Open Mind,
Год журнала:
2024,
Номер
8, С. 1348 - 1368
Опубликована: Янв. 1, 2024
Memory
systems
constantly
confront
the
challenge
of
capturing
both
shared
features
that
connect
experiences
together
and
unique
distinguish
them.
Across
two
experiments,
we
leveraged
a
color
memory
distortion
paradigm
to
investigate
how
handle
this
representational
tension
when
learning
new
information.
Over
thirty-minute
period,
participants
learned
categories
novel
objects,
where
each
feature
was
assigned
particular
color.
While
did
not
differ
in
accurately
they
remembered
these
overall,
inaccurate,
misremembered
(relative
unique)
as
more
similar
category's
average
color,
suggesting
integration
memory.
This
same
rapid
warping
manifested
neural
network
model
trained
on
categories.
The
work
reveals
memories
for
different
are
rapidly
differentially
warped
function
their
roles
category.
Our
representations
of
the
world
need
to
be
stable
enough
support
general
knowledge
but
flexible
incorporate
new
information
as
our
environment
changes.
How
does
human
brain
manage
this
stability-plasticity
trade-off?
We
analyzed
a
large
dataset
in
which
participants
viewed
objects
embedded
thousands
natural
scenes
across
many
fMRI
sessions.
Semantic
item
were
located
by
jointly
leveraging
voxelwise
encoding
model
find
reliable
and
word-embedding
evaluate
semantic
content.
Within
medial
temporal
lobe,
hippocampal
subfield
CA1,
parahippocampal
cortex,
perirhinal
cortex
gradually
drifted
period
multiple
months.
Whole-brain
analyses
revealed
gradient
plasticity
with
drift
more
evident
anterior
than
posterior
areas.
On
short
timescales,
rapid
was
observed
only
such
that
co-occurrence
statistics
warped
within
single
session.
Together,
results
suggest
solves
trade-off
through
regions.
Open Mind,
Год журнала:
2024,
Номер
8, С. 1348 - 1368
Опубликована: Янв. 1, 2024
Memory
systems
constantly
confront
the
challenge
of
capturing
both
shared
features
that
connect
experiences
together
and
unique
distinguish
them.
Across
two
experiments,
we
leveraged
a
color
memory
distortion
paradigm
to
investigate
how
handle
this
representational
tension
when
learning
new
information.
Over
thirty-minute
period,
participants
learned
categories
novel
objects,
where
each
feature
was
assigned
particular
color.
While
did
not
differ
in
accurately
they
remembered
these
overall,
inaccurate,
misremembered
(relative
unique)
as
more
similar
category's
average
color,
suggesting
integration
memory.
This
same
rapid
warping
manifested
neural
network
model
trained
on
categories.
The
work
reveals
memories
for
different
are
rapidly
differentially
warped
function
their
roles
category.