50 Years Ago
Reading aids for the blind have
so far involved the use of intact
sensory pathways and have
progressed little beyond Braille
and tape-recorded “talking-books”.
Both these systems are quite
expensive … and both are slow in
terms of information transfer to the
reader … At a recent meeting of the
Physiological Society, Brindley and
Lewin demonstrated a device for
stimulating the visual cortex of man
directly … Essentially it consists
of an array of radio receivers,
encapsulated in silicone rubber and
screwed to the skull … Activation
of a receiver stimulated the cortex:
transmission was in the form of
a train of short (200 μs) pulses …
it does at least seem feasible to
transmit visual information directly
to the central visual pathways of the
From Nature 8 June 1968
100 Years Ago
It happened last week that about
1 lb. of fresh lamb was put into
an oven at night in order that it
might be cooked by morning on
the “hay-box” principle. It was
in a casserole, with a little water.
Similar treatment in the same oven
on previous occasions had been
very successful. At about 5 a.m.
the casserole was examined, and
the broth was found to be very well
tasted, and the whole smelt fresh
and good, but the meat when tested
with a fork was not tender, and
the fat (of which there was a good
deal) was entirely unmelted. The
casserole was returned to the oven
(then quite cool) and taken out
again after breakfast. The contents
were then found to be smelling
most offensively, as if extremely
“high”. The fat was melted. The
meat and broth were judged quite
unfit for human food. I wonder if
any of your readers would explain
this curious development.
From Nature 6 June 1918
are generated by multiple thalamic inputs
that have temporally different responses to
the stimulus (Fig. 1). Thalamic inputs that
respond slowly to visual stimuli generate slow
responses in cortical regions, whereas those
responding faster generate fast responses.
Lien and Scanziani’s results, taken together
with previous work
, raise the interesting pos-
sibility that cortical direction selectivity is gen-
erated through a common mechanism — the
convergence of temporally diverse thalamic
inputs — in rodents, cats and primates. But as
with all research, some questions remain open.
For instance, the authors focus their study
on the middle layers of the visual cortex,
which receive the bulk of the thalamic input
As Lien and Scanziani show, many thalamic
inputs in these middle cortical layers are not
direction selective, but their combined activity
is. It remains unclear whether thalamic inputs
that target other cortical layers (or serve other
functions) can encode direction selectivity
through different mechanisms. For example,
neurons in the superficial layers of the cor-
tex might derive their direction selectivity
from thalamic neurons that are themselves
It is also known that thalamic inputs to the
visual cortex are arranged by their receptive-
field position — inputs that have receptive fields
close to one another in the field of view are clus-
tered together. However, it is not yet known
whether the thalamic inputs are also arranged
according to their temporal properties. If so, this
could explain why spatial position and direction
preference tend to change together in different
neurons across the visual–cortical map
Whatever the answers are, it is becoming
increasingly clear that the visual cortex gen-
erates stimulus selectivity, such as prefer-
ences for direction and orientation, through
thalamo–cortical convergence. Lien and
Scanziani’s work shows that this mechanism
is better preserved across mammals than was
Jose Manuel Alonso is in the College of
Optometry, State University of New York,
New York, New York 10036, USA.
1. Hubel, D. H. & Wiesel, T. N. J. Physiol. 160, 106–154
2. Lien, A. D. & Scanziani, M. Nature 558, 80–86
3. Alonso, J.-M., Usrey, W. M. & Reid, R. C. J. Neurosci.
21, 4002–4015 (2001).
4. Ferster, D., Chung, S. & Wheat, H. Nature 380,
5. Saul, A. B. & Humphrey, A. L. J. Neurophysiol. 64,
6. Saul, A. B. & Humphrey, A. L. J. Neurophysiol. 68,
7. Stanley, G. B. et al. J. Neurosci. 32, 9073–9088
8. Reid, R. C., Soodak, R. E. & Shapley, R. M.
J. Neurophysiol. 66, 505–529 (1991).
9. Livingstone, M. S. Neuron 20, 509–526 (1998).
10. McLean, J. & Palmer, L. A. Vision Res. 29, 675–679
11. Lorente de No, R. In Physiology of the Nervous
System (ed. Fulton, J.) 291–340 (Oxford Univ. Press,
12. Cruz-Martín, A. et al. Nature 507, 358–361 (2014).
13. Kremkow, J., Jin, J., Wang, Y. & Alonso, J. M. Nature
533, 52–57 (2016).
This article was published online on 23 May 2018.
Two artificial synapses
are better than one
Emerging nanoelectronic devices could revolutionize artificial neural networks, but
their hardware implementations lag behind those of their software counterparts. An
approach has been developed that tips the scales in their favour. S A .60
GINA C. ADAM
nspired by the brain’s neural networks,
scientists have for decades tried to
construct electronic circuits that can
process large amounts of data. However, it
has been difficult to achieve energy-efficient
implementations of artificial neurons and
synapses (connections between neurons).
On page 60, Ambrogio et al.
report an arti-
ficial neural network containing more than
200,000 synapses that can classify complex
collections of images. The authors’ work dem-
onstrates that hardware-based neural networks
that use emerging nanoelectronic devices
can perform as well as can software-based
networks running on ordinary computers,
while consuming much less power.
Artificial neural networks are not
programmed in the same way as conventional
computers. Just as humans learn from experi-
ence, these networks acquire their functions
from data obtained during a training process.
Image classification, which involves learning
and memory, requires thousands of artificial
synapses. The states (electrical properties) of
these synapses need to be programmed quickly
and then retained for future network operation.
Nanoscale synaptic devices that have
programmable electrical resistance, such
7 JUNE 2018 | VOL 558 | NATURE | 39
NEWS & VIEWS RESEARCH