bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2024,
Volume and Issue:
unknown
Published: June 10, 2024
Abstract
Novelty
detection,
also
known
as
familiarity
discrimination
or
recognition
memory,
refers
to
the
ability
distinguish
whether
a
stimulus
has
been
seen
before.
It
hypothesized
that
novelty
detection
can
naturally
arise
within
networks
store
memory
learn
efficient
neural
representation,
because
these
already
information
on
familiar
stimuli.
However,
computational
models
instantiating
this
hypothesis
have
not
shown
reproduce
high
capacity
of
human
so
it
is
unclear
if
feasible.
This
paper
demonstrates
predictive
coding,
which
an
established
model
previously
effectively
support
representation
learning
and
discriminate
with
capacity.
Predictive
coding
includes
neurons
encoding
prediction
errors,
we
show
produce
higher
activity
for
novel
stimuli,
be
decoded
from
their
activity.
Moreover,
hierarchical
uniquely
perform
at
varying
abstraction
levels
across
hierarchy,
i.e.,
they
detect
both
low-level
features,
higher-level
objects.
Overall,
unify
associative
single
framework.
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2023,
Volume and Issue:
unknown
Published: Jan. 24, 2023
Mammalian
cortex
features
a
vast
diversity
of
neuronal
cell
types,
each
with
characteristic
anatomical,
molecular
and
functional
properties.
Synaptic
connectivity
powerfully
shapes
how
type
participates
in
the
cortical
circuit,
but
mapping
rules
at
resolution
distinct
types
remains
difficult.
Here,
we
used
millimeter-scale
volumetric
electron
microscopy
1
to
investigate
all
inhibitory
neurons
across
densely-segmented
population
1352
cells
spanning
layers
mouse
visual
cortex,
producing
wiring
diagram
connections
more
than
70,000
synapses.
Taking
data-driven
approach
inspired
by
classical
neuroanatomy,
classified
based
on
relative
targeting
dendritic
compartments
other
developed
novel
classification
excitatory
morphological
synaptic
input
The
between
revealed
class
disinhibitory
specialist
basket
cells,
addition
familiar
subclasses.
Analysis
onto
found
widespread
specificity,
many
interneurons
exhibiting
differential
certain
subpopulations
spatially
intermingled
potential
targets.
Inhibitory
was
organized
into
“motif
groups,”
diverse
sets
that
collectively
target
both
perisomatic
same
Collectively,
our
analysis
identified
new
organizing
principles
for
inhibition
will
serve
as
foundation
linking
modern
multimodal
atlases
diagram.
Prediction
errors
are
differences
between
expected
and
actual
sensory
input
thought
to
be
key
computational
signals
that
drive
learning
related
plasticity.
One
way
prediction
could
is
by
activating
neuromodulatory
systems
gate
The
catecholaminergic
locus
coeruleus
(LC)
a
major
system
involved
in
neuronal
plasticity
the
cortex.
Using
two-photon
calcium
imaging
mice
exploring
virtual
environment,
we
found
activity
of
LC
axons
cortex
correlated
with
magnitude
unsigned
visuomotor
errors.
response
profiles
were
similar
both
motor
visual
cortical
areas,
indicating
broadcast
throughout
dorsal
While
layer
2/3
primary
cortex,
optogenetic
stimulation
facilitated
stimulus-specific
suppression
responses
during
locomotion.
This
-
induced
minutes
recapitulated
effect
on
scale
normally
observed
development
across
days.
We
conclude
activity,
facilitates
sensorimotor
consistent
role
modulating
rates.
Nature,
Journal Year:
2025,
Volume and Issue:
640(8058), P. 448 - 458
Published: April 9, 2025
Mammalian
cortex
features
a
vast
diversity
of
neuronal
cell
types,
each
with
characteristic
anatomical,
molecular
and
functional
properties1.
Synaptic
connectivity
shapes
how
type
participates
in
the
cortical
circuit,
but
mapping
rules
at
resolution
distinct
types
remains
difficult.
Here
we
used
millimetre-scale
volumetric
electron
microscopy2
to
investigate
all
inhibitory
neurons
across
densely
segmented
population
1,352
cells
spanning
layers
mouse
visual
cortex,
producing
wiring
diagram
inhibition
more
than
70,000
synapses.
Inspired
by
classical
neuroanatomy,
classified
based
on
targeting
dendritic
compartments
developed
an
excitatory
neuron
classification
reconstructions
whole-cell
maps
synaptic
input.
Single-cell
showed
class
disinhibitory
specialist
that
targets
basket
cells.
Analysis
onto
found
widespread
specificity,
many
interneurons
exhibiting
differential
spatially
intermingled
subpopulations.
Inhibitory
was
organized
into
'motif
groups',
diverse
sets
collectively
target
both
perisomatic
same
targets.
Collectively,
our
analysis
identified
new
organizing
principles
for
will
serve
as
foundation
linking
contemporary
multimodal
atlases
diagram.
bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2024,
Volume and Issue:
unknown
Published: June 10, 2024
The
mammalian
cortex
is
comprised
of
cells
classified
into
types
according
to
shared
properties.
Defining
the
contribution
each
cell
type
processes
guided
by
essential
for
understanding
its
function
in
health
and
disease.
We
used
transcriptomic
epigenomic
cortical
taxonomies
from
mouse
human
define
marker
genes
putative
enhancers
created
a
large
toolkit
transgenic
lines
enhancer
AAVs
selective
targeting
populations.
report
evaluation
fifteen
new
driver
lines,
two
reporter
>800
different
covering
most
subclasses
cells.
tools
reported
here
as
well
scaled
process
tool
creation
modification
enable
diverse
experimental
strategies
towards
brain
function.
Frontiers in Neural Circuits,
Journal Year:
2024,
Volume and Issue:
17
Published: Jan. 8, 2024
If
a
full
visual
percept
can
be
said
to
‘hypothesis’,
so
too
neural
‘prediction’
–
although
the
latter
addresses
one
particular
component
of
image
content
(such
as
3-dimensional
organisation,
interplay
between
lighting
and
surface
colour,
future
trajectory
moving
objects,
on).
And,
because
processing
is
hierarchical,
predictions
generated
at
level
are
conveyed
in
backward
direction
lower
level,
seeking
predict,
fact,
activity
that
prior
stage
processing,
learning
from
errors
signalled
opposite
direction.
This
essence
‘predictive
coding’,
once
an
algorithm
for
information
theoretical
basis
nature
operations
performed
by
cerebral
cortex.
Neural
models
implementation
predictive
coding
invoke
specific
functional
classes
neuron
generating,
transmitting
receiving
predictions,
producing
reciprocal
error
signals.
Also
third
general
class,
‘precision’
neurons,
tasked
with
regulating
magnitude
signals
contingent
upon
confidence
placed
prediction,
i.e.,
reliability
behavioural
utility
sensory
data
it
predicts.
So,
what
ultimate
source
‘prediction’?
The
answer
multifactorial:
knowledge
current
environmental
context
immediate
past,
allied
memory
lifetime
experience
way
world,
doubtless
fine-tuned
evolutionary
history
too.
There
are,
consequence,
numerous
potential
avenues
experimenters
manipulate
subjects’
expectation,
examine
elicited
surprising,
less
surprising
stimuli.
review
focuses
physiology
mouse
monkey
cortex,
summarising
commenting
on
evidence
date,
placing
broader
field.
It
concluded
has
firm
grounding
basic
neuroscience
that,
unsurprisingly,
there
remains
much
learn.
Our
movements
result
in
predictable
sensory
feedback
that
is
often
multimodal.
Based
on
deviations
between
predictions
and
actual
input,
primary
areas
of
cortex
have
been
shown
to
compute
sensorimotor
prediction
errors.
How
errors
one
modality
influence
the
computation
another
still
unclear.
To
investigate
multimodal
mouse
auditory
(ACx),
we
used
a
virtual
environment
experimentally
couple
running
both
self-generated
visual
feedback.
Using
two-photon
microscopy,
first
characterized
responses
layer
2/3
(L2/3)
neurons
sounds,
stimuli,
onsets
found
all
three
stimuli.
Probing
evoked
by
audiomotor
mismatches,
they
closely
resemble
visuomotor
mismatch
(V1).
Finally,
testing
for
cross
modal
coupling
sound
amplitude
flow
speed
running,
were
amplified
when
paired
with
concurrent
mismatches.
results
demonstrate
non-hierarchical
interactions
shape
error
cortical
L2/3.
Our
movements
result
in
predictable
sensory
feedback
that
is
often
multimodal.
Based
on
deviations
between
predictions
and
actual
input,
primary
areas
of
cortex
have
been
shown
to
compute
sensorimotor
prediction
errors.
How
errors
one
modality
influence
the
computation
another
still
unclear.
To
investigate
multimodal
mouse
auditory
cortex,
we
used
a
virtual
environment
experimentally
couple
running
both
self-generated
visual
feedback.
Using
two-photon
microscopy,
first
characterized
responses
layer
2/3
(L2/3)
neurons
sounds,
stimuli,
onsets
found
all
three
stimuli.
Probing
evoked
by
audiomotor
(AM)
mismatches,
they
closely
resemble
visuomotor
(VM)
mismatch
(V1).
Finally,
testing
for
cross
modal
AM
coupling
sound
amplitude
flow
speed
running,
were
amplified
when
paired
with
concurrent
VM
mismatches.
results
demonstrate
non-hierarchical
interactions
shape
error
cortical
L2/3.