Babies make predictions, too

In “The Adventure of Silver Blaze,” about a valuable racehorse that mysteriously disappears, Sherlock Holmes tells the hapless Detective Gregory to note the curious incident of the dog in the nighttime. But, says Gregory, the dog did nothing in the nighttime. That was the curious incident, Holmes replies—the dog didn’t bark on the night of the crime, as you would expect. A new study suggests that as babies start to figure out the world, they think a lot like Sherlock.

People often say that babies are “sponges.” The metaphor reflects a common picture of how the brain works: Information floods into our eyes and ears and soaks into our brains, gradually becoming more abstract and complex. This image of the brain is vividly captured in the “abstract thought zone” of the recent animated Pixar movie “Inside Out”—where three-dimensional experiences are transformed into flat cubist ideas.

But a very different picture, called “predictive coding,” has been making a big splash in neuroscience lately. This picture says that most of the action in the brain comes from the top down. The brain is a prediction machine. It maintains abstract models of the world, and those abstract models generate predictions about what we will see and hear. The brain keeps track of how well those predictions fit with the actual information coming into our eyes and ears, and it notes discrepancies.