William H. Calvin, "Pumping Up Intelligence: Abrupt Climate Jumps and the Evolution of Higher Intellectual Functions during the Ice Ages," in The Evolution of Intelligence, edited by R. J. Sternberg (Erlbaum, 2001), pp. 97-115. See also

There may be minor differences with the printed version.

Webbed Reprint Collection
This 'tree' is really a pyramidal neuron of cerebral cortex.  The axon exiting at bottom goes long distances, eventually splitting up into 10,000 small branchlets to make synapses with other brain cells.
William H. Calvin

University of Washington
Seattle WA 98195-1800 USA

Pumping Up Intelligence


Abrupt Climate Jumps and the Evolution of

Higher Intellectual Functions during the Ice Ages

William H. Calvin
University of Washington
Seattle WA 98195-1800 USA

The title is not a metaphor, though past tense might be better as this chapter is about how each of the many hundred abrupt coolings of the last several million years could have served as a pump stroke, each elevating intelligence a small increment - even though what natural selection was operating on was not intelligence per se.

While we often use the term 'intelligence' to encompass both a broad range of abilities and the efficiency with which they're enacted, it also implies flexibility and creativity, an "ability to slip the bonds of instinct and generate novel solutions to problems" (Gould and Gould 1994, p. 70). Those three pillars of animal intelligence - association, imitation, and insight - are also impressive (Byrne 1994), as are the occasional symbolic (Deacon 1997) and reasoning (Gould & Gould, 1998) abilities. But Piaget (1929; 1952) said that intelligence is what you use when you don't know what to do, when neither innateness nor learning has prepared you for the particular situation.

Intelligence is improvisational. Still, most of the time, not much improvisation is necessary; the individual has encountered somewhat similar situations before, has a repertoire of actions, and simply starts one - and gropes, using feedback's progress reports to guide to the goal. No major planning is needed in most cases, and thus not much in the way of intellectual wherewithal. This suggests a focus on those few behaviors that require an elaborate multistage plan prepared during "get set." An example would be the ballistic movements (hammering, clubbing, throwing, kicking, spitting) where the speed of feedback is so inadequate (they're often over-and-done by the time that progress reports can start modifying the movement), where only near-perfect plans will succeed.

What the mind is often seeking during "get set", I suspect, is coherence - finding a conceptual combination of sensory input, memories, and movement plans that fit together particularly well - though, because of novelty, most will rate less than the "perfect ten" of exact, unambiguous fits. Similar to this is Barlow's (1987) suggestion that intelligence is all about making a guess that discovers some new underlying order. "Guessing well" neatly covers a lot of ground relevant to higher intellectual functions: finding the solution of a problem or the logic of an argument, happening upon an appropriate analogy, creating a pleasing harmony or witty reply, or guessing what's likely to happen next.

Higher Intellectual Functions


Because they all involve pattern-finding and all emphasize human abilities not widely shared with the other apes, I want to restrict myself here to the higher intellectual functions. I will use "intelligence" as a term denoting the speed-and-scale of individual performance of them, not unlike the manner in which much of the variance of the general factor g can be accounted for (Jensen 1992) by the subtests that emphasize speed of performance and the number of items that must be borne in mind simultaneously - as in those multiple choice analogies: A is to B as C is to [D, E, or F] which require six concepts to be simultaneously managed. Closely related is our ability to remember phone numbers long enough to dial them. Many people can retain a seven-digit number for 5-10 seconds, but will resort to writing it down if faced with an out-of-area number or an international one of even greater length. When getting close to your limit, you try to collapse several items into one chunk, so as to make more room (Simon, 1983).

Definitions of "higher intellectual function" vary; I tend to use the phrase to refer to the structured mental abilities such as

• syntax, the structuring schemes of phrases and clauses used to disambiguate sentences longer than a few words, e.g., "I think I saw him leave to go home" has nested embedding involving four verbs. (Syntax is an evolutionary puzzle because there aren't obvious intermediate forms in development or aphasia between short structureless protolanguage sentences and recursive embedding.)

• planning, those speculative structured arrangements, e.g., "Maybe we can go to the country this weekend if I get my work finished, but if I have to work Saturday, then maybe we can go to a movie on Sunday instead." (Squirrels hoarding nuts isn't planning but an innate behavior triggered by longer nights releasing more melatonin from the pineal. Holding an intention for a few hours isn't planning either, and I'd also exclude foraging behaviors that could be explained more simply by choosing between familiar migration routes. Perhaps we should reserve the term for something that requires multiple stages of the move to be assembled in advance of action, rather than organizing the later stages after getting the initial moves in motion, which goal-plus-feedback can accomplish.)

• chains of logic, our prized rationality. But the emphasis here is on novel chains, not routine ones. The most mindless of behaviors may be segued, the completion of one calling forth the next: courtship behavior may be followed by intricate nest building, then a segue into egg laying, then incubation, then the stereotyped parental behaviors. (Kφhler's chimps, that piled up boxes to reach the hanging banana, might qualify under the novel chaining requirement, if simpler explanations can be eliminated.)

• games with arbitrary rules, such as hopscotch. (Both Pan species, chimpanzees and bonobos, have a version of "blind man's bluff" - but it isn't structured.)

• music of a structured sort, such as harmony and counterpunctual themes. (Perhaps not rhythm per se but certainly rhythms within rhythms.)

Obviously, if one relaxes the nested-or-chained structural requirement, there are various primate behaviors that are possible evolutionary precursors.

I am focusing here on structuring because humans exhibit such a large increment in ability over our Pan cousins across these five areas - and because I am interested in whether there is a "common core" of neural machinery that is shared by all such structured behaviors, one where improvements in any one of the five might improve the other four "for free."

Evolutionary arguments often commit the reification fallacy (a "gene for intelligence"). Indeed, we often assume that an abstraction like language implies a real concrete entity such as a "language module." (Separate, of course, from any planning module! And so on, to the balkanization of the mind.) Yes, there is localization of function - and we certainly tend to name a cortical area according to the first of its functions we discover - but multiple functions are commonplace (see chapter six of my Cerebral Code for a discussion of how a neocortical area could alternate between being a narrow specialist and performing as a general-purpose scratch board).

Language in the Multiple Use Context


Multiple uses of a structural entity are common, and a familiar example is the "curb cut" where the steplike curb is locally smoothed into a gentle ramp. What paid for curb cuts was, of course, wheelchair requirements. But as soon as a curb cut is in place, 99 percent of the traffic is for secondary uses such as bicycles, suitcases, baby carriages, grocery carts - none of which would have "paid for it."

A secondary "free" use may, of course, later pay for further improvements (just imagine the skateboarders holding a bake sale to pay for widening!), suggesting that the evolutionary history of higher intellectual function might first emphasize one structured use and later others. If the notion of a "free lunch" offends, note that it is commonly assumed that music is a spare-time use of the language-related parts of the brain, that there was likely little natural selection for four-part harmony via barbershop quartets.

Language is the most defining feature of human intelligence: without the orderly arrangement of verbal ideas permitted by syntax, we might be little more clever than Pan. For a glimpse of life without syntax, consider the Sacks (1989) description of Joseph, an 11-year-old deaf boy. Because he could not hear spoken language and had never been exposed to fluent sign language, Joseph did not have the opportunity to learn syntax during the critical years of early childhood:


"Joseph saw, distinguished, categorized, used; he had no problems with perceptual categorization or generalization, but he could not, it seemed, go much beyond this, hold abstract ideas in mind, reflect, play, plan. He seemed completely literal -- unable to juggle images or hypotheses or possibilities, unable to enter an imaginative or figurative realm....He seemed, like an animal, or an infant, to be stuck in the present, to be confined to literal and immediate perception, though made aware of this by a consciousness that no infant could have."


"Language cortex" isn't just the traditional Broca and Wernicke areas but much of the lateral aspects of the temporal and frontal lobes, plus the parietal lobe areas near the left sylvian fissure (see, for example, Ojemann 1991). Language localizations have a strong overlap with nonlanguage sequential functions such as sound strings and hand-arm sequencing (most aphasics have some form of hand-arm apraxia; see Kimura 1993).

And this overlap brings me to an important point: the use that initially "paid" for the structuring abilities seen in the higher intellectual functions need not be any one of them. The original "wheelchair" analog could, for example, be the structured planning needed for some nonintellectual function, such as the ballistic hand-arm movements used for hammering, clubbing, and throwing.

Throwing is a particularly interesting possibility because targets are located at many different distances and elevations, making each hunting throw a novel situation, quite unlike the more stereotyped dart throws and basketball free throws where long practice can "find the right groove." There is also a premium on being right the first time, as dinner is likely to flee if you miss.

Apes have elementary forms of the rapid arm movements that we're experts with - hammering, clubbing, and throwing - and one can imagine hunting and toolmaking scenarios that, in some settings (more in a moment), were important additions to the basic hominid gathering and scavenging strategies. The evolutionary rewards for individuals having better-than-average projectile hunting skills could thus set the stage for free secondary uses, such as planning on longer time scales, such as logical trains of thought, and perhaps even music and syntax. Some of these, once they had a chance to show their stuff, are exposed enough to natural selection to "pay" for further improvements - and thus improve throwing accuracy, pari passu (Calvin, 1983, 1993, 1996b).

Finding the Right Level


Although it seems to have played little role so far in our modern concepts of intelligence, the concept of levels of organization is a common one in the sciences. Much of guessing well involves finding the right level at which to address a problem, neither too literal nor too abstract - or, sometimes, inventing a new level on the fly.

Levels are best defined by certain functional properties (Calvin & Bickerton, 1999), not anatomy. As an example of four levels, fleece is organized into yarn, which is woven into cloth, which can be arranged into clothing. Each of these levels of organization is transiently stable, with ratchet-like mechanisms that prevent backsliding: fabrics are woven, to prevent their disorganization into so much yarn; yarn is spun, to keep it from backsliding into fleece.

A proper level is also characterized by "causal decoupling" from adjacent levels (Pagels, 1988); it's a "study unto itself." For example, you can weave without understanding how to spin yarn (or make clothing). Chemical bonds illustrate a proper level: Mendeleyev discovered the patterns of the table of elements, and thereby predicted the weights and binding properties of undiscovered elements, without knowing anything about the underlying patterns of electron shells (or the overlying patterns of stereochemistry).

Mental life can pyramid a number of levels, thereby creating structure. Some of the major tasks of early childhood involve discovering four levels of organization in the apparent chaos of the surrounding environment:

• Infants discover phonemes and create standard categories for them; six-month-old Japanese infants can still tell the difference between the English /L/ and /R/ sounds but after another six months of regular exposure to the Japanese phoneme that lies in between them in sound space, the baby will treat the occasional English sounds as mere imperfect versions of the Japanese phoneme (Kuhl et al 1992) and so be set up for later confusing the English words 'rice' and 'lice'.

• With a set of basic speech sounds, babies start discovering longer-duration patterns amid strings of phonemes, averaging nine new words every day during the preschool years.

• Between 18-36 months of age, they start to discover still-longer patterns of words called phrases and clauses, rules such as add -s for plural, add -ed for past tense.

• After syntax, they go on to discover Aristotle's rule about narratives having a beginning, middle, and end (and they then demand bedtime stories with a proper ending).

Indeed, we find it very rewarding to discover half-hidden patterns all through life: that's the basis for the popularity of crossword and jigsaw puzzles. It's why science is so much fun.

Pyramiding four levels in a mere four years is impressive. But levels can also be created on the fly, as we seek an analogy or make a novel abstraction. To spend more time at the more abstract levels in this house of cards, the prior ones have to be sufficiently shored up to prevent backsliding over your concentration span.

From Stratified Stability to Darwinism


But there's no child inside the head to stack up those higher stories of the house of cards, so what self-organizes a higher level? In the simpler physical systems, noise (as in diffusion) can provide the raw material for self-organizing structures (such as crystals). As Bronowski (1973) observed: "The stable units that compose one level or stratum are the raw material for random encounters which produce higher configurations, some of which will chance to be stable..…" If there is an organizational principle in the universe that is even more elementary than Darwin's, it is Bronowski's.

But, as Darwin first realized, competitions between stable alternatives can improve the results, providing a quality bootstrap under certain conditions. Not all of what is loosely called "Darwinian" qualifies, however, as many pruning processes do not have a result that copies and competes. As I have discussed elsewhere (Calvin 1996, 1997), there appear to be six essential features of a recursive Darwinian process:

• It involves a pattern. Classically, this is a string of DNA bases called a gene. But the pattern could be a melody or the brain activity associated with a thought.

• Copies are somehow made of this pattern, as when cells divide or you whistle an overheard tune. Indeed, the unit pattern is defined by what's semi-reliably copied, e.g., the gene's DNA sequence is semi-reliably copied while whole chromosomes or organisms usually are not.

• Patterns occasionally change. Point mutations from cosmic rays may be the best known alterations, but far more common are copying errors and (as in meiosis) shuffling the deck.

• Copying competitions occur for occupation of a limited environmental space. For example, several variant patterns called bluegrass and crabgrass compete for my back yard.

• The relative success of the variants is influenced by a multifaceted environment. For grass, it's the nutrients, water, sunshine, how often it's cut, etc. We sometimes say that the environment "selects" or that there is selective reproduction or selective survival. Darwin called this biasing by the term 'natural selection.'

• The next generation is based on which variants survive to reproductive age and successfully find mates. The high mortality among juveniles makes their environment much more important than that of adults. This means that the surviving variants place their own reproductive bets from a shifted base, rather than the original center of variants at conception (this is what Darwin called an inheritance principle). In this next generation, a spread around the currently successful is again created. Many new variants will be worse than the parental average but some may be even better "fitted" to the environment's collection of features.

From all this, one gets that surprising darwinian drift toward patterns that almost seem designed for their environment. In the cardboard version of darwinian, particular parts such as "natural selection" are often confused with the entire darwinian process, but no one "essential" by itself will suffice. Without all six essentials, the process will shortly grind to a halt.

For example, neural patterning in development (all of that culling of cells and synapses) is an example of a sparse case: just a pattern carved by a multifaceted environment. There is no replication of the pattern, no variation, no population of the pattern to compete with a variant's population, and there's nothing recursive about achieving quality because there's no inheritance principle. It's very useful but it's not a quality bootstrap.

Making Darwinism Fast Enough


Speed is of the essence in behavior, however, and one might reasonably worry about whether a neocortical version of the darwinian process can operate quickly enough to provide an answer within the windows of opportunity afforded by either hunting or social repartee. There are at least four "catalysts" which can greatly speed up evolutionary processes:

• Systematic recombination (crossing over, sex) generates many more variants than do copying errors and the far-rarer point mutations. There's also nonsystematic recombination, such as bacterial conjugation or the conflation of ideas.

• Fluctuating environments (seasons, climate changes, diseases) change the name of the game, shaping up more complex patterns capable of doing well in several environments. For such jack-of-all-trades selection to occur, the climate must change much faster than efficiency adaptations can track it (more in a minute).

• Parcellation (as when rising sea level converts the hilltops of one large island into an archipelago of small islands) typically speeds evolution. It raises the surface-to-volume ratio (or perimeter-to-area ratio) and exposes a higher percentage of the population to the marginal conditions on the margins.

• Local extinctions (as when an island population becomes too small to sustain itself) speed evolution because they create empty niches. The pioneers that rediscover the niche get a series of generations with no competition, enough resources even for the odder variants that would never grow up to reproduce under any competition. For a novel pattern, that could represent the chance to "establish itself" before the next climate change, for which it might prove better suited than the others.

There are also catalysts acting at several removes, as in Darwin's example of how the introduction of cats to an English village could improve the clover in the surrounding countryside: The (i) cats would (ii) eat the mice that (iii) attack the bumble bee nests and, thereby, (iv) allow more flowers to be cross pollinated. Although a Darwinian process will run without these catalysts, using Darwinian creativity often requires some optimization for speed.

Explaining how neocortical circuitry can implement the six essentials and the four catalysts on a milliseconds-to-minutes time scale, thereby facilitating intelligent "get set" groping, lies beyond the scope of this article (see Calvin 1996, 1998b). However, these ten facets of a rapid evolutionary process will be useful in considering how we got our big brains so quickly (2.5 million years is quick). What sped up the slow biological evolution of the rapid neural evolutionary machinery underlying the higher intellectual functions?

Hominid Evolution and the Ice Ages


The earliest known changes in hominids, seen soon after the australopithecines diverged from the other Pan cousins about five million years ago, were rearrangements of hips and knees for upright posture. Brain size (an admittedly inadequate indicator of functional capacities) didn't change very much, remaining in the great ape ballpark. But a number of interesting things all started to happen between three and two million years ago.

• The archaeologists have traced stone tools back that far: the simplest types (the split pebbles which make such good cutting edges for getting through animal hides) go back to about 2.5 million years, with much more elaborate ones developing by 1.5 million years ago. While various mammals use found objects to open shells and the like, simple toolmaking (shatter a rock and select the sharp edges) seems to have been on the rise by 2.5 million years ago.

• The onset of the ice ages has been moved back to about the same time by the paleoclimatologists. Since then, ice sheets have slowly built up. They melt off somewhat more quickly (the rise in sea level takes about 8,000 years) and remain at a minimum for another similar period. The major meltoffs occurred about every 40,000 years - until about 0.7 million years ago, when a 100,000-year cycle became more prominent. It isn't clear what this has to do with African-based hominid evolution, as the average temperatures there only drop about 5°C during the colder periods, enough to create some glaciers on the equatorial volcanos but hardly enough to create a wintertime for animals living in the Rift Valley. All of the ice sheets that formed at higher latitudes were surely unseen by our African ancestors.

• And brain size finally starts to change back about 2.4 million years ago as the australopithecine lineage split off a distinctive Homo lineage with increased cranial capacity (and it kept increasing in size; nothing of the kind is known for other animals). Many animal lineages also split between 3 and 2 million years ago: chimp-bonobo, gibbon-siamang, mastodon-elephant (accelerated speciation is best studied, however, in the antelope and pig lineages).

What do these three things have to do with one another, or with intelligence? Cause and effect? Or merely three independent trains set in motion by the major rearrangement of ocean currents and climate that followed the damming up of the "Old Panama Canal" about 3 million years ago, when North and South American finally joined up and forced the equatorial currents that equilibrated the Atlantic and Pacific Oceans into a long detour around the southern continents?

The Mark Twain Transition Time Principle


The writer Mark Twain once observed that "A round man cannot be expected to fit into a square hole right away. He must have time to modify his shape." Evolution can often track climate changes, selecting for variants with more or less body insulation. Indeed, up until a decade ago, we thought that the ice ages were "glacially slow," that slow changes in the earth's orbit caused gradual cooling, which caused more ice to gradually form, and sea level to gradually lower. No animal lived long enough to realize that climate changes were happening because the change during the lifetime of any one generation was so minuscule.

What happens if the climate changes abruptly, so that adaptations over the generations cannot track it? So that the habitat is largely disrupted (no more customary plants or prey, a different setting for reproduction, etc.), all within one generation's time on earth?

There is a general answer to this (Calvin, 1996b) and a much more specific one. The general answer is that the circumstance provides a selective pressure for versatility, one that counters the usual lean-mean-machine tendencies that reduce unneeded anatomy and behavior in the name of efficiency. Evolutionary theory suggests a tendency towards the latter if the environment remains the same for long enough. But when the habitat changes so drastically in so short a time, only reserve capacity in behavior can solve the problems. Lean mean machines don't survive the downsizings very well. The more versatile may have what it takes.

The more specific answer, however, is grass.

Abrupt Climate Change


Inferring ancient climates can be done from layers of sediments that accumulate in lake and ocean floors. From cores, one can study such proxy climate indicators as pollen types and oxygen isotope ratios, and how they change over time. But there's a problem: worms and bottom-scavenging fish stir the bottom, mixing together hundreds (if not thousands) of years of sediments. Like a moving average of a stock-market index, a smoothed record of paleoclimate can miss some dramatic fluctuations that made and lost fortunes.

When year-by-year high-resolution records became available from Greenland ice cores, where tree-ring-like records can be seen, we became aware that climate could - and frequently did - change quite rapidly. We now know that this is not merely a peculiarity of Greenland, that these were worldwide events in many cases. Unlike the high latitude ice sheets, these abrupt climate shifts affect the tropics as well.

Atop those glacially slow rhythms, abrupt coolings and warmings have often occurred. By 'abrupt' is meant decade transition times; 'often' means every few thousand years; these flips may last for centuries (or even 1500 years) before flipping back, just as abruptly. Obviously, this is not the abrupt climate change which volcanos and Antarctic ice sheet collapse can cause; it's bistable. Ocean circulation seems to have several modes, and when it switches between modes, the abruptness of the transition profoundly affects ecosystems. In the tropics, average annual temperatures fall by 3-5°C. In Europe, it's more like 5-9°C and some high latitude locations cool more than twice as much.

The magnitude of the temperature change isn't the big problem. It's how fast it occurs, Mark Twain's problem. If these coolings were to occur slowly, with temperature ramping down over 500 years, one might expect high altitude plants and animals to slowly move down the hillsides into the valleys below. Each generation of hominids could have continued to make a living in much the way their parents taught them, though their diet mix would shift over the centuries. But with major changes within a decade, other events supervene.

First comes drought (there is much less evaporation from tropical oceans, which also reduces a major greenhouse gas, water vapor). Then, similar to what we saw in the 1997-98 El Niρo, vast fires occur even in tropical forests. After that comes a succession of plants leading, over a few centuries, from grass to forests (of species better suited to the new annual temperature).

For many terrestrial species, this is a trial - and, eventually, an opportunity. Because of vanishing resources, the habitat becomes patchy. There are refugia, where a more-or-less traditional way of life can be maintained, but they don't support very many. These subpopulations inbreed because the resources are too scarce to support a journey to find another subpopulation. But they tend to mix during the downsizing itself, as stray individuals locate remaining subpopulations. Some subpopulations may form up entirely from strays.

Thus an abrupt cooling is likely to provide three (recombination, climate stress, parcellation) of the catalysts that speed up evolutionary processes. The fourth, re-expanding into empty niches, may occur during an abrupt warming (which is accompanied by increased rainfall), where pioneers discover new territories of untapped resources, and have many offspring survive, even the odder ones.

Why Us?


The foregoing is likely true for many mammalian species, not just our ancestors. The remaining great apes likely went through this cycle many times. What was special about hominids?

Eating grass indirectly. In the first few years after the great forest fires, it is a boom time for the remaining grazing animals, with all that grass and all of those succulent shoots. But the waterholes would be scarce, and so whole herds would bunch together to visit the remaining waterholes. They would lose a few peripheral individuals to the predators that lay in wait there.

What was so special about hominid predators lingering around the waterholes? For one thing, upright posture. The grazing animals have innate "search images" for four-legged predators; they keep their distance. But bipeds can get much closer (as my colleague Arnold Towe notes, if you drop down on all fours, grazing animals move away promptly). Upright posture may, if efficient enough, allow hominids to run animals to exhaustion. But clearly, at some stage, projectile predation became a part of the picture (Calvin 1993). Even something so simple as flinging a tree branch can be effective in the context of a tightly-packed herd at a waterhole: the herd immediately starts to wheel around and flee, but the branch lands somewhere in their midst and trips an animal or two. Attempts to get back up are delayed by other animals stampeding past, and injuries will often occur. In any event, the hunters will often have time to run up to a downed animal before it can escape. Chimpanzees love to fling branches and they also covet fresh meat -- but don't seem to have made the connection, perhaps because of lacking savannah waterholes frequented by herds. Following a cool-crash-and-burn downsizing, there are lots of temporary savannahs.

Growth Curves


Chattering between two climate states thus has the potential to speed up evolution, and cool, crash, and burn cycles provide some opportunities that our upright australopithecine ancestors might have exploited. What, however, makes this an important driver for hominid evolution and intelligence, as compared to a bookshelf full of varied suggestions on what might have been behind it all?

Compared, say, to the invention of the carrying bag? The basic idea of a carrying bag must have been very important for both gathering and for small-game hunting. But one cannot reinvent the carrying bag for extra credit. Some inventions can be repeated; for example, the aquatic mammals have all discovered that a small reduction in body hair buys them greater swimming efficiency. Another reduction buys them even more. No matter where along the "growth curve" they are, another increment has additional rewards. (There is, however, a limit: you can only become so naked.) Some growth curves are also steeper than others, faster at driving evolution than slower candidates (which might, like many on that bookshelf of plausible candidates, have done the job in the long run). So the steepness and extent of growth curves are important considerations in sorting out where our intelligence came from.

There are two aspects of the "eat grass indirectly" scenario which have long growth curves, and they involve things where we humans have considerably enhanced abilities over the great apes. They can also be "pumped" in one percent increments to produce many-fold improvements. Neither is that abstraction called 'intelligence' but either could be the wheelchair-like curb cut that gave higher intellectual functions their entry-level jobs.

Cooperation and Group Selection


The "good of the group gene" possibility was dismissed a few decades ago (see Sober and Wilson 1998) on the theoretical grounds that it would, like a leaking tire, backslide. Even if somehow concentrated into a subpopulation yielding a majority of cooperators, you'd still expect that tendencies to share could be swamped by all the non-reciprocating freeloaders, who would out-reproduce the sharers.

If this were the prime consideration, of course, we would also have to conclude that car tires would never work because they all, sooner or later, go flat. We just pump them back up occasionally, and the cool-crash-and-burn cycle suggests both a concentration mechanism and a pump that might allow widespread cooperation to become established for long enough to invent other solutions to the freeloader problem.

There is a certain amount of random recombination of populations during the downsizing, as noted earlier. It is not unlike randomly selecting a jury of 12 from a jury pool of many hundreds. Even if the jury pool is representative (half male, 90 percent right-handed, appropriate racial mixtures), some individual juries are dramatically different (all of one sex or race, all left-handed, and so forth). That's just one of the well-known phenomena of probability theory ("drawing small samples without replacement"). So even if innate tendencies toward cooperation were only prominent in 10 percent of the population, after parcellation many subpopulations would have none and some might have a majority.

These "island" subpopulations are not competing with one another like football teams, thanks to those resource-free gaps; they are fighting the environment in the cool-crash-and-burn case, and most subpopulations disappear with time. When conditions allow populations to re-expand, there will be, after expansion into interbreeding "continental" populations, a higher proportion of those "genes" that helped some subpopulations survive the downsizing.

This, too, is susceptible to the "Why us?" objection as it would seem to apply to many mammalian species, and the answer may lie in what is shared. Groups with a majority of cooperation genes might spend less time arguing (and thereby wasting time which could be spent in locating more food) and fighting (thereby both losing time and risking injury) during the downsizing. This is particularly attractive because of the long growth curve for cooperation, as noted earlier. Or what's shared is language, where in order to realize its benefits, you might need a sizeable proportion of those with beginner's traits.

Or, as Derek Bickerton has proposed (see Calvin & Bickerton 1999), cognitive capacities (mental categories for giver, recipient, beneficiaries, type of action, and so on) to keep rough track of freeloading tendencies might allow "Who owes what to whom" to find another use, namely saying "Who did what to whom." Solving the cheater problem in reciprocal altruism could thus be the "wheelchair" that paid for the argument-structure scheme of handing syntax, until structured language started earning its own way. This by itself would constitute a large step up in intelligence.

Precision Throwing as Curb Cut


Another ape-to-human improvement with a long growth curve is the precision throwing which is so handy for expanding hunting abilities beyond that seen in the other predators. Hunting herd animals also has a link to cooperation, as any one prey animal is usually too much for one hunter to eat; one simply has to give away most of it as the chimpanzees do ("tolerated scrounging") and hope for reciprocation when someone else gets lucky. As Frans de Waal (1996) observed:

If carnivory was indeed the catalyst for the evolution of sharing, it is hard to escape the conclusion that human morality is steeped in animal blood. When we give money to begging strangers, ship food to starving people, or vote for measures that benefit the poor, we follow impulses shaped since the time our ancestors began to cluster around meat possessors. At the center of the original circle, we find a prize hard to get but desired by many... this small, sympathetic circle grew steadily to encompass all of humanity -- if not in practice then at least in principle.... Given the circle's proposed origin, it is profoundly ironic that its expansion should culminate in a plea for vegetarianism.

And, of course, hunting was one of the only solutions to an environment where, for a few centuries, you either had to eat grass or eat an animal that ate grass.

The side-of-the-barn accuracy needed for flinging branches into waterhole herds may not have much of a growth curve by itself (it doesn't matter which one you trip), but it could have gotten hunters onto the bottom on the precision-throwing growth curve. Being able to hit smaller herds has an even higher payoff. So does throwing from farther away (herds eventually become wary), which also reduces risk to the hunter. One can use other projectiles, such as rocks. One can become accurate enough to hit lone animals. Then there are spears, and their augmentation by launching sticks, and so forth. Each improvement has an additional payoff: more days that you and your dependents can eat nice sterile, high-calorie, low-toxicity fresh meat. (Cooking has made the world much safer for vegetarians and scavengers.)

It's easy to see how natural selection could have repeatedly improved throwing, but what does it have to do with higher intellectual function? As noted, all of the ballistic movements require much detailed planning during "get set" as feedback is too slow. While many ballistic movements have some payoffs even when stereotyped, the hominid hunter cannot function like the frog throwing its tongue when a fly is heading into its "gunsight." There is no standard throw because of the "approach distance" problem; each throw is a somewhat novel problem in both elevation and range, even if using a standard projectile size and weight.

And, beyond planning and versatility demands, there is the problem of timing jitter (Calvin 1983). If you throw at a rabbit-sized target a car-length away, and release 5 milliseconds too soon, you'll overshoot the target. At two car lengths away, the launch window shrinks eight-fold. Redundant motor programs, each with independent noise, can solve this double-the-distance problem by using 64 times as many motor programs in parallel.

Even the four-fold increase in the number of neocortical neurons during hominid evolution cannot solve this jitter problem (by itself, it would only buy a 25 percent increase in approach distance). Many-fold increases in parallelism can only be done by temporarily borrowing helpers from association cortex during "get set."

I have developed this jitter-reduction idea elsewhere (Calvin 1983, 1993, 1996b). What's important for intelligence is to recall Kimura's (1993) result from aphasics, that most also suffered from hand-arm sequencing problems when confronted with novel sequencing tasks (apraxia) - and to recall Ojemann and Mateer's (1979) result from the perisylvian core of language cortex, about the overlap of nonlanguage sensory and motor sequencing tasks. If there is a common neural sequencing machinery for mouth-and-face, hand-and-arm, sensory-and-motor, language cortex is an obvious candidate (Calvin & Ojemann 1994).

And this can explain how a curb cut paid for by natural selection for precision throwing might greatly augment planning on other time scales. The structured aspect of the higher intellectual functions could easily arise from the nested embedding aspect of throwing: the shoulder motion is atop the forwards motion of the trunk; planning the elbow rotation needs to similarly work from the velocity of the upper arm; the wrist flip needs to be planned in light of the prediction of all those compounded motions controlling the lower arm's velocity, and so forth. All of the coordination must be done in advance, tweaking the parameters to find one of the dozens of possible combinations that will hit the target amidst a sea of solutions that will miss. This is nested embedding of much the same sort as shown in those binary diagrams of phrase structure, the other major way of doing syntax (see Calvin & Bickerton 1999). As you assemble words to find a coherent sentence to speak, you grapple with a problem analogous to novel hand-arm sequencing.

Whatever paid for it in natural selection terms (and I assume it was different things at different times), such a multiple-use neural sequencer would have major implications for the structured higher intellection functions, and thus for intelligence.

The Pump Run by Bistable Climate


Pumping up intelligence is thus a real possibility - even though the natural selection that paid for it may be as remote as wheelchairs are from skateboards. Higher intellectual functions may have some silent, nonintellectual partners, those novel ballistic movements.

Ignoring compound interest considerations for a moment, how many strokes on the pump, and of what size, would be sufficient to produce the many-fold increases in the mental functions that separate the Pan and Homo species? There were several dozen biphasic cooling-warming events (each lasting between 70-1500 years) in the last ice age alone, between 117,000 years and 11,000 years ago. There were dozens of ice ages and, although the high-time-resolution records do not extend to cover them yet, cores with thousand-year resolution can pick up the longer-lasting ones. These longer-lasting biphasic events have now been tracked back to 1.1 million years ago. From what we know of the oceanographic mechanisms (see Broecker 1997, 1999; Calvin 1998a), I would guess that some will be found between 2 and 3 million years ago -- but not in the period before the Isthmus of Panama forced the major detours in ocean currents.

So there are hundreds of events of pretty much the same type each time: abrupt cooling, crashing populations, and burning ecosystems. This suggests that a one percent increment each time might be sufficient. However, there were likely many more events (and so even smaller increments might suffice) because the typical abrupt cooling or warming chattered between modes like an old fluorescent light tube before finally settling down in the new mode. Typically, there would be a century where the temperature and rainfall whipped back and forth between modern and ice-age values a few times, where vast storms churned a lot of dust into the atmosphere (the isotopic signature of the dust in the Greenland cores suggests that much came from the Great Gobi Desert). Such flickering climate would have run the population contraction-expansion cycle a few times within a single "madhouse century."

This type of pumping and multiple use shows how big steps up in functionality (say, from unstructured protolanguage to structured syntax) can arise from a series of small changes in nonintellectual functions. It may be that something else from that bookshelf of plausible suggestions will prove to run the evolutionary ratchet more quickly than my combination of grass, throwing, and cooperation. But if we are to ever give an explanation for how an ape can turn into a human, we will likely have to address the profound challenges and unusual opportunities given our ancestors by the fickle climate.


During the two decades that this theory has been under construction, I have profited from numerous suggestions. The most recent extensions of the theory have benefitted much from a stay at the Rockefeller Foundation's Bellagio Center and from workshops organized by the LaJolla Origins of Humans group (sponsored by the Preuss Foundation and the Mathers Foundation) and by the Center for Human Evolution at the Foundation for the Future.

Afterword: The Future's Intelligence Test for Humans

It has been 8,200 years since an abrupt cooling of even half the magnitude discussed here (the Little Ice Age starting about 700 years ago was an order of magnitude smaller). Everything we know about the geophysical mechanisms (see Broecker 1999, Calvin 1998a) suggests that another one could easily happen - indeed, that our greenhouse-effect warming could trigger an abrupt cooling in several different ways.

Because such a cooling would occur too quickly for us to make readjustments in agricultural productivity and associated supply lines, it would be a potentially civilization-shattering affair, likely to cause a population crash far worse than those seen in the wars and plagues of history.

The best understood part of the flip-flop tendencies involves what happens to the warm Gulf Stream waters, with the flow of about a hundred Amazon Rivers, once they split off Ireland into the two major branches of the North Atlantic Current. They sink to the depths of the Greenland-Norwegian Sea and the Labrador Sea because so much evaporation takes place (warming up the cold dry winds from Canada, and eventually Europe, so that it is unlike Canada and Siberia) that the surface waters become cold and hypersaline - and therefore more dense than the underlying waters. At some sinking sites, giant whirlpools 15 km in diameter can be found, carrying surface waters down into the depths. Routinely flushing the cold waters in this manner makes room for more warm waters to flow far north.

But this sinking mechanism can fail if fresh water accumulates on the surface, diluting the dense waters. The increased rainfall that occurs with global warming causes more rain to wall into the oceans at the high latitudes. Ordinarily, rain falling into the ocean is not a problem -- but at these sites in the Labrador and Greenland-Norwegian Seas, it can be catastrophic. So can meltwater from nearby Greenland ice cap, especially when it comes out in surges. By shutting down the high-latitude parts of this "Nordic Heat Pump," these consequences of global warming can abrupt change Europe's climate. If Europe's agriculture reverted to the productivity of Canada's (at the same latitudes but lacking a preheating for winds off the Pacific Ocean), 22 out of 23 Europeans would starve.

The surprise was that it isn't just Europe that gets hit hard. Most of the habitable parts of the world have similarly cooled during past episodes. Another failure would cause a population crash that would take much of civilization with it, all within a decade.

Ways to postpone such a climatic shift are conceivable, however -- cloud-seeding to create rain shadows in critical locations are just one possibility. Although we can't do much about everyday weather or greenhouse warming, we may nonetheless be able to stabilize the climate enough to prevent an abrupt cooling.

Devising a long-term scheme for stabilizing the flushing mechanism has now become one of the major tasks of our civilization, essential to prevent a drastic downsizing whose wars over food would leave a world where everyone hated their neighbors for good reasons. Human levels of intelligence allow us both foresight and rational planning. Civilization has enormously expanded our horizons, allowing us to look far into the past and learn from it. But it remains to be seen whether humans are capable of passing this intelligence test that the climate sets for us.

End Notes

Barlow, H. B. (1987). Intelligence:  the art of good guessing.  In R. L. Gregory (ed.), Oxford Companion to the Mind., pp. 381-383.  Oxford:  Oxford University Press.

Broecker, W. S. (1997). Thermohaline Circulation, the Achilles Heel of Our Climate System: Will Man-Made CO2 Upset the Current Balance? Science 278:1582 - 1588.   Available:  

Broecker, W. S. (1999).  What If the Conveyor Were to Shut Down? Reflections on a Possible Outcome of the Great Global Experiment.  GSA Today 9(1):1-7 (January). Available:  

Bronowski, J. (1973). The Ascent of Man.  Boston: Little, Brown.  

Byrne, R. W. (1994).  The evolution of intelligence, pp. 223-265.  In P. J. B. Slater, T. R. Halliday (eds.), Behaviour and Evolution.  Cambridge:  Cambridge University Press.  

Calvin, W. H. (1983).  A stone's throw and its launch window:  timing precision and its implications for language and hominid brains.  Journal of Theoretical Biology, 104, 121-135. Available:  

Calvin, W. H. (1987).  The brain as a Darwin machine.  Nature, 330, 33-34.  Available:

Calvin, W. H. (1993).  The unitary hypothesis:  A common neural circuitry for novel manipulations, language, plan-ahead, and throwing?  In K. R. Gibson & T. Ingold (Eds.) Tools, Language, and Cognition in Human Evolution.  New York:  Cambridge University Press, pp. 230-250. Available:  

Calvin, W. H. (1996a).  The Cerebral Code.  MIT Press.  Available:

Calvin, W. H. (1996b).  How Brains Think.  New York: Basic Books.  Available:  

Calvin, W. H. (1997).  The Six Essentials? Minimal Requirements for the Darwinian Bootstrapping of Quality.  Journal of Memetics - Evolutionary Models of Information Transmission, 1, at  

Calvin, W. H. (1998a).   The great climate flip-flop.  The Atlantic Monthly 281(1):47-64.    Available:  

Calvin, W. H. (1998b).  Competing for consciousness: A Darwinian mechanism at an appropriate level of explanation.  Journal of Consciousness Studies.  5(4):389-404.    Available:  

Calvin, W. H. (1998c).  The emergence of intelligence. Scientific American Presents 9(4):44-51 (November 1998).  Available:  

Calvin, W. H., and Bickerton, D. (2000).  Lingua ex machina: Reconciling Darwin and Chomsky with the Human Brain.  MIT Press.  Available:

Calvin, W. H., & Ojemann, G. A. (1994).  Conversations with Neil's Brain:  The Neural Nature of Thought and Language.  Reading MA: Addison-Wesley.  Available:

Deacon, T. (1997).  The Symbolic Species:  The Co-Evolution of Language and the Brain.  W. W. Norton.

de Waal, F. (1996).  Good Natured: The Origins of Right and Wrong. Harvard University Press.

Gould, J. L., & Gould, C. G. (1994).  The Animal Mind.  New York: Scientific American Library.

Gould, J. L., & Gould, C. G. (1998).  Reasoning in animals.  Scientific American Presents 9(4):52-59.

Jensen, A. R. (1992).  Understanding g in terms of information processing.  Educational Psychology Review 4:271-308.

Kimura, D. (1993).  Neuromotor Mechanisms in Human Communication. Oxford University Press.

Kuhl, P. K., Williams, K. A., Lacerda, F., Stevens, K. N. & Lindblom, B. (1992).  Linguistic experience alters phonetic perception in infants by 6 months of age.  Science 255:606-608.

Ojemann, G. (1991).  Cortical organization of language.  Journal of Neuroscience 11:2281-2287.

Ojemann, G., Mateer, C. (1979).  Human language cortex:  localization of memory, syntax, and sequential motor-phoneme identification systems.  Science 205:1401-1403.

Pagels, H. (1988).  The Dreams of Reason.  Basic Books.

Piaget, J. (1929; 1952). The Origins of Intelligence in Children. New York: International Universities Press (translation of Naissance de l'intelligence chez l'enfant), last chapter.  

Sacks, O. (1989).  Seeing Voices.  Berkeley: University of California Press.

Simon, H.A. (1983).  Reason in Human Affairs. Palo Alto CA:  Stanford University Press.

Sober, E., Wilson, D. S. (1998).  Unto Others: The Evolution and Psychology of Unselfish Behavior.  Harvard University Press. || Home Page || Calvin publication list || The Bookshelf || February 2002