viewing the admirable energy with which “intelligence” is tested
quantitatively by hundreds of able psychologists, one is almost
reminded of Fechner’s time. In this case, it is true, there seems to
be a positive result for practical purposes at least. Apparently
something like a crude total ability for certain achievements is
measured by those tests which have been invented and applied since
Binet and Simon. And since the test-scores show a rather satisfactory
correlation with certain general abilities in study or other work
of subsequent life, we have in these tests a first approach towards
a practical prognosis. Still there is a grave danger in this very
success. Do we know or do we learn by those tests what processes and
factors, which are masked by the gross scores, coöperate in the test
performance? We do not know, nor do we learn much, concerning these
factors. Figuratively speaking, a given total score may mean: degree 3
of “intelligence,” together with degree 1 of “accuracy,” with degree
4 of “ambition” and degree 3 of “quickness of fatigue,” or it may
mean “intelligence” 6, “accuracy” 2, “ambition” 1, and “quickness of
fatigue” 4,--and so forth.[5] Quite a number of different factors may
combine in various proportions in order to give the same I.Q. And that
fact matters, even for practical purposes. In the educational situation
a teacher should treat the child according to the actual nature and
strength of the various factors constituting the total I.Q. of a given
child. This is not a new criticism, of course, but it must be repeated
for the sake of science. We seem to be too well satisfied sometimes by
our testing enterprise because it is a quantitative procedure, and as
such it appeals to us as thoroughly scientific. This is superficial,
however. If we compare the nature of testing with the main traits of
our ideal, viz., the procedure of physics, we find a difference which
may be decisive. What does the physicist ask when he must face a new
field of research? In such cases his questions have been: Is light an
oscillatory process? If it is, does it oscillate in the direction of
propagation or vertically to it? Is magnetism produced by the magnetic
fields of elementary currents in molecular structures? Why does
surface tension produce _regular_ forms of liquids and liquid films?
How can the spectrum of one element contain thousands of different
lines? This is the type of question with which we find the physicist
occupied. They are his fundamental problems, in the treatment of
which quantitative work appears at definite stages as an aid toward
the solution, eventuating finally in the determination of laws, but
directed and governed throughout by those questions about the _nature_
of states and processes. What are the phenomena? Are they of one type
or another? Such are the main problems of experimental science, in the
_service_ of which we find measurement. If now we ask ourselves what
problems about the processes underlying intelligent behavior we are
solving by our tests, not very many of us will have an answer ready.
But some will answer that from now on, intelligence shall be defined
as the X which is measured by those tests and that measuring is more
scientific than grubbing among the unknown functions of the nervous
system. The fact is that we are not in the habit of asking questions
about underlying processes in psychology, questions similar to those
in which the whole interest of physicists is centered. Instead of
imitating the very kernel of physics, we assume merely the outer,
quantitative form of exact science. Think of a physicist interested
in all types of motors who would restrict his study of them to the
following tests to be applied to all the various types: measurement of
their volume, of temperature at their surface, of ionization of the
air in their neighborhood, of their actual frequency of revolution,
and of the total weight of the bodies; who would calculate from these
data an average “power-coefficient” for each of them, define “power”
by the method, never ask a question about the working process in their
interior, and remain satisfied by this quantitative procedure for
many years. I know I am exaggerating in some respects if I compare
intelligence testing with such an attitude. I do it intentionally in
order to call attention to the main point, viz., that admiration for
quantitative method leads us almost exclusively to those research tasks
which immediately afford us an opportunity for measurement. Given the
natural narrowness of human interest, the next consequence will be our
failure to see those problems which do not lend themselves immediately
to quantitative investigation, problems which may be, however, the more
essential and the more scientific, in a deeper meaning of that word.
In such problems qualitative observation and experimentation would
constitute the first step; but since in adult physics (as it appears
from without) these methods seem to be despised, we dare not undertake
these fundamental tasks. Wishing to be utterly scientific, we may
just miss those opportunities for research which, in our stage as an
immature science, would be the most scientific of all. The nature and
the actual development of our problems should decide about the methods
to be followed. What actually happens, however, is that an imposing
method blurs our view of the subject-matter. This method then begins
to decide what problems we shall see. I do not object in principle to
testing as a preliminary mode of diagnosis and prognosis. But just as
the engineer learns from the investigations of processes in physics,
even practical diagnosis in applied psychology will never go very far
without an eager interest in the inner nature of intelligent behavior.
Eventually, if all psychologists occupied with “intelligence” are
engineers in the field of testing, who shall investigate the “physics”
of intelligence, without which we shall never have a true and sensible
engineering?
Occasionally, in animal psychology, the situation has tended to
become somewhat similar to that just described. In experimentation on
animals almost the only quantitative method is statistical. We do not
measure in the proper meaning of that term, but we count positive and
negative cases at least. In order to have cases which are comparable,
and therefore may be counted, we put our animals in situations where
their behavior is restricted to a few types of reaction. These we count
and the result of our experiments is given in frequencies enumerating
these few possible alternative reactions. I do not criticize the method
as such; I have used it myself. But we should never forget that, when
used exclusively, it may make our knowledge of behavior narrower. With
a given problem in mind we exclude a great many behavior possibilities
by our choice of experimental conditions. We do not see them any
more. Some psychologists will at least observe the remaining forms of
behavior during the experiment as an aid for the interpretation of
quantitative results; but others will not, because in their opinion
what is not quantitative cannot be scientific. In this case, what is
left of the material are abstract figures. Even then the procedure may
be valuable, if the experimenter is one of those who always find new
and productive problems for experimentation. For most of us, however,
conservatism may become the consequence of such an attitude, since,
admiring figures and curves, we keep away from the true source of new
ideas and new problems in our youthful science: i.e., from a broad
outlook upon our subject-matter.
Once more I venture to defend broad and qualitative information as a
necessary supplement to quantitative work. Otherwise behavioristic
psychology might become as sterile as supposedly it is exact. Too
great an interest in present quantitative methods and in their
standardization is not a hopeful response to the fact that the
development of psychology depends upon the discovery of _new_ methods
and problems in the future, and not upon the monotonous repetition
of a few standardized methods which fit only our primitive problems.
If the reader says that behavioristic psychology does not need
this advice, I reply that Watson has been criticized because his
well-known observations on children were not made in the sanctified
form of quantitatively controlled experiments. I do not think that
those observations give quite an adequate account of autochthonous
reactions and of learning in children; but, on the other hand, they
reveal more interesting facts than we might have learnt from columns
of abstract figures. An eminent psychologist was friendly enough to
say a few approving words about my own work on intelligent behavior in
apes. Still, he added, you have missed the main point by not applying
the statistical method. To my mind this means an inability to see
the problems which I had tried to treat in a preliminary way. Those
problems concerned the characteristic forms of intelligent behavior in
a special type of animal, as influenced by various situations. What
may be valuable in those observations would disappear in an abstract
statistical treatment of the “results.” Am I right then in warning
against a one-sided glorification of quantitative procedure _as such_?
Let us apply it, but not before we know by qualitative analysis where
to find good problems for quantitative research.
A certain broadmindedness and a clear appreciation of our present
predicament would be wholesome, then, in the application of
behavioristic methods. There is more in a scientific method than
observing, experimenting and measuring. In physics, at least, observing
and measuring usually result from definite questions, and these
questions are preliminary forms of hypotheses concerning unknown
properties of nature. Observed but enigmatic aspects of nature may
be deduced and explained from assumptions about unknown parts of it.
Taken as such, however, an assumption is not what the physicist wants.
Fortunately, a concrete assumption made in order to explain certain
observations will always lead to consequences other than those facts
for which it was made. These new consequences are to be tested in
further observation and experimentation. What is the hidden basis of
electrolytical conduction? Arrhenius makes the hypothesis of molecules
being dissociated into independent ions. By their charge conduction
is explained. But from their assumed independence definite conclusions
are to be drawn about the optical properties of the electrolyte. These
are tested in new experiments. Thus, productive method contains a bold
hypothesis as essentially as it requires adequate observation, and
the growth of physics is a constant oscillation from the first to the
second and vice versa.
In behavioristic psychology we cannot complain about a lack of
unknown parts of the organic systems we study to which to apply
productive hypotheses. We know a little about the elementary effects
of stimulation upon the sense organs of our subjects; we observe their
responsive behavior when it becomes overt. But between the two terms of
the sensori-motor circuit there is more _terra incognita_ than was on
the map of Africa sixty years ago. If behavior is to be understood as
depending upon inner organization as well as outer conditions, we must
try to imagine the modes and traits of these inner processes, which are
either started from without by environmental stimulation or aroused
intra-organically by inner dynamics. For those who know the history of
physics and wish to emulate it intelligently in psychology, this task
of finding productive assumptions about the hidden parts of behavior
will probably appear the most important of all. The whole future
development of psychology depends upon it. Here all the creative force
of behaviorism ought to be concentrated in a fine emulation of physics.
Behaviorism’s critical attitude toward introspection and direct
experience is an absolutely negative feature of the movement, at a
time when _positive_ ideas are needed. If I feel a little disappointed
by the work of behaviorism, the reason is not so much a certain
innocence in its treatment of direct experience and in its imitation
of adult physics, but its astounding sterility in the development of
productive concepts about functions underlying observable behavior.
As an imitation of physics it is scarcely a satisfactory achievement
for the behaviorist to have taken the old concept of reflex action
from physiology (including the reflexes of inner secretion) and to
give us no further comprehension into the formation of new individual
behavior than is offered by his concepts of positive and negative
“conditioning.” Why should behaviorism be so utterly negativistic in
its characteristic statements? “Thou shalt not acknowledge direct
experience in science” is the first commandment and “Thou shalt not
conceive of other functions but reflexes and conditioned reflexes” is
the second. I do not see how the behaviorist can reconcile this creed
with our actual knowledge of organic processes and of behavior. Nor
do I understand why, from the standpoint of “exactness,” the organism
should be conceived as such a crude and poor affair. In its effort to
imitate the technical procedure of physics, behaviorism often shows a
stubborn narrowness; the same attitude becomes quite striking in the
exclusion of all but two types of processes from its assumptions about
the inner side of behavior. What a strange conservatism and dogmatism
in a revolutionary school! Even now, as an adult science, physics
is allowed to have at least one new idea about the atom every year.
Though _our_ science is so very young, most behaviorists do not even
dream of any possible change in the nature of their two fundamental
ideas. The truth was revealed to them in its perfection at the birth of
behavioristic psychology.
The two behavioristic concepts in question show one remarkable
property: If you compare them with various types of processes in the
inorganic world of the physicist, you will find that even simple
physical systems are by far richer in the variety of their kinds of
function than is the nervous system of man in the eyes of a radical
behaviorist. A soap-bubble does not show us reflexes, it is true;
therefore we cannot expect to find conditioned reflexes in it.
Nevertheless those functional properties which the soap-bubble does
exhibit are decidedly superior in some respects to the monotony of
reflexes and conditioned reflexes. The same thing holds for innumerable
structures in the inorganic world. Sometimes their behavior will
at once remind us of the behavior of organisms, although they lack
reflexes and conditioned reflexes. However, by the standard of
exactness as the behaviorist understands it, it would be heresy
to follow this hint. His imitation of physics excludes most of the
functional possibilities given in physics itself. In organisms there
shall be nothing but reflexes. Some members of the school even are
beginning to protest cautiously against this strange situation.
It is not probable that an observer, looking upon human and animal
behavior without prejudices, would find reflexes and conditioned
reflexes as the most natural, or as the only, types of function by
which his observations might be explained. For those, however, who are
deeply convinced that the theory of original and acquired reflexes is
the whole truth about the nervous system, there is no real incentive
for the further observation of natural behavior, since they do not feel
the need of any new information or new functional concepts. On the
other hand, the restriction of observation to counting the frequency of
the few reactions retained in standard experiments will protect those
conservative concepts; and so one narrowness keeps the other alive.
But even with a more impartial interest in all the various forms of
behavior, how shall we find more productive concepts, if the gap
between the observable stimulating environment and observable overt
responses is as large as we find it in the present stage of physiology?
Of course we shall use whatever hint may be given by nerve physiology
and endocrinology. But even the more recent discoveries about the
all or none law and the metabolism in nervous activity do not give us
just that basis which we need for our purposes. In such a situation
anything will be valuable upon which we may base a working hypothesis.
In a critical predicament even a stone or a chair may become a weapon,
though the strategist would despise it in exact warfare. There is no
serious danger in developing a working hypothesis upon a somewhat
wavering ground, since, in empirical science, such an assumption
will be tested and continually corrected in further observation and
experimentation. Either it serves as a path toward new explorations,
its consequences being verified by success, or it is shown to lead us
the wrong way. In the first case no one will have scruples about the
previous legitimation of its birth; in the second, we shall discard it
as soon as observed facts are against it, or as soon as we find it bare
of real productive power.
* * * * *
The basis upon which I propose to build a working hypothesis has
much in its favor. In some respects it is just as solid as objective
observation is in behavioristic psychology, and in others, it is
certainly not insecure enough to raise serious doubts.
In an experiment in behavioristic psychology the “direct experience”
of the subject does not play any rôle. If he has such experience I as
experimenter do not assume that it has the slightest influence upon
the course of the physiological events. My assumptions about those
events must be able to explain thoroughly the overt behavior of the
physical organism of the subject. Direct experience is not a “force”
interfering with the chain of physical causation. (The opposite
hypothesis is made by dualists. But I do not think that it, on the
other hand, is productive, or that it illuminates the innermost nature
of things.)
I cannot exclude _my own_ direct experience in observing behavior,
as I cannot avoid dealing with direct experience even when employing
the most “indirect” methods of physics. Moreover in behavioristic
psychology I have to use objective experience in a great variety of
forms which no longer are material for observation in adult physics.
If I can rely upon objective experience for my statements about the
behavior of the subject I observe objectively in a behavioristic
experiment, why should I hesitate to utilize it to develop a working
hypothesis in the following manner?
Suppose I am used as a subject in a behavioristic experiment. With me
as a subject, behavioristic research would have the same difficulty in
developing a working hypothesis about the hidden processes underlying
my behavior. But in this case I can help to build a bridge over the
“large gap” mentioned above. During much of my own behavior I have
direct experience. Now there is no doubt that this experience depends
upon some of those processes about which we wish to formulate a
working hypothesis. It is more than probable that, under these
circumstances, I may use my direct experience in order to guess
about these processes. It is granted that direct experience does not
accompany _all_ the unobservable events which produce the behavior
observable from without; hence the hypothesis I am about to offer has
a limited application. But in our crucial situation we have not much
to choose. Confident as we are of further help from physiology, we are
so utterly in the dark at present that it would be pure folly not to
profit by any help that we can get for our work.
Let me declare that I do not propose to “introspect” in the technical
meaning of the word. The same kind of simple and naïve statements about
experience which I make as an observer of other people, of animals,
of instruments and so forth, shall now be used for another purpose.
Objective experience depends upon physical events outside my organism
and upon physiological events _in_ it concurrently. As depending upon
physical events _outside_ my organism, objective experience leads
to the construction of the surrounding physical world; as depending
upon physiological events _in_ my organism, it gives me hints about
these processes. _There is no reason at all why the construction of
physiological processes directly underlying experience should be
impossible, if experience allows us the construction of a physical
world outside, which is related to it much less intimately._
Obviously, I must have some leading principle in making my inferences
from given properties of direct experience to the properties of
concomitant physiological processes. In a more special form such a
principle was established by Hering many years ago. Its content is
the following: My various experiences can be ordered systematically
according to the sorts and degrees of similarity found between them,
exactly as animals are ordered in zoölogy and plants in botany.
The processes upon which these experiences depend are not known
directly; but if I knew their properties, I should be able to order
them as systematically according to similarities as I do order the
experiences actually. It is possible to imagine a great many different
relationships between the two systematic orders, that of experiences
and that of physiological processes. But I should have ever so much
difficulty in trying to relate definite experiences to definite
processes so long as I failed to assume one specific relationship
between the two orders, viz., that of _congruence or isomorphism in
their systematic properties_. This principle is sometimes formulated
more explicitly in a number of “psychophysical axioms.”[6] Here it will
suffice to give some examples of its application.
A sound of given qualitative properties is produced in various degrees
of experienced intensity or loudness. The systematic order of all these
different experienced intensities may be represented by a straight
line, which means that, following that order, we have the impression
of moving continually in the same direction. What may correspond to
loudnesses in the underlying processes? The principle decides that
whatever the concrete nature of these processes, the physiological
fundaments of all the experienced degrees of loudness must show the
same order as these show themselves, i.e., that of a straight line.
More especially: if a definite loudness is situated between two
other loudnesses in the systematic order of experiences, the process
corresponding to the first shall be between the processes corresponding
to the two others in the order of underlying physiological events. That
gives the congruence or isomorphism between the two systems, which I
mentioned above.
At the present time the all-or-none law does not allow us to choose
“intensity of nervous activity” as a physiological correlate
of experienced degrees of loudness. But our principle could be
equally well applied if frequency or density (number) of elementary
currents were taken as the correlate of loudness.
As another example, colors in their relation to underlying processes
have been considered so thoroughly by G. E. Müller,[7] that there is no
need of repeating his theory here. In some respects his assumptions
go far beyond the principle we are discussing here, insofar as he
draws conclusions about the _retinal_ processes, whereas the principle
_as such_ applies only to the processes underlying visual experience
directly. In his hypothesis it also happens that these processes
are chemical reactions. Such a transgression of the principle is
perfectly sound, however, and shows the productive force of it. If
there is congruence between the system of color experiences and that of
underlying processes, not all kinds of physical events will appear fit
to give a system of so much variety (of so many “dimensions”) as the
system of colors is found to possess. Chemical reactions may easily be
the only type of event which can be taken into consideration here. So
the principle of systematic congruence serves to restrict the number
of those physical processes which we may draw upon in developing the
concrete hypothesis about the “large gap.”
On the other hand, we have begun to apply the principle itself in a
much more general, and at the same time much more concrete, form than
either Hering or Müller would have done. As used by these authors it is
based upon the _logical_ order in which we arrange certain experiences
after abstracting them from their context and judging about their
similarities. It attributes a congruent logical order to the underlying
processes, similarly abstracted from their dynamic context like dead
specimens in a museum. But what about the real concrete order of
experiences, which itself is experienced? At this moment I have before
me three white points on a black surface, one in the middle of the
field and the others in symmetrical positions on both sides of the
center one. This is an order again; not an abstract, logical order,
however, but a concrete, experienced order. As experienced it depends
upon physiological processes in my organism, so that some feature of
a physiological function must correspond to it. And applied to this
concrete order our principle says that correlated with the visually
experienced symmetry there is an homologous symmetry of dynamic context
in the underlying processes. Or, in the same example: one point is seen
_between_ the others, this relation being experienced exactly as the
white of the point is. Again in the underlying processes there must
be something functional which corresponds to that “between.” Applied
to the _concrete_ order our principle claims that the experienced
“between” is accompanied by a functional “between” in the concrete
dynamic context of concurrent physiological events. If the principle
is applied strictly in this manner to _all_ cases of spatial order in
experience, it will lead to the general statement that _all experienced
order in space is a true representation of a corresponding order in the
underlying dynamical context of physiological processes_.
This is a bold hypothesis. It has far-reaching consequences,
if it is taken seriously enough. Its significance will become
clearer in the following chapters where we shall also discuss some
misinterpretations which may easily arise. For the present I will
offer another application of the same principle. The temporal position
of one experience as “between” two others is frequently experienced,
in the same way that spatial “betweenness” is. Now as far as it is
experienced, time must have a functional correlate in underlying
physiological processes no less than experienced space. Again, then,
the temporal “between” is correlated with a functional “between” in
the concrete dynamic context of physiological events. And if, in this
manner, we apply the principle throughout, we will arrive at the
proposition that _experienced order in time is a true representation of
the corresponding concrete order in the underlying dynamical context_.