Human memory is composed of at least two types of memory; implicit and explicit. Explicit memory refers to our conscious attempt at retrieving a past experience, whereas implicit memory refers to unconscious retrieval of information. Previous research on implicit and explicit memory has revealed that recall of studied words is affected by factors including list length, encoding context, and target set size (e.g., Gee, 1997; Nelson, McKinney, Gee, & Janczura, 1998; Nelson, Schreiber, & McEvoy; 1992). The experiment presented here will examine each of these variables in the context of the same experiment. The purpose of combining these variables is to evaluate the predictions of a specific model of implicit and explicit memory; the PIER model (Process Implicit and Explicit Representations), which was developed by Nelson, Schreiber, and McEvoy (1992).
Words in long term memory are related to other words, and the number of these associates of a given word tend to vary from word to word. For example, as you can see from Figure 1, the word CAT has only three associates in long term memory whereas the word FARMER has 19 words associated to it in long term memory. These words make up the associative sets of the words CAT and FARMER. Previous research has shown that target words with smaller sets of associates are more likely to be recalled than words with larger sets of associates (Nelson, et al., 1992). This phenomenon is called the target set size effect.
The effects of target set size vary with the type of encoding condition. Target set size effects are evident for a variety of different encoding processes and testing conditions (Nelson, et al., 1992). For the extralist cued recall task, targets are shown in isolation at study. Cue words are later presented to prompt recall of the studied target. A target with a large set size has a smaller probability of being sampled from among itsą associates because there are more competitors. Therefore, a target with a small set size has a greater probability for recall in the extralist cued recall task (Nelson, Gee, & Schreiber, 1992).
In the intralist cued recall task, meaningfully related cue-target pairs are presented together at study. The cue word from the pair is then presented to facilitate recall of the target word. Previous research has indicated that the set size effect is reduced or eliminated in the intralist cued recall task.
According to the PIER model, meaning set size effects are reduced in the intralist task because the cue creates a context that activates only the relevant associates of the target. (Nelson, et al., 1992). Thus, the cue determines which associates to activate, thus inhibiting irrelevant associates. As a result of this process, a target with a large set size effectively becomes a word with a small set size (Nelson, Gee, & Schreiber, 1992). Thus set size effects are reduced in the intralist task because there is no effective difference between targets with small and large set sizes. This is in comparison to using an extralist task, in which context does not effectively alter the number of active associates related to the target. Target set size effects decline when meaningfully related context words are presented at study, whether or not the cue word is later used for recall (Nelson, et al., 1992).
Critics of the PIER model argue that the reduction or elimination of the set size effect in the intralist task is really a result of a ceiling effect. A ceiling effect occurs when a variable reaches its highest possible value because the task is too simple. This is consistent with the finding that recall levels are much higher in the intralist relative to the extralist task.
The present research examines this interaction between set size and type of task to determine if it is due to the explanation offered by the PIER model or if it is a result of a ceiling effect. A third variable of list length has been added to investigate whether this interaction is due to a ceiling effect or occurs as a result of the impact of encoding context. The length of the word lists that were used in this study were manipulated at three levels of 20, 24, and 28 items. The basic expectation is that recall performance will be higher for shorter lists relative to longer lists. The point of manipulating list length is to examine the presence and absence of a ceiling effect. In the short list condition performance should be at or near ceiling whereas in the long list condition performance should not approach ceiling. If the set size effect is reduced or eliminated in the long list condition then the PIER model explanation is supported, however the presence of the set size effect in this condition will offer credence to the ceiling effect argument.
Preliminary results indicate that the main effect of task (intralist vs. extralist) was significant F(1, 75) = 174.66, p < .05. As was expected, performance in the intralist task (M = .85, SD = .14) was better than in the extralist task (M = .54, SD = .17). The main effect of set size was also significant F(1, 75) = 32.57, p < .05. Again, as predicted, words with small sets of associates (M = .75, SD = .19) were recalled better than those with large sets (M = .63, SD = .24). A significant interaction between set size and task was also found F(1, 75) = 7.51, p < .05. As you can see from Figure 2, the set size effect is reduced in the intralist task as compared to the extralist task. This finding is consistent with previous research (Nelson, et al., 1992). No other effects were significant.
If the reduction or elimination of the set size effect in the intralist task is due to a celing effect, then the interaction between set size and list length should be significant. This interaction would have taken the form of a set size effect in the long, but not in the short list length conditions. This critical interaction did not approach significance. Therefore, these preliminary results support the argument made by the PIER model that the set size effect should be reduced in all list length conditions. Data collection is currently ongoing, and this issue cannot be completely resolved until the data collection is complete. However at this point, it appears that the explanation offered by the PIER model is accurate.
References
Nelson, D.L., Gee, N.R., & Schreiber, T.A. (1992). Sentence encoding and implicitly activated memories. Memory & Cognition, 20, 643-654.
Nelson, D.L., McKinney, V.M., Gee, N.R., & Janczura, G.A. (1998). Interpreting the influence of implicitly activated memories on recall and recognition. Psychological Review, 105, 299-324.
Nelson, D.L., Schreiber, T.A., & McEvoy, C.L. (1992). Processing implicit and explicit representations. Psychological Review, 99, 322-348.
Gee, N.R. (1997). Implicit memory and word ambiguity. Journal of Memory and Language, 36, 253-275.