HANDBOOK OFPSYCHOPHYSIOLOGY

Edited by

JOHN T. CACIOPPOUniversity of Chicago

LOUIS G. TASSINARYTexas A & M University

GARY G. BERNTSONOhio State University

PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGEThe Pitt Building, Trumpington Street, Cambridge, United Kingdom

CAMBRIDGE UNIVERSITY PRESSThe Edinburgh Building, Cambridge CB2 2RU, UK www.cup.cam.ac.uk40 West 20th Street, New York, NY 10011-4211, USA www.cup.org10 Stamford Road, Oakleigh, Melbourne 3166, AustraliaRuiz de Alarcón 13, 28014 Madrid, Spain

© Cambridge University Press 2000

This book is in copyright. Subject to statutory exception andto the provisions of relevant collective licensing agreements,no reproduction of any part may take place withoutthe written permission of Cambridge University Press.

First published 2000

Printed in the United States of America

Typefaces Sabon 9.75/12 and Univers System AMS-TEX [FH]

A catalog record for this book is available from the British Library

Library of Congress Cataloging in Publication DataHandbook of psychophysiology / edited by John T. Cacioppo, Louis G.Tassinary, Gary G. Berntson. – 2nd ed.p. cm.Rev. ed. of: Principles of psychophysiology. 1990.Includes bibliographical references and index.ISBN 0-521-62634-X (hardback)1. Psychophysiology handbooks, manuals, etc. I. Cacioppo, JohnT. II. Tassinary, Louis G. III. Berntson, Gary G. IV. Principlesof psychophysiology.[DNLM: 1. Psychophysiology. WL 103 H2363 2000]QP360.P7515 2000612.8 – dc21DNLM/DLCfor Library of Congress 99-36718

CIP

ISBN 0 521 62634 X hardback

CONTENTS

Preface page ix

List of Contributors xii

Part One. Introduction

1 PSYCHOPHYSIOLOGICAL SCIENCE 3CACIOPPO, TASSINARY, & BERNTSON

Part Two. Foundations

2 HUMAN ELECTROENCEPHALOGRAPHY 27DAVIDSON, JACKSON, & LARSON

3 EVENT-RELATED BRAIN POTENTIALS 53FABIANI, GRATTON, & COLES

4 POSITRON EMISSION TOMOGRAPHY AND FUNCTIONAL MAGNETICRESONANCE IMAGING 85REIMAN, LANE, VAN PETTEN, & BANDETTINI

5 NEUROPSYCHOLOGY AND BEHAVIORAL NEUROLOGY 119TRANEL & DAMASIO

6 THE PUPILLARY SYSTEM 142BEATTY & LUCERO-WAGONER

7 THE SKELETOMOTOR SYSTEM: SURFACE ELECTROMYOGRAPHY 163TASSINARY & CACIOPPO

8 THE ELECTRODERMAL SYSTEM 200DAWSON, SCHELL, & FILION

9 CARDIOVASCULAR PSYCHOPHYSIOLOGY 224BROWNLEY, HURWITZ, & SCHNEIDERMAN

10 RESPIRATION 265HARVER & LORIG

11 THE GASTROINTESTINAL SYSTEM 294STERN, KOCH, & MUTH

12 THE SEXUAL RESPONSE SYSTEM 315GEER & JANSSEN

13 STRESS HORMONES IN PSYCHOPHYSIOLOGICAL RESEARCH: EMOTIONAL,BEHAVIORAL, AND COGNITIVE IMPLICATIONS 342LOVALLO & THOMAS

v

vi CONTENTS

14 REPRODUCTIVE HORMONES 368SNOWDON & ZIEGLER

15 PSYCHOLOGICAL MODULATION OF CELLULAR IMMUNITY 397UCHINO, KIECOLT-GLASER, & GLASER

16 PSYCHOSOCIAL FACTORS AND HUMORAL IMMUNITY 425SANDERS, ICIEK, & KASPROWICZ

Part Three. Processes

17 FROM HOMEOSTASIS TO ALLODYNAMIC REGULATION 459BERNTSON & CACIOPPO

18 INTEROCEPTION 482DWORKIN

19 MOTOR PREPARATION 507BRUNIA & VAN BOXTEL

20 COGNITION AND THE AUTONOMIC NERVOUS SYSTEM: ORIENTING,ANTICIPATION, AND CONDITIONING 533ÖHMAN, HAMM, & HUGDAHL

21 LANGUAGE 576KUTAS, FEDERMEIER, COULSON, KING, & MÜNTE

22 EMOTION AND MOTIVATION 602BRADLEY

23 INTERPERSONAL PROCESSES 643GARDNER, GABRIEL, & DIEKMAN

24 DEVELOPMENTAL PSYCHOPHYSIOLOGY: CONCEPTUAL ANDMETHODOLOGICAL PERSPECTIVES 665FOX, SCHMIDT, & HENDERSON

25 SLEEP AND DREAMING 687PIVIK

Part Four. Applications

26 PSYCHOPHYSIOLOGY IN THE STUDY OF PSYCHOPATHOLOGY 719KELLER, HICKS, & MILLER

27 PSYCHOPHYSIOLOGICAL APPLICATIONS IN CLINICAL HEALTHPSYCHOLOGY 751STONEY & LENTINO

28 THE DETECTION OF DECEPTION 772IACONO

29 APPLICATIONS OF PSYCHOPHYSIOLOGY TO HUMAN FACTORS 794KRAMER & WEBER

30 ENVIRONMENTAL PSYCHOPHYSIOLOGY 815PARSONS & HARTIG

Part Five. Methodology

31 PSYCHOMETRICS 849STRUBE

32 SALIENT METHOD, DESIGN, AND ANALYSIS CONCERNS 870JENNINGS & STINE

33 BIOSIGNAL PROCESSING 900GRATTON

CONTENTS vii

34 DYNAMIC MODELING 924GREGSON & PRESSING

Part Six. Appendices

35 GENERAL LABORATORY SAFETY 951GREENE, TURETSKY, & KOHLER

36 FUNCTIONAL MRI: BACKGROUND, METHODOLOGY, LIMITS, ANDIMPLEMENTATION 978BANDETTINI, BIRN, & DONAHUE

Index 1015

PSYCHOPHYSIOLOGICAL SCIENCE

JOHN T. CACIOPPO, LOUIS G. TASSINARY, & GARY G. BERNTSON

What is the relationship between the mind and brain? Whyis it that in intense fear and anger one’s bodily processesfeel as if they have a mind of their own? Do feelings andthoughts affect physiological processes and health? Canone learn to control physiological responses and, if so,does the control of these responses or the associated feed-back from the responses modulate cognition and emotion?What is the basis of consciousness? How exactly does oneperson’s mood spread to others? Psychophysiology is a fieldconcerned with just such questions. In psychophysiology,the mind is viewed as having a physical substrate. Con-trary to unidirectional reductionism, however, the mind isviewed as being more comprehensible if the structural andfunctional aspects of this physical substrate are consideredin conjunction with broader aspects of the functional out-puts of that substrate, including verbal, behavioral, andcontextual data. As such, psychophysiology is an interdis-ciplinary field that emphasizes multiple levels of analysis.

It is an exciting time in the field of psychophysiology.Investigations of elementary physiological events in normalthinking, feeling, and interacting individuals are now fea-sible, and new techniques are providing windows throughwhich psychological processes can be viewed unobtrusively.Instrumentation now makes it possible for investigatorsto explore the selective activation of discrete parts of thebrain during particular psychological operations in nor-mal individuals. Developments in ambulatory recordingand computing equipment make it possible to measure pe-ripheral physiological and endocrinological responses innaturalistic as well as laboratory settings. And the emer-gence of accepted recording standards, signal representa-tion, and multivariate statistical analyses have facilitatedstudy of the interrelationships among autonomic, somatic,brain, endocrinologic, and immunologic processes. How-ever, the views from these windows are clear only becauseof the deliberate efforts of knowledgeable investigators.

Knowledge and principles of physiological mechanisms,biometric and psychometric properties of the measures,statistical representation and analysis of multivariate data,and the structure of psychophysiological inference are im-portant if veridical information is to be extracted frompsychophysiological data. These are among the topics cov-ered in depth in this Handbook.

Objections have been raised to the notion that the mindcan be understood any better by including investigation ofits physiological manifestations. For instance, some havesuggested that changes in consciousness are sufficient toexplain social behavior, and that it is more important tostudy how people perceive and react to events than toreduce consciousness to physiological events (e.g. Allport1947; Caldwell 1994). Consciousness has been classified ina variety of ways but is typically thought to include changesin awareness of emotions, needs, motives, cognitions, orexpectations (cf. Kipnis 1997). However, concerns aboutthe fragility of theoretical structures based on self-reportdata have been expressed for more than half a century(Cook & Selltiz 1964; Gutek 1978; Turner & Martin 1984).As Kipnis (1997) noted:

Unlike the physical and biological sciences, however, theseclassification systems describe dynamic mental processes thatcannot be directly seen, touched, or measured. . . . Despite thesplendor of these visions about how the mind works, theseclassifications are particularly vulnerable to challenge and re-placement. This is because, as I said, there are no empiricalprocedures used to verify their independent existence. (p. 207)

Others have argued that we do not know, and are not soonto discover, the physical counterparts of psychological pro-cesses. Kipnis (1997), for instance, regarded the obstaclesto discovering the physical counterparts to these dynamicalprocesses as being overwhelming at present and thereforeargued to explain “social behavior in relatively objective

John T. Cacioppo, Louis G. Tassinary, and Gary G. Berntson (Eds.), Handbook of Psychophysiology, 2nd ed. © Cambridge University Press 2000.Printed in the United States of America. ISBN 62634X. All rights reserved.

3

4 CACIOPPO, TASSINARY, & BERNTSON

changes in technology, rather than in subjective changes inmental states” (p. 208).

There are undoubtedly psychological, social, and cul-tural phenomena whose secrets are not yet amenable tophysiological analyses. Yet elementary cognitive and be-havioral processes are being illuminated by psychophysi-ological methods, and more complex cognitive, affective,social, and clinical phenomena are increasingly succumbingto psychophysiological inquiries. Where most psychophys-iologists differ from Kipnis (1997), therefore, is not in theimportance placed on considering objective environmentalfactors and contexts, but rather in the value placed onpursuing the physical manifestations of psychological pro-cesses and of using biological data to inspire and constrainpsychological research and theory.

Consider the range of human experiences that resultsfrom exposure to light. The human eye is a sensory recep-tor organ that is sensitive to only a narrow band of theelectromagnetic spectrum. One could refer to this limitedband of electromagnetic energy as the “psychological” do-main, since electromagnetic phenomena falling within thisband have clear and obvious effects on human experience.However, the visible light band is but a small part of abroader, coherent spectrum of electromagnetic energy thatcan also influence human experience and behavior and thatcan be revealed through the use of specialized equipmentto extend the reaches of the human senses. Exposure toultraviolet light, although imperceptible to the human eye,can alter visual perception by damaging receptors in theeye, or it can influence behavior indirectly by producingsunburn; exposure to infrared light, which is also imper-ceptible to the human eye, can affect thermal sensationsas well as body temperature. Knowledge of the broaderorganization of the electromagnetic spectrum and its rela-tionship to human vision also makes it possible to designspecialized instruments that render ultraviolet and infraredlight visible to the human eye. These developments, andour understanding of the effects of electromagnetic energyon various aspects of human behavior, would not be likelyif methods were limited to verbal reports and observationslimited to the band of visible light. An illustration is the“blindsight” of patients with visual cortical damage, whocan accurately respond to light without conscious aware-ness (Weiskrantz 1986).

Recent studies further indicate that visual perceptionis not simply a faithful representation of the image castby a stimulus on the retina; rather, it is a neural im-age formed by neural mechanisms over the entire visualpathway. When the visual features of an external stimu-lus produce the experience of a different external stimulus,we call it an illusion. This label does not explain the dis-crepancy between the external stimulus and the consciousexperience of this stimulus, but recent psychophysiologi-cal investigations demonstrate how such a discrepancy mayoccur. Tootell et al. (1995), for instance, used functional

magnetic resonance imaging (fMRI) to observe brain activ-ity as individuals watched an expanding grid on a screenand again as individuals viewed the grid when it stoppedexpanding. When the grid stops expanding, individualsexperience an illusion that makes it appear as if the grid iscontracting rather than sitting motionless on the screen, anillusion known as the waterfall effect. Tootell et al. (1995)found that activity in the medial temporal (MT) region ofthe visual system continued during this illusion. These re-sults, combined with animal studies in which stimulationof neurons in the MT region affected visual perception, areconsistent with the notion that the pattern of neuronal ac-tivity in visual pathways (rather than the external stimulusper se) gives rise to the visual perception of movement.

Studies of the psychophysiology of cold provide anotherillustration of how consciousness can fit within a broadertheoretical context, the nature of which would not be evi-dent by an exclusive focus on verbal reports of feelingstates. Physiological processes are temperature-sensitive,and various mechanisms have evolved to maintain corebody temperature within the narrow ranges necessary fornormal functioning. Most of these mechanisms (e.g., de-creased eccrine gland activity to reduce cooling throughevaporation, peripheral vasoconstriction to shunt heatedblood away from the surface of the skin, shivering to pro-duce additional body heat) are initiated and can operateentirely outside the range of consciousness. If automaticadjustments to raise body temperature prove insufficient,the individual becomes aware of the cold and of the needfor willful action to raise body temperature. Mentation canalso foster behavioral flexibility and creative adaptation. Inthis example an individual’s actions could take any of avariety of forms, ranging from donning additional clothingto moving to a warmer environment. Thus, consciousnessis not a divine endowment that orchestrates the organism’sactions from a site on high but instead reflects brain pro-cesses that evolved because they confer adaptive advantageto the species. Consciousness may facilitate our abilityto establish goals, engage in exploratory behavior, antic-ipate and predict rewards and punishments, contemplatethe likely outcomes of behavioral options, learn vicari-ously, engage in counterfactual reasoning, and more. Butrather than viewing consciousness as the ultimate achieve-ment of the human species that can only be understoodunto itself, most psychophysiologists view consciousness asan important but narrow band of influences, instantiatedin brain processes and relevant to the governance of humanexperience and behavior. In this regard, the approach fol-lowed by contemporary psychophysiologists is reminiscentof that first adopted by seventeenth-century astronomers:

Nothing could be more obvious than that the earth is stableand unmoving, and that we are the center of the universe.Modern Western science takes its beginning from the denial ofthis commonsense axiom. . . . Common sense, the foundation

PSYCHOPHYSIOLOGICAL SCIENCE 5

of everyday life, could no longer serve for the governanceof the world. When “scientific” knowledge, the sophisticatedproduct of complicated instruments and subtle calculations,provided unimpeachable truths, things were no longer as theyseemed. (Boorstin 1983, p. 294)

Accordingly, psychophysiological research has provided in-sights into almost every facet of human nature, from theattention and behavior of the neonate to memory andemotions in the elderly. This book is about these insightsand advances – what they are, the methods by which theycame about, and the conceptualizations that are guidingprogress toward future advances in the discipline.

Historically, the study of psychophysiological phenom-ena has been susceptible to “easy generalizations, philo-sophical pitfalls, and influences from extrascientific quar-ters” (Harrington 1987, p. 5). Our objectives in thischapter are to define psychophysiology, review briefly ma-jor historical events in the evolution of psychophysiologicalinference, outline a taxonomy of logical relationships be-tween psychological constructs and physiological events,and specify a scheme for strong inference within each ofthe specified classes of psychophysiological relationships.

The Conceptualization ofPsychophysiology

“The body is the medium of experience and the instrumentof action. Through its actions we shape and organise ourexperiences and distinguish our perceptions of the outsideworld from sensations that arise within the body itself”(Miller 1978, p. 14). Anatomy, physiology, and psychophys-iology are all branches of science organized around bodilysystems whose collective aim is to elucidate the structureand function of the parts of and interrelated systems inthe human body in transactions with the environment.Anatomy is the science of body structure and the relation-ships among structures.

Physiology concerns the study of bodily function orhow the parts of the body work. For both of these dis-ciplines, what constitutes a “body part” varies with thelevel of bodily organization, ranging from the molecularto cellular to tissue to organ to body system to the organ-ism. Thus, the anatomy and physiology of the body areintricately interrelated.

Psychophysiology is intimately related to anatomy andphysiology but is also concerned with psychological phe-nomena – the experience and behavior of organisms inthe physical and social environment. The complexityadded when moving from physiology to psychophysiol-ogy includes both the capacity by symbolic systems ofrepresentation (e.g., language and mathematics) to com-municate and to reflect upon history and experience as wellas the social and cultural influences on physiological re-sponse and behavior. These factors contribute to plasticity,

adaptability, and variability in behavior. Psychology andpsychophysiology share the goal of explaining human ex-perience and behavior, and physiological constructs andprocesses are an explicit and integral component of theo-retical thinking in psychophysiology.

The technical obstacles confronting early studies, theimportance of understanding the physiological systems un-derlying observations, and the diverse goals and interestsof the early investigators in the field fostered a partitioningof the discipline into areas of physiology and measure-ment. The organization of psychophysiology in terms ofunderlying physiological systems, or what can be calledsystemic psychophysiology, remains important today fortheoretical and pedagogical reasons. Physiological systemsprovide the foundation for human processes and behaviorand are often the target of systematic observation. An un-derstanding of the physiological system(s) under study andthe bioelectrical principles underlying the responses beingmeasured contributes to plausible hypotheses, appropri-ate operationalizations, laboratory safety, discriminationof signal from artifact, acquisition and analysis of physio-logical events, legitimate inferences based on the data, andtheoretical advancement.

Psychophysiology – like anatomy, physiology, and psy-chology – is a broad science organized in terms of a the-matic as well as a systemic focus. The organization of psy-chophysiology in terms of topical areas of research can becalled thematic psychophysiology. For instance, cognitivepsychophysiology concerns the relationship between ele-ments of human information processing and physiologicalevents. Social psychophysiology concerns the study of thecognitive, emotional, and behavioral effects of human asso-ciation as related to and revealed by physiological measures,including the reciprocal relationship between physiologicaland social systems. Developmental psychophysiology dealswith ontological changes in psychophysiological relation-ships as well as the study of psychological developmentand aging through noninvasive physiological measurements.Clinical psychophysiology concerns the study of disordersin the organismic–environmental transactions and rangesfrom the assessment of disorders to interventions andtreatments. Environmental psychophysiology elucidates thevagaries of organism–place interdependencies as well as thehealth consequences of design through unobtrusive physio-logical measurements. And applied psychophysiology dealswith the implementation of psychophysiological principlesin practice, such as operant training (“biofeedback”), de-sensitization, relaxation, and the detection of deception.

In each of these areas, the focus of study draws onbut goes beyond the description of the structure or func-tion of cells or organs in order to investigate the organismin transactions with the physical or sociocultural envi-ronment. Some of these areas, such as developmentalpsychophysiology, have counterparts in anatomy and phys-iology but refer to complementary empirical domains that

6 CACIOPPO, TASSINARY, & BERNTSON

focus on human experience and behavior. Others, such associal psychophysiology, have a less direct counterpart inanatomy or physiology because the focus begins beyondthat of an organism in isolation. Yet the influence of socialand cultural factors on physiological structures and func-tions, and their influence as moderators of the effects ofphysical stimuli on physiological structures and functions,leaves little doubt as to the relevance of these factors foranatomy and physiology as well as for psychophysiology(see e.g. Barchas 1976; Cacioppo & Petty 1983; Cacioppo,Petty, & Andersen 1988; Waid 1984). For instance, Meaneyand colleagues (1996) provided evidence that rat pups whoare ignored by their mothers develop a more reactive hy-pothalamic pituitary adrenocortical (HPA) axis than ratpups who are licked and groomed by their mothers.

Psychophysiology is intimately related to anatomy andphysiology, and knowledge of the physiological systemsand responses under study contribute to both theoreti-cal and methodological aspects of psychophysiological re-search. However, as noted by Coles, Gratton, and Gehring(1987), knowledge of the physiological systems, thoughcontributory, is logically neither necessary nor sufficient toascribe psychological meaning to physiological responses.The ascription of psychological meaning to physiologicalresponses ultimately resides in factors such as the quality ofthe experimental design, the psychometric properties of themeasures, and the appropriateness of the data analysis andinterpretation. For instance, although numerous aspectsof the physiological basis of event-related brain potentialsremain uncertain, functional relationships within specificparadigms have been established between elementary cog-nitive operations and components of these potentials bysystematically varying one or more of the former whilemonitoring changes in the latter.

The point is not that either the physiological or the psy-chological perspective is preeminent, but rather that bothare fundamental to psychophysiological inquiries; morespecifically, that physiological and psychological perspec-tives are complementary. Inattention to the logic under-lying psychophysiological inferences simply because one isdealing with observable physiological events is likely tolead either to simple and restricted descriptions of empir-ical relationships or to erroneous interpretations of theserelationships. Similarly, “an aphysiological attitude, suchas is evident in some psychophysiological research, is likelyto lead to misinterpretation of the empirical relationshipsthat are found between psychophysiological measures andpsychological processes or states” (Coles, Donchin, &Porges 1986, pp. ix–x). Thus, whether organized in termsof a systemic or a thematic focus, psychophysiology canbe conceptualized as a natural extension of anatomy andphysiology in the scientific pursuit of understanding hu-man processes and behavior. It is the joint consideration ofphysiological and functional perspectives, however, that isthought to improve operationalization, measurement, and

inference and therefore to enrich research and theory oncognition, emotion, and behavior.

There has been little consensus regarding the formal def-inition of psychophysiology. Some of the early definitionsof the field were in operational terms such as research inwhich the polygraph was used, research published by work-ers in the field, and research on physiological responsesto behavioral manipulations (Ax 1964b; Furedy 1983).Other early definitions were designed to differentiate psy-chophysiology from the older and more established fieldof physiological psychology or psychobiology. Initially,psychophysiology differed from physiological psychologyin the use of humans in contrast to animals as partic-ipants, the manipulation of psychological or behavioralconstructs rather than anatomical structures or physiologi-cal processes, and the measurement of physiological ratherthan behavioral responses (Stern 1964). Although this her-itage is still in evidence, this distinction is often blurredby psychophysiologists who modify physiology with drugsor conditioning procedures and by psychobiologists whomanipulate psychological or behavioral variables and mea-sure physiological outcomes. Contemporary definitions aremore likely to emphasize the mapping of the relation-ships between and mechanisms underlying psychologicaland physiological events (e.g. Ackles, Jennings, & Coles1985; Hugdahl 1995).

A major problem in reaching a consensus has been theneed to give the field direction and identity by distinguish-ing it from other scientific disciplines while not limitingits potential for growth. Operational definitions are un-satisfactory for they do not provide long-term directionfor the field. Definitions of psychophysiology as studies inwhich psychological factors serve as independent variablesand physiological responses serve as dependent variablesdistinguish it from fields such as psychobiology, but theyhave been criticized as being too restrictive (Coles 1988;Furedy 1983). For instance, such definitions exclude stud-ies in which physiological events serve as the independent/blocking variable and human experience or behavior serveas the dependent variable (e.g., the sensorimotor behaviorassociated with manipulations of the physiology via drugsor operant conditioning, or with endogenous changes incardiovascular or electroencephalographic activity), as wellas studies comparing changes in physiological responsesacross known groups (e.g, the cardiovascular reactivity ofoffspring of hypertensive versus normotensive parents).

Moreover, psychophysiology and psychobiology – aswell as behavioral, cognitive, and social neuroscience –share goals, assumptions, experimental paradigms, andsometimes databases, but they differ primarily in terms ofanalytic focus. In psychophysiology, the emphasis is on in-tegrating data from multiple levels of analysis to illuminatepsychological functions and mechanisms rather than phys-iological structures per se. All of these substantive areashave a great deal to contribute to one another, and ideally

PSYCHOPHYSIOLOGICAL SCIENCE 7

this complementarity should not be masked in their defi-nition by the need to distinguish these fields. Indeed, theformulation of structure–function relationships is advancedto the extent that “top down” and “bottom up” informa-tion can be integrated.

At present, for instance, there are fundamental limita-tions in the ability of psychophysiological measures suchas brain imaging to accurately reveal functional localiza-tion (Barinaga 1997; Sarter, Berntson, & Cacioppo 1996a).Even for functions that are in fact localized to specific neu-ral circuits, these circuits may: be diffusely organized orwidely distributed; overlap anatomically or even share com-mon neuronal elements with circuits mediating differentfunctions; or perform different functions (e.g., facilitate orinhibit a given operation) depending on the patterns of in-put and activation associated with different cognitive statesor contexts. These possibilities would clearly complicateefforts to elucidate the cerebral localization of functions.Activation of diffusely organized circuits (or components ofcircuits), for example, may not yield a sufficiently intensesignal to be differentiated from noise or from activity pat-terns of control tasks. This could lead to a failure to detecta localized function. Alternatively, a focal, detected regionwithin a larger undetected area of activation could leadto an overestimate of the degree of functional localization.Similarly, the existence of functionally distinct but overlap-ping circuits could lead to an underestimate of the degreeof functional localization, as the overlapping area would beactivated during multiple cognitive or behavioral contexts.

Perhaps the more problematic possibility is that somecentral circuits may have differential and overlapping func-tions depending on the pattern of activation. A helpfulexample is provided by the research on the functions of“state-setting” systems (Mesulam 1990), particularly the at-tentional functions mediated via cortical cholinergic andnoradrenergic afferents (Sarter et al. 1996b). Research inthis area has resulted in fairly specific hypotheses about(i) the attentional processes that depend on the integrityof cortical cholinergic and noradrenergic afferents and (ii)the key role of attentional dysfunctions based on aberra-tions in activity of these systems in major neuropsychiatricdisorders (Robbins & Everitt 1987; Sarter 1994; Sarter &Bruno 1994). Whereas imaging studies have pointed to cor-tical areas in frontal, temporal, and parietal lobes thatmay be involved in various types of attention (Cohen etal. 1992; Corbetta et al. 1990, 1991; Grossman et al. 1992;Pardo, Fox, & Raichle 1991; Posner et al. 1988; Posner &Petersen 1990), more “bottom up” approaches aim at thedetermination of the specific role of acetylcholine in dif-ferent cortical areas in attentional functions (Metherate &Weinberger 1990).

An understanding of the role of the major afferent pro-jections to cortical areas in attention is facilitated by aconvergence of evidence from imaging studies in humanpatients and from animal studies on the cognitive effects of

manipulations of the activity of cortical inputs (Sarter etal. 1996a). Imaging studies may continue to relate sustainedattention to prefrontal and superior parietal areas (Pardoet al. 1991), for example, and decreases in the activity inthese areas may indeed be associated with the cognitivedecline in dementia (Mielke et al. 1994). In isolation, how-ever, the imaging approach is unlikely to reveal the precisefunctional contribution of individual afferent systems ofthese areas and, more generally, cannot discriminate dif-ferent neuronal activity patterns leading to identical signallevels. The point is not to contrast or even prioritizethe heuristic power of psychophysiological (top-down) orpsychobiological (bottom-up) approaches but rather to il-lustrate the significance of integrating relevant evidencefrom psychophysiology and the neurosciences.

The emergence of areas of research in psychoneuroen-docrinology (Frankenhauser 1983; Grunberg & Singer 1990;Mason 1972), behavioral neurology (Lindsley 1951; Tranel& Damasio 1985), and psychoneuroimmunology (Ader1981; Glaser & Kiecolt-Glaser 1994; Henry & Stephens1977; Jermott & Locke 1984) raises additional questionsabout the scope of psychophysiology. It is important tonote that anatomy and physiology encompass the fields ofneurology, endocrinology, and immunology owing to theircommon goals and assumptions as well as to the embod-iment, in a literal sense, of the nervous, endocrine, andimmunologic systems within the organism.

Psychophysiology, therefore, is based on the assump-tions that human perception, thought, emotion, and actionare embodied phenomena, and that physical (e.g. neuraland hormonal) responses can shed light on human na-ture. The level of analysis in psychophysiology is not onisolated components of the body but rather on organismic–environmental transactions. That is, psychophysiology rep-resents a top-down approach within the neurosciences thatcomplements the bottom-up approach of psychobiology.Thus, psychophysiology can be defined as the scientificstudy of social, psychological, and behavioral phenomenaas related to and revealed through physiological principlesand events in functional organisms.

In the following section, we review some of the majorhistorical developments that have influenced contemporarythinking and research in psychophysiology. As might beexpected from the discussion thus far, many of these earlydevelopments stemmed from studies of human anatomyand physiology.

Historical Developments

Psychophysiology is still quite young as a scientific field.Studies dating back to the turn of the century can befound involving the manipulation of a psychological fac-tor and the measurement of one or more physiologicalresponses (Berger 1929; Darrow 1929; Eng 1925; Jacobson1930; Mosso 1896; Peterson & Jung 1907; Sechenov 1878;

8 CACIOPPO, TASSINARY, & BERNTSON

Tarchanoff 1890; Wenger 1941; Wilder 1931; see also Wood-worth & Schlosberg 1954), and such studies would nowbe considered as falling squarely under the rubric of psy-chophysiology. Chester Darrow (1964), in the inauguralPresidential Address of the Society for Psychophysiologi-cal Research, identified Darwin (1873), Vigoroux (1879),James (1884), and Fere (1888) as among the field’s earliestpioneers. Yet the first scientific periodical devoted ex-clusively to psychophysiological research, the Psychophys-iology Newsletter, was not published until 1955 as anoutgrowth of the Polygraph Newsletter (Ax 1964a). TheSociety for Psychophysiological Research was formed fiveyears later, and the first issue of the scientific journal Psy-chophysiology was published but a quarter century ago.Precisely when psychophysiology emerged as a disciplineis therefore difficult to specify, but it is usually identifiedwith the formation of the Society for PsychophysiologicalResearch in 1960 or with the publication in 1964 of thefirst issue of Psychophysiology (Fowles 1975; Greenfield &Sternbach 1972; Sternbach 1966).

Although psychophysiology as a formal discipline is lessthan 50 years old, interest in interrelationships betweenpsychological and physiological events can be traced asfar back as the early Egyptian and Greek philosopher–scientists. The Greek philosopher Heraclitus (c. 600 b.c.)referred to the mind as an overwhelming space whoseboundaries could never be fully comprehended (Bloom,Lazerson, & Hofstadter 1985). Plato (c. 400 b.c.) sug-gested that rational faculties were located in the head;passions were located in the spinal marrow and, indirectly,the heart; and instincts were located below the diaphragmwhere they influenced the liver. Plato also believed thepsyche and body to be fundamentally different; hence, ob-servations of physiological responses provided no groundsfor inference about the operation of psyche (Stern, Ray,& Davis 1980). Thus, despite the fact that the periph-eral and central nervous system, brain, and viscera wereknown to exist as anatomical entities by the early Greekscientist–philosophers, human nature was dealt with as anoncorporate entity not amenable to empirical study.

In the second century a.d., Galen (c. 130–200) formu-lated a theory of psychophysiological function that woulddominate thought well into the eighteenth century (Brazier1959, 1961; Wu 1984). Hydraulics and mechanics were thetechnology of the times, and aqueducts and sewer systemswere the most notable technological achievements duringthis period. Bloom et al. (1985) suggested: “It is hardlyby accident, then, that Galen believed the important partsof the brain to lie not in the brain’s substance, but in itsfluid-filled cavities” (p. 13). Based on his animal dissectionsand his observations of the variety of fluids that perme-ated the body, Galen postulated that humors (fluids) wereresponsible for all sensation, movement, thoughts, andemotion, and that pathologies – physiological or behav-ioral – were based on humoral disturbances. The role of

bodily organs was to produce or process these humors, andthe nerves, although recognized as instrumental in thoughtand action, were assumed to be part of a hydraulic systemthrough which the humors traveled. Galen’s views becameso deeply entrenched in Western thought that they wentpractically unchallenged for almost 1,500 years.

In the sixteenth century, Jean Fernel (1497–1558) pub-lished the first textbook on physiology, De Naturali ParteMedicinae (1542). According to Brazier (1959), this bookwas well received, and Fernel revised and expanded thebook across numerous editions. The ninth edition of thebook was retitled Medicina, and the first section was enti-tled Physiologia. Although Fernel’s categorization of empir-ical observations was strongly influenced by Galen’s theory,the book “shows dawning recognition of some of the au-tomatic movements which we now know to be reflexlyinitiated” (Brazier 1959, p. 2). This represented a markeddeparture from traditional views that segregated the con-trol of human action and the affairs of the corporeal world.

Studies of human anatomy (e.g. Vesalius 1543/1947) dur-ing this period in history also began to uncover errors inGalen’s descriptions, opening the way for questions of hismethods and of his theory of physiological functioningand symptomatology. Within a century, two additionalevents occurred that had a profound impact on the na-ture of inference in psychophysiology. In 1600, WilliamGilberd (1544–1603) recognized a difference between elec-tricity and magnetism and, more importantly, argued (inhis book, Magnete) that empirical observations and exper-iments should replace “the probable guesses and opinionsof the ordinary professors of philosophy.”

In addition, the reign of authority as the source of an-swers to questions about the basis of human experienceand behavior was challenged by the work of such schol-ars as Galileo, Bacon, and Newton. Galileo (1564–1642)challenged knowledge by authority in matters of science,by which Galileo meant physical sciences and mathemat-ics. He argued that theologians and philosophers had noright to control scientific investigation or theories and thatobservation, experiment, and reason alone could establishphysical truth (Drake 1967). Galileo was also aware of lim-itations of sense data. Concerned with the possibility ofillusion and misinterpretation, Galileo believed that math-ematics alone offered the kind of certainty that could becompletely trusted. Galileo did not extend this reasoningbeyond the physical sciences, but scientific investigations ofthe basis of human experience and behavior benefited fromhis rejection of authority as a source of knowledge aboutphysical reality, his emphasis on the value of skepticism,and his insistence that more could be learned from re-sults that suggested ignorance (disconfirmation) than fromresults that fit preconceptions (confirmation).

Francis Bacon (1561–1626) took the scientific method astep further in Novum Organum (1620/1855), adding in-duction to observation and adding verification to inference.

PSYCHOPHYSIOLOGICAL SCIENCE 9

Bacon was not a scientist, yet he is regarded as a fore-runner of the hypothetico-deductive method (Brazier 1959;Caws 1967). Bacon’s formulation and subsequent workon the logic of scientific inference (cf. Platt 1964; Pop-per 1959/1968) led to the now-familiar sequence underlyingscientific inference: (1) devise alternative hypotheses; (2)devise a crucial experiment with alternative possible out-comes, each of which will disfavor if not exclude one ormore of the hypotheses; (3) execute the experiment toobtain a clear result; and (4) recycle to refine the remain-ing possibilities. Such a scheme was accepted quickly inthe physical sciences, but traditional philosophical and re-ligious views segregating human existence from worldlyevents slowed its acceptance in the study of human physi-ology, experience, and behavior (Brazier 1977; Harrington1987; Mecacci 1979).

William Harvey’s (1578–1657) doctoral dissertation, “DeMotu Cordis” (1628/1931), represented the first major workto use these principles to guide inferences about physiolog-ical functioning, and it also disconfirmed Galen’s principlethat the motion of the blood in the arterial and venoussystems ebbed and flowed independently of one another(except for some leakage in the heart). Pumps were an im-portant technological development during the seventeenthcentury, and Harvey perhaps drew on his observationsof pumps in positing that blood circulated continuouslythrough a circular system, pushed along by the pumpingactions of the heart and directed through and out of theheart by the one-way valves in each chamber of the heart.Galen, in contrast, had posited that blood could flow ineither direction in the veins. To test these competing hy-potheses, Harvey tied a tourniquet above the elbow of hisarm – just tight enough to prevent blood from returning tothe heart through the veins, but not so tight as to preventblood from entering the arm through the arteries. Theveins swelled below but not above the tourniquet, implyingthat the blood could be entering only through the arter-ies and exiting only through the veins (Miller 1978). Avariation on Harvey’s procedure is used in contemporarypsychophysiology to gauge blood flow to vascular beds.

During this period, which coincided with a world nowburgeoning with machines, the human eye was conceivedas functioning like an optical instrument. Images were con-ceived as projected onto the sensory nerves of the retina.Movement was thought to reflect the mechanical actionsof passive ballonlike structures (muscles) that were inflatedor deflated by the nervous fluids or gaseous spirits thattraveled through canals in the nerves. And higher men-tal functions were still considered by many to fall outsidethe rubric of the physical or biological sciences (Bloom etal. 1985; Brazier 1959; Harrington 1987). The writings ofRené Descartes (1596–1650) reflect the presumed divisionbetween the mind and body. The actions of animals wereviewed as reflexive and mechanistic in nature, as were mostof the actions of humans. But humans alone, Descartes

argued, also possess a consciousness of self and of eventsaround them – a consciousness that (like the body) was athing but (unlike the body) was not a thing governed bymaterial principles or connections. This independent entitycalled “mind,” Descartes proposed, presides over volitionfrom the soul’s control tower in the pineal gland locatedat the center of the head:

The soul or mind squeezed the pineal gland this way andthat, nudging the animal fluids in the human brain into thepores or valves, “and according as they enter or even only asthey tend to enter more or less into this or that nerve, theyhave the power of changing the form of the muscle into whichthe nerve is inserted, and by this means making the limbsmove.” (quoted in Jaynes 1973, p. 172)

Shortly following Descartes’ publication of Traite del’Homme (c. 1633), Steno (1638–1686) noted several discrep-ancies between Descartes’ dualistic and largely mechanisticcharacterization of human processes and the extant evi-dence about animal and human physiology. For instance,Steno noted that the pineal gland (the purported bridgebetween the worlds of the human mind and body) existedin animals as well as humans, that the pineal gland didnot have the rich nerve supply implied by Descartes’ the-ory, and that the brain was unnecessary to many animalmovements (cf. Jaynes 1973). Giovanni Borelli (1608–1679)disproved the notion that movement was motivated by theinflation of muscles by a gaseous substance. Borelli con-ducted experiments in which he submerged a strugglinganimal in water, slit its muscles, and looked for the re-lease of bubbles (Brazier 1959). These observations werepublished posthumously in 1680, shortly after the sugges-tion by Francesco Redi that the shock of the electric rayfish was muscular in origin (Basmajian & DeLuca 1985,chap. 1; Wu 1984).

Despite the prevalent belief during this period that thescientific study of animal and human behavior could applyonly to those structures they shared in common (Bloom etal. 1985; Harrington 1987), the foundations laid by the greatseventeenth-century scientist–philosophers encouraged stu-dents of anatomy and physiology in the subsequent cen-tury to discount explanatory appeals to the human soul ormind (Brazier 1959). Consequently, experimental analysesof physiological events and psychological constructs (e.g.,sensation, involuntary and voluntary action) expanded andinspired the application of technological advances to thestudy of psychophysiological questions. For instance, themicroscope was employed (unsuccessfully) in the late sev-enteenth century to examine the prevalent belief that thenerves were small pipes through which nervous fluid flowed.

According to Brazier (1959, 1977), that electricity mightbe the transmitter of nervous action was initially seen asunlikely because, drawing upon the metaphor of electricityrunning down a wire, there was believed to be insufficientinsulation around the nerves to prevent a dissipation of

10 CACIOPPO, TASSINARY, & BERNTSON

the electrical signal. Galvani and Volta’s (c. 1800) experi-ments demonstrated that nerves and muscles were indeedelectrically excitable, and research by Du Bois-Reymond(1849) established that nerves and muscles were electricallypolarized as well as excitable. Based on reaction times,Helmholtz (c. 1850) correctly inferred that nerves and mus-cles were not like wires because they propagated electricalimpulses too slowly. The work that followed ultimatelyverified that neural signals and muscular actions were elec-trical in nature, that these electrical signals were the resultof biochemical reactions within specialized cells, and thatthere was indeed some dissipation of these electrical sig-nals through the body fluids, dissipation that could bedetected noninvasively at the surface of the skin. Specificadvances during the nineteenth and twentieth centuries inpsychophysiological theory and research are discussed inthe remainder of this book. However, the stage had beenset by these early investigators for the scientific study ofpsychophysiological relationships.

Psychophysiological Relationships andPsychophysiological Inference

We praise the “lifetime of study,” but in dozens of cases, inevery field, what was needed was not a lifetime but rathera few short months or weeks of analytical inductive infer-ence. . . . We speak piously of taking measurements and mak-ing small studies that will “add another brick to the templeof science.” Most such bricks just lie around the brickyard.(Platt 1964, p. 351)

The importance of the development of more advancedrecording procedures to scientific progress in psychophysi-ology is clear, as previously unobservable phenomena arerendered observable. Less explicitly studied, but no lessimportant, is the structure of scientific thought aboutpsychophysiological phenomena. For instance, Galen’s no-tions about psychophysiological processes persisted for1,500 years – despite the availability for several centuries ofprocedures for disconfirming his theory – in part becausethe structure of scientific inquiry had not been developedsufficiently (Brazier 1959).

An important form of psychophysiological inference toevolve from the work of Francis Bacon (1620/1855) andGalileo (Drake 1967) is the hypothetico-deductive logicoutlined previously (cf. Platt 1964; Popper 1959/1968). Ifthe data are consistent with only one of the theoretical hy-potheses, then the alternative hypotheses with which theinvestigator began become less plausible. With conceptualreplications to ensure the construct validity, replicability,and generalizability of such a result, a subset of the originalhypotheses can be discarded, and the investigator recylesthrough this sequence. One weakness of this procedureis the myriad sources of variance in psychophysiologicalinvestigations and the stochastic nature of physiologicalevents, resulting in the sometimes poor replicability or

generalizability of the results. A second weakness is theintellectual invention and omniscience required to spec-ify all relevant alternative hypotheses for the phenomenonof interest. Because neither of these shortcomings can beovercome with certitude, progress in the short term canbe slow and uncertain. However, adherence to this se-quence provides grounds for strong inference in the longterm (Platt 1964).

SUBTRACTIVE METHOD INPSYCHOPHYSIOLOGY

Physiological responses are often of interest only to theextent that they allow one to index a psychological pro-cess, state, or stage. A general analytic framework thathas aided the design and interpretation of psychophysio-logical investigations is the “subtractive” method, whichhas been adapted from studies of mental chronometry(Donders 1868; cf. Cacioppo & Petty 1986; Coles 1989).At the simplest level, experimental design begins with anexperimental and a control condition. The experimentalcondition represents the presence of some factor, and thecontrol condition represents the absence of this factor. Theexperimental factor might be selected because it is theoret-ically believed to harbor n information processing stages,and the construction of the control condition is guidedto incorporate n − 1 information processing stages. Dif-ferences between these conditions (e.g., on reaction timemeasures) are thought to reflect the impact (e.g. duration)of the nth information processing stage.

The principle underlying the inclusion of physiologi-cal measures in these designs is twofold: (a) physiologicaldifferences between experimental conditions thought torepresent n and n − 1 processing stages support the theo-retical differentiation of these stages; and (b) the nature ofthe physiological differentiation of experimental conditions(e.g., the physiological signature of a processing stage) mayfurther support a particular psychological characterizationof that information processing stage. According to thesubtractive method, the systematic application of “stagedeletion” makes it possible to deduce the physiological sig-nature of each of the constituent stages underlying somepsychological or behavioral response. For instance, if theexperimental task (n+1 stages) requires 50 msec longer tocomplete than the control task (n stages), this result is con-sistent with the theoretical conception of the experimentaland control tasks differing in one (or more) processingstage(s) as well as with the differential processing stage(s)requiring approximately 50 msec to perform. Similarly, ifthe experimental task is characterized by greater activationof Broca’s area than the control task, this is consistent withboth the theoretical conception of the experimental andcontrol tasks differing in one (or more) processing stage(s)and the differential processing stage(s) relating to languageproduction.

PSYCHOPHYSIOLOGICAL SCIENCE 11

When one of these stages is thought to be responsiblefor the differential impact of two conditions on behavior,analyses of concomitant physiological activity can againbe informative in one of two ways. If the patterns ofphysiological activity resulting from the isolation of pre-sumably identical stages are dissimilar, then the similarityof the stages is challenged even though there may be simi-larities between the subsequent behavioral outcomes. Thegreater the evidence from multiple operationalizations thata particular stage is accompanied by a specific physiolog-ical profile across the ranges of stimuli employed in theinvestigations, the more challenging is the dissimilarity inobtained physiological profiles (Cacioppo & Petty 1986).

If, on the other hand, similar patterns of physiologicalactivity result from the isolation of stages that are hypoth-esized to be identical, convergent evidence is obtained thatthe same fundamental stage is operative. These data donot provide strong evidence that the stages are the same;still, the more peculiar the physiological profile is to agiven stage within a particular experimental context, thegreater the value of the convergent evidence (Cacioppo &Tassinary 1990).

There are two additional issues that should be con-sidered when using a subtractive framework to investigateelementary stages of psychological processes. First, thesubtractive method contains the implicit assumption that astage can be inserted or deleted without changing the na-ture of the other constituent stages. But this method haslong been criticized for ignoring the possibility that manip-ulating a factor to insert or delete a processing stage mightintroduce a completely different processing structure. Us-ing multiple operationalizations to insert or delete a stagemay be helpful but, as outlined in what follows, this doesnot ensure strong logical grounds for inference. If eachoperational insertion or deletion of a stage has the sameeffect, then investigators may have greater confidence thatthese effects are attributable to the conceptual processingstage of interest, or at least that the effects are not an arti-fact of an unintended confound. Parametric studies of eachprocessing stage can also provide important informationabout the range over which a stage manifests as a partic-ular physiological profile, thereby improving investigators’ability to generate appropriate comparison conditions. Afailure to find the same physiological profile across a widerange of a stimulus believed to invoke a given processingstage does not itself indicate whether a new stage is invokedor whether the old stage manifests differently at variouslevels of stimulation; however, it does indicate an importantlimitation when one is interpreting the physiological profilein a subsequent study of this processing stage. As Donchin(1982) suggested, “each hypothesis so tested generates pre-dictions for its own specific range of validity. The observedrelations may or may not be universally applicable” (p. 460).

Second, in order to construct the set of comparisontasks using the subtractive method, one must already

have a clearly articulated hypothesis about the sequenceof events that transpires between stimulus and overt re-sponse. This assumption renders the subtractive methodparticularly useful in testing an existing theory about thestages constituting a psychological process and in deter-mining whether a given stage is among the set constitutingtwo separate processes. Note, however, that confirma-tory evidence can still be questioned by the assertionthat the addition or deletion of a particular stage re-sults in an essentially different set of stages or substages.Again, the inclusion of several experimental and compari-son tasks (i.e., multiple operationalizations) appears to bejudicious.

Although there are important limitations on the assump-tion of simple additive models that make it more useful forstudying some types of processes (see e.g. Sanders 1980), thesubtractive method can also yield data that are helpful inderiving comprehensive, empirically based models of psy-chological processes. Again, assume that the comparisonis between experimental conditions with different impactson behavior. If the physiological profile that differentiatesthese conditions is similar to a distinctive physiologicalprofile that has been found previously to characterize aparticular processing stage, the possibility is raised thatthe same processing stage has been detected in anothercontext. The stronger and more distinctive the link be-tween a physiological profile and a processing stage withinthe ranges of stimuli used, the stronger is this possibility.Similarly, if the sequence of physiological activity differ-entiating two conditions can be modeled by concatenatingthe physiological responses found previously to character-ize even more rudimentary processing types or stages, thena model of a set of stages distinguishing these two condi-tions would be suggested. Again, converging evidence forthis empirically derived model could be marshalled fromother (e.g., observational or verbal) measures obtained inthe experiment and from subsequent experiments.

Whenever a physiological response (or profile) foundpreviously to vary as a function of a psychological pro-cessing stage or state is observed, the possibility is raisedthat the same processing stage or state has been detected.Much of the research in psychophysiology is based on justsuch reasoning. A person might be thought to be anxiousbecause they show physiological activation, inattentive be-cause they show diminished activation, happy because theyshow an attenuated startle response, and so on. However,one cannot logically conclude that a processing stage orstate has definitely been detected simply because a physio-logical response found previously to vary as a function of apsychological processing stage or state has been observed.(The logical flaw in this form of psychophysiological in-ference is termed “affirmation of the consequent.”) In thenext section, we present a general framework for think-ing about relationships between psychological concepts andphysiological events, and we discuss the rules of evidence

12 CACIOPPO, TASSINARY, & BERNTSON

Figure 1. Depiction of logical relations between elements in the psy-chological (9) and physiological (8) domains. Left panel: Linksbetween the psychological elements and individual physiological re-sponses. Middle panel: Links between the psychological elementsand the physiological response pattern. Right panel: Links betweenthe psychological elements and the profile of physiological responsesacross time.

for and the limitations to inference in each (see also Ca-cioppo & Tassinary 1990).

THE PSYCHOLOGICAL ANDPHYSIOLOGICAL DOMAINS

A useful way to construe the potential relationships be-tween psychological and physiological events is to considerthese two types of events as representing independent sets(domains), where a set is defined as a collection of ele-ments that together are considered a whole. Psychologicalevents – by which we mean conceptual variables repre-senting functional aspects of embodied processes – areconceived as constituting one set, which we shall call set9. Physiological (e.g., brain, autonomic, endocrinological)events, by which we mean empirical physical variables, areconceived as constituting another, which we shall call set8.1 Inspection of Figure 1 (upper left panel) reveals thatall elements in the set of psychological events are assumedto have some physiological referent; that is, the mind isviewed as having a physical substrate.2

Focusing first on the top row of Figure 1, the existenceof psychologically irrelevant physiological events (e.g., ran-dom physiological fluctuations; increased electrodermal ac-tivity due to minor variations in body temperature) is ofimportance in psychophysiology for purposes of artifactprevention or elimination. However, such elements withinthe physiological domain can be ignored if nonpsycholog-ical factors have been held constant, their influence on thephysiological responses of interest has been identified and

removed, or they do not overlap with thephysiological event of interest. These objec-tives are achieved through the application ofproper psychophysiological recording tech-niques. The important point here is that theachievement of these objectives simplifies thetask of specifying psychophysiological rela-tionships, in the ideal case, by eliminatingphysiological events that have no direct rel-evance to psychological events (see Figure 1,top row of panel B).

We can now state five general relationsthat might be said to map the elementswithin the domain 9 of psychological eventsto elements within the domain 8 of physi-ological events (see Figure 2); these may belisted as follows:

1. A one-to-one relation, such that an element in thepsychological set is associated with one and only oneelement in the physiological set and vice versa.

2. A one-to-many relation, meaning that an element inthe psychological domain is associated with a subset ofelements in the physiological domain.

3. A many-to-one relation, meaning that two or morepsychological elements are associated with the samephysiological element.

Figure 2. Possible relationships between elements in the psychological(9) and physiological (8) domains.

PSYCHOPHYSIOLOGICAL SCIENCE 13

4. A many-to-many relation, meaning that two or morepsychological elements are associated with the same (oran overlapping) subset of elements in the physiologicaldomain.

5. A null relation, meaning there is no association be-tween an element in the psychological domain and thatin the physiological domain.3

Of these possible relations, only the first and third al-low a formal specification of psychological elements as afunction of physiological elements (Coombs, Dawes, &Tversky 1970, pp. 351–71). This is important because, aswe have noted, psychophysiological research typically in-volves the manipulation of (or blocking on) elements in thepsychological domain and the measurement of elementsin the physiological domain. The grounds for theoreti-cal interpretations in psychophysiology can therefore bestrengthened if a way can be found to specify the relation-ship between the elements within 9 and 8 in terms ofone-to-one or, at worst, many-to-one relationships.

PHYSIOLOGICAL ELEMENTS AS SPATIAL ANDTEMPORAL RESPONSE PROFILES

If a one-to-one relation exists then an element in thepsychological set is associated with one and only one ele-ment in the physiological set. Such relationships providestrong grounds for inference but are not common in psy-chophysiology (Cacioppo & Tassinary 1990; Coles et al.1987; Donchin 1982). The functional opposite of the one-to-one relation is the null relation, which means that theelement in the physiological domain is unrelated to (andhence harbors no information about) the element in thepsychological domain.

Another form of relation between elements in the psy-chological and physiological domains is the one-to-many,meaning that an element in the psychological domain isassociated with a subset of elements in the physiologi-cal domain. Note that one-to-many relations between setsof psychological and physiological elements can be greatlysimplified by reducing them to one-to-one relations as fol-lows. Define a second set 8′ of physiological elements suchthat any subset of physiological elements associated withthe psychological element is replaced by a single elementin 8′, which now represents a physiological syndrome orresponse pattern. Thus, one-to-one and one-to-many re-lations between elements in the psychological domain 9

and elements in the physiological domain 8 both be-come one-to-one relations between 9 and 8′ (see Figure 1,panel B).

The remaining relations that can exist between elementsin these domains are the many-to-one, meaning that twoor more psychological elements are associated with thesame physiological element; and the many-to-many, mean-ing that two or more psychological elements in 9 areassociated with two or more of the same elements in 8

(see Figure 1, lower row of panel A). These relations canalso be simplified with a few changes in how one concep-tualizes an element in the physiological domain.

As before, a new set 8′ of physiological elements isdefined such that any subset of physiological elements asso-ciated with one or more psychological elements is replacedby a new element representing a profile of physiologicalresponses. Thus, the elements in 8′ again represent aspecific pattern or syndrome of physiological events, andmany-to-many relations between elements in 9 and 8 canbe reduced to many-to-one (or, in some cases, one-to-one)relations by viewing elements within the physiological do-main as representing singular physiological responses andphysiological response syndromes.

However, such a reconceptualization may not be suf-ficient to cast all of the psychophysiological relations ofinterest in terms of one-to-one or many-to-one relations.This may still not be a problem, because the set 8′

of physiological elements can be redefined such that theforms of physiological events as they unfold over timeare also considered to yield yet another set of physio-logical elements, 8′′ (Figure 1, panel C; cf. Cacioppo,Marshall-Goodell, & Dorfman 1983). By including bothtemporal and spatial information regarding the physiolog-ical events in the definition of the elements in 8′′, manycomplex psychophysiological relationships can be reducedto one-to-one or many-to-one relations.

ASYMMETRIES IN PSYCHOPHYSIOLOGICALINFERENCE

Research in which psychological or behavioral factorsserve as the independent (or blocking) variables and physio-logical structures or events serve as the dependent variablecan be conceptualized as investigating the P(8 |9), the“probability of 8 given 9.” Research in which physiologi-cal structures or events serve as the independent (or block-ing) variables and psychological or behavioral factors serveas the dependent variable can, in contrast, be conceptual-ized as investigating the P(9 |8).

For example, when differences in brain images or physi-ological events (8) are found in contrasts of tasks that arethought to differ only in one or more cognitive functions(9), the data are often interpreted prematurely as showingthat brain structure (or event) 8 is associated with cogni-tive function 9. These data are also treated as revealingmuch the same information that would have been obtainedhad brain structure (or event) 8 been stimulated or ablatedand a consequent change in cognitive function 9 observed.This form of interpretation reflects the explicit assumptionthat there is a fundamental localizability of specific cog-nitive operations, as well as the implicit assumption thatthere is an isomorphism between 8 and 9 (Sarter et al.1996a). Causal hypotheses regarding a specific brain struc-ture or process (8) underlying a cognitive operation (9)

14 CACIOPPO, TASSINARY, & BERNTSON

are of the form 9 = f(8); they necessarily imply that 8is always followed by 9 but do not imply that 9 is al-ways preceded by 8.4 Furthermore, brain events (8) maybe of interest to the extent that they index a cognitive op-eration or state, so that the inferences are of the form 9 =f(8) rather than not-9 = f(not-8). Thus, an aim of psy-chophysiology can generally be specified by the conditionalprobability of 9, given 8:

P(9 |8) = 1.

The typical structure of the imaging study (and mostpsychophysiological studies) can be represented by the con-ditional probability of 8, given 9:

P(8 |9) = x.That is, psychophysiological measures provide informa-

tion about 8 as a function of 9, but the conditionalprobabilities are equivalent (P(9 |8) = P(8 |9)) if andonly if there is a one-to-one relationship between physi-ological event 8 and cognitive function(s) or operation(s)9 (Cacioppo & Tassinary 1990). Interpreting psychophys-iological studies of the form P(8 |9) as equivalent topsychophysiological studies of the form P(9 |8) is notmisleading if one is dealing with one-to-one relationships.However, this is an assumption that must be tested. As-suming that brain areas become active during specificoperations, for instance, reflects a one-to-one relationshipand may hinder progress in psychophysiology unless thisassumption is tested or its implications for inference rec-ognized. These interpretations should be based on morethan mere consistency with the imaging data; ideally, evi-dence would also be presented demonstrating that theP(not-9,8) = 0 in the assessment context (in which case8 would be a marker of 9) or across contexts (in whichcase the relationship between 8 and 9 would be invari-ant). Strong inferences would then be possible.

Approaches such as stimulation and ablation studiesprovide complementary rather than redundant informationto traditional psychophysiological studies, in which phys-iological variables serve as dependent measures. This isbecause stimulation and ablation studies bear on the rela-tionship P(9 |8), whereas studies in which physiologicalvariables serve as dependent measures provide informationabout P(8 |9). Despite the formal parallelism between theexpression P(8 |9) and P(9 |8), there is a fundamentalasymmetry in the heuristic power of studies aimed at thedemonstration of P(9 |8) versus P(8 |9). The causal roleof 8 in process 9 can be examined by direct experimentalmanipulation of the physiology. The loss of cognitive func-tion by inactivation of neuronal processes (by mechanical,thermal, or neurochemical means) can serve to establish 8as causal for function 9. Addressing a more complex av-enue of research – facilitation of cognitive processes (e.g.,attention or memory) by electrical or neurochemical brain

activation – can further establish that 8 is a sufficientcondition for influencing process 9 (Sarter et al. 1996a).Whether the observed relationships have import in normal,functioning humans is informed by studies in which 8 asa function of 9 is examined.

Although both relationships P(9 |8) and P(8 |9) canbe studied experimentally, the complexity implied in stud-ies aimed at demonstrating P(8 |9) is based on the factthat experimental alteration of cognitive function 9 nec-essarily alters brain activities 8′,8, . . . (including those,say 8, that underlie 9). Thus, although a psychologicalcontext may yield concurrent manifestation in 9 and 8′,the causal linkage remains in question because an alter-nate (even undetected) brain event 8 could be causallymediating both 9 and 8′. In this case, excluding 8′ asa necessary and/or sufficient condition for the observed9 may require a tedious and exhaustive evaluation of thealternative hypothesis not-9 = f(not-8′). Note that thisconclusion does not challenge the importance and utility ofevidence generated by imaging methods, but it does pointto fundamental limitations in the strength of the infer-ence deduced from experimental approaches aimed solelyat the demonstration of P(8 |9). The integration of meth-ods and data from bottom-up and top-down approachesprovides a means of circumventing some of the thornierinterpretive problems of either approach alone, therebypermitting strong inferences in psychophysiology even inareas in which hypothesis-driven research is not yet possi-ble (Sarter et al. 1996a).

AN ILLUSTRATION

As Gould (1985) noted:

We often think, naively, that missing data are the primaryimpediments to intellectual progress – just find the right factsand all problems will dissipate. But barriers are often deeperand more abstract in thought. We must have access to theright metaphor, not only the requisite information. Revolu-tionary thinkers are not, primarily, gatherers of facts, butweavers of new intellectual structures. (Essay 9)

It may be useful to illustrate some of these points us-ing a simple physical metaphor in which the bases of amultiply determined outcome are known. Briefly, let 8represent the heater and 9 the temperature in a house.In the context of psychophysiology, the heater parallels aneural mechanism and the temperature represents the cog-nitive manifestation of the operation of this mechanism.Although the heater and the temperature are conceptuallydistinct, the operation of the heater represents a physicalbasis for the temperature in the house. Thus, 9 = f(8).A bottom-up approach (i.e. P(9 |8)) makes clear certaindetails about the relationship between 9 and 8, whereas atop-down approach (i.e. P(8 |9)) clarifies others. For in-stance, when the activity of the heater is manipulated (8

PSYCHOPHYSIOLOGICAL SCIENCE 15

is stimulated or lesioned), a change in the temperature inthe house (9) results. This represents a bottom-up ap-proach to investigating the physical substrates of cognitivephenomena. That manipulating the activity of the heaterproduces a change in the temperature in the house can beexpressed as P(9 |8) > 0. Note that the P(9 |8) neednot equal 1 for 8 to be a physical substrate of 9. This isbecause, in our illustration, there are other physical mech-anisms that can affect the temperature in the house (9),such as the outside temperature (81) and the amount ofdirect sunlight in the house (82). That is, there is a lackof complete isomorphism specifiable (at least initially) be-tween the regulated variable (9) and a physical basis (8).

In any given context, the temperature in the house maybe influenced by any or all of these physical mechanisms.If the outside temperature or the amount of direct sunlighthappens to vary when the heater is activated, then the tem-perature may not covary perfectly with the activation ofthe heater (i.e., P(9 |8) < 1) even though the temperatureis, at least in part, a function of the operation of the heater(i.e., P(9 |8) > 0). If the outside temperature and amountof direct sunlight are constant or are perfectly correlatedwith the activation of the heater, then the temperature inthe house and the activity of the heater may covary per-fectly (P(9 |8) = 1). In the context of psychophysiology,this is analogous to a brain lesion study accounting forsome of the variance (P(9 |8) > 0) or all of the variance(P(9 |8) = 1) in the cognitive measure used in the study.The latter result does not imply that the lesioned brain re-gion is a necessary component, just as temperature in thehouse covarying perfectly with heater activation does notmean that no other physical mechanisms could also influ-ence the temperature. Thus, as long as P(9 |8) > 0, 8can be considered a predictor (or component) of 9; thatP(9 |8) = 1 does not imply that 8 is the only (or a nec-essary) cause of 9.

Also evident in this metaphor are the asymmetry be-tween P(9 |8) and P(8 |9) and the interpretive problemsthat may result when simply assuming P(9 |8) = P(8 |9).As outlined previously, the former term represents varia-tions in temperature in the house given variations in theactivity of the heater, whereas P(8 |9) represents the ac-tivity of the heater given variations in the temperature inthe house. Although one would expect to find P(8 |9) >0 in some contexts, the fact that the temperature in thehouse increases reliably when the heater is activated doesnot necessarily imply that changes in the temperature inthe house will be associated with variations in the activityof the heater. In the winter months, changes in the temper-ature in the house may be associated with correspondingchanges in the activity of the heater. However, in anothercontext (e.g., the summer months), the heater may be uni-formly inactive yet housing temperature continues to varyowing to the operation of other physical factors (e.g., out-side temperature 81 or exposure 82 to direct sunlight).

Thus, the finding that P(8 |9) = 0 means not that 8 hasno role in 9 but only that 8 has no role in 9 in that con-text. In the context of brain imaging studies, areas thatare not found to become active as a function of a cognitiveoperation may nevertheless be part of a physical substratefor that cognitive operation ( just as a heater may remaina part of the physical mechanism for the temperature in ahouse).

The preceding example illustrates why one would notwant to exclude a brain area as potentially relevant to acognitive operation based solely on the area not being il-luminated in a brain image as a function of the cognitiveoperation. The converse also holds; that is, a brain areathat is illuminated as a function of a cognitive operationmay or may not contribute meaningfully to the produc-tion of the cognitive operation. Consider a light emittingdiode (LED) on a thermostat (which we will call 8′ ) thatilluminates when the heater (8) is operating. In this case,P(8 |9) = P(8 ′ |9) > 0. That is, the LED represents aphysical element that would show the same covariationwith the temperature in the house as would the operationof the heater – as long as a top-down approach were used.Were the complementary bottom-up approach to be used,it would become obvious that disconnecting (lesioning) theheater has effects on the temperature in the house whereasdisconnecting (or directly activating) the LED has none.

The metaphor also illustrates the “categorical” error:the intuitively appealing notion that the organization ofcognitive phenomena maps in a one-to-one fashion intothe organization of the underlying neural substrates. Thetemperature of the house, for instance, does not map intoa single “temperature center” in the house; rather, it isdetermined by several different physical mechanisms. Onemight argue that the problem is in how 9 is being de-fined, that the temperature in the house is not the mostuseful way of conceptualizing the phenomenon. One so-lution may be to specify the heater-related contributionto temperature (reconceptualize the function dimension, 9as 9 ′ ), but this is not the only or even the best solu-tion. Suppose, for instance, that the system is designedto maintain the temperature of the house within a nar-row range. The problem in such a system may not be inthe definition of 9 but in the full delineation of its physi-cal basis (8,81,82, and interactions) in order to achievea closer approximation between functional concepts andmodels of mechanisms. Although we anticipate that someone-to-one mappings between 8 and 9 may ultimatelybe achieved, reaching this ultimate aim requires a recogni-tion of the preliminary state of our knowledge at presentand the attendant implications for strong inference. Giventhe complementary nature of the data from brain imagingand from direct stimulation and lesion studies, progressin psychophysiology should be fostered by an integrationrather than a progressive segregation of these approachesand literatures.

16 CACIOPPO, TASSINARY, & BERNTSON

Four Categories of PsychophysiologicalRelationships

As illustrated in the preceding section, relations betweenelements in the psychological and physiological domainscannot be assumed to hold across situations or individuals.Indeed, elements in the psychological domain are delim-ited in the subtractive method in part by holding constantother processes that might differentiate the comparisontasks. Such a procedure is not unique to psychophysiologyor to the subtractive method, since many psychological andmedical tests involve constructing specific assessment con-texts in order to achieve interpretable results. The interpre-tation of a blood glucose test, for instance, can rest on theassumption that the individual fasted prior to the onset ofthe test. Only under this circumstance can the amount ofglucose measured in the blood across time be used to indexthe body’s ability to regulate the level of blood sugar (Guy-ton 1971). Here, the relationship between the physiologicaldata and theoretical construct is said to have a limitedrange of validity because the relationship is clear only incertain well-prescribed assessment contexts (Cacioppo &Petty 1986; Donchin 1982). The notion of limited rangesof validity thus raises the possibility that a wide range ofcomplex relationships between psychological and physio-logical phenomena might be specifiable in simpler, moreinterpretable forms within specific assessment contexts.

In order to clarify these issues, it is useful to concep-tualize psychophysiological relationships in terms of a 2(one-to-one vs. many-to-one) by 2 (situation-specific vs.cross-situational) taxonomy. The specific families (i.e. cat-egories) of psychophysiological relationships that can be

Figure 3. Taxonomy of psychophysiological relationships.

derived from this taxonomy are depicted in Figure 3. Thecriterial attributes for, and theoretical utility in, establishingeach of these categories are specified in the two dimensionsillustrated in Figure 3; causal attributes of the relationships,and whether the relationships are naturally occurring orartificially induced, constitute yet other (orthogonal) di-mensions and are explicitly excluded here for didacticpurposes. For instance, the category in Figure 3 labeled“psychophysiological concomitants” refers only to the con-ditions and implications of covariation and is not intendedto discriminate between instances in which the psychologi-cal factor is causal in the physiological response, vice versa,or a third variable causes both. In the sections that fol-low, each type of psychophysiological relationship and thenature of the inferences that each suggests are outlined.

PSYCHOPHYSIOLOGICAL OUTCOMES

In the idealized case, an outcome is defined as a many-to-one, situation-specific (context-dependent) relationshipbetween 9 and 8. Establishing that a physiological re-sponse (i.e., an element in 8) varies as a function of apsychological change (i.e., an element in 9) means thatone is dealing at the very least with an outcome relation-ship between these elements. Note that this is often thefirst attribute of a psychophysiological relationship that isestablished in laboratory practice. Whether the physiolog-ical response follows changes in the psychological eventacross situations (i.e., is context-independent) or whetherthe response profile follows only changes in the event (i.e.,is isomorphic) is not typically addressed initially. Hence, agiven psychophysiological relationship may appear to be anoutcome but subsequently be identified as being a markeras the question of isomorphy is examined; a relationshipthat appears to be an outcome may subsequently be reclas-sified as being a concomitant once the range of validity isexamined; and a relationship that appears to be a marker(or concomitant) may emerge as an invariant upon study-ing the generalizability (or isomorphy) of the relationship.However, this progression is not problematic in terms ofcausing erroneous inferences; as we shall see, any logicalinference based on the assumption that one is dealing withan outcome relationship holds as well for marker, con-comitant, or invariant psychophysiological relationships.

Despite serving as the most elemental psychophysiolog-ical relationship, the outcome can provide the basis forstrong inferences. Specifically, if two psychological modelsdiffer in predictions regarding one or more physiologi-cal outcomes, then the logic of the experimental designallows theoretical inferences to be drawn based on psy-chophysiological outcomes alone. That is, a psychophysio-logical outcome enables systematic inferences to be drawnabout psychological constructs and relationships based onhypothetico-deductive logic. Of course, no single opera-tionalization of the constructs in a crucial experiment is

PSYCHOPHYSIOLOGICAL SCIENCE 17

likely to convince the adherents of both theories. If multi-ple operationalizations of the theoretical constructs resultin the same physiological outcome, however, then strongtheoretical inferences can be justified.

The identification of a physiological response profilethat differentiates the psychological element of interest issufficient to infer the absence of one or more psychologicalelements, but it does not provide logical grounds to inferanything about the presence of a psychological element.Hence, the identification of psychophysiological outcomescan be valuable in disproving theoretical predictions, butthey are problematic as indices of elements in the psycho-logical domain. This caveat is often noted in discussionsof the scientific method and is perhaps equally often vio-lated in scientific practice (Platt 1964). Skin conductance,for instance, has been a major dependent measure in psy-chological research because emotional arousal is thoughtto lead to increased skin conductance (Prokasy & Raskin1973). Similarly, EMG (electromyographic) activity overthe forehead region has been a frequent target measurein relaxation biofeedback because tension has been foundto increase EMG activity over this region (Stroebel &Glueck 1978). Yet as noted in the previous section, sim-ply knowing that manipulating a particular element in thepsychological domain leads to a particular response in thephysiological domain does not logically enable one to inferanything about the former based on observations of thelatter, because one does not know what other antecedentsmight have led to the observed physiological response.Procedures such as holding constant any variations in theelements in the psychological domain that are not of in-terest, measuring these elements in addition to those ofimmediate theoretical interest to determine which elementsare likely to have caused the observed changes in physiolog-ical response, and excluding those physiological responsesbelieved to covary with these irrelevant elements all rep-resent attempts to reduce many-to-one relationships toone-to-one relationships (i.e., going from psychophysiolog-ical outcomes to psychophysiological markers; see Figure3). Such procedures clearly strengthen the grounds for psy-chophysiological inference, but they neither assure that allrelevant factors have been identified and controlled norprovide a means for quantifying the extent of other influ-ences on psychophysiological responding.

For example, consider what can be expected if the prob-ability of a physiological element, P(8), is greater than theprobability of the psychological element of interest, P(9).Since this implies that P(9,8) | P(9) > P(9,8) | P(8),it can be seen that P(8 |9) > P(9 |8) and hence that re-search based only on outcome relationships would resultin an overestimation of the presence of the psychologicalelement.

We should emphasize that these probabilities are simplya way of thinking more rigorously about psychophysiolog-ical relations; one still needs to be cognizant that these

relationships and probabilities may vary across situations.Indeed, comparisons of these probabilities across assess-ment contexts can provide a means of determining theindividual or situational specificity of a psychophysiolog-ical relation. Before proceeding to this dimension of thetaxonomy outlined in Figure 3, we elaborate further onpsychophysiological relations within a specific assessmentcontext when viewed within the framework of conditionalprobabilities. In particular, as P(9,8) approaches 1 andP(not-9,8) approaches 0, the element in the physiologi-cal domain can be described as being an ideal marker ofthe element in the psychological domain.

PSYCHOPHYSIOLOGICAL MARKERS

In its idealized form, a psychophysiological markeris defined as a one-to-one, situation-specific (context-dependent) relationship between abstract events 9 and8 (see Figure 3). The psychophysiological marker relationassumes only that the occurrence of one event (usually aphysiological response, parameter of a response, or profileof responses) predicts the occurrence of the other (usuallya psychological event) within a given context. Thus, mark-ers are characterized by limited ranges of validity. Sucha relationship may reflect a natural connection betweenpsychological and physiological elements in a particularmeasurement situation, or it may reflect an artificially in-duced (e.g., classically conditioned) association betweenthese elements. It is important to recognize that mini-mal violations of isomorphism between 9 and 8 withina given assessment context can nevertheless yield a use-ful (although imperfect) marker when viewed in terms ofconditional probabilities.

Markers can vary in their specificity and sensitivity. Themore distinctive the form of the physiological responseand/or the pattern of associated physiological responses,the greater is the likelihood of achieving a one-to-onerelationship between the physiological events and the psy-chological construct – and the wider may be the range ofvalidity of the relationship thereby established. This is be-cause the utility of an element in 8 to index an elementin 9 is generally strengthened by defining the physiologi-cal element so as to minimize its occurrence in the absenceof the element in the psychological domain.

In terms of sensitivity, a psychophysiological markermay simply signal the occurrence or nonoccurrence of apsychological process or event and possess no informa-tion about the temporal or amplitude properties of theevent in a specific assessment context. At the other ex-treme, a psychophysiological marker may be related in aprescribed assessment context to the psychological eventby some well-defined temporal function, so that the mea-sure can be used to delineate the onset and offset of theepisode of interest, and/or it may vary in amplitude andthus reflect the intensity of the psychological event.

18 CACIOPPO, TASSINARY, & BERNTSON

In sum, markers represent a fundamental relationshipbetween elements in the psychological and physiologicaldomains that enables an inference to be drawn about thenature of the former given measurement of the latter. Themajor requirements in establishing a response as a markerare: (1) demonstrating that the presence of the target re-sponse reliably predicts the specific construct of interest; (2)demonstrating that the presence of the target response isinsensitive to (e.g., uncorrelated with) the presence or ab-sence of other constructs; and (3) specifying the boundaryconditions for the validity for the relationship. The term“tracer” can be viewed as synonymous with “marker,” foreach refers to a measure so strictly associated with a partic-ular organismic–environmental condition that its presenceis indicative of the presence of this condition. On the otherhand, the term “indicant” is more generic and includes in-variants, markers, and concomitants, since each allows theprediction of 9 given 8. We turn next to a description ofconcomitants.

PSYCHOPHYSIOLOGICAL CONCOMITANTS

A psychophysiological concomitant (or correlate), inits idealized form, is defined as a many-to-one, cross-situational (context-independent) association between ab-stract events 9 and 8 (see Figure 3). That is, the search forpsychophysiological concomitants assumes there is a cross-situational covariation between specific elements in thepsychological and physiological domains. The assumptionof a psychophysiological concomitant is less restrictive thanthe assumption of invariance in that one-to-one correspon-dence is not required, although the relationship tends tobe more informative given a stronger association betweenelements in the psychological and physiological domains.

Consider, for instance, the observation that pupillary re-sponses vary as a function of individuals’ attitudes towardvisually presented stimuli – an observation followed by theconclusion that pupillary response is a “correlate” of peo-ple’s attitudes (Hess 1965; Metalis & Hess 1982). However,evidence of variation in a target physiological response asa function of a manipulated (or naturally varying) psycho-logical event establishes only an outcome relation, whichis necessary but insufficient for the establishment of a psy-chophysiological concomitant or correlate.

First, the manipulation of the same psychological ele-ment (e.g., attitudes) in another context (e.g., using au-ditory rather than visual stimuli) may alter or eliminatethe covariation between the psychological and physiologi-cal elements if the latter is evoked either by a stimulus thathad been fortuitously or intentionally correlated with thepsychological element in the initial measurement contextor by a noncriterial attribute of the psychological elementthat does not generalize across situations. For instance,the attitude–pupil-size hypothesis has not been supportedusing nonpictorial (e.g., auditory, tactile) stimuli, where

it is possible to control the numerous light-reflex–relatedvariables that can confound studies using pictorial stimuli(Goldwater 1972). It is possible, in several of the stud-ies showing a statistical covariation between attitudes andpupillary response, that the mean luminance of individu-als’ selected fixations varied inversely with their attitudestoward the visual stimulus (Janisee 1977; Janisee & Peavler1974; Woodmansee 1970).

Second, the manipulation of the same psychologicalelement in another situation may alter or eliminate thecovariation between the psychological and physiologicalelements if the latter is evoked not only by variations inthe psychological element but also by variations in oneor more additional factors that are introduced in (or area fundamental constituent of) the new measurement con-text. Tranel, Fowles, and Damasio (1985), for instance,demonstrated that the presentation of familiar faces (e.g.,famous politicians, actors) evoked larger skin conductanceresponses (SCRs) than did the presentation of unfamiliarfaces. This finding, and the procedure and set of stimuliemployed, were subsequently used in a study of patientswith prosopagnosia (an inability to recognize visually thefaces of persons previously known) to demonstrate that thepatients can discriminate autonomically between the famil-iar and unfamiliar faces despite the absence of any aware-ness of this knowledge. Thus, the first study established apsychophysiological relationship in a specific measurementcontext, and the second study capitalized on this rela-tionship. However, to conclude that a psychophysiologicalconcomitant had been established between familiarity andSCRs would mean that the same relationship should holdacross situations and stimuli (i.e., the relationship wouldbe context-independent). Yet ample psychophysiologicalresearch has demonstrated a psychophysiological outcomeopposite to that specified by Tranel et al. (1985) – namely,that novel or unusual stimuli can evoke larger SCRs thanfamiliar stimuli (see e.g. Landis 1930; Lynn 1966; Stern-bach 1966). Hence, it is safe to conclude that the relationbetween stimulus familiarity and skin conductance shouldnot be thought of as a psychophysiological concomitant.5

Unfortunately, evidence of faulty reasoning based onthe premature assumption of a true psychophysiologicalcorrelate (or invariant) is all too easy to find:

I find in going through the literature that the psychogal-vanic reflex has been elicited by the following varieties ofstimuli . . . sensations and perceptions of any sense modal-ity (sight, sounds, taste, etc.), associations (words, thoughts,etc.), mental work or effort, attentive movements or attitudes,imagination and ideas, tickling, painful or nocive stimuli,variations in respiratory movements or rate, suggestion andhypnosis, emotional behavior (fighting, crying, etc.), relatingdreams, college examinations, and so forth. . . . Forty investi-gators hold that it is specific to, or a measure of, emotion ofthe affective qualities; ten others state that it is not necessar-ily of an emotional or affective nature; twelve men hold that

PSYCHOPHYSIOLOGICAL SCIENCE 19

it is somehow to be identified with conation, volition, or at-tention, while five hold very definitely that it is nonvoluntary;twenty-one authorities state that it goes with one or anotherof the mental processes; eight state that it is the concomitantof all sensation and perception; five have called it an indicatorof conflict and suppression; while four others have used it asan index of character, personality, or temperament. (Landis1930, p. 391)

The hindrances to scientific advances, it would seem, stemnot so much from impenetrable psychophysiological rela-tionships as from a failure to recognize the nature of theserelationships and their limitations to induction.

As in the case of a psychophysiological marker, the em-pirical establishment of a psychophysiological concomitantlogically allows an investigator to make a probability state-ment about the absence or presence (if not the timing andmagnitude) of a particular element in the psychologicaldomain when the target physiological element is observed.It is important to emphasize, however, that the estimateof the strength of the covariation used in such inferencesshould not come solely from evidence that manipulated orplanned variations of an element in 9 are associated withcorresponding changes in an element in 8. Measurementsof the physiological response each time the psychologicalelement is manipulated or changes can lead to an over-estimate of the strength of this relationship and hence toerroneous inferences about the psychological element basedon the physiological response. This overestimation occursto the extent that there are changes in the physiological re-sponse not attributable to variations in the psychologicalelement of interest. Hence, except when one is dealing withan invariant relationship, establishing that the manipula-tion of a psychological element leads cross-situationally toa particular physiological response or profile of responses isnot logically sufficient to infer that the physiological eventwill be a strong predictor of the psychological element ofinterest; base-rate information about the occurrence of thephysiological event across situations must also be consid-ered. This is sometimes done in practice by quantifyingthe natural covariation between elements in the psycho-logical and physiological domains and by examining thereplicability of the observed covariation across situations.

PSYCHOPHYSIOLOGICAL INVARIANTS

The idealized invariant relationship refers to an isomor-phic (one-to-one), context-independent (cross-situational)association (see Figure 3). To say that there is an in-variant relationship therefore implies that: (1) a particularelement in 8 is present if and only if a specific elementin 9 is present; (2) the specific element in 9 is presentif and only if the corresponding element in 8 is present;and (3) the relation between 9 and 8 preserves all rel-evant arithmetical (algebraic) operations. Moreover, onlyin the case of invariants does P(9 |8) = P(8 |9) and

P(not-9,8) = P(not-8,9) = 0. This means that the logi-cal error of affirmation of the consequent is not a problemin psychophysiological inferences based on an invariantrelation. Hence, the establishment of an invariant relation-ship between a pair of elements from the psychologicaland the physiological domains provides a strong basisfor psychophysiological inference. Unfortunately, invariantrelationships are often assumed rather than formally es-tablished; as we have argued, such an approach leads toerroneous psychophysiological inferences and vacuous the-oretical advances.

It has been suggested occasionally that the psychophys-iological enterprise is concerned with invariant relation-ships. As we have seen, the search to establish one-to-onepsychophysiological relationships is important. Moreover,as noted by Stevens (1951):

The scientist is usually looking for invariance whether heknows it or not. Whenever he discovers a functional relationbetween two variables his next question follows naturally:under what conditions does it hold? In other words, underwhat transformation is the relation invariant? The quest forinvariant relations is essentially the aspiration toward general-ity, and in psychology, as in physics, the principles that havewide application are those we prize. (p. 20)

It should be emphasized, however, that evidence for in-variance should be gathered rather than assumed, and thatthe utility of psychophysiological analyses does not rest en-tirely with invariant relationships (Cacioppo & Petty 1985;Donchin 1982). Without this recognition, the establishmentof any dissociation between the physiological measure andpsychological element of interest invalidates not only thepurported psychophysiological relationship but also theutility of a psychophysiological analysis. However, as out-lined in the preceding sections of this chapter (and in thechapters to follow), psychophysiology need not be concep-tualized as offering only mappings of context-independent,one-to-one relationships in order to advance our under-standing of human processes and behavior.

In summary, the minimum assumption underlying thepsychophysiological enterprise is that psychological andbehavioral processes unfold as organismic–environmentaltransactions and hence have physiological manifestations,ramifications, or reflections (Cacioppo & Petty 1986). Al-though invariant psychophysiological relationships offerthe greatest generality, physiological concomitants, mark-ers, and outcomes also can provide important and some-times otherwise unattainable information about elementsin the psychological domain. These points hold for theneurosciences, as well. In laboratory practice, the initialstep is often to establish that variations in a psychologicalelement are associated with a physiological change, therebyestablishing that the psychophysiological relationship is, atleast, an outcome. Knowledge that changes in an elementin the psychological domain are associated with changes in

20 CACIOPPO, TASSINARY, & BERNTSON

a physiological response or profile neither assures that theresponse will serve as a marker for the psychological state(since the converse of a statement does not follow logicallyfrom the statement) nor that the response is a concomitantor invariant of the psychological state (since the responsemay occur in only certain situations or individuals, or mayoccur for a large number of reasons besides changes in theparticular psychological state). Nevertheless, both formsof reasoning outlined in this chapter can provide a strongfoundation for psychophysiological inferences about behav-ioral processes.

Conclusion

Psychophysiology is based on the dual assumptions that(i) human perception, thought, emotion, and action areembodied and embedded phenomena and that (ii) theresponses of the corporeal brain and body contain infor-mation that – in an appropriate experimental design –can shed light on human processes. The level of analysisin psychophysiology is not on isolated components of thebody but rather on organismic–environmental transactions,with reference to both physical and sociocultural environ-ments. Psychophysiology is therefore, like anatomy andphysiology, a branch of science organized around bodilysystems whose collective aim is to elucidate the structureand function of the parts of and interrelated systems in thehuman body in transactions with the environment. Likepsychology, however, psychophysiology is concerned witha broader level of inquiry than anatomy and physiologyand can be organized in terms of both a thematic and asystemic focus. For instance, the social and inferential ele-ments as well as the physical elements of psychophysiologyare discussed in the following chapters.

The importance of developing more advanced record-ing procedures to scientific progress in psychophysiology isclear, as previously unobservable phenomena are renderedtangible. However, advanced recording procedures aloneare not sufficient for progress in the field. The theoreticalspecification of a psychophysiological relationship necessar-ily involves reaching into the unknown and hence requiresintellectual invention and systematic efforts to minimizebias and error. Psychological theorizing based on knownphysiological and anatomical facts, exploratory researchand pilot testing, and classic psychometric approachescan contribute in important ways by generating testablehypotheses about a psychophysiological relationship. Itshould be equally clear, however, that the scientific effec-tiveness of psychophysiological analyses does not derivelogically from physiologizing or from the measurement oforganismic rather than (or in addition to) verbal or chrono-metric responses. Its great value stems from the stimulationof interesting hypotheses and from the fact that, when anexperiment agrees with a prediction about orchestrated ac-tions of the organism, a great many alternative hypotheses

may be excluded. The study of physiological mechanismsand techniques can sharpen our thinking and reduce the er-ror of our conceptualizations and measurements. Althoughnecessary and important, one should bear in mind thatsuch studies are means rather than ends in psychophysi-ology. Little is gained, for instance, by simply generatingan increasingly lengthy list of correlates between specificpsychological variables and additional psychophysiologicalmeasures.

A scientific theory is a description of causal interre-lations. Psychophysiological correlations are not causal.Thus, in scientific theories, psychophysiological correla-tions are monstrosities. This does not mean that suchcorrelations have no part in science. They are, in fact, theinstruments by which the psychologist may test his or hertheories (Gardiner, Metcalf, & Beebe-Center 1937, p. 385).

Thus, in order to further theoretical thinking, thischapter has outlined a taxonomy of psychophysiologicalrelations and suggested a scheme for strong inference basedon these relationships. This formulation can help addressthe following questions:

1. How does one select the appropriate variable(s) forstudy?

2. How detailed or refined should be the measurement ofthe selected variables?

3. How can situational and individual variability in psy-chophysiological relationships be integrated into theo-retical thinking about psychophysiological relationships?

4. How can physiological measures be used in a rigorousfashion to index psychological factors?

However, the ultimate value of the proposed way ofthinking about psychophysiological relationships rests onits effectiveness in guiding psychophysiological inferencethrough the channels of judgmental fallacies. As Leonardoda Vinci (c. 1510) noted: “Experience does not ever err, itis only your judgment that errs in promising itself resultswhich are not caused by your experiments.”

NOTES

1. In our discussions here, the psychological domain is coex-tensive with the conceptual domain and the physiologicaldomain is coextensive with the empirical domain (Ca-cioppo & Tassinary 1990).

2. The identity thesis states that there is a physical coun-terpart to every subjective or psychological event (Smart1959). Importantly, the identity thesis does not imply thatthe relationship between physical and subjective events isone-to-one (i.e., invariant). Within the context of psy-chophysiology, for instance, the identity thesis does notnecessarily imply that the physiological representation willbe one-to-one in that: (a) there will be one and only onephysiological mechanism able to produce a given psycho-logical phenomenon; (b) a given psychological event will

PSYCHOPHYSIOLOGICAL SCIENCE 21

be associated with, or reducible to, a single isolated phys-iological response rather than a syndrome or pattern ofresponses; (c) a given relationship between a psychologi-cal event and a physiological response is constant acrosstime, situations, or individuals; (d) every physiological re-sponse has specific psychological significance or meaning;or (e) the organization and representation of psychologicalphenomena at a physiological level will mirror what sub-jectively appears to be elementary or unique psychologicaloperations (e.g., beliefs, memories, images).

3. Both the many-to-many and the null relation may result inrandom scatter plots when measuring the natural covaria-tion between elements in the psychological and physiologi-cal domains. However, these relations can be distinguishedempirically by manipulating the psychological factors andquantifying the change in physiological response (and viceversa). The scatter plot between this psychological factorand physiological response should remain random in thecase of a null relation between them, but not if they arepart of a many-to-many relation.

4. See Cacioppo and Berntson (1992) for a discussion of thedistinction between parallel and convergent multiple deter-minism.

5. Nevertheless, the application of the psychophysiologicaloutcome and assessment context developed by Tranel et al.(1985) in the study of prosopagnosics by Tranel and Dama-sio (1985) illustrates the scientific value of psychophysiolog-ical investigations even when the relationship between ele-ments in the psychological and physiological domains holdsonly within highly circumscribed assessment contexts.

REFERENCES

Ader, R. (1981). Psychoneuroimmunology. New York: AcademicPress.

Ackles, P. K., Jennings, J. R., & Coles, M. G. H. (1985). Ad-vances in Psychophysiology, vol. 1. Greenwich, CT: JAI.

Allport, G. (1947). Scientific models and human morals. Psycho-logical Review, 54, 182–92.

Ax, A. F. (1964a). Editorial. Psychophysiology, 1, 1–3.Ax, A. F. (1964b). Goals and methods of psychophysiology. Psy-

chophysiology, 1, 8–25.Bacon, F. (1620/1855). Novum organum (translated into English

by T. Kitchin). Oxford University Press.Barchas, P. R. (1976). Physiological sociology: Interface of socio-

logical and biological processes. Annual Review of Sociology,299–333.

Barinaga, M. (1997). What makes brain neurons run? Science,276, 196–8.

Basmajian, J. V., & De Luca, C. J. (1985). Muscles Alive: TheirFunctions Revealed by Electromyography, 5th ed. Baltimore:Williams & Wilkins.

Berger, H. (1929). Uber das Elektrenkephalogramm des Men-schen [On the electroencephalogram of man]. Archiv fürPsychiatrie und Nervenkrankheiten, 87, 551–3. [Reprinted inEnglish in Porges & Coles 1976.]

Bloom, F. E., Lazerson, A., & Hofstadter, L. (1985). Brain,Mind, and Behavior. New York: Freeman.

Boorstin, D. J. (1983). The Discoverers: A History of Man’sSearch to Know His World and Himself. London: Dent.

Brazier, M. A. (1959). The historical development of neurophys-iology. In J. Field (Ed.), Handbook of Physiology. Section I:Neurophysiology, vol. I, pp. 1–58. Washington, DC: AmericanPhysiological Society.

Brazier, M. A. (1961). A History of the Electrical Activity of theBrain. London: Pitman.

Brazier, M. A. (1977). Electrical Activity of the Nervous System,4th ed. Baltimore: Williams & Wilkins.

Cacioppo, J. T., & Berntson, G. G. (1992). Social psychologicalcontributions to the decade of the brain: Doctrine of multi-level analysis. American Psychologist, 47, 1019–28.

Cacioppo, J. T., Marshall-Goodell, B., & Dorfman, D. D. (1983).Skeletal muscular patterning: Topographical analysis of theintegrated electromyogram. Psychophysiology, 20, 269–83.

Cacioppo, J. T., & Petty, R. E. (1983). Social Psychophysiology:A Sourcebook. New York: Guilford.

Cacioppo, J. T., & Petty, R. E. (1985). Physiological responsesand advertising effect: Is the cup half full or half empty? Psy-chology and Marketing, 2, 115–26.

Cacioppo, J. T., & Petty, R. E. (1986). Social processes. In M. G.H. Coles, E. Donchin, & S. Porges (Eds.), Psychophysiology:Systems, Processes, and Applications, pp. 646–79. New York:Guilford.

Cacioppo, J. T., Petty, R. E., & Andersen, B. L. (1988). Socialpsychophysiology as a paradigm. In H. L. Wagner (Ed.) SocialPsychophysiology and Emotion: Theory and Clinical Applica-tions, pp. 273–94. Chichester, U.K.: Wiley.

Cacioppo, J. T., & Tassinary, L. G. (1990). Inferring psy-chological significance from physiological signals. AmericanPsychologist, 45, 16–28.

Caldwell, A. B. (1994). Simultaneous multilevel analysis. Ameri-can Psychologist, 49, 144–5.

Caws, P. (1967). Scientific method. In P. Edwards (Ed.), The En-cyclopedia of Philosophy, pp. 339–43. New York: Macmillan.

Cohen, R. M., Semple, W. E., Gross, M., King, A. C., & Nor-dahl, T. E. (1992). Metabolic brain pattern of sustained audi-tory discrimination. Experimental Brain Research, 92, 165–72.

Coles, M. G. H. (1988). Editorial. Psychophysiology, 25, 1–3.Coles, M. G. H. (1989). Modern mind-brain reading: Psy-

chophysiology, physiology, and cognition. Psychophysiology,26, 251–69.

Coles, M. G. H., Donchin, E., & Porges, S. W. (1986). Psy-chophysiology: Systems, Processes, and Applications. NewYork: Guilford.

Coles, M. G. H., Gratton, G., & Gehring, W. J. (1987). Theoryin cognitive psychophysiology. Journal of Psychophysiology, 1,13–16.

Cook, S. W., & Selltiz, C. (1964). A multiple-indicator approachto attitude measurement. Psychological Bulletin, 62, 36–55.

Coombs, C. H., Dawes, R. M., & Tversky, A. (1970). Mathe-matical Psychology: An Elementary Introduction. EnglewoodCliffs, NJ: Prentice-Hall.

Corbetta, M., Miezin, F., Dobmeyer, S., Shulman, G. L., et al.(1990). Attentional modulation of neural processing of shape,color, and velocity in humans. Science, 248, 1556–9.

Corbetta, M., Miezin, F. M., Dobmeyer, S., Shulman, G. L., &Petersen, S. E. (1991). Selective and divided attention duringvisual discriminations of shape, color, and speed: Functional

22 CACIOPPO, TASSINARY, & BERNTSON

anatomy by positron emission tomography. Journal of Neu-roscience, 11, 2383–2402.

Darrow, C. W. (1929). Differences in the physiological reactionsto sensory and ideational stimuli. Psychological Bulletin, 26,185–201.

Darrow, C. W. (1964). Psychophysiology, yesterday, today andtomorrow. Psychophysiology, 1, 4–7.

Darwin, C. (1873).The Expression of the Emotions in Man and An-imals. New York: Appleton. [Original work published in 1872.]

Donchin, E. (1982). The relevance of dissociations and the irrel-evance of dissociationism: A reply to Schwartz and Pritchard.Psychophysiology, 19, 457–63.

Donders, F. C. (1868). Die schnelligkeit psychischer Prozesse.Archive für Anatomie und Psychologie, 657–81.

Drake, S. (1967). Galileo Galilei. In P. Edwards (Ed.), The Ency-clopedia of Philosophy, p. 262. New York: Macmillan.

Eng, H. (1925). Experimental Investigation into the EmotionalLife of the Child Compared with That of the Adult. London:Oxford University Press.

Fere, C. (1888). Notes on changes in electrical resistance underthe effect of sensory stimulation and emotion. Comptes Ren-dus des Seances de la Societe de Biologie, 5, 217–19. [Reprintedin English in Porges & Coles 1976.]

Fernel, J. (1542). De naturali parte medicinae. Paris: Simon deColies. [Cited in Brazier 1959.]

Fowles, D. C. (1975). Clinical Applications of Psychophysiology.New York: Columbia University Press.

Frankenhauser, M. (1983). The sympathetic-adrenal and pitui-tary-adrenal response to challenge: Comparison betweensexes. In T. Dembroski, T. Schmidt, & G. Blumchen (Eds.),Biobehavioral Bases of Coronary Heart Disease, pp. 91–105.Basel: Karger.

Furedy, J. J. (1983). Operational, analogical and genuine def-initions of psychophysiology. International Journal of Psy-chophysiology, 1, 13–19.

Gardiner, H. M., Metcalf, R. C., & Beebe-Center, J. G. (1937).Feeling and Emotion: A History of Theories. New York: Amer-ican Book Company.

Glaser, R., & Kiecolt-Glaser, J. K. (1994). Handbook of HumanStress and Immunity. San Diego: Academic Press.

Goldwater, B. C. (1972). Psychological significance of pupillarymovements. Psychological Bulletin, 77, 340–55.

Gould, S. J. (1985). The Flamingo’s Smile: Reflections in NaturalHistory. New York: Norton.

Greenfield, N. S., & Sternbach, R. A. (1972). Handbook of Psy-chophysiology. New York: Holt, Rinehart & Winston.

Grossman, M., Crino, P., Reivich, M., Stern, M. B., & Hur-tig, H. I. (1992). Attention and sentence processing deficitsin Parkinson’s disease: The role of anterior cingulate cortex.Cerebral Cortex, 2, 513–25.

Grunberg, N. E., & Singer, J. E. (1990). Biochemical measure-ment. In J. T. Cacioppo & L. G. Tassinary (Eds.), Principlesof Psychophysiology, pp. 149–76. Cambridge University Press.

Gutek, B. A. (1978). On the accuracy of retrospective attitudinaldata. Public Opinion Quarterly, 42, 390–401.

Guyton, A. C. (1971). Textbook of Medical Physiology, 4th ed.Philadelphia: W. B. Saunders.

Harrington, A. (1987). Medicine, Mind, and the Double Brain:Study in Nineteenth-Century Thought. Princeton, NJ: Prince-ton University Press.

Harvey, W. (1628/1931). Exercitatio anatomica de motu cordiset sanguinis in animalibus. Frankfurt: Fitzeri. Translated intoEnglish by C. D. Leake. Springfield, IL: Thomas.

Henry, J. P., & Stephens, P. M. (1977). Stress, Health, and theSocial Environment. New York: Springer-Verlag.

Hess, E. H. (1965). Attitude and pupil size. Scientific American,212, 46–54.

Hugdahl, K. (1995). Psychophysiology: The Mind–Body Perspec-tive. Cambridge, MA: Harvard University Press.

Jacobson, E. (1930). Electrical measurements of neuromuscularstates during mental activities: III. Visual imagination and rec-ollection. American Journal of Physiology, 95, 694–702.

James, W. (1884). What is an emotion? Mind, 9, 188–205.Janisee, M. P. (1977). Pupillometry: The Psychology of the Pupil-

lary Response. Washington, DC: Hemisphere.Janisee, M. P. & Peavler, W. S. (1974). Pupillary research today:

Emotion in the eye. Psychology Today, 7, 60–3.Jaynes, J. (1973). The problem of animate motion in the seven-

teenth century. In M. Henle, J. Jaynes, & J. J. Sullivan (Eds.),Historical Conceptions of Psychology, pp. 166–79. New York:Springer.

Jermott, J. B. III, & Locke, S. E. (1984). Psychosocial factors, im-munologic mediation, and human susceptibility to infectiousdiseases: How much do we know? Psychological Bulletin, 95,78–108.

Kipnis, D. (1997). Ghosts, taxonomies, and social psychology.American Psychologist, 52, 205–11.

Landis, C. (1930). Psychology and the psychogalvanic reflex. Psy-chological Review, 37, 381–98.

Lindsley, D. B. (1951). Emotion. In S. S. Stevens (Ed.), Hand-book of Experimental Psychology, pp. 473–516. New York:Wiley.

Lynn, R. (1966). Attention, Arousal, and the Orientation Reac-tion. Oxford, U.K.: Pergamon.

Mason, J. W. (1972). Organization of psychoendocrine mech-anisms: A review and reconsideration of research. In N. S.Greenfield & R. A. Sternbach (Eds.), Handbook of Psy-chophysiology, pp. 3–124. New York: Holt, Rinehart &Winston.

Meaney, M. J., Bhatnagar, S., Larocque, S., McCormick, C. M.,Shanks, N., Sharma, S., Smythe, J., Viau, V., & Plotsky, P.M. (1996). Early environment and the development of indi-vidual differences in the hypothalamic-pituitary-adrenal stressresponse. In C. R. Pfeffer (Ed.), Severe Stress and Mental Dis-turbance in Children, pp. 85–127. Washington, DC: AmericanPsychiatric Press.

Mecacci, L. (1979). Brain and History: The Relationship be-tween Neurophysiology and Psychology in Soviet Research.New York: Brunner/Mazel.

Mesulam, M. M. (1990). Large-scale neurocognitive networksand distributed processing for attention, language, and mem-ory. Annals of Neurology, 28, 597–613.

Metalis, S. A., & Hess, E. H. (1982). Pupillary response/semanticdifferential scale relationships. Journal of Research in Person-ality, 16, 201–16.

Metherate, R., & Weinberger, N. M. (1990). Cholinergic modula-tion of responses to single tones reduces tone-specific receptivefield alterations in cat auditory cortex. Synapse, 6, 133–45.

Mielke, R., Pietrzyk, U., Jacobs, A., Fink, G. R., Ichimiya, A.,Kessler, J., Herholz, K., & Heiss, W. D. (1994). HMPAO SPET

PSYCHOPHYSIOLOGICAL SCIENCE 23

and FDG PET in Alzheimer’s disease and vascular dementia:Comparison of perfusion and metabolic pattern. EuropeanJournal of Nuclear Medicine, 21, 1052–60.

Miller, J. (1978). The Body in Question. New York: RandomHouse.

Mosso, A. (1896). Fear (translated by E. Lough & F. Riesow).New York: Longrans, Green.

Pardo, J. V., Fox, P. T., & Raichle, M. E. (1991). Localization ofa human system for sustained attention by positron emissiontomography. Nature, 349, 61–4.

Peterson, F., & Jung, C. G. (1907). Psychophysical investigationswith the galvanometer and pneumograph in normal and in-sane individuals. Brain, 30, 153–218.

Platt, J. R. (1964). Strong inference. Science, 146, 347–53.Popper, K. R. (1959/1968). The Logic of Scientific Discovery.

New York: Harper & Row.Porges, S. W., & Coles, M. G. H. (Eds.) (1976). Psychophysiol-

ogy. Stroudsburg, PA: Dowden, Hutchinson & Ross.Posner, M. I., & Petersen, S. E. (1990). The attention system of

the human brain. Annual Review of Neuroscience, 13, 25–42.Posner, M. I., Petersen, S. E., Fox, P. T., & Raichle, M. E. (1988).

Localization of cognitive operations in the human brain. Sci-ence, 240, 1627–31.

Prokasy, W. F., & Raskin, D. C. (1973). Electrodermal Activityin Psychological Research. New York: Academic Press.

Robbins, T. W., & Everitt, B. J. (1987). Comparative functions ofthe central noradrenergic, dopaminergic and cholinergic sys-tems. Neuropharmacology, 26, 893–901.

Sanders, A. F. (1980). Stage analysis of reaction processes. In G.E. Stelmach & J. Requin (Eds.), Tutorials in Motor Behavior.Amsterdam: North-Holland.

Sarter, M. F. (1994). Neuronal mechanisms of the attentionaldysfunctions in senile dementia and schizophrenia: Two sidesof the same coin? Psychopharmacology, 114, 539–50.

Sarter, M., Berntson, G. G., & Cacioppo, J. T. (1996a). Brainimaging and cognitive neuroscience: Towards strong inferencein attributing function to structure. American Psychologist,51, 13–21.

Sarter, M., & Bruno, J. P. (1994). Cognitive functions of corti-cal ACh: Lessons from studies on trans-synaptic modulationof activated efflux. Trends in Neurosciences, 17, 217–21.

Sarter, M., Bruno, J. P., Givens, B. S., Moore, H., McGaughy,J., & McMahon, K. (1996b). Neuronal mechanisms of drugeffects on cognition: Cognitive activity as a necessary inter-vening variable. Cognitive Brain Research 3, 329–43.

Sechenov, I. M. (1878). Elements of Thought. Reprinted in R. J.Herrnstein & E. G. Boring (Eds.) (1965), A Source Book inthe History of Psychology. Cambridge, MA: Harvard Univer-sity Press.

Smart, J. J. C. (1959). Sensations and brain processes. Philosoph-ical Review, 68, 141–56.

Stern, J. A. (1964). Toward a definition of psychophysiology.Psychophysiology, 1, 90–1.

Stern, R. M., Ray, W. J., & Davis, C. M. (1980). Psychophysio-logical Recording. New York: Oxford University Press.

Sternbach, R. A. (1966). Principles of Psychophysiology. NewYork: Academic Press.

Stevens, S. S. (1951). Handbook of Experimental Psychology.New York: Wiley.

Stroebel, C. F., & Glueck, B. C. (1978). Passive meditation:Subjective, clinical, and electrographic comparison with bio-feedback. In G. E. Schwartz & D. Shapiro (Eds.), Con-sciousness and Self-Regulation, vol. 2, pp. 401–28. New York:Plenum.

Tarchanoff, J. (1890). Galvanic phenomena in the human skinduring stimulation of the sensory organs and during vari-ous forms of mental activity. Pflügers Archiv für die gesammtePhysiologie des Menschen und der Tiere, 46, 46–55. [Reprintedin English in Porges & Coles 1976.]

Tootell, R. B. H., et al. (1995). Visual motion after effect inhuman cortical area MT revealed by functional magneticresonance imaging. Nature, 375, 139.

Tranel, D., & Damasio, A. R. (1985). Knowledge without aware-ness: An autonomic index of facial recognition by prosopag-nosics. Science, 228, 1453–4.

Tranel, D., Fowles, D. C., & Damasio, A. R. (1985). Electro-dermal discrimination of familiar and unfamiliar faces: Amethodology. Psychophysiology, 22, 403–8.

Turner, C. F., & Martin, E. (1984). Surveying Subjective Phe-nomena. New York: Russell Sage Foundation.

Vesalius, A. (1543/1947). De humani corporis fabrica. Basle:Oporinus. Translated into English by J. B. de C. M. Saunders& C. D. O’Malley. New York: Schuman.

Vigoroux, R. (1879). Sur le role de la resistance electrique destissues dans l’electro-diagnostic. Comptes Rendes Societe deBiologie, 31, 336–9. [Cited in Brazier 1959.]

Waid,W. M. (1984). Sociophysiology. New York: Springer-Verlag.Weiskrantz, L. (1986). Blindsight: A Case History and Implica-

tions. Oxford University Press.Wenger, M. A. (1941). The measurement of individual differences

in autonomic balance. Psychosomatic Medicine, 3, 427–34.Wilder, J. (1931). The “law of initial values,” a neglected bi-

ological law and its significance for research and practice.Zeitschrift für die gesammte Neurologie und Psychiatrie, 137,317–24. [Reprinted in English in Porges & Coles 1976.]

Woodmansee, J. J. (1970). The pupil response as a measure ofsocial attitudes. In G. F. Summers (Ed.), Attitude Measure-ment. Chicago: Rand McNally.

Woodworth, R. S., & Schlosberg, H. (1954). Experimental Psy-chology. New York: Holt, Rinehart & Winston.

Wu, C. H. (1984). Electric fish and the discovery of animal elec-tricity. American Scientist, 72, 598–607.