ABOVE: A blue-and-yellow macaw (Ara ararauna), a core land bird © ISTOCK.COM, BORYAK

Mammals and birds have dramatically more neurons in their forebrain and cerebellum than reptiles, and neuron numbers have scaled up significantly only four times in more than 300 million years of brain evolution in the clade that includes reptiles, birds, and mammals, according to a study published in PNAS on March 7. Instead of brain volume, which has long been used as a proxy for brain complexity, the study’s authors used the number of neurons typically found in species’ brains as an indicator of smarts.

“Reptile brains are smaller than the brains of birds or mammals of similar body size, but just how much smaller and how the size difference translates into differences in behavior and cognition is a problem that has eluded scientists for a long time,” Enrique Font, a zoologist and ethologist at the Universidad de Valencia in Spain who was not involved in the study, writes in an email to The Scientist. “This is an important paper that goes a long way to explaining the differences in brain size/structure among different groups of amniotic vertebrates.”

Brain size is often used as an indicator of cognitive capacity. “But that’s a very crude proxy because the composition of brains differs from taxon to taxon. So it is much more precise to count cells, and neurons specifically,” Pavel Němec, a study coauthor and evolutionary biologist at Charles University in Prague, Czech Republic, tells The Scientist. Němec adds that it would be even better to combine neuronal number with the number of synapses to estimate complexity, “but we currently don’t have a tool to measure the number of synapses precisely, and definitely not across many species.” 

In the study, the researchers used the isotopic fractionator, a method developed by Suzana Herculano-Houzel in 2005 that quantifies neuron number quickly and cheaply by homogenizing brain structures and labelling intact nuclei. With the isotopic fractionator, Němec and his coauthors counted neurons in the forebrain, cerebellum, and “rest of brain” in bird and reptile species, and compared them to the same measures of neuron numbers in mammalian brains, drawn mostly from literature published by Herculano-Houzel.

In previous work, Němec and colleagues showed that birds have high neuronal densities. “They basically compensate, with these densely packed neurons, [for] the fact that they have relatively small brains in absolute terms, but they have just as many neurons as mammals,” he says. But they didn’t know whether that was true of reptiles as well. In the new study, the researchers found that reptiles have very low neuronal densities, with an average neuron number 20 times lower than that of birds or mammals of similar body size. 

See “Altruism in Birds? Magpies Have Outwitted Scientists by Helping Each Other Remove Tracking Devices

With a phylogenetic analysis in the current study, the researchers show that the relationship between neuron number and brain size changed in a major way only four times in the evolution of land vertebrates. “With the appearance of birds and mammals, brains not only enlarged, but also density increased a lot,” says Němec. Within mammals, previous studies had established that primates have higher neuronal density. Within birds, the new study finds that so-called core land birds, a group that includes woodpeckers, falcons, and parrots, also have relatively large brain sizes and densities in the brain. “The appearance of birds and mammals, and within these groups independently the two crown groups [core land birds and primates] . . . increased processing power significantly,” says Němec. “One highly surprising finding is that it was actually very rare, such occasions. We expected that it would be changing within evolution, going up and down all the time. This is partly true, but these really big changes are extremely rare.” 

With the appearance of birds and mammals, brains not only enlarged, but also density increased a lot.

—Pavel Němec, Charles University

Based on the findings, Němec draws a clear distinction between reptiles and the more neuron-dense birds and mammals. Reptiles “have adopted an economic way of living, and it’s not compatible with a large brain because neurons are metabolically demanding,” he says. By contrast, “birds and mammals have adopted a very different strategy, an expensive way of living with large brains and high cognitive abilities.” On average, birds and mammals have 20 times more neurons than reptilian species, he says, with even more in primates and core land birds. 

Barbara Finlay, a cognitive neuroscientist at Cornell University who was not involved in this study, says that the researchers present a “useful piece of information,” particularly basic data long missing about reptiles. However, she questions whether neuron numbers—or any other single factor—in isolation can really be a proxy for computational power. “Counting up numbers does not equal cognition,” she tells The Scientist. 

Additional information about the brain’s morphology and connectivity, as well as the way different types of neurons are packed into a brain region, would improve brain power estimates, Finlay says. “Brain mass has many aspects that anchor its computing power. Since neurons vary widely in size and synaptic density across structures and species, the number of synapses, the organization of single regions, the overall network structure of the brain and brain energy consumption are all important,” she adds in an email to The Scientist. 

She also points out that the cortex is less neuron-dense than the cerebellum, but has a range of functions, which she suggests shows that counting neurons gives an incomplete picture of cognition. If the cerebellum is lost or damaged, motor coordination may become poor, she notes; by contrast, losing the cerebral cortex results in, among other things, “the loss of visual guidance of movement and recognition of objects, all of language, speech, facial and emotional recognition, initiating and planning ranging from motor skills to life-course strategies, and moral understanding,” Finlay writes. 

“Approaching brain evolution analysis from the more granular approach using neuron number has its advantages,” Sean O’Hara, a researcher in social and cognitive evolution at the University of Salford in the UK who was not involved in the study, writes in an email to The Scientist. It could, he says, be particularly useful in some exceptional cases: “For example, selection for small size may occur in rapid-flying animals living in complex three-dimensional habitats as increased flying manoeuvrability will be favoured. Although absolute brain size or regional brain size may decrease under such circumstances, one wouldn’t expect neuron density to do so.”

“Neuron number is certainly an important variable, but my guess is that it will be most useful if combined with other variables, such as relative brain size,” Font writes. Němec agrees that additional variables would be useful. “You might have different sizes of neurons, different numbers of cortical connections. But this data is simply not available and certainly not available for many species.” In studying vertebrate brain evolution on a large scale, neuron number would be a good proxy, he argues. “If an animal has billions of neurons, it’s definitely more clever than an animal that has millions of neurons. But I would not say that it is a very tight correlation.”