Academic journal article Memory & Cognition

Is Memory for Stimulus Magnitude Bayesian?

Academic journal article Memory & Cognition

Is Memory for Stimulus Magnitude Bayesian?

Article excerpt

This study was designed to determine whether memory for stimulus values is a Bayesian weighting of the magnitude of a stimulus and the central tendency of an exemplar's category (Huttenlocher, Hedges, & Vevea, 2000). In five experiments, participants reproduced the remembered size of a geometric figure drawn from one of two categories whose means for size differed. Reproductions were biased toward the mean of the combined distribution rather than the mean of either category. Reproductions were also influenced by the size of the stimulus on the preceding trial. Neither of these results is entirely consistent with the view that recollections are partially constructed from a consideration of the long-run probabilities established by category membership.

Experience tells us that significant discrepancies can exist between our memories and actual events. For example, many of us have had the experience of walking to our cars only to find that we did not park where we thought we had. A common idea is that these errors are not random but are in fact a result of systematic distortions (Bartlett, 1932). In many cases, these distortions occur because we mistakenly believe that a particular event "followed the norm." Thus, we may walk to the wrong part of a parking lot because we typically park there, even though we did not park there today.

Recently, Huttenlocher and her colleagues (Huttenlocher & Hedges, 1992; Huttenlocher, Hedges, & Duncan, 1991; Huttenlocher, Hedges, & Vevea, 2000) have advanced an explanation of this phenomenon. According to their category adjustment model, we utilize two sources of information to reconstruct our memories for particular events or objects. One source is knowledge about the value of the stimulus that we encountered. They argue that instead of thinking about this source of knowledge as exact, we should think of it as a distribution of values centered near the actual value that was encountered. The second source is knowledge about the central tendency of the category of events or objects from which this stimulus was sampled.

The claim of this theory is that we make use of categorical information because it generally improves the accuracy of our memories. According to the theory, factors such as time or interfering events decrease the reliability of our memories because they increase the spread of the values around the actual value of a previously encountered stimulus. Under these conditions, we can improve our accuracy by weighting our inexact representation of an individual item by the mean of the category to which it belongs because we are more likely to encounter stimuli with values close to the mean of the category than stimuli with relatively rare or extreme values. Within the framework of Bayes's theorem, this knowledge about the central tendency of a category provides information about the prior odds of a stimulus value. For example, if I saw a tall individual, my memory for the height of this individual might be represented by a range of values from 6 ft 1 in. to 6 ft 5 in. Although it is possible that I did in fact see a 6 ft 5 in. individual, it is much more likely that I saw someone whose height was closer to 6 ft 1 in. because more people are 6 ft 1 in. than 6 ft 5 in. This difference in the prior probability of the two heights means that the accuracy of my memory could be improved over the long run if I moved my estimate closer to 6 ft 1 in. because I am simply much more likely to encounter 6 ft 1 in individuals in the first place.

The observation that memory for stimulus magnitude is frequently less extreme than the direct perception of these magnitudes has a long history in the judgment literature and is well established for a variety of tasks. For example, the psychophysical function derived from magnitude estimates of remembered stimuli is frequently flatter than the corresponding psychophysical function for estimates based on direct perception (Kerst & Howard, 1978; Moyer, Bradley, Sorensen, Whiting, & Mansfield, 1978). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.