<s>
Neural	B-Algorithm
Engineering	I-Algorithm
Object	I-Algorithm
(	O
Nengo	B-Algorithm
)	O
is	O
a	O
graphical	O
and	O
scripting	O
software	O
for	O
simulating	O
large-scale	O
neural	O
systems	O
.	O
</s>
<s>
As	O
the	O
neural	B-Algorithm
network	I-Algorithm
software	I-Algorithm
Nengo	B-Algorithm
is	O
a	O
tool	O
for	O
modelling	O
neural	O
networks	O
with	O
applications	O
in	O
cognitive	O
science	O
,	O
psychology	O
,	O
artificial	B-Application
intelligence	I-Application
and	O
neuroscience	O
.	O
</s>
<s>
Some	O
form	O
of	O
Nengo	B-Algorithm
has	O
existed	O
since	O
2003	O
.	O
</s>
<s>
Originally	O
developed	O
as	O
a	O
Matlab	B-Language
script	O
under	O
the	O
name	O
NESim	O
(	O
Neural	O
Engineering	O
Simulator	O
)	O
,	O
it	O
was	O
later	O
moved	O
to	O
a	O
Java	B-Language
implementation	O
under	O
the	O
name	O
NEO	O
,	O
and	O
then	O
eventually	O
Nengo	B-Algorithm
.	O
</s>
<s>
The	O
first	O
three	O
generations	O
of	O
Nengo	B-Algorithm
developed	O
with	O
a	O
focus	O
on	O
developing	O
a	O
powerful	O
modelling	O
tool	O
with	O
a	O
simple	O
interface	O
,	O
and	O
scripting	O
system	O
.	O
</s>
<s>
This	O
most	O
recent	O
iteration	O
of	O
Nengo	B-Algorithm
defines	O
a	O
specific	O
Python-based	O
scripting	O
API	O
with	O
back-ends	O
targeting	O
Numpy	O
,	O
OpenCL	O
and	O
Neuromorphic	O
hardware	O
such	O
as	O
Spinnaker	O
.	O
</s>
<s>
Nengo	B-Algorithm
uses	O
a	O
custom	O
license	O
which	O
allows	O
free	O
personal	O
and	O
research	O
use	O
,	O
but	O
a	O
paid	O
license	O
is	O
required	O
for	O
commercial	O
purposes	O
.	O
</s>
<s>
Nengo	B-Algorithm
is	O
built	O
upon	O
two	O
theoretic	O
underpinnings	O
,	O
the	O
Neural	O
Engineering	O
Framework	O
(	O
NEF	O
)	O
and	O
the	O
Semantic	O
Pointer	O
Architecture	O
(	O
SPA	O
)	O
.	O
</s>
<s>
Nengo	B-Algorithm
differs	O
primarily	O
from	O
other	O
modelling	O
software	O
in	O
the	O
way	O
it	O
models	O
connections	O
between	O
neurons	O
and	O
their	O
strengths	O
.	O
</s>
<s>
Using	O
the	O
NEF	O
,	O
Nengo	B-Algorithm
allows	O
defining	O
connection	O
weights	O
between	O
populations	O
of	O
spiking	O
neurons	O
by	O
specifying	O
the	O
function	O
to	O
be	O
computed	O
,	O
instead	O
of	O
forcing	O
the	O
weights	O
to	O
be	O
set	O
manually	O
,	O
or	O
use	O
a	O
learning	O
rule	O
to	O
configure	O
them	O
from	O
a	O
random	O
start	O
.	O
</s>
<s>
That	O
being	O
said	O
,	O
these	O
aforementioned	O
traditional	O
modelling	O
methods	O
are	O
still	O
available	O
in	O
Nengo	B-Algorithm
.	O
</s>
<s>
To	O
represent	O
symbols	O
in	O
Nengo	B-Algorithm
,	O
SPA	O
is	O
used	O
.	O
</s>
<s>
In	O
Nengo	B-Algorithm
,	O
these	O
are	O
presented	O
as	O
vectors	O
with	O
a	O
set	O
of	O
operations	O
associated	O
to	O
them	O
.	O
</s>
<s>
Notable	O
developments	O
accomplished	O
using	O
the	O
Nengo	B-Algorithm
software	O
have	O
occurred	O
in	O
many	O
fields	O
,	O
and	O
Nengo	B-Algorithm
has	O
been	O
used	O
and	O
cited	O
in	O
over	O
100	O
publications	O
.	O
</s>
<s>
An	O
important	O
development	O
to	O
note	O
is	O
Spaun	B-Architecture
,	O
a	O
network	O
of	O
6.6	O
million	O
artificial	O
spiking	O
neurons	O
(	O
a	O
small	O
number	O
compared	O
to	O
the	O
number	O
in	O
the	O
human	O
brain	O
)	O
,	O
which	O
uses	O
groups	O
of	O
these	O
neurons	O
to	O
complete	O
cognitive	O
tasks	O
via	O
flexible	O
coordination	O
.	O
</s>
<s>
Spaun	B-Architecture
is	O
the	O
world	O
's	O
largest	O
functional	O
brain	O
model	O
,	O
and	O
can	O
be	O
used	O
to	O
test	O
hypotheses	O
in	O
neuroscience	O
.	O
</s>
