Compulsivity is linked to suboptimal choice variability but unaltered reinforcement learning under uncertainty
Nature Mental Health,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 6, 2025
Язык: Английский
Computational processes of simultaneous learning of stochasticity and volatility in humans
Nature Communications,
Год журнала:
2024,
Номер
15(1)
Опубликована: Окт. 21, 2024
Making
adaptive
decisions
requires
predicting
outcomes,
and
this
in
turn
adapting
to
uncertain
environments.
This
study
explores
computational
challenges
distinguishing
two
types
of
noise
influencing
predictions:
volatility
stochasticity.
Volatility
refers
diffusion
latent
causes,
requiring
a
higher
learning
rate,
while
stochasticity
introduces
moment-to-moment
observation
reduces
rate.
Dissociating
these
effects
is
challenging
as
both
increase
the
variance
observations.
Previous
research
examined
factors
mostly
separately,
but
it
remains
unclear
whether
how
humans
dissociate
them
when
they
are
played
off
against
one
another.
In
large-scale
experiments,
through
behavioral
prediction
task
modeling,
we
report
evidence
dissociating
solely
based
on
their
We
observed
contrasting
rates,
consistent
with
statistical
principles.
These
results
model
that
estimates
by
balancing
dueling
effects.
Adaptive
difficult
noisy
environments,
yet
people
often
succeed.
Here,
authors
show
do
between
easily
confused
noise—volatility
stochasticity—which
require
opposite
adjustments
learning.
Язык: Английский
Nonlinear modulation of human exploration by distinct sources of uncertainty
bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2025,
Номер
unknown
Опубликована: Апрель 21, 2025
Abstract
Decision-making
in
uncertain
environments
requires
balancing
exploration
and
exploitation,
with
typically
assumed
to
increase
monotonically
uncertainty.
Challenging
this
prevailing
assumption,
we
demonstrate
a
more
complex
relationship
by
decomposing
environmental
uncertainty
into
volatility
(systematic
change
reward
contingencies,
learnable)
stochasticity
(random
noise
observations,
unlearnable).
Across
two
behavioral
experiments
(N=1001,
N=747)
using
probabilistic
task,
find
robust
U-shaped
between
the
volatility-to-stochasticity
(
v
/
s
)
ratio
exploratory
behavior,
participants
exploring
when
either
or
dominates.
Remarkably,
pattern
extends
real-world
financial
as
demonstrated
through
analysis
of
five
years
S&P
500
stock
market
data,
where
portfolio
diversity
(a
proxy
for
exploration)
shows
same
price
movements
driven
fundamental
factors,
e.g.,
economic
shifts)
relative
trading
fluctuations
from
activity
unrelated
fundamentals).
These
findings
reveal
how
humans
adaptively
modulate
strategies
based
on
qualitative
composition
uncertainty,
optimal
performance
occurring
at
intermediate
ratios.
This
nonlinear
has
important
implications
understanding
decision-making
across
domains
arises
multiple
sources.
Язык: Английский
Reward-based option competition in human dorsal stream and transition from stochastic exploration to exploitation in continuous space
Science Advances,
Год журнала:
2024,
Номер
10(8)
Опубликована: Фев. 23, 2024
Primates
exploring
and
exploiting
a
continuous
sensorimotor
space
rely
on
dynamic
maps
in
the
dorsal
stream.
Two
complementary
perspectives
exist
how
these
encode
rewards.
Reinforcement
learning
models
integrate
rewards
incrementally
over
time,
efficiently
resolving
exploration/exploitation
dilemma.
Working
memory
buffer
explain
rapid
plasticity
of
parietal
but
lack
plausible
policy.
The
reinforcement
model
presented
here
unifies
both
accounts,
enabling
rapid,
information-compressing
map
updates
efficient
transition
from
exploration
to
exploitation.
As
predicted
by
our
model,
activity
human
frontoparietal
stream
regions,
not
MT+,
tracks
number
competing
options,
as
preferred
options
are
selectively
maintained
map,
while
spatiotemporally
distant
alternatives
compressed
out.
When
valuable
new
uncovered,
posterior
β
1
/α
oscillations
desynchronize
within
0.4
0.7
s,
consistent
with
option
encoding
-stabilized
subpopulations.
Together,
outcomes
matching
locally
cached
reward
representations
rapidly
update
maps,
biasing
choices
toward
often-sampled,
rewarded
options.
Язык: Английский
Characterizing the role of unpredictability within different dimensions of early life adversity
Development and Psychopathology,
Год журнала:
2024,
Номер
unknown, С. 1 - 15
Опубликована: Окт. 1, 2024
Abstract
Dimensional
models
of
early
life
adversity
highlight
the
distinct
roles
deprivation
and
threat
in
shaping
neurocognitive
development
mental
health.
However,
relatively
little
is
known
about
role
unpredictability
within
each
dimension.
We
estimated
both
average
levels
of,
temporal
exposure
during
adolescence
a
high-risk,
longitudinal
sample
1354
youth
(Pathways
to
Desistance
study).
then
related
these
estimates
later
psychological
distress,
Antisocial
Borderline
personality
traits,
tested
whether
any
effects
are
mediated
by
future
orientation.
High
were
found
be
associated
with
worse
health
on
all
three
outcomes,
but
only
traits
decreased
orientation,
pattern
consistent
evolutionary
psychopathology.
Unpredictability
proved
increased
distress
higher
number
There
was
some
evidence
buffering
against
detrimental
developmental
levels.
Our
results
suggest
that
different
dimensions
adversity.
Язык: Английский
Computational processes of simultaneous learning of stochasticity and volatility in humans
Опубликована: Июль 13, 2023
Adapting
to
uncertain
environments
is
crucial
for
survival.
This
study
explores
computational
challenges
in
distinguishing
two
types
of
noise:
volatility
and
stochasticity.
Volatility
refers
diffusion
noise
latent
causes,
requiring
a
higher
learning
rate,
while
stochasticity
introduces
moment-to-moment
observation
reduces
rate.
For
the
learner,
dissociating
their
effects
on
one’s
observations
challenging
because
they
both
increase
variance
observations.
Previous
research
examined
these
factors
separately,
but
it
remains
unclear
whether
how
humans
dissociate
them.
In
large-scale
experiments,
through
novel
behavioral
tasks
modeling,
we
report
compelling
evidence
solely
based
We
observed
contrasting
rates,
consistent
with
statistical
principles.
These
results
are
model
that
estimates
by
balancing
dueling
effects,
not
number
other
models
fail
make
this
distinction.
elucidates
processes
behind
adaptive
environments.
Язык: Английский
Fluctuations in sequential many-alternative decisions reveal strategies beyond immediate reward maximisation
Опубликована: Фев. 16, 2024
Humans
are
strategic
animals.
We
constantly
make
prospective
choices,
allocating
limited
resources
in
situations
of
uncertain,
future
outcomes.
The
management
our
finite
monthly
budget,
financial
investments,
or
the
allocation
time
to
different
questions
an
exam
just
a
few
examples.
In
these
scenarios,
both
decision-making
and
resource
tend
fluctuate
over
even
under
invariable
set
constraints.
However,
it
is
unclear
whether
fluctuations
affect
performance
they
underlie
additional
objectives
beyond
pure
reward
maximisation.
address
using
breadth-depth
dilemma,
novel
ecological
protocol
where
participants
engage
sequential
multiple-choice
scenarios
characterised
by
capacity.
designed
two
experimental
environments.
one
environment,
optimal
performance,
formalised
with
ideal
allocator
model,
associated
homogeneous
across
consecutive
choices.
contrast,
other
environment
entails
that
fluctuating
leads
greater
expected
rewards.
Our
study
evaluates
participants'
adherence
measures
as
deviation
from
allocations.
results
revealed
participants’
behaviour
fluctuates
more
than
optimal,
but
critically,
behavioural
adapt
available
capacity
environmental
context.
Moreover,
findings
unveil
pronounced
strategies,
such
save-for-later
history-dependent
choice,
further
implying
strategies
contribute
decision
variability.
An
extension
model
showed
characteristic
excess
fluctuation
driven
entropy
seeking,
pursuit
information-gain
risk
avoidance.
Although
having
modest
impact
on
may
reflect
advantageous
behaviours
long
run
ever
changing
real-world
Язык: Английский
The relationship between temporal discounting and foraging
Current Psychology,
Год журнала:
2024,
Номер
unknown
Опубликована: Сен. 19, 2024
Язык: Английский
Prediction of Potential Evapotranspiration via Machine Learning and Deep Learning for Sustainable Water Management in the Murat River Basin
Sustainability,
Год журнала:
2024,
Номер
16(24), С. 11077 - 11077
Опубликована: Дек. 17, 2024
Potential
evapotranspiration
(PET)
is
a
significant
factor
contributing
to
water
loss
in
hydrological
systems,
making
it
critical
area
of
research.
However,
accurately
calculating
and
measuring
PET
remains
challenging
due
the
limited
availability
comprehensive
data.
This
study
presents
detailed
sustainable
model
for
predicting
using
Thornthwaite
equation,
which
requires
only
mean
monthly
temperature
(Tmean)
latitude,
with
calculations
performed
R-Studio.
A
geographic
information
system
(GIS)
was
employed
interpolate
meteorological
data,
ensuring
coverage
all
sub-basins
within
Murat
River
basin,
area.
Additionally,
Python
libraries
were
utilized
implement
artificial
intelligence-driven
models,
incorporating
both
machine
learning
deep
techniques.
The
harnesses
power
intelligence
(AI),
applying
through
convolutional
neural
network
(CNN)
techniques,
including
support
vector
(SVM)
random
forest
(RF).
results
demonstrate
promising
performance
across
models.
For
CNN,
coefficient
determination
(R2)
varied
from
96.2
98.7%,
squared
error
(MSE)
ranged
0.287
0.408,
root
(RMSE)
between
0.541
0.649.
SVM,
R2
94.5
95.6%,
MSE
0.981
1.013,
RMSE
0.990
1.014.
RF
showed
best
performance,
achieving
an
100%,
values
0.326
0.640,
corresponding
0.571
0.800.
climate
topography
data
used
algorithms
consistent,
indicate
that
outperforms
others.
Consequently,
model’s
superior
accuracy
highlights
its
potential
as
reliable
tool
prediction,
supporting
informed
decision-making
resource
planning.
By
leveraging
GIS,
AI,
learning,
this
enhances
modeling
methodologies,
addressing
management
challenges
promoting
practices
face
change
limitations.
Язык: Английский