Epistemic status: exploratory

If the Earth moves, why can’t we feel it? And why do object dropped from high places fall vertically instead of to the side (since the Earth is supposed to have moved)? These are not stupid questions. If you were first introduced to the idea of the Earth that moves around a background of thinking that it was immobile, they would surely be a point of contention. And were for many thinkers the reaction to Copernicus's physics, and to the Greek thinkers he was inspired by.

So why shouldn't that refute heliocentrism? Galileo argued, in a move that has been vindicated by the last 400 years of success in physics, that there’s a distinction between absolute movement and relative movement, and that we can only detect the latter. So the stone falls horizontally because it is the only relative movement with regard to us and the Earth; it's also following Earth on its trajectory, but we do too. And we don't feel the movement of the Earth because it's not a movement relative to us but a movement with us.

Paul Feyerabend, the anarchist philosopher of science, describes this move as Galileo replacing what Feyerabend calls a natural interpretation with another: the intuitive, naive correspondence between a movement and its observation is replaced with only the distinguishability of relative movements.

After reading about it, I now see this idea of switching natural interpretations in many fascinating places in the history of science. This is a short post expanding on it and analyzing the concept through a few of these examples.

Revealing Natural Interpretations

Here’s how Feyerabend introduces natural interpretations in Against Method:

(All quotes are from Against Method)

Making the additional simplifying assumption, we can now distinguish between sensations and those 'mental operations which follow so closely upon the senses', and which are so firmly connected with their reactions that a separation is difficult to achieve. Considering the origin and the effect of such operations, I shall call them natural interpretations.

So they are not the observation themselves, but the almost instantaneous and unconscious implications that are drawn from the observations. Which can come from internalizing a model of the world — a theory of physics for example.

Feyerabend uses this notion to discuss what Galileo does in his dialogue, and how he deals with some of the main criticisms of Copernicus’s view — namely the horizontal fall of a stone.

According to the Copernican view as presupposed in the tower argument the motion of a falling stone should be 'mixed straight-and-circular'. By the 'motion of the stone' is meant not its motion relative to some visible mark in the visual field of the observer, or its observed motion, but rather its motion in the solar system or in (absolute) space, i.e. its real motion. The familiar facts appealed to in the argument present a different kind of motion, a simple vertical motion. This refutes the Copernican hypothesis only if the concept of motion that occurs in the observation statement is the same as the concept of motion that occurs in the Copernican prediction. The observation statement 'the stone is falling straight down' must, therefore, refer to a movement in (absolute) space. It must refer to a real motion.

[...] Producing an observation statement, then, consists of two very different psychological events: (1) a clear and unambiguous sensation and (2) a clear and unambiguous connection between this sensation and parts of a language. This is the way in which the sensation is made to speak. Do the sensations in the above argument speak the language of real motion?

They speak the language of real motion in the context of 17th century everyday thought. At least, this is what Galileo tells us. He tells us that the everyday thinking of the time assumes the 'operative' character of all motion, or, to use well-known philosophical terms, it assumes a naive realism with respect to motion: except for occasional and unavoidable illusions, apparent motion is identical with real (absolute) motion. Of course, this distinction is not explicitly drawn. One does not first distinguish the apparent motion from the real motion and then connect the two by a correspondence rule. One rather describes, perceives, acts towards motion as if it were already the real thing. [...] However, there are paradigmatic cases in which it is psychologically very difficult, if not plainly impossible, to admit deception. It is from these paradigmatic cases, and not from the exceptions, that naive realism with respect to motion derives its strength. [...] The motion of the stone in the tower argument, or the alleged motion of the earth, is such a paradigmatic case. How could one possibly be unaware of the swift motion of a large bulk of matter such as the earth is supposed to be! How could one possibly be unaware of the fact that the falling stone traces a vastly extended trajectory through space!

By introducing the relativity of motion and the law of circular inertia, Galileo resolves that discrepancy by making the explicit separation between absolute and relative motion: he replaces one natural interpretation with another, more theoretical and abstract.

Galileo's first step, in his joint examination of the Copernican doctrine and of a familiar but hidden natural interpretation, consists therefore in replacing the latter by a different interpretation. In other words, he introduces a new observation language.

What I like about this example is that it finally clarified the theory-ladenness of observation to me. The latter is Kuhn (and Feyerabend actually) claim that observation depends on theories, and so there are no objective facts that can be used to judge all theories against. I never got what Kuhn was talking about, until I read the passage above. It’s not just that the choice of which observation to make depends on the theory; it’s that the natural interpretation influences almost automatically what an observation means or should look like, in a way that can be hard to realize except from outside its pull.

Consideration of all these circumstances, of observation terms, sensory core, auxiliary sciences, background speculation, suggest that a theory may be inconsistent with the evidence, not because it is incorrect, but because the evidence is contaminated. The theory is threatened because the evidence either contains unanalysed sensations which only partly correspond to external processes, or because it is presented in terms of antiquated views, or because it is evaluated with the help of backward auxiliary subjects. The Copernican theory was in trouble for all these reasons.

More Examples

I see this move in many places in the history of science: a new model/approach is introduced, which explains a lot but contradicts obvious and fundamental observations. Part of the work involved in pushing the new theory is in the switch of natural interpretations.

  • Special and General Relativity remove the absoluteness of time. At first glance, this is incoherent with many previous observations where time behave as if absolute. The switch in natural interpretation introduces time dilation that is too small to be noticed in the previous observation.
  • Evolution introduces the process of natural selection for the generation of new species. At first glance, this is incoherent with the observation that humans don't create new species by breeding (a pointed form of selection). The switch in natural interpretation is that the scale of our human observation is far smaller than the scale of time used by natural selection to make new species and fashion the natural world.[1]
  • Computability introduces what Dennett calls Turing's strange inversion: that to solve problems like mathematically sophisticated human computers, you only needed the composition of simple, trivial, non-understanding components. At first glance, this is incoherent with the observations that solving these problems required comprehension and sophistication from the human computers, and that basic machines and mechanisms were at the time unable to do such feats. The switch in natural interpretation is that the scale of the mechanisms then was far smaller (both in number of components and speed) than the one needed for solving complex problems.[2]
  • Quantum Mechanics gives a whole new way of modelling and manipulating microscopic states and particles. At first (and second and third) glance, this new approach contradicts so many of the previous observations of macroscopic behavior. The switch in natural interpretation introduces decoherence (or your other favorite interpretation).

Generator or Justification?

The examples above seems to separate into two groups: the ones where the switch of natural interpretations is itself a generator of the theory (Relativity, Computability) or if it comes in the defense of the theory (Copernican physics, Evolution).

In the former group, switching to the new natural interpretation is the source of the model and the insights. It's either enshrined into the theory itself (Relativity) or implicit in its direct consequences. This is the case where I have trouble seeing how the new theory/model could have even emerged without the switch of natural interpretations.

On the other hand, for both Galileo's argument about relativity of movement and Evolution's scales, these were apparently downstream of the conception of the theory. I might be wrong because I'm neither an expert in Galileo or Darwin, but my understanding of the situation is that Galileo introduced this new natural interpretation to defend Copernicus, so years after the inception of the theory; and Darwin noticed the problem himself after formulating his initial thoughts on natural selection, and latched on contemporary geological work as evidence that evolution had the time needed.

Quantum Mechanics doesn't really fit into this duality (pun intended). The new natural interpretation preceded and caused to some extent the theory, but the full-fledged formulation required the natural interpretation for its defense and there was (and still is) debates about this interpretation.

When Are You Allowed to Replace Natural Interpretation?

That being said, it’s clear that we can’t just always replace a natural interpretation with another. If the move was always allowed, then we could fit any theory whatsoever to the evidence.

All these switch eventually become testable (as they pay rent), yet it might be years or even centuries before the whole apparatus necessary for checking them exists. In the mean time, how can we get early evidence on the productivity of such switches?

Here the two categories above might help: the natural interpretation will be judged differently if it is a generator or a justification. Notably, the power of a theory to explain its object is evidence for the validity or at least relevance of a generator switch in natural interpretation, but not for a justification one.

In additional, simplicity sounds like it plays a big role in making productive replacement. All the examples above include a fundamental but simple replacement of natural interpretation (which don't hardcode the desired conclusion), a switch that does a lot for predicting phenomena and dealing with the apparent incoherence with little input. Even in the original case of Galileo, it doesn’t take much to remove the incoherence.

This first example also points that the new natural interpretation could be justified by specific instances where it's already accepted. The relativity of movement is justified by using intuitive examples like people on boats. It's not necessary though, since both Quantum Mechanics and Relativity don’t have this feature (their new natural interpretation is not intuitive at all, even if it makes sense on reflection[3]).

  1. ^

    This requires Earth to be old enough for natural selection to create the current biological world, which was indeed at the center of debates between Darwin and Kelvin.

  2. ^

    In a nice analogy to the other case of strange inversion (Darwin), many people still have trouble accepting this switch in natural interpretation.

  3. ^

    I'm quite confident of this for Quantum Mechanics, but good old Albert might disagree with me on Relativity.

New to LessWrong?

New Comment