Study suggests our brains crunch physics like video games

A study conducted at the Massachusetts Institute of Technology proposes that human brains include an "intuitive physics engine" that acts like a video game engine and allows us to understand things like when a stack of dishes is about to topple.

In "Simulation as an engine of physical scene understanding," Peter W. Battaglia, Jessica B. Hamrick and Joshua B. Tenenbaum describe the mental model we use to "make robust and fast inferences in complex natural scenes where crucial information is unobserved." The paper was published in volume 110 of Proceedings of the National Academy of Sciences (PNAS).

As explained in the research's Supporting Information (PDF link), to test the phenomenon, researchers presented their subjects with a computer simulation of stacked 3D blocks. In one test, blocks were added to the stack, and subjects had to answer the question "Will this tower fall?" and rate the likelihood on a scale of 1 to 7, a range that spanned from "definitely will not fall" to "definitely will fall." Related experiments asked participants to determine in which direction the tower will fall and how far blocks would travel if the table on which they rested were to be bumped.

Definitely will not fall. Definitely will fall.

Based on their analysis of the data collected, the authors theorize that we use our past observations to understand how objects will behave, which allows us to make assumptions about real-world physics even when our knowledge of the situation is incomplete.

"We propose a model based on an 'intuitive physics engine,' a cognitive mechanism similar to computer engines that simulate rich physics in video games and graphics, but that uses approximate, probabilistic simulations to make robust and fast inferences in complex natural scenes where crucial information is unobserved," the paper's abstract reads. "This single model fits data from five distinct psychophysical tasks, captures several illusions and biases, and explains core aspects of human mental models and common-sense reasoning that are instrumental to how humans understand their everyday world."

Unlike a video game physics engine, the human mental model uses past data to infer quickly what will happen in the future, whether that's in Angry Birds, Boom Blox or while holding a plastic bag that's about to rip.

"What happened in the past, what will happen — the way you form those judgments is based on these simulations supported by an intuitive physics engine," Battaglia, told Co.Design. "It's a very powerful way of making these predictions."