Learning and Memory 55coherent theory. Important influences of Gestalt notions can
be seen in Hebb’s theory. He is a connectionist but in a mod-
ern sense: Connections must underlie brain organization but
there is no need for them to be in series.
One concept in Hebb’s book has come to loom large (too
large perhaps) in modern cognitive-computational neuro-
science—the Hebb synapse:
When an axon of Cell A is near enough to excite a cell B and re-
peatedly or persistently takes part in firing it, some growth
process or metabolic change takes place in one or both cells such
that A’s efficiency, as one of the cells firing B, is increased.
(1949, p. 62)Lashley’s pessimistic conclusions in his 1929 monograph
put a real but temporary damper on the field concerned with
brain substrates of memory. But other major traditions were
developing. Perhaps the most important of these was the in-
fluence of Pavlov. His writings were not readily available to
Western scientists, particularly Americans, until the publica-
tion of the English translation of his monumental work Con-
ditioned Reflexesin 1927. It is probably fair to say this is the
most important single book ever published in the field of be-
havioral neuroscience. Pavlov developed a vast and coherent
body of empirical results characterizing the phenomena of
conditioned responses, what he termed “psychic reflexes.”
He argued that the mind could be fully understood by analy-
sis of the higher order learned reflexes and their brain sub-
strates. As an example of his influence, Clark Hull, in his
Principles of Behavior(1943), wrote as though he were a
student of Pavlov.
W. Horsley Gantt, an American physician, worked with
Pavlov for several years and then established a Pavlovian
laboratory at Johns Hopkins. He trained several young psy-
chologists, including Roger Loucks and Wulf Brogden, who
became very influential in the field. Perhaps the most impor-
tant modern behavioral analyses of Pavlovian conditioning
are the works of Robert Rescorla and Allan Wagner (1972).
Although Pavlov worked with salivary secretion, most
studies of classical conditioning in the West tended to utilize
skeletal muscle response, à la Bechterev. Particularly pro-
ductive have been Pavlovian conditioning of discrete
skeletal reflexes (e.g., the eyeblink response), characterized
behaviorally by Isadore Gormezano and Allan Wagner and
analyzed neuronally by Richard Thompson and his many stu-
dents, showing localization of the basic memory trace to the
cerebellum (Thompson, 1986). Masao Ito and associates in
Tokyo had discovered the phenomenon of long-term depres-
sion (LTD) in the cerebellar cortex (see Ito, 1984). Repeated
conjunctive stimulation of the two major inputs to the
cerebellum, mossy-parallel fibers and climbing fibers, yields
a long-lasting decrease in the excitability of parallel fibers—
Purkinje neuron synapses. Ito developed considerable evi-
dence that this cerebellar process underlies plasticity of the
vestibular-ocular reflex. Thompson and associates developed
evidence, particularly using genetically altered mice, that
cerebellar cortical LTD is one of the mechanisms underly-
ing classical conditioning of eyeblink and other discrete
responses.
Fear conditioning was characterized behaviorally by Neal
Miller and analyzed neuronally by several groups, particu-
larly Michael Davis (1992), Joseph LeDoux (2000), and
Michael Fanselow (1994), and their many students. They
showed that at least for classical conditioning of fear, the es-
sential structure is the amygdala, which may contain the basic
memory trace for this form of learning (but see just below).
The process of LTP may serve to code the amygdalar fear
memory.
Duncan’s discovery in 1949 of the effects of electrocon-
vulsive shock on retention of simple habits in the rat began
the modern field of memory consolidation. Hebb and Gerard
were quick to point out the implication of two memory
processes, one transient and fragile and the other more per-
manent and impervious. James McGaugh and his associates
(1989) have done the classic work on the psychobiology of
memory consolidation. He and his colleagues demonstrated
memory facilitation with drugs and showed that these effects
were direct and not due to possible reinforcement effects of
the drugs (and similarly for ECS impairment).
The amygdala is critical for instrumental learning of fear.
McGaugh and his associates demonstrated that for both pas-
sive and active avoidance learning (animals must either not
respond, or respond quickly, to avoid shock) amygdala le-
sions made immediately after training abolished the learned
fear. Surprisingly, if these same lesions were made a week
after training, learned fear was not abolished, consistent with
a process of consolidation (see McGaugh, 2000). The appar-
ent difference in the role of the amygdala in classical and in-
strumental learning of fear is a major area of research today.
Chemical approaches to learning and memory are recent.
The possibility that protein molecules and RNA might serve
to code memory was suggested some years ago by pioneers
such as Gerard and Halstead. The RNA hypothesis was taken
up by Hyden and associates in Sweden and by several groups
in America. An unfortunate by-product of this approach was
the “transfer of memory” by RNA. These experiments, done
by investigators who shall remain nameless, in the end could
not be replicated.
At the same time, several very productive lines of investi-
gation of neurochemical and neuroanatomical substrates of