Experiment 2: Sternberg’s short-term memory model



                The dual-memory theory is a popular model of human memory, which differentiates two fundamental types of memory, short-term memory (STM) and long-term memory (LTM). It is assumed that STM stores information for only a brief time, about 10 seconds. Furthermore, STM is thought to have a limited capacity, about 7 +/- 2 items. Long-term memory is assumed to store information indefinitely, and is thought to have an indefinitely large storage capacity.



                The two memory systems have a dynamic relationship with each other, as depicted in Figure 1. Incoming information, such as the sounds made by someone speaking, cause memories of the meanings of words to be retrieved, and in this way the speaker’s message is understood. As the memories of the words are retrieved, they occupy short-term memory.







Short-term Memory


Long-term Memory


Incoming Information









Fig. 1. Dual-memory model. Two memory systems are hypothesized, short-term memory (STM) and long-term memory (LTM). It is hypothesized that incoming information is only briefly stored in STM, unless it is rehearsed. Otherwise, it is transferred to LTM or displaced.


                Some of what the speaker says may be transferred to LTM, that is stored permanently. Exactly what is stored in LTM, or remembered, depends on many variables, including the relevance or importance of the incoming information to the listener.


                Incoming information may also be REHEARSED by the listener, that is, the listener may consciously repeat the information. Rehearsal has two important functions: (1) it will prolong storage of the information to LTM. For example, if you call the operator for someone’s telephone number, and you are given the number, you may rehearse the number until you have time to write it down or dial the number. Will you remember the number five minutes later? If so, you transferred the number to LTM. If not, as depicted in Fig. 1, the information has been DISPLACED or forgotten.



                How long does it take to RETRIEVE information from short-term memory? If, for example, you are presented with a series of five numbers, 5-2-7-9-4, will it take you longer to decide whether another number, 6, was or was not one of the numbers you had just seen than it would if you were presented with a series of two numbers, 3-1?


                Sternberg (1966) conducted such an experiment and found that as the length of the series increases the length of time required to scan short-term memory and decide whether a number is present in STM increases.



                The present experiment is a replication of Sternberg’s experiment. It will be hypothesized that if subjects see numbers one at a time, in sets ranging in length from one to six digits, and are then asked whether a probe digit was in the set, their mean decision time increases linearly with the length of the set.




                The subjects will be you and other members in your lab section.



                The experiment will be carried out on an IBM computer with a color monitor. The program used in the experiment as written by Dr. L. Gilden and John Zhu.




1.  The subject will be required to sit at the keyboard with his/her left index finger on the ‘N’ key and his right index finger on the ‘/’ key. The task involves viewing sets of the digits 0 through 9 presented one at a time in the center of the computer monitor, and then pressing the ‘N’ or the ‘/’ key.


  1. A set will consist of random presentation of one to six digits. After each set a cue symbol, #, will be presented, followed by a probe digit. The subject will then be required to press the ‘N’ key AS QUICKLY AS POSSIBLE, if the test digit was not in the set, and the ‘/’ key AS QUICKLY AS POSSIBLE, if the test digit was in the set. After each response the subjects will be informed whether the response was or was not correct.


  1. Following the response to the probe digit, the subjects will enter on the key at the right side of the keyboard the numbers in the set just presented.


  1. A practice set of 24 trails will be conducted first, to insure that the subjects are familiar with the procedure. Then, after a five minute rest during which time the subjects read a book, they will be given a set of 24 TEST trials.



The presentation of the sets of digits will be done using the following parameters: a digit will be presented for 1.2 seconds followed by a 0.5 second pause, until all the digits in the set have been presented. The cue symbol, ‘#’, will then appear for 2 seconds, followed by the probe digit.


The size of the set is the independent variable, and the subjects’ decision times to the test digit is the dependent variable.


The computer will also be keeping track of the subjects’ recall of the number in each set. This recall also constitutes a dependent variable.




                Begin with the results section with an evidence report such as the following: when subjects saw numbers one at a time, in sets ranging in length from one to six digits, and were then asked whether a test digit was in the set, their mean decision time increased (or did not increase) linearly with the length of the set. (This statement will have to be written after the following calculations have been made.)


  1. Examine the data sheet and determine whether any subject’s recall percentage was less than 75%. If so, do not include his or her data in your calculations. State how many subjects’ data were used and excluded.


  1. Calculate the mean percent recall for all the subjects. Report these results.


  1. Calculate the mean decision time in milliseconds (msec) for all the subjects for each set size. Give a general description of these data.


  1. Make a scatter diagram of the SET SIZE (on the X axis) and the associated mean DECISION TIME in msec on the Y axis). Then, use a straight edge and draw a line through the dots, so that there is approximately equal distance for dots above and below the line. In mathcad, you can do so using regression function to derive the function for generating the line.


  1. Calculate the correlation coefficient for these data.


  1. Report  your own data, and compare them tot eh group data.


  1. Calculate the difference between the mean decsion time for the set with one digit and the set with six digits. Divide the difference by 5. This is a rough approximation of the time to search for an item in short-term memory. Report this result in msec.




 Begin the discussion section with a general summary of your results.


Answer all of the following questions.

  1. Describe your experience as a subject in the experiment.
  2. How can a linear relationship between number of items in short-term memory and scanning time be explained, if the linear relationship exists?
  3. What does the time required to engage in the search process for each item in short-term memory indicate about the way memory scanning occurs? (Suggest an analogy using scanning an array of actual objects).
  4. If you were asked to imagine the route to travel from a class in SB to the student union and to your home, which imaginary trip will take more time? Why? Approximately how much time would be required to imagine each step along each route?
  5. Describe in some detail a similar experiment which would shed further light on how people retrieve information from short-term memory.
  6. What were some problems with the present experiment that should be controlled when the experiment is conducted again?



1. Sternberg, S. (!966). High-speed scanning in human memory. SCIENCE, 153, 652-654.