Friday, August 12, 2016

This is Your Brain on Physics

Like the physics engine in a video game that brings to life car crashes, nosedives, touchdown passes, and other physical events, humans may have a kind of “physics engine” in the brain that helps us survive. After all, even non-physicists quickly swerve to miss an oncoming car, duck to avoid being hit, and reflexively catch falling objects.

New research published early this week in the Proceedings of the National Academy of Science supports the physics engine model, and shows that our intuitive understanding of physics is closely linked to the regions of the brain that help us prepare to take action.

The location of the 'physics engine' in the brain is highlighted in color in this illustration.
Image Credit: Jason Fischer/JHU
Composed of cognition experts, the research team includes Jason Fischer, John G. Mikhael, Joshua Tenenbaum, and Nancy Kanwisher. All of them are affiliated with MIT; Fisher is also with Johns Hopkins and Mikhael is with Harvard Medical School.

Rather than attempting to get every detail right, like a physics teacher might require, software-powered physics engines take shortcuts. With this approach they can get most things just about right, and they can do it quickly. This way, games can generate the outcome of a car crashing into a race barrier in real-time. Might this also be how the brain works?

This idea has been explored from other angles, but the new study is the first one to look at the areas of the brain involved in interpreting physical events. The study involved four experiments, each building on the previous one. All of them took place while participants were undergoing fMRIs (functional magnetic resonance imaging). fMRI is a brain scanning technique that measures and maps brain activity.

Experiment 1
To see if/how brain activity differs when you judge a scenario using physics intuition versus judging it visually, twelve people watched movies showing unstable block towers. Participants were asked to judge either where the blocks would land if the tower toppled, or whether the tower had more blue or yellow blocks. (You can test your brain’s physics engine with sample videos on YouTube.)

The result? Several areas of the brain responded more strongly when participants predicted how the blocks would fall. The researchers called these “regions of interest.”

Experiment 2
To explore whether these regions of interest were specific to physics intuition or just reflected the process of making predictions or using spatial reasoning, the participants watched movies starring two dots. The dots interacted either like billiard balls following the laws of physics, or like people interacting socially (for example, one could chase the other). After eight seconds one dot disappeared. Two seconds later it reappeared. Participants predicted the path of the invisible dot during those two seconds and reported whether they were correct.

The result: Five “regions of interest” responded significantly more when the prediction required physics intuition. Some areas of the brain were active in both cases, likely those involved in general problem-solving, prediction, and spatial reasoning.

Experiment 3
To explore whether looking at a physics event is enough to cause a strong reaction in the five regions of interest, participants watched short movies with varying amounts of physics content (like a ball rolling). They weren’t asked to perform any tasks, just to watch.
The result? The more physics content in a movie, the stronger the responses in the regions of interest.

Experiment 4
So far so good, but what if these areas just correspond to parts of the brain involved in lots of different kinds of activities? The brain has several regions like this, together called the multiple demand network. The multiple demand network usually responds stronger to more difficult tasks, so the researchers had the participants from the first two experiments perform tasks not requiring physics intuition that had varying degrees of difficulty.

The result: The areas of brain activity overlapped with the regions of interest, but other areas responded as well.

A detailed analysis of the fourth experiment indicates that the areas of the brain involved in physical intuition are a subset of the multiple demand network. In fact, they seem to be the same areas involved in motor planning and tool use—the ones we use to figure out how to respond to a situation.

Why is physics intuition so closely related to action planning? The authors propose two possibilities. Babies create physical models of the word as they learn to crawl, push, pull, and in general practice their motor skills. Maybe physics intuition and action planning just grow up together. Or, maybe it’s because in order to plan any kind of physical action (run, catch, duck), we need to have a good functional understanding of physics in the first place. Hopefully future work will shed light on the whys and hows.

Now if only we had some quantum mechanics intuition too...

Kendra Redmond

No comments:

Post a Comment