Abstract--Letter displays were shown in five different sizes, 12pt, 18pt, 27pt, 60pt, and 135pt. Twenty subjects were divided into two groups. One group was cued to size change of the letters and one was not. The subjects that were cued before the size change showed no greater accuracy in answers than did the non-cued group. Both subject groups gave more accurate answers for size extreamities indicating that information decay may not be as prominant in very small or large objects.
In 1960, Sperling hypothesized that more is seen than can be remembered . To test this he used a tachistoscope and projected a number of stimulus cards for 50 milliseconds . These cards displayed an increasing numbers of letters and numbers. Sperling determined that subjects experienced a fast decay of information. In fact, his subjects could only remember the information for a few tenths of a second. Irene T. Armstrong and D. J. K. Mewhort completed a study that furthered Sperling's in 1995 that analyzed the encoding and retrieval failure of rapid-serial-visual-presentation displays. They used letter lists that were presented for 130 milliseconds . Decay was examined through letter lists with or with out replicating letters. They determined that this decay is due to retrieval error (Armstrong and Mewhort, 1995). Is it possible that size change of the stimulus could effects this decay?
Twenty subjects, eighteen who are currently enrolled as students at Alma College participated in this test. The subjects were tested with Super Lab, a computer program, on a Macintosh G3 using tachistoscopy. These 20 subjects were divided into two groups. a cued and a non-cued group. The cued group was given a fixation point and a box , which showed the outliers of the stimuli before it was shown. The Non-Cued group was only given a fixation point (See Figure One).
There were five sizes of stimuli presented, three letter sets per size. The different sizes were: 12pt, 18pt, 27pt, 60pt, and 135pt. The order of appearence of the letter sets was randomized for every subject. When the subjects sat down in front of the computer they were instructed to rest their forehead on a ball hanging from the ceiling. This ensured the subject s head was exactly 22.5 inches away from the computer screen. Each letter set incured a neccessary attention area for the subject as shown in Table One. The experiment started with written instructions in the computer program (See Figure Two). Once finished reading the subjects were encouraged to ask questions. As soon as the subject felt they understood the directions they pressed the space bar to bring up either a fixation point (Non-cued group) or a fixation point and box (Cued Group) (See step one in Figure Three).
This fixation stimuli remained on the screen untill the subject pressed the space bar for a second time. After the subject pressed the space bar the first letter set was shown for one hundred milliseconds, the optimal time shown by Velliky in 1976. The computer, immediately after the stimuli had been shown, sounded a toned voice that said high , mid , or low with appropriate pitch. After this the computer circled back to a fixation stimulus and waited for the space bar to be pressed so a second letter set could be given. This method continued throughout the whole of the experiment, fifteen letter sets were given in random order and sorted later by hand. After each set the subjects were instructed to say out loud what letters they believed were present in the indicated row (See Figure Three). A recorder sat next to them and documented the answers with pen and paper. This recorder encouraged subjects to guess at letters when they were uncertain, but if the subject insisted they did not know they were not forced to guess. The results for this experiment were tablulated and sorted by hand.
Figure Three: Representation of order of events in the experiment (Non-cued group's test would have been exactly the same but without the box and only the fixation point.
Figure Five: Scores of subjects per point size with both fixation point and box. (Cued Group).
In this experiment subjects in both cued and non-cued groups did better at size extremities than they did at middle sized letters. This discovery disproved the researchers hypothesis that the accuracy of the subjects would increase to a certain point size and then decrease when the letter grew too large. It was hypothesized that the cued group because of their fore warning of size change would have more accurate answers than the non qued group. A study done by Sears and Pylyshyn determined that attentional tracking was more accurate when objects were targeted (2000). In this experiment there was no greater accuracy when the subject was warned of a size change than when the subject was left unwarned. It can be concluded that size extreamities somehow have greater resistence to decay in iconic memory.
During the experimentation, it grew obvious that the subjects would likely answer correctly to stimuli that they had been familiar with or have had some form of association based on prior knowledge. The letters were selected at random, which created some combinations that form common three letter acronyms and/or word segments. There were five variants of the stimuli. For example, a subject would recognize and remember N B A or F B I more frequently and consistently than three letters that are not an acronym or commonly seen. Repeating this experiment with common letter combinations would probably yield supporting data to this notion. But since letters can be associated in this way, experimenting with alternative stimuli might provide a way to avoid this bias with letters. Perhaps using numbers or even basic shapes, like triangles, squares, and circles, might be the way to eliminate any associative recognition.
In the future this experiment should be done with more subjects who have been trained prior to taking the real experiment.
An interesting experiment that would build off this one would be to take Armstrong s and MewhortÕs study with iconic memory and information decay and see if by changing the size of there different lettersets if the sets with replicating letters would more accurately recalled by the subjects.
Aboud, Hishan. Super Lab [Computer Program] Cedrus Corporation. 1991.
Armstrong, Irene T. and Mewhort D. J. K. Repetition Deficit in Rapid-Serial-Visual -Presentation Displays: Encoding Failure or Retrieval Failure?. Journal of Experimental Psychology: Human Perception and Performance 21.5 (1995): 1044-1052
Feldman, Dan and Gagnon, Jim. Stat View [Computer Program] Berkeley, CA . Abacus Concepts 1991.
Sperling, George. The information available in brief visual presentations. Psychological Monographs: General and Applied 74.11 (1960): 1-28
Sears, Christopher R.; Pylyshyn, Zenon W. Multiple Object Tracking and Attentional Processing. Canadian Journal of Experimental Psychology 54.1 (2000):1-14
Velliky, Robert S. (1976) Iconic Storage: A functional comparison between retarded, normal and borderline subjects. Master Thesis.