Friday, May 06, 2005

Do Inherited Beliefs and Conjectures Affect Inference?


And sometimes the inherited beliefs and conjectures that influence inference are more than 2000 years old:

Ian Fisher, Rome Journal: Some Say His Digging Up of a Legend Is Just a Myth, NYTimes Online Edition (May 6, 2005):
[A]ndrea Carandini of Italy's most renowned archaeologists ... has discovered something extraordinary underneath the tightly packed ruins of the Roman Forum: a palace that he believes belonged to the first king of Rome, who just maybe was actually named Romulus.


[I]n the two decades that Dr. Carandini, 68, has excavated in and around the Palatine Hill, the epicenter of successive generations of Roman rulers, he has without doubt attracted a fair share of skeptics.


[Dr. Carandini] says his latest discoveries show the myth to be quite possibly true, even if the king's name was not necessarily Romulus ....

Others say that in his two decades at the site, Dr. Carandini has sometimes worked backward from myth to explain what he has found, rather than waiting for evidence to emerge from the finds themselves.

Note, this is not simply an instance in which Dr. Carandini and his critics have different prior beliefs. This is equally a situation in which Dr. Carandini's prior beliefs give the archeological evidence that he finds a different flavor in his mind's eye than it has in his critics' minds' eyes.
  • "critics minds' eyes"? Now, really, Tillers!
  • Monday, May 02, 2005

    Seminars in Edinburgh: Law, Probability, and Risk

    Please take note of the following announcement of two transdisciplinary seminars on law, probability, and risk (venue: University of Edinburgh):
    The following dates are now confirmed for the last two LPR seminars:

    1. Criminology organised by Sarah Armstrong: October 14th and 15th 2005 (Friday and Saturday).

    2. Evidence evaluation organised by Colin Aitken: December 2nd and 3rd 2005 (Friday and Saturday).

    Details of the seminar programme, sponsored by [the Economic and Social Research Council], are available at

    If anyone is interested in attending either or both of these seminars, please let Sarah Armstrong ( know.

    There is no registration fee.

    Best wishes,

    Colin Aitken
    C.G.G. Aitken
    School of Mathematics
    The King's Buildings Fax: (0)131 650 6553
    The University of Edinburgh, EH9 3JZ Scotland
    E-mail: cgga (at)

    Aha! I have found a blurb about the mysterious "2029 Project." See Institute for Alternative Futures This is a biomedical research project supported by Pfizer.

    More on the Architecture of the Brain

    Thomas R. Insel, Nora D. Volkow, Ting-Kai Li, James F. Battey, Story C. Landis, Neuroscience Networks: Data-sharing in an Information Age, PLoS Biol 1(1): e7 (Essay, October 13, 2003)(© 2003 Public Library of Science. This is an open-access article distributed under the terms of the Public Library of Science Open-Access License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.):
    In parallel to the worldwide effort to map the human genome, investigators in neuroscience have used a range of techniques to map the brain. The efforts share some superficial similarities: the genome has 3 × [10 to the 9th power] bases and the human brain has roughly 100 × [10 to the 9th power] neurons; both the genome and the brain have embedded modules of functional units (genes versus circuits) that can be mapped in space; and localization of both genes and circuits requires computational power that can be distributed across laboratories. But the analogy breaks down quickly. Whereas fundamental genome data can be addressed as unidimensional text of four letters in varying order, a comprehensive map of the brain includes molecular, cellular, system, and behavioral data—all of which are dynamic, interacting, and interdependent. For example, brain circuitry is organized in three-dimensional space constantly changing in time, with each neuron having [10 to the 3rd power] – [10 to the 4th power] synapses and with many of those synapses capable of plasticity that may, in turn, have significant functional consequences.

    Brain as Computer?

    Some work in artificial intelligence (a/k/a computational intelligence) and in computer science has tried to make hay out of the analogy between computers and brains. Recent work in the biology of the brain, however, suggests that if the brain is a computer, the brain-computer is quite unlike the digital computer (this we already knew!) and -- moreover! -- that the architecture of the brain is also rather unlike computers that make use of parallel processing. Furthermore, some people in the field -- the field of the biology of the brain -- now think that at the perceptual level brains use sensory inputs to make predictions about the attributes of the thing that is being perceived: they suspect that the brain takes a few sensory inputs, does a bit of processing (but initially not much), and then proceeds by searching in memory for similar or analogous mental constructs based on similar sets or constellations of sensory inputs. This sequence of operations is not necessarily the end of the process but such a sequence, I gather, is the way (some people think) the brain gets things started, and quickly so. See the pdf manuscript by an anonymous author, Advances in Cellular and Systems Knowledge of the Nervous System(Feb. 9, 2005), document self-identified as part of something called "Nervous System 2029 Project."

    If You Look for Surprising Coincidences Long Enough, You Will Find Them

    Books: The Long, Strange Journey of Einstein's Brain Online Text for Morning Edition, April 18, 2005:
    In another peer review, Terence Hines of Pace University, writing in Experimental Neurology in 1998, echoed Kantha's criticisms. Diamond's study, he wrote, "is so seriously flawed that its conclusions should not be accepted." Hines's complaint is the same one lodged against Paul Broca by Stephen Jay Gould in The Mismeasure of Man.

    If you look long enough, if you measure a sufficient number of attributes, and if you are highly selective, you can eventually find statistical evidence to support or defeat any claim.