Fish in an electric maze

For many, this is the most familiar image of animal behaviour:

The rodent in the maze is such a well-known image, you can find countless cartoons alluding to it. And with good reason: we've learned a lot by teaching animals to perform these kinds of maze running tasks. By tweaking the task, we can learn much about the kinds of cues that animals pay attention to in learning.

Walten and Moller put a new spin on this old classic by using a different kind of animal: electric fish. Fish do not have reputations as being good at navigation. You only have to think about the old, busted myth that goldfish only have a 5 second memory. But, to some degree, it might make a certain amount of sense if fish were bad at the navigating: after all, there aren't that many permanent positions in the water.

With many animals, the trick is finding a task that motivates them. This species, Mormyrus rume, doesn't like to be out in light and bare tanks, much preferring small, dark, hidey-holes. Indeed, the authors say that the animal has a home base that it roams away from, and has to learn its way back. When an electric fish was placed at one end of an open tank with a nice dark spot they could escape into, they quickly learned how to get there.

The interesting thing here is that the electric fish sees the world though a few different sense, including one for which we have no equivalent. The fish can see the maze. The fish can also sense the way electricity "flows" through the maze.

The authors trained the fish, letting them use either vision or electrosense during the learning phase.

The maze was dead simple; we really wouldn’t even consider it a maze, given that there were no turns or decision points where the fish had to remember, “Do I got left, or right?” Basically, the maze was just a long corridor that went back and forth.

In some versions of the maze, the authors added extra visual cues: big “signs” sitting outside the tank. In others, they made the maze more distinct electrically by placing aluminum or plastic cubes at key points in the maze.

To test what they fish were doing, they “knocked out” one sense at a time. They removed vision by letting the fish do the maze in the dark. With a little surgery, they could knock out the fish’s electosense.

After the fish learned the maze (intact animals got to about half a minute within 3 days; those ), they took the walls and the shelter away. There literally wasn’t a maze any more – the fish could just swim straight to the place the hole had been if they wanted to.

These are a few possibilities as to what the fish might do.

• It might just swim straight towards the place where “home” used to be; this would indicate it had a sort of mental “map” of the tank.
• It might actually do the whole turning back and forth, indicating it had memorized a pattern of movement.
• It might be completely discombobulated and not be able to find anything.

When the visual cues were left outside the maze, the fish still tended to meander a bit. If you moved the starting point, though, the fish would not meander – they would treat it as a new situation. All of which suggests the fish, even though they can see, aren’t paying attention to those visual cues.

When the electrical cues remained in the maze, the fish tended to use them. If the authors threw in some new ones, the fish got very lost. The most interesting situation was when the electrically important cues were removed: the fish zipped straight to the target. This suggests that the fish had formed a map, but only fell back on it when the old dependable cues weren’t available any more.

Suffice to say, what’s going on when these fish are learning these mazes is fairly complicated. They fish mostly use electrosensing, but learn better if they can see. They will use electrical cues if they’re there, but have other ways to find their way if they don’t. I’ve left out a batch of other variations and controls and such that Walton and Moller performed.

The maze learning behaviour is just the first tip-toe to looking at the neurobiology, though. The authors don’t have some very specific hypotheses about a new form of protein kinase C that they expect to find in the brain of these fish...

Reference

Walton AG, & Moller P (2010). Maze learning and recall in a weakly electric fish, Mormyrus
rume proboscirostris Boulenger (Mormyridae, Teleostei) Ethology : 10.1111/j.1439-0310.2010.01807.x

Rat photo by 19melissa68 at Flickr. Used under a Creative Commons license.