On 07/12/2011 12:36 PM, Invisible wrote:

> So let's hear it for Physarum polycephalum!

This stuff is weird in that it's large enough to see with the naked eye 
(indeed, it can be multiple centimetres across), and yet it's a 
single-celled organism.

Reading further, it turns out that it has multiple nuclei, but only a 
single cell wall. In other words, it's a multicellular organism, just 
without the internal cell walls.

That doesn't make its movement any less amazing, mind you. The other 
night I was watching time-lapse photography of it moving on TV. One of 
the links above shows a lovely photograph of the fractal networks it 
produces. And as it moves it sort of pulsates.

Being a "slime mould", it likes moisture and avoids dryness. It avoids 
light, presumably because that might dessicate it. It feeds by slowly 
spreading out in all directions, until it finds something edible. Then a 
blob of slime engulfs the food, and the rest of the fan dwindles to 
nothing, leaving only a network of thin tubes connecting the food pieces 
together.

In one experiment, they arranged pieces of food in the same positions as 
the cities around Tokyo. The mould spread out, found them all, and 
produced a network of tubes remarkably similar to the transport network 
around Tokyo. Of course, the transport network was designed by a huge 
team of engineers with enormous brains which consume vast amounts of 
power. The slime mould did the same thing without any brain cells at 
all. (Indeed, the entire network is one continuous macroscopic cell.)

In another experiment, they used the slime mould to control a robot. 
It's not clear what they were actually trying to achieve, or what part 
the mould actually played. The only Internet documentation I could find 
claimed that the mould avoids light, and they used that to make the 
robot avoid light. Personally I don't see how having a big complex 
microprocessor delegate a trivially implementable function to an organic 
entity is particularly "useful", but hey.

Slime moulds can also navigate mazes. Combined with the transport 
network result, this demonstrates that the organism clearly has 
computational abilities. But unlike digital computers, this thing is 
massively parallel. There may yet be some way in which that could be 
useful. Then again, it has a top speed of 1 cm per hour. So we're 
looking at problems that currently take insane amounts of time, but can 
be attacked in a very parallel way. Protein folding, perhaps. Or weather 
simulation. Something like that.

The organism also has the ability to learn from past events and 
anticipate future ones. It's not surprising that an organism should 
evolve such an ability. It is obviously desirable. What it surprising is 
that a single-celled organism is sufficiently sophisticated to implement 
it. (From the documentation I've seen, it's not clear to me how much 
complexity it can cope with. It's still impressive though.)

Post a reply to this message