Sunday, July 17, 2016

Suzana Herculano-Houzel's The Human Advantage is a good book with some important insights hidden behind a fairly dry and dense presentation on "how I made these important discoveries." Herculano-Houzel (by her own account) pioneered a technique for determining the number of neurons in brain tissue, and managed (through a fair number of mildly interesting adventures) to bring together samples of many different primate, rodent, and other mammalian brains in order to work out the scaling laws that govern how brains and neuron counts grow with body mass in different tissues across different lineages. She shows a lot of graphs and charts to demonstrate that (with two exceptions) for most mammals, neuron counts scale up with an exponent of .5 with body mass, but in primates, the scaling factor is .8. If neurons have to be added in order to increase intelligence, this means that primates have a huge advantage. In order to get smarter, brains and neuron count have to increase. Larger bodies are necessary in order to sustain a larger brain, and if neuron count or neuron density is the limiting factor in intelligence, then you want to be able to pack more neurons into a denser brain in order not to require an enormous body.

Once Herculano-Houzel has established the basic scaling laws, she delves into the economics of maintaining a sufficiently neuron rich brain. The comparative scaling laws mean that as you look at bigger and bigger species of mammals, the neuron counts increase with the square root of the body mass. As primate species get larger, their neuron count increases much more quickly, which means that for a given body size, a primate is capable of supporting a larger brain than would a mammal. The cross-over point where the two lineages have comparable neuron sizes are with body sizes in the 10 gram range. At any larger size, if you compare a primate with another similar sized mammal, the primate is probably smarter.

There is a separate literature showing the energy requirements for many particular species. Herculano-Houzel used her new numbers on neurons for rodent and primate species to show that there is a direct correlation between the number of neurons and the amount of glucose consumed per minute by the brain. For humans (and others in the Homo lineage, being able to take advantage of the primate scaling laws gives a big boost, but you still have to find a way to ingest sufficient calories to afford the bigger brain.

Humans have two main advantages on this score compared to other primates. Walking on two legs is much more energy efficient than knuckle-walking like other (primarily arboreal) primates or on four legs. This increases the range over which foragers could range, and also freed up hands for gathering and carrying. It's not clear what originally drove bipedalism in the homo line, but it occurs at the same branch point that leads to the massive growth in cranial capacity.

The other big human advantage is cooking. I've seen discussions before that cooking increases the efficiency of digestion, and led to our shorter digestive tract, which allowed us to switch energy resources from digestion to our brains, but Herculano-Houzel points out that even before control of fire, other kinds of preparation (chopping and mashing for example) reduce the energy required for digestion. The anthropological evidence for food preparation goes back much further than the evidence of cooking, and significantly after bipedalism. The earliest evidence of eating meat is swiftly followed by anatomic adaptations to a more efficient diet, which is quickly followed by better tools, and the then bigger brains. Part of the evolutionary adaptation for bigger brains included smaller jaws.

Evidence of tool use and manufacture date back to 3.3 MYA (Million Years Ago). This date was recently pushed back from 2.6 MYA. These tools were simple flint knives. Archaeologists wouldn't count rocks that were used for pounding, since they are impossible to distinguish from unworked rocks. The flint knives would have been useful for cutting up meat, which would make it more digestible, and is necessary in order to survive with smaller jaws. Presumably, eating primitively processed foods had to become habitual before later evolutionary steps that relied on it would have survived in the population. The archaeological evidence gives the following timeline:

  1. 4.4 MYA: bipedalism appears
  2. 3.3 MYA: earliest tool use
  3. 2.5 MYA: eating meat
  4. 2.4 MYA: beginning of the reduction in size of the jaw
  5. 1.9 MYA: smaller gut is clearly present
  6. 1.7 MYA to 300 KYA: The Acheulean hand axe
  7. 1.5 MYA-100 KYA: start of the increase in cranial capacity
  8. 1 MYA: Clear indications of cooking

Another tantalizing clue is that the taste for cooked food may pre-date adoption of the habit. Herculano-Houzel refers to two studies that show that chimpanzees have a very strong preference for the taste of cooked food over raw. I don't know whether this has been investigated in other lineages, but if so, (even if it's just the body innately being able to detect foods that are provide big efficiency gains) it provides a boost for any lineage that can figure out how to reliably prepare foods--once you start, it would be an easy habit to keep, providing that the right food sources and tools are accessible.

Earlier, I mentioned that there are two exceptions to the laws regulating the number of neurons in primates and in all other mammals. The first is gorillas, which have brains and neuron counts much closer to those of other mammals rather than those expected of a primate. This fact about gorillas has been throwing off the results of previous researchers, who could only measure brain capacity. They concluded that the rules for primates would be the same as for other mammals, and argued that it was humans that were outliers. Once you plot the detailed data from small and medium primates and compare to mammals, it's easy to see a different trend line applies, and that humans fit on the primate line and gorillas do not. The other exception is elephants. (Herculano-Houzel has an entertaining section about her adventures getting elephant brains to analyze.) Elephants have brains whose size follows the standard scaling rule for mammals. They're huge, and they have huge brains. But their neurons are distributed very differently from all other species. 98% of the neurons are in the cerebellum, while the normal number doesn't get much above 80%. So elephants have big brains and a lot of neurons, but this explains why they're not even smarter than us, presuming neurons in the cerebral cortex are the thing that matters most.

Anyway, the later clues about cooking and bipedalism only added to my reaction that this work may provide an improved answer to the Fermi paradox. Herculano-Houzel doesn't appear to have data about the brains of animals beyond mammals, but if all the mammals outside of primates share a common scaling factor, then that's an indication that it's hard to evolve intelligence given the standard energy budget. It takes a special trick (which didn't have an immediate obviously benefit in the small primates in which it evolved) which was only discovered in one previously obscure branch of the mammal family tree to enable the efficient scaling that allows bodies to grow large enough to support brains supporting enough neurons to enable tool use. This enables (with other accidents like bipedalism and prepared food appearing in the same lineage) the feedback cycle that led to our massive growth in intelligence.

I've never been very worried by the argument that says the Fermi paradox indicates that there's a Great Filter, and if we can't figure out what the hard step was in our past, we should expect to encounter a hurdle in our future that has stopped other species from getting to space. The Human Advantage makes me even more sanguine. It's hard to evolve an intelligent species. There are a lot of happy accidents in our past, and the likely number of extra-terrestrial species in our light cone may be smaller than we thought. It would be nice to see more data showing the scaling laws that apply outside the primates (and in the cetaceans, which she didn't give much data about). I'll be surprised if any of them show divergent scaling progression compared to baseline mammals.


Peter McCluskey said...

I don't think it's much of an accident that bipedalism and cooking evolved in the same ape lineage. The Secret of our Success suggests that the main accident which distinguished our ancestors from other apes was leaving trees, to spend most of their time on the ground. Most distinctive human features seem somewhat likely given that one accident.

It's hard to evaluate how much intelligence depends on the primate brain-scaling innovation. Cetaceans appear to be fairly close to having enough neurons to produce civilization (because marine animals can get more food than land animals can). Also, it seems puzzling that evolution didn't stumble on other ways to create brains with many neurons, such as brains that drastically reduce energy use per neuron when nothing important is happening.

So I'm more concerned about the great filter than you are.

Robert Ayers said...

Thanks for the interesting review! Got me to review my brain nomenclature :-)

I wonder what is special about the elephants? There have been many recent tales of intelligence and feelings in elephants. Maybe they aren't good at abstract reasoning but very good at concrete reasoning?

Robert Ayers said...

I wrote "maybe elephants aren't good at abstract reasoning but ..." Probably wrong.

At the very end of the article on moving elephants to Malawi, we see:
Unlike many other species, Vickery said, the first thing that a darted mother elephant does on recovering is look for her calf.

And I recall reading mebbie a year ago an article about a baby elephant that had died, and the mother stayed there and kept trying to stand the dead baby upright ...

I wonder if the book tries to tell us about the significance of the brain results -- and whether they have been replicated.

Peter McCluskey said...

The long-finned pilot whale has 37.2 billion cortical neurons, versus 21 billion for humans.