“The large camera array works the way you would track actors for making Gollum [in the ‘Lord of the Rings’] or other types of animation,” Marshall said. “We use it to precisely measure the position of their head and limbs and then there’s a lot of analysis on top of that that enables the detection of behaviors and the isolation of other components [like behavioral organization].”

Marshall spent months researching the constraints and advantages of existing movement technologies before settling on motion capture. He then spent six months figuring out what type of marker would stay attached for long periods of time.

“Traditional markers from Hollywood are made of foam, and that wasn’t going to fly with rats,” Marshall said.

Working with local veterinarians, the team designed custom body piercings made of specialized reflective glass and attached them to 20 locations on the rats’ bodies. With the markers in place, they let the rats explore a naturalistic area and tracked their movements using the cameras 24/7 for weeks.

The researchers mapped out natural behaviors such as grooming, rearing, and walking and showed how those movements are organized into structured patterns, like the arrangement of words into sentences.

The team then looked at how those behaviors and patterns changed in response to two stimulants, caffeine and amphetamine. As expected, both drugs caused the rats to move around more, but they did so in different ways. Caffeine amped them up, but they explored their cage normally. Amphetamine, on the other hand, shifted their behavior in novel ways and made them quite disturbed, Marshall said. They ran around in repeated, sequential patterns.


Movement monitor

An open-source AI tool for studying movement across behaviors and species

From fins to limbs and water to land

Harvard scientists reconstruct the evolution of terrestrial movement in early tetrapods

When the team studied rats with a form of autism, the data showed another surprise.

The scientists saw the rats perform abnormal grooming patterns that hadn’t been described before. Grooming pattern alterations could be an important indicator used to model repetitive movements observed in people with autism, but they traditionally have been difficult to measure. The scientists say detecting these types of subtle and precise behavioral deficits are important in getting a better handle on many diseases and could be one of the prime uses of CAPTURE.

Other efforts to expand CAPTURE include combining their data with recordings of neural activity to map the relationship between brain signals and behavior across the full set of natural movements a rat performs. They are also working with Google DeepMind to use CAPTURE to help model how the brain produces behavior, and potentially to make new advances in artificial intelligence.

“We’re only going to go deeper,” Marshall said.

This research was supported with funding from the Helen Hay Whitney Foundation, the National Institutes of Health, the National Science Foundation, and the Segal Family Foundation.