Frontiers in Computational Neuroscience,
Journal Year:
2018,
Volume and Issue:
12
Published: Nov. 13, 2018
To
infer
the
causes
of
its
sensations,
brain
must
call
on
a
generative
(predictive)
model.
This
necessitates
passing
local
messages
between
populations
neurons
to
update
beliefs
about
hidden
variables
in
world
beyond
sensory
samples.
It
also
entails
inferences
how
we
will
act.
Active
inference
is
principled
framework
that
frames
perception
and
action
as
approximate
Bayesian
inference.
has
been
successful
accounting
for
wide
range
physiological
behavioural
phenomena.
Recently,
process
theory
emerged
attempts
relate
their
neurobiological
substrates.
In
this
paper,
review
develop
anatomical
aspects
theory.
We
argue
form
models
required
constrains
way
which
regions
connect
one
another.
Specifically,
neuronal
representing
variable
receive
input
from
Markov
blanket
variable.
illustrate
idea
four
different
domains:
perception,
planning,
attention,
movement.
doing
so,
attempt
show
appealing
enables
us
account
architectures.
Ultimately,
committing
an
ensures
can
empirical
hypotheses
be
tested
using
neuroimaging,
neuropsychological,
electrophysiological
experiments.
Cerebral Cortex,
Journal Year:
2015,
Volume and Issue:
25(12), P. 4812 - 4827
Published: Aug. 12, 2015
On
the
50th
anniversary
of
Norman
Geschwind's
seminal
paper
entitled
'Disconnexion
syndrome
in
animal
and
man',
we
pay
tribute
to
his
ideas
by
applying
contemporary
tractography
methods
understand
white
matter
disconnection
3
classic
cases
that
made
history
behavioral
neurology.
We
first
documented
locus
extent
brain
lesion
from
computerized
tomography
Phineas
Gage's
skull
magnetic
resonance
images
Louis
Victor
Leborgne's
brain,
Broca's
patient,
Henry
Gustave
Molaison.
then
applied
reconstructed
lesions
an
atlas
connections
obtained
diffusion
129
healthy
adults.
Our
results
showed
all
patients,
disruption
extended
projecting
areas
distant
lesion.
confirmed
damaged
tracts
link
neuroscience
are
considered
functionally
engaged
for
tasks
related
emotion
decision-making
(Gage),
language
production
(Leborgne),
declarative
memory
(Molaison).
findings
suggest
even
historic
should
be
reappraised
within
a
framework
whose
principles
were
plainly
established
associationist
schools
last
2
centuries.
Physiological Reviews,
Journal Year:
2020,
Volume and Issue:
100(3), P. 1181 - 1228
Published: Feb. 20, 2020
For
more
than
one
century,
brain
processing
was
mainly
thought
in
a
localizationist
framework,
which
given
function
underpinned
by
discrete,
isolated
cortical
area,
and
with
similar
cerebral
organization
across
individuals.
However,
advances
mapping
techniques
humans
have
provided
new
insights
into
the
organizational
principles
of
anatomo-functional
architecture.
Here,
we
review
recent
findings
gained
from
neuroimaging,
electrophysiological,
as
well
lesion
studies.
Based
on
these
data
connectome,
challenge
traditional,
outdated
view
propose
an
alternative
meta-networking
theory.
This
model
holds
that
complex
cognitions
behaviors
arise
spatiotemporal
integration
distributed
but
relatively
specialized
networks
underlying
conation
cognition
(e.g.,
language,
spatial
cognition).
Dynamic
interactions
between
such
circuits
result
perpetual
succession
equilibrium
states,
opening
door
to
considerable
interindividual
behavioral
variability
neuroplastic
phenomena.
Indeed,
underlies
uniquely
human
propensity
learn
abilities,
also
explains
how
postlesional
reshaping
can
lead
some
degrees
functional
compensation
brain-damaged
patients.
We
discuss
major
implications
this
approach
fundamental
neurosciences
for
clinical
developments,
especially
neurology,
psychiatry,
neurorehabilitation,
restorative
neurosurgery.
Frontiers in Computational Neuroscience,
Journal Year:
2018,
Volume and Issue:
12
Published: Nov. 13, 2018
To
infer
the
causes
of
its
sensations,
brain
must
call
on
a
generative
(predictive)
model.
This
necessitates
passing
local
messages
between
populations
neurons
to
update
beliefs
about
hidden
variables
in
world
beyond
sensory
samples.
It
also
entails
inferences
how
we
will
act.
Active
inference
is
principled
framework
that
frames
perception
and
action
as
approximate
Bayesian
inference.
has
been
successful
accounting
for
wide
range
physiological
behavioural
phenomena.
Recently,
process
theory
emerged
attempts
relate
their
neurobiological
substrates.
In
this
paper,
review
develop
anatomical
aspects
theory.
We
argue
form
models
required
constrains
way
which
regions
connect
one
another.
Specifically,
neuronal
representing
variable
receive
input
from
Markov
blanket
variable.
illustrate
idea
four
different
domains:
perception,
planning,
attention,
movement.
doing
so,
attempt
show
appealing
enables
us
account
architectures.
Ultimately,
committing
an
ensures
can
empirical
hypotheses
be
tested
using
neuroimaging,
neuropsychological,
electrophysiological
experiments.