Dedication
Preface
Dedication The Quad-L Project is
dedicated to
my two mentors Kenneth W. Spence & Neal E. Miller
Preface Like the proverbial old soldier
who "never dies but just slowly fades away", the domain of Animal Learning
and Motivation
(AL&M) has drifted from its earlier role as a central figure to
its current background role in the mosaic science of psychology.
Once-controversial names such as Guthrie, Hull, Spence,
and Tolman have become historical figures and classic issues such
as awareness, contiguity, continuity, and emergentism have been
placed in new contexts. Where it was simply taken for granted
that learning is the essential process distinguishing organisms
from other systems in the universe, with motivation being the
essential process that both guides and energizes behavior, the
terms "learning" and "motivation" are rarely used today. The waning of recognition has not been without protestations
to the contrary. In 1972, I chose the title, "The Psychology of
Animal Learning and Now," to focus on some of the many important
practical implications of basic research (Logan, 1972). More
recently Domjan (1987), Rescorla (1987, 1988), and Skinner (1987)
have pronounced the domain of AL&M as being not just alive and
well, but relevant in contemporary Psychology. Despite the
cogency of these statements, the displacement of AL&M has been
inexorable. My own belief is that a major reason for AL&M falling from
favor is internal to the domain. I think that the development of
a science follows a Piagetian-like course. A body of knowledge
gives rise to a conceptual structure (theory/system) that serves
to organize and, in some sense, "explain" the observations. New
knowledge is assimilated into this structure, sometimes undergoing
some distortions in the process. Totally incompatible facts are
dismissed until the need for a different conceptual structure can
no longer be denied. This new structure must accomodate all
available knowledge. Thereupon, the development process is
repeated. This view is similar to, but importantly different from the
notion of scientific "revolutions" (Kuhn, 1970). At least as I
understand it, a change in "paradigm" renders the pre-existing
knowledge base obsolete and no longer of contemporary interest.
Of course, anyone can arbitrarily identify the variables of interest
to her or him. But so long as some psychologists retain an interest
in behavior, AL&M is an invaluable source of information. The
problem is that AL&M is also an overwhelming source of
information. The reason that AL&M is responsible for its own decline is that
it was caught up in the information explosion that followed World
War II. New findings were hard to disseminate, much less
assimilate. Had AL&M been able to maintain an integrated image
with a conceptual core, it might have held on to its status.
But the domain itself disintegrated into pockets of research activity and
miniature models with limited generality. The grand theories of
learning could not evolve fast enough. The grand theories were based on a one-directional, completely
deterministic model: the goal was to specify the independent
variables of which the dependent variables were functions. This
model may be appropriate for classical conditioning because the
events are presented according to a schedule without regard to the
organism's behavior, but in all of the other learning paradigms,
the variables are mutually interdependent. Thus, apparent dilemmas
such as the "avoidance paradox" arose because of the failure to
distinguish the omission of an independent event (classical
extinction) from the omission of an interdependent event
(instrumental avoidance). It would be foolish to think that one could somehow patch up
this and other deficiences and thereby return AL&M to a place of
revered dominance in the field. At the empirical level, Tolman
(1938) was over-zealous when he wrote that "everything of importance
in psychology can be understood by studying the behavior of a rat
at a choice point," and Hull (1952) was overly optimistic when he
wrote that "we are even now well on the way to a general theory of
social behavior." AL&M may be alive and well but it is not likely
to live up to such promises. Nevertheless, I believe that the database generated by the
unbridled enthusiasm for AL&M remains an important resource of
information, and the conceptualizations remain a source of insight
toward a general theory. The challenge is to increase the
accessability of this information by organizing it in some systematic
fashion. Otherwise, the fate of countless hours of thought and
research effort by many dedicated scientists will be a "dusty death"
on the shelves of university libraries. The Quad-L was conceived
in the hope of forming one link between the archives and
contemporary thought. Introduction
It is generally contended that science is public and
cumulative.
We can certainly agree that private knowledge is not a part of science
and we can probably agree that "good" scientific research is based, at
least in part, on the extant publicly available knowledge. But the
more general dicta are not fully realized in the Experimental
Psychology of Learning and Motivation. Scientific knowledge of Psychology is public in the literal
sense of having been published in archival books and journals that are
stored in university libraries. But in view of the burgeoning
volume
of the relevant literature, accessing those archives has become
extremely tedious. Were the discipline truly cumulative, this
problem
would not exist. By "truly" cumulative, I mean that each new
addition
to the fundamental store of knowledge is integrated within the already
existing framework, thus building a solid foundation for the expanding
pyramid of information held together by increasingly abstract conceptual
analyses. Given such a state of affairs, mastery of the
contemporary
systems would automatically subsume by implication the detailed knowledge
in the area and there would be little need, except as an historical
exercise, to study the earlier publications from which the present
systems developed. I do not perceive Psychology as being cumulative in the above
sense,
nor public in the sense of being widely known by most psychologists.
Some people (e.g., Finkelman, 1978) have argued that our failure to
achieve these desiderata belie out status as a science.
Alternatively,
the empirical data base of the field may simply have grown too fast,
with a large number of people generating such an enormous wealth of
knowledge that traditional means of processing information have been
inadequate. The Quad-L (Logan's Literature and Laws of Learning) project
provides a method of summarizing scientific information in a format
that might facilitate the kind of insightful reorganization that will
lead to a cumulation of knowledge in the ideal sense. Of course,
this
is only a hope, but I do have a conviction: True integration can only
be achieved by many skills under one skull, not under many skulls.
An extraordinary breadth of knowledge must be assimilated by any
person who would aspire to organize our field into a true
discipline.
The Quad-L at least outlines the dimensions of the task. The QUAD-L Literature: The Quad-L project has three basic objectives. One of these
is
to compile a reasonably complete bibliography of the Experimental
Psychology of Learning and Motivation within the domain bounded by
classical conditioning at the one end and spatial learning at the
other.
The domain could be described as "animal learning and motivation," if
one recognizes that humans are animals and restricts the coverage to
learning paradigms that are applicable to all animals.
Specifically,
uniquely human contexts, such as verbal paired associates learning,
concept formation, and problem solving, are currently excluded from
consideration. The domain of interest is also the one that is
historically identified with the grand theories of learning such as
those of Guthrie, Hull, and Tolman and the atheoretical approach of
Skinner. Database: The second objective is to associate with each of these references
one or more one-line statements that provide a "root-abstract" of that
publication. Ideally, the "power" of the "root" will be quite high,
so
that the resulting root-abstract will capture the essential substance
of an empirical finding or the critical essence of a conceptual
proposition. Root-abstracts are directed at a reasonably
knowledgeable
audience in the sense of presuming familiarity with the basic procedures
and conceptualizations pertaining to these areas. Laws: The third objective is to code the root-abstracts in such a way
that
they can be sorted into subsets corresponding to classes of experimental
operations and conceptual issues. All of the information is stored
in a
computer in such a way that not only can any errors of omission or
commission be corrected, but new information can be added as it becomes
available. Boundary Conditions A very critical feature of any system in science is the
specification
of its boundary conditions. Because the conceptual domain of
science is
indefinitely large, and because the conditions under which scientific
principles may operate are indeed infinite, a manageable systematic
analysis necessarily pertains only to some limited range. At least
in
principle, these limitations should be explicit and definite. Most
importantly, the boundary conditions should be stated in terms of the
permissible experimental operations and the allowable conditions of
those operations, rather that in terms of outcomes. Although these
points are obvious in the abstract, in actual practice, they are not
infrequently ignored. It should be clear that, for an empirical system of the present
kind, the boundary conditions are almost entirely matters of
convenience.
That is to say, the domain of "Animal Learning and Motivation" as
previously described has fuzzy boundaries and the ideal of making these
decisions in a completely atheoretical manner has undoubtedly not been
achieved. My ultimate purpose in this endeavor should not be
hidden;
I believe that a reductive theory involving the introduction of
hypothetical constructs will prove to be not only useful, but necessary
to make the myriads of empirical phenomena comprehensible. Because
I
am personally committed to such an orientation, there undoubtedly are
instances where I have included material that fits with my theoretical
hunches and excluded material because of theoretical intuitions. I
know
of no way to guarantee objectivity in these regards. But there is
this
consolation: Others are free to add new information into the system and
thereby set the challenge to develop a theory that will also encompass
those phenomena. Accordingly, this is at best a first approximation
to
the ideal systematic empirical analysis. In order to make the domain finite, I have a temporal limitation
as one of the boundary conditions: only publications between 1900 and
2000 are included. Even so, there are many relevant publications
that
remain to be added to the system, and a great many root abstracts
remain to be written. Furthermore, the aforementioned coding
system
has not yet been fully implemented in the materials. I hope that
anyone
viewing the system at this time will be an optimist and be encouraged
that the glass is now almost half full, rather than focusing on
the obvious fact that it is still at least half empty! Postscript: The enterprise that the Quad-l attempts to identify,
summarize, and systematize is based on several working hypotheses. One
of these is determinism, the contention that the behavior of organisms
is a part of nature and, as such, is subject to natural laws. A
corrolary of this hypothesis is that these laws can be inferred from
the systematic experimental and conceptual analysis of the methods of
natural science. A second hypothesis is evolution, the contention that all extant
organisms have developed in relation to their ecological niche and are
to some extent unique. A corrolary of this hypothesis is that there
are commonalities at some level of abstraction among the laws of the
behavior of different animal species. A further hypothesis is that at least the human organism is
capable of covert behavior such as feeling, thinking, and reasoning.
A corrolary of this hypothesis is that such behavior obeys the same
basic laws as those of overt behavior. These are called "working" hypotheses because there is no way
to test them directly; one can only assume them to be true and proceed
with the awareness that they may not apply in all situations. As a
possible example, consider the fact that I personally believe that
there must be, in some sense, a Higher Power that gives, again in
some sense, meaning and purpose to existence. Because of my cultural
heritage, I call this Higher Power, God, but the corresponding term
in most of the world's religions could equally be used. However,
my God is not, to the best of my knowledge, the same as the Higher
Power of any of the organized religions, differing for example from
most Christian faiths because my God would never intervene in such
a way as to violate natural laws. This means that a so-called
miracle will either be understood by science or is an illusion,
an exaggeration, or a fantasy. My purpose is not to engage in a religious debate. Rather, I want
simply to explore the basis for such a belief. More particularly,
because religious beliefs must be acquired, one can ask whether the
principles implied by the Quad-L suffice to account for one's
belief (or non-belief) in a Higher Power. I called this a "postscript" because I had earlier identified
the domain of the Quad-L as "basic learning processes" and hence
not including (presumably) uniquely human behavior. Hence, this
would be an exercise in extending the domain beyond its initial
boundaries. Perhaps prematurely. After reviewing the domain identified in the Index, I think
the most likely relevant basic principle of learning is imitation
(observational learning.) This is reasonable because, as a result
of aperiodic attendance at Sunday School as a youth and Church as
an adult, I have certainly been exposed to a number of people who
profess a similar belief. Perhaps one of them impressed me
sufficiently that I have attempted to imitate his or her faith. There are two difficulties with this interpretation of
the origin of my belief in an eternal, omnipotent God. First,
imitation is only possible when the requisite behaviors are
already in one's repertoire. Watching someone else do something
(e.g., wiggle an ear) does not bestow the capability to perform
that act. Hence, one must still account in some other way for
the initial acquisition of a religious belief. Secondly, beliefs are, by their very nature, unobservable
to others. There is thus no way that I could have imitated the
actual inner value system of someone else. To be sure, one can
sing religious songs, read religious passages, or repeat religious
prayers, but doing these things does not require a true belief.
Accordingly, although the seed my belief may have been planted
through observation, it must have been acquired through some
other means. Similar problems arise with other interpretations based
on basic principles of learning. For example, for beliefs to be
treated as an instance of classical conditioning, one would have
to identify some stimulus event that naturally elicits a belief
in a Higher Power. Although it is said that there is no such person
as an atheist in a foxhole, I cannot profess to have had a true
relevation. And if a belief is treated as an operant, God cannot
reinforce a belief unless such a "response" is emitted. Others may ultimately acccount for religious beliefs in
scientific terms, but I have to entertain the possiblity that some
beliefs are the result of a true "free" choice that is uniquely
human and that is not mimicked in any overt behavior. Hence, they
may violate all of the aforementioned working hypotheses. If so,
the boundary conditions of the Quad-L are further restricted.(/P> Actually, the perceptive person will note that almost all of
the references in the Quad-L as currently constructed concern
associative lerning. There is very little attention paid to "pure"
perceptual learning or "pure" motor learning. Associative learning
simply presumes that both perception and performance are already
available to the organism. There is much to be done if science is
to encompass all forms of human behavior. Consider the conundrum posed by the attempt to prove the
logical impossibility of a science of human behavior by contending
that the behavior of an anti-predictably-motivated person could not
be predicted by science. Because science is public, such a person
always knows the scientific prediction and could foil any system.
Part of the resolution of this conundrum is to recognize first
that the person must know that his or her behavior is being
predicted and then must value proving the point more than
that associated with conforming to the scientific prediction.
Scientific knowledge is limited and does not presume to deal
with unanswerable questions. December 31, 2000
who convinced me that the better we understand non-human Animal
Learning and
Motivation, the better we can understand human behavior.