4

The social structure of

psychological experimentation

How the Leipzig research community organized itself

In the last section of chapter 2 we took note of the fact that the establish-
ment of psychological laboratorios involved the institutionalization of cer-
tain social arrangements. Psychological experimentation became a
collaborative effort dependent on a división of labor among individuals
who carried out different functions in the experimental situation. More
specifically, a broad distinction emerged between those who acted as the
human source of psychological data, the experimental subjects, and those
who manipulated the experimental conditions, the actual experimenters.
Although this distinction was essentially a response to the practical exi-
gencies of late nineteenth-century “brass instruments” experimentation
and not the product of deep reflection, it rapidly became traditionalized.
To carry out sophisticated experiments with relatively complex apparatus,
one now had to assign the function of data source and the function of
experimental manipulation to different people.
What was not noticed at the time, or indeed for almost a century after-
ward, was that this arrangement created a special kind of social system -
one of psychological experimentation. The interaction between subjects
and experimenters was regulated by a system of social constraints that set
strict limits to what passed between them. Their communication in the
experimental situation was governed by the roles they had assumed and
was hedged around by taken-for-granted prescriptions and proscriptions.
However, the specific features of this social system were not necessarily
fixed. The basic división of labor between experimenters and subjects still
left much room for local variation. For instance, there was nothing in the
practical requirements of psychological experimentation that dictated a
permanent separation of experimenter and subject roles. The practical

49
50 Constructing the subject
exigencies that had led to a separation of these roles did not preclude the
possibility that the same person might take one or the other role on different
occasions. All that experimental practice entailed was the convenience of
separating these roles for a particular experimental session; it did not
necessarily decree that people could not exchange these roles from session
to session. However, other factors might enter the picture to convert the
occasional separation of experimenter and subject roles into a permanent
separation. Whether this in fact happened is a matter for historical inves-
tigation.
What kind of factor might be expected to have an effect on the per-
manence of the división of labor in experimental situations? One such
factor is quite obvious. Psychological experiments are not conducted in a
social vacuum. 1 The participants in such experiments are not social blanks
but enter the experimental situation with an already established social
identity. It is of course impossible to insulate the social situation of the
experiment from the rest of social life to the point where the participants’
social status outside the experiment has absolutely no bearing on their
interaction within the experimental situation. Who gets to play experi
menter and who subject may have something to do with the roles that
individuals play outside the experimental situation. Similarly, the possi
bility of exchanging experimental roles may depend on the social identity
of those who are occupying those roles.
Apart from the likely influence of social factors that are clearly external
to the experimental situation, there is the question of how the distribuí ion
of roles in the face-to-face laboratory situation ties in with other activities
that form a necessary part of the social practice of investigation. The direct
interaction of experimenters and subjects in the laboratory is only one part
of the research process. Activity in the laboratory is embedded in a pen
umbra of other activities, like theoretical conceptualization, data analysis,
and writing a research report. The fact that in the contemporary psychology
experiment these activities are assumed to form part of the experimenter
role should not lead us to presuppose that this was always the case. The
domain of what necessarily belonged to the experimenter role and what
to the subject role has in fact been subject to historical variation. Let us
therefore return to Wundt’s Leipzig laboratory to see how subject and
experimenter roles related to various other functions entailed by the social
practice of psychological investigation.
When we peruse the research reports published in Wundt’s house organ,
the Philosophische Studien, we find that the person who had authored the
published account of the experiment had not necessarily functioned as the
experimenter. For instance, Mehner published a paper on an experiment
in which he had functioned solely as the only subject while two other
persons had functioned as experimenters at various times. 2 Moreover, func-
tioning as a source of psychological data was considered incompatible with
 Social structure of experimentation 51

functioning as a source of theoretical conceptualization. During the first

years of his laboratory, Wundt appeared regularly as a subject or data
source in the experiments published by his students, although he also
contributed much of the theory underlying these experiments. 3 Interest-
ingly, Wundt does not seem to have taken the role of experimenter. This
suggests that the role of psychological data source was considered to require
more psychological sophistication than the role of experimenter. The role
of experimental subject was quite compatible with Wundt’s status as head
of the laboratory, but the role of experimenter could be left to his students.
If the role of functioning as source of psychological data was quite com
patible with the functions of a scientific investigator, there was clearly no
reason to make a permanent distinction between experimenter and subject
roles in the laboratory situation. Indeed, it was not uncommon for Wundt’s
students to altérnate with one another as stimulus administrators and as
sources of data, sometimes in the same experiment. 4 The roles of subject
and experimenter were not rigidly segregated, so that the same person
could occupy both at different times. Their differentiation was regarded
as a matter of practical convenience, and most participants in the laboratory
situation could play either or both roles equally well.
The participants in these experiments clearly saw themselves as engaged
in a common enterprise, in which all the participants were regarded as
collaborators, including the person who happened to be functioning as the
experimental subject at any particular time. Characteristically, authors of
experimental articles would sometimes refer to the experimental subjects
as co-workers (Mitarbeiter), especially when giving their ñames as part of
the published account of the investigation.5 Moreover, functioning as sub
jects and experimenters for one another would usually be part of a rela-
tionship that extended beyond the experimental situation. Those who
participated in these situations generally knew each other as fellow stu
dents, as friends or friends of friends, or as professor and student. The
experimental situation was not based on the interaction of strangers.
What we find in Wundt’s Leipzig laboratory is a distinctly collaborative
style of interaction among the participants in the experimental situation,
not a situation characterized by fixed role and status differentials. (Of
course, Wundt himself occupied a special position, but once he entered
the experimental situation as a subject his responses were tabulated along
with everyone else’s and were not given any special importance.) In these
early experiments the interaction of investigators with their human data
sources has not been clearly separated from the interaction among the
investigators. Most of the individuals who functioned as subjects in these
experiments were members of the research community. Not all the subjects
who appear in the research reports of the Leipzig group were students of
psychology, but those who were not seem to have had a definite interest
in the subject and generally some experience of psychological experimen-
52 Constructing the subject
tation. So instead of a two-tiered social structure in which there is a clear
distinction between one social interchange with experimental subjects and
another with colleagues, there is a less differentiated situation in which,
to a large extent, the scientific community is its own data source.
This style of psychological experimentation appears to have resulted from
the convergence of a particular tradition of scholarly work with a special
kind of research goal. The scholarly tradition was that associated with the
Germán nineteenth-century university system. It was based on linking the
training of an intel·lectual elite with the systematic production of new knowl-
edge in the context of a collaborative research enterprise.6 The research
goal involved the analysis of generalized processes characteristic of the
normal, mature human mind. Psychological experimentation was designed
to investigate this object much as physiological experimentation was de
signed to analyze biological processes involved in the functioning of normal,
mature organisms.
But just as the practice of introspection had helped to construct the
object it was meant to investigate, so the new practice of psychological
experimentation constructed its own object. Experimental subjects were
not studied as individual persons but as examples that displayed certain
common human characteristics. That is why the role of subject could be
assumed by any member of the research community. They did not represent
themselves but their common mental processes. These “elementary” men
tal processes, as Wundt called them, were assumed to be natural objects
that could be studied independently of the whole personality. All that was
necessary were the restricted conditions of the laboratory and a certain
preparation of the subject. That preparation involved a specific kind of
background and a suitably collaborative attitude. But this bringing together
of restricted conditions of stimulation and response - of psychologically
sophisticated background and of a readiness to play the subject role -
amounted to a careful construction of the very object that the research
program was designed to investigate. This did not make it any less “real,”
but it did mean that it would have to compete for attention with other
psychological objects constructed in different ways.

Alternative models of psychological investigation

The model of experimentation inaugurated at Leipzig had a special sig-
nificance for the early development of psychology as an acadèmic discipline
with scientific pretensions, but it was not the only model of psychological
experimentation available at the time. The clinical experiment had goals
and social arrangements of quite a different order. The first major research
program that depended on this type of experimentation involved the ex
perimental study of hypnotic phenomena. At exactly the same time that
Wundt’s Leipzig laboratory was getting under way, a group of French
 Social structure of experimentation 53

investigators embarked on the systematic use of experimental hypnosis as

a tool of psychological research. 7 In their studies various psychological
functions were investigated under conditions of experimentally induced
hypnosis. Now, contrary to the practice in Leipzig, there was no inter-
change of experimental roles among the participants in these French studies
but a clear and permanent distinction between experimenters and individ
uals experimented on. Experimenters remained experimenters and hyp-
notic subjects remained hypnotic subjects. Moreover, there was a glaring
status difference between these male scientists and their generally female
lay subjects. The distribution of social functions approximated the pattern
of the typical modern experiment, with the function of providing the data
source being strictly segregated in an experimental subject role/
By 1890, Binet had turned from experiments on hypnotized subjects to
experiments on infants.9 This was possible without altering the essential
social structure of the experimental situation. For the Wundtian experi
ment, on the other hand, this kind of extensión of scope was not possible.
The clinical experiment had emerged in a medical context. Those who
functioned as subjects in these experiments were identified by labels such
as “hysterics” or “somnambulists.” Where normal or “healthy” subjects
were used, it was for purposes of comparison with diagnosed clinical cases
who were the essential target of the research. The experimenters were
individuals with a medical background. Before experimental sessions be-
gan, the experimenter and the subject were already linked in a physician-
patient relationship, and the essential features of this relationship were
simply continued into the experimental situation. The whole situation was
defined in medical terms. A crucial feature of this definition was the un-
derstanding that the psychological States and phenomena under study were
something that the subject or patient underwent or suffered.
In the context of the clinical experiment, we find the first consistent
usage of the term “subject” in experimental psychology. These medically
oriented experimenters quite spontaneously refer to the patients they ex
perimented on as “subjects” (sujet), because that term had long been in
use to desígnate a living being that was the object of medical care or
naturalistic observation. This usage goes back at least to Buffon in the
eighteenth century. Before that a “subject” was a corpse used for purposes
of anatomical dissection, and by the early nineteenth century one spoke
of patients as being good or bad subjects for surgery. 10 When hypnosis
carne to be seen as an essentially medical matter, which was certainly the
case in the Paris of the 1880s, there was nothing more natural than to
extend an already established linguistic usage to yet another object of
medical scrutiny. However, within the medical context we immediately get
the formulation “healthy subject” (sujet sains)" when it is a matter of
comparing the performances of normal and abnormal individuals. From
this it is a very short step to the generalized use of the term subject to
54 Constructing the subject
refer to any individual under psychological investigation, a step quickly
taken by Binet and others.
The English term subject had acquired similar medical connotations as
its French equivalent. In the eighteenth century it was used to refer to a
corpse employed for anatomical dissection and by the middle of the nine-
teenth century it could also mean “ a person who presents himself for or
undergoes medical or surgical treatment” - henee its use in the context of
hypnosis. 12 In the English-language psychological literature, the first uses
of the term subject occur in the context of experiments involving the hyp-
notic State. 13 It is interesting to note that at this time Stanley Hall uses the
term subject in the context of his work in the area of experimental hypnosis
but switches to the terms “percipient” and “observer” when reporting on
his experimental work with normal individuals. 14
J . McKeen Cattell appears to have been the first to use the English term
subject in describing a psychological experiment involving a normal adult
human data source. However, he is by no means sure of his ground, for
in an 1889 paper coauthored by him we find the formulation “an observer
or subject,” with “subject” in inverted commas. 15 Putting the term subject
in inverted commas, when used in this context, was quite common in the
English literature of this time, indicating that this usage was of recent
origin, possibly influenced by French models. 16
Thus the earliest years of experimental psychology were marked by the
emergence of two very different models of the psychological experiment
as a social situation. The Leipzig model involved a high degree of fluidity
in the allocation of social functions in the experimental situation. While
participants in this situation could assume both the role of experimenter
and that of subject, the latter role was, if anything, more important than
the former. In the clinical experiment, by contrast, experimenter and sub
ject roles were rigidly segregated. The experimenter was clearly in charge,
and only he was fully informed about the significance of the experiment.
Experimenter and subject roles were not exchangeable. The objeets of
study that these two experimental situations were designed to display were
clearly different. The clinical experiment was meant to display the effects
of an abnormal condition for the benefit of the informed experimenter and
those who were able to identify with him. On the other hand, the Leipzig
experiment was meant to display universal processes that characterized
all normal minds and whose significance was therefore open to all. In the
one case the object of investigation presupposed the asymmetry of the
experimenter-subject relation, but in the other case it did not.
Although these two models of psychological investigation constituted the
types whose fundamental divergence holds the greatest theoretical interest,
several other models had begun to emerge before the end of the nineteenth
century. The earliest of these first appeared in England. In 1884, at the
International Health Exhibition in London, Francis Galton set up a lab-
 Social structure of experimentaron 55

oratory for “testing” the “mental faculties” of members of the public. Here
we have an investigative situation whose social structure is clearly different
from that of either the Leipzig or the Paris model. The individuals inves-
tigated were not medically stigmatized but presumed to be ordinary mem
bers of the public. Yet their role and that of the investigator were definitely
not exchangeable, and between these roles there was a clear status differ-
ence. The investigator was supposed to have some kind of expert knowl-
edge about the individual tested, and he was willing to share this knowledge
upon payment of a fee. Galton charged every person who availed himself
of the Services of his laboratory the sum of threepence in return for which
that individual received a card containing the results of the measurements
that had been made on him or her. The relationship operated on a “fee
for Service” principie, but the Service was clearly not thought of as being
medical in character. Galton referred to the individuals who presented
themselves for testing not as subjects but as “applicants.”
There was apparently no lack of “applicants” for Galton’s Services, over
9,000 having been tested by the time the exhibition closed. 17 This seems
to indícate that there was something not altogether unfamiliar about the
kind of interaction that Galton’s laboratory invited, and that the situation
was sufficiently meaningful to a large number of people, not only to induce
them to particípate but also to pay money for the privilege. Competitive
school examinations would certainly have provided a general social model
for anthropometric testing. But beyond this it is possible that a more specific
social model for Galton’s anthropometric laboratory was provided by the
practice of phrenology, which, though by then discredited, had been widely
relied upon a generation earlier. Phrenologists had long offered individuals
the Service of informing them about their “mental faculties” on the basis
of measurements performed upon them. Some of Galton’s “applicants”
may not have regarded his Service as being so very different from that
provided by a familiar model. Even Queen Victoria had consulted a phre-
nologist about her children, and as a young man Galton himself had gone
to a phrenologist and taken his report very seriously. 18 1 am not of course
suggesting an explicit resolve to imítate the phrenologist’s practice, but
rather the operation of an analogous social practice as an implicit model
that made Galton’s innovative procedure immediately acceptable.
A comparison of the social structure of Galton’s investigative practice
with that of the Leipzig or the clinical experiment reveáis some divergent
features. The utilitarian and contractual elements of Galton’s practice are
particularly striking. Galton was offering to provide a Service against pay
ment, a Service that was apparently considered of possible valué to his
subjects. What he contracted to provide his subjects with was Information
about their relative performance on specific tasks believed to reflect im
portant abilities. 19 His subjects would have been interested in this infor-
mation for the same reasons that they or their parents were interested in
56 Constructing the subject
the Information provided by phrenologists: In a society in which the social
career of individuals depended on their marketable skills any “scientific”
(i.e., believed to be objective and reliable) information pertaining to these
skills was not only of possible instrumental valué to the possessors of those
skills but was also likely to be relevant to their self-image and their desire
for self-improvement. 20
From Galton’s own point of view the major hoped-for return on this
large-scale investment in “anthropometric measurement” was certainly not
the money but a body of information that might ultimately be useful in
connection with his eugenics program. The practical implementation of
this social program of selective human breeding would have been facilitated
by a good data base on human ability. Galton’s interests in this research
situation were just as practical as those of his subjects, the difference being
that while they were interested in their plans for individual advancement
he was interested in social planning and its rational foundation.

Three models of practice compared

The object of investigation that Galton and his subjects created through
their structured interaction in the testing situation was very different from
the objects constituted by either the Leipzig or the clinical experiment,
although it had elements in common with both. What Galton’s testing
situation produced was essentially a set of individual performances that
could be compared with each other. They had to be individual perfor
mances - collaborative performances were not countenanced in this situ
ation. At least the performances were defined as individual, and the fact
that they were the product of a collaboration between the anthropometrist
and his subjects was not allowed to enter into the definition. The perfor
mances therefore defined characteristics of independent, socially isolated
individuals and these characteristics were designated as “abilities.” An
ability was what a person could do on his own, and the object of interest
was either the individual defined as an assembly of such abilities or the
distribution of performance abilities in a population. The latter was Gal
ton’s primary object of interest, the former that of his subjects.
Locating the object of research within an isolated individual person was
of course a basic common feature of all forms of psychological investigation;
in fact, this constituted it as psychological, rather than some other kind of
investigation. However, the line of inquiry inaugurated by Galton was the
most uncompromising in this respect. As we saw in the previous chapter,
Wundt was uncomfortable about the way in which the experimental method
isolates the individual from the social-historical context, so that he limited
the reach of this method to the “simpler” mental processes. The French
clinical experimentalists were interested in the interindividual process of
suggestion, which they mostly regarded as a fundamental condition for the
 Social structure of experimentation 57

appearance of the abnormal States they wanted to investígate. But Galton’s

anthropometry was quite radical in its conceptual severing of the links
between an individual’s performance and the social conditions of that per
formance. It accomplished this by defining individual performances as an
expression of innate biological factors, thereby sealing them off from any
possibility of social influence.
The potential practical appeal of Galton’s anthropometric method was
based on three features. First was the radical individualism, which we have
just noted meant that it claimed to be dealing with stable and unalterable
individual characteristics that owed nothing to social conditions. This pro-
vided the necessary ground for supposing that the performance of individ
uals in the test situation could be used as a guide to their performance in
natural situations outside the laboratory.
Second, there was the partly implícit element of interindividual com-
petition built into the method. Of interest was not just one individual’s
performance but the relative standing of that individual in comparison with
others. For this type of investigative practice any one episode of experi-
menter-subject interaction only obtained its significance from the fact that
it was part of a series of such interactions, each of them with a different
subject. In the other forms of investigative practice single laboratory ep-
isodes were also often part of a series, but here the series was constituted
by a variation of conditions with the same subject. In the Galtonian ap-
proach the series of experimental episodes was essentially defined by the
differences among the subjects. Of course, duplication of an experiment
with different subjects was known both in the Leipzig laboratory and in
clinical experimentation. But this was merely a replication of the experi
mental episode for the purpose of checking on the reliability of the ob-
servations made. For the practice of anthropometry, on the other hand,
the set of experimental episodes with different subjects formed a statisti-
cally linked series, and it was this series, rather than any of the individual
episodes, that formed the essential unit in this form of investigative prac
tice. Only in this way was it possible to generate the desired knowledge
object: a set of performance norms against which individuals could be
compared.
Whereas the Leipzig style of psychological experimentation only made
use of the división of labor between experimental and subject roles, the
Galtonian approach added to this the multiplication of subjects as an in
herent and necessary component of its method. This was closely connected
to a third feature of its practical appeal, namely, the statistical nature of
the Information it yielded. To be of practical valué, the comparison of
individual performances had to be unambiguous, so that rational individual
or social policy decisions could be based on it. The way to achieve this
was by assigning quantitative valúes to performances, thus allowing each
individual performance to be precisely placed in relation to every other
58 Constructing the subject
performance. Therefore, the anthropometric approach was designed to
yield knowledge that was inherently statistical in nature. For the Galtonian
investigator the individual subject was ultimately “ a statistic.” This made
him a very different kind of knowledge object than the “case” of clinical
experimentation or the exemplar of a generalized human mind, as in the
Leipzig experiments.
These differences in the status of the knowledge object were linked to
certain features of the interaction between investigators and their human
subjects. What was of particular relevance here was the bearing of the
prior relationship between experimenters and subjects on the social system
of experimenter-subject interaction. In Galtonian practice investigators
and their subjects met as strangers, contracting to collaborate for a brief
period of experimentation or testing. In the clinical experiment investi
gators and subjects generally extended an existing pattern of physician-
patient relationships into the experimental situation. In the Leipzig labo-
ratory, as we have seen, the participants were fellow investigators whose
common interests and friendships often extended across several
experiments.
Thus, in each case there was a certain fit between the kind of knowledge
object sought for and the nature of the experimental interaction that was
used to produce it: The superficial interaction of strangers in a limited,
contractually defined relationship was the perfect vehicle for the production
of a knowledge object whose construction required a series of brief en-
counters with a succession of human subjects. 21 Similarly, the collaborative
interaction of colleagues in exchangeable subject and experimenter roles
was nicely suited to the construction of a knowledge object defined in terms
of common basic processes in a generalized adult mind. Finally, the phe-
nomena produced by clinical experimentation depended in large measure
on a tacit agreement among the participants to extend the medical context
into the experimental situation.
Each type of investigative situation constitutes a coherent pattern of
theory and practice in which it is possible to distinguish three kinds of
interdependent factors. First of all, there is the factor of custom which
provides a set of shared taken-for-granted meanings and expectations
without which the interaction of the participants in the research situation
would not proceed along certain predictable paths. Much of this pre-
understanding among the participants depends of course on general cultural
meanings, but what is of more interest in the present context are the local
variations in these patterns. We have distinguished several varieties of
custom that played a role in the emergence of independent styles of psy-
chological investigation in different places. These were the customs of the
Germán nineteenth-century university research institute, of scientific med
ical investigation, of competitive school examinations, and possibly of phre-
nological assessment. Here were the sources of models of social interaction
 Social structure of experimentation 59

that could be readily adapted to the purposes of psychological research.

The existence of these models meant that psychological research could get
started without having to face the impossible task of inventing totally new
social forms. However, once it did get successfully started the new practice
led to new variants of the oíd forms and established its own traditions.
The actual practices of psychological investigation, though derived from
certain customary patterns, were not identical with these patterns and
therefore constituted a second distinguishable element in the emerging
local styles. A third element was constituted by the diverging knowledge
interests that have already been alluded to. The three original forms of
psychological research practice were by no means interested in the same
kind of knowledge. There was a world of difference between knowledge
of elementary processes in the generalized human mind, knowledge of
pathological States, and knowledge of individual performance comparisons.
From the beginning, psychological investigators pursued several distinct
knowledge goals and worked within investigative situations that were ap-
propriate to those goals. However, it would be quite inaccurate to think
of their adoption of appropriate investigative situations as a product of
delibérate rational choice. The knowledge goals that these investigators
pursued were as much rooted in certain traditions as were the practices
that they adopted. Intel·lectual interests and the practices that realized those
interests were both generally absorbed as a single cultural complex of
theory and practice. Each such complex then led to the construction of the
kind of knowledge object it had posited.
Modern psychology began with several different models of what a psy
chological investigation might look like, and the differences among these
models went rather deep. Psychological investigation only existed in a
number of different historical incarnations. Although in certain locations
a particular model of investigation might achieve an overwhelming pre-
dominance for a considerable period, this should not mislead us into equat-
ing a particular form of experimentation with experimentation as such.
Qüestions about the scope and limits of experimentation in psychology
have usually presupposed a particular variant of experimental practice.
However, it is historically more meaningful to compare the implications
of different forms of practice.

Social psychological consequences

One set of qüestions that is suggested by the historical analysis of psycho
logical experiments concerns the social psychology of such experiments. If
there were different models of psychological investigation, with different
patterns of social interaction among the participants, then each such model
is likely to have entailed its own social psychological problems. In historical
perspective it is inadequate to take one type of investigative situation as
60 Constructing the subject

Figure 4.1. Wilhelm Wundt and assistants at the Leipzig laboratory (photograph
ca. 1910)

V *
* 'V' 4
•

Figure 4.2. A medical demonstration of hypnosis and grand hysteria; from a paint-
ing by André Pierre Brouillet, The School of Jean-Martin Charcot
 Social structure of exper imentation 61

Figure 4.3. Galton’s anthropometric laboratory in London

representing the psychological experiment and to generalize from the so

cial psychological factors in this situation to all types of psychological
experimentation.
One also needs to distinguish between the intended and the unintended
social features of experimentation. This is where the term “artifact” can
be seriously misleading. 22 All results of psychological experiments are social
artifacts in the sense that they are produced by the participants in certain
specially constructed social situations. But insofar as this production cor-
responds to the intentions and plans of those in charge of the experiments,
it is not thought of as artifactual. That epithet is commonly reserved for
the unavoidable unintended social features of experimentation - for ex-
ample, the suggestive influence of the experimenter’s hopes and expecta-
tions, or the inappropriate anxieties of the subject. However, just as the
planned social features of experimentation vary considerably in different
models of the experimental situation, so do the unplanned features. It is
not unlikely that each basic model would have been prey to a different set
of unintended social psychological disturbances.
In the clinical experiment a major source of such disturbances would
have to be sought in the profound status differential that rigidly separated
experimenters and subjects. This status differential meant that the exper-
62 Constructing the subject
imental situation was, among other things, an encounter between individ
uals who were very unequal in power. Moreover, this power inequality
was an essential feature of the experimental situation, because the subject’s
role was that of providing the material on which the experimenter could
make his observations and test his hypotheses. The subject had to play the
role of a biological organism or medical preparation if the experimental
scenario was to be completed successfully. Finding oneself in this inferior
position, the subject could be expected to be affected by the stresses and
anxieties that people in such positions commonly experience. These sub-
jects might have been expected to experience some anxiety about receiving
a favorable evaluation from the experimenter and to worry about possible
harmful consequences to themselves. A situation that they did not control
and only half understood would have aroused anxiety in any case.
However, from the viewpoint of the outcome of the experiment, these
anxieties were probably less serious in themselves than the various coping
strategies in which the subjects, as persons in inferior and relatively pow-
erless positions, could be expected to indulge. 23 These strategies might vary
all the way from apathetic reactions to strenuous collusion with the inves-
tigator to give him the kind of performance he wanted. In any particular
case one would not know how much of what was displayed to the inves-
tigator was due to peculiarities of specific coping styles and how much was
due to the factors in which the investigator was really interested.
At this point we should not overlook the fact that the social structure
of experimental situations may have had unintended social psychological
consequences for the experimenter’s as well as for the subject’s role. People
in positions of power and control are quite apt to drift into arrogance and
callousness in the absence of appropriate institutional safeguards. Insofar
as the social situation of the experiment is marked by a clear power dif-
ferential, the psychological reactions of the more and the less powerful
role incumbents are likely to be complementary to one another. So the
performance the subject puts on for the benefit of the investigator is not
recognized for what it is because the investigator does not think of the
subject as a human agent but simply as material for the demonstration of
his pet ideas. The extreme illustration of this outcome was provided by
Charcot’s notorious públic demonstrations of hysteria at the Salpétriere. 24
These hardly qualified as experiments, yet the social dynamics that they
involved only represented gross forms of processes that had their subtler
counterparts in every clinical experiment.
Although the investigative style pioneered by Galton would have shared
some of the social psychological problems of the clinical experiment, the
less-intense involvement of the participants and the introduction of a cer-
tain contractual reciprocity would tend to mitigate these problems. How
ever, the kind of experimenter-subject relationship that this type of study
prometed had potential problems of its own. In particular, the divergent
 Social structure of experimentation 63

interests that experimenters and subjects had in the outcome of the ex

perimental situation, combined with the superficiality of their contact, con-
tained the seeds of trouble. The subject’s performance would be rendered
in the Service of aims that were not those of the investigator and the nature
of their relationship was not such as to allow much opportunity for mutual
explanations. This was surely a situation in which misunderstanding was
likely to flourish. Such misunderstanding would take the form of divergent
definitions of the experimental situation, and henee of the experimental
task, by experimenter and subject. Although less significant for many of
the tasks that Galton employed, it would become a more and more serious
problem as Galton’s successors increased the psychological complexity and
potential ambiguity of experimental tasks.
Participants in the Leipzig model of psychological experimentation were
much better protected against problems of this kind, for their collegial
relationship and the exchangeability of experimenter and subject roles
would ensure that misunderstanding was kept at a minimum. Ñor would
this style of research be plagued by the psychological consequences of
extreme status inequality in the manner of the clinical experiment. How-
ever, the Leipzig model was not free of unwanted social psychological side
effeets of its own. The most noteworthy of these derived from the same
source as its strength, namely, the cióse mutual understanding that devel-
oped among the participants. This made it relatively easy for observations
to be reported that depended entirely on the conventions and tacit agree-
ments among the members of specific research groups. Such observations
were then unverifiable by anyone else. The reports from Titchener’s Cor
neli laboratory on the nonexistence of imageless thought provided the
elassie example of this outcome. 25 It has been observed that the problems
of “introspective psychology” did not arise on the level of comparing the
introspective reports of individuals with each other but on the level of
reconciling the conflicting claims of different laboratories. Historically,
these problems had less to do with the intrinsic limitations of introspection
than they had to do with the social psychological context in which intro
spection was employed - namely, the context provided by the Leipzig
model of experimentation.
With the advantage of hindsight, it is easy to see that there was no
prescription for arranging the social conditions of psychological experi
mentation that was free of unintended social psychological side effeets.
Psychological research was only able to collect its material by putting whole
persons into social situations. Even though this research was originally
interested neither in the persons ñor in the social situation but in certain
isolated intraindividual, processes, it could isolate these processes only
conceptually. In practice, the research had to contend with the persons
who were the carriers of these processes and with the social situations that
were necessary to evoke the phenomena of interest. But that inevitably
64 Constructing the subject
led to the appearance of various side effects, which were unintended and
for a very long time not even recognized. However, the precise nature of
these effects would vary with the model of investigative practice that was
adopted. There was no single unflawed procedure for exposing the truth,
only a variety of practices with different sources of error. Any estímate of
the relative seriousness of these sources of error would have to depend on
the kind of knowledge goal one had adopted.

The eclipse of the Leipzig model

The last three chapters have presented an analytic account of the origins
of modern psychology as a form of investigative practice. In order to
prepare the transition to the rest of this book, which will deal mainly with
twentieth-century developments, let us take a brief look at certain historical
developments beyond the earliest period of experimental psychology.
A major development was the rapid adoption of all the major techniques
of psychological investigation by American investigators. The Leipzig
model of investigative practice was replicated by many of those who had
spent some time working with Wundt, most notably by Titchener at Cor
neli. But what made Titchener somewhat unusual was the fact that he
never ceased to insist that the Leipzig model was the only one on which
a genuine Science of mental life could and should be based. Most of the
early American investigators were much less dogmàtic about their choice
of methods, and some were extremely eclectic. G . S. Hall, for example,
carried out different pioneering studies that followed all three of the models
we have distinguished. 26 Cattell was particularly active in promoting the
style of investigation that had been begun by Galton. 27 Clinical experi-
mentation had its adherents too. 28 Early American psychology presented
a mixture of investigative styles with few centers showing exclusive com-
mitment to one style, although some had obvious preferences. However,
in the course of the first half of the twentieth century a particular synthetic
type of investigative practice developed and carne to domínate American
psychology. That development will be traced in the following chapters.
The model of investigative practice that proved least viable in the twen
tieth century was the Leipzig model. Because this model has a specific
feature, which is always unambiguously reported in published research
reports, it is possible to follow its decline rather accurately by means of a
content analysis of psychological journals. The feature in question involves
the exchange of experimenter and subject roles among at least some of
the participants in the same experiment. Highly characteristic of the Leipzig
model, the feature is clearly absent in clinical experimentation and in the
type of investigation pioneered by Galton.
Concentrating on this feature as an index of the relative prevalence of
the Leipzig model has the practical advantage that it is readily and reliably
 Social structure of experimentation 65

Table 4.1. Percentage of empirical studies in which there was an exchange

of experimenter and subject roles

Journal title 1894-1896 1909-1912 1924-1926 1934-1936

Philosophische Studien 32 79" _ _ _b
—
American Journal of Psychology 38 28 20 8
Psychological Review 33 28 N.C. C
Archiv für die gesamte
Psychologie — 43 28 16
Zeitschrift für angewandte
Psychologie — 4 N.C.
Journal of Educational
Psychology — 2 0 0
Psychological Monographs — 18 18 7
Journal of Experimental
Psychology — — 4 6
Journal of Applied Psychology — — 0 0
Psychologische Forschung — — 38 31

"The second entry for Philosophische Studien is based on the new series of that
journal, called Psychologische Studien.
' Dashes indícate periods before or after the appearance of a journal.
r
An entry of N.C. means that coding of the journal was discontinued because no
significant information was likely to result; for example, in the case of Psychological
Review the proportion of codable empirical articles went from low to zero.

coded in published research reports. It should be noted, however, that this

does not provide an accurate estímate of the absolute prevalence of the
model, because it misses the cases where research colleagues exchange
roles in different experiments but not in the same experiment. This practice
was especially prevalent in the early days of experimental psychology, but
picking it up from the research literature is time-consuming and hardly
justified by the intrinsic interest of the results. There are also other features
of the Leipzig model that are missed. However, it is the fluctuation in the
relative popularity of the Leipzig model over the years that is historically
more interesting, and this can be assessed by the simplified Índex as already
described.
Empirical articles published in major American and Germán psycho
logical journals were therefore coded in terms of the presence or absence
of an exchange of experimenter and subject roles in the same study. (The
sampling of journal volumes for different time periods, and the rules gov-
erning the inclusión of articles in the coding process are described in the
Appendix.) Table 4.1 shows the results of this content analysis.
Although the simplified Índex used underestimates the use of the Leipzig
66 Constructing the subject
model, this underestimation tends to be greatest when the use of this model
is relatively high, especially in the earliest period. So the simplification in
coding dilutes rather than accentuates the overall trend. This trend never-
theless emerges quite clearly. It involves an overall decline in the practice
of exchanging experimenter and subject roles. Where continuity of pub-
lication allows a comparison of the same journal over several decades, the
practice generally declines steadily with each time sample. The best doc-
umented case is that of the American Journal of Psychology where inci-
dence of the practice gradually declines from 31 to 8 percent over a forty-
year period. These figures do not suggest a sudden abandonment of the
practice so much as a steady deterioration in its relative frequency of
adoption. Nor does this general trend appear to be a specifically American
phenomenon, because among the Germán journals, the one that is not
associated with a particular school, Archiv für die gesamte Psychologie,
shows the same trend.
It does seem as though the kind of experimental practice represented in
this Índex was always somewhat more popular in Germán than in American
psychology. In any time period the Germán figures are always higher than
the American. By far the highest incidence (79 percent) of this pattern of
research is to be found in Psychologische Studien, the Leipzig successor to
Wundt’s Philosophische Studien. But the Gestalt journal, Psychologische
Forschung, also maintains a relatively high incidence on the Índex at a time
when the practice of exchanging roles is being virtually abandoned by
everyone else. At no time after 1890 did the original Leipzig style of
arranging psychological experiments really domínate the discipline. At best
it was one model among others. Never dominant, it had dwindled virtually
to insignificance in American psychology during the period preceding
World War II.
For applied psychology, whether American or Germán, it never had the
slightest appeal, as is shown by the figures for the relevant journals, Zeit-
schrift für angewandte Psychologie, Journal of Applied Psychology, and
Journal of Educational Psychology. Applied psychology had committed
itself to knowledge goals that were unlikely to be advanced by the kind of
investigative practice associated with Wundt’s laboratory. What it was after
was knowledge that could be quickly utilized by agencies of social control
so as to make their work more efficient and more rationally defensible.
Knowledge that led to behavioral prediction suited this purpose, but knowl
edge obtained in situations where the participants collaboratively explored
the structure of their experience did not. Thorndike, a major pioneer of
the new applied psychology, saw this very clearly when he complained that
some of the procedures of acadèmic experimental psychology “were not
to aid the experimenter to know what the subject did, but to aid the subject
to know what he experienced.”29 The kind of knowledge that Thorndike
and others who shared his interests were after was predicated on a fun-
 Social structure of experimentation 67

damental asymmetry between the subject and the object of knowiedge.

This asymmetry was not merely epistemic, however, but also social, be-
cause it was accepted that it was to be useful to the subject (or those for
whom he acted) in controlling the object. What was desired was knowiedge
of individuals as the objects of intervention rather than as the subjects of
experience. Experimental situations in which there was an exchange of
roles between experimenters and subjects might well have seemed irrele-
vant to this type of knowiedge goal.
As more and more acadèmic psychologists, especially in the United
States, adopted knowiedge goals that were similar to those that had char-
acterized applied psychology from the beginning, the Leipzig model was
pushed even further into the background. It survived, on a relatively much
reduced scale, only in the area of research on sensation and perception,
the area in which it had also registered its first successes in the previous
century.30 Thus, as psychology’s knowiedge goals changed, its preferred
mixture of investigative practices changed too.
Neither clinical experimentation ñor the style of investigation pioneered
by Galton was left unaffected by these changes. On the contrary, the
Galtonian approach in particular underwent a dynamic development dur-
ing the first half of the twentieth century. But this affected the nature of
the knowiedge produced rather than the social parameters of the
experimenter-subject relationship. To analyze these developments, we will
have to turn from the social interior of the investigative situation to its
social exterior - that is, to the relations of investigators with each other
and with the wider society. The making of a psychological knowiedge
product involved more than just the interaction of experimenters and sub
jects. It involved matters of conceptual form, of appropriate packaging,
and of dissemination through approved channels. Twentieth-century de
velopments in investigative practice mostly affected these aspects, and it
is to them that we must now turn.