Abstract the participants to determine which way the


This study
assessed the accuracy and reaction times of two groups of typically-developed (individuals
that have met all the developmental milestones) adult participants, using a
motion coherence task. The 116 participants were split into two groups based on
their autism trait score (high vs. low), determined using the Autism Spectrum
Quotient. The motion coherence task required the participants to determine
which way the majority of a group of dots were moving in a random dot
Kinematogram. The high autism trait group showed significantly less accuracy
than the low autism trait group (meaning they have a high motion coherence
threshold), but no difference in reaction times was detected. The findings
suggest that individuals with autistic traits may struggle to process/detect coherent
motion. Future work should investigate where on the autistic spectrum motion
coherence begins to deteriorate and whether the change is abrupt or gradual, so
to better provide educational support for autistic individuals.  


Carpenter and El-Ghoroury (2017) define autism as a severe neurodevelopmental
disorder that characterizes itself by affecting a person’s social functioning
and communication, and causing repetitive behaviour patterns.

One of the
drivers of these effects is a decrease in visual perception ability, which can
present itself as a form of agnosia or the inability to detect whole shapes.
The perception difference of autistic individuals and ‘normal’ individuals was
investigated using various visual stimuli in either whole or partial form e.g.
whole shapes and shapes in multiple parts (Shah and Frith, 1993). The
participants with autism performed significantly better than the control group
(normal) when examining the whole shapes, suggesting that they can see the finer
details in whole shapes better than those without autism. This finding was the
catalyst for the development of the Weak Central Coherence Theory (Firth,
2003), which states that individuals with autism focus on the local information
(finer details) rather than global information (overall picture).

One way to
measure how well individuals view global and local information is using a
motion coherence task (i.e. random dot Kinematogram). During a motion coherence
task, participants view a group of moving dots and must judge which direction
the majority are moving in. A participant’s motion coherence threshold is
determined from the minimum number of moving dots that they can still correctly
identify as moving in a single direction, amongst a large group of randomly/erratically
moving dots. Milne et al. (2002)
conducted a motion coherence task with a control group and a group of autistic
children. The autistic group had a higher motion coherence threshold than the
control group, leading to the suggestion that individuals with autism perform
worse in motion coherence tasks than non-autistic individuals. Manning (2015),
however, has challenged this theory after conducting a similar experiment (with
a similar age group) to that of `Milne et
al. (2002), which concluded that children with autism could tolerate wider
variability when judging the direction of the majority of the dots. Meaning
that autistic children should perform better when participating in tasks
connected to motion coherence than typically-developed children.

studies have focused on comparing formally diagnosed autistic individuals with
a control (non-autistic) group. To date no study has looked for perception
differences in a typically-developed population sample, which left a gap in the
knowledge for the current study to fill, by starting to determine where
significant perception differences begin to appear on the autistic spectrum. In
accordance with the majority of previous studies using autistic individuals,
the current study predicted that motion coherence discrimination will be
significantly worse for typically-developed individuals who have higher autism
trait scores. The two specific hypotheses tested were that accuracy will be
significantly lower for individuals with higher autism scores, and reaction
times will be significantly longer for individuals with higher autism scores.
Both hypotheses were tested using an Autism Spectrum Quotient and a Motion
Coherence Task.




participants consisted of 116 typically-developed, first-year-psychology
students from a UK university, with a mean age of 20.7 years (SD = 5.3), who
were required to participate as a condition of their psychology degree course.
Each participant used the last three digits of their student number so the
experimenter could track the results, and therefore the results for each test were


took the Autism Spectrum Quotient (ASQ) using an online server, and a motion
coherence task (random dot Kinematogram) conducted using a piece of
experimental software called PsychoPy (Peirce, 2007).            

Experimental Design

experiment was a between subject’s design, and had one independent variable
(the group of participants) with two levels (high and low autism traits), and
two dependent variables (test accuracy and reaction time). Both dependent
variables were measured simultaneously. 


First, all
participants completed the ASQ and then were given the instructions for the
Motion Coherence Task prior to the practice trials. All participants were
instructed to press the left or right arrow keys of a computer keyboard,
corresponding to the direction they see the majority of a large (40+) group of
dots moving in. Each participant completed 80 practice trials. Each trial began
with a brief fixation cross (present for 200 ?s) to focus the participant’s
attention, before the dot stimuli were presented (for 1500 ?s per trial).      

participants then moved on to the experimental task, which consisted of five
groups of 40 trials (200 trials in total). The software recorded whether the
participant was correct (accuracy) and how much time it took them to respond to
the stimuli (in ?s).

Data Analyses

The scores
from the Autism Spectrum Quotient (ASQ) were ordered and divided into three
groups: lowest 1/3 of scores, middle 1/3 of scores and highest 1/3of scores.
The middle 1/3 of the scores were then discarded, leaving the high autism trait
group (highest 1/3 of scores) and low autism trait group (lowest 1/3 of
scores). Outlying data points, determined as those laying too far from the
mean, were excluded. Independent samples two-tailed t-tests were used to
determine whether the high and low autism trait groups scored differently in
both their accuracy and response times. All means are expressed ± standard



There was
a significant decrease in accuracy for the high autism trait group (74.86 ±
12.33) compared to that of the low autism trait group (81.10 ± 11.25)
conditions; t(114) = 2.83, p = 0.005. In contrast, no significant difference in
response time scores was found between groups; low autism trait group (623.14 ±
112.54 ?s) and high autism trait group (654.44 ± 134.03 ?s) conditions; t(114)
= 1.35, p = 0.117.



participants with higher autism traits had less accuracy than the lower autism
trait group, and therefore have a higher motion coherence threshold. These
findings support the first hypothesis; accuracy will be lower for individuals
with higher autism scores. There was, however, no detectable effect of autism
score on reaction times, which is contrary to the second hypothesis tested
(reaction times will be significantly longer for individuals with higher autism
scores). The results support previous work that found children with autism had
less accuracy when completing a motion coherence task than the
typically-developed children (Milne et al,
2002), but expands the concept, and therefore doesn’t support Manning’s (2015)
theory that children with autism perform better. Specifically, motion coherence
thresholds may be affected by autism score within typically developed
individuals and not just a trait of autistic individuals per se.                                                                                                                                                                    

?n effect
on response time, specifically an increase in response time for the high autism
trait group, would have indicated a difference in attention span (Robertson et al, 2014). The reason why autism
scores were only found to affect accuracy and not response times here could be
because both of the groups tested were typically-developed, meaning there
should be no major difference between the two (Robertson et al, 2014). The results could suggest that response is binary,
whereas accuracy is a continuous spectrum.  

findings of this study could also be the result of the methodology used, for
example, it is relatively unknown how reliable the ASQ is when differentiating
between Autism, Asperger’s, and Attention Deficit and Hyperactivity Disorder
(Sizoo et al, 2009). It was found
that out of 129 participants tested, using the Autism Spectrum Quotient, only
73% of classifications were correct (when the results were compared to the participant’s
former diagnosis’s), indicating that the ASQ may lack resolution (Sizoo et al, 2009). However, as the
participants in the current experiment were all typically-developed and the ASQ
wasn’t being used as a diagnostic tool, and the distance between the ASQ scores
of the two groups was broad (achieved by disregarding participants that
produced the middle third of scores), the results are less likely to be
impacted by the resolution limitations of the ASQ.

research could expand on this study by comparing a typically-developed, high
autism trait group with a diagnosed autistic group to try and determine where
on the autistic spectrum accuracy dramatically changes. This could be useful
knowledge in providing adequate support for children with autism in education
settings, by broadening the understanding of how they view the world. In
summary, a significant decrease in accuracy was found for the high autism trait
group, but no effect of ASQ scores was found for the reaction times. Therefore,
the first hypothesis is supported, but the second experimental hypothesis must
be rejected. The data collected supports previous theories e.g. Milne et al. (2002), but the ASQ may not
always accurately define autism level. Nonetheless, the data collected provides
a good foundation for future research to build on.  



Frith, U.
(2003). Autism: Explaining the enigma. Oxford: Blackwells. Retrieved from

C., Tibber, M. S., Charman, T., Dakin, S. C., & Pellicano, E. (2015).
Enhanced integration of motion coherence thresholds in children with autism.
Journal of Neuroscience, 35(18), 6979-6986.

Milne, E.,
Swetten, J., Hansen, P., Campbell, R., Jeffries, H., & Plaisted, K. (2002).
High motion coherence thresholds in children with autism. Journal of Child
Psychology and Psychiatry, and Allied Disciplines, 43(2), 255-263.

J.W. (2007). PsychoPy – Psychophysics software in Python.  Journal of Neuroscience Methods, 162(1-2),
8-13. https://doi.org/10.1016/j.jneumeth.2006.11.017

C. E., Thomas, C., Kravitz, D. J., Wallace, G. L., Baron-Cohen, S., Martin, A.,
& Baker, C. I. (2014). Global motion perception deficits in autism are
reflected as early as primary visual cortex. Brain, 137(9), 2588-2599.

Shah, A.,
& Frith, U. (1993). Why Do Autistic Individuals Show Superior Performance
on the Block Design Task? Journal of Child Psychology and Psychiatry, 34(8),
1351-1364. doi:10.1111/j.1469-7610.1993.tb02095.x

Sizoo, B.
B., Brink, W. V., Eenige, M. G., Koeter, M. W., Wijngaarden-Cremers, P. J.,
& Gaag, R. J. (2009). Using the Autism-Spectrum Quotient to Discriminate
Autism Spectrum Disorder from ADHD in Adult Patients With and Without Comorbid
Substance Use Disorder. Journal of Autism and Developmental Disorders, 39(9),
1291-1297. doi:10.1007/s10803-009-0743-2

L., Carpenter, L. A., & El-Ghoroury, N. H. (2017, January). Diagnosing and
managing autism spectrum disorder (ASD). Retrieved December 15, 2017, from

Written by