[LINK] "Brainless creature solves problems with memories of slime"

[rfmcdonald]

Not Exactly Rocket Science's Ed Yong reports on a new paper (Reid, Latty, Dussutour & Beekman, "Slime mold uses an externalized spatial “memory” to navigate in complex environments", in PNAS) reports on how the slime mold Physarum polycephalum has a remarkable sort of external memory. Yes, a slime mold.

Together with Audrey Dussutour from the CNRS in France, [Chris Reid from the University of Sydney] found that Physarum strongly avoids ground that it has already laced with slime. “We have been researching the problem-solving abilities of the slime mould for a few years, and we had noticed that it seemed to avoid areas it had been before,” says Reid.

He proved this reticence by providing Physarum with a Y-shaped arena, where one arm was covered in slime and another was not. It almost always chose to forge new ground. This isn’t a reflex – it’s more of a choice. If there aren’t any slime-free options available, then Phsyarum will willingly re-tread a slimy path.

Next, Reid and Dussutour confronted the slime mould with a U-shaped trap – a challenge commonly used to test autonomous robots. Physarum starts at the top of a jelly-filled Petri dish, and its mission is to reach a goal at the bottom. The goal is rich in sugar, which diffuses through the jelly and creates a gradient that Physarum can track. But in the way, there’s a U-shaped barrier. If the plasmodium heads straight to its goal, it will get stuck. It needs to navigate around the U, and it can do so if it uses its slime to avoid areas it has been to before. Without the slime, it wanders blindly.

Reid and Dussutour placed the slime mould on different Petri dishes—some had blank jelly and others had jelly already coated with slime, which masked Physarum’s own trails. In the blank-jelly dishes, almost all the slime moulds eventually reached their goal, in an average time of just under 60 hours. By contrast, in the pre-slimed dishes, just a third of the moulds reached the goals, and those that did took around 75 hours. Unable to sense their own trails, they spent more than 10 times as long going over places they had previously travelled.

When we think about navigating the world, we might initially think of our own maps, or of the impressive migrations undertaken by familiar animals. But navigation can happen without much brainpower. Social insects like ants can create efficient trails linking their nests and sources of food, by laying down trails of pheromones in their wake. As more of them reach the food, they add their own pheromones, making the routes even more attractive to other ants. If the trails aren’t reinforced, the pheromones evaporate. That’s exactly like human hikers, whose collective feet trample effective trails into the landscape, while allowing boring or inefficient trails to overgrow.

The slime mould takes these ideas even further – it can navigate without a brain at all, using chemicals that it deposits onto the environment.

Engineers have hit upon similar solutions when trying to programme independent robots that can find their own way around. It’s too complicated to equip the bots with a pre-constructed map, or to programme them to build such maps as they explore. At the other end of the complexity scale, if the robots simple react to their local environment, they can move about easily enough but often get trapped by obstacles, like a fly buzzing at a window. But add a simple imperative to “avoid the past” and suddenly these machines become much better at skirting round obstacles, and navigating complicated environments.

The slime mould achieves the same feat by secreting slime, and then avoiding it. Reid suspects that such external memories preceded internal ones, and ancient creatures used them to solve problems, in a similar way to slime moulds, long before neurons and brains evolved. “This has been hypothesised before, but no one has been able to provide any evidence to support the theory, until now,” he says.