Differences in Information-Seeking Behavior Between Typically-Developing Children and Children with Autism Spectrum Disorder

Autism spectrum disorder (ASD) is characterized by restricted, repetitive
behaviors and interests (RRBs) which results in a reduced tendency to explore
and examine their surrounding environments. These atypical patterns of
exploration might lead autistic children to miss or avoid learning
opportunities offered by the environment. Recently, it has been suggested that
typically-developing (TD) individuals tailor their attention to maximize the
information they can obtain from the environment. Infants look longer to
stimuli that offer more opportunity for learning. Likewise, TD adults engage
longer in environments where they can learn the most from, and their
exploration of new environments is guided by how much information they expect
to gain from them. However, it remains unclear whether the explorative behavior
of autistic individuals is similarly guided by learning progress. Our study
aims to elucidate which mechanisms underlie the explorative behavior of
autistic children, and whether underlying differences in these mechanisms might
explain some of the atypical exploration patterns commonly reported in ASD. We
hypothesize that learning progress might guide exploratory behavior of children
with autism differently compared to TD children and that children with autism
might remain for longer in environments even as learning progress decreases.

The main objective of this study is to investigate whether there are
differences in the way factors such as learning progress (LP) maximization and
prediction error (PE) minimization drive the exploratory behavior of children
with autism compared to TD children. Using a behavioral task [motion-pattern
learning task] implemented as a tablet game and computational modelling
techniques we want to investigate how ASD and TD children learn in different
offered environments and the learning opportunities they seek. The second
objective of this study is to relate potential differences in the way LP and PE
drive exploration in children with ASD in relation to their particular score
profiles of repetitive, restricted behaviors.

The proposed study is a behavioral study in which children will be presented
with a motion-pattern learning task with different environments containing
learnable sequences that will vary in the amount of learning progress children
can make. Children*s behavioral responses will be recorded to assess accuracy
(i.e., prediction error: PE), duration of play and environment selection. A
hierarchical reinforcement learning model (Velázquez et al., 2019) will be used
to quantify children*s learning-related measures for task engagement and future
environment choice such as learning progress (LP), expected learning progress
(ELP) and expected prediction error (EPE) on the basis of children*s
performance measures. Based on these learning-related measures, we will test
our predictions on what drives children*s engagement and continuous play with
an environment or switch to a different one. As well as what explains what they
chose to explore next.

Our study requires one visit to the Baby & Child Research Center (BRC) at the
Donders Centre for Cognition, with a time investment of about one hour. We will
present children with a behavioral task (touch-game on a tablet) and record
their responses. Additionally to the experimental task, during the test
session, children will undergo an IQ test (duration of 50 minutes). Previous to
the test session, parents will be asked to complete three standardized
questionnaires which will be available to them in online form through
LimeSurvey (duration of around 30 minutes). The risks associated with these
measurements are negligible, no adverse events are expected, and the burden for
the participants is considered to be minimal. Stimuli have been chosen to be
interesting and pleasurable for children.

We chose to include minors of 6-8 years old in our study because of several
reasons. First, ASD is a developmental condition which is characterized by its
early onset. Although experienced clinicians are able to reliably diagnose ASD
in children as young as 2 years old, the average age when children are
identified with ASD in Netherlands is of 56-116 months (van*t Hof et al.,
2021). We are interested in finding out how emerging autistic traits impact
curiosity-driven exploration in a developing population. In addition, several
studies have pointed at developmental changes in the manifestation of RRBs in
ASD children, where motor and sensory stereotypies decrease with age, but more
complex RRBs such as insistence of sameness, restricted interests and
ritualistic behaviors become more prevalent as age increases (Bishop et al.,
2006; Esbensen et al., 2009). We thus decided to test a sample of children in
an age range (6 to 8 years old) where a clinical ASD diagnosis is already
present, as well as the manifestation of more complex, high-order RRBs.
Finally, this age group is also easily accessible for our research, which
enables us to recruit a sample size that will allow us to come to reliable
conclusions.