Why human memory is hugely dependent on probability and selective recall

People only remember things they expect to have relevance to in the future, scientists say. An experiment looking at memory formation shows people are not able to remember features associated with specific objects unless they were specifically asked to do so.

Published in the journal Cognition, the study comes up with a new theory for how memory is formed. In modern psychology, two explanations are usually given to explain this process. The first is that people store all the information regarding an object in their working memory. The second opposing theory is that memory is a selective process, and people actually only remember some aspects of the object.

In this latest research, the scientists worked on the basis of this second hypothesis, to come up with their own conclusions. They reveal a specific selective process through which memory only stores elements that the brain considers to be important for the future.

Balls and surprise test

Sixty participants were recruited and asked to watch videos of two balls being thrown between multiple players. They were instructed to count the number of passes made with one of the balls
– the 'target ball'. The other ball was intended as a distraction from this task.

"Participants have memories of the colour of both balls, but those memories aren't attached specifically to the target ball or the distraction ball," explains Hui Chen, the study's first author.

Their explanation is simple: because the participants were only told to count passes, and did not expect to be asked about colours, they did not store the information in their memory.

Four control trials followed the surprise question, during which people were asked to report the number of passes as well as the colour of the target ball. This time the percentage of participants who did not give the right colour was much lower, amounting only to 14%. This time, the memory stored the information regarding colour, because individuals were expecting to be asked about it.

"What we're showing is that attention is not enough to ensure accurate memory. You need some kind of expectation that attributing certain features to the object is important," concludes co-author Brad Wyble.