Okay, if you haven't watched the video yet, go back and do it, otherwise you're not going to have a clue what I'm talking about. I'll wait.

All done? Good.

To start off, imagine creating a goldfish-chase game for your cat. The goldfish swim along the screen (which should be placed on the floor) and as the cat pounces, they scatter. If kitty is accurate enough, he gets one.

Now consider what happens when you map the behaviours of the cat to certain applications that affect its environment. Turn the screen into a gray-scale gradient, white on one side and black on the other. As the cat walks from light to dark, the room lights dim, and they brighten when it walks from dark to light. Now kitty can control the lights.

There's an important distinction to be made here between this interface and, say, training your cat to swipe at a light switch. Animals aren't good at abstractions, nor do they have the same sense of cause and effect as we do. You could fool yourself into thinking that your cat understands that flicking the light switch causes the lights to go on or off, but that's not what's going on. Your cat thinks in sequences: run-jump-swatbeigelightswitch-lightchanges. Paint your light switch a different colour, and the cat will no longer try to use it; it would have to learn a whole new behaviour, run-jump-swatblacklightswitch-lightchanges. That's as abstract as a cat will get. They also perceive movement better than us, but detail less well--so it's hard for them to even see a light switch much less tell what its positions mean.

The multitouch interface is perfectly suited to such sequential, non-causal thinking, and it can be optimized for the sensory system of non-humans. The grayscale gradient, for instance.

So you could design a system that lets your cat take care of its own needs on a whole new level. Do you actually know whether your pet cares if the curtains are open or the lights on? How would you? It's not going to work to design a system to communicate with your cat about these things, because they just don't operate that way. Nor can you create an interface that abstracts features of the environment into symbols, like icons, and expect your cat to use it.

But say there's a little picture of cat kibble on the screen (jiggling, so your cat can see it). After feeding time the picture gets small, and gradually grows until it's feeding time again. By digging at the screen or nosing it your cat can bring the kibble closer, possibly speeding up the next feeding if he's ravenous. With the right kind of flex and resistance in the system, you could end up with something where the cat's feeding time optimizes itself, and the cat gets fed when he's properly hungry but not before, and not after.

You could use such interfaces with farm animals as well, say, horses or even chickens. The screen is just a responsive spot in the environment, after all; if the application is properly designed there's no abstraction required to use it. The result might be an augmentation of our animals' minds that we've never imagined. If Temple Grandin is right and animals are as smart as we are in their particular domains of interest, such an application could be a revelation to us. It might even make us all vegetarians (at least until they start manufacturing meat from stem cells).

Don't be surprised, though, if your cat uses the the product of billions of dollars and millions of man-hours of human ingenuity to do nothing but chase goldfish until he gets dizzy and falls down.

Cats will be cats.

<  Finally Some Common Sense | The Finite Universe  >