Information Processing in Animals

Memory Mechanisms

N. E. Spear, R. R. Miller

• Publisher: Psychology Press
• Year: 2014
• ISBN:
  9781317757696
• Edition: 1
• Topic: Psychology
• Subtopic: History & Theory in Psychology
• Index: Psychology

First published in 1982. During the past fifty years, dramatic changes have occurred in the use of laboratory animals to study learning and memory. Yet the basic reasons for this research, diverse as they are, have not changed. At one extreme is the need for relatively direct application of findings with animal models to medical or educational problems of humans; at the other extreme, the quest for understanding animal behavior for its own sake. It is probably fair to say that no chapters in this book represent either of these extremes, although in each case the author's purposes can be said to be like those of some scientists working in this area fifty years ago. In contrast to this continuity of purpose, the approach that scientists now take in this area of study is really quite different from that of most or all scientists in the 1930s.

1
SOP: A Model of Automatic Memory Processing in Animal Behavior

Allan R. Wagner
Yale University

Current information-processing theories of memory phenomena emphasize an interplay of “automatic” and “controlled” processing. The exact nature of the distinction between the two kinds of processing can be somewhat different in the hands of different theorists (Posner & Snyder, 1975; Shiffrin & Schneider, 1977). But the contrast generally addresses the degree to which processing is assumed to proceed mechanically according to relatively invariant operating characteristics and stable “structural” features of the system (automatic processing), versus the degree to which processing is assumed to call upon flexible routines that may be “volitionally” exercised in a task-specific manner under the attention of an executive monitor (controlled processing).

Given this distinction, it is commonplace to assume that the human subject is unique in the richness, pervasiveness, and plasticity of its controlled processing, limiting the direct generalizability of observed memory phenomena between the human and other species. On the same grounds, however, one could assume that the infrahuman subject may present us with a clearer visualization of examples of automatic processing, unembellished by contributions of controlled processing. Whether or not automatic processing is the same in detail in all species, we may be instructed by comparative investigations as to its likely forms in biological systems (including the human) and perhaps be constrained in what need be attributed to more complex, controlled processing. This is, of course, only a rephrasing, in modem terminology, of one of the major strategic rationalizations concerning the investigation of animal memory (Hobhouse, 1915; Morgan, 1894).

I begin with these remarks as they help to announce the intended scope of the theoretical speculations that follow. A model of a memory system is presented that is meant to be a theory of automatic processing. The acronym, ‘SOP,’ seemed fitting to suggest that it is concerned with what may be presumed to be the standard operating procedures in memory. Flexible routines of controlled processing could be envisioned to interact with the mechanics of the model. But the model itself carefully ignores such potential complications. This scope is consistent with the data that the model was primarily designed to address, i.e., the learning and retention of infrahuman subjects in circumstances of simple conditioning and habituation.

The model is a formalization and extension of a set of assumptions that my students and I have promoted over the last several years concerning the variability in “rehearsal” and the consequence of “priming of short-term memory” (Pfautz & Wagner, 1976; Terry & Wagner, 1975; Wagner, 1976, 1978, 1979; Wagner, Rudy, & Whitlow, 1973; Wagner & Terry, 1975). The nature of these assumptions is better articulated in due course. But it may help to introduce the present theoretical effort to hint briefly at them here and try to convey some sense of why the development of a more formal model such as SOP was deemed necessary (if not desirable) at this juncture.

Variable-Rehearsal (Priming) Theory

The central thesis has been that the immediate behavior, as well as the learning, occasioned by an external stimulus is dependent on the course of “active” representation of the stimulus in a short-term memory (STM) store. The more active the representation at some moment, or the longer active during some episode, the more vigorous will be the subject’s immediate response, the more persistent will be evidence of memory for the stimulus, and the more associative learning may be witnessed between the stimulus and other contemporaneous events (Wagner, 1976, 1978). Added to this thesis was the notion that two states of activation in STM could be distinguished, an especially active state (dubbed “rehearsal,” after the terminology of Atkinson & Shiffrin, 1968) and a less active state (given no special label). Then, it was importantly assumed that, whereas the presentation of a stimulus de nova normally would lead to a state of rehearsal of its representation in STM, the presentation of a stimulus that was already represented, or “primed,” in STM would not provoke this specially active state (Wagner, 1976, 1978).

Following this reasoning, it was possible to rationalize a considerable number of phenomena, in which a stimulus may be seen to be less effective than it otherwise would be, if it is preceded by a recent instance of the same stimulus or by some associatively related stimulus. The prevention of stimulus “rehearsal” due to the priming of STM could account for such phenomena as the short-term refractorylike effect in studies of habituation (Davis, 1969; Whitlow, 1975), the “conditioned diminution of the UR” in Pavlovian conditioning (Kimble & Ost, 1961; Kimmel, 1966), the less persistent memory of signaled versus unsignaled samples in a short-term memory paradigm (Terry & Wagner, 1975), and the apparent decrease in the associative learning occasioned by a signaled US in studies of “blocking” in Pavlovian conditioning (Kamin, 1969).

Problems of Indeterminancy

By this time there are a considerable number of analytical studies (e.g., Best & Gemberling, 1977; Pfautz & Wagner, 1976; Terry, 1976; Wagner, 1978) that further encourage the basic propositions. Nonetheless, it must be acknowledged that certain of the critical phenomena that have been listed are embraced by the formulation more in the sense of being allowed than being necessarily predicted.

Consider, for example, the phenomenon of the “conditioned diminution of the UR,” which is a reduction in the amplitude of the response following a US, when preceded by an associated CS than when presented alone. It is supposed that responding at the time of a US is determined finally by the degree of activation of the US representation that is provoked in STM. If a CS produces some measure of US representation prior to the US occurrence, it is assumed that the US itself will be less likely to produce a “rehearsed” representation and thus be less likely to provoke its usual response. But what is indexed in behavior is presumably not just this effect but the combined action of the CS and US. Will the total US representation in STM be less, immediately following a CS-US sequence, than following a US-alone presentation? Only on the possibility that what US representation in STM is instigated by the CS to provoke responding (as witnessed, for example, in some CR) is more than offset by the loss in contribution of US representation in STM by the US. The priming theory allows that this may happen and that a “conditioned diminution of the UR” may then be observed, but it does not necessarily predict it.

The same reasoning can be followed in regard to a phenomenon reported by Terry and Wagner (1975). A US was used as a sample in a short-term memory paradigm. If a US were presented, then one reinforcement contingency was in effect for a vibrotactual cue presented 2, 4, 8, or 16 sec later. If a US were not presented, then an opposite contingency was in effect. With such training, subjects learned to respond differentially to the vibrotactual cue depending on the prior presence versus absence of a US, and performed less accurately the longer the interval between the sample and the vibrotactual cue. The major observation was that if the US were preceded on test trials by a CS with which it had otherwise been paired, there was a reduction in its behavioral control at the time of the vibrotactual cue. An “expected” or “primed” US did not appear to be as memorable several seconds later as an “unexpected” or “unprimed” US. Terry and Wagner (1975), however, also reported an important ancillary observation. The CS that was otherwise paired with the sample US, when presented alone as a substitute sample, controlled delayed responding in the same direction, only to a lesser degree than did the US. It appeared that either the CS or the US could provoke a representation of the US in STM. If a CS → US sequence was less memorable than a US alone, it had to be because the positive contribution by the CS was more than offset by the attendant loss in contribution by the US. The priming theory clearly allows that this may happen but does not necessarily predict it.

These two example cases are not unusual.¹ Our theorizing has emphasized a particular presumptive effect of priming of short-term memory, a decreased likelihood of rehearsal that should decrease evidence of stimulus processing. In many circumstances the same manipulations of priming should be expected to have other effects that would appear to complement stimulus processing. Although we can be encouraged that the former effect has been detected (Kimble & Ost, 1961; Terry & Wagner, 1975) in spite of the latter possibilities, a determinate theory must acknowledge both. SOP is an attempt to do this. Because the more compreh...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Preface
• Prologue: Reminiscences
• 1. SOP: A Model of Automatic Memory Processing in Animal Behavior
• 2. Differences in Adaptiveness Between Classically Conditioned Responses and Instrumentally Acquired Responses
• 3. Within-Event Learning in Pavlovian Conditioning
• 4. Long-Delay Conditioning and Instrumental Learning: Some New Findings
• 5. Actions and Habits: Variations in Associative Representations During Instrumental Learning
• 6. Working Memory and the Temporal Map
• 7. Directed Forgetting in Animals
• 8. Short-Term Memory in the Pigeon
• 9. Studies of Long-Term Memory in the Pigeon
• 10. Postacquisition Modification of Memory
• 11. Mechanisms of Cue-Induced Retention Enhancement
• 12. Extending the Domain of Memory Retrieval
• Author Index
• Subject Index