Royal Society Open Science,
Год журнала:
2025,
Номер
12(4)
Опубликована: Апрель 1, 2025
We
humans
are
capable
of
solving
challenging
planning
problems,
but
the
range
adaptive
strategies
that
we
use
to
address
them
is
not
yet
fully
characterized.
Here,
designed
a
series
problem-solving
tasks
require
at
different
depths.
After
systematically
comparing
performance
participants
and
models,
found
when
facing
problems
certain
number
subgoals
(from
1
8),
make
an
their
cognitive
resources—namely,
they
tend
select
initial
plan
having
minimum
required
depth,
rather
than
selecting
same
depth
for
all
problems.
These
results
support
view
as
bounded
rational
process,
which
adapts
costly
resources
task
demands.
Neuron,
Год журнала:
2022,
Номер
110(23), С. 3952 - 3969.e8
Опубликована: Окт. 12, 2022
Integrating
neurons
into
digital
systems
may
enable
performance
infeasible
with
silicon
alone.
Here,
we
develop
DishBrain,
a
system
that
harnesses
the
inherent
adaptive
computation
of
in
structured
environment.
In
vitro
neural
networks
from
human
or
rodent
origins
are
integrated
silico
computing
via
high-density
multielectrode
array.
Through
electrophysiological
stimulation
and
recording,
cultures
embedded
simulated
game-world,
mimicking
arcade
game
"Pong."
Applying
implications
theory
active
inference
free
energy
principle,
find
apparent
learning
within
five
minutes
real-time
gameplay
not
observed
control
conditions.
Further
experiments
demonstrate
importance
closed-loop
feedback
eliciting
over
time.
Cultures
display
ability
to
self-organize
activity
goal-directed
manner
response
sparse
sensory
information
about
consequences
their
actions,
which
term
synthetic
biological
intelligence.
Future
applications
provide
further
insights
cellular
correlates
The
predictive
nature
of
the
hippocampus
is
thought
to
be
useful
for
memory-guided
cognitive
behaviors.
Inspired
by
reinforcement
learning
literature,
this
notion
has
been
formalized
as
a
map
called
successor
representation
(SR).
SR
captures
number
observations
about
hippocampal
activity.
However,
algorithm
does
not
provide
neural
mechanism
how
such
representations
arise.
Here,
we
show
dynamics
recurrent
network
naturally
calculate
when
synaptic
weights
match
transition
probability
matrix.
Interestingly,
horizon
can
flexibly
modulated
simply
changing
gain.
We
derive
simple,
biologically
plausible
rules
learn
in
network.
test
our
model
with
realistic
inputs
and
data
recorded
during
random
foraging.
Taken
together,
results
suggest
that
more
accessible
circuits
than
previously
support
broad
range
functions.
Cell Reports,
Год журнала:
2024,
Номер
43(2), С. 113671 - 113671
Опубликована: Янв. 26, 2024
Outside
of
the
laboratory,
animals
behave
in
spaces
where
they
can
transition
between
open
areas
and
coverage
as
interact
with
others.
Replicating
these
conditions
laboratory
be
difficult
to
control
record.
This
has
led
a
dominance
relatively
simple,
static
behavioral
paradigms
that
reduce
ethological
relevance
behaviors
may
alter
engagement
cognitive
processes
such
planning
decision-making.
Therefore,
we
developed
method
for
controllable,
repeatable
interactions
others
reconfigurable
space.
Mice
navigate
large
honeycomb
lattice
adjustable
obstacles
an
autonomous
robot
coupled
their
actions.
We
illustrate
system
using
pseudo-predator,
delivering
airpuffs
mice.
The
combination
mobile
threat
elicits
diverse
set
behaviors,
increased
path
diversity,
peeking,
baiting,
providing
explore
ethologically
relevant
laboratory.
Nature Neuroscience,
Год журнала:
2024,
Номер
27(7), С. 1340 - 1348
Опубликована: Июнь 7, 2024
When
faced
with
a
novel
situation,
people
often
spend
substantial
periods
of
time
contemplating
possible
futures.
For
such
planning
to
be
rational,
the
benefits
behavior
must
compensate
for
spent
thinking.
Here,
we
capture
these
features
by
developing
neural
network
model
where
itself
is
controlled
prefrontal
cortex.
This
consists
meta-reinforcement
learning
agent
augmented
ability
plan
sampling
imagined
action
sequences
from
its
own
policy,
which
call
'rollouts'.
In
spatial
navigation
task,
learns
when
it
beneficial,
provides
normative
explanation
empirical
variability
in
human
thinking
times.
Additionally,
patterns
policy
rollouts
used
artificial
closely
resemble
rodent
hippocampal
replays.
Our
work
theory
how
brain
could
implement
through
prefrontal-hippocampal
interactions,
replays
are
triggered
by-and
adaptively
affect-prefrontal
dynamics.
The Innovation,
Год журнала:
2025,
Номер
6(2), С. 100755 - 100755
Опубликована: Янв. 6, 2025
Chimpanzees
(Pan
troglodytes)
are
one
of
humans'
closest
living
relatives,
making
them
the
most
directly
relevant
comparison
point
for
understanding
human
brain
evolution.
Zeroing
in
on
differences
connectivity
between
humans
and
chimpanzees
can
provide
key
insights
into
specific
evolutionary
changes
that
might
have
occurred
along
lineage.
However,
such
comparisons
hindered
by
absence
cross-species
atlases
established
within
same
framework.
To
address
this
gap,
we
developed
Chimpanzee
Brainnetome
Atlas
(ChimpBNA)
using
a
connectivity-based
parcellation
Leveraging
new
resource,
found
substantial
divergence
patterns
two
species
across
association
cortices,
notably
lateral
temporal
dorsolateral
prefrontal
cortex.
These
deviate
sharply
from
pattern
cortical
expansion
observed
when
comparing
to
chimpanzees,
highlighting
more
complex
nuanced
evolution
than
previously
recognized.
Additionally,
identified
regions
displaying
connectional
asymmetries
differed
species,
likely
resulting
divergence.
Genes
highly
expressed
divergent
connectivities
were
enriched
cell
types
crucial
projection
circuits
synapse
formation,
whose
pronounced
expression
hint
at
genetic
influences
neural
circuit
development,
function,
Our
study
provides
fine-scale
chimpanzee
atlas
highlights
chimpanzee-human
rigorous
comparative
manner.
In
addition,
these
results
suggest
potential
gene
correlates
species-specific
linking
neuroimaging
data,
offering
human-unique
cognitive
capabilities.
Trends in Cognitive Sciences,
Год журнала:
2022,
Номер
27(1), С. 65 - 80
Опубликована: Ноя. 26, 2022
Colombian
drug
lord
Pablo
Escobar,
while
on
the
run,
purportedly
burned
two
million
dollars
in
banknotes
to
keep
his
daughter
warm.
A
stark
reminder
that,
life,
circumstances
and
goals
can
quickly
change,
forcing
us
reassess
modify
our
values
on-the-fly.
Studies
decision-making
neuroeconomics
have
often
implicitly
equated
value
reward,
emphasising
hedonic
automatic
aspect
of
computation,
overlooking
its
functional
(concept-like)
nature.
Here
we
outline
computational
biological
principles
that
enable
brain
compute
usefulness
an
option
or
action
by
creating
abstractions
flexibly
adapt
changing
goals.
We
present
different
algorithmic
architectures,
comparing
ideas
from
artificial
intelligence
(AI)
cognitive
neuroscience
with
psychological
theories
and,
when
possible,
drawing
parallels.
Nature Communications,
Год журнала:
2023,
Номер
14(1)
Опубликована: Май 23, 2023
Abstract
Birds
of
the
crow
family
adapt
food-caching
strategies
to
anticipated
needs
at
time
cache
recovery
and
rely
on
memory
what,
where
when
previous
caching
events
recover
their
hidden
food.
It
is
unclear
if
this
behavior
can
be
explained
by
simple
associative
learning
or
it
relies
higher
cognitive
processes
like
mental
time-travel.
We
present
a
computational
model
propose
neural
implementation
behavior.
The
has
hunger
variables
for
motivational
control,
reward-modulated
update
retrieval
policies
an
network
remembering
with
consolidation
mechanism
flexible
decoding
age
memory.
Our
methodology
formalizing
experimental
protocols
transferable
other
domains
facilitates
evaluation
experiment
design.
Here,
we
show
that
memory-augmented,
reinforcement
without
time-travel
sufficient
explain
results
28
behavioral
experiments
birds.