Cognition

TAKE A LOOK AT THE FOLLOWING ANALYSIS OF THE CLASSIC STERNBERG STUDY:

Sternberg, S. (1966). High-speed scanning in human memory, Science, 53, 652-654.

Sternberg set out to verify the existence of serial processing of information in the human mind by way of high-speed scanning of memory. He believed the subjects' mental representations of sequences to be functionally equivalent to the actual visual percept of the sequence. Specifically, the mean time subjects took to determine whether of not a particular test symbol was present was analyzed. Subjects were shown beforehand a short sequence of symbols to be memorized. Following sufficient time to memorize the sequence, the subjects in Sternberg�s experiment were shown a test stimulus and asked to identify whether that particular symbol was a part of the previous sequence. The serial processing Sternberg believed to exist is human scanning of objects is defined as the process of observing each item one by one and utilizing all resources on each item.
This study of high-speed scanning in human memory refers to only recent memory and thus the analysis of previous research primarily focused on short-term memory. The study of short-term memory is analyzed because their study is looking specifically at symbolic information in recent memory and we know that according to the Waugh and Norman (1965) traditional model of memory, short-term memory most likely refers to their primary memory store which, �holds temporary information currently in use,� (Sternberg, 158). However, it is also probable that Sternberg did indeed know of the Atkinson and Shiffrin model, (to be published in 1968), which holds short-term memory as, �the hypothetical construct of a short-term store, which has modest capacity (for only 7 items, give or take 2), has a duration of storing information for only a number of seconds unless strategies (ie. Rehearsal) are used for keeping information in the short-term store for longer periods of time,� (Sternberg, 521).
Although previous research with short-term memory has revealed some specific reasons why humans often fail to remember particular bits of information, there still hadn�t been any research done offering clues to explain how error-free performance may occur within some short-term memory tasks. Similarly, information indicating the specific processes in charge of short-term retrieval had yet to be determined. �One reason for the neglect of retrieval mechanisms may be the implicit assumption that a short time after several items have been memorized, they can be immediately and simultaneously available for expression in recall or in other responses, rather than having to be retrieved first,� (Sternberg, 652). It seems that experimenters automatically leave out a step in memory processing because of our innate, human tendency to oversimplify. Therefore, without recognizing retrieval as a necessary step toward recall, the previous studies of short-term memory have completely disregarded an entire area of research. �This is to assume the equivalence of the �span of immediate memory,� (the number of items that can be recalled without error) and the �momentary capacity of consciousness,� (the number of items immediately available)." The objective of the Sternberg experiment was to prove that there are two distinct dynamics working in recall and that previous research assumptions do not recognize.
Upon designing his experiment, Sternberg hypothesized that �the latency of the response will reveal something about the process by which the information is retrieved,� (Sternberg, 652). Sternberg and his colleuges were especially interested in how the number of items stored in memory would effect the recall-response time. It seems that if there does exist an internal-serial comparison process, (as mentioned in the abstract), their subsequent data should reveal a linear response in relation to the length of the sequence stored in memory. The subjects were initially given a short series of symbols to be memorized. After the series had been placed into memory, the subjects were then shown a randomly selected test stimulus. The test stimulus was either one of the symbols placed into memory or a novel symbol, not seen previously. Two levers were provided and depending upon whether they believed the test stimulus to be a previous symbol, the subjects either pulled one or the other to give their answer. Response latency was operationally defined by Sternberg as the time from the onset of the test stimulus to the occurrence of the response. Sternberg is careful to note that in this particular experimental model, �it is the identity of the symbols in the series, but not their order, that is relevant to the binary response,� (pulling one of two levers). (Sternberg, 652.)
The symbols chosen for memorization in Sternberg�s experiment were the first 10 digits in our numerical counting system. Basic digits were chosen as test symbols particularly because they were �well learned and highly discriminable to all subjects,� (Sternberg, 652). Each trial of experiment #1 consisted of a random series of digits shown to the subjects which could range anywhere between 1 and 6 different digits. These were displayed one at a time, regardless of how many digits were in a particular series. The chosen lengths for each series to be given to the subjects were randomly assigned. Following the series displayed, there was a 2.0 second delay, a warning signal given, and then the subsequent test digit would appear that the subject responded to. Once the test digit came into view, the subjects were to pull the appropriate lever. Once the lever was pulled, a feedback light would glow, letting the subjects know whether there response was correct. Finally, the last task asked of his subjects was to recall, (to the best of their ability), the series of symbols in order.
Sternberg varied the responses of subjects by equally administering questions requiring a positive response with those requiring a negative response. Additionally, each subject had 24 practice trials and 144 test trails. �Feedback and payoffs were designed to encourage subjects to respond as rapidly as possible while maintaining a low error-rate�The lower-error rates justify the assumption that on a typical trial, the series of symbols in memory was the same as the series of symbols presented,� (Sternberg, 352). The above statement by Sternberg, to me, appears slightly flawed because he is justifying an assumption with an assumption. In Particular, whose to say that the practice effects don't account solely for low-error rates? I believe that there is a possibility that other internal strategy mechanisms, (especially rehearsal), facilitate the optimization in performance and thus decrease errors. He doesn't consider that feedback and payoffs may be working differently, (other than visually representing symbols in the subjects mind), to facilitate lower error-rates.
Sternberg found that 99.4 percent of the variance of the overall mean response latencies could be accounted for in the linear regression during data analysis. The nearly perfect linearity proves that the present findings exceed significant probability of error. Second, Sternberg found that �the latency of a response depends, in part, on the relative frequency with which it is required.� I believe this to possibly signify that with frequent responses, fatigue may result and thus increase response latency.
In constructing experiment #1, Sternberg made adjustments to the experimental procedure which eliminated most confounds, and thus facilitated experiment #2 by making it a type of control. Experiment #2 only analyzes affirmative responses by giving subjects test stimuli that were in the preceding digit series. This was done because the response time may theoretically be longer for negative responses and shorter for positive responses, (We see, however, that this was not the case). Together, these experiments lead to Sternberg�s serial-exhaustive conclusion regarding human memory scanning of images.
As in experiment #1, a trial for experiment #2 consisted of a warning signal after being shown the digit-series, followed by the test digit. The subjects would then respond, and finally, a feedback light would indicate whether the response, indicated by the subjects, was correct. The subjects were given 60 practice trials and 120 test trials. Since experiment #2 only administered correct test digit after each digit-series, Sternberg was able to compare this linear regression and slope to those results containing both from experiment #1. The linear regression of mean response latencies and the slope in experiment #2 were remarkably similar to those in the first experiment. Therefore, "a positive set in experiment 2 apparently plays the same role as a series of symbols presented in experiment #1, both corresponding to a set of symbols in memory and used in the selection of a response," (Sternberg, 353). Sternberg believes these findings to be indicative of some type of internal representation of the test stimuli, which are used to compare a subsequent test stimulus. To Sternberg, this internal comparison is done in linear succession because of the highly significant linearity of mean-latencies in response times. When a subject analyzed negative responses, they must continue making the comparison for the full length of the symbolic representation�"If positive responses were initiated as soon as a match had occurred (as in a self-terminating search), the mean number of comparisons on positive trials would be (S + 1) / 2 rather than S," (Sternberg, 353). This is because, once a match is made, the search process would come to an end if a self-terminating search technique was used. This would indicate that the "latency function for positive responses would have 1/2 the slope of the functions for negative responses," (Sternberg, 353). We know this was not the case, and in fact both experiments were equal in terms of linear regression and slope. "The equality of the observed slopes shows, instead, that the scanning process is exhaustive, even when a match has occurred, scanning continues through the entire series," (Sternberg, 353).

LINK TO STERNBERGS ORIGINAL PAPER!!