auditory nerve and interpreted as sound.
The same principle is now being applied to vision. The Optobionic chip contains 3 500 microscopic solar cells. Light entering the eye creates electricity in the cells and stimulates the optic nerve.
But the Optobionic chip is a sub-retinal chip. Epiretinal implants, by contrast, actually fit on top of the retina and stimulate it with amplified signals often supplied by an external source such as a video recorder. Patients hooked up to such devices have been able to identify colours and letter shapes. Unfortunately, it is estimated that 15 years of research will be needed before such devices become useful.
Such research often seems to intrude crudely into the mysteries of life. Yet in a few cases it has served to turn up results of great poetry.
Scientists at Chicago University led by Daniel Margoliash hooked up monitors to the brains of zebra finches and discovered that the tiny birds dream, and what is more, dream of the songs they will sing when awake. The patterns of firing neurons in the brains of a slumbering zebra finch closely resemble those in the brain of a zebra finch gripped by the ecstasy of song.
US scientists have developed implants that can be permanently inserted directly into the motor cortex of the brain. They contain tiny radio transmitters, so the need to be permanently wired up has been eliminated. Simply by thinking, disabled patients can select icons on a computer screen by moving a cursor.
Of course, such research can be controversial. Moravec believes it is leading to nothing less than the day in which our minds can be uploaded into machines. This is by now such a hackneyed feature of science fiction that invoking it almost spoils the novelty of the research from which it may ultimately develop.
Yet Moravec is taken seriously because he has a good track record on such predictions. In 1988 he predicted the current state of the robotic art and the proliferation of personal computers. In 100 years time, he believes, the lines between minds, humans and machines will have become irretrievably blurred.
We’re getting closer to our machines, into them and they into us. We can only hope that they’ll see some point to keeping us around, or that psychiatry develops apace with the software.
ENDS