Autism, MR, and Intelligence
August 21, 2007 by Kristina Chew, PhD
Filed under Health
A few days ago I wrote about how, when a different test for intelligence (Raven’s Progressive Matrices) was used to evaluate autistic children, they scored significantly higher, than when a more “traditional” test was used (the Wechsler). An interesting observation about intelligence and autism is made by psychiatrist Glen Elliot in an August 19th article in the San Francisco Chronicle about Bryna Siegel, director of UCSF’s autism clinic at the Langley Porter Psychiatric Institute:
…..mentally retarded people usually achieve superior self-sufficiency than autistic people, even if their IQ is lower. A retarded person with an IQ of 65 can learn to ride a bus, but an autistic person with an IQ of 80 might struggle to do this, Elliott says. If the bus arrives a few minutes late or if it pulls up a few feet away from the curb, he might not get on the bus. That’s because autistic people frequently fail to see what Elliott calls “the essential features of what is required.” It is a trait that was evocatively depicted in “Rain Man,” when Dustin Hoffman’s character suddenly freezes midcrosswalk because the “don’t walk” light had started to flash.
Being “smart” doesn’t always correlate with being able to manage the basics of day to day to life. Measuring intelligence is not a simple business: I can know my son is smart, and that teaching him to take the bus on his own is not going to be easy (but when it comes to swimming in the ocean, even on a rainy day, there’s no one savvier in the waves than Charlie).
“A retarded person with an IQ of 65 can learn to ride a bus, but an autistic person with an IQ of 80 might struggle to do this, Elliott says.”
An autistic person with an IQ of 140 might struggle to do that. I don’t think I’m exagerating too much.
IQ is really a construct that started with Binet’s invention of it. It’s not meant to measure life skills, but educational needs. Even if there’s some sort of correlation between IQ and some skills, it’s not surprising that it fails to predict a subset of people’s ability to navigate the (NT) world.
There are tests to measure self-adaptive skills, such as the Vineland Adaptive Behavior Scales.
I’ve been riding the bus most of my life and I still have troubles with it. In addition to the issue with being on time or not and getting close enough to the curb, there’s also the following.
Buses are noisy, sometimes crowded, unpredictable sources of sensory overload. I used to own a walkman when I was a teen, just so I could stand to be on a bus. I’m thankful that I managed to master driving (that was a challenge I was 24 when I finally managed to pass the test) and as such never needed to take the bus again.
“A few days ago I wrote about how, when a different test for intelligence (Raven’s Progressive Matrices) was used to evaluate autistic children, they scored significantly higher, than when a more “traditional” test was used (the Wechsler).”
I’ve written about this elsewhere (Mike’s blog), but I don’t know how to link the comment.
What I will say here is that it is not wise to compare at test that is highly saturated in ‘g’ (such as the Raven Progressive Matrices, or the Test of Non-verbal Intelligence, or even the Universial Non-verbal Intelligence Test) with a test that is in fact a battery of subtests which each tap into different aspects of intellectual ability, such as the Wechsler Scales (all of them: WPPSI, WISC, WAIS &WASI) or the Kaufman suite of tests (K-ABC & K-AIT), or others with more than two subtests on which test scores are elicited. This is because many of the subtests on these ‘multi-dimensional’ assessment tools tap into aspects of ability that are not so highly saturated in ‘g’ (this being the signifier for Spearman’s ‘general intelligence factor’). In a sense, it is no better than Lenny Schafer comparing ‘autism oranges and autism apples’ (as he notoriously did some years ago). A task that is highly saturated in ‘g’ will – almost by necessity – allow a higher score, since it does not tap into a narrow aspect of ability: rather, it allows for a wider contribution from ‘g’ that enables the testee to complete a task involving many aspects of ability.
So, in the case of autistics, it is likely that we would score highly on any ‘g’-saturated test of general ability (such as the RPM, or the UNIT or the TONI-3, or just the Matrix Reasoning subtest of the Wechsler suite).
FWIW, though, my general impression is that Bryna Siegel tends to be extremely negative re autistic people’s abilities. I read her book eight or nine years ago and it terrified me about her child.
My own feeling is that higher intelligence does allow for more learning, so in general it’s more adaptive than lower intelligence (which goes along with my other basic belief re my kids: they have to be taught many things you don’t have to teach other kids, but they’re also well capable of learning). And I also think that these tests are unreliable, tap into factors besides intelligence, and culturally biased.
Also? Not every autistic person does well w/things like Raven’s Matrices. A subset of peopel w/AS has very high verbal abilities, not so much with the nonverbal reasoning.
Um, that would be “terrified me about MY child.” I’m a nice person, but not THAT sensitive to other people’s lives.
“My own feeling is that higher intelligence does allow for more learning, so in general it’s more adaptive than lower intelligence”
I don’t think there’s evidence it’s more adaptive. It’s possible that an IQ of 120 is about as adaptive as an IQ of 80. (By adaptive I mean the potential to produce offspring many generations down the line). In support of this, see:
http://www.gnxp.com/blog/2007/04/intercourse-and-intelligence.php
Siegel emphasizes that “acceptance” is the main quality parents need—the reporter expecting her to talk about resilience, persistence, etc.. And I think there is something to it though “acceptance” too often seems to passive—-and it really is not, at all.
Both RPM and Wechsler scales are tests of general intelligence which have been routinely compared in both typical and atypical non-autistic populations.
For Emily, our paper did not include AS people.
Klin et al. (2003) noted that autistic adults with normal measured intelligence had the measured adaptive abilities (via the VABS) of typical 4-year-olds, but some were remarkably self-sufficient. Possibly, behaviour which is adaptive for non-autistics (the only kind measured by the VABS) is not necessarily adaptive (in the sense of leading to self-sufficiency) for autistics.
“Both RPM and Wechsler scales are tests of general intelligence which have been routinely compared in both typical and atypical non-autistic populations.”
There may be comparisons to check for reliability and validity issues on overall general intellectual ability scores. But it doesn’t make sense to say that one is better for assessing a particular group of people: they are designed for different assessment needs. RPM is good for any group of people when a quick assessment of general intellectual ability is needed (such as in occupational psychology, for example, when one is assessing candidates for particular jobs and a full WAIS administration isn’t necessary). But the Wechsler Scales have their own utility when it comes to assessing some of the more specific aspects of ability that are not covered in a more simply constructed test such as the RPM. Such test require training in their administration and in the interpretation of their results (and understanding their construction), whereas the RPM does not require as much training.
Wechsler and RPM scores have been compared in various typical and atypical poplulations for many purposes, and there are multiple other considerations, both in typical and atypical populations (e.g., deaf children). As I’ve written elsewhere, in our paper (which is about intelligence in autism), on RPM
“… we report a very high individual item difficulty correlation (near unity) between autistics and non-autistics. As we point out, this makes it likely that RPM is measuring in autistics what it measures in non-autistics. This contrasts with the uneven autistic subtest profile in Wechsler scales, and factor analytic attempts to compare autistic and non-autistic performance on Wechsler scales (e.g., Goldstein et al., 2006).”
And far from being “simple”, RPM is recognized as the most complex and general single test of intelligence (Snow et al., 1984).
“And far from being “simple”, RPM is recognized as the most complex and general single test of intelligence”
It is simple in that it has one ’sub’-test. That makes it simple, as opposed to the more complicated ones (such as those that map on a number of sub-tests). It is also ’simple’ in that it does not assess the individual facets of intellectual ability that are used in completing the tasks.
RPM is not a subtest and, as a test of intelligence, is complex and general rather than simple and specific.
I KNOW what RPM is and the issue of subtests was to compare the nature of it and the Wechsler Scales in terms of what they do.
In actual fact, the PRM IS a subtest these days. The Psychological Corporation publishes the two together as the Raven Progressive Matices and Vocabulary Scales (or at least they will be doing from 2008 onwards).
http://www.harcourt-uk.com/product.aspx?n=1343&s=1492&cat=1356&skey=3694
In such a package, the RPM is indeed a subtest within a two-subtest battery. It isn’t complex. It is a simple test (for the reasons I have stated). Certainly, it is general – which I have already said myself – and does not separate out any of the individual skills required in order to complete the tasks it sets. So it cannot tell the assessor what is happening at the cognitive level. It reveals a single score, not a whole range of scores that can tell the assessor a lot more about what is happening regarding perception, general learning and comprehension, attentional ability and other individual aspects of intellectual functioning.
You’re only arguing to disagree with anything and everything I say. That’s your choice. But I’m the one who studied educational psychology (and who studied test construction as part of it!).
I’m satisfied with my understanding of what these tests are about. I’m not convinced of yours yet.
What RPM (the most complex and general single test of intelligence) measures has been extensively studied “at the cognitive level,” as have the cognitive and neural mechanisms involved in general and fluid intelligence.
During the holiday I have had a chance to catch up on some reading which seemed to touch on the topics of mental retardation, the labels attached to people, autism, assessment and how even the most well-designed instruments, because of their nature of “objective” measurement of ability/deficits may miss the “whole” of a person and that which gives life meaning from his/her point of view.
Maybe it’s an opinion of one, but I would recommend reading, “The World of the Simple”, pp.171-233, in “The Man Who Mistook His Wife for a Hat and Other Clinical Tales”, by Oliver Sacks. I think that there is much to reflect on.
I read that a long time ago—-good to return to. Any other suggestions?
Learning To Listen: Positive Approaches and People With Difficult Behavior. Herbert Lovett (1996). Baltimore: Paul H. Brookes Publishing Co. 269 pp.
Finding A Way Toward. Everyday Lives. The Contribution of Person Centered Planning. John O’Brien & Herbert Lovett.
http://thechp.syr.edu/everyday.pdf