Escaping flatland, or: How fish remember where they’re going

What do you do when you’re trying to get somewhere, and you get conflicting information? The person in the passenger’s seat swears he knows which way to turn, but the GPS unit in your car tells you to go the opposite direction. Problematic. Indeed, possibly relationship-changing depending on who’s in the passenger seat.

Still, humans generally have it easy when it comes to navigating. Our decisions are usually “turn left” or “turn right.” We are close to flatlanders when we are moving around our environment. Animals animals that fly, climb, or swim have to take the vertical dimension into account when getting around much more than humans do.

A new paper by Robert Holbrook and Theresa de Perera look at how fish remember positions in space. They put small tropical tetra, Astyanax fasciatus (shown below) through their, um, paces? (can fish pace?) in a classic Y-maze. One arm ends in a bare cul de sac. The other end of the rewards the fish with food and the sight of another individual of the same species. (This species of fish has an eyeless variety found in caves, but that’s not what was studied here.)

Contrary to urban legend, fish do have memories and can learn – even the much maligned goldfish. The Y-maze used in these experiments rotated, so they could train the fish to turn left or right to get to the reward, or swim up or down to get to the reward. The fish learned these two tasks at the same rates.

The authors then trained fish when the Y-maze was rotated at a 45° angle, so their choices to get the reward were, say, “up and left” or “down and right.” The fish learned this task, too. But when the maze was flattened out, the fish continued to make the correct left or right turn. But when the maze was straightened vertically, the fish couldn’t pick the correct up or down turn when the maze was oriented vertically.

The last experiment was to train the fish when the maze was slanted at 45°, then, instead of straightening in, turn the maze to the other 45° angle. Based on the experiment above, where the fish remembered the left or right turn, you might expect they would turn left or right, depending on which direction they had been rewarded. But the fish, if trained to swim up and left, would swim up and right after the maze was rotated. If trained to swim down and left, they would swim down and right after the maze was rotated.

In this experiment, the vertical dimension had primacy, not the horizontal. But the both directions weren’t the same to the fish. Fish trained to swim up took longer to learn the task than those trained to swim down.

The authors don’t have a ready explanation for why the horizontal “left / right” memory gets primacy in one experiment but not the other. Their bigger point is that the two components are processed by the fish independently of each other. This would probably be quite interesting to neuroethologists studying spatial navigation, who could probably start to look at what sorts of neural circuits might handle these two lines of information.

A concern with this paper is the small sample sizes in the experiments. They used only four or five fish for each one. Would it have been so hard to get to double digits, guys? Then at at least you could use some more powerful statistics. Combined with the the apparent ease of conducting the experiments, it makes this paper feels very slight (much like another recent paper in this journal).

Reference

Holbrook, R., & Burt de Perera, T. (2009). Separate encoding of vertical and horizontal components of space during orientation in fish Animal Behaviour DOI: 10.1016/j.anbehav.2009.03.021