Comparing the fidelity of perception, short-term memory, and long-term memory: Evidence for highly detailed long-term memory representations

Alvarez, Konkle, Brady, Gill, & Oliva

Recently we demonstrated that visual long-term memory (LTM) can store thousands of objects with remarkable fidelity, but it remains unclear how the fidelity of LTM compares to the fidelity of short-term memory (STM) or online visual perception. We used color as a case study to quantify the fidelity of LTM, STM, and perception using pictures of real world objects. LTM: 180 real objects were presented one at a time, with their color space randomly rotated. Afterwards, observers adjusted the color of each test item to match the original. STM: 3 objects were simultaneously presented. After a brief blank interval, color memory for all three items was sequentially tested. PERCEPTION: 2 copies of the same object appeared side by side, and observers adjusted the color of the item on the right to match the color of the other item. In all experiments, error was measured as the distance on the color wheel between the chosen color and the actual color of the object. Using a mixture modeling method (Zhang and Luck, 2008) we quantified two parameters of performance: the variability of internal representations of color (SD), and the probability of guessing (PG). As expected, the fidelity of color representations was better for perception (SD = 6.2°) than STM (SD = 18.4°, p p = .56). However, the probability of guessing rose dramatically from 8% in STM to 56% in LTM. These results indicate that observers are much more likely to have no memory at all for the color of an object in LTM, but that when the color is remembered (about half the time), it is represented with the same fidelity as just a few objects briefly stored in STM.




Alvarez, G., A., Konkle, T., Brady, T. F., Gill, J., & Oliva, A. (2009). Comparing the fidelity of perception, short-term memory, and long-term memory: Evidence for highly detailed long-term memory representations Journal of Vision. 9(8): 584; doi:10.1167/9.8.584.