A discussion on cognitive tests in high-functioning autism


For all of us, our behavior is what is visible; it is what defines us in the eyes of others. Yet, behavior is only the end result of multiple internal processes that – in our own eyes – define our identity. As a high-functioning autistic individual and neuroscientist, I experience two distinct discrepancies in my interactions with others. First, my rich internal world cannot be adequately expressed by my behavioral repertoire. Second, the social response of neurotypical people often mismatches my intended behavioral output. My curiosity becomes aggression. My concern becomes skepticism. It is disorienting and painful. A very common experience among autistic people is that of a lifetime of repeated misunderstandings, and a lifetime of being severely underestimated.

One setting where it is particularly important that neurotypical and autistic people understand each other is during a neuropsychological evaluation. Cognitive tests are designed to tap into specific functions of the brain, such as the ability to organize or remember information. To work, these tests generally require intact sensory and motor systems, and they need to be given in a setting where the subject is minimally affected by stress, noise or performance anxiety. There are many potential confounds in testing autistic people, such as sensory sensitivities, coordination deficits and problems with novelty. 

Here, I will focus on a less recognized challenge: the difference in thinking style between autistic and neurotypical people.

Highly logical, detail-oriented information processing can provide an enormous advantage academically or professionally, but in other venues it can be a barrier. Few people are unaffected by the highly artificial social setting of a cognitive evaluation, but for an autistic person the interactions can come across as outright bizarre. Kate, who recently wrote about her intensely negative experience of a research study, described feeling humiliated by the scripted questions and “silly” test materials. These are her words regarding a test involving judgments about the “trustworthiness” of a series of faces:

"I told the tester that I did not know any of these men so how could I possibly know if they were trustworthy — and anyway, of all racial and ethnic groups, white men have proven to me to be the very least trustworthy, so I didn’t trust any of them. But I still was asked to flip through every one of the fifty pictures, and say, ‘no, not trustworthy,’ because I didn’t know them. This was absurd."

It is not that difficult to understand both sides. Tests are standardized; there is little flexibility. The scripting is the same for everyone, autistic or not – that is part of what makes them valid. But there are aspects of the autistic mind that are poorly compatible with the way cognitive tests are designed. 

Or, rather: 

There are aspects of cognitive test design that may prevent them from accessing the autistic mind. 

Kate was not trying to be nitpicky; she was expressing her honest confusion. And note that, in this case, on this test, there was no evidence of a purely social deficit, even though her test scores must have been abysmal.

I believe there is a very autism-specific confound in many neuropsychological tests: the performance of an autistic person may reflect a hyper-logical thinking pattern rather than an inability to read faces, empathize, or whatever the test is measuring. This is a true barrier to understanding the autistic mind and a potential confound in research studies attempting to assess cognitive function in autistic people.

When I went through neuropsychological testing in the mid-nineties I felt the same type of confusion, although I was unable to articulate it at the time. I was presented with cards with drawings of people. They were not real people. They did not feel anything. The stories I could potentially make up about the drawings were literally unlimited. I could pick any scenario, ranging from a shallow interpretation of the immediately obvious visual information, to a dark twisted Kafkaesque fantasy that would surely strand me with some ominous psychiatric diagnosis. I had a literally unlimited repertoire of interpretations that could all be considered realistic, given that nothing about this situation was realistic. 

Performing the test required significant mental acrobatics to streamline my mind into some acceptable output. Again, the interpretation of my behavioral output did not match my internal reality.

The consequences of hyper-logical thinking are not limited to tests of imagination or social function. Give me instructions for any test and I will probably see ambiguities that are invisible to the neurotypical individual. If you tell me to name “each of the objects”, and then point at only one, my brain struggles to switch from “Each object” to “Whatever object you point at”. If you want me to pair a cartoon dog with one out of four cartoon objects, my brain will immediately form associations for all the objects. 

I have to model the neurotypical mind, and realize that you want me to simplify this exercise to a stereotypical association between a dog and its bone. 

To pass the test, I must wrestle my complex mind into petty generalizations. That is not something I am good at, and it sure can be disabling, but it does not necessarily reflect a deficit in the function that the test aims to isolate. And to characterize this possible confound, it will be necessary to start a dialogue between scientists and people with autism.