Abstract
Human memory consists of sensory memory (SM), short-term memory (STM), and long-term memory (LTM). SM enables a large capacity, but decays rapidly. STM has limited capacity, but lasts longer. The traditional view of these memory systems resembles a leaky hourglass, the large top and bottom portions representing the large capacities of SM and LTM, whereas the narrow portion in the middle represents the limited capacity of STM. The “leak” in the top part of the hourglass depicts the rapid decay of the contents of SM. However, recently, it was shown that major bottlenecks for motion processing exist prior to STM, and the “leaky hourglass” model was replaced by a “leaky flask” model with a narrower top part to capture bottlenecks prior to STM. The leaky flask model was based on data from one study, and the first goal of the current paper was to test if the leaky flask model would generalize by using a different set of data. The second goal of the paper was to explore various block diagram models for memory systems and determine the one best supported by the data. We expressed these block diagram models in terms of statistical mixture models and, by using the Bayesian information criterion (BIC), found that a model with four components, viz., SM, attention, STM, and guessing, provided the best fit to our data. In summary, we generalized previous findings about early qualitative and quantitative bottlenecks, as expressed in the leaky flask model and showed that a four-process model can provide a good explanation for how visual information is processed and stored in memory.
Highlights
It is a challenge to understand how the continuous stream of information available to the visual system is processed in sensory memory (SM), short-term memory (STM), and long-term memory (LTM)
Among the different report types, there were two kinds that were of particular interest for this part of the study: single report (SR), i.e., the one item randomly selected by the computer for reporting; the first item in the full report condition (FR1), i.e., the first item that the participant chose to report in each trial
The statistics showed that single report condition (SR) and FR1 were not significantly different across set size (F = 0.772, p = 0.444, ηP2 = 0.205), which means the involvement of memory capacity was limited or negligible, as one would expect from a cue delay of zero
Summary
It is a challenge to understand how the continuous stream of information available to the visual system is processed in sensory memory (SM), short-term memory (STM), and long-term memory (LTM). STM lasts longer with a limited capacity [5–18], constituting a bottleneck in information flow [9,10,19,20]. The red items in the figure represent information, which is first encoded with a large capacity encoding stage and stored in SM. Whereas SM’s capacity is large, as depicted by the broad top of the hourglass, information is lost at a rapid pace, which in turn is shown as a leak from SM. In classical models of memory, a single temporal decay process is used to explain the temporal dependence of the quality and quantity of information in memory. As the representations of items decay in time, the quality of information becomes degraded and the number of items available for storage becomes smaller
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
