April 18, 2009 | David F. Coppedge

How Valid Is Computerized Dinosaur Reconstruction?

Can you reconstruct a dinosaur on a computer?  Of course you can.  The question is how accurate it reflects something no one has ever seen.  Live Science told about Peter Falkingham at the University of Manchester who is using “genetic algorithms” and simulating evolution to figure out how dinosaurs walked.
    A dinosaur’s gait doesn’t just jump out of the bones.  Even if you have a whole skeleton, there’s a lot of uncertainty if you don’t know where the muscles attached and how long they were.  Falkingham models different attachment points in his computer and sees if they allow the animal to walk.  “Initial attempts to randomly decipher which pattern of muscle activation works best result almost always in the animal falling on its face,” he explained to Live Science.  Assuming falling headlong was not their normal behavior, Falkingham let the algorithms “evolve” such that a computerized T. rex, for instance, could enjoy a good life.  The genetic algorithms employed “can alter themselves and evolve, and so run pattern after pattern until they get improvements,” reporter Charles Q. Choi explained.  Both scientist and reporter think they are simulating evolution:

Eventually, they evolve a pattern of muscle activation with a stable gait and the dinosaur can walk, run, chase or graze, Falkingham said.  Assuming natural selection evolves the best possible solution as well, the modeled animal should move similar to its now extinct counterpart.  Indeed, they have achieved similar top speeds and gaits with computer versions of humans, emus and ostriches as in reality.

Combining the computer model with data from dinosaur tracks can help present a unified picture of dinosaur life – as long as one takes into account the difficulties in interpreting tracks:

The problem with tracks is that they can be very hard to interpret, as the number of variables involved with how tracks form “is staggering,” Falkingham explained.  “Is the sediment made of tiny clay particles that stick together, or larger sand particles that roll over?  What is the water content, which can help particles stick together, but if you put in too much, it pushes particles apart?  What is the strength, elasticity and compressibility of the soil?  And what happens when you have layers of sediment?  The impressions left behind at lower layers can be very different from the ones left on the surface.
    Physically recreating each potential scenario with a real box of mud is extraordinarily time-consuming and difficult to repeat accurately, so this is where computer simulation comes in.

Running hundreds of simulations with supercomputers produces workable solutions – and surprises.  Falkingham discovered another dubious assumption about tracks made by webbed feet:

Sometimes the experiments can throw up unexpected results.  Falkingham added.  For instance, when he was once simulating wet, sloppy mud to see how an extinct bird walked – findings that could shed light on how birds evolved from dinosaurs – once the result was a webbed footprint, even though the foot itself was not webbed.  The virtual soil had been pushed up between the toes, before collapsing into a platform-like structure that, in a fossil track, could be interpreted as the impression left by a webbed foot.
    The soft parts of animals, such as webbing, only rarely gets preserved as fossil, so much of the evidence for the evolutionary history of webbed-footed-birds comes from tracks, Falkingham said.  These results call for careful reinterpretation of webbed footprints.

Falkingham is working next on modeling four-footed dinosaurs.  Readers can make their own assessments of the scientific validity of virtual dinosaurs walking on virtual soil in a virtual world.

Computer simulations confer degrees of probability – not certainty – on their conclusions.  Presumably, the model gains credibility points if it can accurately simulate the stride and tracks of a living bird or animal.  Even so, the problem of underdetermination of theories by data always leaves room for possibilities that a working model contains flaws that cancel each other out, or were rigged to guarantee the result a theory needed.  These and other problems mean that models cannot be extrapolated carelessly into unknown cases.
    Falkingham and Choi won Stupid Evolution Quote of the Week not for trying to understand dinosaur physics, but for “Assuming natural selection evolves the best possible solution” for anything, and for assuming their findings “could shed light on how birds evolved from dinosaurs.”  Any scientist who assumes that chance plus natural law could find a solution to anything is not thinking straight.  They might just as well assume that chance and natural law evolved the supercomputer that ran their genetic algorithms.  That’s artificial (intelligently-designed) selection, not natural selection.  Waltzing over assumptions is a sure way to fall on one’s face in logical mud.

(Visited 5 times, 1 visits today)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.