“direct experience,” “consciousness” and so forth, appears very strange
to me. They do not generally show an undue interest in epistemological
considerations. It is just _one_ question which suddenly arouses their
interest: How can I know about the direct experience of others? I shall
never have a definite proof of the validity of such knowledge. But
physics, that is another matter! There we are safe! The behaviorist,
however, forgets that it is a truism in epistemology that I shall
never be able to “prove” conclusively the existence of an independent
physical world. As an extreme purist I might argue this point exactly
as the behaviorist disputes the assumption of direct experience in
others. Somehow it does not occur to the behaviorist to apply his
epistemological criticism to the assumption of the physical world. He
does not say: Thou shalt not work upon a physical world, the existence
of which will always remain a mere assumption. On the contrary, he
assumes its reality with all the healthy naïveté which he lacks in
psychology. The reason is, perhaps, that the achievements of physical
science are imposing and have become the ideal of behaviorism. But
to the mind of a methodological purist that cannot be taken as a
satisfactory proof of the existence of the physical world. Of course,
personally and practically I am as convinced of its existence as any
behaviorist has ever been, and I am fully aware of the fact that
sciences may and must believe and postulate, where the epistemologist,
if he likes, may doubt. But then I shall believe and postulate the
direct experience of others, as well as the existence of physical
entities, if this makes my procedure simpler and more productive. And
I feel the more justified in this attitude, since I have seen that all
work I may do in physics is founded on direct experience and that,
therefore, the enormous superiority of physics over psychology in
accomplishment, cannot follow from any differences in this respect.
At this moment I see the behaviorist smiling ironically: With all his
medieval philosophy Mr. Köhler will never make any progress against
the sound scientific basis of behaviorism. I answer that this basis
of behaviorism is as philosophical as anything could be: it is purely
epistemological. In this connection the only difference between the
behaviorist and me is one of completeness: the behaviorist sees a
single theorem of epistemology and, as an extreme purist, he dwells
exclusively on this one point, ignoring the context from which it has
been taken. I am aware of this context; it is stated in the foregoing
argument; and I find myself therefore opposed to behaviorism, or any
other one-sided and impractical purism in science.
BIBLIOGRAPHY
W. S. Hunter: _Human Behavior_. 1928.
K. Koffka: _The Growth of the Mind_. 1924. 2nd edition, 1928.
J. B. Watson: in _Psychologies of 1925_ (ed. by C. Murchison).
A. P. Weiss: _A Theoretical Basis of Human Behavior_. 1925.
II
_Psychology as a Young Science_
IN the first chapter I became slightly autobiographical in order to
show why I have found it impossible to become a behaviorist. If I am
ever to become one, some adherent of that school will have to effect
the conversion. I do not myself see the way.
So far direct experience has been shown to be the raw material of both
physics and psychology. If despite this fact the physical sciences are
so far ahead of psychology in their development, what shall be our
method of imitating their achievements?
Contemporary physics has a real advantage in the careful selection of
the experiences which play a decisive rôle in the crucial moments of
physical research. Of course, the physicist ignores all “_subjective_
experience,” as defined in the previous chapter, as far as possible,
because apparently it has nothing to do with the properties of the
physical world. All the difficulties of psychologists who try to
observe and describe subjective experience are simply avoided by the
physical sciences.
But the selection and exclusion does not stop there. If most
“objective experience” was taken as evidence of physical existences
and physical events in early investigations, a more critical view of
the situation tended to eliminate large parts of this material as
well. The process of selection has become particularly severe, since
the advancement of science has made it possible almost everywhere
to transform qualitative observation into quantitative measurement.
At the present time almost all physical measurements are made in an
extremely indirect way. We do not observe an experienced counterpart of
the physical variable in question; we observe a related but different
object the virtue of which is that it lends itself to more exact
statement and is shown, or supposed, to be in a well-known functional
relationship to the physical variable. Of all possible objective
experience nothing seems better to fulfill the requirements of the
physicist than the localization of a single line (a needle) on a scale
of other lines, especially if this localization becomes the coincidence
of the first line (needle) with one of the others (or inclusion between
two which are near to each other). The range of direct experiences
used in measurement has now reached a minimum value, indeed. It almost
seems as if the same scale and the same needle were used universally;
it tells a hundred altogether different stories about the physical
world. It means “atmospheres” or “volts,” or “ampères” or “temperature”
and so forth, as the case may be. In addition to observing a definite
coincidence, and controlling the connection of the apparatus with the
system to be investigated, the physicist has only to recognize certain
words and figures on the scale. That is all the direct experience which
enters into the procedure. There is not much opportunity for serious
blundering. Even so simple a property as physical size is not measured
directly. No physicist measures the length of an object by comparing it
with the experienced length of a standard object directly. That would
not be accurate enough. Rather he prefers the method of coinciding
lines or points. Defining physical length by this method, he measures
the length of a line by observing the coincidence of its ends with
certain points on a scale.
If now we ask ourselves whether we shall imitate the procedure of the
physical sciences in psychology, two answers are to be given to our
question, because that procedure has two aspects. In the first place,
the procedure involves making statements about physical systems by
means of objective experience. As the behavior of men and animals
is observed in our objective experience--the direct experience of
the _subjects_ not playing any rôle in the matter--such a study is
perfectly legitimate and undoubtedly will be much more extended in
the future. Though it existed before the days of behaviorism, this
school is fundamentally right in praising the advantages of objective
procedure as against prevailing introspection. Although behaviorists
have gone too far through failing to appreciate that, even in the
objective method, direct experience remains the raw material as “the
objective experience” of an observer, their error is of no practical
importance as long as the second part of their answer to our question
is proper. Unfortunately, behaviorism has here tended to take the wrong
position.
As we have seen, in present physics objective procedure is
characterized by the use of a small group of selected experiences
and, consequently, by the exclusion of the great variety of objective
experience, the exclusion being made to satisfy the requirements of
quantitative measurement. Shall we do the same thing in psychology
as a science of behavior? Evidently that will depend upon the nature
of behavior itself. It is difficult to judge about a method as such.
A method is not good or bad in itself. It is good if it is properly
adapted to the essential aspects of my problems and my material;
and it is bad if it lacks regard for those essentials or misdirects
research. Therefore, what has been shown to be an excellent procedure
in one science or for some problems may be altogether useless or
even a hindrance in another science or for other problems. In this
respect, behavior is easily seen to have distinctly different phases
and to give the psychologist different tasks correspondingly. Wherever
indirect and quantitative methods can be applied in conformity
with the task, they should be applied. Since C. P. Richter and his
collaborators at the Phipps Psychiatric Clinic have now discovered a
way in which the various drives of animals, and their temporal changes,
can be investigated by registering amounts of general or of special
activity,[4] we shall be keenly interested in the future development of
the method. It is the correct procedure in cases in which total amounts
of activity in relation to outer and inner conditions can yield us
valuable information.
But what about other cases where either our problems are not of the
quantitative type or we have no way as yet to transfer our observations
from the direct experiences with which we are concerned to a special
region, indirectly connected with the first, but better adapted to
finely accurate statement and registration? Obviously, qualitative
types of behavior are not less important than quantitative differences
of some sort. Where we know about the qualitative varieties which
occur in behavior and where, in a concrete case, we are sure about
the special type before us under observation, we may be interested
in finding ways of quantitative measurement. But the discrimination
of qualitative types must be primarily accomplished. During the
observation of a puppy under a given set of conditions, this important
question will frequently arise: whether certain behavior of the animal
represents playful activity or “serious” reaction to the situation.
Such a question does not imply “consciousness” in the puppy; it refers
to some very characteristic differences in what is being objectively
observed. The difference is one of “quality” of behavior. Again, while
observing a man in a somewhat critical situation, it may be decisive to
observe whether he talks to us in a “steady” or in a “shaky” voice. At
the present time this is necessarily a qualitative discrimination; even
should we find in the future a method to measure the steadiness of a
voice, this method would forever presuppose that in direct observation
we know what is meant by unsteadiness as a definite characteristic of
the voice.
In a similar way, we are restricted in the application of indirect
methods to many other forms of behavior. I take it that the
behaviorists hold that we can investigate the “emotional behavior”
of subjects without being involved with their “consciousness” or
“subjective experience.” Certainly, psychologists have tried to
transfer observation in this case to a region of more accurate
registration and measurement. Much work has been done to develop
pneumographic, plethysmographic, galvanographic methods, and so forth.
The result is not altogether encouraging, however; for again, our
interpretation of the curves recorded by all these apparatus depends
greatly upon simultaneous _direct_ observation; by no means do we
feel justified to draw conclusions from the curves alone. All these
procedures seem to be more problematic in themselves at the present
time than an aid for the solution of psychological problems. In most
cases it is easier to see “anger” directly in a subject’s attitude than
to state and measure, let us say, the adrenin in his blood.
Why does this difficulty beset behavioristic psychology, and not occur
in physics? The answer is simple enough: Physics is an old science and
psychology is in its infancy. Over centuries physicists have succeeded
in eliminating a great many forms of more direct and qualitative
observation, relying upon a small number of indirect and very accurate
observations as decisive in experimentation. Their success depended
on previously acquired knowledge about the physical world. Most
indirect methods and sound measurement presuppose a vast background of
knowledge. Qualitative observation, experimentation and analysis have
provided the primary knowledge which enabled the physicist to discover
where he could measure an important factor, and what physical relation
makes an indirect observation capable of indicating occurrences not
well observable directly. Oersted had to discover the deflection of
a magnet in the neighborhood of an electric current before exact
measurement of intensities of current became possible. His was a
qualitative, direct observation; indirect and quantitative procedure
was the fruit of it. Even in our days, Röntgen did not _measure_ at the
start of his investigation of X-rays. He had to discover and to analyze
their properties in qualitative experimentation first of all. Later on
his rays could become a means of measuring the constants of crystals.
Too easily do we forget the simple truth that in their origins as well
as at the opening of a new field of special investigation later on,
the natural sciences, physics, geology, biology, rely upon qualitative
statements almost completely. Certainly, the quantitative and the
accurate indirect methods are the most imposing features of exact
sciences _now_ when we review them or consider them superficially. But
we ought to be aware of the fact that in most cases this procedure is
a mere refinement and end result of the underlying free and direct
observation, without which there would not have been a basis upon
which to build the refined superstructure. In the eighteenth century
Cavendish measured the resistances of different materials by comparing
the shocks he received in his arm when conductors of those materials
were used to connect him with a battery, the other pole of which was in
his other hand! Was that improper? On the contrary, it was perfectly
sound in a young science, which was trying to find that first knowledge
of facts upon which more refined methods may afterwards be founded.
From this it follows that where we have a good quantitative problem
in psychology and an accurate method of measuring and registering, as
in the case of Richter’s excellent experiments, we can immediately
apply the quantitative and indirect procedure, as used in physics. The
problems which Galileo attacked in the seventeenth century could be
solved quantitatively at once, the qualitative experience of everyday
life having sufficiently provided the necessary basis. But for the
majority of psychological problems this is not the case. Where do
we have that first more or less qualitative knowledge of important
functional relationships in psychology which might become the basis
for indirect and exact measurement? It does not exist. Since the
development of more “exact” methods presupposes its existence, our
main task must be the gathering of that knowledge. In most cases
our preliminary advance in this direction will have to be crude and
qualitative. Whoever protests that conclusion in the name of purism
does not understand our actual situation in psychology; he sees neither
the nature of, nor the historical background prerequisite for, special
quantitative methods. If we wish to imitate the physical sciences,
we must not imitate them in their contemporary, most developed form;
we must imitate them in their historical youth, when their state of
development was comparable to our own at the present time. Otherwise
we should behave like boys who try to copy the imposing manners of
full-grown men without understanding their _raison d’être_, also
without seeing that in development one cannot jump over intermediate
and preliminary phases. A survey of the history of physics is certainly
illuminating. Let us imitate the natural sciences, but intelligently!
Behavior is enormously rich in forms and nuances. Only acknowledging
this wealth, and studying it directly as it is given in all its
fascinating varieties, shall we become able gradually to find those
forms of more quantitative, and perhaps more accurate, procedure
which may become as adequate for us as are the methods of physics in
its realm. At present, and in this broader historical perspective,
qualitative observation and analysis may be, in a sense, more exact,
i.e., adequate to our subject-matter, than much blind measurement. We
shall press forward towards more refined methods, of course; but owing
to our situation as beginners, we can go forward only through the use
of less refined methods for the time being.
If organisms were more similar to the systems investigated in
physics, a great many methods of the physicist might be introduced
in our science exactly as the physicist uses them. But the
similarity is not very considerable, practically. One of the
greatest advantages which makes the physicist’s work so much
easier, is the simplicity of his systems. And his systems are
simple because in physics the experimenter determines their main
properties himself. He prepares them, more or less. I am far from
believing that organic processes are of a supernatural kind. On
the contrary, the most startling difference between the organism
and a simple system investigated by the physicist is the enormous
number of physical and chemical processes which, depending upon
each other in the most intricate manner, occur in the organism
simultaneously. And we are utterly unable to create simple organic
systems for elementary study. An amœba is a more complicated
system than all known systems of the inorganic world. We also know
that, in studying the properties of a nerve-muscle-preparation,
we are not investigating “a part” of natural behavior; because,
physiologically, its properties are not only simpler but also
radically different in some respects from what they would be if the
same nerve and muscle were contributing to normal behavior. It is
the _whole_ organism, as some behaviorists have rightly said, the
behavior of which is our subject-matter. But in the whole organism
one can seldom follow the change of one variable alone as the
consequence of one change in outer conditions. The change of one
factor usually involves concomitant changes in a great many others
and these affect the first retroactively. Now, isolation of changes
and reduction of variables are the great artifices by which the
physicist makes exact measurement possible in his simple systems.
Since this procedure does not work in our case, since we have to
take the organism more or less as it is, any kind of observation
that gives us an expression of its total attitude and activity, of
the integrated tendencies and states which it undergoes, will be
the right observation to begin with. Such is typical behavior as
we see (or hear) it before us, changing, with the conditions we
give our subject, in innumerable phases, and much too attractive
to be already excluded for the sake of narrowing ideals, taken
prematurely from quite another field of research.
In the meantime, however, the outward aspect of physics has been too
seductive. Since experimental psychology first became a science, every
now and then a wave of blind imitation has swept it off its feet.
Fechner himself was the first to copy _adult_ physics when psychology
was an embryo. He seems to have been convinced that measuring in itself
would make a science out of psychology. We have seen the result: If
flowers are impossible without a root and a stem, measuring, which is
fruitful only as the most refined consequence of previous qualitative
observation and experimentation, necessarily becomes a dead routine
without it. Hundreds of thousands of quantitative psychophysical
experiments have been made almost in vain, because no one knew just
what he was measuring or what were the processes upon which the whole
procedure was built. In Fechner’s own day, however, psychology came
into existence as a science, not as a result of his psychophysics, but
rather casually and _in spite_ of the premature quantitative program.