Tacit Knowing and the Challenge to Science

There are a number of ways to read Polanyi. For example, there are some historical and political angles to The Tacit Dimension that relate to the pursuit of science and knowledge and to restrictions on freedom by the state or by other entities to pursue those activities. Polanyi also worked during a period of the 20th century that saw great upheaval, massive wars, and a rise in totalitarianism that in some ways mirrors the kind of nationalism and demagoguery that is showing its face today across the world. It would be interesting to explore these aspects of Polanyi’s work, and especially reasons for some of the influence Polanyi’s ideas have had on some neoliberal thinkers, but for now, let’s simply bracket out those aspects and focus on what the knowledge management (KM) literature has taken off with—the idea that “we can know more than what we can tell.”

Each of us will have our own interests in where KM intersects with our areas of pursuit. For me, it is hard to escape how tacit knowing impacts the pursuit of science and, inclusive of that, scholarly communication and the conduct of scientific discourse. And this is because, unlike some skeptics in the scholarly communication literature, I hold the view that science is only as successful as the scholarly publishing system. That is, our primary way to communicate scientific knowledge is through the written text, the documentation, and if our system for disseminating that documentation is unhealthy, then our pursuit of science, globally and writ-large, suffers.

With this in mind, we can relate the implications attached to tacit knowing to some modern developments. There is an increasing number of researchers, scientists, and librarians who actively pursue a thing called open science. Some of the arguments put forth in favor of a process that opens science includes claims that an open science results in a more efficient and productive system that takes advantage of web and internet technologies, is able to operate with better transparency, is better at re-using other scientists’ data, and is better at attributing other scientists’ work. In some ways, we might say that an open science is the full, or nearly full, realization of the inherent norms of scientific practice.

Within this rhetoric among the proponents of open science lies the idea that an open science results in a science that is more scientific. The key idea here is that science practiced openly will be more available for critique, review, judgment, and so forth and as such, its epistemological claims can more easily be falsified or verified via, usually, reproducibility and replication tests. Thus, and at the very least, there are ideas inherent in the open science movement that imply that science is not science unless it is open and it is not true unless it is reproducible, in a very mechanical sense. But, these are questionable and testable claims, which demand a path for moving forward—a path that requires of us to test the validity of open science and the premises upon which rest our ability to document and codify the processes involved in the conduct of scientific pursuits. That is, to say that open science is better science requires us to test that claim, scientifically.

Thus, there are significant epistemological and also commensurability issues that are simply not acknowledged by the open science community. Pertinent to this is one basic issue: the criterion of demarcation. That is, what criterion do we use to identify science from what is not science (or pseudoscience)? For Polanyi, if “we can know more than we can tell” means there are limits to what we can communicate, and open science is about being better at communicating scientific work, then even under a better model of scholarly and scientific communication there will always be an upper bound limit on what can be falsified, verified, or reproduced via scientific documentation. This means there will never be a complete guarantee that scientific claims can be trusted via the scholarly communication system. There is always something intrinsic in our scientific knowing that is beyond what we can tell.

What is also not fully acknowledged are other ideas about the demarcation of science—issues related to problem-solving (ala Kuhn) and problem-finding (ala Merton). Here we can refer to, I think, one of the best passages (pp. 64-66) of Polanyi’s book, the ending of which he writes:

Thus the scientific interest—or scientific value—of a contribution is formed by three factors: its exactitude, its systematic importance, and the intrinsic interest of its subject matter (p. 66).

In other words, Polanyi argues that science is demarcated not just by its truthfulness, its coherence, but also by whether it is interesting in the right theoretical way. This brings us back to Merton’s statement about the transmission of not just scientific knowing to the next generation of scientists, but also of scientific taste. Let me re-quote Merton:

The role of outstanding scientists in influencing younger associates is repeatedly emphasized in the interviews with [Nobel] laureates. Almost invariably they lay great emphasis on the importance of problem-finding, not only problem-solving. They uniformly express the strong conviction that what matters most in their work is a developing sense of taste, of judgment, in seizing upon problems that are of fundamental importance (p. 453).

I am not a skeptic about scientific documentation, but tacit knowing raises very interesting and serious challenges about the conduct and dissemination of science, and as well as any kind of knowledge that must be taught and passed from one generation to the next, from teacher to student, or from co-worker to co-worker. This is the task that Polanyi has placed before us.


Emergence and Tacit Knowing

There is a tendency among some or among many, depending on who you count, to reduce wholes to their parts when explaining the wholes. We see this kind of thing when, for example, some scientists or philosophers seek to reduce human consciousness to a specific physical location in the brain (see scientism, more generally). This kind of thinking has extraordinary implications, for if all higher order things (e.g., people) can be reduced to their lower level, basic physical parts (i.e., the wholes are merely summative of the physical aspects that comprise them), then all problems at the whole can be addressed simply by attending to the parts, especially the tangible, mechanistic parts. But is that necessarily true? For example, can all mental issues be addressed simply by attending to the physical processes or components in the brain, or to borrow an example from Polanyi, can we fix someone’s grammar simply by fixing that person’s vocabulary, since the grammar of a language is simply made up of its parts, the words? And then, can we attend to someone’s style of writing simply by attending to that person’s grammar, and so on?

This way at explaining things (and viewing the world — i.e., defining reality or what is real [the ontological]) is common across all domains of knowledge and areas of practice. Another example: several years ago I was doing historical work at an institutional archives and reading some annual library reports from around the mid-20th century. One of the common problems that the head librarian described in those reports concerned his administration’s view of the library as nothing more than a warehouse of books. As he described it, the administration at his academic institution ignored (or even failed to see) the complexities attached to managing and using a library, and as a result, repeatedly failed to invest in the library and the librarians who operated it (the practical implication of a reductionist viewpoint). For the head librarian, the library was more than a warehouse of books; in the process of acquiring, describing, managing, shelving, circulating, using, and so forth, and by virtue of the material (books, serials, etc.) that was being attended to in those processes, something emerged or came into existence that was beyond a basic warehouse. And the thing that emerged was as real as any of its constitutive parts (e.g., books and shelves), even if it could not be reduced to those parts. Thus, even though that administration would agree that the library was a real place, it seemed that for them, the librarian might have said, it was only real as “cobblestones” are real.

And since I regard the significance of a thing as more important than its tangibility, I shall say that minds and problems are more real than cobblestones (p. 33).

Polanyi’s point in his lecture on emergence is simply this: higher order things (technically, coherent things), cannot be explained by or reduced to their constitutive parts (“the whole is greater than the sum of its parts”) even if those parts are necessary for the whole to exist. Similarly, grammar requires a vocabulary, but cannot be reduced to a vocabulary. The reverse is true, too. A vocabulary cannot dictate a specific grammar, just as a grammar cannot dictate, or determine, a style of writing:

Take two points. (1) Tacit knowing of a coherent entity relies on our awareness of the particulars of the entity for attending to it; and (2) if we switch our attention to the particulars, this function of the particulars is canceled and we lose sight of the entity to which we had attended. The ontological counterpoint [my emphasis] of this would be (1) that the principles controlling a comprehensive entity would be found to rely for their operations on laws governing the particulars of the entity themselves; and (2) that at the same time the laws governing the particulars in themselves would never account for the organizing principles of a higher entity which they form (p. 34).

Polanyi, in all of this, is making a case for tacit knowing (the distal) as something that cannot simply be explained by or reduced to explicit knowledge (the proximate). In the final lecture, he will address some scientific consequences of this position — issues that relate, in some respects, to problem-finding.

Personal Knowledge and Science

In this post, I want to address the personal aspects of tacit knowing. Here, Polanyi writes that:

The declared aim of modern science is to establish a strictly detached, objective knowledge. Any falling short of this ideal is accepted only as a temporary imperfection, which we must aim at eliminating. But suppose that tacit thought forms an indispensable part of all knowledge, then the ideal of eliminating all personal elements of knowledge would, in effect, aim at the destruction of all knowledge. The ideal of exact science would turn out to be fundamentally misleading and possibly a source of devastating fallacies (p. 20).

On the next page, he writes:

It is commonplace that all research must start from a problem. Research can be successful only if the problem is good; it can be original only if the problem is original. But how can one see a problem, any problem, let alone a good and original problem? For to see a problem is to see something that is hidden. It is to have an intimation of the coherence of hitherto not comprehended particulars (p. 21).

These passages highlight an important issue in knowledge production, aka, science. This is the issue of problem-finding (discovering the question to ask …), as opposed to problem-solving (… that produces the knowledge).

Most of the world sometimes appears focused on the explicit aspects of knowledge production. This might be because these acts are more tangible (hands-on) and because they are often the direct objects of scientific study itself (think of the focus on methodology and on the development of data collection instruments among social scientists and scientists). As such, we see in current political debates, in K-12 educational practices, and so forth, a major emphasis on the analytical methods needed to produce knowledge (e.g., the so-called scientific method). But this focus is one-sided. It is true that methodological concerns are important, and this is likely why, as an example, scientific articles contain entire sections dedicated to Methodology or Methods. It is because these sections serve a very important purpose; they provide a way to communicate the explicit aspects of scientific research—aspects that afford or enable others to either reproduce or replicate scientific studies from afar, by way of the document that communicates the explicit knowledge to do so (although we may argue that the background knowledge required to read these documents raises a whole host of tacit-related issues).

But what about the problem-finding issue? How does someone know what research question to ask? Polanyi is saying that this is a tacit issue. We sense the “coherence” of a thing or a system without knowing the particulars of the thing or system, and doing that allows us to start asking about the particulars. And thus problem-finding, a key but often ignored aspect of the scientific process, is completely dependent on tacit knowing. Since tacit knowing is personal (when we make a thing explicit, we make an object of it, and therefore, impersonal), then science depends on the personal. In fact, its foundations rest on the personal.

There’s some key empirical evidence to back this up. A little after Polanyi’s The Tacit Dimension was published, Robert Merton, the sociologist of science, published a paper called “The Matthew Effect in Science”. He wrote:

The role of outstanding scientists in influencing younger associates is repeatedly emphasized in the interviews with [Nobel] laureates. Almost invariably they lay great emphasis on the importance of problem-finding, not only problem-solving. They uniformly express the strong conviction that what matters most in their work is a developing sense of taste, of judgment, in seizing upon problems that are of fundamental importance (p. 453).

All science is, ergo, personal, and the better scientists are doing the personal (taste, judgment, etc.) better. What are the implications?

The Functional Relation in Tacit Knowing

When many philosophers speak about epistemology, they speak about how we know things and how we provide justification for what we know, and quite a bit of this epistemological work is spent on a certain kind of knowledge. This knowledge is the kind of knowledge that we articulate (make explicit) when we say a thing propositionally, like I know that January is the first month of the year in the US.(1) Such propositional knowledge is also sometimes referred to as know-that knowledge, and is often structured in a particularly grammatical way, at least for demonstration and analysis: e.g., S knows that P, or a subject knows that predicate.

Knowledge under such a view is generally defined as justified true belief (JTB). The description is nicely stated in the SEP article linked to above, but in short, we can only know things that are true (T) and not things that are false, we can only know things that we believe (B) and not things we do not believe, and we can only know things if we have justification (J) for them and not know things by luck. If all of these conditions are sufficiently met, then we have knowledge of the know-that type. For example, if we fail to meet the condition of belief that January is the first month of the year in the US (i.e., we do not believe that January is the first month of the year), then we do not carry that knowledge around with us. Or, that is knowledge we do not have, even if it is true, even if there is justification for it, and even if we are aware that such a proposition has been made by others (we might believe they are lying to us).

Polanyi does not ignore know-that (explicit) knowledge. In fact, he clearly states that this kind of knowledge and the kind that he is interested in discussing (tacit knowledge) are strongly intertwined, but he is primarily interested in understanding the latter (know-how) because he thinks it is important in understanding science, what science is, and what makes good science. We will come back to more of this science issue later.

In the meantime, this background information is important to have when learning about tacit knowing and knowledge. This is because it helps provide some understanding of how knowledge is understood, analyzed, and researched in a certain way in the knowledge management literature, which rests upon, often without attribution (see McCain on Obliteration by Incorporation), Polanyi’s concepts. Most of that literature (that I have read, at least) does not address tacit knowledge in much detail but to say that it exists and to posit ways in which it can be converted into explicit, codified knowledge, if it can be.

Let us reflect on what Polanyi is then doing in this first lecture from The Tacit Dimension. Polanyi articulates four aspects of tacit knowing: functional, phenomenal, semantic, and ontological. We will address the last three later, but it is important to spend a bit of time on the functional aspect because it is very key to understanding the rest.

The functional aspect lays out how we know a thing tacitly. Specifically, we know a thing tacitly by attending to some related, explicit thing. His example: people who are given electrical shocks when presented with certain nonsense syllables know when to expect the shocks even when they cannot say what specific nonsense syllables are triggering the shocks. That is, they attend to the shocks and by attending to them, attend from the particulars that cause them (the nonsense syllables). This attending from is still anchored to the particulars (the nonsense syllables) and so there is a relation. That is, the particulars are indeed known in this attending from. How else, Polanyi argues, would subjects in this test know when to expect the corresponding shocks? But even though the subjects know the particular nonsense syllables that trigger the shocks, they can not tell so explicitly (in their minds or verbally). They know more than they can tell.

One of the key ways that I have thought about this relation and how it fits within the knowledge management literature is by thinking about the process of writing computer programs. In this process, the computer programmer must (tacitly) make things explicit when writing instructions so that the computer can correctly execute the software code. If the programmer fails to be explicit in writing these instructions, often painstakingly explicit, then the computer will not be able to process those instructions. Unlike people, computers, then, seem to be purely explicit-knowledge-machines. They are, apparently, entirely dependent on know-that knowledge and on the functional and other relations that compose that type of knowledge.

Yet, facial recognition, autonomous vehicles, and robots that walk, jump, and flip seem to challenge that analogy simply because these abilities require a different sort of software instruction that cannot be all explicit or codified. Remember that Polanyi uses the facial recognition example to explain tacit knowledge, but if computers are capable of recognizing faces, and they do so without detailed propositions about the structure of all human faces on the planet or that will be on the planet in the future, then they either must not be explicit-knowledge-machines, or there must be a way of converting all tacit knowledge into explicit knowledge. If the latter, then Polanyi is not correct and we can tell all that we know even if we have not yet told everything that we know.

There is more to this though. The functional relation in tacit knowing is only one of the relations. There is still the phenomenal, the semantic, and the ontological to deal with, and we must still reckon with the very prescient discussions about the body and about dwelling that Polanyi writes about—discussions that will have some major implications on topics such as human computer interaction and other practical, applied issues. Thus — to be continued.

(1): Amartya Sen addresses this background work nicely in the preface to the Tacit Dimension.

Tacit Knowledge Challenges Scientific Communication

Let me write a bit about why I find tacit knowledge to be such an important concept as well as some of the motivations that I find very compelling in the discussions on the topic.

Since, most of the time, we cannot watch, interact, participate, or study in other people’s labs, on their computers, with them on their field work, and so forth as they conduct scientific investigations and experiments, we depend on the scientist’s ability to report about the processes, environments, and methods used when investigating and experimenting. As a result, science depends on good scientific communication, and good scientific communication depends on documenting the steps, materials, methods, and other activities and qualities that are a part of that process (and also on writing well).

However, Polanyi’s (1966/2009) claim that “we can know more than we can tell,” what he refers to as tacit knowledge, raises important dilemmas. In short, the assumption in the knowledge management literature, that we can know more than we can tell, does hold, but there is often disagreement about the type of knowledge that will forever be tacit and the type of knowledge that has the potential to be explicit but has not yet become so. (Nonaka (1994) was one of the first successful people to begin categorizing these differences.) Specifically, then, some problems we face with tacit knowledge, with this knowing more than telling problem, is whether tacit knowledge can be made explicit (can it be codified?); if so, is it tacit knowledge (or just implicit in some way); whether there will always be some kind of knowledge that will forever be immune to documentation; and whether, among other things, it’s worth documenting, or attempting to, certain kinds of tacit or implicit knowledge (because, for example, the cost is high in some important way or the cognitive load of too much documentation is too great (information overload), etc.)?

Transparency is not a new norm for science. Historians and sociologists of science have written about early motivations to disseminate scientific reporting, and Robert Merton’s insights about the norms of science touch on this (in particular, the communism norm reflects this in particular, at least in part as it focuses on findings more so). However, what’s interesting as of late is that communication and information technologies, such as the web and the internet, have rekindled the debate because these technologies provide an enhanced ability to document, in more detail and in more thoroughness, a greater part of the scientific process. The assumption, then, is that more and better information and documentation of the scientific process and not merely the findings, the argument goes, will lead to a better way of doing science and thus more reliable science.

These assumptions, however, need more testing. First, the same epistemological questions raised above apply to scientists; that is, what do scientists know more than they can tell in the process of doing scientific research? How much of and of what kind of their tacit knowledge needs to be made explicit, if it can be? If some tacit knowledge cannot be made explicit, does this type of knowledge present any true obstacles to reproducibility (akin to the Gettier problem–reproducibility by luck and not by true justification)?

Second, once those epistemological questions are surveyed, then the assumptions about the communication and practice of science need to be addressed. For example, is the quality of the scientific enterprise a function of its transparency; that is, its openness and the openness of its various aspects? For example, how important are some of these composite parts to science:

If so, what are the maxima and minima, so to speak, of this function? Or, if science is a function of transparency, made possible by documentation, what are the minima or maxima points of transparency needed for a healthy scientific enterprise?

It turns out that these problems hold true across the spectrum of organized human activity and not just the scientific enterprise (this is why the business/management literature has been able to take off with these knowledge management dilemmas). This makes sense and gives some credence to Susan Haack’s view that scientific knowledge is neither epistemologically privileged nor ill-founded. Generally, it’s simply more rigorous.

Haack, S. (2007). Defending science—within reason: Between scientism and cynicism. Amherst, NY: Prometheus Books. http://www.worldcat.org/oclc/52377534

Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization Science, 5(1), 14-37. http://www.jstor.org/stable/10.2307/2635068

Polanyi, M. (1966/2009). The tacit dimension. Chicago: University of Chicago Press. (Original work published 1966) http://www.worldcat.org/isbn/0226672980