What if your ‘lizard brain’ never existed? Scientists challenge a 70-year-old theory and uncover the wiring battle that may have shaped human intelligence

Published in Science Advances, the study draws on comparisons of biological brains and experiments on artificial neural networks to suggest a new framework for understanding the evolution of intelligence.

For decades, the human brain was like some evolutionary skyscraper: ancient instincts on the bottom, emotions in the middle, and sophisticated reasoning on the top. That’s a good reason why we sometimes feel stuck between impulse and logic. But the evolutionary story of the brain may be far messier. New research from Georgia Tech challenges the popular notion of a primitive “lizard brain” sitting below a newer, rational brain. Instead, scientists suggest, it may have been a competition between fundamentally different ways of wiring neural circuits, with limited brain space forcing different systems to compete, which influenced brain evolution. Published in Science Advances, the study draws on comparisons of biological brains and experiments on artificial neural networks to suggest a new framework for understanding the evolution of intelligence. The old model was devised in the 1950s and perceived the brain in evolutionary layers, with basic bodily functions at the bottom, followed by an emotion-driven reptilian brain and finally the complex neocortex associated with human reasoning. But researchers now say this view is overly simplistic. The neocortex is involved in functions including vision, perception and reasoning. The so-called limbic system, often loosely described as the “reptilian brain,” is much more complicated. Its components play a role in memory, smell, navigation and emotional regulation and other functions. Rather than looking at the regions as separate structures, the Georgia Tech team examined how those regions change together across species.

A concerted evolutionary change

They discovered that if one part of the limbic system was relatively large, then other parts of the limbic system were likely to be larger as well. At the same time, the neocortex generally occupied less of the brain. That coordinated pattern suggests these regions may function as interconnected systems whose sizes shift together during evolution. This finding raised another question: what could drive this trade-off? Perhaps the answer lies in the wiring of the brain before birth.

Two different ways to wire a brain

Neocortical circuits often are arranged as spatial maps. Adjacent parts of the body correspond to parts of the brain, and visual and auditory information can also be organized in terms of spatial relationships. The limbic system goes the other route. Its connections are more distributed, like a barcode, with specific patterns of neural activity representing complex information such as smells or memories. To see if these differences were fundamental or just learned through experience, the researchers used artificial neural networks. Networks with localized connections performed well on vision, sound and touch tasks. Distributed wiring, meanwhile, was better suited to smell recognition and memory tasks.

An evolutionary struggle for brain space

The team then modeled how these two wiring strategies would compete for scarce neural resources. The system in an artificial environment that rewarded smell, grew while the simulated neocortex shrank. When vision became more important, the balance shifted. The model shows differences seen in real animals. The nine-banded armadillo has a relatively large limbic system, which is heavily reliant on smell. Squirrel monkeys are highly visually dependent and their brains are dominated by the neocortex. And in 182 species, the scientists saw evidence that evolution isn’t just adding new layers of intelligence. Instead, it may be allocating scarce neural real estate to whatever wiring strategy best suits an animal’s surroundings.

What this could mean for AI

The finding could have implications beyond understanding the evolution of biology. Contemporary artificial intelligence systems usually rely on large scale training data. But biological brains start out with a lot of prewired architecture and then combine that built-in structure with experience. Recreating some of that architecture in artificial neural networks could one day lead to artificial intelligence systems that learn with less data and consume less energy. Thus, the research suggests an alternative view of intelligence, not as a simple hierarchy of ancient and modern brain regions, but as the result of an evolutionary balancing act between different ways of processing information. The “lizard brain” is a catchy metaphor, but the truth appears to be far more complicated. Evolution didn’t just build a smarter brain on top of an older one. It may have been all the time negotiating what kind of wiring was worth the space.