Researching memory coding and generalisation

Maria Wimber (University of Birmingham, UK) has been leading some interesting research into possible reasons behind generalisation in memory encoding. One of the studies involved inviting participants to memorise pairs of unrelated images e.g. a mountain paired with a kiwi fruit.

Two days later the participants were linked up to an MRI machine to measure brain activity during recall. They were then shown one of the images from a pair and asked to recall the other image. In NLP terms, this was a test of a mild anchoring (e.g. low level stimulus).

Participants on average, were 79% likely to recall the correct category the associated object belonged to when shown the stimulus picture. Using the example above, when shown the image of the mountain people might remember the associated image was fruit but not the specific fruit, in this case kiwi.

The MRI scans showed an increase in brain activity in the neocortex during the recall test that had similarities to the pattern of activity seen during sleep. According to Wimber, the pattern observed in both sleep and her study is likely to be connected with memory formation and consolidation.

The study goes into more depth about these processes however from an NLP perspective what is even more interesting is Wimber’s conclusions from a second follow up study. She created baselines using scans for an association exercise (images and words) and then again during recall.

The study looks at what appears to be activity in the visual processing centre 300 milliseconds before activity in the neocortex, during recall this process is reversed. Wimber hypothesised from the data that such activity results in abstract generalisations encoded in the neocortex without detail so that in recall it is a generalisation that is remembered.

Studying the data from scans and self-report experience of participants led Wimber to suggest some possible reasons. She suggests that abstract generalities are more advantageous than specifics in creating useful reactions and behaviours. The example she gives is where someone is bitten by a small white dog in a park. The generalisation may be that a free running dog (in any location) might bite.

Reading this study does leave me with questions. It does appear that there is a leap between concrete data of brain activity and what it might mean. I can accept these leaps as useful and can see that we could make links to our NLP map. What do you think?

Reference article: “Why your brain forgets details”, author Chelsea Whyte, New Scientist, 15thDecember 2018.