joeyo

Most insects do not exhibit the classic signs of pain responses (the new findings discussed in this New Yorker article, notwithstanding). For example insects generally don't groom or guard an amputated limb. This puzzling (lack of) response can be explained as being aligned with their reproductive strategy: they reach breeding age quickly and die soon after. Thus, it's better not to waste energy avoiding limb loss for a future that won't happen.

Despite this lack of sentience, insects can be quite intelligent and learn complex cues and behaviors. Other invertebrates that look superficially like insects, prawns for example, have quite different life cycles and lifespans and often do exhibit signs of pain / sentience.

Basically pain/sentience emerge when there is a reason for the organism to protect the body from damage, and does not evolve (or is subsequently lost!) when there are more important short term goals. One wonders, for example, whether salmon experience pain when they fling their bodies up rivers, over and onto rocks, damaging them horribly in the process: all for purposes of spawning.

  > Iteration itself isn’t inherently bad. It’s just that the objective
  > function usually isn’t what we want from a scientific perspective.

Science: Is treatment A better than treatment B?

Engineering: I would like to make a better treatment B.

Iteration is harmful for the first goal yet essential for the second. I work in an applied science/engineering field where both perspectives exist. (and are necessary!) Which specific path is taken for any given experiment or analysis will depends on which goal one is trying to achieve. Conflict will sometimes arise when it's not clear which of these two objectives is the important one.

A reason it's so hard to learn to produce these novel sounds, I would argue, is because the learner literally cannot hear the differences at first. It's only after learning (i.e. when the qualia starts to change) that production of the new sounds becomes possible.

One can think of other similar examples in the context of expert performance: a sonar operator can hear sounds in his headphones that most (at first) cannot; an artist can distinguish colors that the novice cannot, etc.

If you buy this argument, that learning can affect perception/qualia, then it's a fairly small leap to imagine how qualia itself might also be learned ex nihilo.

What the Kalman filter equations don't tell you, is how to estimate the parameters of such an observation model. You either have to write it down from first principles, estimate it from "fully observed" data (if you have such a luxury), estimate it (up to a rotation) using expectation maximization (EM), or guess.

  By three methods we may learn wisdom: First, by reflection, which is noblest; 
  Second, by imitation, which is easiest; and third by experience, which is the 
  bitterest.