The problems of representationalism
In this section, my aim is to discuss some of the problematic consequences that representationalism and the substitution theory of representations are associated with. In the subsequent section my claim is that we can construe an alternative suggestion, which manages to give a satisfactory account of historiography and its main knowledge contributions, but which avoids the problems of representationalism.
Before commencing a detailed critique, it is instructive to take a few steps back and consider the concept of 'representation' in general. What does it mean? On the most fundamental level, 'representation' is a two- place relation, creating a link between two variables: one that represents the other that is thus represented. There are several ways in which to understand this relation as we have already seen. The following are arguably the most obvious ones: (1) 'representation' can imply that one object is a copy of another in some sense; (2) 'representation' can be taken to mean that one object is a substitution of another; (3) 'representation' can be understood yet more widely to state that one object stands for or symbolizes another even though the former is not a replacement of the latter. In the case of historiography, the idea that a historical text could be a copy has been dismissed by the narrativist school. I concur with this. The text seems to have qualities and structures that historical reality lacks, something that White emphasized (White 1984, 24-259). Ankersmit aptly spoke about a 'morphological or structural difference' between the past and a narration (1983, 82).
Ankersmit's and White's argument that representation in the form of resemblance and morphological similarity does not work is well in tune with recent debates in the philosophy of science. One might say that the theories that rely on similarity and isomorphism are indeed the ones that try to capture the celebrated and criticized intuition that A is a representation of a target B if and only if A constitutes a mirror image of B, and it appears that the concept of representation in terms of similarity and isomorphism is in deep trouble also more generally in the philosophy of science. Mauricio Suarez (2003) argues that 'no theory that attempts to reduce scientific representation to similarity or isomorphism will succeed' (2003, 241).3
There are of course many suggestions for how the concept of representation should be modified. The pragmatically oriented philosophers share the belief that what seems initially attractive in the similarity and isomorphism accounts, that is, that the representation depends on the facts of the world alone and guarantees in this way the objectivity of scientific representation, is untenable.4 Their idea is that it is necessary to include the purposes, views or interests of enquirers in the account. Suarez speaks about the 'essential directionality of representation' and suggests that 'a necessary condition for A to represent B is that consideration of A leads an informed and competent inquirer to consider B' (2003, 237).5 It is notable that even the modified accounts of representation entail a one-to-one relation between the representation and the represented.6 This is challenging in the case of historiography. If A is a text on the Renaissance, B should be the historical phenomenon Renaissance. The point of the narrativist philosophy of historiography is that that the text constructs the historical phenomenon on the narrative level, which is thus something that cannot be discovered. In this case then, A constitutes B and therefore A can only refer to B, that is, to itself. And further, if B is accepted as a colligatory concept, it is does not seem clear whether it is possible to consider it as an independently existing single entity that can be referred to in the way that a name refers to a person, for instance. Now, it can still be maintained that A, the text, leads to a consideration of B, although A constructs B. Namely, A can lead one to think about events colligated under B, such as some paintings, furniture, ideas or scholars of the period, etc. The crux of the matter is that the text cannot lead to a consideration of the pre-con- structed historical phenomenon, because no higher-level phenomenon, such as the Renaissance, that is constructed of lower-level objects is independent of the historian's act. The discussion concerning the nature and reality of constructed phenomena will be left for Chapters 6 and 10, but it can already be said that the reality of either the Renaissance or the events it colligates are not in doubt. Further, the judgment on whether a representation could be isomorphic with or similar to its target object naturally depends on both what is identified as a 'representation' and what is seen to be the 'target object'. The subsequent chapters consider both these questions in detail.
We thus come to Ankersmit's idea that representation is substitution, something that is suggested by the meaning of the word 're-presentation'. His idea differs from that of the pragmatically oriented philosophers of science. According to Ankersmit, every painting or historical representation contains something unique, some 'aspect(s)' or 'presented(s)', which was also (in Historical Representation) called their 'personality'. Further, the postulation of an 'aspect' turns the two-place relation into a three-place relation. It is true that Ankersmit is not alone in suggesting that a simple two-place relation between 'representation' and the world as given fails to explicate the concept of representation adequately. The difference is that while most other suggestions would include something like purposes or agents (scientists) in the relational schema (e.g. Giere 2004), Ankersmit adds an extra object or object-like entity into the equation.
On first sight, Ankersmit's 'presenteds' or 'aspects' appear to be a cunning solution for retaining the representationalist account when it is clear that resemblance, similarity and isomorphic theories of representation fail. Provided that historical representation cannot be directly about the past, the problem is to know what these aspects are that historical representations are about. What kind of entities they are? Are they additional abstract objects in their own right? Does their postulation help us to understand historical texts better than normal interpre- tational practices?
One line of interpretation suggests that 'presenteds' are indeed independently existing entities and that they reside in their own abstract world. Consider the following paragraphs:
All objects of interpretation drag along with them the roots they have in the (imagined) reality they represent. (Ankersmit 2012, 52)
There is always an (imagined) world or reality that representations, whether texts or paintings, are "about" and of which they are more or less "true" (where I take the word "true" here in the vaguest possible way); and whoever interprets texts and paintings without taking this into account will inevitably be like a sailor without a compass. (Ankersmit 2012, 52)
It is worth noting that the reality of the past itself is not at stake here, despite the connotations associated with the word 'imaginary', since historical reality is one part of Ankersmit's tri-partite account. One should thus avoid attaching any psychological connotations to this notion of 'imaginary', If we read these passages at face value, they would seem to suggest that the world of 'presenteds' forms an abstract world of its own. One might see this world as parallel to the realm of Platonic heaven, Fregean propositions or a Popperian Third World, that is, entities that really exist but in a non-material and an ahistorical form. Perhaps Fregean propositions and concepts provide the best analogy for 'aspects' and 'presenteds' as they also connect linguistic entities and the external world. Frege claimed that one should not assume that our subject terms refer directly to entities in the world or that their meaning is given directly by their external references (in contrast to the causal theorists of reference, such as Kripke (1980) and Putnam (1975)). For it seems entirely possible that the same person believes that the 'evening star' is not the same as the 'morning star', even though both refer to the same object, that is, the planet Venus. Frege thought that the 'senses' (or 'concepts') of these expressions are different although their reference is not. We can make similar observations in the context of belief ascriptions. Jack may well believe that 'Clark Kent didn't save the world' but that 'Superman did save the world' without any contradiction if we assume that there are two different propositions to which these sentences refer and that Jack does not know that Clark Kent is Superman. The world of senses and propositions as a fully disembodied world outside history would seem to help us to make sense of the said between our language and the actual embodied material world.
Now, perhaps Ankersmit can be seen as joining this venerable tradition of the philosophers of language with his suggestion that between textual representations and historical reality there are similar disembodied 'aspects' or 'presenteds'.7 Frege's suggestion was that these abstract entities are needed to make sense of the said when the content of the said differs, but when the references in the external world remain the same. The appropriate test for this analogy would then be to see whether the postulation of a 'third world' has similar benefits for the intelligibility of our historical discourse. Namely, in the case that two historians refer to the same historical (material) world with different kinds of representations, one could not make sense of a historian's interpretations, the said, without postulating two 'presenteds' as the corresponding abstract entities. I must admit that I find this interpretation interesting. Further, it may perhaps be seen to link to a relatively recent discussion on abstract social ontologies.8
However, the 'two-world' suggestion is not without problems either philosophically or as a textual interpretation of Ankersmit. To begin with the philosophical problems: a metaphysical implication is that the postulation of 'aspects' duplicates historical ontology. That is, we would have, first, the world of representations and, second, the world of 'presenteds'. There would thus be two levels and two ontologically separate worlds. On one occasion Ankersmit writes that:
Each representation, then, carries its own represented or aspect9 along with itself - much in the way that we are each accompanied by our shadow on a sunny day - and each of these representeds is indissolubly linked to one, and only one, particular representation corresponding to it. (Ankersmit 2012, 72)
In the ontological interpretation, this would seem to mean that we view ourselves and our shadows as objects of equal standing. Therefore, if we think that both 'representation' and 'presenteds' are objectively existing entities, we introduce a two-tier ontology (in addition to actual historical reality, the third tier). It thus seems that the 'being about' postulation entails that there is a (perhaps linguistic) level of representations and that there is a separate level of objects, presenteds that the former are about. Each historiographical representation would give a birth to a new independent abstract object. In brief, the 'being about' stipulation forces us to duplicate and inflate our ontology.
The upshot is that although the substitution theory of 'representation' is a theory about substitution, or the identity10 of the past and present, it unwittingly suggests some kind of correspondence relation between subject and object, that is, between the one that re-presents (such as language, text or 'representation') and the one that is re-presented. The analogy between bodies and shadows mentioned above entails that there should be such a relation of correspondence. This suggestion is perfectly intelligible when we pause to consider this relation. A shadow can be expected to be isomorphic with the body; it reflects the two hands, one head, two legs, etc. that the body possesses. This analysis provides us another view onto the question of how and why the language of 'representations' inadvertently returns us to the realist 'language game' of mirroring, even though the world of objects mirrored is not the 'real' human-independent world of the realist but the imaginary world of historiographers.11 An ontological interpretation of the abstract world would commit us to some kind of realism in idealism: it requires one to postulate a real world - albeit that this real world is dependent on the human mind and its creations, at least until they are created.12 Perhaps, after their creation, 'presenteds' do gain independence from the human mind but even then they would not exist in material or physical form but as a world of thought products.
It may be necessary to simply bite the bullet here despite such metaphysical complications if the postulation of an extra realm is required to make sense of historiography and its content. In this case one would expect clear practical gains from the postulation. I leave the discussion of the pragmatic value for the following section. Now it is necessary to return to the textual interpretation of Ankersmit's philosophy and ask whether he actually commits himself to the existence of an abstract independently existing (imaginary) world of 'presenteds'. Some of the passages displayed above seem to suggest so, but in actuality Ankersmit denies this and, in the end, puts forward a 'one-world solution'. He states that 'aspects' or 'presenteds' have the same ontological status as past reality itself: 'a representation's presented is an aspect of things and hence part of the world itself' (Ankersmit, 2012, 105). Ankersmit's view that they belong to the same 'inventory of the world' and that there is no 'ontological hierarchy' was already mentioned earlier (p. 48; Ankersmit 1998, 50). An analogy that clarifies the status and meaning of 'aspects' is that of the front and back of a person, which are both part of one and the same person.13 It is striking that this suggestion entails the cumulative nature of historical knowledge. If all historical books, say, about Napoleon present us with an aspect of Napoleon, can we think that all the aspects increase our knowledge of Napoleon in such a way that we gradually move toward a more complete picture of Napoleon? The 'complete picture' can of course be understood as an ideal limit, perhaps never actually reached. Nevertheless, there would thus seem to be genuine accumulation of historical knowledge in the form of collecting more aspects or presenteds of some historical phenomenon. Some expressions of the later Ankersmit indicate that this is indeed possible. Note the following illustration concerning the role of aspects through reference to a geographical discovery: 'Think of the first explorers of the American continent in the early 1500s: each of them discovered only an aspect of the continent, but then mapmakers pulled all their individual discoveries together; only then did a new thing come into being, namely, the American continent' (2012, 155). Might one thus view historical objects as undiscovered continents that we come to understand more and more in the course of time and by virtue of diligent exploration? Ankersmit says that 'nobody will doubt that there is progress in the discipline of historical writing: we know far more about the past than ever before'. Even more revealingly, he indicates that one day we might reach 'a universal consensus' on historical phenomena such as the Renaissance. This would be the situation in which 'the historical (propositional) truth about the Renaissance would have been discovered' (2012, 84-85, 228).
The prospect of the accumulation of narrative knowledge in this sense is surprising and in stark contrast with what Ankersmit argued in his earlier texts. At the time when Ankersmit still identified himself as a
'postmodernist', he compared the increasing pace with which historical interpretations are produced to 'intellectual alcoholism' (1989, 138) and agreed with Jonathan Culler's (1983, 90) statement that in historiography, 'paradoxically, the more powerful and authoritative an interpretation, the more writing it generates' (Ankersmit 1986, 25). He thus did not expect that historical writing would bring an end to historical debate and result in consensual clarity and definitive knowledge regarding some historical phenomenon (Ankersmit 1985, 25; similarly White 1978, 89). The later Ankersmit is clearly more optimistic about this. If 'presenteds' are inherent parts of an object presented, that is, 'aspects', then adding more of them arguably yields a more accurate account of the object itself. Indeed, Ankersmit defines 'representational truth' 'as what the world, or its objects, reveal to us in terms of its aspects' (Ankersmit 2012, 107).
We come to the following picture. Representations are about 'presenteds' or 'aspects', which themselves belong to the inventory of the historical world, that is, the represented reality, itself. I must admit that I feel uneasy with the realist and cumulativist undertones that this constellation implies. First, if aspects are part of historical reality itself, this would seem to be a reversal of the early Ankersmit's commitment (in Narrative Logic) to the colligatory nature of historiography. Second, it is not clear in what sense aspects could be part of the past itself. What does it mean to say, for example, that an aspect of Napoleon (as re-presented in a book of history) is a part of the historical world itself? Was it not that an aspect is a constitution and colligatory contribution by the historian? It seems to me that any particular interpretation owes more to the 'subject-side' than is recognized here, as I will explain in Chapters 6 and 10. And it is not clear how a historian could discover a 'part-of-the-world aspect'. In relation to this, one is puzzled as to whether the number of aspects is limited. Further, as Ankersmit expresses it,14 if representations are about 'aspects' then it is difficult to say what they are 'aspects of'. Does one need to commit to a pre-given set of objects (in the historical world) about which all historiography and their interpretations (via their aspects) are? The cumulative view creates an expectation that historiographical inquiry comes to an end and that all aspects of an inquiry are re-presented in that end. Finally, although there is no commitment to the 'third world' of presenteds, the constellation reproduces the subject-object dichotomy, that is, the basic conceptual opposition between representation and represented. For Ankersmit, represented reality is 'objectively given' and an interpretation is 'an activity of the subject' (Ankersmit 2012, 51). I won't go further into the metaphysical intricacies of what these objects might be, but it is obvious that much more philosophical groundwork is needed to make the view viable. At the very least one would expect clear practical benefits to emerge from this kind of problematic postulation.
In the next section I outline a principled alternative to the representa- tionalist account. Before that there is still one more task ahead, however. At the beginning of this section I suggested that one possible sense of 'representation' is that one object stands for another. An example of 'standing for' could be a member of parliament, who 'stands for me' in the parliament, but is nevertheless not a substitution of me. Another way to express this idea is to think of two religious groups and their symbolism. Typically an image of a saint stands for the saint in the absence of the real thing, but it cannot arguably be said to be its full-bloodied substitution. One still needs and craves the real saint. By contrast, the urge to destroy the images of God by iconoclasts may be said to reflect the fear that the images would take the place of God as a full substitution. Baudrilland writes that iconoclasts predicted the 'omnipotence of simulacra': 'that God himself was never anything but his own simulacrum' (1994, 4). They thus feared the death of divine referential and that the worshipping of God is transferred to the veneration of the simulacra or image of him (cf. Ankersmit 1994, 189-192).
Could we apply the 'standing for' sense of representation and see historiography in representationalist terms in this fashion? Could we say that a historical text 'stands for' the past? Undoubtedly we could, but this is too easy a solution at the same time. Namely, anything can be said to 'stand for' anything, if we just stipulate so. The meaningfulness of that stipulation is another matter. If I state that my coffee mug stands for an omnipotent being, my claim would be understandable linguistically, but would probably be taken as senseless nevertheless. I accept that a historical text 'is about' the past in some loose manner, but it does not make good sense to claim that a historical text stands for the past. In any literal sense, a text is a rather different kind of thing to the past, which makes it an unsuitable symbol of the latter. Most importantly, this postulation of 'representation' does not in any case change the substantial issue, which is that a historical text cannot be isomorphic with historical reality. They are structurally very different, as they are by their qualities. How could non-representationalism help us here?