Autism should have disappeared. If we accept Charles Darwin’s reasoning, as most biologists do, the incidence of autism in the human population should keep dropping. Traits that promote genetic survival thrive, whereas those that hinder reproduction disappear. Many autistic people do not bear children, so autistic gene variants should be selected out of the gene pool. Instead, autism has either persisted or increased (increase is a matter of debate) to a rate of one out of every sixty-eight births. What’s up?
One explanation might be that autism results from a de novo mutation in a few genes. The mutation is new to this generation and will disappear after this generation. After all, every person has thousands of “mistakes” in their genes, but most make no difference in the long run. Periodically, however, one-time mutations do make a difference, as the extra chromosome responsible for Down syndrome demonstrates.
But autism does not work this way. We know that many of the gene variants contributing to autism are not rare, one-time mutations. Autism results from a combination of genes, hundreds or possibly thousands, and many of them are inherited. These gene variants’ persistence in the human genome implies one crucial fact: they must serve some useful purpose, some purpose that enhances reproductive success.
Two professors at Yale University believe they have identified the purpose. R. Polimanti and J. Gelernter of the departments of psychiatry and genetics looked at genetic data on five psychiatric conditions: autism (ASD), attention deficit hyperactivity disorder, major depressive disorder, schizophrenia, and bipolar disorder. Other researchers had already identified gene variants that seem to connect with each of these conditions, and the Yale scientists analyzed the data to determine whether the relevant genes were dying out or becoming more stable, or "fixed." Only autism and schizophrenia showed a high likelihood of being associated with advantageous (in an evolutionary sense) genes, and only autism “survived” all of the statistical tests they ran. Polimanti and Gelernter hypothesize that “certain ASD risk alleles [gene variants] were under positive selection during human evolution due to their involvement in neurogenesis and cognitive ability.” In other words, some of the gene variants associated with autism are also significantly associated with high intelligence. “Smart” genes are advantageous from an evolutionary standpoint, so they persist.
Hooray! Their conclusion validates my belief: research intended to eradicate autism is misguided. I've argued repeatedly that humans do not know enough to understand what we destroy along with the “unattractive” traits we seek to eliminate. And now here come these two scientists saying that many of the variants for autism also contribute to high cognitive capacity. Nobody knows how these genes interact to cause autism or intelligence. Nobody understands what other genes are relevant in causing a person to present with autism, high intelligence, or both. We do not know many things.
We can start, though, by acknowledging two well-accepted ideas. The first, which I presented already, is that gene variants that hang on through many generations hang on for a reason. The second is that the Law of Unintended Consequences holds for biology as much as it does for geopolitics (Recall Donald Rumsfeld’s famous reference to “unknown unknowns.”). Actions have results that are not part of the actor’s purpose. We are now developing the capacity to cut out strings of DNA that we do not like using CRISPR-9 technology. Such targeted cutting may be a good idea for cancer, assuming the relevant gene has been correctly identified, but neurological characteristics are polygenic, meaning they involve the interaction of many genes. And many of those genes seem to play a role in more than one character trait. Message to scientists: watch out what you wish for, because you will certainly find yourself in a world of unknown unknowns—situations you never considered—when you alter any environment.
Looking beyond the paper’s thesis, a few questions emerge. First of all, autism is often associated with low cognitive ability. (The Center for Disease Control’s most recent estimate suggests that about sixty percent of autistic individuals have some cognitive impairment, although scoring for the autistic population is notoriously unreliable.) How do genes that code for high intelligence also code for low intelligence? The short answer is that no one knows. A slightly longer answer is that other genes, besides those identified here, must be implicated in cognitive capacity. A longer answer, one hypothesized by the biologist B. Crespi, is that autism reflects an imbalance in the development of components of intelligence. Large brain size, cortical thickness and many other traits are associated with both high intelligence and autism, but high cognitive capacity seems to also require neural connections that stretch across long distances within the brain. Often autistic people have higher than average neural activity, but it is relatively isolated within sections of the brain. “Some or most components of intelligence are increased, but in such a way that overall performance is often reduced.” Perhaps this line of reasoning will be validated by further research.
Another question the article elicits is: what do we mean by intelligence? As I’ve noted before, Howard Gardner changed our understanding of intelligence when he proposed the concept of multiple intelligences. The authors of the study use expressions of cognitive capacity that include years of schooling, college completion, childhood IQ, and openness to experience. Crespi considers these characteristics as well, though he adds several others. He then uses them further to hypothesize “that autism represents most broadly a disorder of high intelligence (and low imagination), and schizophrenia a disorder of high imagination (and low intelligence).” He goes on to suggest that more males are autistic because men are better at systematization, the key skill in these measures of intelligence. Empathy and imagination do not count as useful measures of intelligence. From my perspective, the measures seem to reflect a biased and somewhat limited understanding of intelligence. However, I readily grant that my own understanding of the intelligence measures being used is somewhat limited, so I leave open the distinct possibility that I have misinterpreted the research.
Returning to the study’s thesis though, I strongly endorse the line of research. Examining the evolutionary rationale for autism (and for schizophrenia, which has been shown in other research to have strong evolutionary “roots”) prompts us to ask ethical questions and policy questions from a vantage point that honors autism. It’s time for our thinking to evolve.
*Special thanks to Dr. Gopal Thinakaran, neurobiologist (and friend), for walking me through GWAS interpretation and positive selection. Special thanks to Dr. Jeff Campbell, economist (and husband), for walking me through the statistical significance scores. All mistakes in interpretation are entirely my own.