Abstract
Research on visual short-term memory buffers often relies on probing a location that was recently occupied by stimulus (e.g. letter, color patch or oriented bar). This method of probing stimuli in visual memory implicitly assumes that the spatial position of a stimulus is encoded along with other stimulus properties. However, this assumption may not always be correct, and when it is not memory capacity estimations will be biased. We present the results of 4 experiments where we compare the effectiveness of feature and spatial memory probes in masked brief-exposure memory tasks. We used 3 different stimulus sets: 1) oriented bars, 2) letters, and 3) silhouette figures of common items and animals. All stimuli were presented in unique colors. We compared performance in a 2AFC orientation judgment and precision orientation judgments of bar stimuli, and identification of letter and figure stimuli. The results show that a feature-probe (color) is more effective than a spatial probe when subjects report the orientation of a bar stimulus with high-precision, but did not benefit performance when subjects reported the coarse orientation of the same stimulus (lean left or right). Conversely, subject performed better when visual memory was probed with spatial probes during a letter identification task, but figure identification was not affected by probe type. This suggests that visual short-term memory has a degree of flexibility in terms of which stimulus properties are encoded. It highlights the need to consider which type of probe will maximize the retrieval from visual memory to get the best available approximation of absolute memory capacity. Failing to consider this aspect of the research design may systematically underestimate the capacity of memory buffers.
Meeting abstract presented at VSS 2018