Monday, April 29, 2013

Ben Thomas Blog: Web Portals and Mouse Mazes: How Your Brain Sorts the World

We're drowning in information. Every day, a whopping 2.5 quintillion bytes of new data appears across the Internet -- the tweets alone contain more textual data than your hard drive can hold. So we rely on search engines and RSS aggregators to track down and organize the data that's most useful to us, in much the same way as our ancestors relied on encyclopedias and almanacs.

This need to organize data isn't even a new phenomenon on the Internet. From the earliest days of the World Wide Web, the homepages of Yahoo and AOL (among others) provided "portals" to useful pages on other sites -- vouching, in effect, for these other pages' relevance and legitimacy.
Perhaps more surprising is the fact that, even in the age of Google, portal pages continue to proliferate.

From BuzzFeed's up-to-the-minute meme roundups to Reddit's constantly updated rankings of popular links, curated content continues to shape our Internet voyages.

In fact, these websites about websites reflect a central principle in human psychology: Our instinct for generalizing; for grouping our experiences under headings, then grouping those headings under bigger headings.

Not only is this ability a sign of intelligence; it's also crucial to the organization of our brains -- and to how we interact with the world around us.

The principle is known as "invariant representation," and it's as easy to explain as it's baffling to study.

Take vision, for example; your ability to see. The simplest visual areas of your brain -- the areas where signals from your eyes arrive earliest -- respond only to basic components of shapes, such as edges, corners and contrasts.

The neural signals triggered by those shape-components then get passed up to more complex areas, which respond to larger groups of visual signals, like those encoding whole shapes or movements.

This process continues all the way up through the brain's chain of communication, until eventually a signal arrives at a group of nerve cells that responds only to faces, or to views of scenes, or even -- believe it or not -- to cute things.

As a general rule, cell groups that sit higher in the hierarchy tend to respond to more complex or specific concepts -- so a group of cells that responds to cats might respond to a cat you see strutting by the window, to the sound of a cat's claws on your countertop, or even -- in some cases -- to the memory of the cat you had in high school.

This is what scientists mean when they talk about "invariant representation" -- the more sensory signals your brain learns to associate with cats (or with a particular cat), the more neural pathways can lead to the activation of that "cat" cell group, and the more persistent the concept of "cat" becomes in your memory.

This is a super-simplified version of the process, as I'm sure you can tell. There's probably no group of cells in your brain that responds only to cats and to nothing else; nor is it likely that any one particular brain cell, or group of cells, responds only to memories of your childhood cat.

Still, the process of generalized category representation is one that scientists have discovered again and again, all throughout the cerebrum, from the brain's relatively simple visual layers all the way up to abstract-planning areas like the prefrontal cortex. From the cellular level up, our brains are evolved to categorize.

Why are we so good at forming categories? The short answer is, complex categories help us make long-term, large scale predictions about the world.

A mouse can learn the layout of a maze, but -- as far as we know -- it doesn't suddenly pause in the middle of maze #317 and wonder why it has to keep running mazes every day. A mouse's brain probably isn't physically able to build categories that big, so its understanding of the past -- and its predictions about the future -- just can't reach that far.

Unlike a mouse's brain, yours is amazingly talented at building bigger, more complex, more precise categories. Just about every category in your mind was forged by your brain's trial-and-error attempts to predict the future based on your past.

Even your brain's "invariant" representations aren't necessarily immune to modification -- they can change quite a bit as you learn new details about the world. What's more, every representation in your brain is deeply intertwined with others -- which is why it's hard to picture the ocean without imagining the sound of crashing waves, or to say the words to your favorite song without thinking of its melody too.

It wasn't too long ago that it seemed we'd always need human brains to categorize the information around us. More than ever before, though, search engines are learning algorithmically from our behavior, tailoring their results to our preferred corners of reality.

Advertisers are starting to predict what we're likely to buy before we even know we want it. It's plausible that within our lifetimes, software programs will learn to pick up on data patterns whose existence we'd never imagined. And what happens beyond that could be hard for even the smartest human brains to predict.

Ben Thomas is an author, journalist, inventor and independent researcher who studies consciousness and the brain. A lifelong lover of all things mysterious and unexplained, he weaves tales from the
frontiers of science into videos, podcasts and unique multimedia events. Lots more of his work is available at .


No comments:

Post a Comment