How complex is a lemon?

Early on​ in his Philosophical Investigations, Wittgenstein imagines an interlocutor who claims that every word in language signifies something – where ‘signifies’ means something like ‘names an object’. Wittgenstein gives an indirect assessment of that claim by discussing a second, analogous one. He imagines someone saying: ‘All tools modify something.’ Sounds plausible: a hammer patently modifies the position of a nail, and a saw the shape of a board. But what is modified by a glue pot or a ruler? Well, with a bit of imagination and a lot of intellectual charity, we could find something they might be said to modify – the temperature of the glue perhaps, or our knowledge of a thing’s length? But, Wittgenstein asks, what would be gained by this assimilation of expressions?

The problem is that each step we take in extending the reach of ‘modification’ to accommodate apparently resistant cases forces us to attenuate the meaning it has when we apply it specifically to hammers and saws. If temperature and a mental state are to qualify as possible objects of modification just as easily as the position of a nail or the shape of a board, our definition of what counts as an act of modification will become increasingly loose and baggy, and so calling something a ‘modifier’ will tell us less and less about how it actually works. At the limit, we may be able to accommodate every imaginable kind of tool within our ‘modification’ theory, but only at the price of evacuating the theory of any content. Rather than revealing that the underlying essence of tools is that they modify something, we have simply recast ‘modifier’ as a notational equivalent for ‘tool’. As Wittgenstein puts it, we have said nothing whatever.

This is the trouble with philosophical theories of everything, whether it’s everything in a given domain (tools, say, or language), or absolutely everything. It isn’t so much that their generality increases the risk of incorporating erroneous claims; it’s that they risk failing to make a claim at all. This isn’t necessarily a problem for scientific theories of everything: they have a specific purpose – to unify gravity, electromagnetism, and the strong and the weak nuclear forces – that concerns differences within the domain of material reality. As Graham Harman rightly points out, this kind of endeavour is problematic only when its proponents present it as a genuinely universal theory of absolutely everything, for this amounts to presenting a project in physics as if it were ipso facto an exercise in metaphysics – an account of the structure of reality as such (not just physical reality). Unfortunately, Harman also thinks that philosophy can provide what physics cannot. His object-oriented ontology (OOO) is explicitly presented as an original and compelling metaphysical theory of everything, which can serve as one central contribution to a broader but loosely-woven movement sometimes (though controversially) called ‘speculative realism’, to which the writings of Ray Brassier, Quentin Meillassoux, Levi Bryant and Timothy Morton also belong. Since speculative realism is one of a number of philosophical projects competing to occupy the dominant position in contemporary European thought vacated by the death of Derrida, and has acquired some influence in cognate areas of culture such as art and architecture, it’s worth asking whether Harman’s theory of everything avoids the familiar problems encountered by its predecessors.

If someone claims that his ontology is object-oriented, he implies that other ontological approaches are not; then everything hangs on what is meant by ‘an object’. Harman’s opening attempts to specify what he means by it look promising. He tells us that physical objects are only one kind of object, and that for OOO objects exist at numerous different scales and in many different modes. Thus, for OOO objects include the electron, the Dutch East India Company and the galaxy; fictional entities such as Sherlock Holmes or the conspiracy to assassinate President Kennedy; and forests and married couples (whose status as objects in their own right is not impugned by the objecthood of the individual trees and people that make them up). For Harman, an object is simply anything that is resistant to reduction: anything whose autonomous, unitary nature and existence cannot be explained away by being reduced to its component parts, or to its effects on the world, or to its relations with other objects. So OOO defines itself by contrast with philosophical theories that have reductionist tendencies of any kind: it is committed to a non-exclusionary ontology by virtue of the apparently minimal demands it places on any candidate for objecthood.

On one way of taking these claims, OOO has a lot going for it. For it appears to embody a healthy principle of ontological charity: begin by assuming that the entities which appear to populate a given domain really do exist, rather than assuming that they must be explained away. And one would expect its proponents also to assume that the mode of existence of the entities in one domain need not resemble that of entities in another. If the distinctive nature of a marriage cannot be reduced to the distinctive nature of an individual human being (because marriages have properties that people can’t have), then why should the distinctive nature of a marriage resemble that of a forest? And indeed, when Harman turns his attention to social theory, he treats the Dutch East India Company and the American Civil War as valid objects of sociohistorical investigation, with each having properties that differentiate it from the other (as joint-stock companies differ from conflicts), as well as sharing properties that differentiate both from electrons and from forests.1

Even at this early stage, however, we might worry that Harman’s strategic orientation towards objects is less distinctive than it seems. It’s not clear, for example, that his ontological open-mindedness in social theory generates anything terribly original in the way of insights into the nature and history of its distinctive objects. His disquisition on the American Civil War in Object-Oriented Ontology is of course only an illustrative sketch; and it is certainly enough to show that his approach raises serious questions for his ontological competitors. But it is equally plain that those competitors are other ontologists rather than historians: what interests Harman is primarily devising new ways to conceptualise the objects of historical investigation.

In the case of the Civil War, this finds expression in his use of neologisms such as ‘symbiosis’. This term derives from biology, where it can refer to a living entity’s transformation by virtue of its incorporation of another such entity. Given OOO’s generous conception of an object, Harman feels licensed to apply the term to the realm of history, by claiming for example that Grant’s promotion to Union commander was the final symbiosis of the war. By this, he means that it was the last significant internal transformation of that historical object, and one effected by that object’s incorporation of another such object (presumably the object ‘Grant’s promotion’). But apart from a certain rhetorical frisson conjured up by the suggestion that historical phenomena have a life-devouring life of their own, it’s hard to see how conceptualising matters in this way actually helps the working historian to determine whether Grant’s promotion really was the war’s last significant development, as distinct from giving her a dramatic new vocabulary in which to articulate both the problem and its solution.

More important, maintaining one’s initial respect for a given domain’s apparent ontology depends on actually doing the hard, detailed work needed to show that it is truly indispensable. So OOO must allow for the possibility that this work will not succeed. It may turn out that the apparently distinctive population of a given domain can be reduced or explained away; then those ‘objects’ won’t count as OOO objects. So the real range and variety of object-oriented ontology is always hostage to its reductionist opponents’ ability to make good on their aspirations; and that means that OOO cannot in fact be defined in contrast with reductionism, but only with unjustified or failed reductionism (exactly the kind that reductionists would also reject). Being a reductionist and being a proponent of OOO are thus perfectly compatible: the sole distinguishing feature of OOO turns out to be nothing of the kind.

What we’re left with is the following principle: don’t assume in advance that the unitary phenomena that appear to populate any given domain of reality are reducible, but don’t assume in advance that they’re not. In other words, don’t let your assumptions get in the way of seeing what is really there. Good advice, Polonius, but hardly a radical theoretical breakthrough.

Harman doesn’t exactly own up to this. He does, however, say that adherence to a generous or flat ontology is only OOO’s starting point, and that it would be profoundly disappointing if that were all there was to it. His readers are quickly whisked to a very different theoretical terminus. According to Harman’s final view, there are in fact only two kinds of object: real and sensual, each with its own properties, or qualities. In other words, he advocates a four-fold ontology of real objects, real qualities, sensual objects and sensual qualities, and he further claims that there are four kinds of rift or tension between objects and their qualities so understood.

Real objects present themselves to us as possessed of various qualities that distinguish them from other objects, many of which can fluctuate without compromising our conviction in the continued existence of the object (as when a green tomato turns red). Intellectual effort can identify which of these properties cannot so fluctuate: they are the ones that determine the essence of the object – what makes it the kind of thing it is. But since this is an intellectual construction from the deliverances of specifically human senses, what it really gives us is the sensual object – an object that exists only as the correlate of an act of consciousness. The real object, the apple itself rather than the apple as we encounter it, together with the real properties it possesses, thus withdraws behind sensual objects and their sensual properties.

Emphasising this withdrawal is supposed to make OOO strongly realistic, affirming the genuinely independent existence of objects in a way that Harman thinks no other approach can match. But it is equally important to him that the real and the sensual are internally related in various ways, and so form a unified if conflictual ontological field that is of universal application. The real object must have real qualities if its difference from other real objects is to be realised; it registers its presence by virtue of the sensual qualities it emits to and for human consciousness; the sensual object takes shape by virtue of our intellectual operations on those emissions; and that object’s experiential constancy intimates the necessarily inaccessible presence of the real properties, and so of the real object whose properties they are.

This four-fold ontology grounds Harman’s account of space and time, although he barely sketches in how it does so here.2 But it also generates two other, more richly elaborated, metaphysical proposals that are worth dwelling on. The first results from the apparently innocuous claim that a truly realistic ontology must not privilege relations between real objects and subjects over relations between real objects. But if all such relations are on an equal footing, and real objects withdraw as thoroughly from other real objects as they do from subjects, then – Harman claims – real objects in relation to other real objects must also encounter some intermediate, inadequate proxy for their real natures. The fire may not think about the cotton it burns, or feel guilt about its effect on that cotton, but these are mere differences of degree with the subject-object case. Since the fire can no more make direct contact with the real cotton than a subject can, that contact must be mediated somehow, and Harman is willing to call what results from it a sensual object. That result is, according to OOO, simply a more complex object created by the encounter between two real objects, just as a subject’s perception of an object is itself a new object (containing the perceiver and the perceived object as parts). Harman is reluctant to call this position ‘panpsychism’; but that is simply because his claim is not that to exist is to perceive, but rather that to relate is to perceive, which he thinks is better called ‘polypsychism’.

Harman’s second proposal concerns aesthetics, which he presents as the root of all philosophy, and as driven by insights he derived from the Spanish philosopher Ortega y Gasset. On this account, metaphor is the heart of art; and the heart of metaphor is a complex transaction between subject, real object and sensual object. Suppose a poet says: ‘The cypress is the ghost of a dead flame.’ This metaphor is experienced by the reader as positing a revelatory identity between (some of) the sensory qualities of a flame and the reality of the cypress itself. But since Harman (unlike Ortega) thinks the real cypress is necessarily inaccessible, he believes that we can only make sense of this by positing some other real object that is always present in any aesthetic experience: the subject of that experience, the reader. It is we who stand in for the absent cypress and support its freshly-anointed flame-qualities; we are method actors playing a cypress playing a flame. Harman concludes that metaphor is inherently theatrical: the experience of a good one amounts to the generation of a new complex object formed from the reader posing as a cypress and the sensory qualities of flame. But this object is not a cognitive phenomenon, not the advent of knowledge about a pre-existing object. It is neither the cypress nor the reader.

It’s amazing how quickly we’ve moved from an anodyne reminder that we should check our ontological privileging to an enthusiastic investment in polypsychism and in art as metaphysically revelatory metaphorical theatricality. It’s also quite striking that a supposedly realist philosophy, which rejects from the outset any form of idealism (which typically claims that objective reality must be understood in relation to subjectivity), finds itself making the human subject the primary object and actor in art, and reinterpreting relations between objects as primitive versions of the kind of relations subjects have with objects. But even if we take each position on its own terms, neither looks attractive.

Harman’s polypsychism rests on a refusal to privilege subject-object relations over object-object relations, which he equates with the belief that all relations are on an equal ontological footing. But one could believe that object-object relations are all real, genuinely existent phenomena (that objects really do encounter other objects as well as subjects, that they disclose aspects of their nature in both cases, and that both kinds of encounter are equally worthy of the philosopher’s attention), and also believe that the two types of encounter are very different in kind, and in particular that subject-object encounters have ontological possibilities involving sense and intellect that object-object encounters lack. Identifying essential differences between kinds of relation doesn’t discredit either kind; and if the ontological possibilities are more numerous and richer in one case, then acknowledging that fact does not in itself amount to privileging that kind of relation.

As for the theatricality of metaphor: does Harman’s account affirm the inaccessibility of real objects or not? First, he tells us that the real cypress cannot be accessible to us because it’s real; then that the subject, which is a real object, is always genuinely and truly present in aesthetic experience, and so can substitute for the absent real cypress; and then that the only real object involved in such experiences is not the subject but the newly amalgamated object of which it is a part. And if this new object is not a way of gaining access to any pre-existing object, how can it be the root of all philosophy: that is, provide a privileged vantage point for apprehending the underlying reality of things? But if it does offer such insight, then surely reality must be accessible after all.

Such unclarities return us to the larger puzzle of how we got from a quietist conception of an object as whatever resists reductionism to the claim that there are exactly two kinds of object and two kinds of quality that yield precisely four distinct but internally related rifts that together generate the flat, pluralist ontology of everyday life. The crucial motivation is Harman’s belief that we can properly acknowledge the reality of real objects only if we deny their accessibility to human subjects; for this is what leads him to introduce his second category of sensual objects, and to define them as necessarily veiling real objects. But why should the accessibility of objects threaten their reality? After all, we ordinarily acknowledge a variety of limitations in our modes of access to the real world. We think that the apple on my desk continues to exist when I’m not perceiving it; that I may misperceive some of its attributes under certain conditions, and that others may or may not be in a position to correct me; that no matter how much knowledge of it we acquire, there may be more to discover; that our best current theory of its nature may turn out to be wrong, either in small details or large (if, for example, a paradigm shift occurs in the plant sciences, revealing that we have radically misconceived its nature); and even that some aspects of reality (material, psychological, moral) might resist our understanding indefinitely. These ideas add up to what looks like a pretty robust acknowledgment of the autonomy and independence of real objects; but they also allow for the possibility that we can sometimes grasp their real properties and nature. Although we will often get it wrong, we can also have good reason to believe that at least some of our representations of this apple are accurate at least some of the time, and that – with a fair wind, great care and a huge amount of collective theoretical and practical labour – we can come to know a great deal about it.

Harman never explains why this kind of sober acknowledgment of our finitude as knowers should not satisfy our realist impulse; why allowing even the possibility of genuine knowledge of a real object, however qualified and conditioned and subject to revision it may be, is a threat to its independent reality. But what seems to worry him above all is the thought that all our modes of access to objects are necessarily indirect.

This is not how we normally use the phrase ‘indirect mode of access’. If my belief that a particular apple is green is based on what someone else tells me, then my access to it might well be said to be indirect; and this is worth saying, since reliance on testimony raises additional specific (though not insuperable) problems in grounding knowledge-claims. But calling testimony an indirect mode of access has force only if other modes of access count as direct (and so don’t generate the same problems, even if they face real but not insuperable challenges of their own). In this case, the obvious candidate would be seeing the apple, right in front of my properly functioning eyes in good light from the window. Moreover, we don’t assume that an indirect mode of access to objects is incapable of disclosing the truth about things simply because it’s indirect. Indirect access is still access, and can’t testimony sometimes be accurate?

Harman nevertheless believes that neither sort of encounter really is direct, and indeed that no conceivable mode of human access to objects could be anything other than indirect. And he assumes that if that is so, these encounters cannot offer genuine access to the objects themselves. He is led in this direction because he thinks that – as he puts it in various places – to encounter an object in any way is to translate that object, hence to transform it, hence to distort it; as a consequence, he claims, we are everywhere and always surrounded by fictions:

Any real orange or lemon, as I perceive it, is a vast oversimplification of the real citrus objects in the world that are submitted to rough translation by the human senses and human brain … All of the objects we experience are merely fictions: simplified models of the far more complex objects that continue to exist when I turn my head away from them, not to mention when I sleep or die.

So many dubious ideas are compressed into these remarks that it’s hard to know where to start unpicking them. To begin with, even if our purported model of a lemon is too simple to capture its full complexity, a simplified picture of an object can still get some things about it right, just as a rough translation can capture some aspects of a remark’s meaning perfectly well. A simplification leaves things out, but it doesn’t necessarily introduce things that are not there, so it need not be either a distortion or a falsification; and it can always be further complicated so that it captures more of its object. Or does Harman think there could be no such thing as a fully accurate model of an object because the object is infinitely complex and there cannot be an infinitely complex model? Setting aside worries about what ‘infinite complexity’ might mean, this would amount to the bizarre idea that anything less than an absolutely complete representation of an object must absolutely misrepresent it.

If we shouldn’t equate ‘simplified’ with ‘false’, why should we equate it with ‘fictional’? Harman has already offered Sherlock Holmes as his model of a fictional being; so if our model of a lemon really were a fiction, then surely nothing at all would correspond to it in the independently existing physical world (so it wouldn’t even be a rough translation of something in that world). For on the face of it, Sherlock Holmes is neither a vast oversimplification of a real fictional object, nor a rough translation of one, nor a distortion or falsification of one. Indeed, it’s hard to see how Harman’s supposedly universal distinction between real and sensual objects even applies in Holmes’s case. Do we have fictional representations of this fictional being, and if so what are the real properties of the real (fictional) being of which they are (fictional) representations? Is there a reality to him that withdraws definitively beyond our various means of access to him?

Harman’s policy of treating ‘simplified-translated-distorted-false-fictional’ as a chain of synonyms conflates importantly distinct ideas, and thereby implicates the respectable ideas towards the beginning of the chain with the flaws we associate with those towards the end. But never mind the gratuitous introduction of fictionality as the chain’s final prejudicial link; why are we talking about ‘oversimplified models’ of lemons in the first place – as opposed, say, to our best available conception of this real citrus object?

The passage I have quoted, with its emphasis on the senses and the brain, strongly suggests that Harman is endorsing a natural scientific account of how human perception operates, then drawing metaphysical conclusions from it. In doing so, however, he implicitly privileges the case of material objects, rather than maintaining the assumption of ontological plurality to which he claims to adhere. He also treats this theory as a truthful account of the reality of human perception, when by his own lights it must be entirely fictional. And even if that account were true, it does not self-evidently legitimate the metaphysical conclusions he draws from it. Many philosophers would deny that human beings are really model-building theorists of their own experience at every moment of their lives, as distinct from, say, engaging in such refined intellectual practices only in laboratories and research institutes. Indeed, for the phenomenologists Harman most respects, the assumption that modelling or representing is our most basic relation to the world is a distinctively modern fallacy. And more generally, erecting metaphysics on the basis of physics is an exercise in naturalistic reductionism: it is a version of the scientism that Harman begins his book by denouncing.

There may be another influence lurking in the background here: Nietzsche, and in particular his early essay ‘On Truth and Lying in an Extra-Moral Sense’, which has had such a fateful influence on the philosophical traditions in France and Germany to which Harman is indebted. Nietzsche blends scientific analyses of perception with Kantian metaphysics to suggest that because the impacts of material objects on our senses generate a chain of different kinds of physical and mental event (from retinal images to nervous impulses to brain activity to mental representations), their initial input might be thought of as undergoing translation between a series of languages. This is why he claims that truth is a mobile army of metaphors: because our representations of the world are the result of multiple mediation, assuming their veracity involves imputing an identity between a series of phenomena that are patently not identical (just as our earlier metaphor posited an identity between the cypress and the flame). Hence, they cannot but falsify the reality they purport to represent.

Unfortunately,​ Nietzsche’s metaphorical transposition of the idea of a metaphor from the domain of language to that of human perception and thought occludes more than it reveals. For it couldn’t possibly be the case that each element in the sequence of events initiated by an object’s impact on a subject is, or has to be, identical with its predecessor in order to do its work. If that were so, there wouldn’t be a sequence of distinct events to do the work; and if that sequence normally results in representations of the world that allow us to move around successfully within it, then presumably it preserves the crucial information across these transformations. Likewise, the notion of a representation declares on its face that it re-presents its objects in the medium of a language, rather than claiming to be identical with those objects. ‘This lemon has one leaf on its stalk’ may be true or false. If it’s true, that won’t be because the claim and the fruit are one and the same; and if it’s false, that won’t be because it is non-identical with its object – it will be because it misrepresents the stalk.

Harman actually regards Heidegger as his primary ‘Continental’ authority for both the four-fold ontology of the object and the inaccessibility of the real. Heidegger’s later work certainly offers a ready (if deeply obscure) source for the figure of the four-fold; and he always emphasised that something about the real Being of things necessarily withdraws from us. However, Harman tries to derive this latter insight solely from Heidegger’s early ontology of tools, and in particular from his valid but modest point that tools function smoothly as such primarily by virtue of sinking into the background when they are employed, thereby allowing the end-product or goal of the relevant activity to occupy the foreground.3

But this self-concealing tendency does n0t on the face of it lead Heidegger to think that he is constitutionally debarred from giving a detailed analysis of the ontology of equipment, or of the non-objectual meaning-informed environment within which alone (he claims) tools can exist. And nothing in that analysis entails that genuine knowledge of the hammer, regarded as a material object as opposed to a tool, is impossible. On the contrary, that’s the kind of knowledge of it that we naturally seek when, for example, a hammer is damaged, and so can no longer function smoothly as a hammer until it is repaired. What Heidegger is trying to show is that the environmental conditions which make it possible for us to grasp a hammer either as a tool or as a material object cannot themselves be analysed in the terms appropriate to such encounters. At this deeper ontological level we do indeed find an enigmatic dialectic between disclosure and withdrawal. But that dialectic is what makes it possible for us to encounter objects as they are (or to fail to); it does not demonstrate that we can never do so.

These are deeply contested issues, of course. But Heidegger is at least not as straightforwardly vulnerable as Harman is to the following objection. If real objects and their properties are essentially inaccessible to us, how could we grasp – let alone articulate – an object-oriented ontology that not only posits their existence but delineates their turbulent four-fold nature? Harman never addresses this worry directly; but his defence can’t be that OOO is itself an exercise in metaphorical rather than descriptive language use. For insofar as the OOO view of metaphor is internally stable, it disavows the idea of metaphorical penetration of reality’s sensuous veils, regarding the experience induced by a metaphor as just one more complex object (hence no more ontologically revealing than states of knowledge are).

Whatever its proximate sources, the picture shaping Harman’s thinking only needs to be plainly articulated for its peculiarity to become apparent. His conviction about the inherent inaccessibility of reality seems ultimately to rest on assuming that genuine knowledge of an object would have to take the form of becoming wholly and fully identical with that object: fusing with it, actually realising the theatrical method-acting aspiration that he claims is internal to the experience of metaphor (but also claims is not truly realisable). This is a strangely idealist assumption for a putatively realist ontology, and one which would abolish the independent reality of both subject and object if it were realised. But of course, knowledge cannot rest on any such assumption, because it is a sheer fantasy. It evacuates the idea of knowledge of any content, since neither we nor Harman have any idea what it would mean for a subject to become the object of her knowledge.

In cleaving to this fantasy, Harman implies that the fact we humans must acquire knowledge of the real world, and so must employ a variety of physical, mental and cultural means or media in so doing, engenders a kind of metaphysical incarceration. OOO treats the (embodied, rational, social) reality of human animals as if it were a mode of exile from reality – as if the human condition were a punishment. Rather than despising that condition or despairing over it, philosophy carried out in a truly realistic spirit should aspire to get it properly into focus.