Computer algorithm, MRI used to tap memory

Brain scans identify which short film clip volunteers were remembering, which could help researchers better understand memory recording patterns and, ultimately, dementia.

Elizabeth Armstrong Moore
Elizabeth Armstrong Moore is based in Portland, Oregon, and has written for Wired, The Christian Science Monitor, and public radio. Her semi-obscure hobbies include climbing, billiards, board games that take up a lot of space, and piano.
Elizabeth Armstrong Moore
2 min read

A computer algorithm, functional Magnetic Resonance Imaging (fMRI), and neuroscientists working together have been able to identify what people are remembering by measuring blood flow levels, according to new research out of the University College London.

First, a group of 10 volunteers (average age 21) was shown three very short (as in 7 seconds) films, each of a woman on a city street doing a simple task, such as mailing a letter. Then, each of the volunteers was placed inside an fMRI scanner and asked to recall each film, first in a specific order, then at random.

One of the three short films showed this woman mailing a letter. Wellcome Trust Centre for Neuroimaging/UCL

Using the scanner to measure changes in the brain's blood flow and a computer algorithm, researchers were able to identify which short film each person was remembering at a level the study's lead author describes in a news release to be "significantly above what would be expected by chance."

"This suggests that our memories are recorded in a regular pattern," says Martin Chadwick, who conducted the research at the University College London's Wellcome Trust Centre for Neuroimaging. The article appears in the journal Current Biology.

The researchers homed in on the medial temporal lobe, a region of the brain thought to be most involved in episodic memory. The computer algorithm performed best when analyzing the hippocampus, which the team has already studied.

Across all participants, the rear right, front left, and front right areas of the hippocampus appeared to be consistently involved. While it remains unclear exactly what role the front two regions play, the rear right was found in a previous study--identifying where a person was standing in a virtual reality simulation--to be where spatial information is recorded.

"Now that we are developing a clearer picture of how our memories are stored, we hope to examine how they are affected by time, the aging process and by brain injury," says Eleanor Maguire, who helmed the study as an extension of last year's work on spatial memory.

It's arguable whether the algorithm's accuracy is good enough to celebrate, since it identified the film being thought about correctly less than half of the time (40 to 45 percent, to be precise). But that is better than 33 percent, which is the rate at which a blind guess between three films would be accurate.

Probably the most exciting discovery is that the memory traces associated with each film were consistent throughout not only the study, but from one volunteer to the next, suggesting that memories may in fact have some sort of fixed, identifiable pattern.