Monitoring Brain Activity During Studying to Predict Test Performance
A research team led by Laura Matzen at Sandia National Laboratories in Albuqurque, NM has demonstrated that it is possible to predict how well people will remember information by monitoring their brain activity while studying. Matzen’s team monitored test volunteers with electroencephalography (EEG) sensors to make accurate predictions. Why bother making a prediction if the result will show how well someone remembered the information anyways? Matzen brought up this example, “if you had someone learning new material and you were recording the EEG, you might be able to tell them, ‘You’re going to forget this, you should study this again,’ or tell them, ‘OK, you got it and go on to the next thing.” Essentially providing a real-time performance metric, the applications of which many students would appreciate.
The team monitored test subjects’ brain activity while they studied word lists, then used the EEG data collected during the trial to predict who would remember the most information. Researchers had a baseline of what brain activity looked like for good and poor memory performance, so they knew the average percentage of correct answers under various conditions. The computer model predicted five of 23 people tested would perform best, based on their EEG scans. And the model was correct, they remembered 72 percent of the words on average, compared to 45 percent for everyone else.
This study is part of Matzen’s overarching research goal to understand the Difference Related to Subsequent Memory, or Dm Effect. The Dm effect is a measure of brain activity that can distinguish remembered items from forgotten ones. A measurable difference would give cognitive neuroscientists a way to test hypotheses about how information is encoded in memory. Matzen is interested in not only what causes the effect, but also how to change it; she wants to discover how different methods of training can help people performing at different levels. That’s why the second half of this study was done, to predict who would benefit most from memory training.
This second half of the study tested different types of memory training to see how they changed participants’ memory performance and brain activity. This study, still in its preliminary stages, aims to find out whether recording partcipants’ brain activity while they use their natural approach to studying can predict what kind of training would work best for them. The computer model from the first half of the study was used to predict who would perform best on the memory tasks, and after memory training, the high performers did even better.
90 volunteers spent 9 to 16 hours over five weeks in the memory training study. The first half provided a baseline for how well they remembered words or images. Most then underwent memory training for three weeks and were retested. The control group received no training, one group practiced mental imagery strategy, thinking up vivid images to remember words and pictures, and the final group went through working memory training to increase how much information they could handle at a time. Each volunteer, shut into a sound-proof booth, watched a screen that flashed words or images for one second, interrupted with periodic quizzes on how well the person remembered what was shown.
The test was divided into five sections, each about 20 minutes long and testing a different type of memory. The first, middle, and last sections consisted of single nouns. During quizzes, volunteers hit buttons for yes or no to whether they’d seen the word before. The other two sections combined adjectives and nouns or pairs of unrelated drawings, and volunteers were tested on what they remembered. The image section tested associative memory or memory for two unrelated things, which according to Matzen is the most difficult because it links arbitrary relationships.
When performance was compared before and after training, the control group did not change, but the mental imagery group’s performance improved on three of the five tasks. “Imagery is a really powerful strategy for grouping things and making them more memorable,” Matzen said.
The working memory group did worse on four of the five tasks after training. Volunteers trained on working memory, remembering information for brief periods, improved on the task they had trained on, but that training did not carry over to other tasks. Matzen believes the difference between the two groups boils down to strategy: The imagery training group learned a strategy, while the working memory training group simply tried to push the limits of memory capacity.
While the imagery group did better overall, they made more mistakes than the other groups when tested on “lures” that were similar to, but not the same as, items they had memorized. “They study things like ‘strong adhesive’ and ‘secret password,’ and then I might test them on ‘strong password,’ which they didn’t see, but they saw both parts of it,” Matzen said. “The people who have done the imagery training make many more mistakes on the recombinations that keep the same concept. If something kind of fits with their mental image they’ll say yes to it even if it’s not quite what they saw before.”
What’s next? Matzen and the Center for the Advanced Study of Language at the University of Maryland plan to study tasks that measure cognitive flexibility and how it relates to training performance, working on understanding and affecting the Dm affect.