Experiment 1. Iconic memory
19 participants took part in the first experiment. All of them had normal vision and hearing (did not have any reported vision or hearing impairments).
Apparatus and procedures
The participants were placed in front of the computers in a large room. Brightness and contrast of the monitors were adjusted to produce maximally sharp image. The participants were able to adjust the parameters of their monitors, and were able to locate themselves in order to reduce any glare. The experiment took part in the day time, and the lights were turned off to ensure optimal conditions for the participants.
The participants passed through 120 practice trials, and then were tested through 4 blocks of trials, every block included 30 trials. 12 letters were randomly chosen from 21 consonants for each trial. A matrix consisting of three rows and four columns of randomized 12 letters was shown for 100 ms to the participant. In the “full”¯ block, the participants had to report as many letters as they could remember; in the other three blocks, the participants heard a tone (low, middle or high), and had to report the letters from the associated row. The difference between the blocks was the delay between the letters and the tone (0, 150 and 500 ms). Counterbalancing was performed through assigning random order of blocks for every subject.
Independent variable in this research was the delay between showing the picture and the tone (0, 150 and 500 ms), along with immediate memory (“full”¯ block). The dependent variable was the percentage of correctly named letters.
Mean percentage of letters correctly recalled constituted 0.49 for full report, 0.47 for sound delay 0 ms (no delay), 0.44 for delay of 150 ms, and 0.43 for the delay of 500 ms. Thus, mean number of letters correctly recalled is about 5-6 letters for full report; for the situation when sound follows the image without a delay, 2 letters are likely to be correctly recalled, and 2 letters or less when the delay increases. This is aligned with Sperling’s findings regarding the properties of iconic memory: the icon vanishes over time quickly, and partial reporting shows that approximately 6 numbers were recalled from the whole matrix for the 0 ms delay block.
Fig. 1 illustrates the changes in the proportion of correctly recalled numbers as function of block type (delay and instruction differences).
ANOVA analysis showed that although the hypothesis regarding the means and effect of sound delay on iconic memory was confirmed, the difference in the percentage of correctly entered letters was not statistically significant: F(2,54) = 0.271, p=.764. Post hoc LSD tests showed that there were no statistical differences between group means. Most likely, such result is conditioned by the fact that a limited number of participants took part in the experiment, and these participants might not represent a valid sample of the population.
Overall, the results of the analysis have confirmed Sperling’s findings: the highest proportion of correct letters was witnessed for full report block, and there was a notable decay in the number of correctly recalled letters with the increase of sound delay. The findings of Averbach and Coriell (1961) regarding the mean number of letters stored in the iconic memory were also confirmed.
However, ANOVA analysis showed that the differences between groups could not be regarded as statistically significant. The lack of significance for the results of this experiment can be explained by several factors.
First of all, the choice of the measurement of the dependent variable might be affecting the results. According to Sperling, it is optimal to use average number of letters correct in the partial report multiplied by the number of non-overlapping partial reports as a measure of correctly entered letters for comparison with full report (Sperling, 1960). Secondly, the results might have been affected by monotony: participants’ attention could have become weaker due to 120 practice trials followed by 120 experimental trials.
One more possible cause of lack of significance for ANOVA results is that the participants might not have been trained enough to show statistically significant results: in this experiment it is important that participants are located in the appropriate setting, and are appropriately trained.
The research can be further developed and improved by increasing the number of participants to at least 50 persons, training them not immediately before the experiment, but several days prior to the experiment, and improving the measurement of the dependent variable in accordance with Sperling’s recommendations (Sperling, 1960).