Chapter 21: The Perfect Storm

Core Idea

Human brains stayed chimp-sized for millions of years after our split from chimpanzees (around seven million years ago). Then, starting around two and a half million years ago, they became over three times larger. A dying forest pushed an ape onto the savannah, into meat eating, tool making, cooking, and premature births. Language raises a puzzle that no earlier breakthrough did. It helps only when others use it honestly, so liars and freeloaders should defeat it. Bennett’s answer is that language began among kin, as parents taught children to make tools. Then gossip plus punishment of cheaters made altruism toward non-kin pay. That started a feedback loop of gossip, altruism, and language skill, which kept pushing brains bigger. Language, altruism, and cruelty came out of the same loop.

Frameworks Introduced

• The ecological story, step by step:
  1. Until ten million years ago eastern Africa was dense forest. Tectonic shifts built the Great Rift Valley, which cut off ocean moisture. The forest died back into a mosaic of tree patches and grassland.
  2. Around six million years ago the mountains split the apes. Figure 21.2 is a map with a dotted line along the rift. West-side apes (a knuckle-walking chimp, in the forest band to the west) stayed largely unchanged and became chimpanzees. East-side apes (an upright walker) became human.
  3. By four million years ago they walked on two legs but had chimp-sized brains and no extra tool use: “upright-walking chimpanzees”.
  4. Around 2.5 million years ago they became scavengers. Up to 30 percent of their diet was meat, against about 10 percent for chimps. They made Oldowan tools in three steps: find a hard hammerstone, find a more fragile core (quartz, obsidian, or basalt), and smash them to get sharp flakes and a chopper (Figure 21.3 shows a hammerstone striking a core, with a flake falling off).
  5. About five hundred thousand years later came Homo erectus, an apex predator and hypercarnivore with about 85 percent meat in its diet. Its brain was twice the size of the upright ancestor from a million years before. It made stone hand axes and had a body built for throwing (a chimp throws at only about twenty miles per hour, and a preadolescent human almost three times as fast). It also had a body built for endurance running (hairless skin, sweat glands, long legs), and possibly used persistence hunting.
  6. Its jaw and gut shrank. Richard Wrangham’s explanation is that H. erectus invented cooking, which lets animals absorb 30 percent more nutrients. Over 50 percent of people on fully raw diets become temporarily infertile. The calorie surplus could be spent on bigger brains.
• The obstetric dilemma (Sherwood Washburn): big brains must pass through hips narrowed by walking upright. The human solution is premature birth, followed by twelve years of brain growth. This pushed H. erectus toward mostly monogamous pair-bonding, active fathers, and possibly grandmothering (only orcas and humans have females that stop reproducing long before death).

Species	Percent of adult brain size at birth	Time until full brain size achieved
Human	28 percent	12 years
Chimpanzee	36 percent	6 years
Macaque	70 percent	3 years

• The altruism problem of language:
  • When to use: to test any story of how a group-benefiting trait evolved.
  • How: genes appear in individuals, not groups, so ask whether a single carrier gains. Simple group selection, the idea that evolution favors whatever is good for the species, fails. Defecting in language, by lying or withholding, pays the individual. A group full of liars could even favor those immune to words.
  • Two kinds of altruism do evolve. Kin selection works because a child or sibling shares a given gene with 50 percent probability, a grandchild 25 percent, and a cousin 12.5 percent. Most animal altruism is kin selection: vervet alarm calls near family, beehives of sisters. Reciprocal altruism (“I’ll scratch your back if you scratch mine”) works only if defectors are detected and punished, as with red-winged blackbirds that stop defending the nests of neighbors who did not help them.
• The emergence of the human hive mind (Bennett’s proposed solution):
  1. Language starts between parents and children, so that tool-making skills pass on. For H. erectus, “A Homo erectus without a stone hand ax was as doomed as a lion born without teeth.” Simple phrases suffice (“Berries. Home tree”, “Quiet. Danger”, “Mom. Here”). Kin selection explains this with no group selection. The language curriculum is also strongest between parents and children.
  2. Language spreads to non-kin. It stays unstable until gossip appears. Robin Dunbar found that as much as 70 percent of human conversation is gossip. Gossip exposes cheaters (“Did you hear that Billy stole food from Jill?”) and rewards heroes.
  3. A feedback loop forms (Figure 21.5, a circle of three nodes with arrows marked …creates pressure for…). More gossip and punishment of violators creates pressure for more altruism toward non-kin. That creates pressure for more language skill, which creates pressure for more gossip and punishment. The center label is Progressive emergence of language, cruelty, and altruism.
  4. Brains grow on each turn of the loop, because bigger groups need more social tracking, more accumulated ideas need storage, and better-shared simulations reward richer simulations. Meanwhile the ceiling on brain size rises too: better hunting and cooking add calories, and earlier births and longer childhoods allow more brain growth after birth.
  • Failure mode: this is one contested story. Alternatives are listed under Anti-patterns.

Key Concepts

• Runaway growth of the brain: John Ingram’s phrase for the tripling that began over two million years ago. Figure 21.1 plots cranial capacity (log scale) against time. The trend creeps up slowly from 10 million years ago through Australopithecus, then bends sharply upward around 2 million years ago through early Homo and H. erectus to Neanderthals and H. sapiens.
• Wallace’s problem: the difficulty of explaining language by evolution. It is named for Alfred Wallace, who invoked God to explain it, prompting Darwin’s letter: “I hope that you have not murdered too completely your own and my child.” In 1866 the French Academy of Sciences banned publications on language origins.
• Two indisputable milestones (Figure 21.4 timeline): 2.5 million years ago, first humans (Homo habilis), brains start growing, first Oldowan tools. 2 million years ago, H. erectus appears and humans become apex predators. 500,000 years ago, language-ready vocal cords appear in the shared ancestor of H. sapiens and Neanderthals. 100,000 years ago, the first evidence of symbology (fictional sculptures, abstract cave art, nonfunctional jewelry), and H. sapiens is biologically modern.
• Protolanguage: declarative labels and perhaps simple grammar with a narrow range of sounds, which H. erectus probably had.
• Evolutionarily stable strategy: altruism is not one, because freeloaders do better unless they are punished.
• Mutually beneficial arrangements: most group behavior in animals (fish shoals, wildebeest herds), where defecting hurts only the defector.
• Exaptation and spandrel: an exaptation is a trait repurposed later (feathers first insulated, later served flight). A spandrel is a side effect of another trait (male nipples).

Mental Models

• Think of the human paradox (most altruistic to strangers, yet the only species that commits genocide) as one loop, not a coincidence. Our “instinct to punish those who we deem to be moral violators” is the price of our altruism.
• Use the gossip-punishment loop when asking how any cooperative signal survives cheaters: cheap reputation tracking plus real costs for defectors.
• Think of language as the reason a small brain can still be smart. H. floresiensis shrank to under four feet tall with a brain the size of a chimp’s, perhaps smaller, yet kept H. erectus-level tool use.
• Use the perfect-storm view for why language evolved only once, while eyes and wings evolved many times: it needed many unlikely effects interacting together.

Anti-patterns

• Group selection for the species: “genes do not spontaneously appear in groups, they appear in individuals.”
• Explaining language like eyes: an eye helps its owner alone. Language helps only if others use it honestly.
• Brain size as the whole story: H. floresiensis argues against it.
• Treating the story as settled: rivals include cooperative hunting as a mutual benefit (no altruism needed), cooperation evolving before language, Chomsky’s view that language evolved first for inner thinking and was later exapted, and language as a spandrel of musical mating songs. “The debate continues.”

Rules of Thumb (author’s wording)

• “I would happily lay down my life for two brothers or eight cousins.” (J. B. S. Haldane)
• “The essential feature for reciprocal altruism to successfully propagate throughout a group is the detection and punishment of defectors.”
• “The more severe the costs of cheating, the more altruistic it was optimal to behave.”
• “Herein lies both the tragedy and beauty of humanity.”

Worked Example

The freeloader thought experiment, and Bennett’s way out.

1. Suppose 10 percent of group A are altruists who share food locations and teach tool use, and 90 percent are not. A freeloader who takes the lessons but gives nothing back does better, so altruistic genes should not spread.
2. Language makes this worse than in fish shoals. A fish that leaves the shoal is eaten first, but a human who lies gains. “So the presence of language creates a niche for defectors, which eliminates the original value of language.”
3. Step one out: start where kin selection already works. Parents teaching children to make hand axes is ordinary parental investment.
4. Step two out: let gossip spread news of cheaters and heroes, and let groups punish cheaters by withholding help or by harm. Altruism now pays, sharing information pays, and better language makes gossip work better.
5. Supporting clue: H. erectus fossils on Flores. The island lay over thirty miles off Indonesia and was isolated by water for over a million years. Even at the lowest sea levels, reaching it meant crossing twelve miles of open water, and tools there date from as early as one million years ago. This suggests rafts and cumulative, language-based know-how.
6. Outcome: H. erectus left Africa and humans split into at least four species by one hundred thousand years ago (Figure 21.6 family tree: Australopithecus 3.5 million years ago, H. habilis 2.5 million, H. erectus 2 million, branching into H. erectus and H. floresiensis on one side and, around 500,000 years ago, H. sapiens and H. neanderthalensis on the other). H. sapiens left Africa around seventy thousand years ago and was the only human species left by forty thousand years ago.

Key Takeaways

1. Brain growth started about 2.5 million years ago with a shift to meat, tools, and later cooking, not with bipedalism.
2. Premature birth and twelve years of brain growth reshaped parenting toward pair-bonds and grandmothers.
3. Language is the only breakthrough that does not clearly benefit the single individual in whom it first appears, which is Wallace’s problem.
4. Kin selection explains language between parents and children. Gossip plus punishment explains its spread to strangers.
5. The gossip, altruism, and language loop also produced in-group conformity and out-group cruelty.
6. H. erectus probably had protolanguage. Modern-level language existed by one hundred thousand years ago.

Connects To

• Ch 15: Dunbar’s social-brain hypothesis and primate politics, which this loop turned into “a cudgel of coordinated conquest”.
• Ch 15–18: primate grooming and food sharing as reciprocal altruism; theory of mind underlies gossip.
• Ch 19: accumulation of ideas is what bigger brains had to store.
• Ch 20: the parent-child language curriculum.
• Ch 22: the summary of breakthrough #5.