For Ankersmit, the problems with the traditional account of representation strengthen his resolve to stick to the concept of representation and find an alternative formulation for what representation is rather than seek a replacement for it. I agree with him that the problems with representation pose a challenge but, to me, the challenge is to go yet one step further and ask what historiography and its central knowledge contributions are, without making the assumption that they must be representations. In the previous chapter I argued that a commitment to representationalism unites the philosophies of Ankersmit and White. And the beginning of this current chapter discussed how representation- alism (in various formulations) has a long history and goes well beyond these two thinkers, including many who are otherwise extremely critical of Ankersmit and narrativism.15 My view is that it is essential to consider giving up on the pre-analytic representationalist intuition that historical writing is necessarily about creating representations and that the meaningfulness of historical discourse requires that these representations are about some specific (abstract or concrete) corresponding entities that are re-presented.
There are two topics to consider here. One concerns my reasons for claiming that representationalism is a problematic commitment. The second involves the task of signposting a route to an alternative non- representationalist account of historiography. In other words, I wish to outline an alternative way for understanding what historical theses or narratives are - one according to which they are not about any kinds of target objects. The latter topic also revisits the question of what the rationale for historiography is. Both of these themes will be continued and the analysis deepened in subsequent chapters.
According to Ankersmit, the suggestion that each representation is paired with one aspect or presented stems from the logic of historiography. The idea that there is such a logic implies that these notions are needed to understand what historiography is and/or what it produces. The postulation of 'representation' may of course help in this task, but the question is whether the Ankersmitian notion of representation offers the least philosophically problematic and pragmatically most useful account available. Some philosophical problems that a commitment to representations brings were already discussed above. On the other hand, Ankersmit claims something more than that representationalism is pragmatically worthwhile - namely, that without a representational account one cannot provide an adequately comprehensible view of historiography. Representationalism is thus allegedly a necessary position on historiography in the philosophy of historiography. Is it?16
We are faced with the question of the pragmatic value of representations. How is one able to identify an 'aspect' and ensure that we are talking about the correct 'aspect' of a representation? Identification is important when we try to pin down what historian X's representation actually is. How could one check whether a representation 'is about' the correct corresponding 'aspect'? It is instructive to remember Wittgenstein's argument against private language, in which he said that a private language would be unintelligible not only to others but ultimately to its originator as well, as no-one could establish the meaning of its signs due to their non-public nature. There needs to be a possibility of being wrong or right about these aspects, and the criteria cannot be private.
It is notoriously difficult to spell out a historian's, or indeed any writer's message or thesis, and thus disagreements about what that message is are not uncommon. If we imagine that different readers come up with different prima facie justified 'aspects', what then is the status of all these 'aspects'? Are they all wrong, except perhaps one? There does not seem to be any interpretation-independent and other intersubjective way of identifying an 'aspect' except to read the written presentation and try to figure out what its 'presented' is, which takes us back to the original problem. Or would one choose the other horn of the problem, that is, that all interpretations are proper and justified 'aspects' of a representation? This option would render the notion of 'aspect' redundant because one would simply be just offering different interpretations of one representation.
To repeat, Ankersmit has correctly rejected representation as a two-place predicate according to which representation more or less directly reflects historical reality itself. The problems with that are obvious. 'Narratives' or 'representations' of history contain qualities that the past does not have. Further, it seems that they do not refer to any uniquely identifiable entities in the past. The central problem with two-place represen- tationalism is clear. The past and its representations are so different in terms of their qualities that it is senseless to try to pair the two together. Now, it is not entirely clear why one would need the middle variable of the 'presented' or 'aspect' between 'representation' and the past if it has been admitted that representations cannot reflect the past. This variable does not seem to bring any practical benefits in terms of criteria for representational content. This casts doubt on whether the correct logic of historiography has been explicated: if 'presents' or 'aspects' do not help to decipher the meaning of a historiographical text, then they do not seem to be necessary postulations in textual interpretation- Further, philosophically, and particularly with respect to ontology, the added middle layer appears entirely superfluous, and is bound to create additional philosophical puzzles with regard to its ontological status, location, identification, etc. In this respect, my suggestion is to apply the widely accepted maxim of Occam's razor: all things being equal, one should favor a simpler solution over more complex ones. And it is more straightforward and 'logical' to say that historiography is a 'presentational' rather than a 're-presentational' activity once the mirroring metaphor has been rejected. A work of history presents something, or constructs a thesis, but it is not necessary to assume that these products have independently existing counterparts either in a real or in an imaginary world.
The concept of representation appears to imply a capacity to present again something that has a determined form before any representational act. And if it is the case that representation via isomorphism, resemblance and similarity all fail and, further, that there is no determinate object to be re-represented, the concept of representation fails. It is true that many 'revisionist' accounts are more relaxed with regard to isomorphism, resemblance and similarity, but they require a target, an object, nevertheless. If we may stipulate target objects as tolerantly as Suarez (2004) suggests and only require that one be able draw informative inferences, any historiographical interpretation would turn out to be a representation of the 'past'. However, it is obvious that this re-definition makes the notion of 'representation' almost empty of its original meaning. The object category of the 'past', for example, is informationally too broad, whereas more fine-grained stipulations, such as the Renaissance, raise the question of whether the target object is given or is itself historiograpically constructed. Suarez makes a worthy proposal, but it is ultimately an attempt to rescue the concept of representation by removing it from its customary and productive sense. This kind of deflationarinism or minimalism has been fashionable in recent years with regard to truth (e.g. Horwich 1990; Wright 1992), knowledge (e.g. Williams 1996) and meaning (e.g. Horwich 1998), for example, but makes one wonder whether it would instead be more advantageous to reject the idea that such concepts are universally applicable. Why is a particular notion seen as so valuable that the conditions of application need to be stretched to such lengths? It is worth remembering that Nelson Goodman thought that fictional objects - such as the picture of a unicorn - have a 'null denotation'. He argued that such expressions as 'picture of' and 'represents' may be misleading because they have appearance of two-place predicates even in cases when they should be understood as one-place predicates: '"picture of Pickwick" and "represents a unicorn" are better considered unbreakable one-place predicates ... from the fact that P is a picture of or represents a unicorn we cannot infer that there is a something that P is a picture of or represents'(1976, 21). My view is that it is preferable to limit the scope of application, as in the case of Goodman's null denotation, rather than to attempt to make 'representation' an all-inclusive concept.
On one occasion Ankersmit comments on a similar proposal. Assuming that the copy theory of representation is untenable, Ankersmit notes that the 'postmodernist' solution would be to talk only about a presentation that, as such, does not refer to and is not about independent reality - and thus also not about re-presenting anything: 'We cannot properly speak of historical representation at all. For, the term representation requires the presence of an independently given (historical) reality which is, next, represented in and by historical writing. Consequently, as postmodernists often argue, the postmodernist notion of the simulacrum is essentially a going "beyond" or against representation' (Ankersmit 1994, 191-192; cf. Jenkins 2003a, 41). He goes on to say that in this case we would only speak of 'presentation'. However, as we have seen, although Ankersmit aptly analyzes the implications of representationalist vocabulary, he is not ready to sever the one-to-one link between representations and the historical reality that they are about.
The scare-word 'postmodernism' popped up in the quotation and requires some comment. To say that neither knowledge nor historiographical interpretation reflects reality is not to deny that there is something real or a reality. It only says that there are no structural or other (relevant) similarities and no simple referential relation between these. Assuming the risk of making the matter even more complicated, one can still say that historical interpretations are 'about' the historical world in some loose sense. Equally, it is reasonable to say that the phlogiston theories of eighteenth-century chemistry are 'about' the natural world although it is futile to look for the 'phlogiston' that would make statements about it true or false. The claim is thus not that 'presentations' could not refer beyond themselves or beyond a text; it is rather that they fail to refer to any unique corresponding entities. Further, neither does this position lead to a situation in which we could never prioritize different constructions on a principled basis. And certainly this does not lead to an 'anything goes' attitude. I cannot detail now why and how this is so, and therefore, I ask patience from the reader to continue on to the subsequent chapters.
The idea that historiography produces presentations, and not re-presentations, takes one directly to non-representationalism. The main advantage of this move can be seen immediately: it provides liberation from a rigid subject-object dichotomy that forces one to look for clear and determinable objects that historiographical constructions are about. And this in turn enables one to direct attention to philosophical concerns other than those relating to the ontological status and nature of these putative objects. Michael Williams has written that the idea of the 'representational' is symptomatic of Cartesian philosophy and implies that a representation is about or of something: a unicorn, Paris, or the square root of three, (William 2009, 19). It may be better to look for an alternative model to the kind that Dewey called the 'spectator account', at the centre of which representationalism lies. The spectator account implies that 'the thing to be known is something which exists prior to and wholly apart from the act of knowing' (Dewey 1929, 205). It might be said that 'narrative substances' are prime examples of the kinds of entities that do not exist prior to historians' construction. Similarly to Dewey, Richard Rorty defines 'antirepresentationalism' as a position, which 'does not view knowledge as a matter of getting reality right, but rather as a matter of acquiring habits of action for coping with reality' (Rorty 2011b, 1). Emphasis is thus on the process of construction and not on a reflection of what is given in subject- independent reality.
However, I would not characterize the position of this book as 'antirep- resentationalist', even though my intention is to outline a non-represen- tationalist account of historiography with regard to its most important scholarly products. Further, my intention is not to play the role of a general critic of representationalism and the concept of representation aside Rorty (although I also do take the mirroring metaphor to be unhelpful). My aim is more modestly to suggest that it is not reasonable to apply representationalism in the context of historiography with regard to its main knowledge products. To try to make sense of historiography through a rigid scheme of representationalism would be misleading and will only take one further from the central observations: interpretations are inferentially born and are constructions by nature. Although many modifications of representationalism are possible, one should not try to save the concept of representation at all costs, provided that there is a more straightforward 'presentational' option available.
In other words, I do not intend to outline an alternative comprehensive metaphysics and philosophy of language. And I have no general problem with representations, which may indeed work perfectly well in some contexts. My argument is that representationalism is not a reasonable commitment in the case of historiographical theses and interpretations. As a matter of fact, even on Rortyan pragmatist grounds, it seems permissible to continue to use the concept of 'representation' in some contexts - when it appears pragmatically justified and useful language for a particular context.
A potential problem is that if we give up on the idea of 'aspect' or 'presenteds' that are represented in historical writing, we also lose the central disciplinary contribution of historiography, that is, the 'views' or 'messages' that it produces - the narrativist insight of Chapter 1. If a book on Napoleon portrays him as an arrogant ruler, which would be its aspect, then it seems that we miss this and will be left only with the shells without the content, that is, a set of separate descriptions of Napoleon and the actual historical Napoleon, with nothing in between. This worry is understandable but nevertheless groundless. The non-rep- resentationalist solution does not do away with the central messages or theses of historical works. The essence of this suggestion is not to create additional imaginary worlds or assume that there are other kinds of objects that historians' theses must directly refer to. Provided that historiographical theses do not correspond (refer) to historical reality, it does not bring clarity to retain the representationalist language and the (quasi)realist approach, which forces us to match our representations with their 'shadows' in some abstract world. In other words, it is possible to accept that the books of history contain meaningful theses or messages, but that these theses or messages do not refer anywhere.
What is the nature of historiography then? How should we characterize its main products of knowledge if not in representationalist terms? The chapter began with a discussion of the different views of what historiography is and two suggestions were focused on. Many have seen historiography as a science but many others have taken it to be a form of art. I do not position my proposal on this axis because what matters are the discipline-specific features, not disciplinary status as such. I agree with Goldstein that historiography is a way of knowing and that the historian's account must be justified in terms of cognitive and not literary criteria (1976, 176; 181-182). The mistake that Goldstein (and also Tucker 2004) makes is that he thinks that historiographical 'constitution' (Goldstein 1976, 212) could be carried out merely on an evidentiary basis. 'Attending evidence' is certainly required, but it is not enough by itself. The non-representationalist suggestion of this book is that historiography is about reasoning for some theses and that the main contribution of a work of history is to provide an informal argument for or against a given thesis.
That a work of history contains synthesizing theses is something which I agree with the narrativists on, and it is indeed a valuable observation. While Ankersmit and others have suggested that we should see representation as a three-place relation, and some have even proposed a four-place operation (Giere 2004), my proposal is to go in the other direction. I wish to move from the two-place postulation to a flat, one-place proposal with regard to the main cognitive products of historiography. My view is that we can give up on the assumption that there has to be an object that that makes a 'presentation' of history true or false, or an object to which a presentation 'refers' to, or 'is about'. Historical writing contains arguments, or, to say it somewhat differently, a historical presentation in total amounts to an argumentative intervention.
The reader probably already senses how this step changes the vocabulary, and how it re-orients one in theorizing. Historiography, with respect to its higher-order knowledge contributions, is being identified as a discursive practice and not in relation to objects to be portrayed. By the same token it is a rational undertaking. For the narrativist, the historian is a kind of descriptivist storyteller. In my view, the historian is a critical reasoner. An argument (in general) can be said to be comprised of premises and their relations, intermediate conclusions and, ideally, the main conclusion that follows from the reasoning that precedes it. It would be a category mistake to suggest that arguments correspond to anything in historical reality; the past cannot be thought of as being structured like an argument with premises, conclusions and their relations, just as it cannot have narrative qualities, as White pointed out. Instead, arguments function like interventions in historical discourses. In this sense, we might see historiography producing argumentative speech acts with which the author tries do something, which takes me closer to Quentin Skinner's theorizing (e.g. Skinner 2002) than I would have expected a few years ago (cf. Kuukkanen 2008). The 'doing' of history involves bringing about a change in the existing historiographical discourse - perhaps to make people reject a certain thesis or accept it in some modified form, or to persuade them to endorse an altogether different thesis. One should note that the ways of persuasion and reasoning can be very diverse and may also include 'narrative persuasion'. This, in brief, is my suggestion regarding what historical works produce and what their main rationale is. The remainder of the book is intended to convince the reader of this view and provide a detailed account of it. Next, I will analyze how reasoning is manifested in the books of history, contrasting the narrativist view with an argumentative one.