November 2, 2009

  • Orthogonal versus independent: The linear algebra of truth

    Philosophers are quite promiscuous in
    their usage of the word “orthogonal”.For instance: Daniel
    Dennett writes in Freedom Evolves that free will and
    determinism are orthogonal. The Stanford Encyclopedia of Philosophy
    article on “Moral anti-realism” has a section arguing that
    subjectivism and relativism are orthogonal. Accommodationists toward
    religion will often claim that science and religion are orthogonal.

    What these philosophers seem to mean
    is that the concepts are independent,
    that when one changes it is not logically necessary that the other
    change as well. This is no doubt correct; but
    orthogonal is
    a far stronger notion than this, one which has not at all been
    demonstrated.

    Two
    vectors are (linearly)
    independent if
    they are not parallel or antiparallel; similarly, two concepts are
    independent if they are not perfectly correlated or anticorrelated.
    Determinism and free will, subjectivism and relativism, science and
    religion, are surely independent.

    But
    two vectors are only
    orthogonal if
    they are perpendicular, if they share no linear components
    whatsoever. Similarly, two concepts should only be considered
    orthogonal if they are
    completely uncorrelated.
    This is difficult to find in nature, though it does occur—typically
    in completely unrelated domains. Belief in free will is basically
    orthogonal to theoretical claims about dark matter.

    Clearly
    determinism and free will are not
    completely
    uncorrelated
    —in theory and in
    practice beliefs about determinism are closely connected with beliefs
    about free will. Similarly, subjectivism and relativism are closely
    correlated, and belief in science and belief in religion are strongly
    anticorrelated (hence, 95% of the National Academy of Sciences is
    atheist or agnostic).

    It’s
    not a big deal, I suppose; but I think more consistent usage might
    alleviate some confusion. When one person says “orthogonal”
    and another objects that the two concepts aren’t
    completely
    unrelated, and the first
    responds that this isn’t what he meant… basically the two are
    talking past each other until both can agree about what “orthogonal”
    really means.

Post a Comment

Leave a Reply

Your email address will not be published. Required fields are marked *