| Integrative Levels Classification | project | ILC2 schedules | developing schedules | how it works | people | references |
We should welcome Michael Kleineberg to the international debate on the basic theory of knowledge organization, that he has just joined with a brilliant, stimulating paper:
The blind men and the elephant: towards an organization of epistemic contexts
"Knowledge organization", 40: 2013, n. 5, p. 340-362
based on a translation of a conference paper presented at the 13th Meeting of the German ISKO, Potsdam, 19 March 2013Abstract
In the last two decades of knowledge organization (KO) research, there has been an increasing interest in the context-dependent nature of human knowledge. Contextualism maintains that knowledge is not available in a neutral and objective way, but is always interwoven with the process of knowledge production and the prerequisites of the knower. As a first step towards a systematic organization of epistemic contexts, the concept of knowledge will be considered in its ontological (WHAT) and epistemological (WHO) including methodological (HOW) dimensions. In current KO research, however, either the contextualism is not fully implemented (classification-as-ontology) or the ambition for a context-transcending universal KOS seems to have been abandoned (classification-as-epistemology). Based on a combined ontology and epistemology it will be argued that a phenomena-based approach to KO as stipulated by the León Manifesto, for example, requires a revision of the underlying phenomenon concept as a relation between the known object (WHAT) and the knowing subject (WHO), which is constituted by the application of specific methods (HOW). While traditional subject indexing of documents often relies on the organizing principle "levels of being" (WHAT), for a future context indexing, two novel principles are proposed, namely "levels of knowing" (WHO) and "integral methodological pluralism" (HOW).
The paper variously cites works by authors participating in the Integrative Levels Classification project (especially Poli, Szostak, and me) as well as many other sources, to collocate them in a well-informed, wide picture. It acknowledges the relevance of phenomena-based KO and of levels of reality as a model for KO. At the same time, it sees them as only part of what is needed, as they would exceedingly emphasize an objective, "modernist" notion of knowledge while neglecting a more updated "postmodern" constructivist one.
Kleineberg offers a useful discussion of levels of being and their supposedly serial order. As correctly reported, this seriality is questioned by Poli suggesting that the mental and the social levels may be "tangled" rather than one prior to the other: "the underlying idea is that there are no societies without minds, just as there are no minds without corresponding societies" [p. 344]. I personally believe that minds without societies do exist in all non-social higher animals, that is in the vast majority of their species like tigers or eagles, and that this provides a rationale for minds to precede societies in a general KOS. But the main point is probably another. As already acknowledged by the CRG, even the sequential appearance of levels produces a branched rather than linear structure: e.g. rocks are towards the top of a mineral branch, and bacteria arose later not out of rocks, but out of molecules laying at a lower level.
This kind of problems in modeling levels is also mentioned by Kleineberg: "intellectual products like literature and music are not composed of societal beings or material products in the same way as molecules are composed of atoms" [p. 343]. This is true, and stimulates me to state more clearly that the general relationship between two levels is not always one of material integration ("overforming" in Hartmann's term): it is more generally one of existential dependence. Such relatioship should indeed be better distinguished into at least three kinds: integration (e.g. atoms - molecules), aggregation (e.g. atoms - mixtures, cfr. CRG [1969]), representation (e.g. organisms - minds, cfr. Gnoli [Axiomathes 18: 2008, p. 177-192]).
In general, linearization of levels into a series is needed for the practical purposes of sequential browsing of classes in a website menu or a library shelf, but can well be a simplification of a branched and more complex structure. An appropriate tool to model the different kinds of existential dependence could be a digital ontology.
Kleineberg's levels of being also have a structure more complex than the linear one. Indeed, they are modelled as a "Great Nest" [Wilber] emerging out of an "AQAL" Cartesian system of coordinates (interior/exterior x individual/collective) defining four quadrants, each with its own linear series of levels. This is and elegant and original model, although I wonder whether aesthetic reasons prevail over ontological ones in determining its elements: which is the principle leading to choose these two coordinates and to postulate that they are anterior to levels? Their symmetry forces us e.g. to assume that even atoms must have some form of "intentionality"; my way of thinking would rather suggest that intentionality only emerge at a certain level (not necessarily the human one) after which it can well assume several different forms. Evolutionary sequence may be even more relevant with levels of knowing, where the four categories of "behaviour and organism, consciousness and knowing, culture and worldview, and society and environment" would appear to me as laying themselves at different levels, rather than defining four synchronical quadrants.
By the way, the Information Coding Classification is listed together with ILC among "phenomena-based KOSs" and reported to have been inspired by the work of the Classification Research Group: while both claims are true of ILC, they are inaccurate in the case of ICC according to its author [Dahlberg, private email]: indeed ICC is reported to be an independent application of Hartmann's theory of levels, identifying ten areas of knowledge that are subdivided into subject fields corresponding to disciplines rather than phenomena (or than objects of knowledge, as Dahlberg prefers to call them). Anyway, it is true that levels theory has inspired several classifications developed since the 1970s including ICC, BSO, BC2, and ILC.
It is good news that Kleineberg describes ontology and epistemology as two dimensions in KO, thus adopting the same term as Hjørland, me and others, which makes communication easier. In my model [Gnoli, KO 39: 2012, p. 268-275] I also consider further dimensions:
I acknowledge that the epistemological dimension is not well developed yet in ILC, and that knowledge organizers could well investigate it further as urged by Kleineberg (and as started by Szostak with his theories and methods). The epistemological part of Kleineberg's model, where the notion of levels is extended to "levels of knowing", is indeed a very interesting starting point for this.
What we seem to model in different ways is dimension β. To Kleineberg, phenomena should not be seen just as objective entities, but as an intextricable interaction of objects ("what"), knowing subjects ("who") and methods ("how"). Acknowledging some important role for subjects and methods is in agreement with the León Manifesto, and with Szostak's model. It is true that ILC, in representing perspectives as facets of phenomena (0-), privileges the latter [p. 359]; however, this is only a syntactical device, and a different KOS could be developed where phenomena and perspectives are connected by operators of the kind of PRECIS ones, giving equal importance to both their arguments.
The main point is whether perspectives can be separated from phenomena, as in my model above, or not, as claimed by Kleineberg. The subject here is subtle and we should try to avoid misunderstandings. Modernists would separate phenomena from perspectives in admitting "the possibility to have a view from nowhere or a God's eye perspective". I personally acknowledge that any knowledge includes epistemological components, just because it is knowledge. But this does not mean that we cannot refer to the notion of a phenomenon as such, nor that the elephant in itself cannot exist. Indeed, it has most probably existed independently before we knew about it.
This elephant in itself seems to correspond to my dimension α (denied by Vickery who was more positivist than me). Notice, however, that I don't claim that classification-as-ontology organizes reality in α. I acknowledge that KO can only start from phenomena, that is, from the part of reality to which we currently have access through our senses, tools etc. [Gnoli 2012 cit.]. These have collectively led us to identify a phenomenon called "elephants", so that our present KOSs include a class for elephants. I agree that our phenomena are the result of both the resistence of reality and our epistemological means (our "theories", in Popper's terms): so there is a feedback circle between phenomena and perspectives.
As knowledge progresses, new theories can lead us to revise our notion of elephants as a phenomenon, e.g. to split them into distinct species distinguished by some genetic traits of which we were unaware before DNA sequencing techniques. Indeed, the KOSs we develop do evolve: Dewey and UDC are often published in new editions, other systems have been abandoned, and new ones can be conceived. This does not imply that elephants in themselves don't exist. Indeed, while elephants in themselves are in dimension α, what we deal with in KO is knowledge of elephants, which is dimension β. Saying that there is no knowledge of elephants independent from senses and theories is not the same as saying that there are no elephants independent from senses and theories.
Despite the feedback circle between phenomena and perspectives, I do think that it is useful to distinguish between them, rather than saying that phenomena are an undifferentiable mix of ontology and epistemology, none of which is prior to the other, which seems too much a judgment of Solomon. While β are consolidated classes (elephants as we currently believe they are, and will probably continue to do so for some time), γ are the different approaches to those classes, e.g. cladistic taxonomy vs. phenetic taxonomy, or scientific description vs. poetic description. Users of a KOS may well need to search for anything concerning elephants, or for anything concerning cladistics, without being forced to choose since the beginning between cladistic elephants and phenetic elephants...
"The phenomenon of elephants" should not be meant as an a-priori class, like is assumed in postmodernist criticisms [p. 350]. Indeed, we have not just woken up one day suddenly postulating the existence of elephants. We did approach the world starting from our subjective epistemological means, e.g. seeing our first elephant in a zoo or in the wild. But in time we have integrated all available knowledge elements (knowledge provided by our senses, by tools for indirect measuring, by previous knowledge recorded by others, etc.) into a system more and more refined. This system currently includes the phenomenon of elephants as a result. Such knowledge of phenomena is "objective" only in the a-posteriori sense of the title of Popper's book, that is, as it has been challenged by many theories and efforts of scientists, leading to transcend the initial subjective feeling of elephants, that of a naive child who never communicated with others nor studied zoology. Our resulting "neutral description of the elephant" [p. 352] thus is not an a-priori starting point: it is only the a-posteriori result of a long collective effort. And of course this does not guarantee that one day this objective knowledge will be challenged again by some new theory and will have to be updated in our encyclopedias and in our KOSs, like it has been done when discarding such obsolete phenomena as "phlogiston" or "aether" from them.
This dynamic view of knowledge, after all, agrees with Kleineberg's view of levels of knowing (magic, mythic, rational, pluralistic) as not opposed to one another but having developed along a historical evolution (and yes, even along a phylogenetic evolution, cfr. Lorenz and Wuketits's evolutionary epistemology) [p. 353]. Notice that Kleineberg assumes a non-declared progressive notion of knowledge (with which I agree) that may not be shared by all "post-postmodernist" theorists. A view also relevant for accounting the history of KOSs, their chronological dimension [Tennis 2002, cfr. ILC facet 01 for expressing e.g. "elephants as known in the Middle Age"]. I would say that KOSs, rather than including different levels of knowing, are different levels of knowing: indeed, a KOS cannot do more than representing knowledge at the level reached at the time of its creation. KOSs of the mythic level had to be mythic KOSs. While past levels of knowing can be embedded in present KOSs, and this is very promising, future levels will have to be reflected only in future KOSs.
In conclusion, besides these details, I applaud Kleineberg's productive approach aiming at going further than relativistic critics, to attempt an actual organization of the epistemic dimension. This has the potential to produce innovative developments in our field.
The positions of different KO theorists can be described according to which dimension they take as the primary reference for KO:
The recent article by Kleineberg does indeed provide much food for thought. His attempt to bridge the gulf between those who urge only domain analysis and those who seek a truly universal classification is to be applauded. This is something that Claudio and I have tried to do in our own way for some time. I especially liked his argument that the concerns voiced by others regarding ambiguity and subjectivity do not necessarily justify their call for only domain analysis. My mantra of late has been "It is an empirical question," and I believe that we cannot reach such a conclusion on theoretical lines alone.
That mantra applies also to the observation that the boundaries between integrative levels, especially at higher levels, are less easily described and patrolled than we might have hoped for. As Kleineberg notes (the breadth of his research is also to be applauded), Roberto Poli had appreciated this. But boundary problems are hardly unusual in classification. The empirical question is whether they are too severe for the purposes of a particular classification. The success that Claudio, myself and others have had in developing classifications grounded in integrative levels suggests that they are not.
I suppose it is unsurprising that I appreciated Kleineberg's analysis of others but had some qualms about his analysis of the work of Claudio, me, and others who are pursuing universal classification. My main qualm has to do with the reliance on Wilber. Wilber's theories are controversial. Many doubt -- as Claudio has -- that the dimensions that Wilber prioritizes deserve priority. It would be ironic, to say the least, if the best response to Hjorland's concerns with the theory-ladenness of phenomena was to ground a universal classification in a very controversial theory.
I have disagreed with Hjørland [Szostak 2011] regarding his desire to ground classification in the most pessimistic of philosophical concept theories (the theory theory, which argues for the theory-ladenness of all concepts). I have instead urged us to seek to respect as wide a range of philosophical theorizing as possible. I am afraid I have to apply the same guideline to Wilber: we should not ground a classification exclusively in any one theory.
I am not quite clear how Kleineberg's triadic approach to classifying phenomena would work in practice. This I think might be the most useful avenue for future conversation. It seems to me that Claudio and I are trying to do something quite similar. To be sure, when we classify phenomena we may only capture the 'what' in Kleineberg's triad. But Kleineberg appreciates that this 'what' is necessary. When Claudio and I try to classify documents or ideas, we try to capture these also in terms of 'how' (by classifying 'method' applied and also 'theory' applied), and 'who' (we have both discussed how we can in addition to classifying theory applied capture various elements of authorial perspective). So while we do have a one-dimensional approach to classifying phenomena we take (at least) a triadic approach to classifying ideas. I am curious to learn more about both how and why we might apply a triadic approach to phenomena themselves.
Kleineberg's contribution to this ongoing conversation is very much to be welcomed. I hope that we are able to continue this conversation.
The blind elephant: a reply to Claudio Gnoli's comments [PDF]
After discussing Claudio Gnoli's comments [link above] I would like to address some comments offered by Rick Szostak. First of all, I agree that the three of us are trying to do something quite similar, in particular, to develop a KOS with a universal scope covering phenomena as well as theories and methods applied by scholars. In this regard, my main concern is to implement contextualism and perspectivism as they are legitimately emphasized by the various "postmodernist" theories of knowledge. Therefore, I argue that we should reconsider our metatheoretical assumptions as outlined, for example, in the León Manifesto. In other words, I would insist that theoretical considerations are crucial for KO research.
As Szostak rightly suggests, integral theory proposed by Wilber is very controversial. One of the main reasons might be that Wilber writes extensively about spirituality. Indeed, his approach was a long time largely ignored in academic discourse. Personally, I hardly met someone who had ever heard the name of Ken Wilber. Therefore, I did expect the criticism to rely on Wilber's work. In defense of this reliance, I would stress the fact that one might benefit from a discussion of particular theses without adopting the whole approach. For example, Wilber's major work Sex, Ecology and Spirituality [2000] provides a careful analysis of the co-evolution of ontological realms and offers new arguments for the discussion of boundary problems between integrative levels. Since the days of the Classification Research Group, these issues are intensively debated within the field of KO and there still remain some open problems. In spite of the controversial status of Wilber's overall approach, I see no reason why we should ignore such specific arguments addressing these problems. Additionally, in The Blind Elephant I have tried to demonstrate that Wilber's basic assumptions (e.g., panpsychism, co-evolution) might challenge the mainstream view but are based on reasonable conclusions and do not contradict the evidence available. For some of these basic assumptions I have even presented examples of quite similar but independently developed approaches supporting these core ideas. As a metatheory, integral theory provides a bridge between more contextually and more universally oriented approaches to the organization of human knowledge. Therefore, I believe that KO research might benefit from a closer consideration. Sometimes old problems require unexpected solutions. I would not call it "irony" but rather a lesson from history.
However, Szostak emphasizes an important point regarding the theoretical foundation of KO. On the one hand, I agree with him that "we should not ground classification exclusively in any one theory" but rather "respect a wide range of philosophical theorizing". On the other hand, Szostak's [2007, 34] call for "openness" is actually reflected by Wilber's [2006, 16] call for "nonexclusion" as one of three heuristic principles of integral theory (besides "enfoldment" and "enactment", as described in my paper). In fact, this attitude of openness or nonexclusion is the very reason why this approach is labeled "integral theory". Of course, there might be different ways to integrate the multiplicity of perspectives, and Wilber's approach is only one of them, but without a coherent metatheory any attempt to cope with the challenge of epistemic pluralism would end up in eclecticism rather than truly integration. Here is my point: If Wilber's integral theory and his AQAL model are controversial, then let us analyze their potential flaws instead to marginalize these offers from discourse. More recently, the so-called integral movement seems to have indeed an increasing impact on academia. For example, a think tank called Integral Institute (II) has been founded and supported by reputable members such as Francisco Varela, David J. Chalmers, Robert Kegan, Susanne Cook-Greuter, Roger Walsh and many others. In the meantime, there have been established frequent conferences and master programs on integral theory, ongoing debates with other interdisciplinary or "integral" theorists such as Roy Bhaskar or Edgar Morin, and even an academic journal called Journal of Integral Theory and Practice (JITP). Today, Ken Wilber is announced as the most widely translated academic writer in America, and figures like Bill Clinton or Al Gore publicly recommend his books. To avoid misunderstanding, all of this does not mean anything for the validation of integral theory. It still remains a controversial issue. But to be controversial does not mean to be useless for the exploration of new ground.
Szostak's main concern, however, seems to be how one could best respond to Hj�rland's claim of the theory-ladenness of phenomena. I would say the answer should be something like: "Yes, but ...". I believe that Hj�rland is absolutely right about his "theory theory" and I agree with him that the "interpretive turn" (including contextualism and perspectivism) has to be taken seriously in KO theory. The crucial question, however, is what kind of conclusions we should draw from this insight. Hj�rland argues for domain analysis based on pragmatism, and this seems fine to me. But there are many forms of pragmatism on the market. On the one hand, there is a kind of pragmatism leading to relativism such as Ludwig Wittgenstein's linguistic pragmatics or Richard Rorty's neopragmatism. On the other hand, there is a kind of pragmatism avoiding relativism such as Charles S. Peirce's evolutionary pragmaticism, Karl-Otto Apel's transcendental pragmatics or J�rgen Habermas' universal pragmatics. The approach proposed by domain analysis is apparently not at all the "most pessimistic" one since even Hj�rland argues against anti-realism and skepticism. Nevertheless, it seems to me that Hj�rland is somewhat ambiguous on the issue of relativism. While his socially and historically oriented approach clearly acknowledges epistemic pluralism, it is also assumed that some theories do a better job than others. But what exactly does it mean? Most approaches of pragmatism maintain that all knowledge claims can only be judged from within their respective "theory-laden" context. Therefore, a theory from context A cannot do a "better" job as another theory from context B, unless we could obtain a context-transcending or universal measure which enables us to compare different contexts. This seems to be the way followed by Habermas and the historico-genetic research program. As I emphasized in my paper The Blind Men and the Elephant and discussed at length in my reply to Gnoli, we should more carefully distinguish between "content" versus "structure" with regard to human knowledge since only the analysis of the latter would allow us to interrelate divergent perspectives in a non-relativistic way. My impression is that domain analysis is mostly concerned with the surface features of "content" and still lacks adequate analytic tools to investigate the underlying deep features of "structure".
Again, I have to emphasize that my proposed triadic concept of phenomenon as well as the concept of Integrative Levels of Knowing (ILK) do not depend on Wilber's integral theory. This particular approach is simply an example for an application. Although I consider integral theory as one of the most sophisticated approaches to the organization of knowledge, my paper did not exclusively rely on Wilber but referred to the works of Brier, Dux, and Habermas among others. Interestingly, the profound work of the latter seems to be important for all of them, particularly, for Hj�rland's contextually oriented "theory theory" as well as for Szostak's universally oriented theory of interdisciplinarity. This is one more reason why I am still optimistic that KO research is able to develop a more inclusive framework in order to embrace domain analysis within a truly "transdisciplinary integration of knowledge", as Brier would put it. This overall goal, of course, is the driving motivation for my proposal.
Finally, I agree with Szostak that the most useful avenue for future conversation would be to demonstrate the practical relevance of such a triadic and multi-leveled approach to human knowledge.
So, although both Kleineberg and me acknowledge that knowledge is made of both phenomena and perspectives, I seem to be more optimistic than him about the possibility of separating them. Still, the meaning of this separation seems to be quite clear, at least ideally, as we discuss about ontology vs. epistemology, rather than conflating both of them into a single branch.
Our notion of a given phenomenon may change according to our current perspective; but as this is a feedback loop, perspectives in turn are tuned according to new experience of phenomena. As Kleineberg rightly points out, differences and influences between perspectives can be either synchronic or diachronic. An elephant is first perceived as a phenomenon in some naive way, e.g. as one elephant cub is seen from far by a non-African explorer, and after his account a myth develops about elephants in his culture. Then new theories, e.g. after contact with African culture, provide new views about elephants. These views are checked against further experience: adult elephants are observed that are on average greater than that cub, and the notion of elephant changes. So perspectives are determined by phenomena just like the other way around. Indeed, perspectives alone would be frivolous exercises of speculation (one can speculate that talking striped elephants exist in some exoplanet, but this is irrelevant until any clue of them is experienced.)
The notion of elephant in the mythic taxonomy is indeed different from that in the later taxonomy; but it seems quite clear that there has always been something called elephant separated from the available experience of it. Things may be more intricated in less trivial domains, such as particle physics [Szostak 2004]: is Higgs boson only a theoretical construction or an actual phenomenon? (let's build the Large Hadron Collider to check it). Still distinguishing phenomena (as they are currently known) from perspectives is useful for the advancement of KO. Indeed, Kleineberg himself admits that "it seems still useful to consider levels of being and levels of knowing in separate sections", something that is impossible if we say that phenomena and perspectives (β and γ) are the same. Otherwise, how could we achieve Kleineberg's and León purpose of accounting for different perspectives within one same KOS?
Kleineberg follows Hartmann and the general tradition of German philosophy in using the notion of spirit (Geist) as distinct from pshyche, consciousness etc.: higher beings like humans have not just awareness but also "personal spirit". I understand that here this does not necessarily imply a religious framework, but believe that the term may cause problems to readers more familiar with the English-speaking tradition, to which Geist sounds like ghost, so maybe this notion of spirit could be introduced in some more detail in future discussions.
Kleineberg's considerations about the topology of realms, or strata, are very interesting and clear. In general, we have to keep in mind that linear models useful for arranging knowledge items in search interfaces or library shelves may indeed be simplifications of more complex topologies, featuring asymmetrical relations (like in an ordered series) but also branchings and crossings (synergies or co-evolution). Rodrigo De Santis has recently started exploring the representation of such relationships in the form of digital ontologies within the ILC project.
I agree that such main classes as "mind" or "society" may be placed in different ways according to the meaning one gives them, so that in each particular KOS they should be defined more precisely: e.g. society as based on human language in ILC is labeled as "civil society", so to exclude plant associations studied by "phytosociology" or insect colonies studied by "sociobiology": hence it comes after "mind" rather than before it. As for which transitions are more important than others, claiming that the one from life to mind is "incomparably" greater than the other ones [Skrbina] may be questionable and influenced by an anthropocentric viewpoint: what a cell would say about it? :-) Other transitions, like matter to life, cannot be described just as "differences degrees": they are "kinds" too, as they also identify new major classes of phenomena (life, society, culture).
The AQAL model is well described by Kleineberg as an interesting alternative to linear models. Still the ground for its basic axes is not clear. The interior/exterior axis is mostly applied to sentient beings. However, other phenomena such as clouds, dens, or cars also have an interior and an exterior, that would probably not fit this model. So what actually are "interior" and "exterior"? Also, it is not clear to me why Wilber postulates that artifacts have no interior: as they are also material objects, they should have just like in his model molecules and rocks have. Wilber seems to give artifacts little ontological importance, as they would just be byproducts of their authors; however, an artwork can assume different meanings than those intended by its author, which is why it belongs to an autonomous ontic level, acknowledged as "World 3" by Popper and as "objectivated spirit" by Hartmann.
AQAL avoids radical emergentism within each quadrant, as Kleineberg explains. But this does not yield a simpler description of the world, as one still has to explain why the quadrants exist, that is, why psyche, "interior" etc. would exist since the beginning of the cosmos. Overall it seems that a linear sequence of emergences, both weak and strong, is a more economical explanation than four lines each with a series of weak emergences... Anyway, on this road we are ending by doing basic philosophy rather than KO!
Dear Claudio, Michael and Rick, your discussion is very interesting. In the article (Phenomena or noumena? Objective and subjective aspects in knowledge organization) I've just submitted to "Knowledge Organization" journal, I explain why according to me, too, in the human knowledge of reality (and a fortiori in any human attempt to organize it) both objective and subjective aspects are necessarily present. Once this has been admitted, the most natural following step is indeed to wonder (as you have done) if those aspects are or not separable from each other. On this point, however, I have not a definite position (and therefore I have not talked about it in the article) because:
I suspect, therefore, that, even with regard to the issue of separability, the (difficult) road to follow is the one I have foreshadowed in the conclusions of my article, that is to say to resign oneself for now to an alternation (or coexistence) — both in the theoretical studies and in the practical applications of knowledge organization — of the objective and subjective aspects, waiting for the emergence of a really balanced and convincing gnoseological theory of synthesis.
Please feel free to send further comments to ilc@iskoi.org: they can be published here as well.
The blind knowledge organizers and the elephant : working notes on Kleineberg's levels of knowing # 6.1 / Claudio Gnoli — ISKO Italia <http://www.iskoi.org/ilc/elephant.php> : 2013.11.23 - 2016.01.12 -