Retrieval of item frequency information in a continuous memory task

Abstract

The accuracy and response latency of absolute frequency judgments were measured as a function of test lag (the number of intervening items between presentations of a test item) in a continuous memory task. Frequency was varied from one to three presentations in Experiments 1 and 2 and from one to five presentations in Experiment 3. The proportion of correct responses decreased as frequency increased, and correct mean response time tended to increase with frequency. Both accuracy and correct mean response time were found to be largely a function of the most recent test lag. The lag-latency functions were best described by piecewise linear functions, with the breakpoint occurring between lags 1 and 2. Continuous frequency estimation was also shown to improve with extended practice. The implications of the results are discussed with reference to trace strength, numerical-inference, and multiple-trace theories of frequency discrimination.Since Sternberg's (1966) now classic study, response time has become an important dependent variable in the study of retrieval processes in recognition. Response time has been measured in numerous recognition paradigms, and a variety of models have been developed to account for both the accuracy and the latency of recognition performance (e.g., Atkinson & Juola, 1973;Murdock & Anderson, 1975;Pike, Dalgleish, & Wright, 1977;Ratcliff, 1978).In contrast, studies that have investigated the storage and retrieval of item frequency information have not, in general, employed response time as a dependent variable. There are, however, two exceptions. Voss, Vereb, and Bisanz (1975) found that the latency of absolute frequency judgments tended to be an inverted If-shaped function of frequency for frequencies varying from 2 to 32 presentations. Hintzman, Grandy, and Gold (1981) measured response time in a relative judgment of frequency task. They found that response time depended on the frequencies of both the chosen and the unchosen alternatives and that response time was longer the more similar were the frequencies of the two alternatives. Also, in contrast to the results of Voss et al., Howell (1973a) suggested that the measurement of response time for frequency judgments might serve to distinguish between theories of frequency discrimination. However, these theories are not, as yet, sufficiently precise to enable predictions concerning response latency. Hintzman (1976, pp. 59-62), for example, argued that it is not clear what kinds of latency functions the multiple-trace and the proposition-encoding theories predict. Depending on the assumptions made, the multiple-trace theory would appear to be able to accommodate either an increase or a decrease in response time with frequency (Hintzman et al., 1981) The inability of the extant frequency theories to make specific predictions of latency functions does not diminish the potential importance of response time in the study of frequency discrimination. The measurement of response time may serve to provide limits on the nature of the assumptions ...

Toward a synthesis of cognitive biases: How noisy information processing can bias human decision making.

“…The "overshoot" refers to estimates outside of the defined scope of the decision-making exercise (such as illusions) 4 . In line with many empirical studies, Hockley (1984), for example, simply "truncates" or "cuts off" the overshooting estimates (p.230). He deletes 354 answers that referred to a (non-existent) 4 th repetition in his ternary word repetition exercise.…”

“…Table 3 shows the beforementioned empirical finding of Hockley's (1984) word repetitions are overestimated (1.064 and 2.222), while 4 th and 5 th repetitions are underestimated (3.883 and 4.410). The empirically detected transition matrix, however, is not perfectly doubly stochastic, because the columns do not sum up to 1.…”

“…The result is thus only a rough approximation, but it often works quite well for the replication of communication systems in electrical engineering. I discuss the fitting process further in Appendix D. Source: Author, in reference to empirical data from Hockley (1984). Figure 10b shows the fitting limitations in a transition matrix.…”

“…These matrices represent the visual logic of channel graphs (like Figures 1, 2, 4, and 6b), but with the analytical benefit that the well-known tools of matrix algebra can be applied to analyze the given properties. Figure 10b corresponds to the traditional x/y-representation of Hockley's (1984) ternary experiment in Figure 6a, the channel representation in Figure 6b, and the normal noise approximation in figure 10a. All four kinds of representation show the same finding in different ways.…”

“…In an experiment with a very reliable sample size, Hockley (1984) analyzed 47,120 judgments on the repetitions of listed words (Experiment 1, p.230). The presented words were repeated up to three times.…”