Human short-term spatial memory: precision predicts capacity. Banta Lavenex, P., Boujon, V., Ndarugendamwo, A., & Lavenex, P. Cogn Psychol, 77:1–19, 2015.
doi  abstract   bibtex   
Here, we aimed to determine the capacity of human short-term memory for allocentric spatial information in a real-world setting. Young adults were tested on their ability to learn, on a trial-unique basis, and remember over a 1-min interval the location(s) of 1, 3, 5, or 7 illuminating pads, among 23 pads distributed in a 4m x 4m arena surrounded by curtains on three sides. Participants had to walk to and touch the pads with their foot to illuminate the goal locations. In contrast to the predictions from classical slot models of working memory capacity limited to a fixed number of items, i.e., Miller's magical number 7 or Cowan's magical number 4, we found that the number of visited locations to find the goals was consistently about 1.6 times the number of goals, whereas the number of correct choices before erring and the number of errorless trials varied with memory load even when memory load was below the hypothetical memory capacity. In contrast to resource models of visual working memory, we found no evidence that memory resources were evenly distributed among unlimited numbers of items to be remembered. Instead, we found that memory for even one individual location was imprecise, and that memory performance for one location could be used to predict memory performance for multiple locations. Our findings are consistent with a theoretical model suggesting that the precision of the memory for individual locations might determine the capacity of human short-term memory for spatial information.
@Article{BantaLavenex2015,
  author      = {{Banta Lavenex}, Pamela and Boujon, Val\'erie and Ndarugendamwo, Ang\'elique and Lavenex, Pierre},
  journal     = {Cogn Psychol},
  title       = {Human short-term spatial memory: precision predicts capacity.},
  year        = {2015},
  pages       = {1--19},
  volume      = {77},
  abstract    = {Here, we aimed to determine the capacity of human short-term memory
	for allocentric spatial information in a real-world setting. Young
	adults were tested on their ability to learn, on a trial-unique basis,
	and remember over a 1-min interval the location(s) of 1, 3, 5, or
	7 illuminating pads, among 23 pads distributed in a 4m x 4m arena surrounded
	by curtains on three sides. Participants had to walk to and touch
	the pads with their foot to illuminate the goal locations. In contrast
	to the predictions from classical slot models of working memory capacity
	limited to a fixed number of items, i.e., Miller's magical number
	7 or Cowan's magical number 4, we found that the number of visited
	locations to find the goals was consistently about 1.6 times the
	number of goals, whereas the number of correct choices before erring
	and the number of errorless trials varied with memory load even when
	memory load was below the hypothetical memory capacity. In contrast
	to resource models of visual working memory, we found no evidence
	that memory resources were evenly distributed among unlimited numbers
	of items to be remembered. Instead, we found that memory for even
	one individual location was imprecise, and that memory performance
	for one location could be used to predict memory performance for
	multiple locations. Our findings are consistent with a theoretical
	model suggesting that the precision of the memory for individual
	locations might determine the capacity of human short-term memory
	for spatial information.},
  doi         = {10.1016/j.cogpsych.2015.02.001},
  institution = {Fribourg Center for Cognition, University of Fribourg, 1700 Fribourg, Switzerland.},
  language    = {eng},
  medline-pst = {ppublish},
  pmid        = {25727897},
  timestamp   = {2015.05.11},
}

Downloads: 0