The Sternberg Paradigm: Correcting Encoding Latencies in Visual and Auditory Test Designs

Abstract

The Sternberg task is a widely used tool for assessing the working memory performance in vision and cognitive science. It is possible to apply a visual or auditory variant of the Sternberg task to query the memory load. However, previous studies have shown that the subjects' corresponding reaction times differ dependent on the used variant. In this work, we present an experimental approach that is intended to correct the reaction time differences observed between auditory and visual item presentation. We found that the subjects' reaction time offset is related to the encoding speed of a single probe item. After correcting for these individual encoding latencies, differences in the results of both the auditory and visual Sternberg task become non-significant, p=0.252. Thus, an equal task difficulty can be concluded for both variants of item presentation.

Keywords: Sternberg task; differences between visual and auditory encoding; memory load; reaction times; working memory performance.

Figure 1

Schematic illustration of Sternberg’s serial exhaustive scanning theory of working memory. It is hypothesized that observed deviations in RTs between visual and auditory test designs can be explained by fundamental differences in the sensory processing of the respective stimuli at the encoding stage rather than by a general difference in task difficulty.

Figure 2

Experimental Setup. (A) Schematic illustration of the observation chamber and viewing direction. (B) Side view picture of the observation chamber (maintenance door removed) showing a subject focusing the fixation target, which is displayed on the reflective LCD embedded to the center of the chamber’s rear wall. A chin rest is used for positioning and aligning the subject’s head accordingly.

Figure 3

Schematic illustrations of the two Sternberg variants used in this experiment. (A) Visual procedure. (B) Auditory procedure. In both examples, the memory load is 3 with $M=\{5;2;7\}$ and the positive probe $\alpha =7$. In the auditory case, the fixation target is displayed continuously.

Figure 4

Mean reaction times for the visual (open circles) and auditory (crosses) character specific tests. Error bars represent standard errors.

Figure 5

Illustration of the visual and auditory Sternberg task results before and after correcting for character-specific encoding and processing times. Closed circles represent the mean for positive probes, open circles for negative probes. Error bars give the corresponding standard errors. (A) Comparison of the uncorrected Sternberg data. Mean RTs rise linearly with increasing memory load, showing equal slopes, but different y-intercepts for visual compared to auditory stimuli presentation. (B) Comparison of the corrected Sternberg data. Mean RTs still rise linearly with increasing memory load for both methods of stimuli presentation, again showing a similar rate of increase. Differences between the y-intercepts become non-significant.