A general law of evolution
For decades, biologists have known a simple rule: the more abundant a protein, the more slowly it evolves. The logic is straightforward. If a widely used protein changes, the consequences ripple across countless cellular processes, raising the odds of harmful effects. Evolution therefore constrains abundant proteins more tightly.
Starr and Fraser, researchers at Stanford University, asked whether this same principle applies at the level of neuronal cell types. Do the most common brain cells evolve more slowly than rarer ones, simply because errors in them would be costlier for survival?
To test the idea, the team drew on massive single-nucleus RNA sequencing datasets from three different regions of the mammalian neocortex. Together, these datasets covered more than 300,000 neurons from six species, including humans, chimpanzees, gorillas, rhesus macaques, marmosets, and mice.
The result was striking: across species and brain regions, abundant neuronal types conserved their gene expression patterns far more strongly than rare ones. This rule held robustly across multiple datasets, statistical tests, and even when separating excitatory from inhibitory neurons. A universal principle seemed to be at work: common cell types evolve slowly, rare cell types evolve fast.
The human exception
But when the team zoomed in on human brains, they found an exception that rewrites the story.
Our most abundant excitatory cells - layer 2/3 intratelencephalic (IT) neurons - did not show the expected conservation. Instead, in humans, these neurons had undergone accelerated gene expression divergence, evolving much faster than in other primates.
Comparisons of human, chimpanzee, and gorilla brain samples revealed that the acceleration occurred on the human lineage alone. By contrast, chimpanzee neurons followed the expected slow-evolution pattern.
Layer 2/3 IT neurons are central to higher cognition. They connect across cortical regions, integrating signals and supporting complex abilities like language, abstract reasoning, and social intelligence. In most species, they remain tightly conserved. In humans, they broke free of the rule.
Autism genes at the center
The researchers then asked which genes drove this accelerated change. The answer was as surprising as it was telling: many were autism-associated genes.
Human layer 2/3 IT neurons consistently showed down-regulation - lower expression levels - of genes linked to autism spectrum disorder (ASD) and schizophrenia. This was not a random drift. Evidence pointed to polygenic positive selection: coordinated evolutionary pressure favoring lower expression of these genes.
Hybrid human - chimpanzee organoid experiments confirmed the effect was intrinsic to human DNA. When the two genomes were placed in the same cellular environment, the human alleles produced lower gene expression than chimpanzee alleles, proving that the difference was coded into our lineage.
A trade-off at the edge of fragility
This down-regulation comes with consequences. Many autism-linked genes are dosage-sensitive - even modest reductions in their expression can push neuronal circuits into atypical patterns. The study suggests that human evolution tuned these genes downward to stabilize new forms of cognition, but in doing so left our species closer to a fragility threshold.
Above the threshold: the extraordinary expansion of human intelligence. Below the threshold: traits and vulnerabilities associated with autism.
Other lines of evidence reinforced the story. Protein studies showed that synaptic proteins like PSD-95 are present at much lower levels in humans than in chimpanzees or macaques. Genetic analyses confirmed that these expression changes were unlikely to be the result of relaxed constraints; instead, they bore the signature of active selection.
Why this matters
The findings provide some of the strongest evidence to date for a long-standing hypothesis: natural selection for human-specific traits increased our risk of neurodevelopmental disorders.
This fits a broader pattern in evolution. Sickle cell mutations protect against malaria but cause anemia. Variants that sharpen immune defenses can trigger autoimmune diseases. In the same way, the genetic changes that powered human cognition appear to have raised our baseline susceptibility to autism.
The study also highlights how autism may be uniquely human. Traits linked to ASD - such as language differences, theory of mind, and altered social cognition - map directly onto the very capacities that expanded most dramatically in our lineage. Non-human primates show little evidence of autism-like behavior at population levels, reinforcing the idea that autism is entwined with human brain evolution.
Beyond autism: a broader principle
Importantly, the study also advances evolutionary theory itself. By showing that cell type abundance predicts evolutionary constraint, it extends a rule first observed at the protein level to a higher level of biological organization.
In doing so, it suggests a scalable logic of evolution:
- Genes evolve slower when they are highly expressed.
- Cell types evolve slower when they are highly abundant.
- But humans demonstrate that selection can override the rule when adaptive gains are great enough - in this case, producing advanced cognition.
Seven Reflections view
This research forces us to confront an uncomfortable truth: intelligence and vulnerability are not opposites, they are entwined outcomes of the same evolutionary currents.
Autism, seen in this light, is not a deviation from humanity but an echo of how humanity itself was forged. The same changes that made us capable of art, science, and culture also brought us closer to fragility, weaving neurodiversity into the very fabric of our species.
Evolution did not give us brilliance for free. It asked for a price - and part of that price is the higher prevalence of autism in humans.
