Does Temple Grandin’s brain work like a computer or a human?

Yesterday (Oct 23, 2022) I had the pleasure to participate in a panel discussion with Prof. Brenda Smith Myles, Debra Moore and Temple Grandin (at the Autism Cork 2022 online conference). Temple emphasized the importance of ‘exposure‘: introduce children with autism from an early age to as many things as possible that expand their knowledge of the world: “you have to fill their database“. She illustrated this with numerous personal experiences and compared her own learning processes with that of artificial intelligence: “My brain works like an AI program of machine learning“. The example she gave to illustrate this is a well-known example that she has cited for over 30 years, especially how she learned to distinguish dogs from cats. Even then she compared her mind to a computer, saying that her mind is like a web browser: “I don’t have a concept of dog, but I have pictures in my mind of all the dogs I’ve ever met”. In the beginning, Temple distinguished both animal species based on only one criterion: size. Dogs were big, cats were small. Until she came across a small dog. Then her brain had to update the dog model and add new criteria to it. (I described this process earlier chapter on concept development in autism in my book “Autism as Context Blindness”.

What Temple says, is remarkable. AI learning or machine learning tries to imitate human intelligence in learning processes… So, what Temple is really saying is that her brain learns like a computer that imitates how human brains work. That would mean her brain doesn’t work any differently than other people’s. Where is the autism then?

The difference can be found in the difference between conscious and unconscious learning. Indeed, most human learning occurs through the systematic updating of models (or concepts). But those learning processes are largely unconscious. Within the theory of predictive coding we know this as ‘perceptual learning‘. In addition, there is also ‘active learning’, in which people consciously update their model or concept. And therein lies the difference between autism and no autism: Temple’s brain has to update models through conscious ‘active learning’. As she herself said years ago, “I use my intelligence to understand what other people intuitively understand, bringing immense intellectual effort and computational power to bear on matters that others understand with unthinking ease“. (This quote is from a wonderful article Oliver Sacks wrote for the New Yorker in 1993.

Last night Temple emphasized this again with the statement: “I have no working memory“. Of course she has, but what she refers to is the conscious working memory, which is indeed very limited (the formula that all psychologists learn is: 7 minus plus 2). The unconscious brain can process countless more pieces of information.

It is precisely because people with autism (have to) learn more consciously that we have to slow down in the learning processes we devise for them and offer them more learning experiences than we do for people without autism. Machine learning is about feeding the program with a lot of data, so that the computer can generate models for prediction without explicit instructions.

I describe these processes of how the brain builds models or concepts in detail in my newest book “Autism and the predictive brain”.

What it really comes down to, is this: we should focus much more on bottom-up learning than the classic top-down learning that typifies education. In practice, this indeed comes down to what Temple promotes so hard these days: ‘exposure‘. Expose children to many and above all different experiences. During the panel discussion, we (the panel members) applied this to preparing children with autism for hospitalization. We need to offer predictability by giving the child a flexible model (concept) of hospital. And if they have been in hospital before, we should update their model for a next visit because autistic models tend to be absolute rather than relative (the subtitle of my new book…). As in Temple’s brain: dogs were big, cats were small…

I’m delighted to have had exposure time yesterday with some brilliant minds to discuss the autistic brain and how it learns…

One last remark based on the excellent feedback received from Els Van Beneden: this exposure and bottom up learning should happen in safe conditions. We want autistic people to benefit from this and not be stressed or damaged. This has always been obvious to me, but seemingly not to everyone.

One thought on “Does Temple Grandin’s brain work like a computer or a human?

Leave a Reply