Sternberg Search
This lab activity addresses the topic of how information is stored and retrieved from short-term memory using a simple examination of response times. It is based on a search paradigm developed in 1966 by Sternberg, which has since been applied widely to memory research.
The Parallel vs. Serial Debate
Sternberg developed this paradigm at a time when cognitive psychologists were debating two competing models. In parallel models, all information could be retrieved from short-term memory at once so that retrieval time would remain mostly the same no matter the number of items in short-term memory. This is similar to your computer being able to do two or three or four jobs at once because it has multiple “parallel” processors with each being able to handle a separate job simultaneously. An alternative is the serial model in which information is retrieved item by item resulting in increased response times the more items are in short-term memory.
Encoding, Delay, and Probe
The Sternberg paradigm has proven popular because of the simple way it is organized. On each trial, there is an encoding period in which you will see one, three, or five digits, all presented on the screen at once. Examples of these different set sizes could be “3,” “874,” or “27518.” Then following a brief delay, a single probe digit will be presented. So, if you see “874” at encoding, following the delay you could see a positive probe of “7” or a negative probe of “3.” Positive means the probe matches one of the items presented at encoding, whereas negative means the probe is not a digit presented in the encoding set. In this lab activity, you will complete 10 practice trials with feedback followed by 48 experimental trials without feedback.
Response Times by Set Size
At the end of this lab activity, you will be shown your response times on trials where you answered correctly and broken down by positive and negative probes. Typically, response times increase as a function of set size. It takes longer to confirm that, for example, “3” was or was not in a five-size set “45379” than in a three-size set “623.” Usually, the response times for positive probe (target present) and negative probe (target absent) trials are about the same. The first result that response times increase with set size supports a serial model of short-term memory. The serial model predicts that during retrieval one goes through items one by one not simultaneously, resulting in increased response times with increased set size. A parallel model would predict that all items are retrieved at once and there should be no effect of set size on response time. The second result that response times are similar for positive and negative probe trials (i.e., target present vs. target absent) supports what is termed “exhaustive search” in which an entire set is searched even if the target item is detected before the end of the set is reached. An alternative type of search called self-terminating search happens when once an item is found, searching stops. If participants used a self-terminating search, then response times for positive probes should be slightly faster than on negative probe trials because on the probe absent trials the whole list must always be searched. It is intuitive to think that one would do a self-terminating search on the Sternberg paradigm, but similar response times for probe present and probe absent trials suggest otherwise that participants use an exhaustive search.