Green alone does not make a memory: Long-term priming in visual search
Wednesday, April 15, 2015
Imagine the following dystopian job: You are working in a factory, and your task is to look for that matching sock so the pair can be packaged together. To make things worse, you are given another sock every few seconds, together with a new drawer in which to look for its twin. To manage the boredom, you begin to keep track of your performance, and after some time you discover an interesting regularity: whenever the sock you are looking for is the same as the one you searched on the previous trial, your performance is better. This improvement occurs even though the drawers that you have to search on each trial are different from each other.
This improvement in visual search is known as priming and it has been known for a long time. What has been less clear is why this effect occurs. Is it memory for a particular sock from the previous trial? Or is it the fact that the features of the sock—such as its blackness or the number of holes—are still hanging around from the previous trial?
A recent article in the Psychonomic Society’s journal Attention, Perception, & Psychophysics has shed some light on those two different possibilities. Researchers Kruijne and Meeter from the Vrije Universiteit Amsterdam used visual shapes and colors, rather than socks, in several experiments in which participants had to identify a specific target on each trial.
The stimuli are shown in the figure below, with one example for each of the three experiments in the paper:
In each case, the target stimulus is the green(er) diamond at the top, with the bottom tip missing. Participants’ task was to indicate (by pressing the appropriate response key) whether it was the upper or lower tip that was missing from the target diamond. Across trials, the location of the target was obviously different, and the target itself also differed—it could be a red diamond, for example, on some trials and a green diamond on others. Emphasis was on the speed and accuracy with which people could indicate which tip in the target diamond was missing.
In order to induce priming, the experimenters varied the proportion of trials on which the target was the same within a “block” (a continuous stream of200 trials). On some blocks, the target was either green—as in the figure above—or red with equal probability. On other blocks, the target was one color—let’s assume green for consistency with the above figure—on 80% of all trials. It was expected that considerable priming—i.e., a benefit to performance—would accrue when the target was the same across successive trials, and that this priming would be particularly large in the condition in which 80% of targets were of the same color.
The predictions were confirmed. In line with much prior research, there was a benefit whenever the targets on two successive trials were identical.
Of greater interest were some additional, novel effects: For example, once participants had experienced a block of “biased” trials (80% of targets were the same), the biased targets retained a significant advantage for the next 200 trials—that is, the strong priming effect never went away. Even more intriguingly, this happened only when the search was particularly difficult, namely when people had to search for a conjunction of features.
The distinction between singleton and conjunction search requires some more explanation. In the figure above, in the left-most example, the target (the green diamond, remember?) can be located on the basis of deviating from its surrounding distractors by a single feature—namely its color. If you find something that’s green in a sea of diamonds, then you are done. This is called a singleton search.
In the center stimulus in the figure above, the target is the same—but the distractors differ and include some other diamond, in this instance a blue diamond. Hence color is insufficient to detect the target, and neither is shape because there are two diamonds, only one of which is a target. This is called a conjunction search because it is the conjunction of shape and color that defines a target, and no single feature will be sufficient.
So back to the data.
Kruijne and Meeter found the long-lasting priming advantage for conjunction search but not for singleton search. In the singleton case, there was priming when the targets on two successive trials were the same, but the bias manipulation had no additional effect and it did not linger about. In the conjunction search, by contrast, once a target had been biased (by appearing on 80% of all trials in a block), it retained its massive advantage during 200 unbiased trials following it.
Kruijne and Meeter suggest that this long-lasting priming for conjunction search reflected a unique contribution from memory for the particular stimuli—rather than just their features—that is laid down during the bias blocks (when 80% of targets are the same).
To nail this down, they conducted another experiment using the rightmost stimuli in the figure above. Ask yourself whether this is a singleton or conjunction search.
That’s right, it’s a singleton—but it’s a very difficult singleton search because both targets and distractors are green. Kruijne and Meeter reasoned that if difficulty alone were responsible for the long-lasting priming, that it should also be obtained with the green-on-green singleton search.
The data suggested otherwise: Once again, there was no long-term priming even when the singleton search was difficult. There was long-term priming, however, when the search was a conjunction search even when that search was made easier (by re-arranging the distractors; not shown in the figure above).
Taken together, the experiments form a pretty clean picture: When more than a single feature is relevant to a visual task, people tend to lay down seemingly robust memory traces of the stimuli, and they persistently affect performance for a long time. By contrast, when a single feature is sufficient to perform a task, then there is still trial-to-trial (feature-based) priming, but there does not appear to be any long-lasting memory for particular stimuli.
This just goes to show that if you get dressed several hundred times a day, all your socks should differ from each other by more than just one feature. If your sartorial requirements are more modest, then having black socks is just fine.