1
Isolation and Serial Position Effect
Running head: ISOLATION AND THE SERIAL POSITION EFFECT
The Impact of an Isolated Item on the Serial Position Effect
Lisa A. Way
University of Baltimore
Abstract
In an experiment investigating the affect of an isolated item on the serial position effect, the investigators studied the recall of 30 subjects for the first, middle, and last words in presented lists. Fifteen subjects were shown 15 lists of words in which the middle word was highlighted. The other 15 subjects were shown 15 lists of words that did not contain a highlighted word. The data was collected in order to find the number of times each subject recalled the first, middle, and last words in each list for each group. A proportion of the number of times the middle word was recalled to the sum of the number of times the first and last words were recalled was computed for each subject. The results found significantly supported the hypothesis that in the presence of an isolated item the serial position effect would be diminished.
The Impact of an Isolated Item on the Serial Position Effect
This article deals with how the presence of a distinct item among other nondistinct items affects the recall of individuals. This is known as the von Restorff effect, also called the isolation effect. Specifically, this study will focus on the effect that isolation of an item has on serial position recall. It was hypothesized that the isolation effect would override the serial position effect.
The serial position effect has two parts, according to Atkinson and Shiffrin (1968). These two parts are known as the primacy effect and the recency effect. The probabilities of an item being recalled are usually plotted as a function of its serial position, this is known as the serial position curve and is usually a U-shaped curve (Rouder & Gomez, 2001). Atkinson and Shiffrin define the primacy effect as the advantage in recalling the first items in a given list, these items then being stored in long-term memory. This is due to the beginning items being rehearsed more than the ending items (Rundus, 1971), allowing time for these items to be placed into long-term memory. The rehearsal allows for the words to be pulled from long-term memory much easier than words that have not been rehearsed. The recency effect, on the other hand, involves recalling the items at the end of the list at a higher rate than those in the middle of the list. Atkinson and Shiffrin (1968) propose that this effect is due to short-term memory. The items in short-term memory are easily accessible and temporary. Therefore the items at the end of the list are more easily recalled because they are still in the short-term memory store.
The serial position effect has been studied in many past research investigations. Andrade, Werniers, May, and Szmalec (2002) used a visual noise technique to investigate whether it disrupted visual short-term memory involved in serial position recall. With their conclusion that this visual noise did not have an affect on visual short-term memory, they proposed that visual working memory is not analogous to verbal working memory. Similarly, Beaman (2002) investigated the modality effect, which is, “the recall advantage over the last one or two serial positions for verbal items presented auditorily over verbal items presented visually” (Beaman, 2002, p.371). Beaman found there to be many errors in theories that assume this modality effect regardless of the circumstances. A sufficient way to invert the modalities was also found.
There have also been many investigations regarding the nature of the coding of the positions of items for serial recall. Ng and Maybery (2002) studied whether position information is time dependent. Through three different experiments, they concluded that recall for serial order does not depend on temporal information. Their conclusion has been found in many other similar investigations as well.
The isolation effect proposes that an isolated item, in a list of otherwise similar items, would be remembered better than an item in the same relative position in a list in which all items are similar (Main, Leland, & Bartlett, 1998). This effect has been shown in many research investigations.
Cimbalo and Siska (1986) investigated their hypothesis that the isolation effect would be greater when there are more background items. They concluded that there was an “overall list facilitation effect (OLFE)” (Cimbalo & Siska, p.70). In other words this means that when there was an isolated item, in their study, subjects remembered both the isolated item and the surrounding items better. Cimbalo, McQuestion, and Witting (1980) found that overall memory for an entire list containing an isolated item is neither improved nor decreased. They stated that when the list was presented simultaneously the memory for the list was improved, however, when the items in the list were presented successively, the isolated item was recalled much more than the other items in the list.
Main, Leland, and Barlett (1998) specifically investigated the isolation effect to facial feature recognition. They hypothesized that simpler, distinctive stimuli would be recalled better than more complex stimuli. Their findings supported their hypothesis. They found evidence for three properties of one isolated item.
A single, simple, distinctive stimulus in an otherwise cohesive set is remembered better than two or more such stimuli; in a simultaneous-presentation paradigm, with a wide variety of contexts, it leads to better memory for the whole set than do zero, two, or more such stimuli; and it produces better behavioral control than zero, or two, or more such stimuli. (p.204).
Although there have been thorough investigations conducted on both the isolation effect and the serial position effect, few researchers have investigated the effects of one against the other. The present study was designed to incorporate all of the previously discussed research. Given all of the previous knowledge and the theoretical background of the two effects, it was hypothesized that when lists of items are presented to subjects, some containing an isolated item, the distinct item will override the serial position effect of that particular list of items. Specifically, that the isolated item will be recalled better than the other items, thus decreasing the primacy and recency effects.
Method
Participants
The participants in this study were all undergraduate students at the University of Baltimore. There were 15 males and 15 females making the total number of subjects in the study 30. The participants ranged in age from 17 to 35 years (M = 20.25, SD = 2.25).
Materials
The experiment was done in a standard classroom, which contained chairs and desks for the subjects to sit in. Each subject was given a sheet of paper and a pen or pencil. There were thirty lists of words typed out onto transparencies. Each list contained 15 randomly generated words. In half of the lists, the eighth word was highlighted with a yellow highlighter. A projector was used to present the lists of words onto a screen in the front of the classroom. A sheet of paper with a hole cut out of it, large enough for the subjects to see only one word at a time, was used to present the lists. The person displaying the lists of words to the subjects was a randomly selected professor at the University of Baltimore, who was unaware of the research question. This controlled for any experimenter bias effects.
Procedure
All participants were asked to spend a few minutes on a memory task. They were told that they could stop at any time. All subjects willingly agreed to complete the task (Webster, Leland, & Bartlett, 1997). The participants were then randomly assigned to group 1 and group 2, with 15 subjects in each group.
Group 1 was asked to leave the room and wait in the hallway. Once all of group1 were out of the room, the door was shut and group 2 was given a sheet of paper and a pen or a pencil. The professor explained to the participants that they were going to be shown 15 lists of words, one list and one word at a time, and that after each list, they would be asked to write down, on the sheet of paper distributed, as many of the words that they could recall. The projector was turned on. The 15 lists not containing the highlighted word were shown to this group. The piece of paper with the whole cut out of it was used to only allow one word to be seen at a time. Starting at the top of the list, the words were shown and read aloud. Each word was displayed for 5 seconds. When all of the words were displayed, the participants were given 30 seconds to write down, in any order, all of the words that they recalled. The subjects were then shown the remaining lists of words. The same procedure was followed for each list.
Group 2 was then asked to go out into the hallway and group 1 was asked to come into the room. The same procedure was followed for this group. The only difference was that group 1 was shown the 15 lists with the highlighted word. Fifteen lists were shown to each group in order to obtain a more accurate result. When group 1 was finished, group 2 was asked back into the room. The participants were then given a debriefing by the experimenter. The purpose of the study and the studies conditions were discussed.
The data was then recorded. The experimenter recorded the number of times that each subject recalled the first, middle, and last words out of the 15 lists that they were shown. This was done for each subject in both groups. In this study the independent variable was the presence or absence of the highlighted word in the list. The dependent variable was the proportion of the number of times the middle words, also the highlighted words, were recalled to the sum of the number of times the first and last words were recalled. An independent-samples t-test was used to determine if the difference between the proportions in group 1 and group 2 were significant.
Results
A proportion of the number of times the middle words were recalled to the sum of the number of times the first and last words were recalled was found for each subject in each group. As predicted, the proportion of recall was found to be higher for the participants that were shown the lists containing the highlighted word (M = 3.144, SD = 3.8645) than the proportion of recall for the participants that were shown the lists without the highlighted word (M = .371, SD = .2914). These results are shown in Table 1. The results were found to be statistically significant using an alpha level of .05, t(28) = 2.772, p = .05. These results are shown in Table 2.
Figure 1 shows the differences in recall between group 1 and group 2. This graph displays the mean number of times the first, middle, and last words were recalled for each group. It shows the presence of the serial position curve for the recall of group 2. This is compared with the absence of the serial position curve for the recall of group 1. For group 1, shown the lists with the highlighted word, the mean recall for the middle word, or the highlighted word, is much higher than the mean recall for the first and last words. Group 2, shown the lists without the highlighted word, had a higher mean recall for the first and last words than for the middle word. The exact values are shown in Table 3. Figure 1 is used to display the relationship graphically.
Discussion
The hypothesis that the presence of an isolated item would block the serial position effect was supported through the results. In this study, the first and last words were used to define the serial position effect. This idea incorporated Atkinson and Shiffron’s (1968) model of the serial position effect, which stated that the serial position effect was composed of the primacy and recency effects. Thus, the use of the first word in the lists represented the primacy effect and the use of the last word in the lists represented the recency effect. The isolated item was placed in the middle of the list in order to emphasize the results.
Figure 1 clearly shows the difference in recall between group 1 and group 2. The results suggest that when the middle word was highlighted, the first and last words were not recalled as often. When the middle word was not highlighted, the first and last words were recalled much more often than the middle word. The U-shaped serial position curve, which was suggested by Atkinson and Shiffron (1968) and supported by many following researchers including Rouder and Gomez (2001), was seen in the recall of group 2 and it was not seen in the recall of group 1.
One possible explanation for this could pertain to memory strategies. Rundus’ (1971) assumption that the primacy and recency effects are due to the long-term and short-term memory stores can explain the results found for group 2, which portray the serial position effect. In contrast, the results found for group 1 cannot be explained by Rundus’ (1971) assumption. Cimbalo and Siska (1986) suggest that the isolation effect could be due to the encoding processes of memory. They concluded that when the isolation of an item is embedded in an otherwise nondistinct array of items, the isolation provides extra clarity of the item and thus stronger encoding for memory. This may help to explain the results found for group 1, in that the encoding for the highlighted word may have been stronger than that for the first and last word due to its clarity. The stronger encoding for the highlighted word may have blocked the serial position effect. This may be related to the reason that students tend to highlight important readings in their studies. This may help them encode the information better and thus recall it better.
This study did not investigate the affects of the modality used, as in the studies conducted by Beaman (2002) and Andrade, Werniers, May, and Szmalec (2002). These studies investigated the differences and affects of verbal memory and visual memory. The present study controlled for those differences by having the experimenter show as well as read each word aloud one at a time. It is possible that this method strengthened the results by allowing both modalities for the encoding process.
This study also did not investigate the overall list facilitation effect (OLFE) that has been studied by many researchers, including Cimbalo and Siska (1986), and Cimbalo, McQuestion, and Witting (1980). The present study did not take this effect into account when analyzing the results. It did not investigate the idea and therefore concentrated on the highlighted word and not the words in close proximity to it. This may have limited the studies external validity. It cannot support and therefore cannot be generalized to those investigations involving the overall list facilitation effect.
Relating to time for encoding, was an assumption taken from a study conducted by Webster, Leland, and Bartlett (1997). This study states that an isolation, “must be a very simple stimulus, because a complex stimulus would itself require processing time that would detract from the time available for encoding the surrounding material” (Webster, Leland, & Bartlett, 1997, p.404). The present study therefore simply highlighted the isolated word in order to keep the isolation simple and so as not disturb the results of the study. This may have strengthened the internal validity of the study, as well as the external validity by making the results more generalizable. The present study also controlled for the affects of time on recall between the two groups by having the experimenter display each word for 5 seconds. This also strengthened the internal validity of the study.
The internal validity, however, may have been threatened by certain undetermined variables. The experimenter, for example, may not have followed the directions precisely and therefore not allowed exactly 5 seconds between each word. If this time varied, even slightly, it may have had an affect on the results of this study in comparing the two groups. Also it is possible that the experimenter may have paused slightly longer at the highlighted word, thus allowing more time for encoding. If this were to happen the results would have been greatly threatened.
The external validity of the study could also be threatened. It is risky to generalize the results of the present study to situations beyond auditory and visual lists of words. The participants were placed in a classroom setting and not in their natural environments. They knew that they were being tested on their memory ability and may have wanted to do well. Although this also may affect the internal validity of the study, this needs to be taken into account when generalizing the results of the present study to other situations.