Every now and then you come across a scientific hypothesis that is so elegant and powerful in its ability to explain that it just feels right. Yet that doesn't automatically make it right. Even when an elegant hypothesis gets support from experiments, it's not time to declare victory. This is especially true in biology, where causes and effects are all gloriously tangled up with one another. It can take a long time to undo the tangle, and hacking away at it, Gordian-style, won't help get to the answer any faster.

I was reminded of this while reading Andrew Brown's review of A Reason For Everything by Marek Kohn in the Guardian. The book sounds fascinating. Kohn recounts how a small group of English biologists shaped the course of modern evolutionary biology--in particular, by pondering how adaptation through natural selection could account for just about everything in nature. One of the foremost of these thinkers was William Hamilton, who died a few years ago. Brown writes that "even the colours of the leaves on autumn trees around the grave of Bill Hamilton have been given a meaning by evolution - they are so vivid in order to warn parasites that the tree is healthy enough to repel them."

I wrote about Hamilton's leaf-signal hypothesis here. It is one of those beautifully elegant hypotheses, and some studies have even supported Hamilton's idea that the brilliant colors of autumn evolved as a way for trees to tell insects to buzz off. But readers should not have finished reading my post by thinking, "Well, that sews that question up."

Here's why. H. Martin Schaefer and David M. Wilkinson have written a review of the Hamilton hypothesis which has just gone into press in Trends in Ecology and Evolution. They offer a lot of evidence suggesting that Hamilton may have been wrong--or at least may not have captured the whole picture. They show how a completely different process may be responsible for fall colors. Trees may produce them as they prepare for winter.

When leaves die, their nitrogen, phosphorus, and other nutrients get shipped back into their tree. It's a crucial, carefully orchestrated stage in a tree's life; it will survive on these reserves through the winter. In order to pump the nutrients back into the branches, the leaves need a lot of energy, which they have to generate with photosynthesis. That's where the pigments may come in. Pigments act as a sunscreen for leaves, shielding them from harmful UV rays that can shut down their photosynthetic machinery . What's more, as the leaves ship their nutrients back to the tree, they may produce harmful free radicals as a byproduct. It just so happens that pigments are veritable magnets for free radicals.

If the authors are right, then the evidence that seems to support Hamilton's hypothesis might not actually support it at all. For example, researchers have found that birch trees that display brighter leaves grow more vigorously the following year. You could argue that these trees did so well because they could create such strong warning signals, which warded off insects. But perhaps those bright leaves are just a sign that these trees were doing a particularly good job of protecting their leaves as they stored nutrients for the winter--nutrients that made them more vigorous the follwing spring.

Fortunately, evolutionary biologists can do more than just come up with beautiful hypotheses. They can test them. Schaefer and Wilkinson lay out a list of experiments that could discriminate between the leaf-signal hypothesis and the winter-storage hypothesis. It's even possible that evolution has produced fall foliage in order to both ward off insects and ship nutrients out of the leaf. As beautiful as any one hypothesis may be, it's the interplay of different ideas and the experiments that put them to the test that's most beautiful of all. It would not bother Hamilton one bit, I suspect, if it turned out that the leaves that fell on his grave had taken on their autumn colors for an entirely different purpose.

I am sure that in 50 years, we are going to know a lot more about how the mind works. The fusion of psychology and genetics will tell us about how our personality is influenced by our genes, and they'll also show exactly how the environment plays a hand as well. The preliminary evidence is just too impressive to seriously doubt it. Likewise, I am sure that we will have a deeper understanding how our minds have evolved, pinpointing the changes in DNA over the past six million years have given us brains that work very differently than apes. Again, the first results can't help but inspire a lot of hope.

Given where I stand on all this, I would have thought that I'd enjoy Dean Hamer's new book, The God Gene: How Faith is Hard-Wired In Our DNA. The time is ripe, judging from the string of books that have been published in the past few years on the link between religion and biology. I thought that Hamer, a geneticist, might be able to throw some interesting information into the mix, thanks to his expertise in behavioral genetics. The book turned out to be elegant and provocative, and, as I write in my review in the new issue of Scientific American, disappointingly thin on the evidence. From a single study that Hamer hasn't even published yet, he weaves an incredibly elaborate scenario in which faith is an adaptive trait. I wouldn't be surprised if it is the product in some way of natural selection, but now is hardly the time to be writing a book claiming to have figured out its origins--not to mention making appearances on talk shows and the like. Too many links between behavior and genes have already crashed and burned (including some Hamer himself has made).

Update, 9/27: Scientific American has posted the review on their site

The soft spot on a baby's head may be able to tell us when our ancestors first began to speak.

We have tremendously huge brains--six times bigger than the typical brain of a mammal our size. Obviously, that big size brings some fabulous benefits--consciousness, reasoning, and so on. But it has forced a drastic reorganization of the way we grow up. Most primates are born with a brain fairly close to its adult size. A macaque brain, for example, is 70% of adult size at birth. Apes, on the other hand, have bigger brains, and more of their brain growth takes place after birth. A chimpanzee is born with a brain 40% of its adult size, and by the end of its first year it has reached 80% of adult size. Humans have taken this trend to an almost absurd extreme. We are born with brains that are only 25% the size of an adult brain. By the end of our first year, our brains have reached only 50%. Even at age 10, our brains are not done growing, having reached 95% of adult size. For over a decade, in other words, we have newborn brains.

It's likely that this growth pattern evolved as a solution to a paradox of pregnancy. Brains demand huge amounts of energy. If mothers were to give birth to babies with adult-sized brains, they would have to supply their unborn children with a lot more calories in utero. Moreover, childbirth is already a tight fit that can put a mother's life in jeopardy. Expand the baby's head more, and you raise the risks even higher.

Extending the growth of the brain obviously gave us big brains, but it may have endowed us with another gift. All that growth now happened not in the dark confines of the womb, but over the course of years of childhood. Instead of floating in an aminotic sac, children run around, fall off chairs, bang on pots, and see how loud they can scream. (At least mine do.) In other words, they are experiencing what it's like to control their body in the outside world. And because their brains are still developing, they can easily make new connections to learn from these experiences. Some researchers even argue that only after the brains of our ancestors became plastic was it possible for them to begin to use language. After all, language is one of the most important things that children learn, and they do a far better job of learning it than adults do. If scientists could somehow find a marker in hominid fossils that shows how their brains grew, it might be possible to put a date on the origin of language.

That's where the soft spot comes in.

The oldest hominids that look anything like humans first emerged in Africa about 2 million years ago. They were about as tall as us, with long legs and arms, narrow rib cages, flat faces, and small teeth. The earliest of these human-like hominids are known as Homo ergaster, but they rapidly gave rise to a long-lived species called Homo erectus. H. erectus probably originated in Africa, but then burst out of the home continent and spread across Asia to Indonesia and China. The Homo erectus people who stayed behind in Africa are probably our own ancestors. The Asian H. erectus thrived until less than 100,000 years ago. They could make simple stone axes and choppers, and had brains about two-thirds the size of ours.

Paleoanthropologists have found only a single braincase of a baby Homo erectus. It was discovered in Indonesia in 1936, and has since been dated to 1.8 million years old--close to the origin of the species. While scientists have had a long time to study it, they haven't made a lot of progress. One problem is that the fossil lacks jaws or teeth, which can offer clues to the age of a hominid skull. The other problem is that the interior of the braincase was filled with rock, making it hard to chart its anatomy.

In the new issue of Nature, a team of researchers rectified this problem with the help of a CT scanner. They were able to calculate the volume of the child's brain, and then they were able to map the bones of the skull more accurately. As babies grow, the soft spot on their skull closes up and other bones are also rearranged in a predictable sequence. Chimpanzees, our closest living relatives, also close up their skulls in the same pattern, with some small differences in timing. The H. erectus baby, its skull shows, was somewhere between six and eighteen months old. Despite its tender age, the Homo erectus baby had a big brain--84% the size of adult Homo erectus brains as measured in fossil skulls.

A single battered braincase still leaves plenty of room for uncertainty, but it's still a pretty astonishing result. At a year old, this Homo erectus baby was almost finished growing its brain. It spent very little time developing its brain outside the womb, suggesting that it didn't have enough opportunity to develop the sophisticated sort of thinking that modern human children do. If that's true, then it's unlikely it could ever learn to speak. If these researchers are right, then future CT scans of younger hominid skulls should be able to track the rise of our long childhood.

Evolution works on different scales. In a single day, HIV's genetic code changes as it adapts to our ever-adapting immune system. Over the course of decades, the virus can make a successful leap from one species to another (from chimpanzees to humans, for example). Over a few thousand years, humans have adapted to agriculture--an adult tolerance to the lactose in milk, for example. Over a couple million years, the brains of our hominid ancestors have nearly doubled. Sometimes scientists distinguish between these scales by calling small-scale change microevolution and large-scale change macroevolution. Creationists have seized on these terms and used them to build one of their central canards: that they accept microevolution but can then reject macroevolution. That's a bit like accepting microeconomics--how households and firms make decisions and interact in markets--but then denying macroeconomics--how entire societies produce goods, how inflation rises and falls, and so on. Evolutionary biologists debate fiercely about how macroevolutionary change emerges from microevolution. But they continue to find abundant evidence that the two are a package deal.

I was reminded of the interwoven scales of evolution last week when, just before leaving on vacation, I read a wonderful new paper about how the beaks of baby birds develop. As I drove off sans laptop, I was sure that it would be heavily blogged and reported while I was away. But when I returned I found almost complete silence. So I thought I would do my small part to keep this research from disappearing into the data smog.

After all, these baby birds are not just any birds. They belong to a group of some 13 species collectively known as Darwin's finches. Charles Darwin first encountered the birds in 1835 when he visited the Galapagos Islands. He thought at first that they were belonged to various groups of birds, such as wrens and blackbirds. After all, their beaks were dramaticall different from one another--some blunt, some narrow, some curved. Not surprisingly, the birds use these different beaks to get different kinds of food--cracking nuts, drinking nectar, and so on. Darwin was shocked to learn later that all of the birds were finches. He struggled to understand why such an unparalleled diversity of finches existed only on a remote archipelago. That struggle helped lead him to his theory of evolution by natural selection.

As Jonathan Weiner recounts in his excellent The Beak of the Finch, later generations of biologists came back to the Galapagos to study the birds. Living in near isolation, they are a natural experiment in evolution. Today the leading experts on the finches are Peter and Rosemary Grant of Princeton University. They and their colleagues have shown that the birds originate from a few settlers who arrived on the islands two to three million years ago. These founders gave rise to different lineages, each of which adapted to the islands with a special beak shape of its own. This evolutionary change is remarkably fast compared to most other animals, and it continues today. As droughts and heavy rains hit the islands every few years, natural selection favors different beak sizes. Meanwhile, populations of the finches become separated from one another as they develop unique mating songs. Sometimes this divergence produces a new species. In other cases, closely related species may interbreed and fuse back together.

The Grants wondered what sort of mutations were fueling this extraordinary evolution of beaks on the Galapagos. They joined forces with developmental biologists at Harvard to study the genes that build the finch body within the egg--in particular, genes known as growth factors that stimulate cells to divide and differentiate. They found that a gene called bone morphogenetic protein 4 (BMP-4) played a key role. Big-beaked birds such as the ground finch made a lot of BMP-4 early on in development in the cells of their jaws. The slender-beaked cactus finch produces less BMP-4, and does so later. Each species they studied had its own unique pattern of BMP-4 activity, while the other growth factors behaved pretty much the same.

BMP-4 has a number at the end because it belongs to a family of genes. Originally, there was one BMP-like gene, and at some point it was accidentally duplicated. Those copies were duplicated again and again. The copies evolved differences in their sequences, and some eventually mutated into gibberish. It turns out that the first gene of this family evolved a long time ago. A huge range of animals have BMP-like genes, ranging from vertebrates to sea urchins to insects. The genes are so similar that you can destroy the insect version of BMP-4 in a fruit fly, replace it with a frog's BMP-4 gene, and the frog gene will cooperate perfectly well to build a fly. The simplest explanation for this similarity is that all these animals (known as bilaterians) inherited their BMP-like genes from a common ancestor some 700 million years ago. In early bilaterians, BMP-like genes probably helped lay out the front and back of a developing body. In vertebrates, it is active along the abdomen side, where the digestive system grows. Insects run their digestive system along their back, and in insect larva, that's where BMP-like genes are active.

These BMP genes belong to an entire network of body-building genes that have survived for 700 million years. Some of them switch on BMP genes, while others block their activity. And BMP genes in turn switch on and shut down other genes. This network has been borrowed many times in the course of evolution to build new structures in animal bodies. As vertebrates evolved skeletons made of bone, the BMP network took on a new role helping to build it. (BMP encourages bone to grow, and also to heal--making it the object of a lot of interest in medical circles.) But its role was not limited to ribs and vertebrae. As new sorts of vertebrates evolved, the BMP network was coopted yet again. In birds, for example, feathers grow under the guidance of the BMP network. And so to, the Grants and their colleagues have found, do bird beaks.

So here we have a network of genes that has played a major role in evolution at many scales. It emerged as part of an animal toolkit, which could be used to construct bodies as different as that of a fly and a fish. It was then borrowed and redeployed in new ways, building new structures. And because this network controls many other genes, a small tweak to it can produce some significant changes even within a single species. Alter the timing of BMP ever so slightly in a finch's developing beak, and it may be prepared to survive a drought by cracking hard seeds. Thanks to the relative ease by which beaks can evolve, these sorts of generation-to-generation changes have helped Darwin's finches explode into 13 new species over the past couple million years. Micro and macro, in other words, are bound together into one extraordinary whole.

See you September 13.

Today scientists took another step towards creating the sort of simple life forms that may have been the first inhabitants of Earth. I wrote a feature for the June issue of Discover about this group, led by Jack Szostak at Harvard Medical School. Szostak and his colleagues suspect that life started out not with DNA, RNA, and proteins, but just RNA. This primordial RNA not only carried life's genetic code, but also assembled new RNA molecules and did other biochemical jobs. Szostak and others have created conditions in their labs under which today's RNA can evolve into a form able to cary out these primordial tasks. So far, their evolved RNA molecules can assemble short fragments of RNA, using another RNA molecule as a template.

RNA-based life was presumably not just loose genetic goop, but organized into primordial cells. Last year Szostak and his students demonstrated that RNA can spontaneously get inside bubbles made of fatty acids--protocells, in other words. These protocells can grow by absorbing new fatty acids. When Szostak's team pushed the protocells through microscopic pores (as might happen in seafloor rock), the bigger protocells split into smaller ones, each with RNA inside.

Szostak's protocells have to meet three standards in order to become life. They have to carry a genetic code. They have to be able to grow and reproduce. And as they reproduce, they have to evolve. Szostak has already made some important steps towards the first and second standards, and today he and his colleagues moved towards the third. In Science, they report that their protocells compete for the fatty acids necessary to build membranes. They mixed together protocells containing RNA with empty shells. The mere presence of the RNA in a protocell altered physical properties of the membrane, creating tension in the membrane. The empty shells, on the other hand, were more relaxed. As a result, the protocells pulled fatty acids away from nearby empty ones. The protocells with RNA got bigger, while the empty ones got smaller.

Szostak and his colleagues suggest that this competition for membrane material may have driven the evolution of RNA that could replicate itself fast. The faster RNA could replicate itself in a protocell, the more fatty acids it could grab from slower neighbors. And by grabbing fatty acids, it could grow faster and divide faster.

What's interesting about this theory is how simply it is. Szostak isn't bothering with an army of RNA molecules, some specializing in supporting the cell's structure, some specializing in building the membrane, and some specializing in producing new RNA. Even an incredibly simple RNA-based organism that could replicate itself might be able to take advantage of the power of natural selection.

"A world without memory is a world of the present," Alan Lightman wrote in Einstein's Dreams. "The past exists only in books, in documents. In order to know himself, each person carries his own Book of Life, which is filled with the history of his life...Without his Book of Life, a person is a snapshot, a two-dimensional image, a ghost."

Most people would probably agree with Lightman. Most people think that our self -knowledge exists only through the memories we have amassed of our selves. Am I a kind person? Am I gloomy? To answer these sorts of questions, most people would think you have to open up some internal Book of Life. And most people, according to new research, are wrong.

Neuroscientists would call Lightman's Book of Life episodic memory. The human brain has a widespread system of neurons that store away explicit memories of events, which we can recall and describe to others. Some forms of amnesia destroy episodic memories, and sometimes even destroy the capacity to form new ones. In 2002, Stan B. Klein of the University of California at Santa Barbara and his colleagues reported a study they made of an amnesiac known as D.B. D.B. was 75 years old when he had a heart attack and lost his pulse. His heart began to beat after a few minutes, and he left the hospital after a few weeks. But he had suffered brain damage that left him unable to bring to mind anything had done or experienced before the heart attack. Klein then tested D.B.'s self-knowledge. He gave D.B. a list of 60 traits and asked him whether they applied to him not at all, somewhat, quite a bit, or definitely. Then he gave the same questionnaire to D.B.'s daughter, and asked her to use it to describe her fater. D.B.'s choices significantly correlated with his daughter's. D.B.'s Book of Life was locked shut, and yet he still knew himself.

A few other amnesiacs have shown a similar level of self-knowledge, but it's hard to draw too many lessons from them about how normal brains work. So recently Matthew Lieberman of UCLA and his colleagues carried out a brain-scanning study. They wanted to see if they could find different networks in the brain that make self-knowledge possible. They also wanted to see if these networks functioned under different circumstances--for example, when thinking about ourselves in very familiar contexts and unfamiliar ones.

They picked two groups of people to test: soccer players and improv actors. They then came up with a list of words that would apply to each group. (Soccer players: athletic, strong, swift; actors: performer, dramatic, etc.) They also came up with a longer list of words that applied specifically to neither (messy, reliable, etc.). Then they had all the subjects get into an fMRI scanner, look at each word, and decide whether it applied to themselves or not.

The volunteers' brains worked differently in response to different words. Soccer-related words tended to activate a distinctive network in the brains of soccer players, the same one that actor-related words switched on in actors. When they were shown words related to the other group, a different network became active. And, as Lieberman and his colleagues report in an upcoming issue of the Journal of Personality and Social Psychology, it just so happens that they had predicted precisely which two networks would show up in their scans. (Here's the full pdf on Lieberman's web site.)

When people were presented with unfamiliar words, they activated a network Lieberman calls the Reflective system (or C system for short). The Reflective system taps into parts of the brain already known to retrieve episodic memories. It also includes regions that can consciously hold pieces of information in mind. When we are in new circumstances, our sense of our self depends on thinking explicitly about our experiences.

But Lieberman argues that over time, another system takes over. He calls this one the Reflexive system (or X system). This circuit does not include regions involved in episodic memories, such as the hippocampus. Instead, it is an intuition network, tapping into regions that produce quick emotional responses based not on explicit reasoning but on statistical associations. (The picture I show here is a figure from the paper, with the X and C systems mapped out.)

The Reflexive system is slow to form its self-knowledge, because it needs a lot of experiences to form these associations. But it becomes very powerful once it takes shape. A soccer player knows whether he is athletic, strong, or swift without having to open up the Book of Life. He just feels it in his bones. He doesn't feel in his bones whether he is a performer, or dramatic, and so on. Instead, he has to think explicity about his experiences. Now D.B.'s accurate self-knowledge makes sense. His brain damage wiped out his Reflective system, but not his Reflexive system.

This research is fascinating on its own, and even more so when you think about the evolution of the self. Judging from the behavior of humans and apes, I'd guess that the Reflective system seems to be far more developed in us, while apes may share a pretty well developed Reflexive system. Does that mean that a Reflexive self existed before a Reflected one? Is the self we see in the Book of Life a recent innovation sitting an ancient self that we can't put into words? And does that mean that chimpanzees have a Reflexive self? Is that enough of a self to warrant the sort of rights we give to humans because they are aware of themselves?