Possible Marker Found to Predict Long-term Learning
I had a new role, and it gave me an idea.
In 2012 I began work as Associate Chief of Staff for Education at the Veterans Affairs Boston Healthcare System. In this role (which includes the Designated Institutional Official for graduate medical education) I am responsible for the education of the more than 3000 trainees who rotate through VA Boston each year, including over 400 medical students and 1300 residents and fellows. As a researcher who has spent my career examining learning and memory in both healthy individuals and those with memory disorders, my first thought was, "What can I do to improve learning across the medical center?"
Before I could get started, I realized that I needed to have an outcome to measure. I considered surveys of how much trainees either enjoyed their rotations or learned from them, but dismissed that idea. Enjoyment doesn’t necessarily correlate with learning success, and students’ impressions of how much they have learned are often inaccurate1. Testing the students on their knowledge seemed like the right thing, but I ran into another problem: one can perform well on a test but not remember the information long-term1. After cramming for a test, for example, a student may forget the majority of the information in a matter of weeks! In the healthcare professions we must ensure that the knowledge gained by trainees will endure throughout their careers. It was for this reason that I began to look for a marker of long-lasting learning.
We started our exploration using event-related potentials (ERPs), which is simply measuring electroencephalogram (EEG) activity time-locked to a stimulus—such as a remembered word appearing on a computer screen. When you average across a bunch of items, the background EEG activity is averaged away, and you can actually see the brain activity related to remembering (or not remembering!) the item2.
We recorded ERPs on first year medical students taking anatomy for the first time, measuring their responses to anatomical terms at three times: prior to the course, right at the end of the course, and—importantly—6 months later. For each term presented in the three sessions the students reported whether they could define the term (confirmed later by a test), were familiar with the term, or didn’t know it. We then sorted the ERP data obtained immediately at the end of the course for each individual term by their performance at the 6-month timepoint. We found a small left-parietal region, previously associated with detailed recollection of memory, that was actually able to predict whether the student could define the anatomic term 6 months in the future (see figures). This study was published in the Journal of Cognitive Neuroscience. Turk KW, Elshaar AA, Deason RG, Heyworth NC, Nagle C, Frustace B, Flannery S, Zumwalt A, Budson AE. Late Positive Component Event-related Potential Amplitude Predicts Long-term Classroom-based Learning. J Cogn Neurosci. 2018 Sep;30(9):1323-1329. doi: 10.1162/jocn_a_01285.
Legend for left figure: Scalp topographic map showing ERP differences of Can Define minus Familiar responses from 700 to 1000 msec for Session 2 using Session 3 behavioral responses. Legend for right figure: Averaged ERP waveforms from Session 2 separated by behavioral responses at Session 3. Note the separation of all three waves in the middle of the ERP interval.
We are following up this preliminary finding with additional experiments to prove the validity and scope of this marker. We are hopeful that in the future this and other markers of long-lasting learning can be used to rapidly determine which educational techniques best engender education that will last for a lifetime.