Abstract
Everyday cognitive processing employs working memory (WM), the ability to retain relevant information over a short period of time. The capacity of WM has been extensively debated, with standard answers in the range of 4 +/- 3 item. In naturalistic tasks, it has been reported that we choose to hold fewer items in memory, relying on the world as an external memory (Hayhoe et al, 1995, Draschkow et al., 2020). Here, we investigate how much WM capacity is utilized when we encourage, but do not require the use of available capacity. Participants played a version of the ‘memory’ or ‘concentration’ game. Participants turned over ‘cards’ in 6x4 arrays of cards. Each array held an unknown number of pairs (average ~6 pairs per array) and unique set of items. Participants could switch to a new array at will and collected 100 pairs as fast as possible to complete the task. A 15 second pause was imposed between arrays. In this version of the game, participants were alerted when the second member of a pair was uncovered. They then went back to find the first member of the pair as quickly as possible. We measured the number of clicks required to retrieve the first member of the pair. With perfect memory, participants would need just one click. Using no memory, participants could simply reclick each previously uncovered item until the target was found by chance (serial-self terminating search). Data from nine participants were consistent with imprecise memory for ~10 items, no doubt including some storage in longer term memory. Even though participants could have relied on the display as an external memory, they proved willing and able to use deploy more visuo-spatial memory than has been reported for other real-world tasks.