In Personal Knowledge chapter 11 Polanyi elaborates his critique of the "physico-chemical investigations of a machine" by carefully dividing reasons and causes as types of explanations:

Since rules of rightness cannot account for failures, and reasons for doing something can only be given within the context of rules of rightness, it follows that there can be no reasons (int this sense) for a context and to describe the origins of failures invariably as their causes. We can say then that physico-chemical investigations of a machine, carried out with a bearing on its operational principles, can elucidate both the conditions for their success and the causes of their failure. It would be wrong to speak of establishing the physical and chemical 'causes' of success, for the success of a machine is defined by its operational principles, which are not specifiable in physico-chemical terms. If a stratagem succeeds, it does so in accordance with its own premeditated internal reasons; if it fails, this is due to unforeseen external causes.

(Personal Knowledge, 1962, p.349; emphasize mine)

This is preceded by:

We have a solid tangible inanimate object before us... Then let a team of physicists and chemists inspect the object... They will describe the clock precisely in every particular, and in addition, will predict all its possible future configurations. Yet they will never be able to tell us that it is a clock... no physical or chemical observations of clocks will be of any use to a clockmaker, unless such observations are related to the operational principles of a clock, as conditions for their success or causes of their breakdown... Some physical and chemical characteristics of a machine... will be of interest in themselves on certain occasions... But this is about as much as the scientific study of a machine can achieve when pursued in itself, without reference to the principles by which the machine performs its purpose.

(Personal Knowledge, 1962, p.347-348; again emphasize mine)

So basically the gist of Polanyi's criticism is, by invoking the 'rules of rightness', that physical and chemical inquiry cannot account for the 'rightness' (success or failure) of the operation of the machine, but only to the rightness of its intrinsic functions.

But if we, sticking to his clock example, take all the rules of the intrinsic functions of the clock, can we not deduce that if the large hand moves every minute, that function shows us the passing of minutes? Perhaps we won't call this machinery "clock", but simply describe it by its functions "that which tells us the passing of time" (or even simpler "the passing of minutes and hours"). How, then, can we say that the physical investigation cannot account for the operation of the machine? Does Polanyi (if so then bizarrely) disregard the use of logic (even as simple as inference) in these investigations?

Or did I misunderstood Polanyi's reasoning?

  • 2
    I think his argument is valid in that one can not extract teleology out of natural causes, but it is not an argument against physicalism as typically understood. Physicalists themselves happily disclaim teleology as an anthropomorphic manner of speaking, and nature for them does not care what is or is not "of any use to a clockmaker" or what counts as "success". It could, perhaps, be construed as an ethical argument about limitations of natural science.
    – Conifold
    Aug 14, 2020 at 21:51
  • 1
    @Conifold I indeed had difficulty defining in proper terms what he's going out against here. It's in a certain way the physicalist' common premise of reductionism, I presume, and the alleged success of explaining life (if we're expanding it to arguments presented later on in his book) according to the reduction to physico-chemical explanation. But in response to what you say about teleology - even if nature does not care for "the use of x", it wouldn't prevent them from claiming that they can deduce from their investigation what the use of x is. Aug 15, 2020 at 13:14
  • I do not think this is what "reductionism" means, it looks to me like he is pretty much granting the premise of reductionism with "describe the clock precisely in every particular, and in addition, will predict all its possible future configurations" based on physics and chemistry, i.e. by reducing clock to them. Purposes can, of course, be added by hand, and means-end reasoning then performed, but perhaps his point is that adding them goes beyond naturalistic explanations. Peirce makes a somewhat similar point in his semiotic, I think.
    – Conifold
    Aug 16, 2020 at 4:09
  • 1
    A clock does not have any physical functions as a physical system, and neither does a book, it is just a stack of inked paper. As Putnam showed, one can interpret any rock as a Turing machine, but that requires an interpreter. To me, functions "emerging" from causes sounds like a category mistake, it is like redness "emerging" out of a statue. If this boundless emergence is accepted then anything at all can be "explained" by claiming it.
    – Conifold
    Aug 17, 2020 at 4:33
  • 1
    @Conifold The way I read Polanyi here and elsewhere is sticking a finger into the obvious: That the pragmatic and teleological reality has become so much flesh and bone (because it is our primary reality) that proponents of physicalism do not realise the boundaries of their own ability to explain (all of) reality. In fact, the question is a good example for such cases. And even evolutionary biology has to value reproduction of life as success if it is to tell functional from malfunctional mutations. He argues Humean along descriptive vs. normative, that's basically all that is to it.
    – Philip Klöcking
    Sep 15, 2020 at 21:48

2 Answers 2


Yes, it is flawed. Polyani is saying that examining the physics of a clock can never tell you the principles of its operation or its ability to accurately keep time.

This is false.

If we know the physics of the clock (or any other machine) at the level of atoms, we can deduce from this the simplified principles of the clock's operation. The minimum description length of the clock's atomic-level structure will be obtained from a description of the clock that invokes the principles of its operation, e.g. the rates of motion of its hands. To give this minimum description, you would first "paint with a broad brush" by describing the macroscopic shapes of the parts of the clock and their equations of motion, and then you would "paint with a fine brush" by describing how the particular atoms in this particular clock vary from these ideal shapes and principles of operation.

It is similar to the way a statistical analysis of a cloud of data points can be performed. Before you analyze the data, you just have a bunch of data points. But you can look at what trends appear in the data, and fit a line or curve to the data. The curve was not present in the data, but it was implicit in the data. A minimum description of the data could be given by specifying the curve and then specifying the residuals. The property that this curve (+ residuals) is a minimum description of the data, and some other description is not, is a property inherent in the data, not in the person analyzing it.

Another way to look at it is by examining what interpretations the atoms of the clock admit. A clockmaker can interpret the atoms as composing a clock, consisting of specified macroscopic parts in a specified arrangement with a specified evolution over time, with the second hand tracking the seconds and so forth. The clockmaker cannot give the same interpretation to a chair, or to a thermometer; the physics of the chair or thermometer do not match up with the parts of a clock. Thus, the "Property-of-Admitting-a-Clock-Interpretation" (abbreviated PACI) is a function of the arrangement of atoms; the fact that atoms are arranged this way and not some other way gives them PACI. No human is needed to actually assign the clock interpretation to the atoms, and they have PACI even if no human does; we are speaking of an abstract property, PACI, purely of arrangements of atoms.

  • See Conifold's first and last and Philip's comments in the OP. Sep 10, 2021 at 19:36
  • @YechiamWeiss I have. Do you argue that something they said refutes something I said? I would say it's the other way around; PACI shows how the principles of the clock's operation, as described by a clockmaker, arise from just the arrangement of atoms of the clock, without the actual need for the clockmaker. This refutes claims that the principles of the clock's operation go beyond what nature provides.
    – causative
    Sep 10, 2021 at 19:53
  • the point is that PACI is a derivative of a teleological assumption that has to happen outside of the interpretation of the atoms. How can you conclude whether this arrangement is a "proper" one and a different arrangement is not, based on the arrangements themselves alone? I really like the statistical analysis example, because that is one of the most profound examples - you couldn't determine if a set of data is pointing towards a "good" valuation or "bad" valuation without the necessary personal knowledge to make such distinction. Sep 11, 2021 at 9:03
  • In other words - who says that a "malfunction clock" is truly malfunction? Perhaps the seconds hand is supposed to skip 2 seconds at a time. The only one to make that judgment is the clockmaker. Sep 11, 2021 at 9:07
  • 1
    @YechiamWeiss Well, I haven't addressed the valuation of the operational principles here; I have said only that the operational principles themselves (PACI) are a property of the arrangement of atoms in the clock, independent of any clockmaker. Whether we value PACI or not, or value the "property of admitting a 2-second-skipping-clock interpretation" (PA2CI) is another matter.
    – causative
    Sep 11, 2021 at 18:35

Polanyi appears to be arguing several points:

  • Purpose, teleology, has to be in the intention of a builder, and is discerned in the mind of an observer, it is never intrinsic to an object.
  • The logic of a design or mechanism -- how it is constructed to operate and interact with itself and the world, its "operational principles", are a logic or informational feature (IE non-material) that also is not intrinsic to the material of the design or mechanism.
  • "Causes of Success" or failure, are likewise non-material, and are tied to the "operational principles", and "external causes"

Polanyi appears to be arguing a valid point, but does so very unclearly, as it appears he himself is confused about what he is saying.

Purposes, teleology, are abstract objects. They have no mass, nor location, and can exist in between periods of apprehension and consideration by consciousness. Design logic, the "operational principles", are once more abstractions, and again have no mass, location, nor fixed time. Whether something has "success" or "failure" is a judgement call, and has to reference teleology, and the specifics of a situation, and therefore is significantly dependent on abstractions. Causation itself, has neither mass nor location, nor is ti limited in time.

These abstractions that an investigation of a machine must make use of -- are not plausibly material, and serve as a refutation of materialism. This is a valid point, which Polanyi seems to be reaching for, but does not clearly articulate.

They do NOT establish that MINDS are non-material. Hence, they are not actually a valid critique of a physicalist view of mind.

Polanyi is also incorrect in at least one aspect of the third bullet above -- physical investigations of mechanisms, to figure out why they are not operating properly, is a central feature of all mechanical design projects, is a key skill for every repair project, and is the central activity that most medicine consists of. These activities, like almost all in the physical world, rely absolutely upon all sorts of abstract phenomenon, such as causation, purpose, success/failure criteria, etc. This reliance is so intrinsic to pretty much every physical object or physical process, that most of us do not even think about it. But this process IS basically physical PLUS abstract, as opposed to NOT being physical as Polyani asserts.

The philosophers who most clearly articulate that abstractions are non-physical, and thus break the assumptions behind physicalism, would be Frege, Quine, and Popper. Frege and Popper also note thought has no mass nor location, hence is also non-material, ans assert a pluralist ontology as the only way to address this irreducibiltiy between three types of things in our world. All three make the abstraction objection to materialism far more clearly than Polyani does, and do not confuse it with the experience/qualia objection.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .