bioRxiv (Cold Spring Harbor Laboratory),
Journal Year:
2022,
Volume and Issue:
unknown
Published: June 6, 2022
Abstract
Mathematical
operations
have
long
been
regarded
as
a
sparse,
symbolic
process
in
neuroimaging
studies.
In
contrast,
advances
artificial
neural
networks
(ANN)
enabled
extracting
distributed
representations
of
mathematical
operations.
Recent
studies
compared
the
visual,
auditory
and
language
domains
ANNs
biological
(BNNs).
However,
such
relationship
has
not
yet
examined
mathematics.
Here
we
used
fMRI
data
series
problems
with
nine
different
combinations
operators
to
construct
voxel-wise
encoding
models
using
both
sparse
operator
latent
ANN
features.
Representational
similarity
analysis
demonstrated
shared
between
BNN,
an
effect
particularly
evident
intraparietal
sulcus.
Feature-brain
served
reconstruct
representation
based
on
Such
reconstruction
was
more
efficient
when
features
from
deeper
layers.
Moreover,
allowed
decoding
novel
during
model
training
brain
activity.
The
current
study
provides
insights
into
code
underlying
thought.
Nature Communications,
Journal Year:
2022,
Volume and Issue:
13(1)
Published: Nov. 14, 2022
Abstract
Translating
a
perceived
number
into
matching
of
self-generated
actions
is
hallmark
numerical
reasoning
in
humans
and
animals
alike.
To
explore
this
sensorimotor
transformation,
we
trained
crows
to
judge
values
displays
flexibly
plan
perform
pecks.
We
report
selective
neurons
the
crow
telencephalon
that
signaled
impending
actions.
Neuronal
population
activity
during
transformation
period
predicted
whether
mistakenly
planned
fewer
or
more
pecks
than
instructed.
During
both
static
neuronal
code
characterized
by
persistently
number-selective
dynamic
originating
from
carrying
rapidly
changing
information
emerged.
The
findings
indicate
there
are
distinct
functions
abstract
codes
supporting
system.
NeuroImage,
Journal Year:
2022,
Volume and Issue:
254, P. 119123 - 119123
Published: March 20, 2022
The
involvement
of
the
medial
temporal
lobe
(MTL)
in
working
memory
is
controversially
discussed.
Recent
findings
suggest
that
persistent
neural
firing
hippocampus
during
maintenance
verbal
associated
with
workload.
Here,
we
recorded
single
neuron
13
epilepsy
patients
(7
male)
while
they
performed
a
visual
task.
number
colored
squares
stimulus
set
determined
workload
trial.
Performance
was
almost
perfect
for
low
(1
and
2
squares)
dropped
at
high
(4
6
squares),
suggesting
exceeded
capacity.
We
identified
neurons
MTL
showed
period.
More
were
found
trials
correct
compared
to
incorrect
performance.
Maintenance
increased
decreased
entorhinal
cortex
Population
predicted
particularly
Prediction
accuracy
based
on
single-trial
activity
strongest
hippocampus.
data
reflects
domain-general
process
supporting
performance
multiple
items
below
beyond
Persistent
may
be
its
preference
visual-spatial
arrays.
NeuroImage,
Journal Year:
2023,
Volume and Issue:
270, P. 119980 - 119980
Published: Feb. 26, 2023
Mathematical
operations
have
long
been
regarded
as
a
sparse,
symbolic
process
in
neuroimaging
studies.
In
contrast,
advances
artificial
neural
networks
(ANN)
enabled
extracting
distributed
representations
of
mathematical
operations.
Recent
studies
compared
the
visual,
auditory
and
language
domains
ANNs
biological
(BNNs).
However,
such
relationship
has
not
yet
examined
mathematics.
Here
we
hypothesise
that
ANN-based
can
explain
brain
activity
patterns
We
used
fMRI
data
series
problems
with
nine
different
combinations
operators
to
construct
voxel-wise
encoding/decoding
models
using
both
sparse
operator
latent
ANN
features.
Representational
similarity
analysis
demonstrated
shared
between
BNN,
an
effect
particularly
evident
intraparietal
sulcus.
Feature-brain
(FBS)
served
reconstruct
representation
based
on
features
each
cortical
voxel.
Such
reconstruction
was
more
efficient
when
from
deeper
layers.
Moreover,
allowed
decoding
novel
during
model
training
activity.
The
current
study
provides
insights
into
code
underlying
thought.
Heliyon,
Journal Year:
2024,
Volume and Issue:
10(14), P. e34162 - e34162
Published: July 1, 2024
Whether
non-symbolic
encoding
of
quantity
is
predisposed
at
birth
with
dedicated
hard-wired
neural
circuits
debated.
Here
we
presented
newly-hatched
visually
naive
chicks
stimuli
(flashing
dots)
either
identical
or
different
numerousness
(with
a
ratio
1:3)
their
continuous
physical
appearance
(size,
contour
length,
density,
convex
hull)
randomly
changing.
Chicks
spontaneously
tell
apart
the
on
basis
number
elements.
Upon
presentation
fixed
changing
showed
expression
early
gene
c-fos
in
visual
Wulst,
hippocampal
formation,
intermediate
medial
mesopallium,
and
caudal
part
nidopallium
caudolaterale.
The
results
support
hypothesis
that
ability
to
discriminate
quantities
does
not
require
any
specific
instructive
experience
involves
network
several
populations
number-selective
neurons.
Evidence
for
innateness
numerical
cognition
have
implications
both
neurobiology
philosophy
mathematics.
Physiological Reviews,
Journal Year:
2024,
Volume and Issue:
105(1), P. 267 - 314
Published: Aug. 8, 2024
The
human
brain
possesses
neural
networks
and
mechanisms
enabling
the
representation
of
numbers,
basic
arithmetic
operations,
mathematical
reasoning.
Without
ability
to
represent
numerical
quantity
perform
calculations,
our
scientifically
technically
advanced
culture
would
not
exist.
However,
origins
abilities
are
grounded
in
an
intuitive
understanding
deeply
rooted
biology.
Nevertheless,
more
symbolic
skills
require
a
cultural
background
with
formal
education.
In
past
two
decades,
cognitive
neuroscience
has
seen
significant
progress
workings
calculating
through
various
methods
model
systems.
This
review
begins
by
exploring
mental
neuronal
representations
nonsymbolic
then
progresses
acquired
childhood.
During
operations
(addition,
subtraction,
multiplication,
division),
these
processed
transformed
according
rules
principles,
leveraging
different
strategies
types
knowledge
that
can
be
dissociated
brain.
Although
it
was
once
believed
number
processing
calculation
originated
from
language
faculty,
is
now
evident
linguistic
primarily
independently
Understanding
how
healthy
processes
information
crucial
for
gaining
insights
into
debilitating
disorders,
including
conditions
like
acalculia
learning-related
disorders
such
as
developmental
dyscalculia.
