Some quotes and thoughts.
The following line seems to me to be a refutation of Merton's norm of disinterestedness. I don't know if Polanyi read Merton, but it's possible since this work was originally published about 16 years after Merton published “The Normative Structure of Science.”
I start by rejecting the ideal of scientific detachment (p. vii).
The goal of the book:
I want to establish an alternative ideal of knowledge, quite generally (p. vii).
The argument that Polanyi will make and build relates to. The Gestalt is strong here:
Skilful [sic] knowing and doing is performed by subordinating a set of particulars, as clues or tools, to the shaping of a skilful achievement, whether practical or theoretical. We may then be sait to become 'subsidiarily aware' of these particulars within our 'focal awareness' of the coherent entity that we achieve (p. vii).
Here he makes it clear that that he's centering the personal but also anticipating attacks against subjectivity:
Such is the personal participation of the knower in all acts of understanding. But this does not make our understanding subjective (p. vii).
Notes and thoughts on the four chapters in Part One.
This would imply that, of the two forms of knowledge [theoretical knowledge and sensory-based knowledge], we should consider as more objective that which relies to a greater measure on theory rather than on more immediate sensory experience (p. 4).
He's really addressing the physical sciences here, and even though I'd have to reflect on this, I think it would be applicable to the social sciences, too:
Indeed, all theory may be regarded as a kind of map extended over space and time (p. 4).
Objectivity and theory:
A theory on which I rely is therefore objective knowledge in so far as it is not I, but the theory, which is proved right or wrong when I use such knowledge (p. 4).
I'm not sold yet on this, but he's basically arguing that there is a rationality that exists in nature, and that theorizing is the ability to see that rationality that exists in nature. Theorizing (“the theory being placed like a screen between our senses and the things of which our senses”) is mapping work.
We abandon the cruder anthropocentrism of our senses—but only in favour of a more ambitious anthropocentrism of our reason (pp. 4-5).
Another description of the work of theory:
We accept it in the hope of making contact with reality; so that, being really true, our theory may yet show forth its truth through future centuries in ways undreamed of by its authors (p. 5).
This is what he's arguing, and discovery will play an important role in this chapter:
To say that discovery of objective truth in science consists in the apprehension of a rationality which commands our respect and arouses our contemplative admiration; that such discovery, while using the experience of our senses as clues, transcends this experience by embracing the vision of a reality beyond the impressions of our senses, a vision which speaks for itself in guiding us to an even deeper understanding of reality…. the power of science to make contact with reality in nature by recognizing what is rational in nature (pp. 5-6).
So I see now. I was thinking that the subsidiarily aware particulars and the focal awareness concepts applied mostly to skills, but even Polanyi says it applies to theory, so I should have gotten it. But he's applying this here, too. Sensory experiences are the subsidiarily aware particulars and not the focal points. When they are the focal points, this allows reductivism and positivism.
I need to explore this further because it has implications for my work on open science. In short, DC Miller was doing ether research, repeating the Michelson-Morley experiment, and his results would have proved Einstein's theory of relativity wrong if the results were true. Even though Miller's work was adding experimental evidence that relativity theory was incorrect, it was largely disregarded because it was less theoretically rational (in the sense that Polanyi talks about theory above) than Einstein's theory:
The experience of D. C. Miller demonstrates quite plainly the hollowness of the assertion that science is simply based on experiments which anybody can repeat at will. It shows that any critical verification of a scientific statement requires the same powers for recognizing rationality in nature as does the process of scientific discovery [my emphasis], even though it exercises these at a lower level (p. 13).
As a result we are given an account of the scientific method which, having left out the process of discovery on the grounds that it follows no definite method overlooks the process of verification as well, by referring only to examples where no real verification takes place (pp. 13-14).
The main motivation of the book, per Polanyi, and in his “re-evaluation of science” (p. 18):
The purpose of this book is to show that complete objectivity as usually attributed to the exact sciences is a delusion and is in fact a false ideal (p. 18).
I think this is the basic premise of reproducibility arguments in open science, and it's something that Polanyi disagrees with about science itself:
The avowed purpose of the exact sciences is to establish complete intellectual control over experience in terms of precise rules which can be formally set out and empirically tested (p. 18).
Probability introduces random errors and human participation introduces individual biases:
Even the most strictly mechanized procedure leaves something to personal skill in the exercise of which an individual bias may enter (p. 19).
The data that we collect has these personal biases. Discusses Maskeleyne (see more of the story here: https://www.sciencedirect.com/topics/social-sciences/astronomy) and the discrepancies between Maskeleyne's “recording of the passage of stars” and that of his assistant's, Kinnebrook. According to the above URL, this effect is now called “coincidence timing”:
There are consistent individual differences in the degree of “accuracy” in estimating the time of coincidence of the spot and the line (see URL above).
Polanyi concludes from this example and argument that:
We must always assume, therefore, that some trace of a hidden personal bias may systematically affect the result of a series of readings (p. 20).
We have here an essential personal participation of the scientist even in the most exact operation of science (p. 20).
Polanyi makes another important point: scientists do not trash a theory when an anomaly appears. The instead set it aside in the hopes that something down the line will explain the anomaly within the framework of the theory. I think this is how modern physics works, too, with, for example, regards to theories such as string theory.
Here's a really good point that is true just by definition of probability:
Probability statements can never be strictly contradicted by experience … (p. 21).
I like this story about the nature of probable statements:
There is a story of a dog-owner who prided himself on the perfect training of his pet. Whenever he called 'Here! will you come or not!' the dog invariably either came or not. This is exactly how electrons behave when controlled by probability (p. 21).
I think this is helpful way to think about the demarcation issue in science (and pseudo-science) that doesn't rest on Popperian contradiction:
There is, of course, an important sense in which a probability statement can be controverted (though not contradicted) [my emphasis] by the events. If the expectations based on a statement of probability are repeatedly disappointed and the ensuing events appear to have been correspondingly improbable in the light of the anterior statement of their probability, we shall begin to suspect the correctness of this statement. The process of deciding that a certain statistical statement is untenable had indeed been systematically developed by Sir Ronald Fisher in his famous treatise, The Design of Experiments (p. 22).
As a aside, what follows is a nice exposition of Fisher's probability theory.
And then on the personal interpretation of a result by chance or by statistical significance:
This tells us that on the hypothesis (which Fisher calls the null-hypothesis) that the differences in the heights of self-fertilized and cross-fertilized plants are purely accidental, the probability for our actually observed sample having occurred was less than 5 percent. Such a statement entitles us to be surprised by the observed result to the same extent as we would be if we drew a black ball from a sack supposed to contain only 5 black balls in a hundred otherwise indistinguishable balls (p. 23).
Although I like the above example, I think it's a little misleading, too, since the example uses the same number (5%) for effect size (5 out of 100) as it does for experimental probability. That is to say, the probability of an effect observed in an experiment is different than the probability of an experiment being statistically significant rather than random. We can say that in 5% of the cases we'd see such an effect, if we were able to reject the null hypothesis, but then the effect could be 0.75, or something like that. For example, we can say that in a linear regression that has p = 0.05, we might expect a doubling of Y for every unit of X in 1 out of 20 cases.
I think this section is addressing the role of propositions in knowledge. He is making a distinction between the the probability of a statement and of an event:
I conclude, therefore, that in so far as we arrive at probability statements on the lines of the statistical method illustrated by Darwin's or Rhine's investigations, or as made every day about the toss of a coin, these are statements about probable events and not probable statements about events (p. 25).
I think Polanyi here anticipates the statistical 95% problem here:
We may then follow R. A. Fisher in trying to set a specific limit to the improbability which we are prepared to countenance before abandoning the law in question. But since no such rule can be firmly upheld [my emphasis], it merely expresses a personal judgment that is subject to similar variations of confidence as the original probability statement, the validity of which it was intended to test (p. 26).
On the personal nature of (scientific) assertions:
In routine observations, unobstructed by opposition and unworried by doubts, these passions are dormant but not absent; no sincere assertion of fact is essentially unaccompanied by feelings of intellectual satisfaction or of a persuasive desire and a sense of personal responsibility (p. 27).
Statements are not just true/false logical statements but have modality (some indication of its truthiness) without which the statement is meaningless:
If language is to denote speech it must reflect the fact that we never say anything that has not a definite impassioned quality. It should be clear from the modality of a sentence whether it is a question, a command, an invective, a complaint or an allegation of fact. Since an unasserted declaratory sentence could not stand for an allegation of fact, its modality would be unspecified and could therefore denote no spoken sentence (p. 27).
Similarly, I suggest, a sentence itself has only vague significance until supplemented by the symbol defining its modality (p. 28).
But referring back to the previous section:
[it] is not that I make an assertion but that I commit myself to it; it is not the act of my uttering a sentence p that I express .. but the fact that I believe what the sentence p says (p. 28).
This all seems to me to be very Bayesian even though he's referring back to Fisher:
it is of the essence of the scientific method to select for verification hypotheses having a high chance of being true. To select good questions for investigation is the mark of scientific talent, and any theory of inductive inference in which this talent plays no part is a Hamlet without the prince. The same holds for the process of verification. Things are not labelled 'evidence' in nature, but are evidence only to the extent to which they are accepted as such by us as observers (p. 30).
It is a travesty of the scientific method to conceive of it as a process which depends on the speed of accumulating evidence presenting itself automatically in respect to hypotheses selected at random (p. 30).
Lots here that have implications for open science. Refer back to it, but here's something:
Maxims are rules, the correct application of which is part of the art which they govern…. Maxims cannot be understood, still less applied by anyone not already possessing a good practical knowledge of the art…. Another person may use my scientific maxims for the guidance of his inductive inference and yet come to quite different conclusions. It is owing to this manifest ambiguity that maxims can function only–as I have said–within a framework of personal judgment. Once we have accepted our commitment to personal knowledge, we can also face up to the fact that there exist rules which are useful only within the operation of our personal knowing … (p. 31).
No comment on this section now. May come back to it.
Trying to figure out this section. I think the main gist of this section is that there is an ordering to nature–nature is rational. Maybe?
But, as we have seen in the chapter on Objectivity, man has the power to establish real patterns in nature, the reality of which is manifested by the fact that their future implications extend indefinitely beyond the experience which they were originally known to control (p. 37).
There is no chance without order and vice versa:
When I say that an event is governed by chance, I deny that it is governed by order. Any numerical assessment of the probability that a certain event has occurred by chance can be made only with a veiw to the alternative possibility of its being governed by a particular pattern of orderliness (p. 33).
Also, to say that the difference between chance and order is a difference in where we place our attention:
that the appraisal of order is an act of personal knowledge (p. 36).
I like the way that this is framed because it's insightful. Absolute certainty means no randomness:
Statements of probability can be made about random systems and about significantly ordered systems in so far as these are affected by interaction with random systems (p. 37).
Probability statements are therefore always based on an anterior knowledge of randomness (p. 38).
Even though we identify randomness by not identifying patterns:
… by asserting my belief that random systems exist and can be recognized as such, though it is logically impossible to give any precise definition of randomness (p. 38).
Polanyi's “well-know fact” as an assumption here:
that the aim of a skilful performance is achieved by the observance of a set of rules which are not knowing as such to the person following them (p. 49).
But does this tell us exactly how to ride a bicycle? No. You obviously cannot adjust the curvature of your bicycle's path in proportion to the ratio of your unbalance over the square of your speed; and if you could you would fall of the machine, for there are a number of other factors to be taken into account in practice which are left out in the formulation of this rule. Rules of art can be useful, but they do not determine the practice of an art; they are maxims, which can serve as a guide to an art only if they can be integrated into the practical knowledge of the art. They cannot replace this knowledge (p. 50).
This begs the question – what rules of art (or maxims) do scientists follow that serve as guides to the practice of science?
This gets at the heart of my open science dilemma that I'm positing because it challenges the rationality. That is, if the practice of science depends on skills that cannot be fully accounted for, then the practice of science depends on skills that cannot be codified. If something cannot be codified, then this places that thing in the realm of the non-observable, and things in that realm are not scientific.
The fact that skills cannot be fully accounted for in terms of their particulars may lead to serious difficulties in judging whether or not a skilful performance is genuine (p. 50).
I think many of the examples that Polanyi uses are not good examples and are of things that people have overcome in the development of technologies or they are just examples of things that are not of scientific origins. New and better examples should be created.
Also, I'm done with this section but I still am not quite sure what a destructive analysis is or how to do it.
Here's where I think he gets to learning:
An art which cannot be specified in detail cannot be transmitted by prescription, since no prescription for it exists. It can be passed on only by example from master to apprentice. This restricts the range of diffusion to that of personal contacts, and we find accordingly that craftsmanship tends to survive in closely circumscribed local traditions (p. 53).
Here is his bit on science but again followed not by the greatest examples:
Again, while the articulate contents of science are successfully taught all over the world in hundreds of new universities, the unspecifiable art of scientific research has not yet penetrated to many of these (p. 53).
I am not a fan of the wording here or the stated implication/conclusion, but I think there's something to work from here:
To learn by example is to submit to authority. You follow your master because you trust his manner of doing things even when you cannot analyse and account in detail for its effectiveness. By watching the master and emulating his efforts in the presence of his example, the apprentice unconsciously picks up the rules of the art, including those which are not explicitly known to the master himself. These hidden rules can be assimilated only by a person who surrenders himself to that extent uncritically to the imitation of another. A society which wants to preserve a fund of personal knowledge must submit to tradition (p. 53).
Doing is knowing????
What is said of skills applies equally to connoisseurship. The medical diagnostician's skill is as much an art of doing as it is an art of knowing (p. 54).
More on practice:
The large amount of time spent by students of chemistry, biology and medicine in their practical courses shows how greatly these sciences rely on the transmission of skills and connoisseurship from master to apprentice. It offers an impressive demonstration of the extent to which the art of knowing has remained unspecifiable at the very heart of science (p. 55).
This is an important part of tacit knowing theory. He begins this section by stating that what he here writes about is closely related to Gestalt psychology and that he borrows from it, but that it's also different and so he doesn't use it wholesale.
Key terms: subsidiary awareness and focal awareness:
When we use a hammer to drive in a nail, we attend to both nail and hammer, but in a different way. We watch the effect of our strokes on the nail and try to wield the hammer so as to hit the nail most effectively. When we bring down the hammer we do not feel that its handle has struck our palm but that its head has struck the nail. Yet in a sense we are certainly alert to the feeling in our palm and the fingers that hold the hammer. They guide us in handling it effectively, and the degree of attention that we give to the nail is given to the same extent but in a different way to these feelings. The difference may be stated by saying that the latter are not, like the nail, objects of our attention, but instruments of it. They are not watched themselves; we watch something else while keeping in intensely aware of them. I have a subsidiary awareness of the feeling in the palm of my hand which is merged into my focal awareness of my driving the nail (p. 55).
He introduces also the phrases on page 56: knowing how and knowing what.
If our focus tries to the particulars, then we usually fail, like a pianist would if they switched their focus, or become self-conscious, of the specific keys to play in a score.
This introduces another key part of the theory – by becoming focally aware of the particulars, context is destroyed:
This destroys one's sense of the context which alone can smoothly evoke the proper sequence of words, notes, or gestures. Stage fright [ as an example ] is eliminated and fluency recovered if we succeed in casting our mind forward and let it operate with a clear view to the comprehensive activity in which we are primarily interested (p. 56).
The above is another crappy example. He should use more examples from the practice of science.
Another term: logically unspecifiable
We may describe such a performance as logically unspecifiable, for we can show that in a sense the specification of the particulars would logically contradict what is implied in the performance or context in question (p. 56).
This is helpful:
But it is perhaps more appropriate to formulate the contradiction in this case in more general terms, by saying that our attention can hold only one focus at a time and that it would hence be self-contradictory to be both subsidiarily and focally aware of the same particulars at the same time (p. 57).
Regarding meaning, another key concept:
This scheme can be easily reformulated and expanded in terms of meaning. If we discredit the usefulness of a tool, its meaning as a tool is gone. All particulars become meaningless if we lose sight of the pattern which they jointly constitute (p. 57).
In the next paragraph, Polanyi describes another example and refers to objects. I think this term is useful – to shift focus to what we should be only subsidiarily aware of is to make the particulars the object. This is useful in teaching. But to practice, the object is the whole thing. As stated in the quote above, “the patern which they jointly constitute”. This is the collective object.
Key concepts here so far:
Not the point of the section but the phrasing is helpful:
accepts certain meanings by deliberately merging [my emphasis] his awareness of certain particulars into a focal awareness of a whole (p. 57).
In this section, Polanyi discusses two different two types of meanings and two types of wholes. I believe this is an entailment of what I describe above regarding objects but more pointed:
we can easily see that anything that functions effectively within an accredited context has a meaning in that context and that any such context will itself be appreciated as meaningful (p. 58).
Two additional terms:
The tools we use become extensions of our bodies just as the parts of our bodies do:
We accept them existentially by dwelling in them (p. 59).
Okay, I've been critical of Polanyi's use of examples, but here he makes the leap to more relevant ones:
Hammers and probes can be replaced by intellectual tools; think of any interpretive framework and particularly of the formalism in exact sciences (p. 59).
He refers back to his chapter on probability, with reference to the following text. Also, I think Polanyi is scientific epistemology and pyschology as Hume is to logic and epistemology more generally and Kuhn is to sociology of science:
I suggest now that the supposed pre-suppositions of science are so futile because the actual foundations of our scientific beliefs cannot be asserted at all. When we accept a certain set of pre-suppositions and use them in our interpretative framework, we may be said to dwell in them as we do our own body. Their uncritical acceptance for the time being consists in a process of assimilation by which we identify ourselves with them. They are not asserted and cannot be asserted, for asserting can be made only within a framework with which we have identified ourselves for the time being; as they are themselves our ultimate framework, they are essentially inarticulable (p. 60).
On becoming a scientist, but I also gather that this would be true for anyone becoming a part of a community of action or community of practice:
It is by his assimilation of the framework of science that the scientist makes sense of experience. This making sense of experience is a skilful act which impresses the personal participation of the scientist on the resultant knowledge (p. 60).
More phrasing that helps elucidate the theory and here regarding the structure:
The tracing of personal knowledge to its roots in the subsidiary awareness of our body as merged in our focal awareness of external objects [my emphasis], reveals not only the logical structure of personal knowledge but also its dynamic sources (p. 60).
On page 60 – things acquire meaning when they become extensions of ourselves:
In the new scheme which I have just drawn up of the process by which an external thing is given a meaning by which being made to form an extension of ourselves [my emphasis] … (p. 60).
And then on instrumentation – the following is like the Gettier problem, but for Polanyi, knowledge is personal and by personal, embodied:
If a rat accidentally depresses a lever which releases a food pellet it has not used it as a tool; only after the rat has learned to use it for that purpose does the lever become its tool (p. 60).
In addition to the tool as “only one example of the merging of a thing in a whole (or a gestalt) in which it is assigned a subsidiary function and a meaning in respect to something that has our focal attention,” (p. 61), “signs and symbols” can perform this function, too:
Like the tool, the sign or the symbol can be conceived as such only in the eyes of a person who relies on them to achieve or to signify something. This reliance is a personal commitment which is involved in all acts of intelligence by which we integrate some things subsidiarily to the centre of our focal attention (p. 61).
Every act of personal assimilation by which we make a thing form an extension of ourselves through our subsidiary awareness of it, is a commitment of ourselves; a manner of disposing of ourselves (p. 61).
We becoming focally aware:
And again … we become unconscious of the actions by which we achieve our result. This lapse into unconsciousness is accompanied by a newly acquired consciousness of the experiences in question [my emphasis], on the operational plane. It is misleading, therefore, to describe this as the mere result of repetition; it is a structural change achieved by a repeated effort aiming at the instrumentalization of certain things and actions in the service of some purpose (pp. 61-62).
As an example of above: when my wife is knitting, or when I learned to code in R. There's some repetition involved in learning, but it goes beyond that after a while, when it integrates. Think also of the cool Stack Exchange response about using Vim unwell because not grokking vi: Your problem with Vim is that you don't grok vi.
He lays out two kinds of unspecifiability. Not his terminology, but let's call them:
The weak version is when “a set of particulars” that “have subsided into our subsidiary awareness” fade away completely such that we forget about them, and as forgotten, have become unspecifiable (p. 62).
The strong version is related to trial and error, which I should note, is common in science or scientific discovery, and to the process of developing heuristics, which might be a way for the mind to become more efficient in its engagement with trial and error. The tools we use, including our bodies, are used to perform something, and:
If these actions are experienced only subsidiarily, in terms of an achievement to which they contribute, its performance may select fromthem those which the performer fiinds helpful, without ever knowing these as they would appear to him when considered in themselves. This is the usual process of unconscious trial and error by which we feel our way to success and may continue to improve on our success without specifiably knowing how we do it—for we never meet the causes of our success as identifiable things which can be described in terms of classes of which such things are members (p. 62).
Polanyi posits that what distinguishes humans from other intelligent species is “towering superiority of man over the animals is due, paradoxically, to an almost imperceptible advantage in his original, inarticulate faculties” (p. 69).
Thus, our tacit knowing is greater than their tacit knowing.
The situation can be summed up in three points. (1) Man's intellectual superiority is almost entirely due to the use of language. But (2) man's gift of speech cannot itself be due to the use of language and must therefore be due to pre-linguistic advantages. Yes (3) if linguistic clues are excluded, men are found to be only slightly better at solving the kind of problems we set to animals (pp. 69-70).
So it must mean, for Polanyi, that the combination of tacit and explicit is what sets us apart.
The relationship between the tacit, or the informal, and the formal, or the codified. He mixes terms a lot, I think:
Admittedly the scientist's art of knowing, which I have surveyed previously, is on a higher level than the child's or the animals and can be acquired only in conjunction with a knowledge of science as a formal discipline (p. 70).
Let's say that knowledge of science as a formal discipline is gained through instruction and learning, and here we'd lean on the constructivists to flesh that out.
Learning is distinct from instinct:
I suspect for the moment without discussion the usual distinction between the automatic functioning of the organism, including its instinctive performances, and the higher forms of behaviour not specifically included in the animal's native repertoire. Such behaviour will be called learning in which term I shall include also acts of problem solving. Learning will be regarded as a sign of intelligence, in contrast to the functioning of internal organs or to instinctive performances which will be classed as sub-intelligent (p. 71).
It seems like Polanyi will eventually reject behaviorism, but I'm not sure yet. On p. 71-72, he cites Skinner's work as important and describe some of it. But my reading of Polanyi matches what the Gestalt psychologists propose, and they are very against Skinner's behaviorism. As a reminder, see this article, and specifically, page 88:
Rock, I., & Palmer, S. (1990). The Legacy of Gestalt Psychology on JSTOR. Scientific American, 263(6), 84–91. https://doi.org/10.2307/24997014
Yes – I think he begins to argue against behaviorism on the next couple of pages in his discussion of Pavlov and of perception and of attention.
From Rock & Palmer's (1990) article, see bottom, right column on p. 88 and on.
Piaget beginning on page 74.
Going back to the beginning of this section to outline properly Type A, B, and C learning:
Modes of learning belong to three classes. Type A = motility. Type B = sentience. Type C = motility and sentience.
Type A. Trick learning. Motility. Exemplified by Skinner's behaviorism (pp. 71-72).
Type B. Sign-learning: Sentience. Informed by Pavlov but not entirely. Believed Pavlov misinterpreted his experiments. In his disagreement of Pavlov's interpretation of his own experiments, he writes:
… in contrast to Pavlov's description of the process, that in sign-learning the animal is taught to expect an event by recognizing a sign foretelling the event (p. 72).
For Pavlov, the sign and the event equate each other. So the dog equates the sign with food. But for Polanyi, the dog understands the sign as a signal that food is coming: “the animal learning a new sign reorganizes its sensory field by establishing in it a valid and useful coherence between a sign and the event signified by it” (p. 73).
Type C: Latent learning. Inferences can be drawn. Interpretative frameworks are built.
Type C has been described as latent learning, to suggest that in such cases the animal learns something which it can intelligently manifest in more numerous and less predictable ways than the lessons of trick or sign learning (p. 74).
For type A the heuristic act is a contriving; for B an observing; for C an understanding. The routine acts are: for A, the repeating of the trick, for B, the continued responding to a sign, and for C, the solving of a routine problem (p. 76).
Bringing it together:
Our three types of animal learning are primordial forms of three faculties more highly developed in man. Trick-learning may be regarded as an act of invention; sign-learning as an act of observation; latent learning as an act of interpretation. The use of language [note: the explicit, the formal] develops each of these faculties into a distinctive science to which the other two contribute subsidiarily (p. 76).
The inarticulate - articulate:
The transition from inarticulate learning of type C to its articulate counterpart (which I have called Interpretation), has been traced in Piaget's work on the genesis of disciplined thought in children. Eventually, the operational rules implicitly governing the intelligent behaviour of the child growing up to adolescence will comprise a system of logic … (p. 76).