Brain Research, 175 (1979) 233-245
233
Elsevier/North-Holland Biomedical Press
SYNAPSES AS ASSOCIATIVE M E M O R Y ELEMENTS IN THE
HIPPOCAMPAL F O R M A T I O N
WILLIAM B. LEVY* and OSWALD STEWARD
Departments of Neurosttrgery and Physiology, University of Virginia School of Medicflte, Charlottesville,
Va. 22908 (U.S.A.)
(Accepted February 15th, 1979)
SUMMARY
This report analyzes long term potentiation (LTP) and associative interactions
between synapses of the ipsilateral and crossed entorhinal cortical (EC) pathways to
the dentate gyrus (DG). In the anesthetized rat, conditioning stimulation to one ECDG pathway reliably elicits LTP at the ipsilateral synapses, while the synapses of the
collateral, crossed pathway to the contralateral DG do not exhibit LTP. Furthermore,
in the DG ipsilateral to the conditioning stimulation the convergent crossed pathway
from the contralateral side, which had not been itself conditioned, failed to exhibit
heterosynaptic LTP. These results are consistent with a specific 'synaptic' localization
of the changes responsible for LTP, and suggest that some critical number of synapses
must be activated in order to observe LTP.
While the crossed EC-DG projection never exhibited LTP when conditioned
alone, the crossed input could be potentiated under certain circumstances. Specifically, paired conditioning of ipsi- and contralateral inputs by nearly simultaneous
conditioning stimulation of the EC bilaterally results in LTP in the crossed system.
Furthermore, this associatively induced LTP of the crossed system can be reversed by
subsequent conditioning of the ipsilateral system alone. Successive potentiating and
depotentiating sequences are possible using paired and non-paired stimulation
procedures even after lesions which prevent neural loops through the EC. The results
are interpreted as evidence for a 'Hebb' type synapse which has the capability for
erasure. This synaptic type is not appropriate for classical conditioning without
appendant circuitry, but is suited for other forms of associative learning.
* On leave of absence from the Department of Psychology, University of California at Riverside,
California, U.S.A.
234
INTRODUCTION
Some thirty years ago, Hebb 4 proposed that synapses might serve as a substrate
for storing associative information if synaptic efficacy could increase when synapses
were co-active with a postsynaptic cell. Synapses that increase and decrease their
efficacy as a function of convergent co-activation with a postsynaptic cell are required
by several neural models. These models can explain various neuropsychological
processes, including associative learning and recall utilizing non-localized, destruction-resistant engrams 5, the learning of complex motor response patterns s, and the
establishment of normal neuronal response properties during development, such as
occurs in the development of binocularity in the visual cortex 16.
Recently, a population of synapses with characteristics similar to those proposed
by Hebb have been described in the hippocampal formation. The synapses are those
formed by the efferents from the entorhinal cortex (EC) to the dentate gyrus (DG) and
are well known for their ability to exhibit long term potentiation (LTP) of synaptic
efficacy. A recent report adduces that the EC-DG synapses undergo LTP as a function
of co-activity of some critical number of pre-synaptic elements 1° i.e. the associative
property implied by Hebb. The present report further delineates the associative
properties of these synapses. Most notable is the fact that the EC-DG synapses also
exhibit an additional capability of reversing the potentiation when certain subpopulations of synapses remain inactive while other converging elements receive
conditioning stimulation.
These present studies were done utilizing the massive ipsilateral and sparse
crossed projections from the EC to the DG. This pathway offers considerable
advantages as a model system since the minor crossed pathway probably arises as a
collateral of the ipsilateral projection system (see refs. 3, 19), the ipsilateral and crossed
projections (of bilateral origin) terminate in overlapping laminae in the dentate gyrus 3,
and they probably converge upon single granule cells.
The widely spaced origin of ipsilateral and crossed EC projections to a single
DG, coupled with the probable convergence of the two inputs upon single granule
cells, provides unique controls for polysynaptic, heterosynaptic, or non-synaptic
effects that might accompany or be responsible for LTP. Using independent stimulation of the ipsilateral and crossed pathways to the DG, this paper demonstrates that
LTP depends on associative synaptic interactions and can be reversed by nonassociation of synaptic activity. The properties of the EC projection permit a logical
analysis of the site and nature of these associative interactions.
MATERIALS AND METHODS
Experiments were performed on 200-400 g male Sprague-Dawley derived rats
which were obtained from Flow Labs. The animals were anesthetized with chloralose/
urethane (55 mg/kg and 0.2-0.5 g/kg respectively) and received supplemental doses of
urethane during the experiment to maintain anesthesia. In one case a midbrain
transection was performed and no supplemental anesthesia was administered.
235
The electrophysiological methods were comparable to those which have been
previously described18A 9. Evoked responses were recorded via glass micropipettes
filled with 4 M NaC1 and were FM tape-recorded for off line analysis by a Nicolet
Model 1072 transient averager.
Experimental methods
In all experiments, bipolar stainless steel stimulating electrodes were positioned
bilaterally in the angular bundle (8.1 mm posterior to bregma, and 4.4 mm lateral to
the midline 11. The depth of the stimulating electrode was based on the responses
recorded in the dentate gyrus. Stimulus intensity was adjusted in most cases to evoke
maximal responses in the crossed pathway (vide infra) with the minimal stimulation
current. At least one recording electrode was positioned in the ventral leaf of the
dentate gyrus and, in some additional cases, recording electrodes were positioned
bilaterally. Control (and later test) responses to angular bundle stimulation were
established by alternating pulses between the electrodes once every 10 see (i.e. each EC
was stimulated 1/20 see). To maintain a steady state, test stimulation continued
throughout the entire experiment whether or not data were being recorded. The
'conditioning' stimulation consisted of 8 trains (1 train/10 sec) of 8-10 pulses delivered
at a rate of 400 Hz (ref. 2). In some phases of an experiment only a single stimulating
electrode was used to deliver conditioning trains. For the associative conditioning the
stimulator was set so that times of onset of the ipsi- and contralateral synaptic
responses to single pulses occurred approximately simultaneously on one side of the
dentate gyrus (this typically means delaying one train with respect to the other by
approximately 1-2 msec). The effect of conditioning was evaluated by comparing
averages of 4 responses before and after conditioning stimulation, and long term
potentiation (LTP) was defined as a response increase which persisted for at least 15
min.
Experimental rationale
From one set of experimental procedures there come data which address two
distinguishable questions. One question concerns the existence of associative conditioning while the other question concerns the localization and specificity of LTP.
Testing an unconditioned system that converges with the conditioned pathway (i.e. the
crossed system originating on the other side and converging with the ipsilateral
conditioned system) serves as a control for non-specific intra- and extracellular
alterations. Testing the other crossed system (that is, the one unavoidably conditioned
with the ipsilateral system) controls for non-specific changes that might generalize
along an axon. Whether or not these tests serve to demonstrate the specific 'synaptic'
nature of LTP depends on the correctness of the assumptions that the crossed
projections (1) arise as collaterals of the ipsilateral EC-DG afferents and (2) converge
upon granule ceils which are also innervated by the ipsilateral projection system of the
other side. Previous studies utilizing double retrograde labeling methods have
provided strong support for the first assumption, since in animals in which the crossed
projection has been induced to sprout by ipsilateral entorhinal lesions, the crossed
236
projections are collaterals of ipsilateral projections 17. Furthermore, extensive overlap
of ipsilateral and crossed terminal fields in the normal animal suggests that the crossed
projection terminates upon the same granule cell population as the ipsilateral
projection system, and not on some unique population of granule cells.
Experimental protocol
Since there were two stimulating electrodes and (sometimes) two recording
electrodes, several permutations were possible regarding the relative location of the
conditioning and test electrodes. Arbitrarily one stimulating electrode was designated
S1 and the ipsilateral recording electrode R1. The corresponding contralateral
electrodes were designated $2 and R2 (see Fig. I). Control experiments, intended to
demonstrate the specific synaptic nature of LTP and to serve as one control for the
succeeding associative conditioning, required that one side alone be conditioned (i.e.
condition SI). The effects of conditioning SI can then be analyzed in the ipsilateral
pathway of the conditioned side (S 1 -+ R 1), in the crossed pathway of the conditioned
side (S 1 ~ R2), and in the unconditioned, crossed system from the opposite side ($2 -+
R1) which converges with the conditioned ipsilateral path. Notice that two of the tests
(experimental SI ~ RI and control St -+ R2) took place at the same time, used the
same stimulating electrode, but used a different recording electrode. The other control
(test $2 -+ RI) used a different stimulating electrode but used the same recording
electrode as the experimental pathway of the ipsilateral potentiation experiment (S 1
R1). After beginning each animal with these procedures, the experimental pathway of
primary interest shifted from the ipsilateral system S I + R 1 to the crossed system S1
R2. The effect of conditioning $2 on the S 1 -~ R2 response was first analyzed. Paired
conditioning trains (SI + $2) already described above were then applied. This
associative conditioning and subsequent tests were followed by procedures similar to
the initial controls.
In the illustrations, a single graph plots only one test response on the ordinate
but that same graph will indicate the time at which a conditioning train was applied.
Since the conditioning electrode need not be the same as the test electrode, the relevant
conditioning electrode is indicated along the time axis while the test pathway is
indicated to the left of the graph.
Histology
At the termination of the experiment, the animals were sacrificed with an
overdose of sodium pentobarbital (Nembutal) and were perfused transcardially with
10 ~ formalin in saline. The brains were divided midway along the rostrocaudal axis of
the hippocampal formation, and the anterior portions of the brains were sectioned
coronally (to define the position of the recording electrodes), while the posterior portions
of the brains were sectioned horizontally (to define the position of the stimulating
electrodes).
RESULTS
The responses evoked in the dentate gyrus (DG) by entorhinal cortical (EC)
237
stimulation have previously been described 7,19. The ipsilateral pathway evokes
a large amplitude wave which is negative in the dendritic regions of the granule
cells, and positive at the level of the granule cell bodies 7. This response is interpreted as
the extracellular reflection of the EPSP generated in the granule cell population, and
will be termed here the population EPSP in accordance with previous terminology 7. At
sufficiently high stimulus intensities, the population EPSP is interrupted by a spikelike potential which is negative at the cell body layer and positive in the dendritic
regions. This response reflects the summed discharge of the granule cells 7, and is
termed the population spikeL Consistent with its relatively sparse nature, the crossed
projection evokes considerably smaller population EPSPs than the ipsilateral projection system (typically being between 0.5 and 1.0 mV in amplitude), and in normal
animals these population EPSPs are not accompanied by population spikes. Sometimes a waveform which did not change in size or polarity with movement of the
recording electrode was detected concurrently with the initial phase of the crossed
response. Because of its timing, it was assumed that the response was a volume
conducted artifact, perhaps from the other hippocampus.
The population EPSPs were quantified by measuring the initial slope of the
response evoked by ipsilateral EC stimulation (see ref. 7 for a rationalization of the
slope measurements), and the onset to peak amplitude of the response evoked by
contralateral EC stimulation. The peak of the response evoked by the crossed pathway
was defined as the first positive inflection of the negative going population EPSP, to
exclude later polysynaptic responses 7. As demonstrated in the accompanying manuscript la the initial slope and amplitude measures of the population EPSP both yield
approximately comparable estimates of LTP. Peak measurements of the crossed
responses were utilized rather than the initial slope because of the volume conducted
artifacts mentioned above. When the response evoked by the crossed projection was
not obviously contaminated by volume conducted artifacts which did not reverse with
recording electrode movement (see above), slope measurements yielded quite comparable estimates of the potentiation. The results described below were replicated in 6
animals. However, the responses from one animal with bilateral stimulating and recording electrodes were chosen to illustrate the phenomena (for a schematic drawing
of the stimulating and recording arrangements, see Fig. 1).
Effects of unilateral conditioning on the pathways activated by the conditioning
stimulation
As demonstrated by previous studies, conditioning of one EC (S1, Fig. 1) results
in potentiation of the population EPSP and the population spike evoked in the
ipsilateral dentate gyrus (test SI-+R1, uppermost graph of Fig. 1), and such potentiation was consistently observed in the present study providing that the stimulus intensity
was sufficiently high. As an approximate measure, potentiation was observed at
stimulus intensities sufficient to evoke population spikes, although potentiation was
also observed in a few cases in which the conditioning stimulation was delivered at
intensities which failed to evoke a measurable population spike (see ref. 19, Fig. 5).
However, the simultaneous responses evoked by the crossed projections to the opposite
238
test S, -~lt,
-R,
180.
EC
J \+
test S,~R~
EC
EC
J
~ ~ ~
~
t
~1lo
S,"
a
I~e
0
I0
Minutes
20
30
EC
Pre
0
I0
Minutes
20
30
EC
Pre 0
I0
Minutes
20
30
-R2
test S2"~1:11
EC
239
h i p p o c a m p u s (S 1-+R2, Fig. 1) failed to exhibit p o t e n t i a t i o n . P o t e n t i a t i o n o f the r e s p o n ses e v o k e d b y the crossed p a t h w a y was never o b s e r v e d in a n y o f the a n i m a l s a n a l y z e d
(see also ref. 19). Since the c o n t r a l a t e r a l l y p r o j e c t i n g fibers p r o b a b l y arise as collateral
b r a n c h e s o f the ipsilateral p r o j e c t i o n system, the selective p o t e n t i a t i o n o f the
ipsilateral p a t h w a y reinforces previous conclusions t h a t the p o t e n t i a t i o n does n o t
result f r o m a n y excitatory change in the E C cells o f origin, o r any change which results
f r o m increases in efficacy at the site o f the s t i m u l a t i n g electrode.
Effects of unilateral conditioning on a pathway not activated by the conditioning
stimulation
T h e b i l a t e r a l n a t u r e o f the E C p r o j e c t i o n system permits an analysis o f possible
general changes in the p o s t s y n a p t i c cell p o p u l a t i o n as a consequence o f conditioning.
A s illustrated in Fig. 1, p o t e n t i a t i o n o f the ipsilateral p r o j e c t i o n to one side (SI-+R1)
o c c u r r e d w i t h o u t increases in the responses o f the converging crossed p a t h w a y e v o k e d in
t h e same dentate gyrus b y the c o n t r a l a t e r a l E C s t i m u l a t i o n ( S 2 ~ R 1 ) . Indeed, in this a n d
several o t h e r animals, there was actually a decrease ( d o w n to 8 7 i 2 . 5 ~ (/z i S.E.),
n = 4 ) in the a m p l i t u d e o f the responses e v o k e d by the crossed p r o j e c t i o n to the d e n t a t e
gyrus (i.e. S2-+R1) at 15 min after the converging ipsilateral p a t h w a y h a d been p o t e n t iated.
Associative conditioning
The report of McNaughton et al. lo, and our own unpublished observations that
LTP is dependent upon the intensity of stimulation, in combination with the present results in which the robust ipsilateral input exhibited potentiation, while the smaller responses of the contralateral projection did not, suggest that LTP might be dependent
upon the number of converging co-active synapses. This suggestion is given further
credence by the observation that if the crossed projection increases its input to the dentate as a consequence of post-lesion sprouting, it is possible to induce LTPl% The suggestion that LTP depends on the number of co-active synapses during conditioning leads to
a prediction. Specifically, LTP might be inducible in the crossed pathway if it were
Fig. 1. Effect of ipsilateral entorhinal conditioning stimulation (S1) on test pulses to the same ipsilateral (SI -+R1) and on both contralateral responses (SI -+R2 and S2-+RI). The top two graphs plot the
ipsilateral and contralateral responses, respectively, in response to a test pulse delivered at stimulation
site S1. The bottom graph plots the contralateral response to a test pulse delivered at site $2. Recording (R1 and R2) and stimulating (S1 and $2) sites are illustrated in the schematic figure to the left of
each graph. For this animal the subscript 1 refers to the left side of the brain while the subscript 2
refers to the right side of the brain. The response being measured is indicated by the darkened pathway
in the figure and by the test S-+R designation and associated traces. An ascendant arrow in the graph
indicates the time at which a conditioning series (8, 25 msec trains, one train/10 sec) was given. The
location of the conditioning electrode is indicated directly below this arrow. Each descending arrow
and associated lower case letter in the graph indicates the time of the correspondingly labeled response
trace. Test pulses of constant intensity are delivered alternately between S 1 and $2 so that each stimulating electrode is activated once every 20 sec and 10 sec after the other stimulating electrode. Each
point is the average of 4 responses. The control response is determined by the average response value
during the time preceding 0 rain. For the SI-+R1 response, slope measurements are used. For the
SI-+R2 and S2~R1 the initial negative peak was measured.
240
test St-*R2
260-
~/~=
C
v
O'SmV+I
W
~;S,
o
o
"~ 140o~
a
/
lOO. - - ~
................................
I l] II I~l~l~li
10
II II IIi
III
20
II III[
II II II II
30
Minutes
Fig. 2. Effect of paired conditioning stimulation (SI +$2) on the contralateral response (SI -~R2). This
is the same animal as Fig. l, and stimulation and recording sites refer to the placements described in
Fig. 1. The ordinate plots the response to the constant test probe delivered l/20 sec at Sl. The baseline
was established in the time period prior to the combined stimulation and in fact occurred 2 rain after
the last response of Fig. 1. Data are scored as in Fig. l(b). Notice that while neither of the contralateral responses were potentiated by conditioning of S l alone (compare effects of S1 ~ R 2 of Fig. 1(b)
and S2-~RI of Fig. l(c) to Fig. 2), paired conditioning trains at SI + $ 2 result in long term potentiation
of the contralateral response evoked by S1 stimulation.
conditioned concurrently with the massive convergent ipsilateral projection system. A
test of this prediction is illustrated in Fig. 2. As is evident, the paired conditioning
stimulation delivered to S1 and $2 (two paired conditioning trials are illustrated)
resulted in dramatic potentiation of the responses evoked by the crossed projection
(SI-~R2). This induction of LTP in the crossed pathway as a consequence of concurrent activation with the converging ipsilateral projection system was observed in a
total of 6 out of 6 animals, and the median increase in the response over these cases
was 153 % (ranging from 124 to 198 %). These increases of the population EPSP were
also reflected by the positive evoked potentials recorded from stratum granulosum.
Disassociative conditioning
As noted above there was a trend for a small but definite reduction in the size of
the contralateral response when the convergent, ipsilateral input was conditioned prior
to pairing. If the ipsilateral input is able to reduce the size of the convergent crossed
input, then it might be easier to demonstrate this phenomenon in the potentiated
crossed system where a larger amount of depression should be possible. Figure 3
illustrates the analysis of this question in the same animal as illustrated in Figs. 1 and
2. In the left-hand portion of the figure, the crossed response ( S l u R 2 ) is still
potentiated following the second of the paired stimulations. Conditioning S1 or low
intensity conditioning of the ipsilateral pathway ($2 Lo in Fig. 3 was adjusted so as not
241
to evoke a population spike in R2) does not affect the size of the crossed response.
However, the same intensity conditioning train to $2 that previously allowed S l u R 2
to potentiate now totally removes potentiation. This decrease in the potentiated
crossed response following conditioning stimulation of the ipsilateral pathway was
observed in 4 out of 4 cases and, in each case, the responses were reduced to their prepotentiated levels or slightly lower. Further, the systems were capable of being
repotentiated and again depotentiated (see Fig. 3, SluR2). Since the crossed systems
were capable of multiple sequences of potentiation and depotentiation and since
neither additional paired conditioning nor solo conditioning of the crossed input
appeared to reduce the crossed response, it is quite possible that this heterosynaptic
depotentiation of the S l ~ R 2 response represents a reversal of the process that caused
the original potentiation.
Since LTP of the normal crossed pathway occurs only if the ipsilateral pathway
is concurrently activated, there must be some interaction between the activity in the
two systems. To exclude the possibility that such interactions took place at the cells of
origin in the entorhinal cortex, one animal was prepared in which the projections were
separated from their cells of origin by a transection anterior to the entorhinal area.
Stimulating electrodes were positioned in the angular bundle. In this preparation, the
pathway continued to be physiologically operational for the duration of the experiment, and this 'simplified' animal also exhibited all of the phenomena of associative
test ,%~R2
o
(~)
C
0
0
lO0"
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6O
13~) . . . .
i i Ji i
~ l l l J l ~
ill
ii
50
i i i i I i
iiii
60
iiii
iiii
ii
70
ii
ill
i111111111J
80
Minutes
Fig. 3. Effect of various stimulation conditions on the potentiated and depotentiated contralateral test
response S l u R 2 . The data are taken from the same animal as shown in previous figures and in fact
represent a direct continuation of Fig. 2. The ordinate plots the response to the same test pulse delivered at S1 as in Figs. 2, l(a) and l(b). Nomenclature and electrode placements are as in Fig. 1 with the
addendum that $2 (Lo) is a conditioning train using reduced intensity which was subthreshold for a
population spike ($2~R2). All other conditioning trains are the same intensity as used previously.
242
potentiation and de-potentiation as the normal animals. This observation eliminates
any loops through the entorhinal cortex involving the cells of origin of the entorhinal
projection system as the sites of the associative interactions.
DISCUSSION
Mechanisms of LTP
The results of the present study add further support to the hypothesis that LTP is
specific to (and located at or near) the synapses that were activated during the
conditioning. Because of the properties of the system these results eliminate several
alternative interpretations. For example, it has been suggested that changes in the
excitability of the axons (either increases or decreases) could, in part, account for some
forms of short term plastic changes in synaptic responses. Indeed, habituation of the
frog lateral column-ventral root reflex is accompanied by changes in the presynaptic
volley which may mediate the habituation 6. In the case of potentiation, however, there
has been no evidence that increases in afferent excitability accompany the changes in
synaptic responses. Further, accepting the notion that the contralateral projections are
collaterals of the ipsilateral fibers to the dentate (as has been documented in the case of
the sprouted crossed system17), the present results also argue against axonal changes,
for two inputs (ipsilateral and crossed) originating from a single stimulating location
respond differentially to the same conditioning trains. This differential potentiation of
collateral synapses suggests that the changes which mediate potentiation are not
changes in excitability in the EC or at the site of electrical stimulation, since such
changes should be reflected at both branches of a collateral.
Another argument for the specificity and 'synaptic' localization of LTP is based
on the other contralateral input (S2-~RI of Fig. 1). When conditioning stimulation is
delivered to S l, this pathway is not activated, but it does converge within the same DG
with the conditioned and potentiated ipsilateral input (S I-~R 1). Since this convergent
crossed input does not exhibit heterosynaptic potentiation when the ipsilateral input is
potentiated, generalized conductance changes of the extracellular environment are not
good candidates for the cause of LTP. Furthermore, if it can be definitively shown that
single granule cells receive input from both sides, then such control experiments
indicate that LTP is not due to a generalized alteration of granule cell properties (e.g.
resting potential, resting conductance, etc.) or changes mediated via an interneuronlike the basket cells of the DG.
The present paper describes associative potentiation of synapses as a function of
convergent co-activation with other synapses. Such a description resembles Hebb's
suggestion 4 that synaptic efficacy should change as a function of the synapse being coactive with the postsynaptic cell. However, whether or not induction of LTP is
discontinuous, i.e. a threshold phenomenon which is dependent, for example, on the
firing of the postsynaptic cell as predicted by the Hebb model 4, or a continuous
function dependent merely on the amount of synaptic activation remains unanswered.
Indeed, field potentials may be inappropriate for answering this question.
What does appear undeniable is the associative nature of the potentiation, since
243
LTP requires the activation of a sufficient population of synapses. This conclusion is
supported by the present observations that the ipsilateral collaterals are potentiated
while the crossed collaterals (which are much more sparse) are not, and with the
previous observations of McNaughton et al., who demonstrate (1) that LTP requires
some critical amount of activation in terms of the intensity required at the stimulating
electrode9, and (2) that the converging ipsilateral inputs from the medial and lateral
divisions of the entorhinal area can facilitate each other into potentiation1° in the
same manner as the ipsilateral input of the present study permitted the induction of
LTP of the crossed system. As a final piece of evidence to support the notion that LTP
requires sufficient synaptic (or postsynaptic) activation, the companion paper demonstrates that the crossed input (normally incapable of LTP without paired ipsilateral
conditioning) becomes capable of LTP after increasing the density of innervation via
post-lesion sprouting 19.
Significance of reversiblepotentiation
Hebb's postulated associative synapses, while being a corner stone of associative
memory models, could not, without appendant circuitry, perform a basic neuropsychological process such as classical conditioning when this single form of associative
learning is considered to include extinction or contingency type learning discussed by
Rescorla 15. An aspect of the present results which goes beyond Hebb's postulate, but
which should prove to be a very important property, is the capability to depotentiate
a potentiated synapse. We will call synapses which are solely capable of associative
potentiation type I, and synapses which exhibit both potentiation and de-potentiation,
type lI. At least two distinct de-potentiation properties may exist. In one case, a
synapse (designated type IIa) might exist which would potentiate like a type I synapse
and de-potentiate when activated in the absence of its previously co-active convergent
companion synapses (i.e. an extinction-like property). This type IIa synapse would be
capable of classical conditioning without appendant circuitry. The synapse described
in this report, however, is not a type IIa. We refer to the synapses of the EC-DG system
as type IIb synapses. These synapses de-potentiate if they are inactivate while their
previous partners in potentiation are active, in a manner opposite to the type IIa.
Although the type IIb synapses of the EC-DG would not be capable, by themselves, of performing classical conditioning, there exist other forms of associative
learning for which a type IIb synapse is well suited.
For example, suppose that the EC sends highly processed information to the
hippocampus related to spatial cues. A stimulus constellation of table, desk, lamp, and
chair might activate a specific set of EC-DG synapses evoking a specific output pattern
(OFFICE) of granule cell firing. Some of these EC-DG synapses will be convergently
co-active, and these convergent synapses will potentiate. The system will now be
capable of associative recall in the sense that viewing only a fragment of the total
stimulus constellation, for example, the desk, chair and table alone, will evoke a
granule cell output similar to output OFFICE because the potentiated synapses of the
desk, chair and table convergent with the lamp synapses will compensate for the
missing lamp input in so far as granule activation is concerned.
244
U p to this point, synaptic types I, IIa, and lib perform identically. However, if
the table is permanently removed from the office, this cue should no longer aid in the
recall of O F F I C E . Type IIa synapses fail because they do not de-potentiate when the
table is removed, but type lib synapses de-potentiate (for the table) when the office,
without the table, is repeatedly viewed. This is in exact analogy to the way an
ipsilateral input (corresponding to the desk, lamp, and chair) can de-potentiate an
associatively potentiated contralateral input (corresponding to the table input).
On the other hand, not only would a type IIa synapse not de-potentiate at an
appropriate time, but it might also de-potentiate as a result of inappropriate stimulus
conditions, Again consider the associated objects to include the table, desk, lamp and
chair. Seeing an identical table in a quite different setting would cause type IIa table
synapses to de-potentiate inappropriately, since the original table is still physically
associated with the other objects of O F F I C E . On the other hand, type lib synapses
would not show such inappropriate depotentiation.
Although there are certain obvious interpretative (if not outright metaphorical)
liberties in the preceding model, these considerations seem relevant to experimental
paradigms and results of Olton et al. 13, O'Keefe and C o n w a y 12, and R a n c k 14. Future
studies will be required to determine whether synapses which participate in the coding
and perception o f spatial memories are type Ilb or not.
ACKNOWLEDGEMENTS
Supported, in part, by U S P H S Research G r a n t NS12333 and N S F Research
G r a n t BNS 78-10543.
REFERENCES
1 Bliss, T. and Lomo, T., Long-lasting potentiation of synaptic transmission in the dentate area of
the anaesthetized rabbit following stimulation of the perforant path, J. Physiol. (Lond.), 232
(1973) 334-356.
2 Douglas, R. M., Long-lasting synaptic potentiation in the rat dentate gyrus following brief highfrequency stimulation, Brain Research, 126 (1977) 361-365.
3 Goldowitz, D. W., White, F., Steward, O., Cotman, C. W. and Lynch, G. S., Anatomical evidence for a projection from the entorhinal cortex to the contralateral dentate gyrus of the rat, Exp.
Neurol., 47 (1975) 433-441.
4 Hebb, D. O., The Organization of Behavior, Wiley, New York, 1975, pp. 433-441.
5 Kohonen, T., Correlation matrix memories, IEEE Trans. Comput., C-21 (1972) 353-359.
6 Levy, W. B., Cameron, D. E., and Rogers, H. R., Waning axonal conduction in the habituating
frog spinal cord, submitted.
7 Lomo, T., Patterns of activation in a monosynaptic cortical pathway: the perforant path input to
the dentate area of the hippocampal formation, Exp. Brain Res., 12 (1971) 18-45.
8 Marr, D., A theory of cerebellar cortex, J. Physiol. (Lond.), 202 (1969) 437-470.
9 McNaughton, B. L., Dissociation of short- and long-lasting modification of synaptic efficacy at
the terminals of the perforant path, Seventh Annual Meeting of the Society for Neuroscience
(1977), Anaheim, Calif., Abstract.
10 McNaughton, B. L. and Barnes, C. A., Physiological identification and analysis of dentate
granule cell responses to stimulation of the medial and lateral perforant pathways in the rat, J.
comp. Neurol., 175 (1977) 439-454.
I1 McNaughton, B. L., Douglas, R. M. and Goddard, G. V., Synaptic enhancement in fascia
dentata: cooperativity among coactive afferents, Brain Research, 157 (1978) 277-293.
245
12 O'Keefe, J. and Conway, D. H., Hippocampal place units in the freely moving rat: why they fire
where they fire, Exp. Brain Res., 31 (1978) 573-590.
13 Olton, D. S., Branch, M. and Best, P. J., Spatial correlates of hippocampal activity, Exp. NeuroL,
58 (1978) 387-409.
14 Ranck, 5. B., Studies on single neurons in dorsal hippocampal formation and septum in unrestrained rats, Exp. Neurol., 41 (1973) 461-455.
15 Rescorla, R. A., Conditioned inhibition of fear resulting from negative CS-UCS contingencies,
J. comp. physiol. Psychol., 67 (1969) 504-509.
16 Stent, G. S., A physiological mechanism for Hebb's postulate of learning, Proc. nat. Acad. Sci.
(Wash.), 70 (1973) 997-1001.
17 Steward, O. and Vinsant, S. L., Collateral projections of cells in the surviving entorhinal area
which reinnervate the dentate gyrus of the rat following unilateral entorhinal lesions, Brain
Research, 149 (1978) 216-222.
18 White, W. F., Goldowitz, D., Lynch, G. and Cotman, C. W., Electrophysiological analysis of the
projection from the contralateral entorhinal cortex to the dentate gyrus in normal rats, Brain
Research, 114 (1976) 201-209.
19 Wilson, R., Levy, W. B. and Steward, O., Functional effects of lesion-induced plasticity: long
term potentiation in the normal and lesion-induced temporo-dentate circuits, Brain Research, 176
(1979) 65-78.