in a controlled manner. For example, one could
aim to build the magnetic counterparts of
alternating-current electrical circuits. On the
way to turning the study of monopoles into a
proper applied science, it will be necessary to
ask if the basic ideas of dipole fractionalization
that give a spin-ice material its special proper-
ties can be realized in other magnetic settings.
Spin ice may be the first fractionalized magnet
in three dimensions, but surely should not be
the only one. 
Shivaji Sondhi is in the Department of Physics,
Princeton University, Princeton, New Jersey
08544, USA.
e-mail: sondhi@princeton.edu
1. Castelnovo, C., Moessner, R. & Sondhi, S. L. Nature 451,
42–45 (2008).
2. Bramwell, S. T. et al. 461, 956–959 (2009).
3. Onsager, L. J. Chem. Phys. 2, 599–615 (1934).
4. Langevin, P. Ann. Chim. Phys. VII 28, 433 (1903).
5. Fennell, T. et al. Science doi:10.1126/science.1177582 (2009).
6. Morris, D. J. P. et al. Science doi:10.1126/science.1178868
(2009).
NEUROSCIENCE
The inside story on place cells
Douglas Nitz
Neurons known as place cells encode spatial information that is needed
to guide an animal’s movement. Nearly 40 years after these cells were
discovered, neuroscience gets a look at their internal dynamics.
On page 941 of this issue, Harvey et al.
1

describe a novel combination of techniques
that they have used to address unresolved ques-
tions about brain function. The experiments
involve mice navigating within a virtual-reality
setting while intracellular electrophysiological
recordings are made of particular
neurons — pyramidal neurons —
in the hippocampus. The ability
to gather such data adds greatly
to the information that can be
gleaned from the accompanying
extracellular recordings.
The gap in our knowledge
between findings obtained from
intracellular and extracellular
electrophysiology experiments is
closing fast, and the timing could
hardly be better. Today especially,
it is difficult to overstate the
importance of understanding how
the dynamics of electrical activity
within single neurons is related to
firing patterns among collections
of neurons that accompany the
performance of complex tasks.
Such information is necessary to
fully understand the operational
principles of neural networks
that have been newly revealed
by subtle manipulations of their
elements
2
. It is equally relevant
to teams of scientists who are
struggling to develop large-
scale, spiking-neuron models of
the brain that apply to the real
world
3
.
Into this mix comes research
from David Tank’s group, in the
form of the paper by Harvey et al.
1
,
which proves that it is not impos-
sible to examine brain correlates
of higher cognitive processes and at the same
time identify their underlying causes at the
cellular level. The authors’ work unveils the
membrane-potential dynamics of ‘place cells’,
a subtype of pyramidal neuron, whose spike-
firing patterns reflect both the animal’s present
spatial position in the environment and the
specific trajectory taken to reach that posi-
tion. Beyond this impressive technical achieve-
ment is a result that clarifies a basic principle
by which temporal coding of neuronal spike
firing can be realized.
The 1971 discovery of place cells
4
was a
strong early indication of how much would
eventually be learned through new methods
for extracellular single-neuron recordings in
freely behaving animals. After decades of sub-
sequent place-cell recording experiments, there
is a growing consensus about the mechanisms
by which the hippocampus simultaneously
functions to map environmental position and
to generate episodic memories. Critical to this
understanding was the discovery of ‘phase
precession’
5
, wherein place-specific firing of
hippocampal neurons is itself temporally organ-
ized against a background rhythm that takes the
form of theta-frequency (6–10 Hz) oscillations
of the whole hippocampal neuron population.
Here, over short intervals of time (about 125
milliseconds), the firing order for a set of hip-
pocampal place cells with partially overlapping
place fields is found to match the animal’s physi-
cal trajectory corresponding to those fields.
Phase precession stands as perhaps the most
robust example of temporal coding of infor-
mation in the mammalian brain. The means
by which phase precession of place-specific
activity occurs is, at present, a
matter of intense debate.
Harvey et al.
1
provide a power-
ful example of what will be
learned in the decades to come.
The broader promise of the tech-
nique lies in learning exactly how
the myriad incoming synaptic
potentials to any given neuron
are integrated to yield spike-firing
patterns that closely track specific
thoughts, perceptions or actions.
Aside from the intellectual reward
in understanding — at macro-
and microscopic levels — how the
brain functions, some, including
myself, also see this as a prerequi-
site for the development of brain-
based, non-biological devices
capable of autonomous function
in a constantly changing environ-
ment. Neurons condense complex
collections of information arriv-
ing at their synapses into the more
concise messages contained in
their action-potential firing pat-
terns. The mathematical rules
governing such transformations
may well be applied in tomorrow’s
computers and robots.
The experiments themselves
involve a novel integration of
existing techniques, each pre-
senting its own complications.
Combined intracellular and extra-
cellular recordings are obtained
Figure 1 | Mouse navigation in virtual reality. The integration of approaches
used by Harvey et al.
1
allows extracellular as well as intracellular place-specific
activity in the brain to be monitored.
F
.

C
O
L
L
M
A
N

889
NATURE|Vol 461|15 October 2009 NEWS & VIEWS
887-894 News and Views MH IF.indd 889 9/10/09 17:26:32
© 2009 Macmillan Publishers Limited. All rights reserved