causal relationship of interconnected neu-
rons may be necessary to form long-lasting
associations between them.
LTP has been well studied since its initial
discovery in the rabbit hippocampus (4),
which is a region of the brain that plays a
key role in navigation, learning, and memory in humans and other mammals (5). In
rodents, the majority of the hippocampal
CA1 as well as input CA3 pyramidal neurons—collectively known as “place cells”—
show location-specific activity patterns that
become active only in small parts (place
fields) of an environment (6). However, currently, it is still unknown how hippocampal
plasticity—i.e., the neural ability to change
the signal transmission efficiency—
contributes to place cell formation (7).
Previously, it was shown in rodents that
place cell firing can be evoked in silent
CA1 pyramidal neurons in any arbitrary
location within an environment by a brief
intracellular injection of a small depolarizing current that results in a transient
(up to a few seconds) ramplike increase in
the membrane potential, which indicates
that the neuron becomes more excitable
(8). This suggests that CA1 pyramidal neurons must receive a wide range of spatially
tuned inputs from the CA3 area as well as
medial entorhinal regions and that their
ability to respond is determined by the depolarization of their membrane potential.
Consistently, Bittner et al. have shown that
the place cell response is indeed evoked by
ramplike membrane depolarization. The
authors speculate that the induced depolarization triggers a sequence of as yet
unknown processes in CA1 neurons that
potentiates their CA3 inputs during the
entire duration of depolarization, before
and after the firing of the receiving CA1
cell. The ramplike depolarization and the
synaptic potentiation profiles have a complex relationship: Although the origin of
the depolarization is not clear, once it appears, it is shaped by the CA3 inputs during the initial learning stages. The authors
offer a biophysical model to account for
this hypothesis. They test their theory by
investigating the relationship between the
speed of an animal and the ramp width—
i.e., how far the place cell depolarization
ramp extends in space (see the figure).
The Hebbian learning rule predicts that
the ramp width should not depend on the
animal’s speed, whereas BTSP predicts
that ramp width should significantly increase when the mice are running faster,
because a larger number of CA3 place cell
inputs will be potentiated during the same
time as in slower-moving mice. Crucially,
3 s before to 2 s after the standard Hebbian
plasticity window, allowing the CA3 inputs
with no causal or temporal relation to the
receiving cell to become rapidly potenti-
ated. The lack of causality in this type of
learning is puzzling, and its implications
for an animal’s ability to learn new places
and navigate remain an open question. In
fact, it is precisely the causal properties of
LTP—associativity, cooperativity, and input
specificity—that made it so attractive as a
potential mechanism for learning (9): Two
neurons become associated because one
causes the activity of the other.
What would BTSP learning be useful
for? The authors speculate that such a long
time scale plasticity window in principle
allows for the storage of an entire sequence
of events (places traversed) occurring be-
fore and after the place field of each indi-
vidual CA1 cell. In addition, they suggest
that this could result in overrepresentation
of behaviorally important places, such as
reward locations. Conversely, many other
questions remain open. For instance, how
would this BTSP rule operate during navi-
gation in two-dimensional (2D) environ-
ments? Here the animal does not traverse a
sequence of place fields in a fixed order but
approaches the same field from many dif-
ferent directions. Could the potentiation of
CA3 inputs impinging on the CA1 cell after
place field traversal help to build the om-
nidirectional 2D place field? Also, how do
other spatial inputs (e.g., from the medial
entorhinal cortex) contribute to this learn-
ing? Previous studies have shown that CA1
place cell activity remains intact even af-
ter damage to the CA3 region (10). Finally,
what cellular processes regulating neural
plasticity could underlie such a broad time
window of synaptic potentiation? Are they
really qualitatively different from our cur-
rent understanding of cellular processes
underlying LTP (11)? Importantly, what is
the source of the initial depolarization that
creates the place field and generates the
long-duration plasticity in the first place?
The authors have identified an intriguing
new phenomenon that raises the prospect
of bringing physiological plasticity mecha-
nisms and place cell formation into closer
alignment, which eventually will tell us
how we learn and remember new places
and events that happen there. j
1. D. O. Hebb, The Organization of Behavior ( Wiley, 1949).
2. S. Löwel, W. Singer, Science 255, 209 (1992).
3. K. C. Bittner et al ., Science 357, 1033 (2017).
4. T.V.Bliss, T.Lomo, J.Physiol. 232,331(1973).
5. J. O’Keefe, L. Nadel, The Hippocampus as a Cognitive Map
(Oxford Univ. Press, 1978).
6. J.O’Keefe, J.Dostrovsky, Brain Res.34,171(1971).
7. K.J.Jeffery, Hippocampus 7,95(1997).
8. D. Lee, B.-J. Lin, A. K. Lee, Science 337, 849 (2012).
9. T. V. Bliss, G. L. Collingridge, Nature 361, 31 (1993).
10. V. H. Brun et al., Science 296, 2243 (2002).
11. R. G. Morris, E. Anderson, G. S. Lynch, M. Baudry, Nature
319, 774 (1986).
Thalamus Brain stem
Mouse brain Hippocampus
Formation of new
CA1 place cells
Initiation of plateau
potential produces a
ramplike depolarization of
membrane potential driving
CA1 place Eeld Ering in
subsequent trials. The
depolarization ramp in
turn potentiates the CA3
inputs over behaviorally
relevant time scales of a
8 SEPTEMBER 2017 • VOL 357 ISSUE 6355 975
Learning new places
Non-Hebbian behavioral time scale synaptic plasticity (BTSP) spanning a time course of more than
several seconds may underlie CA1 place cell firing properties, implying that a causal relationship between
interconnected neurons may not be necessary to form long lasting associations.