Investigating Sternberg’s Memory search Paradigm

We use cookies to give you the best experience possible. By continuing we’ll assume you’re on board with our cookie policy

Many experiments have been conducted similar to Sternberg’s Memory Search. Some of which supported his claims and others contradicting them. This experiment did both. Supported his claims that the process involved in memory searching was serial not parallel, but the process is self-terminating not exhaustive. However, many other factors that were not considered in this experiment affect the reaction times of the subjects.


Part of Cognitive psychology is to identify and unravel the systems behind mental operations (e.g. matching, recognising, searching, discriminating, thinking etc). Many mental events occur very rapidly thus slight differences in reaction times could reveal differences in performance. Therefore, one of the main methodologies in researching such areas is the use of chronometric measures, which measure reaction times. Usually, the subject is presented with a stimulus displayed on screen and the time that elapses between the stimulus onset and the execution of the response is recorded. Reaction time (the dependant variable) can vary as a function of the type of stimulus presented or as a function of the particular task requirement, and the result is assumed to provide some insight into the mental process involved in the retrieval of information and decision making and perhaps how these processes are linked together.

Research has shown that there are two memory systems that operate in the brain. One that holds information for just a few seconds, known as the Short Term Memory (STM) and one that stores more important information more permanently (the Long Term Memory or LTM). Researchers of the memory believe that if the information temporarily being held in the STM is not transferred to the LTM it will vanish and be forgotten. When the STM was first being discovered and evidence in favour of its existence was accumulated, researchers began to explore its properties. Saul Sternburg began, in 1966, to develop an experimental approach to explore how information was retrieved from the STM.

Subjects were shown a short list of numbers and asked to memorise them. The lists consisted of any number between one to six items and were made up of the numbers 0 to 9. When the subjects were given ample time to put the list to memory a single digit was shown to them, known as the probe number. The probe number was either present in the original memory set or a new number and was therefore absent. The subject then had to react as quickly, but as carefully, as they could to identify whether the probe was present or absent. The reaction time of subjects was recorded and this suggested the time spent searching the STM and retrieving the necessary information to determine whether the probe was part of the memory set. Sternburg varied the number of digits in the memory set and by doing this hypothesised that he could test several theories of STM search.

Sternburg (1966) believed that the selection of a response requires the use of information that is in the memory and the latency of that response will reveal something about the process by which the information is retrieved.

The process of information retrieval was believed to be complied of four primary mental operations; encode: the probe must be identified; search: the mental representation of the probe must be compared to the mental representations of the items in the memory set; decide: a decision must be made as to whether the probe digit matches one of the items in the memory set and finally respond: the appropriate response (yes/no) must be generated, in the case of this experiment one of two buttons being pressed (Sternburg, 1966).

Two interesting properties were discovered by Sternburg from the results of his experiment. The reaction times grew linearly with the increase in the size of the memory set. For each additional item in the memory set subjects took an additional 38 milliseconds to make their response. This implied that the probe was compared to each item in the memory set separately and each comparison took approximately 38 milliseconds. This suggested that if the probe was present in the memory set the process continued to compare the probe to the rest of the memory set even when it had identified that it was present. The process does not stop when the probe had been identified. Each memory comparison operation took a fixed internal processing time. This does not mean, however, that this fixed value represents a fixed property of memory comparison. Factors such as experience and the difficulty of the task will affect it.

Also, when comparing the reaction times for the probe present and the probe absent, Sternburg found no differences in reaction times. This finding was significant as an absent response can only be made after all the items of the memory set stored in the STM have been searched and found not to match the probe item. It would be fair to assume that a probe present trial could terminate as soon as the probe item is matched with the appropriate item in the STM. As a result the probe present trials should have a faster reaction time, but Sternburg’s results contradicted this assumption. He, therefore, found that the function for positive and negative trials were identical. The fact that the functions were parallel implies that a search in STM is always exhaustive and serial as opposed to self-terminating. The cognitive processes responsible for searching STM for a particular item, search through all the items in STM before reporting whether the probe item is in memory or not.

Null Hypothesis: As this experiment is similar to Sternburg’s the hypothesis shall be similar to the results of his experiment. That there will not be any difference between the probe present reaction times and the probe absent reaction times. However, this experiment also involves the use of symbols as a memory set, not just digits, so the prediction will be: there will, also, not be any difference between the reaction times of the digits and the symbols.

Hypothesis: That there are differences between each of the conditions, both between the digits present and absent in both the memory sets of symbols and digits. There will also be a difference between the overall reaction times of the digits and the symbols.


Subjects: The stage two psychology students took part in the experiment. The age group of the subjects was predominantly 19-21. There was mixed gender, however, the majority was female. The outcome of the experiment was unknown to the students at this point.

Apparatus: The computer program was the only apparatus used. The program ran a similar format to that of Sternburg’s experiment, however it was not identical.

The results that were found in this experiment seem to contradict those of Sternburg’s. It is clear from table 2 that the mean reaction times for the memory sets containing digits differ when the probe was present or absent. Similarly between the memory sets containing symbols. Subjects seem to be reacting faster when the probe was present in the memory set and slower when the probe was absent. Overall, differences between the mean reaction times of the digits and the symbols are also apparent. Subjects were slower to react to the symbols than to the digits.

This may be due to digits being instantly recognisable to the subjects whereas each symbol in the memory had to be recognised and memorised then the probe had to be recognised and compared to each of the other symbols stored in the STM. This could be an extra stage in the process only apparent when the stimulus presented to the subjects is not instantly recognisable. The differences in the mean reaction times between the probe present and probe absent suggests a more parallel processing and that the process is in fact self-terminating.

This would contradict Sternburg’s results, however, other researchers have discovered similar results (Van Zandt, T. et al 1993). However, after comparing the reaction times with respect to the number of items in the memory set (figures 3 and 4), it became apparent that there is a linear relationship between the reaction time and the number of elements in the memory set. This would therefore imply that the process is serial, i. e. it compares the probe digit/symbol successively to each item in the memory set.

But from the mean reaction times, it can be assumed that the process is self-terminating. Resulting in a serial processing operation that self-terminates when the probe is matched. This was the same for both the memory sets containing symbols and digits, however the symbols, on the whole took more time to be recognised and led to slower reaction times. This could be interpreted to mean that there are different mental process that occur for different types of stimuli, one for numbers, which seems to be more efficient, and one for symbols which is less efficient.

This could be researched further to discover where the differences lie and how many other different types of stimulus have apparent different mental operations. Other uncontrollable factors could have influenced the reaction times. As this was a repeated measures design it would be fair to assume that practice in the task would have sped up reaction times. As the task was repeated by each subject many times the subjects could have got better at it and therefore reaction times decreased after many repetitions.

The reaction time could have also been affected by the difficulty of the task, this could be another explanation for the differences between the symbols and digits as symbols could be considered as being more difficult. But the stimuli that are presented could be similar, i. e. the numbers 1 and 7, if presented incorrectly could be similar. This may lead to greater levels of inaccuracy within the results of the experiment.

There is a possibility that the subjects themselves could affect the reaction times, forms of dyslexia can affect a subject’s interpretation of numbers, thus making it more difficult to distinguish between separate digits. Similarly with words and other stimuli. From table 4 it seems that the number of elements in the memory set held the most influence upon reaction times with a high variance ratio of 174. 1 being highly significant. This experiment has resulted in a contradiction to one aspect of Sternburg’s result but supports another.

Reaction time was found to increase linearly as a function of set size, the more digits in the memory set the slower the reaction time, which supports Sternburg’s theories suggesting a serial process occurring, however there was a difference in reaction times between the probe being present and the probe being absent, suggesting a self-terminating process. There was also a difference in overall reaction times between the digits and the symbols, suggesting that the task involving symbols was more difficult, or there are separate processes, that differ in efficiency, for digits and symbols.

Tagged In :

Get help with your homework

Haven't found the Essay You Want? Get your custom essay sample For Only $13.90/page

Sarah from CollectifbdpHi there, would you like to get such a paper? How about receiving a customized one?

Check it out