The Encoding of Frequency Information as an Automatic Process: A Quantitative Review

People use information on how often events occur as the basis for many decisions they must make. For example, it has been shown that people will credit a statement as being more true the more frequently they have heard it. Additionally, probability information derived from observations of the frequency of occurrence of events has been shown to play a role in concept formation. One of the primary controversies in the study of how individuals process frequency information revolves around the contention that the encoding of frequency information is an automatic process (one that requires little or no attentional capacity and does not interfere with the performance of other tasks) as opposed to a controlled process (one that requires considerable attentional capacity and does interfere with the performance of other tasks). Considerable research has addressed the issue of whether frequency information is encoded automatically. Much of this research has focused on those variables which should not affect the encoding of frequency information if this process were in fact automatic as asserted by Hasher and Zacks (1979, 1984) (e.g., age of subjects, expectancy of upcoming frequency test, and attentional capacity). The results of research addressing the automaticity of the process required for encoding frequency information in memory have been mixed. Therefore, in the current study, meta-analyses were conducted for each of the variables with the potential to influence the encoding of frequency information (cognitive ability, expected test, strategy, levels of processing, age, capacity, encoding variability, practice, and generation). Within each of the meta-analyses, characteristics of the individual studies which may have mediated that studies results were recorded and analyzed (type of dependent task, type of dependent measure reported, test expectancy, type of stimuli used in the study, and the number of to-be-remembered items in the study list). Results of the meta-analyses indicated that (a) the frequency information encoded when different processing strategies are used is different from the frequency information encoded when no processing strategies, or less beneficial processing strategies, are used, (b) the encoding of frequency information is not invariant to developmental differences later in life, and (c) limiting the encoding capacity of individuals attempting to encode frequency information interferes with this encoding. These results are inconsistent with the automatic view of frequency information encoding. Furthermore, analyses of the study characteristics revealed several problems of methodological interest