A Suggested Principle of Complementarityfor Psychology

In Theory, Not Method

Joseph F. Rychlak

A distinction between the Copenhagen Interpretation andcomplementarity is underscored, with the former basedon methodological effects (i. e., disturbance) and the latteron nonmechanical theoretical understandings of empiricalfindings. Psychology cannot have a Copenhagen Interpre-tation per se because it lacks clear experimental datacomparable to the wave-versus-particlefindings on the na-ture of light. Psychology's problems do not stem frommethod but from the uncertain and confusing grounds thatpredicate its theories. Physikos, Bios, Socius, and Logosare presented as complementary grounds for theoreticalexplanation. It is advised that a theorist try to stay withinjust one ground at a time, although he or she can thenshift to a complementary ground to explain the same targetwithout being inconsistent.

For some time now I have observed colleagues in psy-chology discussing the possibility of our profession'sadopting a principle of complementarity (for recent

examples see Hyland & Kirsch, 1988; Kirsch & Hyland,1987;Snyder, 1988, 1989). The reasons given for wantingto adopt complementarity stem primarily from the verycomplexity of our subject matter—human beings—aswell as the inability we psychologists have to agree on acommon ground on which to base our explanations. I donot think that any psychologist would deny that we havesome rather deep schisms in our approaches to the studyof human experience. But how would a complementarityprinciple help resolve or at least promote tolerance forthe differences that currently exist? To complement is tofill out or make up for what is lacking in a theoreticalexplanation of some targeted item of interest.

It was Niels Bohr who first proposed a complemen-tarity principle for physics, and his model is invokedwhenever this topic is discussed by psychologists. Some-times the phrase Copenhagen Interpretation is used inreferring to this model. In this article I review the evo-lution of the complementarity concept in Bohr's writings,show where it is said to function in the theoretical expla-nations of a physicist, and then present an alternativemodel of complementarity for psychology—one that Ibelieve would both sharpen our theorizing and promotetolerance for conflicting goals in what we want to theorizeabout.

It is not unusual to hear or read that historians ofscience "practically equate the Copenhagen Interpretationand complementarity" (Zukav, 1979, p. 38). ApparentlyNiels Bohr would not have agreed with this equation eventhough he probably contributed to the confounding ofthe two ideas (see Folse, 1985; Moore, 1966; Murdoch,1987). The phrase Copenhagen Interpretation does notappear in Bohr's writings. Strictly speaking, the Copen-hagen Interpretation is restricted to a supposed meth-odological limitation—namely, disturbance (Folse, 1985).The disturbance thesis holds that the reason subatomicphysicists cannot simultaneously measure both the mo-mentum and the position of an electron precisely (i.e.,indeterminacy) is because the instruments used in theprecise measurement of one of these characteristics nec-essarily disturb the other characteristic.

When Bohr first presented the complementarityconcept in 1927 at a Congress in Como, Italy, he didindeed speak of the physicist's observation as disturbingthe atomic system under study. But if indeterminacy weresimply a matter of aligning the proper measuring instru-ment with the object under observation, classical me-chanical theory would still be valid at the subatomic level.Measuring devices would now be intruding on and dis-rupting a mechanical process that could, in principle, betargeted precisely with some alternative, nonintrusiveform of measurement. The problem would be strictlymethodological, although we could not say what thisnonintrusive form of measurement would be like. Butwhat Bohr wanted to convey by his principle of comple-mentarity was that at the subatomic level, physical natureper se can no longer be conceptualized in such Newtonian,mechanistic terms. At the 5th Solvay Conference, heldin Brussels later in 1927, Bohr and Einstein had theirfamous exchange concerning the nature of complemen-tarity. The phrase Copenhagen Interpretation devolvesfrom this congress (Zukav, 1979).

Bohr (1934) was always careful to discuss comple-mentarity in terms of the findings of experimental designs,

Frederick A. King served as action editor for this article.An earlier version of this article was presented to Division 1 on

August 18, 1991, at the 99th Annual Convention of the American Psy-chological Association, San Francisco, CA.

Correspondence concerning this article should be addressed toJoseph F. Rychlak, Department of Psychology, Loyola University ofChicago, 6525 North Sheridan Road, Chicago, IL 60626.

September 1993 • American PsychologistCopyright 1993 by the American Psychological Association, Inc. 0003-066X/93/$2.00Vol. 48, No. 9, 933-942

933

not in terms of the theoretical formulations precedingsuch experiments. This is a crucial point. Bohr and Ein-stein postulated several thought experiments in their ex-change; and because there was this focus on quasi-em-pirical matters, it is easy to see how colleagues could viewthe Copenhagen Interpretation of Bohr's complementar-ity principle as involving instrumental intrusions on thedata rather than conceptual explanations of the data. Butby 1935 Bohr had distanced himself from the disturbancethesis entirely, specifying that complementarity wassomething other than a problem due to the use of grossmeasuring devices to capture minute characteristics ofphysical reality (Folse, 1985). He began at about this timeto speak of a "phenomenal" (p. 156) object of observation.Any one phenomenon could have two complementaryaspects in the measurement of its characteristics—as inthe wave or particle aspects of light, the momentum orposition of the electron, and so on. Hence, in its finalformulation, complementarity was understood as a widerframe of reference within which to conceptualize empir-ical findings (laws of nature, etc.).

Bohr was not endorsing phenomenalism as a theo-retical outlook per se (Folse, 1985). Unlike ImmanuelKant (1781/1952), who proposed a phenomenal theoryin which the mind gave form to all (noumenal) experienceat the outset, Bohr's phenomenalism came into play onlyafter an empirical fact had been documented. It was notthe predicating assumptions giving life to this fact thatconcerned Bohr but rather the remarkable discovery thatmore than one understanding of certain physical factswas not only possible, it was necessary. As Folse observed,"Bohr's profession as a physicist probably blinded himto the possibility that the entire atomic domain which heand his colleagues were attempting to describe by for-mulating quantum theory was but a construct of theirtheories" (p. 229). The main thrust of complementarityis thus to replace the singularity of explanation in classicalmechanics, not to account for the assumptive differencesmade by physicists to guide their work. I would like totake up this point in conjunction with the question: Canpsychology embrace a Bohrian complementarity?

Theory Versus Method inComplementarityA distinction between theory and method strikes me asfundamental to an understanding of complementarity. Iwould define a theory as a "series of two or more con-structions (abstractions, concepts, items, images, etc.)which have been hypothesized, assumed, or even factuallydemonstrated to bear a certain relationship, one to theother" (Rychlak, 1988, p. 520). A method, on the otherhand, is the "means or manner of determining whethera theoretical construct or proposition is [held to be] trueor false. . . . Methods are vehicles for the exercise ofevidence" (pp. 515-516). There are two broad forms ofevidence: (a) procedural evidence, which is based on acoherence theory of truth drawing on plausibility, internalconsistency, tautology, face validity, and so on, and (b)validating evidence, which is based on a correspondence

theory of truth drawing on an empirical test in which asuccession of events is arranged in light of a hypothesis,and a prediction is made to a criterion.

Bohr's challenge was to account for the unexpectedregularities that had shown up in the context of experi-mentation, where validation was being sought. The plau-sibilities of procedural evidence, which enter into math-ematical ratiocinations as well as the design of empirical(and thought) experiments, did not readily accommodatethe strange events being observed at the subatomic levelof experimentation. These findings were valid; they werepredictably reproducible. But they did not make plausiblesense. Frank (1957, chaps. 1 and 2) described this failureof plausibility at the subatomic level as a rupture in thechain that had bound scientific to philosophical under-standing.

Turning now to psychology, it is clear that we lacka valid experimental finding that is so inexplicable as torequire a complementary account of the paradoxical factsafter they have been accrued—as in the wave-particlenature of light. This suggests to me that we cannot havea Bohrian complementarity in psychology. I could try tomanufacture such a paradox in certain findings withinpsychology. For example, I might assert that the remark-able role played by awareness in both classical and operantconditioning reflects such a paradox (see Brewer, 1974).We may claim that people are being shaped by our con-ditioning devices. But when we take the time to talk withour subjects, to view things from their perspective in theexperimental apparatus, we find that conditioning is veryunlikely to occur unless the subject involved has a graspof what is taking place in the sequential apparatus and iswilling to go along with what is being suggested there. Ihave called this a contrasting of the introspective (i. e.,first person or subject's) point of view with the extra-spective (i. e., third person or experimenter's) point ofview concerning what is going on in the experimentaldesign (Rychlak, 1988). From the experimenter's per-spective, subjects are being shaped, and from the subject'sperspective, they are cooperating with what is under sug-gestion (demand) in the experimental design.

Another reflection of such differences in perspectiveis the work of Jones and Nisbett (1971). These investi-gators found that when actors in situations give us theirintrospective understanding of what is taking place, theyinvariably claim that the factors in the circumstances fac-ing them are what causes them to behave as they do. Onthe other hand, observers looking at such actors (whichgives us the extraspective outlook) attribute the cause ofthe actors' behavior to stable personality traits or habitualdispositions to behave in a certain way regardless of cir-cumstances.

Are not such different understandings of the empir-ical data direct parallels to the paradoxes that Bohr hadto deal with in subatomic events? Could we not suggestthat a principle of complementarity is required here topresent the complete action taking place in such exper-imentation? I think not. The problem is that we lack thetension of plausibility being stretched in the findings of

934 September 1993 • American Psychologist

such experiments. A mechanistic explanation of suchfindings is easily achieved (e.g., see Dulany, 1974, on con-ditioning, or Jones & Nisbett, 1971, on their findings).Alternatively, I could readily explain these findings in anonmechanistic or teleological manner (Rychlak, 1988).There is no sense here of a methodological finding—an"after the experimental fact"—demanding that both me-chanist and teleologist accept complementing explana-tions of the facts.

What is required for a Bohrian complementarity tooccur is a clearer explanatory difference of the observeddata. To the physicist, when the facts prove light to be awave function, this is taken as a completely different eventthan when the facts prove light to consist of particles.Wave and particle are totally different groundings for thetheoretical explanations made, hence they describe totallydifferent natural events albeit named as the same physicalconcept (i.e., light). The implausibility of diverse ground-ings for the theoretical description of a single concept iswhat calls for the principle of complementarity. We can-not reduce one theory's grounding (wave) to the other's(particle) or vice versa—yet, there the findings stand! Inpsychology, we have no such methodological finding pop-ping up to stretch our plausibility (procedural evidence)beyond its customary limits. Indeed, as my examplesabove suggest, if psychology is to have a principle of com-plementarity, it will have to function before the meth-odologically proven fact occurs, as an aspect of the theo-rizing to be put to test in the first place. Here is where Ithink some steps can be taken to clarify grounds that doindeed complement each other because they are not re-ducible to each other.

Complementarity in Psychology: FourTheoretical GroundingsThanks to the work of people like Thomas Kuhn (1970)and Robert Watson (1967), we have had a growing ap-preciation of the importance that predicating paradigmsor prescriptions play, not only in framing how we interpretthe findings of our experiments in psychology but inframing what we set out to investigate in the first place.Surely no psychologist today would deny that such as-sumptive influences play a role in his or her theorizing.If we survey current theories in psychology, we are struckby the meaning shared by such concepts as schemata,prototypes, constructs, scripts, plans, and so forth. Allsuch concepts act like Kantian preconceptions or groundson the basis of which the person behaves—including thepsychologist who theorizes, experiments, and interpretsthe resulting data. Strict positivism, which held thatmeanings issue exclusively from below, from the pre-formed substrate of reality, is no longer confidently em-braced by the majority of psychologists.

We know full well that when psychologists conductan experiment, they are necessarily making a preliminaryselection as to which grounding they will be using to con-ceptualize their findings. But the grounds that they mayhave considered and rejected are not therefore invalidwithout assuming yet another ground on which to render

such a negative evaluation (Rychlak, 1980). The groundsprovide a broad background within which theories areframed and then in turn put to test, limited by a certainlogical fallacy.

To reason that if my theory is correct, then the ex-perimental data will array as I predict, in XYZ fashion,and then find that the data do indeed array empiricallyin XYZ fashion does not permit us to conclude that mytheory is necessarily correct. We cannot do so becausethe line of reasoning here is akin to affirming the con-sequent of an if-then logical proposition. This is like say-ing, "If a human being, then a mortal" and affirming thatthe body standing before us is mortal (i. e., affirm theconsequent or second major term of the initial proposi-tion). It would be improper to conclude that this mortalis necessarily a human being because many organismsother than humans are mortal—and could be standingbefore us. In like fashion, as we already have seen in ourexamples of the previous section, two theories drawingon different grounding assumptions (e. g., mechanical,teleological) can account for the same XYZ fact pattern.Indeed, more than one theory framed within the samegrounding (e. g., two mechanical theories) could accountfor this pattern. This is not an issue requiring comple-mentarity, of course, because it is an intrinsic limitationof validation and not a paradoxical finding in the use ofthis form of evidence.

But it is this loss of certainty in validation that seemsto have lent fuel to a growing demoralization among cer-tain critics who question the utility of this form of evi-dence (e.g., see Gergen, 1985, p. 272). These critics inferfrom the logical limitations on validation that a psychol-ogist using this method can prove either nothing at all oranything that she or he wants to. It appears that we arein a period of psychology's history when proposals foralternative methods are mushrooming—qualitative, phe-nomenological, hermeneutic, deconstructive, dialogue-based, narrative-based, and so on. When we look moreclosely at these supposed methods, we learn that theyeither rely on some form of validation in any case or areexclusively based on procedural evidence. I have arguedin favor of retaining the rigors of validation even as weliberalize theorizing in psychology (Rychlak, 1981, 1988).Procedural evidence tests the plausibility of our theories,but as a science we must advance to the additional "con-trol and prediction" testing of validation. Anything lessis anathema to a scientific profession.

So, I would not view these suggestions for differentmethods as reflecting complementarity. All sciences workwithin the limitations of validation. This limitation hasnot prevented scientists from curing polio or putting peo-ple on the moon. Actually, psychologists should be de-lighted with the limitations on positivism, for it meansthat psychological conceptualization will always play arole in the furthering of knowledge. The conceptualizer,the human being that we study, is central to the generationof all knowledge. This conceptualizer must necessarilyaffirm some fundamental basis—a grounding—on whichto make his or her case. We know today that our grounds

September 1993 • American Psychologist 935

are not "out there" in the data but "in here" as assumptiveframeworks. This means that the grounds can vary sig-nificantly. They may, indeed, reflect implausibility whencompared with each other.

If we psychologists can admit that we never haveabsolutely necessary grounds on which to base our the-ories, then we will be in a position to see that alternativegroundings may be complementary. I think that our his-tory can establish a handful of grounds that have beenrepeatedly used in psychological theorizing. If this is true,then it should be possible for each psychologist to statewhat grounding basis he or she is fundamentally affirmingin making a formal theoretical statement. No matter howstrongly we may feel that a colleague errs in selecting acertain ground on which to base the theoretical expla-nation under consideration, it is still possible to examinethe internal structure and clarity of this account, its in-structiveness and relevance for wider issues, and its con-sistency with empirical evidence. We will no longer betrying to explain away (subsume, reduce, etc.) one groundby another but rather focusing and sharpening the theo-retical account being advanced in light of the groundingassumption selected at the outset.

A critic might charge that I favor "anything goes"in the grounding of theoretical explanations. This is notmy intention. The complementarity argued for is of thissort: On what meanings are you grounding your theo-retical ideas? Are your grounds relevant to what you setout to explain? Are you sure that you know where yourassumptive grounding is taking you? How clearly do yourtheoretical concepts actually rest on the ground that youhave selected? Are there any other grounds covertlysneaking in to save the theoretical explanation at crucialpoints? How consistent are your grounded theories witha large body of empirical evidence? Note that we havehere a focus on the theorist's basic assumptions, takingthem for what he or she believes them to be. Hopefullypsychologists will then focus their critical efforts on thetheory being advanced in light of its assumptions, ratherthan judge it unfairly in light of assumptions that it doesnot make in the first place—a practice all too commonin our discipline at present.

Grounding Assumptions forPsychology's ComplementarityI believe that there are four grounds in psychology onwhich to base explanations of behavior: Physikos, Bios,Socius, and Logos. These grounds have been wound—singularly or in combination—into all of our theories inpsychology. This is not an exhaustive list. Other groundsmight be fashioned in the future. But for now I believethat psychology's complementarity should rely on thesefour grounds of equal stature, each capable of solitaryapplication to any explanation. Complementarity is calledfor because these grounding assumptions are no morereconcilable than the wave-particle groundings of physics.Once again, to complement involves filling out or makingup for what is lacking in a theoretical explanation of sometarget. As I am now using the term, there is the further

suggestion that the grounds for any one theoretical ex-planation cannot really be mixed because they lend com-pletely different and even paradoxical meanings to thetarget of interest.

The Physikos ground is taken most directly fromphysical science, where an effort has been made to explaininanimate events in terms of energic processes, such asgravitation, constancy, and conservation. By a process Imean a discernible, repeatable course of action on thebasis of which the targets under description are believedto be patterned so that we recognize them for what theyare. A ground cannot be considered a ground unless ithas a distinctive process of this sort. It is the process thatdistinguishes one ground from another. Each processconveys or produces distinctive contents, which are theitems targeted for explanation. In physics or astronomy,for example, a basic process is that of gravitation. Somepresumed attraction and action at a distance holds theplanets in sequentially patterned orbits as they fall thoughspace within force fields. The planets are contents withinthe gravitational process, targeted for explanation by thisprocess. In psychology, when Clark Hull (1937, p. 2) ar-gued that human behavior is merely a complication ofthe same factors that go to make up the actions of a rain-drop, he was basing his explanation on the Physikos. Atthe most fundamental assumptive level, he recognized nodifference between animate and inanimate substances inmotion.

If such a difference is to be drawn, while continuingto base explanations on the physical substance of animateorganisms, we move to the Bios grounding. Here we findbeing advanced processes like genetics, mediation, andorganic systems. In an article entitled "What PsychologyIs About," Donald Hebb (1974) stated flatly that "Psy-chology is a biological science" (p. 72). Given thisgrounding predication, it followed for Hebb that "Freewill has a physiological basis" (p. 75), which means thathuman agency is a content within a biological process.So long as we ignore the animate-inanimate issue, it isdifficult to distinguish a Bios from a Physikos concept,which is probably why Hull found it so easy to equatethe behaviors of raindrops and people.

If we leave such physical complexities and beginthinking of the person in terms of group relations andcultural influences, we arrive at the Socius grounding.Here, the processes that are invoked include socialization,historicism, and even political collectivism. I would sug-gest that when Kenneth Gergen (1989) asked us to "avoidreducing the social world to the psychological" (p. 427)and to stress exogenous cultural factors in describing be-havior rather than endogenous cognitive factors (Gergen,1985), we were witnessing an argument for theoriesgrounded in the Socius.

The final ground I have discerned in psychologicaltheorizing is the Logos, which draws on the patternedorder of experience to explain things according to pro-cesses like predication, personal construing, or mentalacts in the Brentanoan sense. When George Kelly (1955)defined the construing process as involving the framing

936 September 1993 • American Psychologist

of how two items are alike and by the same token differentfrom a third (p. 112), he was relying exclusively on aprocess in the Logos—which brings about cognitive or-ganizations that we call meaning. To mean is to inten-tionally pattern or simply acknowledge extant organiza-tions of experience that suggest a purpose. The Logos isnot to be found in the physical or the social structures ofexperience—as if it were a content within such pro-cesses—but in the ever shifting patterns of meaningfulrelations that the physical and social structures take onin the conceptual processes of intelligence. Intellect andintelligence take Latin roots from perceiving or gatheringrelations among and between events, resulting in an un-derstanding of meaningful patterns, enabling free-willchoices among them, and so forth. The Logos is funda-mental here.

Explain also devolves from the Latin word planare,which means to flatten or make things level. A psycho-logical principle of complementarity will thereforemake it clear that a theoretical explanation must bebrought down (leveled) to any one of four clear (flat)grounds, each of which has equal status. We are notspeaking of four levels of explanation here. Thegroundings are not to be rank ordered. To complementis not to subsume one ground by another. Zukav (1979)observed that the impact of complementarity on phys-ics was, in effect, "that it does not matter what quantummechanics is about! The important thing is that it worksin all possible experimental situations" (p. 37). I wouldlike to paraphrase this statement by saying that if weaccept the four grounds that I have recommended, itwill not matter which of these bases we select to buildour theory on. So long as what we say is instructive andconsistent with the empirical findings relevant to thetheoretical grounding per se, we will be practicingpsychology.

Also, it will not matter if empirical findings sup-porting one grounding assumption fail to address ex-planations in another grounding assumption. The em-pirical findings in the Bios complement empirical find-ings in the Logos: they do not replace them. If findingsin the Physikos cannot be wound into theoretical ac-counts based on the Socius, this is not necessarily astrike against the Socius. Of course, if a social theorymakes claims about what can happen to people's phys-ical functioning in certain physical climates, then ob-viously these claims must meet with the findings of the-ories that have been drawn from and tested in the Phy-sikos and Bios realms. But just because a social theorydoes not reduce its accounts to biophysical forces—making the Socius a content in the Bios or Physikosrather than a ground in its own right—does not meanthat it is invalid or unfinished as an explanation. Suchreductionism is predicated on a rank ordering of thegrounding levels of explanation, with some consideredmore basic than others (note that this ordering is a Lo-gos hierarchy). Psychological complementarity doesaway with such reductive requirements.

Sample Problems in Psychology'sGroundings

It is not difficult finding examples of problems with thegroundings in psychology—the most notorious of whichwould involve the historical mind-body or, as I wouldphrase it, the Logos-Bios problem. I have always consid-ered Fechner (Hall, 1912) to be the real father of psy-chology, because he theorized—albeit with little accep-tance from colleagues—that these two spheres of humanexperience were of equal stature. Helmholtz's influencewas to win out in psychology, thanks to Wundt's appar-ently half-hearted commitment to elementism (see Blu-menthal, 1975), which seeks to reduce the Logos (content)to the Bios (process). This is still the received view inpsychology because it seems so plausible that people havebrains (Bios) and that they are physical beings (Physikos)with electrochemical actions going on inside them. Whowould.be so rash as to ignore such obvious facts in framingpurely psychological explanations? It is hard for the re-ductionist to accept a complementarity in which to avoidBios accounts in favor of Logos accounts is not therebyto deny the validity of Bios accounts.

I think of Skinner (1974) as an example of someonewho did not wish to confound the actuarial measurementsof observed behavior with functions presumably goingon inside the organism. Skinner always admitted thatthere were things to be observed and even theorized aboutinside the black box. But he also realized that he did notobserve these things in his work as a behavior analyst.All he observed was a pattern of action that he couldmeasure as it unfolded in the behavior of organisms. Itis my belief that Skinner was rejecting efforts to theorizeon the basis of the Physikos-Bios in favor of the Logosand Socius. I think he was correct in his stand on thismatter. Psychologists continually observe people behavingin some patterned way, only to present their observationsin terms of theories postulating all manner of unobservedelectrical circuitry—thereby switching instantaneouslyfrom the Logos or Socius to the Physikos-Bios. Alongwith Skinner, I believe that the fundamental basis of theresultant theoretical account rests much more on theformer than the latter grounds. Subtle shiftings like thiscan blind the theorist to significant misrepresentations inthe resulting explanation.

A classic example of such misrepresentation is Dol-lard and Miller's (1950) valiant attempt to bring Freudiantheory—in my opinion best suited to Logos grounding—under the aegis of the Physikos-Bios formulations ofHullian learning theory. As a result, whereas Freud hadrepression pregnant with meaning—with the censoringof meanings—Dollard and Miller interpreted repressionas a lack of meaning-producing mediators (i. e., cue-pro-ducing responses; Dollard & Miller). Whereas Freud hadthe unconscious sphere of cognition even more knowl-edgeable than the conscious sphere, Dollard and Millerdescribed the unconscious as the unlabeled, hence un-known (stupid) aspects of ongoing cognition and behavior(Dollard & Miller). Of course, Freud cannot be praised

September 1993 • American Psychologist 937

for his theoretical clarity on such matters either. Re-sponding to the Newtonian pressures of his time, he for-mulated his notoriously confusing and arbitrary libidotheory, which can now be seen as an effort to expressLogos concerns in Bios fashion (Freud, 1921/1955).

It should not be thought that disagreements amongtheoretical perspectives in psychology always occur acrossgroundings. It is possible to have alternative theoreticalperspectives share a common grounding. When Titchener(1898) drew the structuralist-functionalist distinction, hewas attacking Logos theorists like Brentano, who he felttoo readily succumbed to unscientific teleological de-scriptions. Titchener was drawing a line separatinggroundings, for he wanted all psychological accounts tobe grounded in the Bios. He was answered by Angell(1907), who favored an interpretation of functionalismrooted not in the Logos but in the same Bios that Titch-ener favored. Because these shifting grounds were neverelucidated, the original issue raised by the structuralism-functionalism debate—attacking teleological theory—waslost to history (see Rychlak, 1988). Even so, there was arift between these two theoretical schools sharing a com-mon ground.

A similar theoretical disagreement occurred whenHull (1943) objected to Tolman's (1932/1967) use of pur-pose, substituting the fractional antedating goal responseas an explanation of such anticipatory behaviors. Bothof these theorists were framing explanations at the Bioslevel, although in truth it might be argued that Tolmanpermitted the Logos to ground his accounts at crucialpoints—a fact that Hull seems to have sensed. Neverthe-less, it is still true that theorists can disagree even thoughthey have accepted common groundings for their expla-nations. What complementarity accomplishes here is tokeep all grounding assumptions clearly above board,sharpening the resulting explanation.

A more recent example of the confusion that canarise in grounding explanations is Sherif's (1987) com-plaint that we do not know precisely what sex refers totoday. Why is this true?

Because the "variable" called sex is like a railroad boxcar: Everyoneknows what it is called and what it is used for, but no one knows whatis inside. Older psychologists had no doubt that it contained "biology"[Bios]. Modern psychologists follow suit, or add culture [Socius], orsubtract biology as well. Result? Utter confusion in almost all discussionsof the variable "sex" or of sex differences, (p. 45)

Complementarity would suggest the following approachto resolve this confusion: Before a psychologist undertakesthe study of sex, he or she should have clearly in mindprecisely on what ground the theoretical account will rest.Although a challenging task, try to stay within onegrounding realm in any one theoretical account. Makethis theoretical grounding explicit in the write-up to fol-low, so that as colleagues read the results of an experimenttesting some aspect of sex, they can frame their under-standing accordingly. Colleagues would then be expectedto acknowledge and take seriously the ground under study,rather than dismissing it out of hand and accounting forthe experimental data in terms of their own pet ground.

The focus of criticism should be on how instructive theground initially selected proves to be in the theory, col-lection of data, and analysis that follows from it.

Returning to the mind-body problem a century afterFechner, we find a modern brain researcher like Granit(1977) suggesting that the central nervous system mustbe understood in terms of purpose and that purpose is apoint of view rather than a physical property of the brain.I would suggest that points of view are best understoodas Logos rather than Bios conceptions. The latter con-ceptions need not be mixed among the former in orderto say something about purpose. Of course, the twogrounds are in fact regularly mixed. For example, afterequating the logic of infant behavior with predictable in-put-output relations, T. G. R. Bower (1989) observedthat "The explanation of input-output relations in termsof intervening logical structures has been going on for along time" (p. 151). Unfortunately this is true. But onsuch an account, logical structures are said to arise with-out a Logos process framing them into existence—actingin turn as input mediators in a Bios-Physikos cyberneticprocess. Once again, the Logos is being framed here as acontent in the Bios, even though there is not one scintillaof evidence to suggest that the latter can influence thelogical patterning of the former in any way (see Penfield,1975).

Sometimes a theorist will combine two grounds, onlyto later drop one without significantly altering his or herfundamental position. This represents excellent evidencefor the likelihood that the theorist was really not relyingon the dropped ground in the first place and that an in-depth analysis of the original theory in question wouldhave established this fact. As one recent example of thisphenomenon, Anderson (1983) initially offered the fol-lowing argument to support his theory of cognitive acti-vation:

There are numerous reasons for believing in a spreading activationmechanism. One is that it corresponds to our understandings of neu-rophysiology. The neurons and their connections can be thought of asa network. The rate of firing of a neuron can be thought of as the levelof activation.. . . The general set of "neurophysiological hunches" thatwe possess is probably an important consideration in many people'sacceptance of spreading activation, (p. 86)

In a book published seven years later, Anderson(1990) reversed his position and suggested that such par-alleling of activated networks with brain structures wasmisguided; for the indefinite future, cognitive psycholo-gists should not be pressured into "a premature insistenceon neural fidelity" (p. 13) in their theorizing (for a similarattitude, see Watkins, 1990).

The gist of Anderson's new position is that we arenot certain what the ultimate brain-processing theory isgoing to be (surely modern conflict theories of such pro-cessing do not support network models; see Popper &Eccles, 1977, p. 243), that different arrangements of theneural-like elements produce the same phenomena andvice versa, and that we can carry on our psychological

938 September 1993 • American Psychologist

studies without a one-to-one matching of network modelsto neuronal brain patterns in any case. All pretense ofBios grounding is lost at this point. The implied questionbecomes: If not the Bios, then what ground is in factunderwriting Anderson's network theory of activation? Ithink it would represent a significant theoretical advanceto get a clear answer to this question.

There is probably no area of study in which ground-ings are more debated than that of emotion. Zajonc andMclntosh (1992) have cautioned that one should notcommit to a theory of emotion that claims universal va-lidity because there are so many assumptive biases activein this area of study, such as focusing exclusively on bio-logical or cultural influences. If we followed a principleof complementarity, there would be no need to seek auniversal theory in the first place. The only question wewould address is, How well do theorists in the area ofemotion who rely exclusively on the Bios (e.g., Papez,1937) convey their assumptive bias, as compared withthose who rely exclusively on the Socius (e.g., Lutz, 1986)?And if it were impossible to bring such grounds togetherinto a single theory, then we would appreciate that thereason this is the case is because different groundings lenddifferent meanings to the target in question, meaningsthat cannot ever be welded together. What we should beafter here is the best possible theory within the stricturesof the ground selected.

As I noted earlier, according to complementarity itdoes not matter what the real nature of an emotion is.What matters is the clarity of the explanation in relationto the targeted experimental data. This clarity is not al-ways achieved. As a case in point, in his network theoryexplanation of mood state-dependent learning, G. H.Bower (1981) relied on a Bios ground, which, as we havenoted, meshes nicely with the Physikos. This compati-bility enabled Bower to note that "electrical energy [Phy-sikos] corresponds to activation [Bios]" (p. 134). But healso described the basic unit of thought as the "propo-sition," which presumably has meaningful subject andpredicate terms linked together by associative ties. I wouldargue that predication is best thought of as a process inthe Logos (Rychlak, 1991). Space considerations do notallow me to analyze Bower's proposition concept in thenecessary detail, but I think much of his account drawson the Logos only to be inappropriately recast into theBios-Physikos. Even if I were wrong in this suggestionand could not deliver on this Skinnerian critique, theeffort expended in this regard could only help to clarifythe theory under scrutiny—where, at the very least, wehave the possible confounding of explanatory grounds.

Just as we accept a colleague's operational definitionof some concept preliminary to his or her gathering datain an experiment, so too can we accept in principle thesingle (i. e., nonconfounded) ground selected by a col-league to account theoretically for some targeted area ofinterest. We then look at the resultant analysis and ex-planation in terms of the chosen ground. If we find aconfusion of grounds mixing into the account, this canbe pointed out, showing if possible what the actual or

most fundamental grounding comes down to in the ex-planation. But if there is a clear analysis and explanation,no matter how we might personally prefer to select an-other grounding, it is not acceptable practice on a com-plementarity principle to reject or ignore the account inquestion. At a metatheoretical level we may debate thegrounds to be used generally in psychology and un-abashedly commit to our favorite. But we should not hiteach other over the head with such biases during the ev-eryday efforts we put into building a scientific discipline.It seems to me that this tolerance of a colleague's groundwould be a nice first step toward the unification of psy-chology that is being sought today by many psychologists(see, e.g., Staats, 1991). We unify through an appreciationof the complementary diversity of our groundings.

Questions Raised by PsychologicalComplementarity

I have had several questions raised by colleagues con-cerning the suggested adoption of a psychological com-plementarity. Here are a half dozen typical examples anda brief reply.

1. Won't we always be using all four of these groundsno matter what we actually write up to explain things?

The important point concerning complementarityis not how many grounds we may consider informallybut how the formal write-up is framed. This final standis what we want to analyze precisely and specifically. Iam frankly dubious about how much actual mixing ofgrounds goes on in the core explanation, the basic pointof view. As in the example from Anderson (1983) quotedearlier, most of the mixing probably occurs at ancillarypoints that are not of crucial relevance to the account inquestion. I suggest that we stop worrying about how thegrounds might or might not fit together. Work on oneground; work on four grounds. Sharpen your thinkingaccordingly when you zero in on the theoretical expla-nation of your target, but limit any one write-up to oneground. Our theoretical accounts become too muddledwhen we begin to slip and slide between groundings.

2. Doesn't it follow that the more grounds used, thericher the theoretical account will be?

No, I no longer believe this to be true. Most of thetime, the account becomes more obscure. I think thatthe intent of reductionism, which sought to bring oneground down to another—the former then being under-stood as a content of the latter—was to simplify expla-nation. All sciences aim at simplifying explanations. WhatBohr faced up to was the fact that we can simplify theoriesdown to more than one ground. There is plenty of workremaining, simplifying explanations in terms of just oneground. There will always be some good and bad theo-rizing going on within any one of the groundings. I believethat when a theorist is mixing two or more grounds intothe same account—as levels of explanation, for exam-ple—it will be relatively easy to show which of these as-sumptions is contributing the crucial meaning to the fin-ished product.

September 1993 • American Psychologist 939

3. Don't the facts or experimental results havesomething to say about what ground should be used toaccount for them? Will all four of these groundings standup equally well to an empirical test?

Clearly, when we approach the explanation of a tar-get (i.e., the item, event, etc., under study), the relevanceof one grounding or another is suggested in the very factsthat our observations record. There is always a precedentframework, a predication that lends the meaning to whatwe select as an area of study in the first place. But, as inthe study of the mentally disadvantaged, for example,there is often great plausibility (procedural evidence) inapplying biological conceptions to our explanation of suchclinical syndromes. Similarly, in theorizing about how itis that clinicians as thinkers frame their hypotheses con-cerning syndromes—not to mention any other form ofhypothesis testing—a logical analysis seems plausible.Why not therefore rely on Bios conceptions in the formercontext and Logos conceptions in the latter context? Thesame psychologist could do this without feeling inconsis-tent. This is not traditional eclecticism because we arenot combining grounds into the same account. The pointis that we remain scientists, that we seek validation when-ever called on to advance evidence for our line of theo-rizing. But as facts never truly speak for themselves (recallthe affirming-the-consequent fallacy), there is a vital roleto be played here by purely theoretical analysis, whichalways brings in the matter of grounds.

The related question about whether or not all fourgroundings will stand up equally to empirical testing isdue to a misunderstanding in the role that a ground plays.It is true that grounds may be thought of as highly abstractmeta-theoretical positions, but as such widely encom-passing predications, they are not precisely under empir-ical testing in the data collection of an experiment. Theyframe the theory that predicts and seeks support fromthese data, but the grounds are never under direct test bythe data. Thus, if human behavior is under empiricalstudy, any evidentiary data (i. e., any lawful regularity insuch behavior) would be targeted by all four of ourgroundings at the same time. The affirming-the-conse-quent fallacy involves the theory under a certain ground-ing being put to empirical test. If such a theory continuallylacks support, it will be discarded in time, but the theoristwould usually frame an alternative explanation under thesame grounding. Alternatively, the theorist might look atthe disconfirming data from a theory framed by a differentgrounding. Strictly speaking, data collections are one stepremoved from the grounds involved.

4. Aren't you trying to gain acceptance for "soft"psychology in your complementarity model? Isn't theLogos what soft psychologists want to use in their expla-nations?

Although I suspect that those who are called softpsychologists might prefer to use the Logos to groundtheir explanations, I am put off by this hard-soft labelingof theoretical grounds. As far as I am concerned, the hardor tough-minded approach in psychology is to be foundin the context of method (proof, experimental design,

etc.) and not in the context of theoretical explanation. Ahard scientific method that is objective can entertaintheoretical hypotheses drawing on any of our fourgroundings. The grounds are selected on the basis of pro-cedural evidence, of course, but this is true of all as-sumptive beliefs including those of mathematics. Plau-sibility is what gets us all going. However, if the methoddictates that only Physikos and Bios groundings are suit-able to it—which would itself be a bias drawn from pro-cedural evidence—then although it may be hard, it is thevery opposite of objective. A truly scientific methodshould never dictate the theories put to it. And if it istrue that theories based on all four groundings may besubjected to validation, then the Logos form of theory isno more soft than any other type of explanation. As Inoted earlier, the kind of soft psychology that worries meis the one that rejects validating evidence in favor of testsbased exclusively on procedural evidence. I want to usethe Logos in my theorizing but also want to use our tra-ditional method of validation.

5. What do you have against theoretical integration?Isn't this the aim of the unity of the sciences, to bring allexplanations under one theoretical umbrella?

I have nothing against theoretical integration whenit works. But I have come to the regrettable conclusionthat this aim for theoretical unity, which is no longerclung to with conviction in other sciences, has paradox-ically become a source of disunity in psychology. I wouldargue that the so-called unity achieved in the natural sci-ences occurred because of the compatibility that we haveseen between the Physikos and the Bios, which easilymeld into each other when the question of life (animatedaction) is not at issue. Our fellow scientists are fully cog-nizant of the importance of the Logos and Socius as com-plements to the traditional groundings of the Physikosand Bios. Physicists today refer to themselves as "parti-cipators" (Zukav, 1979, p. 29) in their work. I believethat such recognition as, indeed, the recognition by Bohrof complementarity itself, requires explanation in termsof a Logos realm. What such scientists have to offer asparticipators is their framing assumptions, their pointsof view, and their logical ordering of the material understudy.

The physical scientists are not going to be upset ifsome of us in psychology base our explanations solely ona Logos grounding. The theoretical physicist, DavidBohm (1985), has postulated an "implicate order" (p.14) that functions both in subatomic fields and in gen-erating meanings in human thought. The theoretical bio-chemist, Rupert Sheldrake (1989), has introduced a re-lated notion of "morphic fields" (p. 198) that lend uni-form behavioral patternings to entire species of animals.Such conceptions are consistent with Logos explanations.

6. Isn't it likely that even though we require differentgroundings today, when we learn more about the variousgroundings in the future we should be able to bring allof our theories together on a "basic" grounding? Is thisnot our real goal in psychology?

940 September 1993 • American Psychologist

This "wait until tomorrow" reassurance strikes meas an effort to retain the status quo of reductive expla-nation. I do not accept the view that simply because wemay frame complementary accounts of behavior we arein some kind of preparadigmatic stage to be rectified intime (Kuhn, 1970). I say there is no tomorrow. We neednot apologize for the fact that the subject matter of ourscience is rich enough to invite more than one theoreticalgrounding. I cannot, of course, deny that maybe somedayone grounding will suffice for all that there is in psy-chology. I doubt this very much. But if the goal of psy-chologists is to adopt complementarity as an intermediatestep, until such time as a single grounding can accountfor all there is to be said about human beings, then I saythis is no real complementarity at all. This would be aninsincere strategy carrying no weight.

Conclusion

I believe that a more open and equal atmosphere is likelyto develop in psychology when we all realize that althoughwe are participators in the same endeavor, we are relyingon complementary groundings for the knowledge we seek.One person's explanatory gain is therefore not the other'sloss. We should encourage and take satisfaction in eachother's new insights given the grounds selected. As longas a colleague's chosen grounding matches the theoreticalexplanation of the empirical data to follow, this will beseen as an advance in knowledge—tentatively held, as inall science. I think that a growing tolerance could be theresult. Psychologists have not been noted for their toler-ance of each other's point of view.

I also believe that a clearer role will emerge for thosepsychologists who are not especially drawn to the con-ducting of research but who have skills in the analysis oftheoretical claims. Such psychologists will play an im-portant role in working out the grounding assumptions,the internal consistency of the resulting theory, and thelegitimacy of claiming certain research findings as sup-porting the theory in question. More psychologists mustdevote themselves to such epistemological efforts. I onlyhope that we can keep intact our traditional method ofcontrol-and-prediction to a criterion. This method of val-idation enables hypotheses to be formulated and testedon the basis of all four groundings. It is this methodolog-ical tie alone that binds us to the family of sciences.

Thus, contrary to the Bohrian model, the comple-mentarity I propose enters at the formulative point of ascientific investigation, even before experiments areframed. As psychologists, we know a little about howpeople cognize, and everything we have learned to datesuggests that framing understanding in terms of foundingassumptions does occur in human theorizing. It is timefor psychology to broaden its base by recognizing andclarifying the grounding assumptions that its proponentshave been using for over a century. There is a good chancethat we will enjoy scholarly benefits and promote unityin the profession as well.

REFERENCES

Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA:Harvard University Press.

Anderson, J. R. (1990). The adaptive character of thought. Hillsdale,NJ: Erlbaum.

Angell, J. R. (1907). The province of functional psychology. PsychologicalReview, 2, 61-91.

Blumenthal, A. L. (1975). A reappraisal of Wilhelm Wundt. AmericanPsychologist, 30, 1081-1088.

Bohm, D. (1985). Unfolding meaning: A weekend of dialogue with DavidBohm. New York: Ark Paperbacks.

Bohr, N. (1934). Atomic theory and the description of nature. Cambridge,England: The University Press.

Bower, G. H. (1981). Mood and memory. American Psychologist, 36,129-148.

Bower, T. G. R. (1989). The rational infant: Learning in infancy. NewYork: Freeman.

Brewer, W. F. (1974). There is no convincing evidence for operant orclassical conditioning in adult humans. In W. B. Weimer & D. S.Palermo (Eds.), Cognition and the symbolic processes (pp. 1-42).Hillsdale, NJ: Erlbaum.

Dollard, J., & Miller, N. E. (1950). Personality and psychotherapy: Ananalysis in terms of learning, thinking, and culture. New York:McGraw-Hill.

Dulany, D. E. (1974). On the support of cognitive theory in oppositionto behavior theory: A methodological problem. In W. B. Weimer &D. S. Palermo (Eds.), Cognition and the symbolic processes (pp. 43-56). Hillsdale, NJ: Erlbaum.

Folse, H. J. (1985). The philosophy of Niels Bohr: The framework ofcomplementarity. Amsterdam: North-Holland.

Frank, P. (1957). Philosophy of science: The link between science andphilosophy. Englewood Cliffs, NJ: Prentice-Hall.

Freud, S. (1955). Suggestion and libido. In J. Strachey (Ed.), The standardedition of the complete psychological works ofSigmund Freud (Vol.18, pp. 88-92). London: Hogarth Press. (Original work published1921)

Gergen, K. J. (1985). The social constructionist movement in modernpsychology. American Psychologist, 40, 266-275.

Gergen, K. J. (1989). Social psychology and the wrong revolution. Eu-ropean Journal of Social Psychology, 19, 463-484.

Granit, J. (1977). The purposive brain. Cambridge, MA: MIT Press.Hall, G. S. (1912). Founders of modern psychology. New York: D. Ap-

pleton.Hebb, D. O. (1974). What psychology is about. American Psychologist,

29, 71-79.Hull, C. L. (1937). Mind, mechanism, and adaptive behavior. Psycho-

logical Review, 44, 1-32.Hull, C. L. (1943). Principles of behavior. New York: Appleton-Century-

Crofts.Hyland, M. E., & Kirsch, I. (1988). Methodological complementarity:

With and without reductionism. The Journal of Mind and Behavior,9. 5-12.

Jones, E. E., & Nisbett, R. E. (1971). The actor and the observer: Divergentperceptions of the causes of behavior. Morristown, NJ: GeneralLearning Press.

Kant, I. (1952). The critique of pure reason. In R. M. Hutchins (Ed.),Great books of the Western world (Vol. 42, pp. 1-250). Chicago: En-cyclopedia Britannica. (Original work published 1781)

Kelly, G. A. (1955). The psychology of personal constructs: A theory ofpersonality (Vol. 1). New York: Norton.

Kirsch, I., & Hyland, M. E. (1987). How thoughts affect the body: Ametatheoretical approach. The Journal of Mind and Behavior, 8, 417-434.

Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.).Chicago: University of Chicago Press.

Lutz, C. (1986). Emotion, thought, and estrangement: Emotion as acultural category. Cultural Anthropology, 1, 405-436.

Moore, R. (1966). Niels Bohr: The man, his science, & the world theychanged. New York: Knopf.

Murdoch, D. (1987). Niels Bohr's philosophy of physics. Cambridge,England: Cambridge University Press.

September 1993 • American Psychologist 941

Papez, J. W. (1937). A proposed mechanism of emotion. Archives ofNeurological Psychiatry, 38, 725-743.

Penfield, W. (1975). The mystery of the mind: A critical study of con-sciousness and the human brain. Princeton, NJ: Princeton UniversityPress.

Popper, K. R., & Eccles, J. C. (1977). The self and its brain. London:Springer International.

Rychlak, J. F. (1980). The false promise of falsification. The Journal ofMind and Behavior, 1, 183-195.

Rychlak, J. F. (1981). A philosophy of science for personality theory (2nded.). Malabar, FL: Krieger.

Rychlak, J. F. (1988). The psychology of rigorous humanism (2nd ed.).New York: New York University Press.

Rychlak, J. F. (1991). Artificial intelligence and human reason: A teleo-logical critique. New York: Columbia University Press.

Sheldrake, R. (1989). The presence of the past: Morphic resonance andhabits of nature. New York: Viking Books.

Sherif, C. W. (1987). Bias in psychology. In S. G. Harding (Ed.), Feminismand methodology: Social science issues (pp. 37-56). Bloomington:University of Indiana Press.

Skinner, B. F. (1974). About behaviorism. New York: Knopf.Snyder, D. M. (1988). On complementarity and causal isomorphism.

The Journal of Mind and Behavior, 9, 1-4.Snyder, D. M. (1989). Causal isomorphism: A concept in search of a

meaning; complementarity and psychology. The Journal of Mind andBehavior, 10, 117-122.

Staats, A. W. (1991). Unified positivism and unification psychology: Fador new field? American Psychologist, 46, 899-912.

Titchener, E. B. (1898). The postulates of a structural psychology. Phil-osophical Review, 7, 449-465.

Tolman, E. C. (1967). Purposive behavior in animals and men. NewYork: Appleton-Century-Crofts. (Original work published 1932)

Watkins, M. J. (1990). Mediationism and the obfuscation of memory.American Psychologist, 45, 328-335.

Watson, R. I. (1967). Psychology: A prescriptive science. American Psy-chologist, 22, 435-443.

Zajonc, R. B., & Mclntosh, D. N. (1992). Emotions research: Somepromising questions and some questionable promises. PsychologicalScience, 3, 70-74.

Zukav, G. (1979). The dancing Wu li masters: An overview of the newphysics. New York: Bantam Books.

942 September 1993 • American Psychologist