I AM A BIOLOGIST, AND I STUDY ROBOTS. BUT AS SOON AS I started describing my research to other people, it was clear I was in trouble. I was speaking to a longtime friend and colleague about a biology grant I’d just gotten from the National Science Foundation to build robots when he stopped me in my tracks. “What do robots have to do with biology?” he asked. I knew then, with the certainty that only dread can provide, that this was an inescapable question—it was the issue that would come up first from now on, every time one of my students or I presented our strange new work to biologists.
What’s the problem? First and foremost, biologists do not study robots. They work on organisms—living things, their environments, and their evolutionary history. They use machines as tools to ascend a rainforest canopy, as instruments to measure biomechanical properties, as modes of transportation to collect fish from a coral reef. As my friend had stated so succinctly, machines in general, and robots in particular, have nothing to do with biology—from his point of view. Not from mine.
I tried to fight back, blurting out the well-rehearsed line that I’d included in the grant proposal: “We use robots to model extinct vertebrates.” With that being not so much an answer as a statement of intent, I got a raised right eyebrow and then a gentle “Well, I hope it works out for you.” Communication over and out.
I needed a better answer.
No matter how cool robots are (to me), their swaggering presence wasn’t enough to justify their usefulness in biology. And this was a problem not only for me—at least I had a grant—but also for the undergraduate researchers in my lab, who would rather avoid being recognized as having been trained by a known kook. So we talked it over until we hit upon a solution that has proven to work for about half the biologists we encounter.
We decided to equate different models of biological systems: those run on computers and those run on robots. Both are machines, after all, and computers are already used in almost every branch of biology, modeling—among myriad other things—neural networks, predator-prey interactions, virus evolution, and perambulating Tyrannosauruses. In fact, “computational biology” is the hot field right now, the bull’s eye on the what-to-be-working-on-if-you-want-a-job-in-academia dartboard.
Robots—mobile ones, anyway—are essentially self-propelled computers. They are machines that run sets of instructions—their software—and produce an output. Certainly, the outputs seem different. Computers output binary bits that we use to represent numbers, and those numbers, in turn, represent everything from screen colors to mathematical formulae to electronic books. Robots output what we recognize as behavior, but underlying it all are the same bits.
That’s not to say there might not be important distinctions between a robot and a computer. Jeff Staten, a senior engineer at IBM, says a robot “is a computer that’s inside out.” Jeff’s point is that computers today are networked and make decisions based on input from other computers most of the time and humans tapping at a keyboard only some of the time. A robot, although it has a computer inside, makes decisions on its own, with information gathered only through its sensors. The messages the robot receives and sends are physical. The mobile robot, unlike most computers, can be autonomous. What autonomous robots have, which computers don’t, is agency.
Agency is what human observers ascribe to anything, organic or artificial, that appears from an external perspective to act on its own. For those of us working in the world of artificial intelligence and cognitive science, an agent can be an organism or a machine. Humans are agents. My dog, Kooka, is an agent. And so long as there is no unseen human pulling the strings of remote control, robots are agents too.
An autonomous robot is an agent using its own sensory inputs to perceive the world, make decisions about how to move, and, in turn, having those movements affect how it perceives the world. This constant feedback between what an agent perceives and how it moves is what my colleague Ken Livingston and I call a perception-action feedback loop. Multiple perception-action loops in an agent can be operating in parallel, working in combination, fusion, or competition. What we observe the agent doing—moving and interacting with its immediate environment—is what we define as behavior.
An agent’s behavior is its computational output. “Behavior emerges from the agent-environment interaction,” as Livingston is wont to say. And behavior, by this definition, is something that autonomous robots have that computers do not. Oops. Have we just defined ourselves into an identity error that invalidates the logic of our first response? No and yes. No, because autonomous robots have, as part of their agency, embedded computers. Yes, because robots are more than simply computers that move.
Autonomous agency is, ultimately, the answer to my colleague’s question, what do robots have to do with biology? They enable us to build models of how organisms behave. Of course, building models raises another question.
It turns out that, much like most biologists think they shouldn’t be studying machines, many think they shouldn’t be studying models, either. One criticism of models or simulations of any kind, instantiated on either a computer or a robot, is that they are, at best, artificial systems that merely copycat the outward behavior of the biological system. Models, the argument goes, fail as true or accurate representations of the underlying causal phenomena because the underlying functional mechanisms are different from those operating in the system of interest. As Norbert Weiner, the founder of the field of cybernetics, is alleged to have said, “The best model of a cat is a cat.” Although this is not quite like saying that to understand cats, you can only study cats, some people do jump to that conclusion.
With this cats-only criticism in mind, my colleagues and I jump to a different conclusion about models: you have to be very careful when you build them. You have to take care to explain what you are trying to do, how you intend to do it, and how you are discriminating between a bad model and a good one. Bad models won’t be anything like cats, and therefore, they will perhaps tell you something about your ability to make a noncat—but little else. Good models, argues Barbara Webb, a biologist at the University of Edinburgh and one of the founders of the field of biorobotics, are those with explicitly defined goals and goals that are attained. Some models are meant to behave like the targeted system. In those cases a close or perfect behavioral match between your robot and its biological target—if it walks like a cat and meows like a cat—means that you have a good model.
As a real example, Sarah Partan, an animal behaviorist at Hampshire College, wanted to study how squirrels respond to the behavior of other squirrels, so she built a robot squirrel that flicks its tail and adjusts its posture. For Partan, a good squirrel model is able to trick the real squirrels into responding to the robot as if the robot were a squirrel. Happily for Partan, the trick worked.
Behaviorally speaking, Partan’s robotic squirrel is a good model of body posture and tail motion. At the same time, it is obviously a bad model of the neuromuscular mechanisms involved in limb motion. However, to model neuromuscular mechanisms wasn’t her intent. If it was, you’d judge the goodness of her neuromuscular mechanisms model not by how well it elicits tail flicks in other squirrels but rather by how closely it matches the underlying functional mechanisms used in muscles and nerves. If the artificial muscles worked like biological ones, generating peak force at only one length and for short periods of time, then she’d have modeled that mechanism accurately, irrespective of whether or not her robotic squirrel can dupe real squirrels.
Different models, then, serve different goals. Webb enumerates seven: emulating behavior and mechanism (both of which we’ve just seen) as well as abstractness, medium, generality, level, and, particularly important for us, whether or not the model tests a hypothesis—that is, an idea that you have about the biological system.
For Webb, if you are interested in any particular aspect of cats and you have learned as much from real cats as you can, then go ahead and model a cat. I think this would be Weiner’s position as well. But if your goal is to learn more about cats, Webb says, avoid the temptation to build something cool just for the sake of building it. There must be a specific target: you shouldn’t just try to build something cat-like. Her criticism is aimed at two fascinating fields, adaptive behavior and artificial life, in which many workers model invented animals, called animats. Although animats illuminate general operating and cognitive principles, Webb argues that the adaptive behavior and artificial life approaches do little to test specific hypotheses about how real animals work. For her, biologists building animats to test biological hypotheses usually walk away empty handed.
Webb’s critique of adaptive behavior and artificial life gets to the heart of my colleague’s skepticism. He intuited two of her points related to the value of any model to a biologist. First, your model must have a specific biological target. Second, your model must be relevant and must enable testing a hypothesis about the targeted system.
Of course, Webb’s critique also raises some problems for our response to the skeptics—that we should be able to use robots because robots are essentially computers. If computer models aren’t necessarily biologically relevant, then our robots may not be either. So our critique of the critiquer became this: he was asking the wrong question, or at least asking a sufficiently vague question to allow literalists like us to misunderstand. Instead of asking, “Why robots?” our skeptic ought to be asking, “What is the scientific purpose of your model, and why is your model in the form of a physically embodied robot?” But then there’s probably a question you’d like to ask of me: how did a biologist ever find himself sticking up for robots? The answer’s quite simple: it was for the love of fish.
I’ve loved fish since I was a child, when I first saw Jacques Cousteau’s underwater world on television. I followed the wake of swimming fish and other aquatic vertebrates through college, graduate school, and, now, into a career. And there was a lot to learn in the flesh. With Steve Wainwright of Duke University and Mark Westneat of the Field Museum, I’ve scuba dived to videotape the propulsive oscillation of two hundred–pound blue marlin. With Mark, Melina Hale of the University of Chicago, and Matt McHenry of the University of California, I’ve outfitted rainbow trout, bowfin, longnose gar, and African bichir with tiny instruments to measure muscle function during escape maneuvers. With Wyatt Korff of the Janelia Farm Research Campus, I’ve used high-speed video to investigate how Wyatt’s trained Amazonian arawana can propel themselves out of the water to catch food in midair. With Lena Koob-Emunds and Tom Koob, I’ve worked at the Mount Desert Biological Laboratory to study the biomechanics of slimy pink hagfish so that we can learn about swimming without a vertebral column. Finally, with Marianne Porter of the University of California, I’ve measured the mechanical properties of the backbones of dogfish sharks to see how skeletons transmit force.
I love real fish, and more than twenty years’ work has shown me much about their shape and structure, how they move, and how they evolved. But real fish only reveal some of their secrets. As with any science, we are limited by what we can and cannot observe and measure. Sometimes we lack an instrument or a technology. At other times we lack the right fish. Take, for example, the giant blue marlin. They die in captivity, so studying them in a lab was impossible. So we went blue-water diving not because we wanted to (it’s very expensive and dangerous work) but because we had no other choice. We had to film blue marlin from a distance and leave some of our questions unanswered.
Well, not entirely. We could’ve studied marlin in other ways, but other questions would’ve been left unanswered. Say you want to know what’s going on inside a blue marlin when it swims. You could, as Barbara Block of Stanford University did, build a team of engineers and physiologists to design tiny instruments that can be implanted quickly in a marlin that you’ve brought alongside a boat using hook and line. These instrument tags carry their own computer, power pack, and broadcast system, collecting data and sending signals back to a ship or satellite. Block can measure the marlin’s body temperature, muscle activity, speed, and depth as it moves freely about its oceanic cabin.
But for all Block’s approach reveals about physiology, the method reveals nothing about the biomechanics of marlin backbones—and that’s what I wanted to study. The backbone, or vertebral column, runs from an animal’s head to tail, and its presence is one of the signatures of the vertebrates, a group of animals to which amphibians, birds, fish, mammals, and reptiles—some fifty-eight thousand species—all belong. In a fish the backbone prevents the body from shortening while also allowing it to bend, and it gives the whole body important mechanical features, such as the ability to store and release energy elastically like a spring. My pursuit of this question is what would ultimately send me headlong into the world of artificial intelligence and robots.
I first met Block—and the blue marlin—back in 1986, when she, as a newly minted PhD from Knut Schmidt-Neilsen’s lab at Duke University, convinced me, a newbie PhD wannabe in Steve Wainwright’s lab, to work on the biomechanics of marlin vertebral columns in the laboratory. Under the guise of buying me a cup of coffee at the Ninth Street Bakery, Block pulled me out of the lab my first day so she could expound the virtues of the marlin.
Of all the fish, she explained in the car, marlin are the best, the fastest, biggest, coolest predators in the sea. “Think tuna are fast?” she asked rhetorically as we pulled into the parking lot. “Well, marlin eat tuna!” As we walked across the street to the bakery, she went for the kill. “Have you seen the vertebral column of a marlin?” she asked, sounding like a minister in the First Church of Poseidon. I knew the proper response: “No, I have not seen the vertebral column of a marlin. What does it look like?”
She introduced me to the mysteries of the marlin’s backbone. “It’s not like a bunch of little bones linked together, like pearls on string, that you see in regular bony fish,” she said. “The vertebral column of a marlin looks like a piece of wood, a long pine board, a one-by-six, with bones overlapping, bones welded together with collagenous connective tissue to form a single, giant spring.” She paused for effect. “And this spring works to store and release energy, the energy that powers the high speeds and spectacular leaps of marlin.”
I shuffled forward in the line, unable to muster words. Only images came to mind: marlin leaping and spinning above white caps, and terrified tuna, swimming for their lives but unable to avoid the explosive charges of the spring-loaded marlin. Block waited for a moment, paid for our coffee and muffins, and guided me to a table. Signaling with her hand for me to eat something, she gave me a chance to return from my reverie. Then she said, knowing the answer, “So. Are you in?” I gushed, “Absolutely!”
Giddiness gave way to the not harsh but practical realities of scientific research. Working in Wainwright’s lab, I spent the next five years chasing after the elusive blue marlin, literally and figuratively. I wanted to measure their vertebral column’s mechanical properties, features like stiffness—related to how much the vertebral column would resist the magnitude of bending and how much spring energy it would store—and energy loss—related to how much the vertebral column would resist the speed of bending and how much energy would be lost as heat. If stiffness is large compared to energy loss, the backbone would be a spring; if the stiffness is relatively small compared to energy loss, the backbone would work as a brake. If I could measure stiffness and energy loss of the vertebral column over a range of motions and speeds, I knew that I could have some idea of what Wainwright calls “mechanical design”—in this case how the mechanical properties of the vertebral column allow it to operate as the blue marlin swims or leaps.
Unable to buy an off-the-shelf marlin-testing machine (they don’t exist), I had to design, build, and calibrate a customized vertebral column bender. My DIY guru for this challenge was Steven Vogel, also at Duke, who helped me brainstorm designs and taught me the difference between a DC brushless motor and a servo one. Once I had a working bending machine in place, Block and Wainwright helped get me and my machine out to the big island of Hawaii and the Pacific Gamefish Research Foundation.
On the Kona side of the island deepwater blue marlin are caught by recreational fishers literally in sight of the steep-sloped volcanic beaches, where, hat in hand, I would beg at the local fish houses for the castaway vertebral columns. Once I had one, I was unable to sleep until I had put each individual motion segment, consisting of two vertebrae and the intervening joint, through a series of mechanical tests. I’d bend the segments with varying frequency and amplitude, just like the marlin would have done as it hit the turbo button to pursue a tuna. To get a sense of what parts of the bone and joint structure helped cause changes in stiffness and energy loss along the column, I also measured the size and shape of each joint and the adjoining vertebrae. Lather, rinse, repeat. After several weeks I had tested the vertebral columns from six different marlin ranging in length from four to seven feet and weighing from thirty-six to more than two hundred pounds.
Back at Duke I began running the raw data through the Newtonian equations of motion that govern the relation between the bending motions the machine imposed on each joint and the bending torque each joint developed in resistance to that imposed motion. Looking over the range of joint positions, bending frequencies, and amplitude, I began seeing some very interesting patterns. The biggest surprise was that the tail, which looks from the video we took of marlin swimming to be the most flexible part of the body, actually has the stiffest part of the vertebral column. Talking with Wainwright, we realized that this was a counterintuitive result only because we were thinking of a jointed column as a series of bony blocks and frictionless hinges. If instead the joints—the hinges—were very stiff because of all of the overlapping bits of bone that Block had talked up, then the joints themselves appeared to be capable of storing energy as they bend.
But were those same joints able to release that spring energy as they unbent? This is where the energy loss came into play, and the marlin played a trick on us again. With simple ideas of springs in our heads, we had been thinking that as the marlin swam faster, increasing the frequency of their tail beats, their vertebral column would become even more spring-like, storing and releasing more elastic energy to match the power that the faster speeds demanded. We expected stiffness to increase and the energy loss to decrease. Just the opposite occurred.
To make sense of these surprises in the biological context of the swimming marlin, we put this information about mechanical properties into a mental, conceptual model of what we thought might be going on inside the marlin. Our guess was that as marlin increased swimming speed, the vertebral column would be adjusting its mechanical behavior, switching gradually from a spring to a spring with a brake. This spring-and-brake mechanism is exactly how the shock absorbers in your car work, with the spring resisting the initial bump, giving way gently, and then returning the wheel to its place on the road. At the same time, the brake, or what we call a dashpot in a shock absorber, uses fluid to dampen the spring’s motion, keeping the spring from bouncing the car vertically after that first bump.
Sounds reasonable, doesn’t it? We’ve only one problem: a backbone in my machine doesn’t necessarily act like one in a dynamically operating animal. That problem is what drove us to Hawaii in search of underwater footage of swimming marlin. We wanted to see how living marlin moved their bodies, how fast they beat their tails, and how much they bend their backbones. Knowing this would let us make a good guess about how the vertebral column is operating during swimming, but it still wouldn’t let us measure the backbone directly as it bent nor evaluate how the muscles responsible for driving the bending do their work. Nor, for that matter, would we be gauging the complex forces from the surrounding water interacting with the undulating body.
The blue marlin is the poster child for problems with the biomechanical approach. And it wasn’t that we were stymied just because we couldn’t bring the fish into the lab to implant measuring devices. Even if we could, problems would remain. For example, when we try to directly measure the forces that bend the vertebral column of living, swimming sharks, we find that the surgery needed to carefully implant the strain gauges on the skeleton disrupts the surrounding muscle, leaving us unsure whether we have changed the way the shark moves. What’s more, the measurements remain somewhat crude: Elizabeth Brainerd and Bryan Nowroozi of Brown University have used real-time CAT scans to show that many of the motions of a fish’s intervertebral joints are subtle enough that they are still difficult to measure accurately.
At this point the best model of a marlin backbone is not a marlin backbone. Because we couldn’t study it any further in the living fish, we were left with three choices. One: quit and do another project. As depressing as that sounds, sometimes it is the only practical alternative. In the hopes of finding a species that works really well for answering a ton of different questions (which would make it a “model organism”), switching species is a common response. Two: try to build a new instrument or experimental procedure to answer the question. For the stubborn and electromechanically minded, this is often a way to work out your frustrations and keep busy while you come to grips with the fact that you really, truly are stuck. Three: build a model of your fish. For those of us who need to keep writing papers so that we can earn tenure and win research grants, this is the way to go—we model.
This may strike you as a cynical way to have backed into modeling. It is, I admit it. So we might as well go through the front door, with a smile, by asking again why a biologist would use robots to study animals. The positive answers are both practical and theoretical. On the practical side we’ve seen that we reach limits with both our instruments and our animals. On the theoretical side some argue for what is called a synthetic approach, a bottom-up philosophy borrowed from engineers that stands in contrast to the biologist’s usual reduce-and-analyze methods: if we can build it, then we understand it.
This synthetic approach underlies what Rohlf Pfeifer and Christian Scheier call embodied cognitive science or embodied artificial intelligence, and it is at the core of the defense of robots I offered earlier: build embodied robots that behave as autonomous agents. The behavior that these agents create can then be understood on the basis of their physical design, programming, and interaction with the physical world. If we can build it, then we understand it.
I should mention that although this synthetic approach with embodied robots is new to biology, physical models have been used to great effect for some time. Vogel, the biomechanics professor from Duke University, pioneered the use of physical models to test ideas about which engineering principles nature is exploiting in organisms. One of Vogel’s and Wainwright’s former students, Mimi Koehl, a biomechanics professor at the University of California, Berkeley, is world famous for building physical models of animals—both living and extinct—to test ideas about the functional principles in operation. Physical models have a lot going for them.
* You can build a simplified version of an organism or its part.
* You can enlarge or reduce the size of the part or the organism.
* You can isolate and change single parts, keeping all else constant.
* You can reconstruct extinct organisms.
For Koehl physical models complement experiments with both real organisms and computer models. Although we’ve talked about the limits of experiments on real organisms, like marlin, it’s worth mentioning briefly here some of the problems with computer models. As many of us have learned, computer models are fantastic when you can represent the biological phenomena you are modeling in well-formed equations or even clunkier but still serviceable numerical recipes. Beautiful is the clean-line output, perhaps cloaked with a surface function painted in a million hues of color, of a computer model to the abstract-art–trained eyes of a scientist. But to get that elegant output, we always have to make, even in the most accurate models, many simplifying assumptions. The trick is to make the right ones.
My first step when experiments with marlin left me at the end of my conceptual tether was to create a computer model. From my biomechanical tests I had derived equations of motion that described the torque needed to bend each joint and the resulting angular motion. My initial assumption was that these equations were sufficient to describe the mechanical behavior of the backbone in a swimming marlin. I simplified the backbone mathematically to a series of equations that were linked by the bending torques that we imposed on one and then another joint. I assumed that muscles and water resistance were acting to create the torques being transmitted up and down the backbone. With these assumptions and simplifications, I created a beautiful animation of an undulating backbone passing waves of bending from head to tail.
I presented this model to the department as the capstone of my PhD research. After the talk a fellow graduate student, Matt Healy, hurried up to me and said, in a concerned hush, “You’ve got a problem. I just heard Vance Tucker say that you may have violated the laws of physics.” This was the equivalent of saying, “A god-like scientist thinks you’ve made a huge mistake and your reputation and career are in immediate jeopardy”—gulp. Tucker, a physiologist working in the physically messy world of bird flight by using brilliant flow-tunnel experiments and engineering theory, was another one of Duke’s biomechanics gurus. I immediately shuffled to his downstairs office.
In response to my knock Tucker looked up from his lab bench and, with a flash of recognition, said, “Come in.” Terse understatement—bad sign. I’m in worse trouble than I thought, I said to myself. I sat down and, unable to bear any longer my impending professional death, pointed right to the sword of Damocles hanging over my head, saying, “I heard that you think I violated the laws of physics?” His response was gracious and careful. Understanding the close proximity of his criticism to my forthcoming dissertation defense, he reminded me that the model I had presented was but one of five chapters in my thesis. Its problems were, by themselves, unlikely to overturn any experimental results because the experiments were independent of the model. And, he offered, he hadn’t seen for himself the model chapter. “But,” he finished, “it appears to me that you’ve created a perpetual motion machine.”
Tucker was right, and I knew it immediately. My various assumptions and simplifications allowed me to create a model that violated the Second Law of Thermodynamics. I had assumed and simplified away energy input and loss so that my backbone, once bent, would continue undulating away forever.
It turns out that this kind of violation of reality is not so uncommon: years later a master of physical modeling and bioinspired design, Charles Pell, would say to me, “Every computer model is doomed to succeed.” Any computer modeler can always create beautiful output even if the physics of the model were wrong. This giant pitfall of the naive computer modeler (ahem, yes, that was me) is the reason that researchers like Pell, Koehl, and Vogel are careful to build physical models. As Pell said to me at another time: “Physical models can’t violate the laws of physics.” If an engineer’s design violates the laws of physics, the machine won’t go on forever: instead, it just won’t go. So we now have a fifth reason to use physical models and not digital ones to understand biological systems. This reason is so important, however, that I will make it point number one:
* You can’t violate the laws of physics.
* You can build a simplified version of an animal.
* You can change the size of the animal.
* You can isolate and change single parts, keeping all else constant.
* You can reconstruct extinct animals.
Do not read this, please, as saying that all computer models violate the laws of physics. Many, many computer models accurately model the physics of the world. You just have to be careful and skilled about which simplifications and assumptions to make.
For my part I’ve realized that the mathematical representation of the biology and physics of swimming fish is, as some say, “nontrivial.” In fact, for a long time many research labs around the world have been working on the hydrodynamics of flexible bodies, like fish, interacting with a surrounding fluid. The applied mathematician James Lighthill was knighted in 1971 for his efforts, which, among other things, describe how a fish creates thrust. Today, teams of biologists, fluid dynamicists, mathematicians, and computer scientists attempt to couple the physics of fluid with that of muscle and connective tissue. Nontrivial, indeed. Robert Root and Chun Wai Liew, both of Lafayette College, and I collaborate on this front, and because of their expertise, I am happy to report that I’m no longer accused of building computer models that violate the laws of physics. In terms of how they represent the physical world, the computer models that we make are not as complex as a robot. Although our computer models of swimming fish are two-dimensional, our fish-like robots are three-dimensional. Our computer models of swimming fish have a lower speed limit, below which Lighthill’s thrust equations don’t work; our fish-like robots can slow down and even stop. Further proof of Rodney Brooks’s dictum: “The world is its own best model.” If you want to be sure that your model hasn’t left out any important physics, the best thing to do is to build it in the real world.
We can now revise our list of reasons to use physical models, here adding reasons based on what we know about autonomous robots. With physically embodied robots built to model animals,
* You can’t violate the laws of physics.
* You can build a simplified version of an animal.
* You can change the size of the animal.
* You can isolate and change single parts, keeping all else constant.
* You can reconstruct extinct animals.
* You can create animal behavior from the interaction of the agent and the world.
* You can test hypotheses about how animals function in terms of biomechanics, behavior, and evolution.
Now you can see why I got interested in physical models to study backbones. But there is one point I have completely ignored so far: the ability to reconstruct the evolution and behavior of extinct organisms.
Consider the complexity of the marlin backbone. It is unusual enough that Block mentioned it in her pitch to get me working on the organism. The point is, of the fifty-eight thousand vertebrates, if we looked at a variety of species, you’d see a great diversity of backbones. Some species have a continuous collagenous rod lacking bones, called a notochord. Some have a series of vertebrae, bones that form the vertebral column. Some have something in between, with what looks like partial vertebrae forming around or along the notochord.
What’s more, we know from the fossil record that our earliest vertebrate ancestors lacked vertebral columns themselves, instead having only a notochord. This continuous axial skeleton evolved earlier in a group of animals known as the chordates. In addition to vertebrates, chordates include living nonvertebrate species, like sea squirts and lancelets. From some group of long-extinct, notochord-bearing chordates, the first vertebrates arose over 530 million years ago. There must have been some problem that being a vertebrate and then having bony vertebrae solved. The question was, what?
I first got into this evolutionary question when I was studying blue marlin. At the time, in the lab of Serge Doroshov at the University of California, Davis, I was studying white sturgeon, big freshwater fish that keep the ancestral notochord as their backbone, even as adults. I would film living ones and subject the backbones of dead ones to the same tests I was using on marlin backbones. The basic hypothesis was simple as can be: vertebral columns, by virtue of possessing rigid bones, would be stiffer in bending than would notochords. Our data suggested that this was correct.
It still didn’t tell us much about the why this trait in marlins evolved or why it did not in sturgeon. As Steve Vogel likes to say: “Biomechanics is about tactics, not strategy.” In other words, biomechanics can tell us about the functional consequences of different structures but not why those different functions may have conferred behavioral and evolutionary advantages to the individuals that possessed them. To make the leap to having anything relevant to say about the evolution of vertebrates, I had to assume (here we go again) that what we learned from two species of fish applied not only to other species of fish but also, in particular, to ancient swimmers like Haikouichthys. These little inch-long jawless fish lived some 530 million years ago and had what looks like little bits of irregularly shaped cartilage blobs on and around its notochord. Revisiting an idea first proposed two centuries ago by Sir Everard Home, Karen Nipper, an undergraduate working in my lab, and I figured that increased stiffness ought to be what you need to swim faster. A stiffer backbone would be a bigger spring, storing more energy that could be used to power the tail.
Knowing that it would be terrifically difficult to measure speed and backbone stiffness in many species (just measuring marlin and sturgeon took me several years), Nipper came up with an easy proxy for backbone stiffness: the number of vertebrae. She also had to find a stand-in for maximum swimming speed, which is notoriously difficult to measure: the swimming fish’s “propulsive wavelength,” roughly the curviness of its body as it swims. Fish with a large propulsive wavelength, like tuna, tend to swim much faster than fish with a small propulsive wavelength, like eels. When we correlated the propulsive wavelength with the number of vertebrae, we found a weak but statistically significant relationship. As the number of vertebrae increased, the propulsive wavelength decreased. Converting this proxy-based result back into our variables of interest, we expected that stiffer backbones would allow their possessors to swim faster than those with floppier backbones.
A complementary approach, known as the phylogenetic approach, pointed us in the same direction. A phylogeny is the branching pattern of ancestor-descendent relationships that describes the evolutionary history of any group of organisms. You can reconstruct these relationships and the timing of evolutionary change by building what is known as a phylogenetic tree—a network that clusters species according to their genealogy. Strictly speaking, a phylogenetic tree is a hypothesis about evolutionary relatedness; it can be tested by collecting new data about the shared features as well as data from newly discovered features and new species. Once you have a tree that is well-supported by a variety of data, you can use it to answer questions about the pattern of evolution. You can map out related features, like notochords and vertebral columns, onto the branches of the tree. You can learn what feature came first, how many different times the feature evolved, and what other traits your feature of interest evolved alongside. This ability to map changes in features, what phylogeneticists call character state evolution, is what makes phylogenetic analysis such a powerful tool.
Using a phylogenetic tree of vertebrates, Tom Koob, a biochemist formerly of the Shriner’s Hospital for Children, and I correlated the pattern of vertebral evolution with changes in swimming behavior. When you map out just the evolution of vertebrae onto a phylogenetic tree of living vertebrates, you get a big surprise: vertebrae appear to have evolved from notochords at least three times. Vertebrae convergently evolved in elasmobranchs (sharks, skates, rays), ray-finned fishes, and tetrapods (amphibians, reptiles, bird, mammals). “Convergent evolution” is a fancy phrase for the same feature—in this case, vertebrae—having evolved independently in different species. Convergent evolution excites the heck out of biologists because it is like naturally repeating an experiment and seeing if you get the same result. Convergent evolution is thus taken as indirect evidence for similar kinds of selection pressures—in different species at different times and places—causing a similar outcome. In the case of vertebrae, they appear to be a good solution to a similar evolutionary problem. But still the question: what is the problem that vertebrae solve?
Thinking fish, fish, fish, Koob and I overlaid on this pattern of convergent vertebral evolution the pattern of changes in swimming behavior. Because we really know so little about swimming speeds and accelerations in vertebrates—which is the same problem that plagued us in the biomechanical analysis—the correlation was weak and, therefore, disappointing. First off, we had to leave out the land-based tetrapods because few adult tetrapods have retained their ancestral fish-like bodies and swimming behaviors. Second, with only elasmobranchs and ray-finned fishes to compare, we only have two large points on the map. Given those caveats, what we think we see on the tree is that vertebrae are correlated with faster swimming. Observations of single species appear to bear this out: frilled sharks with notochords are slow and plodding; mako sharks with vertebrae are some of the fastest fish in the sea; paddlefish with notochords cruise along but are not acrobatic; salmon with vertebrae leap over waterfalls. We were left with the same expectation our biomechanical analysis generated: stiffer backbones would allow their possessors to swim faster than those with floppier backbones.
But this expectation—this prediction—even though it is based on biomechanical and phylogenetic data, isn’t satisfying because it leaves so many questions unanswered. Are the proxies for stiffness and speed reasonable? Is the phylogenetic tree accurate? What other parts of the body, like muscles and shape, influence stiffness and speed? Do we find only a weak correlation because other parts of the species are different too? Would the correlation hold up if we could measure top speeds seen in the wild? Might stiffness also impact other parts of swimming performance, like acceleration and turning? What are the trade-offs in performance with increased speed? And worst of all, these questions don’t even speak to the evolutionary question of the dynamic process of adaptation.
When we ask why vertebral columns evolved from notochords, we are asking about adaptation. For biologists adaptation is the process by which natural selection acts over generational time to alter—to evolve—the characteristics of a population of organisms. Evolution by natural selection—as proposed by Darwin and supported since his time by thousands of experimental and observational tests—happens when the following conditions are met: (1) a feature, like the backbone, varies from individual to individual; (2) genes, at least in part, code the feature and its variations; and (3) the feature’s variations impact how individual organisms behave, survive, and reproduce relative to others in that population. When these three conditions are in place, what we see as we watch a population over time is that some individuals are better at making babies than are others. Because of these individual differences in reproductive output, as individuals and generations die, the population looks different, physically and genetically, from what it once looked like. This change over time is what Darwin called “descent with modification” and what we now call “evolution by natural selection.”
FIGURE 1.1. Evolving robots. Three autonomous, fish-like robots compete with each other for food. Because the swimming mode, sensory system, and brain of these robots are based on the tadpole-shaped larvae of sea squirt chordates, we call them “Tadros,” short for “tadpole robots.” Each Tadro has for its axial skeleton a notochord of differing stiffness. The stiffness of the notochord controls the swimming performance of the Tadro. Stiffness of the notochord is genetically coded and can, therefore, evolve from one generation to the next.
Like a clumsy criminal, adaptation leaves behind many clues in the DNA and anatomy of extinct and living species. But adaptation never leaves behind witnesses or a surveillance tape. Biologists inevitably have to guess at the process of evolution. The best guesses about what went on come from reconstructing the events. Using the clues—the physical evidence—good investigators can piece together a step-by-step sequence of places, agents, and interactions that most likely caused the outcome.
And what can we do to test this sequence? We can build models, let them run, and see if their behavior matches our predictions based on our evolutionary reconstruction. But we can also do one better: let the models evolve. This idea is what would ultimately lead us to invent something my students, collaborators, and I came to call Tadros (Figure 1.1). Starting with those little autonomous robots—not much more than a small computer in a bowl—we were about to embark on a journey of considerable discovery that would help us understand not just what a backbone does for a marlin, but what evolution can do for technology, and what technology can do for our knowledge of the history of life. Which is to say, Tadros themselves would be the best answer to the question: what do robots have to do with biology?