W. H. Calvin's "The antecedents of consciousness"

Calvin, W. H. (1991). The antecedents of consciousness: Evolving the "intelligent" ability to simulate situations and contemplate the consequences of novel courses of action. In: Jean Heidmann and Michael J. Klein (eds) Bioastronomy: The Exploration Broadens. (Springer-Verlag's Lecture Notes in Physics series), pp. 311-319.
copyright 1991 by William H. Calvin


PTN Webbed Reprint Collection
William H. Calvin
University of Washington
Box 351800
Seattle WA 98195-1800 USA
Email || Home Page || publication list

William H. Calvin

"The antecedents of consciousness: Evolving the `intelligent' ability to simulate situations and contemplate the consequences of novel courses of action."

as it appeared in:
Bioastronomy: The Exploration Broadens, edited by Jean Heidmann and Michael J. Klein (Springer-Verlag's Lecture Notes in Physics series), pp. 311-319 (1991).

copyright ©1991 by William H. Calvin

The antecedents of consciousness: Evolving the "intelligent" ability to simulate situations and contemplate the consequences of novel courses of action.

William H. Calvin

University of Washington,
Seattle, Washington 98195-1800 USA


About 1915, Wolfgang Koehler did pioneering studies of problem-solving behaviors in chimpanzees, now found in most introductory psychology texts as the example of insightful behaviors, perhaps near the threshold of human-style intelligence from conscious contemplation. You probably remember the banana hung from the ceiling of a room occupied by a box and a frustrated chimpanzee. As a later observer described it:

The matter gave him no peace, and he returned to it again. Then, suddenly -- and there is no other way to describe it -- his previously gloomy face "lit up." His eyes now moved from the banana to the empty space beneath it on the ground, from this to the box, and back to the space, and from there to the banana. The next moment he gave a cry of joy, and somersaulted over the box in sheer high spirits. Completely assured of his success, he pushed the box below the banana. No man watching him could doubt the existence of a genuine "Aha" experience in anthropoid apes.
In 1927, the philosopher Bertrand Russell wrote his own textbook of psychology. And he made a wry comment:
Animals studied by Americans rush about frantically, with an incredible display of hustle and pep, and at last achieve the desired result by chance. Animals observed by Germans sit still and think, and at last evolve the solution out of their inner consciousness.
I'm not telling that story just because of the analogies to various approaches to SETI -- the people that emphasize random search of the sky vs. those who out of their inner consciousness evolve magic frequencies. Rather, I need to talk about random trial and error as a basis for intelligent behaviors: You don't always have to do your random trial and error in real time, running around and trying this and that, wasting energy and exposed to the hazards of your environment. You can sometimes just do all the trial and error inside your head: that chimp of Koehler's presumably simulated the necessary movements, pieced together the scenario of moving the box and standing atop it, and then acted. The only difference, I suggest, between what Russell characterized as the American and German approaches is whether the random trial-and-error is done on-line or off-line.

Offline trial and error is probably the basic way of solving novel problems and, where fancier procedures such as algorithms are used, trial-and-error was probably the main way of evolving them. We need to know how humans evolved such simulation abilities. And it would be nice to know some alternative paths that an extraterrestrial intelligence might have followed. I'm going to give an example of the neural machinery that intelligence might require, a brief description of how it might have evolved, then discuss hominid evolution more generally and speculate about what might happen in the next century as we better understand the machinery underlying our own higher intellectual functions.

Simulation to Shape Up Novelties

Offline simulation is a powerful technique for exploring complex scenarios of possible movements without risking life and limb. But it requires a lot of neural machinery: To do it rapidly, you need dozens to hundreds of planning buffers, a population of sequences, some of which will rate more highly than others. One can then randomly generate some variants of the top-rated candidate and conduct evaluations of this new generation of scenarios. In this manner, and often in a matter of seconds, we can shape up a plan (for a sentence to speak, or movements to make) that stands a better chance of fitting the situation.

This mental procedure seems quite analogous to Darwinian evolution, where the bodies produced by strings of DNA are evaluated by an environment of prey-predators-pathogens and only the survivors reproduce (with variations caused by a little gene-shuffling as sperm and ova are made). In a matter of millennia, the DNA strings are shaped up to roughly "match" the environment. In the immune response, the shaping up takes only a matter of days; the more successful of the initial array of antibodies reproduce with variations in their amino-acid sequence, some of the new generation are even better matched to the antigen-bearing cells, and so forth.

The mental strings are shaped up against a virtual environment of episodic memories, rather than the often noxious real-time environment of trial-and-error. They may be muscle activation commands, such as those needed to throw a rock or hammer a nut -- or perhaps to command something that even apes don't do, such as kicking a football or dancing at the discotheque. Indeed, we humans have greatly augmented serial-order propensities: we string together words into sentences, we string together musical notes into melodies, we try to create scenarios that explain the past and forecast the future.

Thanks to our mental Darwinism, each of us now has under our control a miniature world, evolving away, making constructs that are unique to our own head. Such Darwin Machines (totally unlike our usual deterministic von Neumann machines) may be a prerequisite to general-purpose intelligence, particularly the kind that might produce a technological civilization with a propensity to communicate.

Evolving a Darwin Machine

What evolutionary situations might serve to augment such serial-order abilities, up to the point that many planning tracks are available and a Darwin Machine might emerge? Certainly the ballistic movements have the right hallmarks: they are the fraction-of-a-second actions such as throwing, clubbing, hammering, kicking (also, as I was reminded the other day by an orangutan, accurate spitting is also a ballistic task -- she was just trying to get my attention, wanting me to come back and groom her some more, rather than studying the pygmy chimp next door).

Because feedback is so slow (0.1 sec is about the fastest), you need a planning buffer for these faster-than-feedback movements. I should caution that this serial-order buffer is a lot fancier than the text buffer in your computer printer: it is more like a roll for a player piano, as there are at least 88 muscles to which one must send activation sequences to hammer or throw smoothly. Furthermore, you may need many such buffers, perhaps not for hammering but certainly for throwing. Tight launch windows demand precision timing that, given jittery neurons, can only be achieved by hitching many timers in tandem. To hit a target twice as far away, you need to reduce your timing jitter 8-fold. You usually get it only by assigning 64 times as many sequencing buffers to the task (imagine a roomful of player pianos, ganged into tandem, trying to sing as a chorus). Note that such massively serial machinery (dozens to hundreds of player piano equivalents) might be capable of secondary use, if not too hard-wired, e.g., a throwing sequencer might be used in the off-hours as a Darwin Machine for stringing words into sentences, spinning scenarios for future possi- bilities.

Many animals plan in a limited sense but they are standard behavioral sequences for their species, e.g., nest building when pregnant or nut gathering when the days shorten. For more novel actions, goal-and-feedback works pretty well; a detailed plan of action is seldom needed, because there is time to make corrections along the way. And most animals can modify standard behaviors by associative learning, in the manner of Pavlov's dogs. No player-piano-like serial buffer is really needed -- and certainly not hundreds, each with the means to judge a candidate sequence against serial-order-coded episodic memories. I simply do not know of any animal behavior that requires even a fraction of the sort of neural machinery needed by a Darwin Machine -- not even the throwing and hammering of chimpanzees seems precise enough.

My argument is that our kind of intelligence arises from a secondary use of neural machinery evolved for its usefulness at the mundane task of throwing accurately and hammering skillfully. And that, as we have gotten better and better at them, there has been something of a jump in language and intelligence as the Darwin Machine secondary uses have emerged.

This is consistent with, though not particularly predicted by, the time course of prehuman evolution. The repeated climatic changes of the Ice Age might have been particularly important in avoiding the stabilities that slow evolution. Upright posture evolved long before the hominid brain started to enlarge beyond ape standards. Brain enlargement started between 2.5 and 2.0 million years ago (the previous major increment occurred more than 30 million years earlier, in the evolution of apes from monkeys). At about 2.5 million years ago is also when archaeological evidence of prolific tool- making is first seen (my personal guess is that this is also when projectile predation got going too, but there is little agreement among anthropologists on the subject). And 2.5 million years ago is also when the Ice Ages started. One suspects that climate change sped up the evolution of those characteristically human skills and big brain (for climate change at 2.5 million years, see Shackleton et al, 1984, Vrba 1985; for arguments on species dates, see Grine 1989).

While the rapid melting of the ice sheets about every 100,000 years is the most obvious feature of the two-dozen Ice Ages, there are also abrupt changes in climate every so often, certainly within this last ice age and probably associated with switches in ocean currents. The North Atlantic Current, which is what keeps Europe warmer and wetter than Canada at the same latitudes, shuts down on occasion. The last such episode was the Younger Dryas event of 11,500 years ago when Europe suddenly cooled and the forests died within decades. The North Atlantic Current apparently resumed about 10,720 years ago -- and over the course of just a few years, as you can see from the increase in rainfall, the decrease in severe storms (that's what dust measures), and warming. The cooling at 11,500 years ago was almost as abrupt.

Evolutionary change can track climate change, so long as climate change happens slowly. But an abrupt change such as the Younger Dryas happens within the time of a single generation, as do the even more rapid changes we associate with droughts or El Niño. Rapid climate changes select, in effect, for versatility: having the body and brain to exist in either climate, able to find food and shelter, able to successfully raise offspring, able to endure the pathogens which come with each climate, etc.

Climate change, via expanding and contracting the range of a species, also helps to make minorities into majorities, time and again. For example, living in the temperate zone year-around requires getting through the winter; in arctic aboriginal groups, we see such adaptations as larger bodies (reduced surface/volume ratio) and an emphasis on hunting grazing animals (grass remains edible year-around) or marine mammals (whose food supply is also not reduced in the manner faced by human gatherers in the dormant season). But arctic and temperate zone populations are often only a small fraction (say, 15 percent) of the world-wide population, and such genetic traits spread very slowly into regions where they are not under natural selection (hunting skills, while handy in the tropics, are hardly essential). Yet an ice sheet that melts back over the course of a few hundred generations serves to increase the temperate zone population; when the ice sheet again advances, the temperate-adapted subspecies is pushed into the subtropics, mixing genes with the main population. If the 15 percent minority had increased to 32 percent during the melt back, they might have remained at 32 percent of the smaller total population. Let this expand-and-compress cycle repeat a few times and the main population may start looking and acting remarkably like the temperate zone subtype.

Prospects and Cautions

Consider the time scale of our own evolution: In only about the last 0.2 percent of the time since the hominid brain started enlarging have we had large-scale social organization; in only a few of the civilizations have technological innovations had any widespread consequences; in only the last dozen generations have we had much in the way of serious science. The radio-based technologies have influenced our thinking about SETI for less than four generations. Computers have been extending our simulation abilities for only several generations.

We have "grown up" very fast, with widening gaps between the abilities of the scientifically illiterate and the sophisticated. Yet I think that, in another century, we will remark on the leap we have made in understanding our own mental processes, and how this greatly expanded our abilities. Consider what happened to transportation in the wake of understanding Newton's physics of moving bodies in several centuries, we went from ox-carts to moon rockets. Consider what happened to communication in the wake of our nineteenth century understanding of electricity and magnetism we went from hand-carried letters to our ability to load up our memory telephones with a 15-digit sequence that dials a distant phone, connects you via communications satellites. At the mere touch of a single button right now, you can rudely awaken someone on the opposite side of the world. Consider what happened to medicine, once the circulation of the blood and the role of microscopic organisms were appreciated in several centuries, we went from purging and leeches to physiologically-based neurosurgery for epilepsy and Parkinson's disease, to persuading lowly bacteria to produce human growth hormone and insulin, merely from snippets of DNA. And computation: from pencil-and-paper to networked supercomputers in about a century. All of which would have seemed like magic to our ancestors, no matter how well-educated they were.

But most of the people in this world (including the 94 percent in the U.S. who lack basic scientific literacy) have no analogies with which to think about how a computer actually functions. Or an aspirin tablet. To many, our high technology is magic, just as a flashlight might appear to a remote tribesman encountering explorers. Because our thinking operates largely via analogies and metaphors, such people are often restricted to the analogy of the give-and-take of social relationships: tit for tat, gifts, deviousness, pleading, placating, flattery. And so they apply such analogies to everything incomprehensible. In the supernaturalistic approach to nature, everything -- say, the weather -- is said to have a spirit; personifying it explains its characteristics and provides humans with a way of influencing it via offerings and pleadings. Personification is still the most widespread "explanation" for mind (we imagine a "little person within" the brain, who acts as voyeur and puppeteer).

One of the key parameters to a Darwin Machine is the "good-enough" judgment. Premature closure -- making up your mind too soon by calling off the search for better scenarios -- is a major hazard, leading to erroneous associations and plans of action that won't work. Premature closure thrives on lack of appropriate mechanistic metaphors; scientists have a wider repertoire of mechanistic metaphors, but it might well be insufficient for interpreting the intellligence behind a message. Yet, once we establish a workable explanation for our thinking and language machinery, we will again see a great augmentation in our capabilities, just from more appropriate metaphors. Imagine what knowledge of the mental machinery will do to augment our sense of self, as we feel in better command of our destinies, better able to chose good courses of action -- able to move beyond the present limits in our abilities to appreciate complicated things, whether they be detailed logical arguments or the complexities of nonlinear systems exhibiting chaos and catastrophe. Rationality will take on an entirely new meaning, once we understand the neural machinery that we use to reason with. That's what another century might bring.

Understanding the mental machinery could make us less susceptible to being manipulated by the powerful. Or perhaps, were the knowledge of the pitfalls not widely known because of poor education, make us more easily manipulated: It would be sad if what our understanding of the mental machinery did was simply to make advertising techniques even more persuasive and pervasive, transferring even more power in our society to those who can afford to buy the best in media exposure. If you think that's a problem, just imagine an alien society that is even a century ahead of us in exploiting knowledge about Darwin Machine mental abilities and pitfalls. Trying to understand them might be difficult without advanced metaphors.

As enthusiastic as I am about the intellectual and technological benefits of the search itself, I remain concerned about how our scientifically illiterate society, with its limited metaphors, might react to exotic communications, should we succeed. Even though personification is understandably a staple of science fiction, social analogies are likely to be totally inappropriate when dealing with alien intelligences. And the more appropriate metaphors with which to imagine an exotic intelligence -- say, from sociobiology, the population thinking of evolutionary biology, neural circuitry, and Darwin Machines -- are still pretty much in their infancy.


References

Calvin, W. H., The River That Flows Uphill: A Journey from the Big Bang to the Big Brain. (Macmillan, 1986).

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

Calvin, W. H., Fast tracks to intelligence. In Bioastronomy -- The Next Steps, edited by G. Marx (Kluwer, 1988). pp. 237-245.

Calvin, W. H., The Cerebral Symphony: Seashore Reflections on the Structure of Consciousness (Bantam, 1989).

Calvin, W. H., The Ascent of Mind: Ice Age Climates and the Evolution of Intelligence (Bantam, 1990).

Dansgaard, W., J. W. C. White, and S. J. Johnsen, The abrupt termination of the Younger Dryas climate event. Nature 339:532-534 (1989).

Grine, F. E. (editor), Evolutionary History of the "Robust" Australopithecines. (Aldine de Gruyter, 1989).

Shackleton, N. J., et al, Oxygen isotope calibration of the onset of ice-rafting and history of glaciation in the North Atlantic region. Nature 307:620-623 (1984).

Vrba, E. S., Ecological and adaptive changes associated with early hominid evolution. In: Ancestors: The Hard Evidence, edited by Eric Delson, pp. 63-71 (Liss, 1985).