Ask our readers: Is knowledge cumulative?

Here’s a question I think about a lot: Is knowledge cumulative?

It strikes me as an important question for how we think about the work that we do. Are we simply making utterances in an ongoing dialog, utterances whose echo will quickly fade, or are we adding to a general storehouse of knowledge which benefits from every additional sand of grain? I think how we answer this question underlies some of the differences between activist anthropology and what we might call “pure research” anthropology.

An activist thinks about her research as an intervention in an ongoing debate. While doing good research is important, even more important is the way that research might impact the contemporary political climate.

The “pure research” anthropologist is willing to spend their time on an obscure topic nobody is much concerned with at the moment because they are contributing to the expansion of human knowledge.

Of course, the reality is more complex than either of these caricatures. Work which is considered progressive at the time might be appropriated by conservative causes, while academic fashions may mean that the contributions of esoteric research lies in obscurity for decades or centuries before once again coming to light (i.e. when Rex blogs about it on Savage Minds).

While the promise of Google Scholar seems to speak to the fantasy of cumulative knowledge, there are some mitigating factors that must be taken into account:

For one thing, the overwhelming and exponentially increasing amount of knowledge means that there is simply more written than any one person (even Rex) can every possibly hope to read. True, the knowledge might be out there to be discovered – but if nobody is reading it what does it matter? The rise of national anthropologies further aggravates this problem, as more and more work is being produced in more and more languages with very little of it ever being translated. Another mitigating factor is the rise of corporate anthropology. More and more work is being done for private consumption and the owners of that knowledge do not wish to share it. Maybe one day it will enter the public domain, but how long will that be? And third, as I already mentioned, what we read is greatly shaped by ideology, politics, fashion, and the influence of a few trendsetters. Entire branches of anthropological knowledge now lie in obscurity because only a few readers will be willing to dig through their outdated paradigms for a few drops of useful data.

Even more importantly, our institutions have a tendency to promote certain kinds of work. For instance, it seems that every few years a new book gets published along the lines of The Bell Curve, catering to the racial ideologies of contemporary American society. This means that even if the general storehouse of knowledge is constantly growing, there is not necessarily a concomitant progress in our understanding of these issues. We seem to have to continually make the same points in new ways: Boas, Montagu, Gould, etc. each refashioning the same basic points. Using new data, to be sure, but it is hard to tell if it is two steps forward and one step backwards, or the other way around?

Finally, even those scholars that promote what might be called a “pure research” approach rarely claim that anthropology can be thought of in the same terms as empirical science. Now, there is no reason an interpretive approach to anthropology would require one to adopt an activist approach – for the very reason that there is no clear connection between any one interpretive position and a particular course of political action; and yet, this position seems to concede that anthropology is little more than an intervention into current discourses. Surely this position implies that the nature of this contemporary intervention is therefore more important than some abstract scientific model in which our research simply adds to a general storehouse of knowledge?

The funny thing is, even though I sound like I come down on one side of this question, I really don’t. I am actually profoundly ambivalent about these questions which is why I’m willing to put my half-assed musings up on the web. At some level I think my position can be described as wishing knowledge were cumulative and acting to realize this ideal (e.g. by advocating Open Access), but not really believing that it is truly attainable.

I’m sure anyone who has ever written an anthropological monograph has thought about these questions, and I’m curious to read what everyone else has to say about the matter!

21 thoughts on “Ask our readers: Is knowledge cumulative?

  1. Hrmm, I’m not exactly clear on what the first point is… “No one can read everything” does not imply “Some work will be unread”. I am in fact quite skeptical that there are any published works that not one single person has read ever. Anyways, as far as I can see, this could at most suggest that knowledge accumulates more slowly than it might, rather than not progressing at all.

    The second point 1) I am not sure what “more and more” means in this context, is that a higher percentage (which I doubt is true) or a higher volume? 2) this argument would apply equally to many fields–lets take engineering. It would basically be impossible if knowledge was not accumulative, and yet most engineering is done in corporate settings. Again, at most knowledge progresses more slowly in thise case (though I don’t think by much and is more than offset by the increased funding–example: engineering).

    The third point (influence of politics, ideology) is interesting, and partially true. I think it’s also a problem (imo bigger) with so-called ‘activist anthropology’ though.

    The last point just seems to say, well if anthropology is part of the humanities, then thats it. But while philosophy and literary criticism are discursive, they are hardly about ‘contemporary intervention’ in the activist sense you have in mind. I’d say what happens is that academics do “interv[ene] into current discourse”, but the result is ‘knowledge’ that enters the general ‘storehouse’. It’s interseting that, perhaps, some (broadly labelling) post-structuralists/post-modernists would dispute that, but the only reason we can even think in those terms (what their position would be, etc) is that we have some knowledge about post-structuralists and their arguments.

    Also, I think “even those scholars that promote what might be called a “pure research” approach rarely claim that anthropology can be thought of in the same terms as empirical science” is a strange idea that ignores biological anthropology and parts of archaeology.

  2. Hrmm, I’m not exactly clear on what the first point is… “No one can read everything” does not imply “Some work will be unread”. I am in fact quite skeptical that there are any published works that not one single person has read ever. Anyways, as far as I can see, this could at most suggest that knowledge accumulates more slowly than it might, rather than not progressing at all.

    If every work is read by at least one person, how does that imply any kind of accumulation of knowledge? For knowledge to accumulate, it must build on past knowledge. My point was that rather than knowledge building up like a mountain, one grain of sand at a time, it seems more like a river, with many various tributaries and streams.

  3. Because its more like a giant wall than a mountain. Bricks need to rest on top of something else, but they don’t need to rest on top of everything else. Or try probably a better analogy: linux. Linux is in fact a kind of “knowledge” and no one could understand everything (even Linus). Bits and pieces are added by many different people, some of it obscure, and they are used, even by people who don’t even know that they use it. I’m not sure what kind of server this place runs on, but I bet it uses, say, POSIX threads. You guys didn’t have to know how they work in order to put this site up, right? But maybe you work with some admin guys, who know a bit about whatever Unix they use, who know some other guys who build Unix distros, who know some other guys who work on thread scheduling in this particular kernel, who know some other guys who wrote and designed the API, etc etc etc. Cavaet: I hope this isn’t running on a Windows server! 🙂

    At any rate, as to the “what does it matter?” question, there is the classical example: Mendel. He didn’t know what use it was and no one read him for 40 years, then all of a sudden it’s rediscovered, and becomes, hrmm, rather important…

  4. How far are you willing to push that analogy? What does it mean to use Durkheim’s work without having read Durkheim? Sure, this is unfortunately standard practice, but is this really the same thing as building PHP software that runs on Apache without knowing how Apache works? In the latter case we don’t really need to know how it works, just that it is there. But can we really talk about knowledge of social theory (or society) in the same terms?

  5. no, it is extensive. knowledge does not accumulate like things, it extends like streams of water, it flows, and through flowing broadens or deepens the the tract through which it passes. knowledge is not an ‘is’ or a ‘have’ because you can quickly lose it, it is maintained. if the water stops flowing, or flows elsewhere, the tract will fade and disappear as the land forgets. think knowledge as process, ala dewey and deleuze and not knowledge as object ala capitalism.

  6. Fair enough. But the point is that you don’t directly “use” pthreads when you are writing PHP either. They are used by others in the kind of “chain of knowledge”. So to rephrase your question in these terms, you would ask well can you use Gluckman, without having read Radcliffe-Brown, without in turn having read Durkheim? I guess the answer is sort of but not really, and not to the extent you can use PHP without having understood apache. On the other hand, the separation is also an ideal with software. If you want to handle the traffic of google or whatever, you probably do need to know about thread scheduling. But you are right, it is just an analogy and it can’t be pushed that far. I still think its a better analogy than the mountain, though 😉

  7. Kerim, the notion that knowledge is cumulative smacks of Dawkins’ notion of the meme. No one, I hope, believes in that idea anymore–in memic terms, the “meme” has gone extinct. But here are two other ways to consider knowledge: The philospher Charles Pierce envisioned that scientific knowledge would at an infinite point coalecse with folk knowledge–basically, our common sense view of the world would accord with the scientific view of the world in the long run. Applied to Google, the more people google “relativity” or “mother’s brother” the more the studied views become fixed in the public consciousness. However, Pierce’s view was already undercut by Newton’s Principia, which is regarded as the intial break between folk-knowledge of physics and abstract physics. The joke goes, on seeing Newton, one student remarked to his friend, “There goes the man who wrote the book that nobody in Europe understands, not even him.”
    I am inclined to see our knowledge becoming more and more abstracted, requiring greater degrees of education with less and less cross-over between disciplines without major effort. It may not appear so from what I have seen on Google, but Anthropology can be highly abstract, requiring a fair bit of education and experience before its major concepts become apparent. But, in closing this point, these abstractions are embedded within arguments, which as students we spent a good deal of time mastering. It is the ability to present and criticise such arguments that constitutes our knowledge of our field.

  8. Fred: Isn\’t saying that knowledge will become more and more abstract simply another way of saying that is is cumulative? Is there any other reason to believe it should only move in this one direction?

    What really interests me is not so much the answer to the question, but how people understand the implications of the answer they choose. If you think knowledge cannot be cumulative, what does that mean for the kind of anthropology you practice?

  9. An interesting distinction you’re making between acitivist and “pure research” anthropologists there. At least in Norway the former really needs to get somewhat more of an acceptance instead of just being labeled as “shallow”.

    In concerns of the cumulative character of research: While it is true that all production is a done in a social fashion (and connecting one name of one individual researcher who with some piece of research is therefore generally in reality somewhat misleading), there is also the research that does not get built upon. And it is not just a random function that determines what is built upon further and what gets forgotten.

    For example, during my studies in Oslo, all research that I ever read abotu came from Norway, the US, Britain and a few select pieces from Sweden/Denmark, Westgermany and France. Now givent hat w could be presented with research from Westgermany, whatever happened to the theories built up in the former East Germany? While I am far from being a Stalinist, it is hard for me to believe that there was absolutely nothing of value in the works of the researchers of the former GDR.

    Another example, for which I can find even less explanations, is the total absence of anything bilt upon the economic analysis of the CEPAL – the Comisión Económica para América Latina y el Caribe. It is generelly taught in Norwegian founding courses of sociology and anthropology that the development models in the earl post war years built upon the ideas of Raul Prebisch, Secretario General of the CEPAL for many years (published CEPAL reviews ), which led to strategies of import substitution. However, when you then come to the oil crisis, somehow only neoliberal models of handling the debt crisis are presented. On top you generally get some culturalists who are supporters of “alternative developments paths” saing that somehow other cultures are not as inclined to develop as our Norwegian.
    When I tried to find out aboput what the CEPAL actually proposed as a solution to the debt crisis in the early eighties, I was rather surprised to find that our university library at the university of Oslo actually didn’t have any of the reviews. The only place in Oslo they would have them was at the house library of the central bureau of statistics. But also their access is restricted and I had to sneak myself in by acting as if I held a position a the university (and getting a guest label printed). The librarian also had a hard time finding the CEPAL reviews, but finally I was lead down some stairs and for the next hours I could copy some of the articles of the CEPAL collection they have in the cellar. Secret knowledge, if you ask me…

  10. >>Isn’t saying that knowledge will become more and more abstract simply another way of saying that is is cumulative?>What really interests me is not so much the answer to the question, but how people understand the implications of the answer they choose. If you think knowledge cannot be cumulative, what does that mean for the kind of anthropology you practice?

  11. >>Isn’t saying that knowledge will become more and more abstract simply another way of saying that is is cumulative? (Sorry, my previous posting was cut off).
    Not necessarily. Abstraction becomes informative of what we know of things when we start to abandon empirical positions which have limited application. For instance, Aristotle’s notion of motion implies F=mv, which makes sense if you roll a ball on the ground. But taking friction into account, the accurate description of motion is F=ma, but this is counter intuitive to everyday experience. In this example, F=ma is abstract as it does not rely on experience; indeed, Newton’s dropping apple is a perfect thought experiment that leads one to see the contradiction in F=mv, and support F=ma. But an understanding of motion, for the student or researcher, is not dependent on retaining an earlier knowledge of motion. In fact, I would suggest the abstraction destroys the earlier view, thereby countering the implication abstractions are cumulative.
    But I do suggest knowledge is redactive. One of the unique aspects of mathematics is that certain problems keep cropping up time and again within different algebras or calculi. In one sense, this shows that the new method has validity as it generates the same old answer, but that it is also valuable as it provides additional power to resolve new problems. I suggest this lends to knowledge a cumulative appearance, but the redactive nature proves validity and while applications increase range.

    >>What really interests me is not so much the answer to the question, but how people understand the implications of the answer they choose. If you think knowledge cannot be cumulative, what does that mean for the kind of anthropology you practice?

    I think what this means is that there is no single correcting path towards knowledge, but that knowledge is always competitive with alternatives. For instance, linguists can choose either a Generative model (Minimalist/Principles and Parameters), a non-linear model which allows for interactions between centres for semantics/phonology and morphology, or Optimality theory (“Emergence of the unmarked”). These are not just competing theories, which can be dealt with in some Popperian sense, but they require commitments that have to be made for personal reasons. Here is where I think I am answering your question: knowledge is personal choice, it involves a degree of commitment, not in believing, but in pursuing.
    (Note, I am not entirely satisfied with this answer, as it seems to blend into Peter Berger’s Social Constructionism).

  12. The concept of knowledge as destructive abstraction strikes me as remarkably Hegelian. It was in this sense that Hegel thought his own mind could embody all of the thought that had previously gone before him. So even though it is destructive, it does strike me as a vision of cumulative knowledge …

  13. Is knowledge becoming more abstract? Or simply more fragmented since the total volume of what might be known has long since reached the point that what any one of us can know becomes a vanishingly small fragment of a constantly expanding universe?

    Isn’t the more interesting question how we respond to this situation, with some sharply restricting input to be able to drill down and become microfield experts, some elaborating grand theories in an effort to still make some sense of the whole, some functioning like honeybees or brokers, flitting from one domain to another carrying ideas that others may bring to fruition?

    One of the hot new books from the Harvard Business School Press is Frans Johansson (2006) The Medici Effect: What Elephants and Epidemics Can Teach Us About Innovation, which focuses on folks who adopt a fourth approach, close study of two apparently disparate domains, placing themselves at an intersection where radically new ideas appear. (One of my favorites is a French telecoms engineer whose interest in ant behavior led to a breakthrough algorithm for routing Internet traffic.)

  14. Reading this, I am reminded of articles I see quite often in the field of Applied Linguistics, in which the collective energies of cognitive science, psychology and computer modeling are deployed in the search for a “language learning model.” The articles frequently end with a rhapsodic return to originary desire: if only we had more studies, more research, more grants, more subjects, more data points, we’d be able to refine our model sufficiently that it would accurately approximate how people learn a language. Other articles take it as their sole purpose to portray a novel visualization of the model, looking like flowcharts with boxes and lines leading the speaker through the deicisions between using the past tense or the present perfect. There is a profound ideological slippage here though in that the accumulation of knowledge is directed towards the end of knowledge – the final accurate model – which we know of course will never materialize otherwise they’d have to shut down whole departments of applied linguistics.

    Models of knowledge therefore act as ideological constructs which promote or detract from various research strategies. Is there some correct model of knowledge? I’m going to come down on the side of ‘probably not.’ For my own purposes, I see knowledge not as the some heap of facts but as the dialectically constituted connections between them. As an anthropological bricoleur, I draw on whatever ideas or notions, anthropological and other, that strike my fancy and seem pertinent to the question at hand. That’s why we can take something like Benedict Anderson’s Imagined Communities and make it into a kind of anthropological knowledge.

    And then if I am asked, is your research contributing to the heap of facts or is it an intervention into contemporary political concerns, I can answer “neither”; I’m just putting ideas out there that someone (maybe even an applied linguist) might find useful some day.

  15. I think it’s an awkward sort of spatial metaphor. I’m uncomfortable with the idea of “knowledge” as a general substance capable of accumulation in some form where its embodiment and method of growth is not strictly specified. Given my pragmatic account of knowledge, I tend to think of knowledge as “useful information understood by a person”. I don’t think that the metaphor fits here because there isn’t really a central reservoir at which knowledge can accumulate uniformly but a lot of little ones (people.) This objection is perhaps just due to my technical use of the term and my general suspicion of impersonal forces understood apart from their specific mechanisms in human interaction. In speaking about knowledge do we include the vast and various various systems which serve to distribute knowledge among specific humans?
    What is the process of accumulation? Is accumulation just more knowledge, or knowledge that builds off of previous knowledge or another method? Does accumulation happen in average knowledge across humanity or specific, measurably high concentrations of knowledge in individuals or in the concentration of distributed knowledge in groups that can make use of it collectively?
    My personal view is that knowledge is in some ways cumulative on average but certainly not uniformly or like in a “great humanist project of betterment” way. We can define groups diachronically and find that they have gained very little over large portions of time, or even lost some knowledge. Some groups have gained quite a bit, though I don’t think they know just all the facts of the previous generation plus more. I think that accumulation is a kind of abstracting process here, in that over time knowledge becomes more easily digestible and integrated as it’s worked through by culture, and perhaps it loses some detail or gains some. When they invented calculus hardly anybody in europe could understand it, now high school students learn a heavily wrought version of it.
    What’s made another difference is of course the changing natures of the way we distribute and produce information, which increases in some groups, the “potential cumulativity” of knowledge by making it more accessable though perhaps in some cases producing a lot of unreliable information (the internet, though don’t take this as an indication that I’m not a huge fan of wikipedia).
    I almost slipped into extending knowledge in the pragmatic direction into another uncomfortable metaphor with technology, institutions, practices and culture but that’s surely not what you meant by the term.

    I prefer a different metaphor, which I haven’t quite worked out. Instead of knowledge as a substance which accumulates on its own, perhaps digestion works better. Some of it is passed over, some is taken in, transformed and reconfigured to provide growing materials for a more complex body?

    PS. I don’t really see the connection to memetics? From my limited understanding of the whole memetics thing, the cumulativity of it was extremely downplayed. Perhaps there was a kind of like, fitness landscape type metaphor, but in general I remember memetics or at least Dawkins’ initial conception of it as a relatively synchronic thing without much concern for the whole underlying structure of knowledge through time.

  16. I appreciate Nelson’s point that certain communities of knowledge producers may create the conditions in which they are able to collectively accumulate certain kinds of knowledge for certain purposes over a limited time period. I think a lot of the recent emphasis on collaborative anthropology is about creating processes within which this can happen.

  17. J.S.Nelson writes,

    Given my pragmatic account of knowledge, I tend to think of knowledge as “useful information understood by a person”.

    I wonder if it mightn’t be analytically useful to restrict a pragmatic definition of knowledge to “useful information understood by two or more persons.” The implication is that information that cannot be shared does not count as knowledge or, conversely, that information becomes knowledge only when information is shared.

  18. It depends on your stance on epistemology, views on social constructionism and there’s a hint of the private language argument somewhere in there. I’m an epistemology person so I’m comfortable talking about knowledge held by only one person even if it is a kind of tree falls in the forest thing. I can see why knowledge that would be shared and thus constitutive of social arrangement would be of a lot more interest to anthro people.

  19. Kerim writes;
    What really interests me is not so much the answer to the question, but how people understand the implications of the answer they choose. If you think knowledge cannot be cumulative, what does that mean for the kind of anthropology you practice?

    I agree the implications are worth investigating. This is like asking the set of all cats. We know house cats, but we include wild cats, and then we include all the cats big and small. This sort of cumulative knowledge is what is being explored here. How does one connect the knowledge in other words. Of course this is inspired by Google.

    Knowledge is accumulative as that is the fundamental of language use in humans. Learning how to hunt with tools in hunter gathers is passed down from generation to generation.

    Text is a treacherous media in relation to connecting the dots to ‘accumulate’ knowledge. The brain uses invariant features of the landscape to ‘know’ in terms of vision for example. The point is that some sort of economy must be imposed upon the gigantic external information landscape. Text though is way way too economical. So for example can we use pictures in a language like way?

    The argument we can’t read everything is an argument about the engineered methods of connecting things to each other. Google is an engineering solution in which spiders index, and massive parallel computing architecture provides speedy access, relatively related material and so on.

    A frame of input, say a standard piece of paper has a certain amount of information it can hold, and can we absorb that? So we ask as well how well we really use the connecting process. So if we cram in more to the frame can we read more information if it is a picture rather than text. The frame also includes time, so that information first appears in the brain system at about 1000th of a second. But what we see is built up until around a fifthieth of a second a whole visual frame is ‘seen’. One eye is roughly 100 million receptors, cycling about every one hundredth of a second so that a second could hold roughly 1000 million different bits of information connected by say the common symmetrical laws of the forces of the universe. Which is immediately reduced to one to one hundred in the visual pathway.

    In this case then the question is a little like Moore’s law about how many transisters will there be packed into a computer chip. Every 2 years roughly double. So the question is about how much connection can we make in a given frame.
    Doyle Saylor

  20. How does one distinguish knowledge from information?

    Is truth a necessary condition for something to be counted as knowledge?

    If knowledge is merely empirical fact, then certainly knowledge is cumulative, since one can count facts. e.g. George Washington was President of the U.S. George Bush is the President of the the U.S. Two facts. Two pieces of knowledge. Facts can be put together to form new facts. Facts about incidents of smoking + facts about incidents of lung cancer are put together to form facts about the relationship between the two.

    There is all kinds of trickiness here though(probably investigated by Wiggenstein), but I don’t think anyone here means the accumulation of facts in a library. On sheets of paper. On hard-drives.. On stone, wood, and the skins of animals. Not some universal library, nor its superset, Borges’s library of Babel. We have to interpret these texts, to obtain knowledge. Information is there, in the medium, but underspecified, requiring decoders and decoding…and interpretation, to become knowledge, to become facts. And so knowledge is a living thing. As a living thing, knowledge is cumulative, but is bounded by our capability to process it, to make it relevant. In storage it is nothing, in activity, it becomes something. But we pour sand through sieves, accumulating but loosing at the same time. To keep it, you have to keep it in circulation.

    As for anthropology’s problem: the problem of accumulation is the problem that no one knows what the facts are.

Comments are closed.