[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: doing computer simulations of locomotion right

From: GSP1954@aol.com

Hutchinson's preference for discussing my criticisms of his work on tyrannos
in the formal literature is reasonable, and it will be interesting to see
what is presented - I'm waiting to see the papers before I put together my
tome on the subject.

There are at least six at present so don't hold your breath. :)

The evolutionary techniques used by Sellars et al better mimic those of
nature than standard biomechanics, and are potentially more effective than
the latter in which people program a computer with what they have concluded
is correct (but may be errant).

I'd actually disagree in this case, although it is an interesting paper. Some of the results contradict reality in humans. For example, note the extremely dorsiflexed ankle joint; that will easily lead to erroneous estimates of muscle force and energy requirements. Trust me, I have the model that is used and it can be shown directly. I don't know if Greg realizes this, but the model they use is the SIMM (Software for Interactive Musculoskeletal Modeling) human model that was developed in the lab I am working in! And it does not seem that Greg realizes that these models use standard biomechanics as their core functions; you cannot get power requirements without it. The original SIMM/SDFast simulations produce excellent results for normal humans but for some reason Sellars et al. are getting odd results. However, the authors are headed in the right direction, using advanced techniques that are far better than the classical approaches used by anthropologists for decades.

A better set of approaches that also match reality is by F. Clay Anderson and Marcus Pandy; search for papers by them on optimization and you'll see a more rigorous approach to locomotor energetics. Clay is also working with me on the SIMM/SDFast models of bipeds, upcoming in our SVP talk on muscle functions and Victor Ng-Thow-Hing's talk (with us) on 3D body dimensions and simulation. I am disappointed by the Sellars et al. results, but hopeful that their models of Lucy and other hominids will develop more rigor. The research team developing them has some great talent and is moving in the right direction with other research.

 So far no one has been
able to show where my conclusions on power requirements in running giants are
in error, yet a number of bright people just do not accept the results
because it's not within their big animals must be slow worldview.

G Paul

I don't believe that big animals must be slow; our elephant work isn't really showing "slow" speeds, although I guess it's all relative, big, semi-big, slow, fast, faster, yada yada ho hum. It's more that I don't buy the 12-20 m/s speeds attributed that are based on methods that I think are superficial, making too many assumptions about form-function correlation/causation, or applying scaling functions that have too much slop in their data. I do not feel we have accurately estimated speed, or for that matter kinematics or limb orientation, for any extinct animal. If folks think "bumblebees can't fly" is a compelling cautionary tale, just think about how many times anatomical inferences have later been shown to be wrong. I'd say it's ten times more often (although attempted much more often too), and difficult to "test" when/how such inferences might be wrong without delving deeper into biomechanics and other methods that incorporate core mechanisms.

I haven't seen the fast-moving giants work substantiated from first principles, and I have my doubts (based on real principles of biomechanics from living animals, not from sweeping generalizations or a static worldview) that it can be shown to work out that way. To satisfy me, it must be solved that way, but not in a back-of-the-envelope fashion (a la RMcNeil Alexander's groundbreaking work), rather in an exhaustive approach with extensive sensitivity analyses of unknown parameters. But who knows; there may be some ways to show that it is possible. That's what we're trying to do, show that it is possible from first principles. So far we have not found a logical solution to the problem that is fully consistent with reality, power requirements or otherwise. The Nature paper was just the first stab at this approach and we'll keep using alternatives/adjustments to see how their results compare. Over the years we'll keep trying, and one way or another progress will be made. I doubt I'll ever be satisfied with any of my or other's work but that's my nature as a skeptical scientist and I'll deal.


John R Hutchinson
NSF Postdoctoral Research Fellow
Biomechanical Engineering Division
Stanford University
Durand 209, BME
Stanford, CA 94305-4038
(650) 736-0804 lab
(415) 871-6437 cell
(650) 725-1587 fax