Tip:
Highlight text to annotate it
X
>>DR. ELLAWADI: What I'm going to be talking about today is category development in children with autism.
And before I start, I want to take some time to acknowledge where this data came from.
Dr. Letitia Naigles of the University of Connecticut was kind enough to share
the data she had collected with me for this analysis. This work came from
Dr. Naigles and Dr. Fein and it was collected by the Child Language Lab at
the University of Connecticut. It was collected by Caitlin Reynolds, Christian Navaro-Torres,
and Janima Piotroski. This was funded by a grant that was awarded to
Letitia Naigles from the National Institute on Deafness and Communication Disorders.
To give you guys just a little bit of background, I'm sure everybody knows this but
it's good to review, autism is a neurodevelopmental disability. It's characterized by
impairments in communication and social interactions, and the presence of
restricted and repetitive stereotype patterns of behaviors, interests, and activities.
The current estimated lifetime per capita cost of an individual with autism is $3.2 million.
The most recent prevalence numbers estimate that one in fifty individuals
has been diagnosed with an autism spectrum disorder.
So, given how prevalent this disorder is, and the fact that communication is a
defining feature of this disorder, it's really important for us to have a really good
understanding of language development in this population.
What makes this even more important is the fact that expressive language by the
age of five is one of the best predictors of outcomes for individuals with autism.
Given that, it really behooves us to figure out where they're looking the same as
typical language learners and then where we see them diverging in their language
development from the typical language learners.
Category development offers a really important place to look. The reason for this is
that category development is a cornerstone in cognitive development
and language development. So if we can have a better understanding
of what category development looks like in this population,
we can really get at something that's fundamental for language learning.
There has been some research that has examined
the categorization skills of individuals with autism.
What's really noticeable about this research is the fact that
there's a wide amount of variability in what they're finding.
Some studies report intact abilities, other studies report mixed findings within
the groups in that research study, and then other studies find that there are poor abilities.
What's really noticeable is not only do we see these mixed results across research
studies- part of this may be explained by the fact that we have different tasks that
are being used, different age groups, there's a variety of confounding factors as to
why we may be getting these different results- but it's also really noticeable,
or notable, that there are mixed findings even within studies.
Some work Letitia Naigles did in her lab looked at categorization skills during a
categorical induction task with adolescents with high functioning autism.
They compared them to a group that had been matched on the basis of language and IQ.
We would expect them to be performing the same if their category structures are
the same as their peers, because they've been matched.
What they did in the categorical induction task is they presented kids with a diverse
set of stimuli and a single set of stimuli or a diverse set and a homogeneous set.
So, the diverse set- and they did use snakes, I'll tell you guys when I was getting
these pictures I got shivers because I hate snakes- with the diverse set, what you can
see is they're really variable, right? The snakes all look somewhat variable.
In the more homogeneous set there's a unifying factor across this group of snakes.
So, for one set they would introduce the diverse set and say, "this group of snakes
has blue eyes and this group of snakes has grey eyes." They would then introduce
a novel snake that the individuals hadn't seen and then say,
"what color eyes does that snake have?" And we would expect that
kids would use that diverse cue because it extends to more exemplars.
What they found was that both the school age adolescents with high functioning
autism and the typical language learners, both extended the diverse properties,
just like we would expect, at a level above chance. But, when they really looked into
the data, yeah, the kids with autism were extending at levels above chance,
but they weren't really being consistent in this extension across the experiment.
Sometimes they would apply it, sometimes they wouldn't apply it
even though it was at a level above chance. Which is really surprising,
they seem to have figured out that this was a really relevant property,
but they're not being systematic in using this across a research study.
And so, when we think about this variable performance, we're really seeing it really
within this population in general. So, the question becomes not only how can we
explain why these kids look different than our typical language learners, but also
why do we see this incredible, variable performance within this population?
Dynamic Systems offers a really nice theoretical framework for thinking about this.
Within dynamic systems, you've got different pieces of information that are coming
together and being softly assembled within a task, to yield a specific performance
within that task. This will make more sense when I go through some examples.
But what your bringing is: you have the knowledge that you're bringing to a task,
and then you also have the properties of the task itself.
One of the ways this has been illustrated- and I think it really gets at this idea of
variable performance and how these two pieces can come together- is when we look
at how little kids, toddlers, are extending labels for objects.
So, the knowledge that kids bring to the task, little toddlers, they know that if you're
presented with a rigid, solid object- so if you see something or are presented with
something that's ball shaped- they know to extend that label on the basis of shape.
This is a very consistent cue for label extension. However, if you present little kids
with something that's a non-solid object, like shaving cream, they know that they
should be extending it based on material. So, what we see is they've got two really
important pieces of knowledge that they're bringing to the task. This is something
we see kids doing consistently, and so it's something we've identified as a stable behavior.
Within dynamic systems, a stable behavior is something that you perform-
you get that same behavior for a task outcome, regardless of how you vary the task.
However, as you can imagine when you think about objects in general,
we have rigid, solid objects, we have non-solid objects, but this really falls on a continuum.
Not all objects are rigid, solid objects and not all objects are non-solid objects.
Researchers have called these objects that fall in between, they've labeled them as
deformable objects. With these deformable objects, there's not a consistent cue for
that class of objects- for what they should be using to generalize the label.
So if you think about paper, paper texture or material becomes a really relevant cue
for generalization. That's how we generalize that label. But if you think about socks,
which are also a deformable object, that material is no longer a relevant cue and,
in that case, shape becomes a really relevant cue. So, with these objects in between,
we see that there isn't a set of consistent cues that kids can use for determining how
they're going to extend that name of that object.
What some researchers have done is they've looked at whether or not you got
variable performance in children, based on the knowledge that they're bringing to
the task and then the property of the task itself. So, what Colunga and Smith found
was that when they presented toddlers with objects that really varied in the degree
of solidity, the more solid that object was, the more likely the child was to extend
that label on the basis of shape. The less solid the object was, the more likely they
were to use a material extension. We see this shift just based on degree of solidity
with how they're extending names.
Another research study that came out of Larissa Samuelson's lab at Iowa, they found
an even- a more robust interaction I want to say, to highlight the point I'm trying to make.
What they did was they presented two different tasks. They had solid objects,
deformable objects, and non-solid objects. They presented the task and some of the children,
when they were asked about the label extension, they were asked in
a forced choice question format and the other group of kids, they were asked in
a yes/no format. So, all that's changing is how you're asking the test question.
What they found is the kids consistently extended on the basis of shape for
the solid objects, they consistently extended on the basis of material for non-solid
objects, but then for the deformable objects, when the children were asked a forced
choice question, then they used shape-based extensions. But when they were asked
yes/no questions it was really variable what they were using.
So this is showing how what kids know, and they're bringing to the task, plus how
you actually set up the task, if there's not a consistent set of cues that children are
paying attention to, you can get really variable performance
in what you're seeing the children do.
>>AUDIENCE MEMBER: Is this with typical kids?
>>DR. ELLAWADI: Yeah, this is all typical kids, typical little toddlers.
So, why I think dynamic system can really help us think about what we're seeing
with autism is we see this incredibly varied performance in tasks where we see
stable performance in typical language learners. Why do we see this variability
across research studies, and then why also do we see this variability within what
kids are doing within a research study itself? I think that becomes a really important
thing to think about when we're thinking about language development in this population.
This is work that was done out of Holly Gastgeb's lab, and they were looking at
the influence of stimulus typicality on categorization abilities. Just like there was that
continuum of solid and non-solid objects, we also have a continuum in all of our
categories, right? In each category that we have, we have very typical or good
representations of a category, and then we have more atypical category members
that may not be such a good representation of that category. So, if you think of
the category of birds, ostriches look very different and are very atypical members,
whereas robins are very typical members. She wanted to see- we know that typical
language learners are more accurate when they see typical, rather than atypical
category members. So, to get out whether or not kids with autism are organizing
their categories the same way, she used the category verification task to investigate
the category skills of school-age, adolescents, and adults with high-functioning autism.
It was really nice because she matched them all to peers based on language.
We can then see, are there differences in how the categories are being structured
when we've equated the two groups for language.
In the task, what they did was they heard the name of a category such as 'dog',
and then they would see a picture on the computer screen. So if they hear 'dog' they
might see something like a couch, and then they press a button if it was correct-
So, 'this is a dog, yes or no?' What she did was she varied the categories by typicality.
Within each category there were atypical category members, somewhat typical
category members, and typical category members to determine how accuracy
and reaction times varied. What they found was, with the school age children both
groups were slower and less accurate on atypical than typical stimuli.
They look like the typical language learners. But, in general, the kids with autism were just slower
than their typically developing peers. And, in this study, the typicality of the stimuli
had a greater effect on the reaction times for the autism group, than the typically
developing group. So, if it was atypical they were slowed even more than typical
language learners, in addition to being slower overall.
When they looked at the results from the adolescents and adults, we see this same
pattern emerging. So, when we get up to school age kids with high functioning
autism, they look really similar to peers they've been matched with.
So, what we wanted to do in this study was to take this task and extend it down to
a younger group of children. For the purpose of this study we were investigating
the categorization abilities of preschoolers with autism. Within this we want to see,
do preschoolers with autism demonstrate the same stimulus typicality effects
as typically developing language learners? Do we see that same effect where they're
responding faster and more accurately to typical, than atypical stimuli?
And then we also wanted to see whether or not there were differences in performance stability
between the children with autism and the typical language learners. I will get more
into this, but this is really coming from that dynamic systems framework,
to not only get a 'how do they look overall?' but what's happening
in that real time performance for these children. These participants were part
of a larger longitudinal study that was examining language development
in children with autism. When they entered the study at visit one,
the children with autism were about three and a half years old, and then they
have been matched to this group of typical language learners on the basis of language.
You can see, a couple years later we still have that significant difference in age,
which we would expect because kids with autism have delayed language development.
But we also see that they've diverged in their development over
the course of time. So, the test of auditory comprehension of language was used
as a measure of language skills, and we see over time that the children who are typical
language learners have significantly better language scores than the children with autism.
We also see that when we look at their IQ scores, as measured by
the differential abilities scale, that once again over time the children who are typical
language learners are scoring significantly better than the children with autism.
Although these groups are different- they've accessed language
and acquired language differently- over the course of a couple years
we can still see whether or not they're patterns look the same.
And so, what we had done is we had used that same task that had been used in Holly
Gastgeb's lab, and this was a category verification task. The children saw four
different categories: cars, birds, chairs and cats. One of the notable things about
the categories that they used in this study is that all of them emerge really early
in typical language development. So, by the age of thirty months, ninety percent
of children have acquired these words and these categories. Now we've got children
who are about six years old, so we would expect even if they're slower language learners,
they've still had a lot of experience with these categories.
Within each category, children saw typical, somewhat typical,
and atypical category exemplars. There is a very rigorous process
that went into determining the typicality. I'm not going to get into that now,
but if you have questions about that I can refer you to Letitia Naigles.
The way the task was set up, it was programmed in E-Prime, and the children
watched the stimuli come up on the computer screen. They heard, "this is a _____",
a picture came up, and then they hit a red switch if it was incorrect and a green switch
if it was correct. I know one speech- two speech pathologists are in the crowd, these
were big mac buttons, they were really big buttons to hit and really nice switches to activate.
There were seventy-two trials in total. What this had done was they presented this
over the course of two sessions. And so, children saw, within each category,
twenty-four category members of varying typicality.
There were also forty-eight foils that were presented.
What I'm going to go through with the analysis is first we'll talk about the stimulus
typicality- with that we're looking at how their accuracy and reaction times differed
based on whether the stimuli was typical, somewhat typical or atypical- and then
to look at performance stability, what we did was we looked at the trial
that had come right before. And so, one of the things in dynamic systems is within a task,
you have the knowledge the child's bringing to the task,
and then you have what's happening in the task itself.
One of the things that can influence performance is what children are seeing that child right before.
What we wanted to see was whether if a child saw a somewhat typical stimuli-
you can think of that as falling in the middle of that continuum- if they see an atypical
exemplar right before it, do they respond differently, is their accuracy and reaction
time affected in contrast to when they see typical stimuli right before it.
So, we're looking at these somewhat typical trials and we're seeing whether or not
we see differences in performance, based on whether an atypical exemplar came
before or typical exemplar came before.
Our preliminary analysis indicated that our data was not normally distributed
and was pretty much violating everything, so all of the stats that I did are nonparametric stats.
What I want to do is talk about the patterns that cross each group
independently first for the stimulus typicality.
When we look at our typically developing group, you can see accuracy is on
the Y-axis and then the different stimuli's on the X-axis. For the typically developing
group, when we looked to see whether we could collapse across visit one and visit
two there's a significant difference in their accuracy for the atypical stimuli
between visit one and visit two, so we didn't collapse those.
Here we have atypical visit one, atypical visit two, somewhat typical, and typical.
What we see is that when they see typical stimuli, they're more accurate than
the somewhat typical stimuli, and they're also more accurate when they're presented
with typical stimuli than atypical stimuli visit one or visit two.
I just want to point out that these are standard error bars so they are a bit larger,
which is why they do overlap a little bit even though we did find these significant differences.
With regards to the somewhat typical stimuli, we see that, in addition, the typically
developing kids were more accurate when they were presented with a somewhat
typical stimuli than the atypical stimuli at visit one and the atypical stimuli at visit two.
What we're seeing is we're seeing this same pattern of results. They're more
accurate responding to typical than atypical stimuli and we see this across the continuum.
The kids with autism are here in the yellow, and you can see atypical has been
collapsed into one visit, because there was not a difference between the two.
When we looked at the results, we saw that for the typical
and somewhat typical stimuli there was not a significant difference in their accuracy.
However, the kids were much more accurate when they saw typical stimuli as compared to
atypical stimuli, and significantly more accurate when they were presented
with somewhat typical stimuli than atypical stimuli.
When we look at the reaction time, what we see is a very similar pattern.
Our typical language learners are faster when they're responding to typical, than somewhat
typical stimuli, when we compare typical and atypical stimuli, and then they're
faster in responding to somewhat typical, than atypical stimuli.
In terms of the latency patterns for the autism group, there was not a significant difference at all.
It didn't matter what type of stimuli they saw, we see that their reaction times did not differ.
We've now got this idea of what the pattern of performance looks like over
the course of the experiments with regard to the typicality. So one of the other things
that I'm really interested in is what does it look like in real time performance,
and are they being influenced by what's happening right before?
Once again, when we look at the typically developing group, we have the somewhat
typical stimuli that was preceded by a typical exemplar, and then when somewhat
stimuli was preceded by an atypical exemplar. You see performance looks
very comparable, there's not a significant difference between the two.
When we look at the kids with autism we see a really different picture.
When the kids with autism saw a somewhat typical stimuli preceded by a typical stimuli,
they're doing okay, but if there's an atypical exemplar that's preceding
the somewhat typical stimuli we see this big drop in their performance.
It's a statistically significant difference in their accuracy, and it's notable because
this somewhat typical stimuli, there's no difference there, the only
difference is what's happening right before they see that exemplar.
When we looked at latency, there's no difference in either group based on what they saw right before.
Just to summarize these results, what we see for our typically developing group is that,
in terms of overall task performance, they're responding much more accurately
and faster to typical than atypical stimuli. We see a pretty similar pattern of results
in terms of accuracy where they're more accurate when they see typical
than atypical stimuli, but we don't see any difference in reaction times here.
In terms of performance stability, there wasn't a change at all in our typically
developing group. It didn't matter what was coming right before,
in terms of accuracy or in terms of latency.
This is different than what we're seeing in the autism group. In the autism group,
what we're seeing is when there's an atypical exemplar that's coming before
a somewhat typical exemplar, we see this drop in performance,
although we don't see a change in their reaction time.
To kind of put this back into the framework of what we had talked about before,
with regards to the previous research using this task, what they had found
was that children were more accurate when presented with typical, than atypical stimuli,
and we can see with our little kits here that they pretty much follow that same pattern
in terms of accuracy. Where we're differing from what they had found with these older
kids, is the younger kids with autism, latency doesn't seem to be effected
by typicality. So the question then becomes, is this something that we see developing
over time, where kids don't yet have those robust connections where we're seeing
that faster response? This is something I think, bears more exploration of why do we
see this difference in latency over the course of development.
With regards to the performance stability, these results were really interesting
I think because they highlight- yeah, we see this same pattern,
but when you really kind of dig in and look at what's happening
over real time, the kids look really different. We see this variability
in the children with autism that we're not seeing with our typical language learners.
If you think about kids with autism, this lack of generalization, this difficulty,
is something that's pretty consistent with the disorder in general.
With regards to typical language learners, it's been hypothesized that children learn
language by tracking the statistical regularities that are present in their environment.
So, this idea of children, or toddlers, in that task I talked about before,
they come in knowing that if I am presented with a solid rigid object I should
pay attention to shape and extend on the basis of shape- work by Larissa Samuelson
and Linda Smith has shown that this really seems to be something that's developed out
of children's statistics. And so, if you look at children's early vocabularies, what we
see is that the majority of their early words are solid count nouns. When they hit
fifty words in their expressive vocabulary, we see the emergence of this shape bias.
We see that the statistics in their early vocabularies support attention to shape,
and then there seems to be some sort of threshold, in terms of when they've accrued
enough information to know, "Yeah, I should be paying attention to this."
This is important because we know that this ability to extend on the basis of shape
really facilitates language development and we see this nice, robust gain in language
development when children begin to demonstrate a shape bias.
When we think about children with autism, the regularities in their long-term experiences
are definitely different than the children with typical development. It may also be
that the regularities that they are extracting are different than the typical language learners.
Some research found that children with autism fail to develop a shape bias
even after they've acquired fifty words in their vocabulary.
So, if the shape bias is set on the statistics that the kids are getting- once again,
they've got the same amount of words in their vocabulary- we would expect them
to learn that shape is a relevant cue, but there's some sort of disassociation
between the words that they're getting and the regularities they're extracting.
We also see that children with autism, in another research study, found that
the children with autism were really variable in their ability to extract the regular
relationships they were presented with. So while typical language learners-
there were a variety of things and cues they could attend to- the typical language learners
were able to track all those regularities, the children with autism in that study weren't.
So it seems like there is something inherently different about how the kids
with autism are learning language. Even when they do have a vocabulary that's
the same size, there's something very different in terms of that vocabulary.
If you think about the fact that they got pushed around, when they saw that atypical stimuli right
before a somewhat typical stimuli, this would suggest that they don't have a firm idea of what
cues they are supposed to be attending to. If you see something atypical right before,
this might shift your attention, leading to poor performance.
What I think is exciting about this is that these findings support other research that
suggests that kids with autism are developing their lexicons differently than typical
language learners. It also suggests that dynamic systems is going to be a really
useful, theoretical framework for thinking about learning in these kids.
Not only do we see these differences in how they're organizing their lexicons,
but how do these differences influence real time performance
in really subtle ways that can actually have very real repercussions.
So, I apologize, I do not have a citation slide up here and I'm happy to email my
citations to anybody who would like them. But that's it.
[audience applause]
>>AUDIENCE MEMBER 1: Allison, did you also look at what defines probability
between other combinations? So, like if they saw atypical and then saw typical?
>>DR. ELLAWADI: You know, we stuck with the somewhat typical because we would expect
that's where you would see the variable performance just because- with the typical
we would expect that to be really entrenched, right? So this is what those category
members should look like, and with the atypical they may have learned that a little
bit more because it's novel. That somewhat typical is where we hypothesize
we would see that shifting. So, we haven't looked at those other areas and that could
be interesting too. These results are interesting but the experiment wasn't actually set
up to look at that. So, it would be interesting to do something where we are set up to
see how performance shifts, where we're really concentrating on that.
>>AUDIENCE MEMBER 1: What I'm wondering is- you talked about shape bias
and not paying attention to the right cues, in that instance from a theoretical standpoint.
But I also wonder how much of it is almost like a response latency? You know,
are they just a little bit slower in processing what's going on? And if that's happening,
I would expect to see a similar influence in regards to the combinations.
>>DR. ELLAWADI: Yeah, that's definitely something that could be a plausible
explanation and something worth looking into.
>>AUDIENCE MEMBER 2: This is actually building on that idea. Did you look at
how performance changed across time, like for the duration of the study?
>>DR. ELLAWADI: No, and that's something we absolutely should do because that
could be really interesting too. I know some study with typical language learners,
typical infants, has found that- they were looking at looking patterns,
and the looking patterns of the strong and the weak learners were the same at the
beginning, and then where you saw them really diverging in the middle.
So that could be something that's happening too, where they look more similar in the
beginning and then over time we see this divergence.
>>AUDIENCE MEMBER 2: Well, sometimes you get practicing effects where
everybody just gets better. Sometimes they get tired and the results crash.
So it would be interesting to see where this falls.
>>AUDIENCE MEMBER 3: Yeah, especially since you're looking at timed response rate.
I wonder if that would really be impacted by that practice effect or just by being tired.
>>DR. ELLAWADI: Yeah, yeah. My guess is that, for those kids with autism, that
latter half of the experiment could prove to be incredibly problematic for them,
just because their retention is not as good as our typical language learners.
So, that would be really interesting too.
>>AUDIENCE MEMBER 4: I have a question.
>>DR. ELLAWADI: Yeah.
>>AUDIENCE MEMBER 4: And this might be your opinion, but talking about
this category differentiation, what do you think that does then for these children's vocabularies?
Because obviously they cant- if they categorize new information
or new things that they're learning incorrectly,
then that's going to do something to their vocabulary.
>>DR. ELLAWADI: I will get on my soapbox a little bit, so I'll apologize. All of the kids
in this study are getting ABA. Are people here familiar with ABA? If you think
about what's happening in ABA, that's a very different language learning experience
than what's happening in typical language development. If we think about
what's happening in terms of statistics, I'm not sure if we're doing this rote learning what
we're actually getting the kids to pay attention to. And I think that might be part of the problem.
I think what's happening is that you're learning these individual pieces
and you're not really pulling out what's relevant.
You have to continue learning pieces and you're not making any of these broader connections that
really- once we see these broader connections, we see this incredibly rapid growth in what kids
are doing. I'm a big fan of the shape bias, so I'll do another example from there.
In Larissa Samuelson's lab they had taught a group of kids who did and did not have
the shape bias. The kids were just coming into the lab and they were just having
some experiences with solid objects. You know, they weren't- it was through play,
they were really having the kids pay attention to shape. Over the course of
the training there was a control group that just came into the lab, and then this group
that was playing with solid objects, and they were saying, "This is a dax, this is a dax,
this isn't a dax", to draw attention to the shape. And so, we see a growth in their
language over time. The group of kids that actually got that solid object experience,
they developed a shape bias, and then not only that but the control groups
vocabulary grew by, I think, fifty or sixty percent. There was a 125% growth in
the group that actually got the shape bias. So, if we're just teaching these, you know,
"This is a car, this is a car, this is a car", and we're not saying, "These are the relevant
pieces of information", then in terms of learning you're not going to be able
to become an independent learner in some ways where you can look at that and say,
"I know how to extend this, I'm pulling out those regularities that matter for learning."
I think it becomes a really big issue and it just doesn't enable them to be learning
independently in a way that we would want them to. They're not learning the things
that will facilitate more learning. That's not to knock ABA in any way, shape,
or form, but I think we just haven't thought about what are we teaching these kids
to pay attention to and what are we teaching them that's relevant?
I kind of went all over the place, did I answer your question?
>>AUDIENCE MEMBER 4: No, that's good. It's funny because I've done a lot of ABA
throughout college and getting my masters and I can't tell you how may times
I've handed a child stack of cards and said, "Put these in the right categories", and they're
just supposed to put them in the right categories but we never really discussed why,
or what shapes they were, or what made that a member of that category or not.
>>DR. ELLAWADI: Yeah, it's really interesting to think about what are we really
teaching them. And I think we also don't do this in speech therapy, right? This is just
another way of thinking about language learning in general, and I think it can yield-
and it's great because then the kids have those categories, but how stable are those categories?
Are we going to see that generalization that we want to see? Here, if you
are able to be shifted around that much, your performance
is going to be really variable over time, which can be problematic.
>>AUDIENCE MEMBER 5: Um, how much time was there between visit one and visit two?
>>DR. ELLAWADI: I can double check with this, I think just a couple of days.
>>AUDIENCE MEMBER 5: Why do you think you saw a difference?
>>DR. ELLAWADI: With the typicals? I do not know. I was looking a the data
and I was like, "What is going on?" I'm not sure, I'm really not sure.
>>AUDIENCE MEMBER 5: There was obviously different stimuli.
>>DR. ELLAWADI: It was different stimuli. They only saw half of the stimuli at visit
one and half at visit two, so I'm not sure what engendered that difference that we saw.
I don't know. It's interesting and I have no explanation for it, at all.
They weren't getting reinforcement at all during the task,
so I'm not sure why their performance changed.
>>AUDIENCE MEMBER 6: How much of the differences that you see do you think
are naturally occurring and how much is a result ABA?
>>DR. ELLAWADI: I don't know. Would these kids have any language if they
weren't getting ABA? I don't know. Right?
So - I think one of the things that can be really interesting is,
do we see differences, you know,in this real-time performance?
and then how kids are developing their lexicon based on the type of intervention
that they're getting. Because we know other types of intervention are effective.
Most kids are getting ABA right now because that has the biggest evidence base
behind it, and I know they were trying to recruit to do 'Hannon More Than
Words' which is more of a naturalistic teaching approach out in Boston.
Parents didn't want to participate because they wanted ABA. So it's really, it's interesting.
I don't know if we would see the same thing happening if they
were getting a different type of intervention or not, but I think that's
something that definitely warrants further exploration, especially since we're trying
to figure out,"what can we do to get the best" you know- "how can we help these
kids the most" and it may be that slower acquisition but more robust acquisition
may be better over the long time - you know the longer period then just kind
of getting these words in - and I'm not sure.
Yep?
>>AUDIENCE MEMBER 7: How much do you think that the limited interests core -
like that aspect of the disability - plays into how they group words or images?
>>DR. ELLAWADI: For this task or in general?
>>AUDIENCE MEMBER 7: It can be more in general.
>>DR. ELLAWADI: Um. That's a really hard question because they definitely have
limited interests in some areas but then we also see them be - some kids are real
'experts' in other areas. There are kids that become experts
on dinosaurs or experts on trains, and so they kind of have this incredible depth
of knowledge in this one particular area.
I would - my guess is . . .just like any of us -
if it's not something that we're that interested in, we won't learn as much about it.
And I think that could be part of it too. That's why I liked the categories they
choose for this task - is they're really early-developing categories and they're
something kids should have had experience with. Right? So they see chairs everyday,
so even if they might not be that interested in chairs this is that ongoing thing that they're seeing.
The same with cars. We know that's something that can be
of higher interest to kids with autism. And then the two birds or something that you've
probably seen on a very regular basis. Cats we may see a difference there.
I'm not really sure? Because I think it really depends on the kid.
>>AUDIENCE MEMBER 7: I was trying to think of some of my kids I was studying
over an increased school age but - there were some computer programs we used
in the classroom and some images they would self-stimph and I'm thinking -
I know you didn't have any latency . . . affects - but I'm thinking, "Well maybe if they
were things that were more interesting" - like you said, the cars,
they might spend more time on those than birds and than
not have anything to do with typicality or atypicality.
>>DR. ELLAWADI: Mmm-hmm. So another way we could pull this out too is the animate versus
the inanimate to see if there's a difference in how the kids are responding to those.
With this sample of kids, we're currently coding some other data and I can tell you-
they're doing a 'looking paradigm' and the kids with autism just do
very different looking than the typical learners - a lot more 'out of the corner of the eye' -
it's just very different qualitatively which is interesting.
>>FACILLITATOR: Alright well if there are no more questions we do have the salad
and pizza and cupcakes over here. So, if everyone would like to spend some time,
maybe mingle a little bit. If you have any questions
you can ask Allison personally and. . . eat!