|posted 24 August 2003|
William H. Calvin, "Our Precious Prototype: Logic hasn’t been debugged by evolution." Highlands Forum (DARPA) talk (27 July 2003).See also http://WilliamCalvin.com/2003/DARPA.htm
This lecture is available as webbed powerpoint (don't even try unless you have a broadband connection).
After it loads, click on the "Slide Show" link at lower right.
Remember to hit the mouse or space bar to advance slides.
There are several different levels of organization in the brain,
including cell-level computation and circuit-level emergent
properties. You can have half a million categories and a large
number of relations between categories, e.g., transforming language
into sentences such as "this is bigger". There are also relations
between relations, e.g., "bigger is better". We have the ability to
plan and create in advance as we analyze a situation; for example,
we can create coherent sentences that we have never spoken before.
My focus here will be on the evolution of the higher orders on this
Examples of multifunctionality abound in our world. For example curb cuts that were designed for disabled people (their paid use) are now used by skateboarders, people with luggage, bicycles, shopping carts (free uses). As the original models evolve (e.g., wheelchairs begin to climb stairs), the original paid uses will become obsolete. Curb cuts at airports created suitcase queues; this secondary use eventually paid for making curb cuts wider. Thus, we aren't always dealing with what appears on the surface. We may think that the "current best use" is what evolution is all about – as when we say that big brains are all about intelligence -- but it is often about discovering the free uses.
Bonobos and chimpanzees shared a common ancestor about 2-3 million years ago. Their vocalizations in the lab are not repeated in nature. In the lab, they can repeat and learn hundreds of words, create two-word sentences, and understand longer ones. However, so far, they do not learn syntactical structure.
If we plot brain size over 3 million years, we note that brain size begins to enlarge at a more rapid rate during the ice ages. Today, the average human brain is 3 pints; the average chimp brain is one pint. 400,000 years ago, tool-making became much more complex as prehumans began to prepare tools from prepared cores. 250,000 years ago, blade technology began to emerge. These tools are well-established by 120,000 years ago. Anatomically modern Homo sapiens appeared 150-200,000 years ago, but their anatomically modern brain did not become behaviorally modern until 50,000 years ago. About this time, humans began making patterns in materials (e.g. inscribed ochre 77,000 years ago), decorative beads (52,000 years ago), and so on. Around the same time, there was an explosion of Homo sapiens out of Africa and into Central Asia. 40,000 years ago, the migration continued into East Asia, South Asia, and Australia. The migration also went to Europe and displaced the Neanderthals. This time marked the appearance of very fine tools, portable art (40,000 years ago) and cave art/wall art 35,000 years ago – which indicates the emergence of creative intelligence. Thus the "behaviorally modern" emerges at the end of evolution, i.e. in very recent times.
It is not clear whether there was one or many behavioral adaptations underlying this creative explosion. Evolutionary psychology tends to focus on separate evolutionary paths for each little behavioral trait. And some could hold the key to the nothers; for example, apes do not mimic or teach. The emergence of a whole suite of "modern" behaviors in the last 1% of the time since the common ancestor with chimps suggests that some common neural machinery with many uses was what happened. From a common ancestry, humans developed new uses for old circuitry, i.e., new uses for sharing and planning such as logic, contingency planning, and so on. For example, the act of throwing is a novel function that requires coherent planning. There is a payoff for throwing twice as far, or twice as fast. The brain areas are premotor and prefrontal, and they show overlap with other functionality such as language.
Like the curb cuts, planning ballistic movements like throwing and hammering may have "paid" for neural circuitry that has "free" secondary uses, such as planning a long sentence to speak, or planning a week-long agenda. Our behaviorally modern explosion included consciousness, planning in depth, symbolism, creativity, and language with structure. Higher intellectual functions all include structured syntax, planning, logic, games, music, coherence-finding (i.e. looking for patterns) – these are all examples of complex thought.
There is sometimes a danger when we find new uses for old things, because they are untested for these new uses. Biological evolution does not always perfect things, it just leads to new "products". Logic is a way of avoiding uncertainty, but we can also learn to live with uncertainty. Most people deal with it through reductionism – the problem is that in so doing, they often exclude other systems. In other words, strong belief systems are problematic because they are blinkered systems of logic that display a half-baked, overconfident rationality.
In sum, the task at hand is to find the right level at which to address the problem!
Questions and Comments from Participants
Question: According to your presentation, planning movement tracks with language in the brain. Has this concept of functionality been tested, e.g. through magnetic resonance?
Bill Calvin: No, this has not been tested, because of the way in which brain surgery occurs. A good case study would be the throwing example. Patients are operated on their left brain while lying on their right side. Therefore, they can't throw with their right hand… perhaps we could ask the patients to imagine throwing? Areas of the brain overlap in producing facial and arm sequencing. Aphasic patients have problems with hand/arm sequencing tasks.
Question: If we extend Bill's graph and look at the evolution in size and functionality of the brain, what are the next cuts we can expect in that curve? What is our brain designed for? What are the secondary and tertiary uses?
Calvin: Size is not correlated with evolutionary landmarks such as tool-making and other innovations. Human brains grow over time but their growth does not parallel a growth in intelligence and creativity, which mostly emerge in the last 50,000 years. Beyond the size issue, the question is how well we handle the upper levels of organization and how much we can handle at once. We can be educated to hold more things in our minds simultaneously. Education also applies to the speed of decisionmaking, and also to fallacies. As we begin to understand the levels of organization in the brain from a neurological point of view, we can educate ourselves better.
Q: The notion of education for better decisionmaking has huge implications for the DoD. Is it harder to design a cell or a multicellular organism in the learning process? Was there a long run-up to the behaviorally modern like there was to the Cambrian explosion?
Q: Learning happened long before we learned to speak.
Calvin: Yes. Two-year-olds acquire 6-9 new words every day, but they discover structure in sentences much later, which leads to plurals, past tenses, etc. This happens long before they learn to tie their shoelaces.
Q: There is syntax in genomic command structures – is that language?
Calvin: It is certainly command and control. ...
Question: If we apply engineering tools, can we begin to catalogue species? Can we play with evolutionary rates, i.e., produce evolution in a test tube? For example, we can derive new phenotypes from stem cells through genetic engineering. If the science of neurology is overlaid with evolution, can we change expectations about rate and process?
Calvin: Hox genes control segmentation; lots of modularity can emerge.
Question: You mentioned that catalysts and pressures can lead to brain upgrades. What are contemporary catalysts and pressures that can take our brains to a higher level? This is interesting because physiological evolutions occur, for example high-speed games lead to faster thumb-twitch speeds, etc.
Calvin: The classic example is kids learning a language. Patterns are discerned in a local language and this leads to a rewiring of the brain. There isn't a whole lot of data available to answer your question authoritatively, unfortunately. But it is clear that the brain is the most plastic in a child's preschool years. After puberty, it is hard to speak a language without a foreign accent. The games we give to kids can change a lot of things, but not necessarily for the better.
Comment: Evolution is mostly variation followed by selection. It follows genetic algorithms where variation is engineered to select for a specific response. The current instantiation of something in our body doesn't tell us anything about the history of how that instantiation came about. We observe changes, but we do not understand the evolutionary processes, which are all about how nerve cells communicate, synaptic changes, which genes are turned on in the nucleus, and so on.
Calvin: There is an example in evolutionary systems: breathing receptors exist for oxygen, but most of the modulation is done through carbon dioxide receptors.
Question: Simple organisms are hardwired and have a fixed number of neurons. Humans and mammals have "grow and prune" neural systems, such as are exhibited when learning languages very young. When did this system arise?
Calvin: This model of overproduction and pruning back is widespread in biology, especially in the brain. What we don't know, as Henri discussed, is whether brain cells have 80% permanent synapses and 20% turnover – we don't have that clear a picture yet. Cells can physiologically behave as face cells one week and thumb cells the following week – there are many examples of dual wiring. There is quite a bit of plasticity in the brain. Cutting a peripheral nerve can activate a whole slew of dormant synapses.
Calvin: Networks can be massively shut down and can paralyze and organization; therefore, it is important to study the emergent properties of networks. The rapid evolution of technology can have negative economic effects if it is not treated carefully.
copyright ©2003 by William H. Calvin