Mind & Body

Why Do You Sometimes Miss What's Right in Front of You?

Confession time. Some of us at the Curiosity offices have been known to use the flashlight function on our phones...while looking for our phones. And we're willing to bet that something similar has happened to you, too. As it turns out, it's not so uncommon to be so completely unaware of your surroundings. No, really, it's not. Now, where did we put that laptop...?

The Eyes (Don't) Have It

Share the knowledge!

Before you read any further, watch the video above.

Weird, right? It's called inattentional blindness, and it's what happens when you encounter something in a place you aren't at all expecting. No matter how strange, blatant, or eye-catching it is, our brains just don't want to notice things where we don't think they belong. That's why it's so easy to miss the bear amid the basketballs, and why you don't notice your keys when they're right by the door instead of their usual spot on your bedside table.

Here's a classic example that spotlights exactly how pronounced this effect can be. Taking their cue from the old saying about money growing on trees, researchers made their own money tree by attaching dollar bills to a tree on the quad at Western Washington University. The dollars were attached to a branch that extended out over the walking path, and the spot was chosen specifically because researchers had seen many students have to duck to avoid it or push it out of their way. And that remained true after the dollars were attached — but hardly anybody noticed.

There's another way that our brains can gloss over the details of a scene, though it's not likely to occur outside of a research scenario. Even if the object is in the place that we expect it to be, our brains will sometimes gloss over it if it's too large. In one example, people could easily spot the toothbrush on this bathroom counter, but missed the GIANT toothbrush right behind it. On first blush, it seems like an unfair test, because if somebody tells you to find the toothbrush, they aren't going to be looking for something four feet long. But here's the rub: tell an artificial intelligence to find a toothbrush, and it'll identify brushes big and small, no problem.

A Helpful Flaw

So does that mean that AI systems are better than humans and other animals at spotting what they're looking for? Actually, it's kind of the opposite. Our brains automatically jump to conclusions when given tasks like "find the toothbrush", and most of the time, those conclusions are right (or at least, well-informed). That makes us very good at finding and identifying things under normal conditions, and not so hot when their size and location have been tweaked. But while a Deep Learning network might be able to see the giant toothbrush easier than we can, it will also end up mistaking a vaguely toothbrush-shaped floor lamp for its target. You know what that means: "Terminator 2" could have been a lot shorter if they'd just set up John Connor mannequins all over Los Angeles.

Written by Reuben Westmaas November 15, 2017