of Clustering in the Recall of Randomly Arranged Associates · W. A. Bousfield et al. The Journal of Psychology. Volume 36, – Issue 1. Bousfield, W.A. BousfieldThe occurrence of clustering in the recall of randomly arranged associates. Journal of General Psychology, 49 (), pp. Psychol., 49 (), pp. Google Scholar. Bousfield et al., W.A. Bousfield, B.H. Cohen, G.A. WhitmarshAssociative clustering in the recall of words.
Interpreting semantic clustering effects in free recall
The content is solely the responsibility of the authors and does not necessarily represent the official views of our supporting organizations. We quantify the degree of semantic clustering using the semantic clustering score Polyn et al.
Although the similarity values produced by each of these myriad similarity metrics are somewhat related, the pairwise correlations between the measures tend to be surprisingly low. University of Pennsylvania; Philadelphia, PA: Two measures of semantic similarity A.
Each dot corresponds to a single comparison between two words. The primacy, recency, and temporal clustering effects may be measured objectively by examining the relative probabilities of bkusfield or transitioning between items that appeared at each serial position on a studied list.
Because this procedure ensures that each recall will be followed by the most similar word that is yet to be recalled, by definition it will maximize the semantic bousfild score according to g p. We first divided the distributions of LSA-derived pairwise similarity values into equally sized bins the centers of the bins are plotted along the x -coordinate.
First, it is important to use multiple measures of semantic similarity if one is to obtain an accurate estimate of whether participants are semantically clustering their recalls. This process continues until the k th word is recalled. We then measure the degree of semantic clustering according to a different similarity metric, f.
The full distributions of similarity values derived from the two metrics are shown in Figure 1Panels A and B.
Distribution of the pairwise LSA-derived semantic similarity values for the words shown in Table 1. We order the words in the pool by their semantic similarity according to g p to i 1. As defined above, the semantic clustering score according to metric g p is maximized i. We next generate a percentile score by comparing the semantic similarity value boousfield to the next item in the recall sequence with the rest of the distribution.
In most free recall studies, g p is unknown. Associative clustering during recall.
Word association spaces for predicting semantic similarity effects in episodic memory. Journal of Experimental Psychology. Support Center Support Center.
There was a problem providing the content you requested
Rather, a near-ceiling clustering score may reflect the specific sequence of words presented to the participant, or the specific structure of the experiment. We found that the mean semantic clustering score was 0. However, the techniques developed here are equally applicable to arbitrary choices of n and k. Bpusfield oscillations distinguish true from bousfied memories.
National Center for Biotechnology InformationU. We expect that these biases are related to the form of the semantic similarity distributions derived from each measure see Fig. LSA represents one technique for deriving similarity values bousfiield automated text processing. Each simulated participant encounters many word lists, and we simulate a sequence of recalls after each studied list.
We have focused on a single semantic clustering 19553, the semantic clustering score Polyn et al. Given that the clustering bousfidld obtained using any given model of semantic similarity are likely to be only noisy reflections of any true patterns in the data, one should use multiple models of semantic similarity whenever possible.
The serial position effect of free recall. The same 5, randomly chosen item lists were used in both panels. Distribution of the pairwise WAS-derived semantic similarity values for the same words. Behavior Research Methods, Instruments and Computers. Specifically, we calculate the proportion of the possible similarity values that the observed value is greater than, since strong semantic clustering will cause the observed similarity values to be larger than average.