Perception, Cognition, and Development

Interactional Analyses

T. J. Tighe, B. E. Shepp

• Publisher: Psychology Press
• Year: 2019
• Print ISBN: 9780898592542
• eBook ISBN: 9781317738138
• Subtopic: Cognitive Psychology & Cognition

This volume is based on a conference held at Dartmouth College's Minary Conference Center in Holdemess, New Hampshire, June 4 -7 , 1981. The conference brought together a number of investigators whose separate lines of inquiry bear in significant ways on the relationships among perception, cognition, and development. The purpose was to consider interactions among these basic processes not only as a critical facet of the research programs of the participants but also as a central conceptual problem for current theoretical psychology. First published in 1983. Routledge is an imprint of Taylor & Francis, an informa company.

1	Asymmetric Interactions of Stimulus Dimensions in Perceptual Information Processing

Wendell R. Garner

Yale University

Introduction

For more than the past decade much of my research has been focused on the nature of the dimensions that generate the stimuli we use in our experiments and, more specifically, on the nature of the interactions between these dimensions. Most of this research (mine as well as that of others) has assumed or experimentally demonstrated an equivalent status of the two or more dimensions used in generating a stimulus; that is to say, any interactions between the stimulus dimensions are assumed or shown to be symmetric. There is an increased interest, however, in those types of stimulus dimensions where the roles of the two or more dimensions are not equivalent, where the interactions between the dimensions are asymmetric. Much of what we know about different types of symmetric dimensional interactions, at least for adults, I have summarized else-where (Gamer, 1974b, 1976) and I shall refer to this material only when necessary to clarify or to contrast with the idea of asymmetric dimensional interaction. Instead, the purpose of this chapter is to summarize some of what we know about asymmetric dimensional interactions, to clarify some differences between various types of asymmetric interactions, and to suggest other types of asymmetric interaction that may exist.

As in writing any report, I found it necessary to make some delimiting decisions about the topics to be covered. Current information processing research is filled with research on processing levels or stages, and these ideas imply an asymmetry of processing at least and, therefore, potentially an asymmetry of the properties of the stimuli that are processed. So the more general topic of asymmetric information processing, including hierarchical and contingent processing, would encompass most of current human information processing research. I have limited my topic to those types of stimuli that can be defined as generated from dimensions without even being too concerned about distinctions between dimensions and features (Gamer, 1978a). Thus I remain interested primarily in the nature of the stimulus properties. This decision, however, requires me to omit material that may at some later time interrelate with the material I shall discuss here. Certainly all the research on hierarchical processing of words and sentences is necessarily omitted, and that is a very large area of research indeed. Another area of research and theorizing that I regret omitting is that concerning asymmetric similarity. To illustrate, Rosch (1975) showed that a nonreference stimulus is judged more similar to a reference stimulus (such as a vertical line or a prototypical color) than the converse, and Tversky (1977) provided a theoretical framework within which such asymmetric similarities might exist, as did Krumhansl (1978). The reason I especially regret omitting this material is that the stimuli used in much of this research are sufficiently simple (i.e., artificial) so that it might be possible to find a connection between asymmetric similarities and some of my ideas.

I shall discuss interactions involving stimulus dimensions. In doing so, I shall consider two major types of interaction: (1) those involving primarily asymmetry of a process without any necessary equivalent asymmetry in the logical structure of the stimuli; and (2) those involving primarily an asymmetry in the logical structure of the stimulus or stimulus set, without any necessary presumption that an experimental outcome will show an equivalent asymmetry. As always with a dichotomization, the flow between the two categories of interaction will be free, because when we are interested in process, we are also interested in the stimulus properties that lead to process differences, and when we are interested in stimulus structure, we are interested in the processing consequences of the differences in structure. Nevertheless, the distinction between asymmetries of process and asymmetries of structure will provide a useful working system. There are many other ways of categorizing the types of issue, and Treisman (1979), in a paper concerned with levels of processing, has made a most useful distinction between sequential and temporal ordering of levels. So there are other distinctions that can usefully be made, and the ones I make here are chosen primarily because they easily fit the types of stimuli and experiments I want to discuss.

ASYMMETRIES OF PROCESS

I shall discuss three types of process asymmetry in which stimulus dimensions are important: selective attention, response interference, and name encoding. Although in each case it is possible to argue that there is a logical stimulus structure that determines the process asymmetry, it seems more likely that the asymmetries are due to processing differences rather than to the logical relations between dimensions.

Selective Attention to Dimensions

When I first began investigations of dimensional interaction with information processing tasks (Gamer & Felfoldy, 1970), it seemed necessary to distinguish only between two types of interaction: integral and separable. Integral dimensions were those that showed an increase in speed of processing when the two dimensions were correlated and a failure of selective attention when orthogonal stimulus sets were used with classification required of one dimension as relevant with the other dimension varying irrelevantly. Separable dimensions were those that showed no improvement in processing speed when the dimensions were correlated but did allow selective attention to both dimensions. In all cases examined in that report the results were symmetric in that both of a pair of dimensions showed either the redundancy gain or selective attention. The world then seemed simple and dichotomous with regard to this issue, but it did not remain so very long. By 1974, (Gamer, 1974b) it was clear that there were at least some kinds of dimensions for which the information processing results were asymmetrical, and I introduced the idea of asymmetric integrality or separability. By 1976 (Gamer, 1976), I had added the idea of asymmetric configural dimensions. In both cases, the critical processing task that led to asymmetry was the selective attention task, with selective attention being possible for one dimension but not the other.

Linguistic Dimensions. The experiments that first led to the need to consider asymmetry of selective attention were by Day and Wood (1972) and by Wood (1974, 1975), and in both cases the stimuli were auditory, with one dimension involving a linguistic property and the other involving a nonlinguistic property. I shall describe in particular the stimuli and results of Wood (1975). He used four stimuli, as usual in these experiments, which were formed from the orthogonal combinations of a consonant sound and pitch, with the vowel sound being constant. The four stimuli were bae-high pitch, bae-low pitch, gae-high pitch, and gae-low pitch. When discrete reaction times were measured for pairs of stimuli differing on a single dimension, reaction time for discrimination of pitch was 411 msec and that for discrimination of the consonant was 416 msec. Thus these two control measures showed that the two dimensions were equally discriminable, a very important fact to establish in experiments of this kind. When the classification task was used, reaction time for pitch as the relevant dimension with consonant as the irrelevant dimension rose very slightly to 413 msec, not a meaningful increase. When, however, the consonant was the relevant dimension, with pitch now the irrelevant dimension, reaction time rose to 467 msec. In other words, subjects could selectively attend to pitch, ignoring variations in the consonant; but they could not selectively attend to the consonant, ignoring irrelevant variations in pitch.

In a discussion of these and other results, Wood (1974) makes clear that no simple serial or parallel model can explain the totality of his results, especially because in his 1974 experiment a decrease in reaction time was obtained when these two dimensions were used in a correlated fashion. However, as far as the selective attention results are concerned, some form of explanation of the asymmetric results that considers that the linguistic property is processed at a higher level than pitch is not unreasonable. Thus in terms of levels of processing, a property processed at a lower level can be selectively attended from one processed at a higher level, but a property processed at a higher level will be interfered with by one processed at a lower level.

Alternatively, however, there is at least the possibility of an explanation in terms of the logical structure of the stimulus properties (Gamer, 1974b, p. 136ff), and that is based on the necessary relations between the two types of dimensions. An auditory stimulus with a consonant and vowel must have a pitch, whereas an auditory stimulus with a pitch need not have a linguistic property. This asymmetry of the required logical relations between the two dimensions may be all that is necessary to explain the asymmetric selective attention. On the other hand, the logical properties of the stimuli may in fact determine processing levels, in which case the two explanations would be quite consistent.

Although it is fairly comfortable to interpret the Wood (1974) experiments as indicating a difference in processing levels, pitch being processed more peripherally, other experimental results with auditory stimuli serve to remind us that answers are not simple or easy. Blechner, Day, and Cutting (1976) found the same asymmetry in selective attention, but with stimulus dimensions that are not easily described as phonetic and nonphonetic. Their stimuli were tones, and their dimensions were intensity of tone and rise time to maximum intensity of the tones. Irrelevant variations in intensity interfered with judgments of rise time, but irrelevant variations in rise time did not interfere with judgments of intensity. No ready explanation of this result in terms of levels of processing is available, although the data do not, of course, in any way vitiate such an explanation.

Another series of experiments (Eimas, Tartter, Miller, & Keuthen, 1978) found asymmetric selective attention with stimuli and dimensions that have no obvious difference in levels of processing. Their stimuli were consonant-vowel syllables, the two dimensions being place of articulation and manner of articulation. As an example, the syllables ba, da, va, and za were used in one experiment. The syllables ba and va are labials, and da and za are alveolar, thus differing in place of articulation. The syllables ba and da have stop consonants, whereas the syllables va and za have fricative consonants, thus differing in manner of articulation. In several experiments these authors found that, although both dimensions prevented selective attention, the amount of interference was much greater when manner varied irrelevantly with judgments of place than the converse. Still further, in one of the very few experiments to manipulate relative discriminability of the dimensions, they found that discriminability differences did not alter the relative interference effects of the two dimensions. As with the Blechner et al. (1976) study, these results do not rule out an interpretation in terms of differences in levels, but the authors chose to interpret their results as indicating the kind of concurrent-contingent processing that Turvey (1973) suggested as an explanation of visual pattern masking. The idea is that processing of both dimensions (or levels) goes on concurrently, at least overlapping in time, but that the processing of one dimension is contingent on processing of the other; such a relationship leads to an asymmetry of selective attention but does not allow perfect selective attention for either dimension. This idea of concurrent-contingent processing is related to the discussion on page 23ff in which I suggest ways of generating stimulus sets that actually require contingent processing for successful performance, in some cases concurrently.

Visual Dimensions. Because of the possibility that the necessary logical relations of dimensional existence could account for the asymmetry of selective attention, I (Gamer, 1974b) suggested that possibly color and form would provide the same kind of asymmetry of selective attention, on the grounds that color can exist without a form, but a form must have a color to define its boundaries. Thus irrelevant color variations would be expected to interfere with selective attention to form, but irrelevant form variation would not be expected to prevent selective attention to color. Furthermore, Gottwald and I (Gottwald & Gamer, 1972) had done a card-sorting experiment requiring selective attention to form or to color and had found a small difference indicating that selective attention to color might be easier.

Schroeder (1976; data also reported in Posner, 1978, p. 69) carried out an experiment with color and form, using both the classification and discrimination techniques that I have used in such studies and the same–different task made so popular by Posner. With the same–different procedure, evidence for selective attention comes primarily from the results for the same response. If a subject is instructed, for example, to say same when the two stimuli are the same color, then an actual pair of stimuli requiring the response same might be identical in both color and form or might be identical in color but different in form. If it takes longer to respond same when the pair of stimuli are different in form than when they are identical in form, then ther...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• Table of Contents
• Preface
• 1. Asymmetric Interactions of Stimulus Dimensions in Perceptual Information Processing
• 2. The Analyzability of Multidimensional Objects: Some Constraints on Perceived Structure, the Development of Perceived Structure, and Attention
• 3. Holistic and Analytic Modes in Perceptual and Cognitive Development
• 4. Stimulus Preferences As Structural Features
• 5. Labeling, Overtraining, and Levels of Function
• 6. Components in the Hypothesis-Testing Strategies of Young Children
• 7. Structural Principles in Categorization
• 8. Intuitive Physics: Understanding and Learning of Physical Relations
• 9. The Perception and Use of Information by Good and Poor Readers
• 10. Some Issues on the Relation between Perceptual and Cognitive Development
• 11. Commentary on the Development of Perception and Cognition
• 12. Categorization, Perception, and Learning
• Author Index
• Subject Index