Abstract
A long-standing tradition of memory research is to describe memory as a serial process, in which memory successively accumulates new information (Sternberg, 1969; Shiffrin & Schneider, 1974; Farrell & Lewandowsky, 2002). Memory also has been described as including a flexible system that can modify stored information (Baddely, 1986; Morris & Jones, 1990; Woodman, Vogel, & Luck, 2012). When we change a memory, do we successively overlap new information on old information or do we actually replace old information with new information? To test this idea, we asked participants to memorize sample arrays of 1 to 5 colored dots and to perform three different tasks. In the rehearsing task, an asterisk appeared in one of the sample array locations, and participants had to rehearse the color at that location. In the adding task, a new colored dot appeared in a new location, and participants had to add the new colored dot into their memory. In the changing task, a new colored dot appeared in one of the sample array locations, and participants had to change their memory at that location to the color of the new colored dot. Finally, participants reported all colors in their most recent memory by choosing the colors from an array of all possible colors. Performance was impaired as set size increased (F(4,16)=39.858, p<.001). Rehearsing performance was better than adding (F(1,19)=9, p <.008), showing a cost of forming a new memory. However, adding performance was better than changing (F(1,19)=12.617, p<.003) and this difference became larger at a larger set size (interaction F(4,16)=5.37, p<.007), showing an additional cost of changing a memory. These results were replicated in a separate experiment blocking each task. The findings show that changing a memory is not the same as forming a new memory, but instead is a more demanding process.
Meeting abstract presented at VSS 2013