An
animal
entering
a
new
environment
typically
faces
three
challenges:
explore
the
space
for
resources,
memorize
their
locations,
and
navigate
towards
those
targets
as
needed.
Here
we
propose
neural
algorithm
that
can
solve
all
these
problems
operates
reliably
in
diverse
complex
environments.
At
its
core,
mechanism
makes
use
of
behavioral
module
common
to
motile
animals,
namely
ability
follow
an
odor
source.
We
show
how
brain
learn
generate
internal
“virtual
odors”
guide
any
location
interest.
This
endotaxis
be
implemented
with
simple
3-layer
circuit
using
only
biologically
realistic
structures
learning
rules.
Several
components
this
scheme
are
found
brains
from
insects
humans.
Nature
may
have
evolved
general
search
navigation
on
ancient
backbone
chemotaxis.
Current Biology,
Год журнала:
2022,
Номер
32(17), С. 3676 - 3689.e5
Опубликована: Июль 20, 2022
tested
humans,
rats,
and
RL
agents
on
a
novel
modular
maze
d
Humans
rats
were
remarkably
similar
in
their
choice
of
trajectories
Both
species
most
to
utilizing
SR
also
displayed
features
model-based
planning
early
trials
Progress in Neurobiology,
Год журнала:
2024,
Номер
240, С. 102653 - 102653
Опубликована: Июль 2, 2024
We
present
here
a
view
of
the
firing
patterns
hippocampal
cells
that
is
contrary,
both
functionally
and
anatomically,
to
conventional
wisdom.
argue
hippocampus
responds
efference
copies
goals
encoded
elsewhere;
it
uses
these
detect
resolve
conflict
or
interference
between
in
general.
While
can
involve
space,
do
not
encode
spatial
(or
other
special
types
of)
memory,
as
such.
also
transverse
circuits
operate
an
essentially
homogeneous
way
along
its
length.
The
apparently
different
functions
parts
(e.g.
memory
retrieval
versus
anxiety)
result
from
(situational/motivational)
inputs
on
which
those
perform
same
fundamental
computational
operations.
On
this
view,
key
role
iterative
adjustment,
via
Papez-like
circuits,
synaptic
weights
cell
assemblies
elsewhere.
The
hippocampus
is
thought
to
enable
the
encoding
and
retrieval
of
ongoing
experience,
organization
that
experience
into
structured
representations
like
contexts,
maps,
schemas,
use
these
structures
plan
for
future.
A
central
goal
understand
what
core
computations
supporting
functions
are,
how
are
realized
in
collective
action
single
neurons.
potential
access
point
this
issue
provided
by
‘splitter
cells’,
hippocampal
neurons
fire
differentially
on
overlapping
segment
trajectories
differ
their
past
and/or
However,
literature
splitter
cells
has
been
fragmented
confusing,
owing
differences
terminology,
behavioral
tasks,
analysis
methods
across
studies.
In
review,
we
synthesize
consistent
findings
from
literature,
establish
a
common
set
terms,
translate
between
single-cell
ensemble
perspectives.
Most
importantly,
examine
combined
through
lens
two
major
theoretical
ideas
about
function:
representation
temporal
context
latent
state
inference.
We
find
unique
signature
properties
each
models
necessary
account
data,
but
neither
theory,
itself,
explains
all
its
features.
Specifically,
gradedness
signal
strong
support
context,
hard
explain
using
models,
while
flexibility
task-dependence
naturally
accounted
inference,
poses
challenge
otherwise.
These
theories
suggest
number
avenues
future
work,
believe
application
timely
informative
domain
testing
refining
function.
Annals of the New York Academy of Sciences,
Год журнала:
2024,
Номер
1534(1), С. 45 - 68
Опубликована: Март 25, 2024
Abstract
This
paper
considers
neural
representation
through
the
lens
of
active
inference,
a
normative
framework
for
understanding
brain
function.
It
delves
into
how
living
organisms
employ
generative
models
to
minimize
discrepancy
between
predictions
and
observations
(as
scored
with
variational
free
energy).
The
ensuing
analysis
suggests
that
learns
navigate
world
adaptively,
not
(or
solely)
understand
it.
Different
may
possess
an
array
models,
spanning
from
those
support
action‐perception
cycles
underwrite
planning
imagination;
namely,
explicit
entail
variables
predicting
concurrent
sensations,
like
objects,
faces,
or
people—to
action‐oriented
predict
action
outcomes.
then
elucidates
belief
dynamics
might
link
implications
different
types
agent's
cognitive
capabilities
in
relation
its
ecological
niche.
concludes
open
questions
regarding
evolution
development
advanced
abilities—and
gradual
transition
pragmatic
detached
representations.
on
offer
foregrounds
diverse
roles
play
processes
representation.
Proceedings of the National Academy of Sciences,
Год журнала:
2025,
Номер
122(4)
Опубликована: Янв. 23, 2025
Efficient
planning
is
a
distinctive
hallmark
of
intelligence
in
humans,
who
routinely
make
rapid
inferences
over
complex
world
contexts.
However,
studies
investigating
how
humans
accomplish
this
tend
to
focus
on
naive
participants
engaged
simplistic
tasks
with
small
state
spaces,
which
do
not
reflect
the
intricacy,
ecological
validity,
and
human
specialization
real-world
planning.
In
study,
we
examine
street-by-street
route
London
taxi
drivers
navigating
across
more
than
26,000
streets
(United
Kingdom).
We
explore
unfolded
dynamically
different
phases
journey
construction
identify
theoretic
principles
by
these
expert
planners
rationally
precache
decisions
at
prioritized
environment
states
an
early
phase
process.
particular,
find
that
measures
path
complexity
predict
mental
sampling
prioritization
dynamics
independent
alternative
derived
from
real
spatial
context
being
navigated.
Our
data
provide
evidence
for
complexity-driven
remote
access
within
internal
models
precaching
during
very
large
structured
spaces.
Nature Communications,
Год журнала:
2023,
Номер
14(1)
Опубликована: Сен. 20, 2023
Neural
processing
of
a
desired
moving
direction
requires
the
continuous
comparison
between
current
heading
and
goal
direction.
While
neural
basis
underlying
is
well-studied,
coding
remains
unclear
in
insects.
Here,
we
used
tetrode
recordings
tethered
flying
monarch
butterflies
to
unravel
how
represented
insect
brain.
recording,
maintained
robust
directions
relative
virtual
sun.
By
resetting
their
directions,
found
neurons
whose
spatial
tuning
was
tightly
linked
directions.
Importantly,
unaffected
when
changed
after
compass
perturbations,
showing
that
these
specifically
encode
Overall,
here
discovered
invertebrate
goal-direction
share
functional
similarities
cells
reported
mammals.
Our
results
give
insights
into
evolutionarily
conserved
principles
goal-directed
orientation
animals.
Neuron,
Год журнала:
2023,
Номер
111(23), С. 3885 - 3899.e6
Опубликована: Сен. 18, 2023
Humans
can
navigate
flexibly
to
meet
their
goals.
Here,
we
asked
how
the
neural
representation
of
allocentric
space
is
distorted
by
goal-directed
behavior.
Participants
navigated
an
agent
two
successive
goal
locations
in
a
grid
world
environment
comprising
four
interlinked
rooms,
with
contextual
cue
indicating
conditional
dependence
one
location
on
another.
Examining
geometry
which
room
and
context
were
encoded
fMRI
signals,
found
that
map-like
representations
emerged
both
hippocampus
neocortex.
Cognitive
maps
orbitofrontal
cortices
compressed
so
cued
as
goals
coded
together
state
space,
these
distortions
predicted
successful
learning.
This
effect
was
captured
computational
model
current
prospective
are
jointly
place
code,
providing
theory
warp
macroscopic
signals.
As
we
learn,
dynamic
memory
processes
build
structured
knowledge
across
our
experiences.
Such
enables
the
formation
of
internal
models
world
that
use
to
plan,
make
decisions,
and
act.
Recent
theorizing
posits
mnemonic
mechanisms
differentiation
integration
–
which
at
one
level
may
seem
be
odds
both
contribute
emergence
knowledge.
We
tested
this
possibility
using
fMRI
as
human
participants
learned
navigate
within
local
global
virtual
environments
over
course
3
days.
Pattern
similarity
analyses
on
entorhinal
cortical
hippocampal
patterns
revealed
evidence
work
concurrently
environmental
representations,
variability
in
relates
differences
navigation
efficiency.
These
results
offer
new
insights
into
neural
machinery
underlying
translate
experiences
allows
us
achieve
goals.