Short-term Visual Information Forgetting (PLE: Memory)

A.H.C. van der Heijden

• Publisher: Taylor & Francis
• Year: 2014
• Print ISBN: 9781138996151
• eBook ISBN: 9781317749028
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

When this title was originally published in 1981, the information processing approach to perception and memory was dominant in experimental psychology, and the research reported here had major implications for future development. After exploring the shortcomings of earlier work in this field, the author develops a new model which he shows to be capable of accounting for a variety of experimental data connected with human information processing, visual perception and attention.

The central theme which is discussed is how we select relevant and discard irrelevant information. The basic assumption is that all incoming information is identified, that is, it reaches and activates the appropriate lexical entries. A piece of identified information is described as a unit consisting of three distinguishable codes: a visual code, a lexical or semantic code and a motor or action code. Identified information decays fast, so selective attention operates by selecting those units which have to be saved from this rapid decay. In a sense, therefore, the human information processor is described as struggling against forgetting.

1 Whole report as a function of exposure duration

Summary

In this chapter it is investigated whether processing time or short-term memory (STM) capacity sets a limit in a task in which observers look at a briefly exposed array of numbers and try to write down as many of these numbers as possible.

Mackworth (1962, 1963a) reported results obtained in a similar type of experiment and showed that processing time is a crucial variable: the number of elements reported showed a linear increase with exposure duration. Broadbent (1971) makes this observation one of the cornerstones on which his filter notion in visual information processing rests:

We therefore have the evidence we wanted, that the selective process is limited in its rate of working rather than in the number of items it can pick off; that … we have a buffer store lasting a fixed length of time, and a subsequent serial process which can extract one item after another from the buffer until the limit of time is exceeded. (Broadbent, 1971, p. 173)

Sperling (1960) reports the results of a similar experiment and shows that the exposure duration has a negligible effect on the number of elements reported.

The results of the experiment reported in this chapter showed an increase, but not a linear increase in the number of elements reported with exposure duration. A search of the literature showed that often a non-linear relation between number of elements reported and exposure duration is found.

The results of the experiment reported and of other results reported in the literature can be described with a simple equation. Contrary to the linear function fitted by Mackworth, the non-linear function derived from the equation gives estimates of the buffer duration that are in accord with independent estimates of buffer duration in the visual case.

The equation suggests a two-process model for this type of task: on the one hand, the elements on the stimulus card are processed at a rate of ‘b’ elements per unit of time, and on the other hand, during each unit of time a proportion of the elements already identified disappear from STM, or cannot be retrieved.

Besides a number of advantages that this interpretation has in common with other interpretations of the equation, it also seemed to have two specific advantages. First, it presents only a minor deviation from Broadbent’s (1971) point of view. It remains possible to interpret parameter b as reflecting the serial extracting of items from the buffer. The only new feature added is the forgetting of elements already extracted and identified. Second, this interpretation applies to the whole range of exposure durations, that is, there is no need for postulating different processes for different ranges of exposure durations.

This interpretation was taken as a background for further theorizing in chapters 2 and 3.

One problem with this interpretation is that it leads to STM notions that appear rather obsolete in the light of recent advances in STM research. In an appendix a number of alternative interpretations are given that are possibly more compatible with recent theories on STM, and that at least cannot be ruled out given the evidence presented in this chapter.

Introduction

In a visual whole report experiment, an observer is shown a display containing a number of elements, for example, digits or letters, and is asked to name or write down as many of the elements as possible after the exposure. When the exposure duration is such that only a single glance is possible (that is, no shift in fixation is possible during the exposure), only a limited number of elements presented are reported by the observer. Woodworth and Schlosberg (1954) and Neisser (1967) mention an average of between four and seven elements.

This limit on the number of items reported after a brief exposure can be seen as just another manifestation of a general limitation of human memory, ‘the span of immediate memory’, and is then called ‘the span of attention’, ‘the span of apprehension’ (cf. Miller, 1956; Neisser, 1967) or ‘the span of immediate memory’ (Sperling, 1960, 1963). With this interpretation, the upper limit on the number of elements reported after a brief exposure only depends on the capacity of a memory that has to hold the results of processing the display, until written or oral reporting takes place. This explanation will be called the ‘span hypothesis’. Proponents of this explanation regard exposure duration as a variable of minor importance. Only if the time that the visual information is available for processing is artificially reduced to times much shorter than the duration of a fixation, an effect of exposure duration is to be expected.

Another explanation of the limit on the number of elements reported with brief exposures may be called the ‘processing time hypothesis’. This states that with brief exposures there is not enough time to transfer information about the elements to a postcategorical level. The number of elements reported only depends on the time for which the visual information is present and the rate of processing the information. Only in the case of extended exposures do the capacity limitations of the memory that has to hold the names of the elements have an effect on the number of elements reported.

In the early 1960s, two independent attempts were made to incorporate the ‘whole report’ data in more general information processing models. Both explanations were maintained and investigated with rather different results. Sperling (Sperling, 1960, 1963; Averbach and Sperling, 1961) was the prime advocate of the ‘span hypothesis’ and gathered much evidence in favour of this explanation, whilst Mackworth (Mackworth, 1962, 1963a) advocated the ‘processing time hypothesis’ and carried out a number of experiments that firmly supported this point of view. (Later on, both investigators (Sperling, 1967, 1970; Mackworth, 1972) changed some of their interpretations of their data. For ease of exposition and as we are mainly concerned with the results of their experiments we will confine ourselves to the data and interpretations mentioned in the earlier papers.)

There are three points both investigators agreed upon.

1 Visual information from brief exposures is not only available for processing during the exposure, but persists for some time after the exposure as a ‘rapidly fading visual image of the stimulus’ (Sperling, 1960, p. 26), or as a ‘visual image’ (Mackworth, 1962, 1963a). Upon arrival, visual information is stored in VIS (visual information storage), which ‘acts as a buffer which quickly attains and holds much information to permit its slow utilization later’ (Sperling, 1963, p. 22). Subsequently, Neisser (1967) gave the term ‘iconic memory’ or ‘icon’ to this rapidly decaying visual persistence from which a subject could read information as if the stimulus was still present.

2 During stimulus exposure and icon, items are ‘read’ or processed one after another and transferred to a more endurable memory. This memory was named ‘immediate memory’ (Sperling, 1960), ‘Auditory Information Store’ or ‘AIS’ (Sperling, 1963), ‘the memory trace’ (Mackworth, 1962), and ‘post-perceptual immediate-memory’ (Mackworth, 1963a).

3 Only during the first interval of time that the stimulus information is available for processing is the rate of increase of the number of elements in the more endurable memory constant. After that interval, the processing rate decreases (Sperling, 1963) or a reduction in the efficiency of information intake in that memory will take place (Mackworth, 1962, 1963a).

To conclude, there is a general agreement on the outline of a model for visual information processing in whole report tasks. Visual information is stored in a buffer memory and outlasts objective exposure. Information about the elements in the display is serially transferred to a more durable memory.

There is, however, a remarkable disagreement between the two investigators about the parameters of this model, namely, the duration of the icon, the rate of processing the elements, and the interval of time that the processing rate is constant. These parameters, except Sperling’s estimate of icon duration, are derived from whole report experiments in which the relation between number of elements reported and exposure time is investigated. (Sperling’s estimate of icon duration is based upon the results of his well-known partial report experiments. Recent evaluations of this method are given by Dick (1974) and Coltheart (1975a). Mackworth (1963b) gives additional estimates of icon duration using methods other than whole report.) Of course, this type of experiment provides the most direct way for deciding between the ‘span hypothesis’ and the ‘processing time hypothesis’.

Two types of whole report experiments with varying exposure duration were used. Mackworth (1962) (and also Mackworth, 1963a, exp. 3, and Sperling, 1960, exp. 2) had the display followed by a homogeneous post-exposure field. Sperling (1963, p. 25) (see also Sperling, 1960, p. 24) and Averbach and Sperling (1961, p. 202) presented data obtained from an experiment in which the display was followed by a patterned post-exposure field, consisting of densely scattered pieces of letters. In Sperling’s opinion, such a ‘noise’ field erases the icon, leaving only visual information storage during exposure.

With both procedures it was found that the first part of the function relating the number of elements reported to the exposure time, was linear, or, could be characterized by

X = aT + b,

where X equals the number of elements reported, and Τ equals t...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• Acknowledgments
• Figures
• Tables
• Author’s note
• Introduction
• 1 Whole report as a function of exposure duration
• 2 The processing of multi-element visual displays
• 3 The selection of information within a fixation
• 4 The ‘Stroop test’ and the ‘modified visual probe experiments’
• 5 The logogen model and the Stroop phenomenon
• 6 A tentative model for performance in the ‘modified Stroop test’
• 7 Further speculations and reinterpretations
• Bibliography
• Author index
• Subject index