Revisiting the Notion of Vicarious Cause: Allure, Metaphor, and Realism in Object-Oriented Ontology

1 Introduction

Contemporary continental philosophers seldom discuss causation as it is in itself because they tend to remain committed to the claim that one cannot examine the nature of the world independently of the way it is correlated with human modes of access. For instance, Immanuel Kant famously listed “causality” as one of the categories of understanding, and his successors such as Husserl and Heidegger effectively treated the question of mind-independent reality as a pseudo-problem nullified by the more interesting question of human access. And philosophers who do not subscribe to correlationism still tend to take the fact of causal relations for granted. The most significant exception to these generalisations is Graham Harman, whose Object-Oriented Ontology (OOO) presents a speculative, “democratic,” analysis of causation where philosophy has the remit to attempt to understand what causal relations are independent of their relations to humans or gods. Nonetheless, for Harman, objects on their own are causally alienated from one another, such that real causation between two objects must be a function of a third superordinate object containing the two. In this way, Harman’s OOO resembles a version of “occasionalism,” since the superordinate objects are clearly analogous to the occasionalist’s God acting as a sole causal mediator.

The goal of this study is to reconstruct Harman’s concept of vicarious cause and present its major motivations as well as the tensions inherent in Harman’s attempts to elucidate it. Along the way, we will show how the three Harmanian concepts of sincerity, allure, and metaphor are key to understanding vicarious causation. Our objective here is not to rebut or refute Harman’s thinking about causality, but to reassess it in a way that we hope encourages further development. To do this, we first explain how OOO differentiates between what it calls “local” and “global” occasionalism. We then turn to Harman’s early model of allure and later models of metaphor, showing how an object-oriented epistemology is needed to supplement the account of “sincerity,” and how his notion of “allure” speculatively extends into a new object-oriented form of teleological causation. We then further develop his account of metaphor to the point where we can raise fruitful questions about how to integrate it into Harman’s metaphysics.

2 Local and Global Occasionalism

Harman’s object-oriented approach rests on the defence of his fundamental claim that each and every real object is thoroughly irreducible and autonomous or “vacuum sealed,” to the effect that all forms of direct causal contact are excluded at the outset.^[1] This line of reasoning not only cuts against the phenomenological tradition in continental philosophy, but also against the grain of new materialism, as well as assemblage and actor-network theory since all of these movements emphasise the primacy of an entity’s inter-objective relations. For most contemporary continental philosophers, direct causal relations are at best assumed at the outset rather than having to be explained. Against the backdrop of such positions, Harman’s view has been criticised – including by those who otherwise share OOO’s enthusiasm for objects – for its alleged view that objects do not interact at all, thereby leading Harman into the impasse of inter-objective solipsism, a monadology without windows but also without the mediational help of God.^[2] We believe this critique misrepresents Harman’s actual view for, as we shall show in some detail, he has often insisted that the clear fact of change and alteration necessarily entails that objects must be capable of (indirect) affect, for otherwise every entity would simply reside in its own everlastingly isolated ambit.^[3] Something close to a dialetheia or true contradiction is at work here; while it is true that objects do not interact, it is simultaneously also true that they do, albeit indirectly.^[4]

This non-relational relation prompts Harman to express his own view of the question of causality by wedging his specific position between those he dubs “occasionalism” and “reverse occasionalism” or “scepticism.”^[5] While occasionalism has a long and venerable history,^[6] it is worth stressing that we are here concerned with Harman’s own interpretation of this term rather than its historical genesis. With this caveat in place, we can then note that Harman defines “occasionalism” broadly in terms of the view that separate substances do not possess any causal power; their only means of interaction is through God acting as the sole mediator.^[7] It is worth stressing again that Harman defines “occasionalism” here in a broader sense than its classical use by taking it to include any philosophical position which assumes a divine being as the global mediator between entities. Understood in this specific manner, Descartes, Spinoza, and Leibniz would then also be characterised as occasionalists in Harman’s sense of the term, since they all make use of God in order to account for the interaction between entities, albeit in decidedly different ways.^[8]

What Harman calls “reverse occasionalism” or “scepticism,” on the other hand, essentially inverts the occasionalist’s focus on God in order to bestow all mediational causal power to the human mind.^[9] On this view, David Hume can be seen to put forward an inverted form of occasionalism in that he confers all causal mediation to the human mind in the form of “customary conjunction,” while Immanuel Kant follows the same line of reasoning by listing the concept of cause as a category of the understanding. Thus, and always in the context of Harman’s characterisation, occasionalists appreciate the fundamental problem of communication between discrete substances, yet solve the problem by fiat in positing a transcendent divine being a sole causal mediator. Conversely, reverse occasionalists represent the anthropocentric flipside of occasionalism in that they replace God with human when it comes to matters of causal relation.

This brief analysis of Harman’s take on occasionalism and scepticism is important for our purposes in that it serves to show that Harman’s alternative model of causality concurrently conserves and discards elements of these views. More specifically, he preserves the mediational aspect present in both positions,^[10] while also rejecting the “global” recourse to an almighty entity – whether it be God or the human – acting as the sole locus for interaction.^[11] Harman thus proposes a “local” solution by asserting that any entity is capable of supplying the necessary preconditions for a mediated interaction between real objects.^[12] Thus, as a prelude to the coming sections, we might then summarily describe Harman’s “vicarious” model of causation in terms of the view that any two real objects are able to indirectly (or vicariously) interact through a local third term acting as both a mediator and medium for relations.^[13]

3 Sincerity

As we have already stated, OOO is sometimes critiqued for its denial of change and affect, largely due to the assumption that Harman’s early use of phrases such as “absolute withdrawal” and “vacuum-sealed objects” implies the outright denial of relationality.^[14] Such terms require qualification, for it is in a sense true that Harman denies all forms of direct relational accounts according to which relationality is at best assumed in advance, or at worst taken to be the sole defining feature of all entities. Nevertheless, OOO also identifies two specific modalities of indirect contact between entities, namely sincerity and allure.

Harman claims that his specific use of sincerity is intended as another name for what Husserl calls “intentionality.”^[15] It might, however, be more accurate to describe it as both a radical modification and expansion of Husserl’s original model. Two features of the latter’s account of intentionality stand out in view of our current purposes. First, according to Harman, Husserl transposes the classical object/content distinction into the “immanent sphere of consciousness”^[16] when he argues that every intentional act aims at an immanent yet unified intentional – or what Harman calls “sensual” – object, which is only partially given in perception at any point in time through specific partial profiles or “adumbrations” (Abschattungen) – the equivalent of Harman’s “sensual qualities.” Second, Husserl describes intentionality as an a priori “correlation” between two poles, namely an ego or act (noetic) pole and its corresponding object (noematic) pole. Furthermore, Harman often stresses that this intentional relation may be said to be simultaneously both “one and two.”^[17] On the one hand, the intentional relation is one to the extent that it fastens both the subjective and objective poles into a cohesive whole from the beginning. Harman dubs this aspect of intentionality its “adhesive” function. On the other hand, intentionality is simultaneously also two in that the object/act poles of the relation remain distinguishable rather than seamlessly and indistinctly fusing together. Harman calls this aspect of intentionality its “selective” side,^[18] insofar as it works to bring to the fore – or “select” – specific objects while allowing others to fade into the background.

Combining all these features gives us the fundamentals of Harman’s account of sincerity which, at its most rudimentary level, he describes as the “contact between a real object and a sensual one” existing on the interior of a larger intentional whole.^[19] This description of sincerity crucially retains the two aforementioned Husserlian features of intentionality, but Harman also expands sincerity beyond the realm of the human, and hence beyond Husserl’s notion of intentionality. Moving away from “onto-taxonomy”^[20] and towards a more democratic or flat ontology, he insists that sincerity is in fact not exclusive to human consciousness. Rather, the latter should be seen as an admittedly more complex feature of a rudimentary form of sincerity.^[21] Sincerity therefore involves the contact between any real object and a sensual object surrounded by a loose array of sensual qualities, contained in the interior of a total relation which, for Harman, is in turn an object in its own right.

Harman argues that sincerity presents a situation in which the specific sensual qualities “encrust” the specific sensual object, while also claiming that the said qualities must “emanate from a real object no less than a sensual one, because even though such qualities are obviously attached to a sensual object at any given moment, they are the sole way in which the withdrawn [real objects] become present in consciousness.”^[22] It seems clear that in the state of sincerity, a real object RO¹ in contact with sensual object and qualities SO²/SQ² is not in contact with the withdrawn object (RO²) which emanates the sensual qualities, since it relates only to the sensual translation of the said object. In “Object, Reduction, and Emergence,” Niki Young thus is able to define Harmanian sincerity as a state in which a real object intends another real one, but nevertheless only relates to a sensual façade. Even if one might suspect that Harman himself might have some qualms with this description, it remains helpful because it emphasises that the real always forms the basis for the sensual, even if this fact never comes to the fore in those relations which fall under the umbrella term of sincerity.

In some of Harman’s earlier work, such as The Quadruple Object, it can become easy to lose sight of whether he is talking about a sensual object simpliciter or the real object emanating that sensual object. This is because they are in some sense the same object. That is, while the sensual dog I perceive is in some sense only the “dog for me,” when I talk about the dog I’m pointing to, I mean to be talking about the real dog. It is easy to clarify all of this with a little bit of notation. In sincerity, a sensual object (and its qualities, though those will drop out for the purpose of our discussion at this point) is in some sense the glue between two real objects, one emanating and one intending. So we need to merely resort to ordered superscripts to denote this. Consider:

[RO ^x SO ^x,y ] means that RO ^x is emanating SO ^x,y for RO ^y , or alternatively

[SO ^x,y RO ^y ] means that RO ^y is intending RO ^x .

The only order that matters here is that of the superscripts in the sensual objects; [SO ^x,y RO ^y ] means the same as [RO ^y SO ^x,y ], as the key insight is that when we read “SO ^x,y ” we read the superscripts right to left, meaning that real object x is emanating the sensual object for real object y. If we wanted to express this with theatrical metaphors we would write:

[RO ^x SO ^x,y ] means that RO ^x is performing SO ^x,y for RO ^y , or alternatively

[SO ^x,y RO ^y ] means that RO ^y is an audience for RO ^x ’s performance. The sensual object here is the performance itself.

To use the objects above, [RO¹ SO^1,2] and [SO^1,2 RO²] are both ways of denoting that RO¹ is emanating/performing SO^1,2 for intender/audience RO². However, to the extent that our brackets are used as part of a notation to denote a new entity, and perhaps new object, they would not name the same entity as each one contains the same sensual object but a different real object.

It is exactly in this manner that (now that we have represented the basic connections between real and sensual objects) we can represent Harman’s occasionalist claim that each instance of sincerity requires a third superordinate object containing the original two objects, Harman’s democratised occasionalist “third term.” We can denote this form of superordinate object making sincerity possible thusly.

RO³[SO^1,2 RO²]. Here RO³ is the superordinate object, analogous to an Aristotelian formal cause or Occasionalist deity.

As far as we are aware, Harman does not specify whether different superordinate real objects are needed for each nexus of a real object and sensual object it is perceiving/intending. Consider the special case of sincerity when we are talking about perception by creatures which clearly have a mental life. A dog perceives many things at once. For Harman, the real dog is perceiving multiple sensual objects at the same time. Does one occasionalist superordinate object cover all such cases, that is the dog’s entire sensorium? Not in the sense of the one occasionalist God for the entire cosmos, but rather in the sense of one superordinate object per real intender. For example, where the dog is RO², SO^1,2 is the sensual object emanated by a real squirrel RO¹ and intended by the dog, and SO^4,2 is the sensual object emanated by a real tree RO⁴ that the dog also sees, do we have RO³[SO^1,2 RO²] and RO³[SO^4,2 RO²], with RO³ doing the stand in for the occasionalist’s God in both cases, or do we need separate superordinate objects for each nexus? We strongly suspect that Harman does not intend to proliferate superordinate objects quite this much, if only because of the sense in which his theory can be seen as a speculative extension of Kant’s reverse occasionalism. So in what follows, we will assume that the same superordinate object can contain one real and multiple sensual objects, though our substantive arguments do not rest on this assumption.

Likewise, and more importantly, Harman only ever talks about the need for a superordinate object with respect to an intender/audience real object and performance sensual object, never with respect to an emanator/performer real object and performance. But it might be the case that by the same logic he also would need RO⁴[RO¹ SO^1,2], a superordinate object binding performer and performance.

4 The Containment Problem

In his most recent book, Harman tentatively accepts a criticism made by Jon Cogburn. First, the criticism:

The OOO model requires that what happens on the interior of the third or “superordinate” object (Cogburn’s term) involves the confrontation between the real me and the sensual tree (and that in cases where the tree also makes contact with me, there is a different but closely related superordinate object). But Cogburn notes a possible difficulty here. The containing object must be in some sort of relation with the two subordinate objects – one real, one sensual – on its interior. But since the containing object is by definition real, and ex hypothesi also makes contact with the real interior one, this flouts the OOO principle that two real objects can never be in direct relation. But if we suppose for this reason that both objects on the interior can only be sensual, this would make it impossible for them to relate at all, given that two sensual objects for OOO cannot make direct contact either, but must be mediated by a real one.^[23]

Harman goes on to consider, and tentatively endorse, a solution where the contained, subordinate real object is itself a sensual object for the superordinate object. He concludes thusly:

In short, containment too is a mediated relation rather than a direct one. The fact that I inhabit the interior of the superordinate object does not absolve of it dealing with me in indirect fashion, any more than water is absolved from touching hydrogen and oxygen indirectly.^[24]

Here we have to reinterpret our symbolism. Remember that RO³[SO^1,2 RO²] means that RO³ contains RO² and the sensual object that RO¹ is performing for RO². But with Harman’s revision, in this context, RO² is now a sensual object emanated from RO² and intended by RO³, which in our notation is SO^2,3. So RO³[SO^1,2 RO²] is, strictly speaking, always RO³[SO^1,2 SO^2,3].

Two brief comments before we show in the next section why this is so philosophically important. First, note that Harman’s solution is perfectly general. For any markers, a, b, and c there is a simple algorithm to replace every instance of RO ^a [SO ^b,c RO ^c ] with RO ^a [SO ^b,c SO ^c,a ]. Just turn the final “R” into an “S” and add the superscript for the superordinate object. Second, Harman’s solution is reversible, entailing that an important asymmetry has not been lost. From the first half of [SO^1,2 SO^2,3] (e.g. when we have a case of RO⁴[SO^1,2 SO^2,3]) alone we can infer [SO^1,2 RO²]. That is, we can infer from SO^1,2 alone that RO² is the recipient of a sensual object emanating from RO¹. Since this very asymmetry is central to Harman’s account of metaphor, Harman’s revised version of sincerity would not fit his overall account of causality where the asymmetry is lost in making containment itself a relation between the real and the sensual. That is, we strongly suspect that Harman’s original picture where the superordinate object contains both a real and sensual object was motivated by the desire to keep the asymmetry, precisely because the asymmetry is necessary for his account of metaphor. For instance, and to borrow Harman’s example, the metaphorical content of the claim that “the cypress is the flame” is not the same as that of “the flame is the cypress.” But our notation shows conclusively that the asymmetry is preserved with Harman’s recent emendation of the view.^[25]

5 Allure

We can now attend to Harman’s pregnant concept of “allure,” which philosophically can be seen as a combination of “to allude” and “to lure.” This captures for Harman both the sense in which real objects are not perfect mirrors of language (so at best we allude to them) and the sense in which the things themselves are the sources of our normative assessment (when we are doing it right, we are attentive to the manner in which they lure us). This latter is the most important with respect to understanding why allure is a part of Harman’s theory of causation. If sincerity allows us to recapitulate normal judgements of efficient causation in terms of the occasionalist’s sort of formal causation, allure allows us to recapitulate teleological causation. When I sit in silence before the mountain, when I respect the mountain’s autonomy as something that does not exist merely for human consumption, I am responding correctly to, as Alphonso Lingis would say, the world’s imperatives,^[26] or rather the imperative of this mountain, which is prior to language. To the extent that allure is a relation between non-human entities, Object-Oriented Ontology gives us a fully re-enchanted world.

As we have already shown, for Harman, sincerity represents an elementary form of contact which “must always be present before deeper contact [namely allure] is made.”^[27] In other words, the transformation of sincerity to allure can only come about through a disruption of the former, somewhat analogous to the way in Thomas Kuhn “normal science” is disrupted by a revolutionary moment leading to a paradigm shift.^[28] Allure contrasts with sincerity in that it is a “special and intermittent” state of affairs.^[29] It is therefore occasional, and this is in turn to be understood in two senses. First, allure is an occasional occurrence in the sense that it only occurs via the disruption of the everyday relation of sincerity. Second, allure is occasional in the stricter sense of mediated, since the mechanism only allows two real objects to make contact via sensual qualities acting as causal envoys. We may then note that the process by which sincerity is converted into allure consists of a double activity. First, allure disturbs the routine fusion between a sensual object and its qualities, producing a fission whereby the sensual qualities break loose of the sensual object to which they would normally be subservient.

In other words, and as Harman puts it in Guerrilla Metaphysics, allure produces an “interference … in the usual relation between a concealed sensual object and its visible symptoms’, thereby creating ‘a strife between an object and its own [sensual] qualities.”^[30] As we have seen earlier, Harman’s critique of over mining entails that there can be no free-floating qualities subsisting independently of an object. For this reason, the second activity of allure involves the attraction of sensual qualities towards the “withdrawn” real object in such a way that they subtly allude or point to its being, but “without making its inner life directly present.”^[31] In other words, the second activity of allure consists in its ability to draw one real object toward another absent one through the medium of sensual qualities which fill in for its absence. For Harman, allure thus always refers to “the presence of objects to each other in their absent form,” in such a way that it translates “already recognized objects into a more potent form.”^[32] In this way, we now see in the context of Harman’s fourfold metaphysics exactly how allure is the amalgam of two related terms, the first being to lure, to attract, or bring something closer to the extent that the qualities now point towards the withdrawn real object, rather than having the object emanate the qualities as is the case with sincerity. Nevertheless, because allure can never bring the withdrawn object into direct presence, allure can also refer to allusion in the sense of pointing at something without mistaking its reality for its absent presence.

For Harman, this dual mechanism of allure allows for an intending real object to obliquely establish a connection with another emanating real object via the medium of sensual qualities which separate from a sensual object in order to then point to the emanating real one; “allure,” claims Harman, “is the fission of sensual objects, replacing them with real ones.”^[33] In this manner, allure transforms the total intentional relation (containment) into a new genuine emergent entity composed of a mediated connection between two real objects.^[34]

Throughout his work, Harman provides a number of examples of the mechanism of allure in human experience. The most prominent and consistent paradigm of the role allure plays in causation is the example of metaphor, where Harman suggests that something analogous to what goes on in metaphorical language is at work in (vicarious) causation itself.^[35] With this in view, we now turn to a discussion of metaphor in OOO.

6 Metaphor

At the level of epistemology, metaphor is for Harman a clear case study of how conceptually mediated cognition accesses non-conceptual reality, and in good speculative fashion it is also an instance of relations between non-human objects. As we will show, in Harman’s earlier writing, the connection between epistemology and metaphysics is clear, but as he’s developed his account it becomes less so. Our task here is merely to say enough about Harman’s view of metaphor applied to epistemology to show that it is plausible and novel and in doing so motivate the project of speculatively extending it in a way where it might be seen as part of the causal glue of the non-human universe, reflecting relations and processes in non-human reality just as allure, as a potential relation between any two objects, can now be seen as in fact the teleological glue of the cosmos. But even this process will require connecting metaphor with allure in a project that extends this one. Again, we take what we say here to be part of the necessary ground clearing for that project to succeed.

It seems to us that there are two versions of the example of metaphor at work in Harman’s writings. The first comes from his second work Guerrilla Metaphysics in 2005. In this early work, Harman seems to be suggesting that we can figure out what is going on at the level of the real by analogy with what goes on in metaphorical language. While we do not have the space to venture into all the intricacies of his early account of metaphor, it can be relatively easily summarised in one paragraph using the metaphor “the cypress is a flame,” the example Harman adapts from the work of Jose Ortega.

In Harman’s esteem, the literal use of the word “cypress” or “flame” in everyday language directs one’s intention to a particular sensual object with specific qualities. In other words, cases of everyday linguistic use are analogous to the sincerity of perception to the extent that they are both “doomed merely to point at the inner execution of things … without ever reaching full intimate union with their being.”^[36] Contrastingly, the use of a metaphor such as “the cypress is a flame” is evidently not a clear case of literal language or analogy between the two entities. Instead, the metaphor suddenly brings two previously asymmetrical objects into proximity, such that the flame qualities disengage from the flame object to which they are normally subservient, and are subsequently grafted onto an underlying cypress object (the subject). A metaphor therefore coerces the user or listener of the metaphor “to live executantly a new object” which is “neither quite tree nor quite fire, but a vaporous hybrid of both.”^[37]

An analogous case may be noted in Harman’s oft-cited example of fire burning cotton at the level of the real. Given Harman’s claim that entities are “withdrawn” from one another, it follows that fire is unable to directly encounter the real being of cotton. Instead, the fire encounters a “sensual” stand-in displaying the qualities which are relevant to it – its flammability for instance. Occasionally, the “proximity without fusion” which Harman dubs sincerity allows for the sensual qualities of cotton to allude to its withdrawn reality in such a way that a new emergent third object (“burning cotton ball”) is formed. Once this emergent connection is established, this new total object is then able to have retroactive effects on its respective parts. The event called “burning” is therefore nothing more than the ex post facto effect of a joint “connection” between fire and cotton through which the flammable qualities of fire are grafted onto the object cotton.^[38] This account is all well and good, and it tallies nicely with our account of vicarious causation previously discussed.

Harman however seems to shift away from this original position just described towards the year 2018, with the publication of his Object-Oriented Ontology: A New Theory of Everything. In this work, the example of metaphor seems to be employed less by way of analogy, and shifts more towards the theatricality of metaphorical language. The modified account of metaphor is discussed in great detail in his later in Art + Objects (2019a). However, we believe that the most succinct version of this amended account of metaphor is found in his introductory book to Speculative Realism.^[39] This new position of metaphor may be paraphrased as follows: in a metaphor such as “the cypress is a flame,” the cypress plays the role of an object, while the flame supplies the qualities. Yet Harman explains that the latter is clearly sensual qualities rather than real ones, for he emphasises that real flame qualities withdraw from presence no less than real objects. Nevertheless, the cypress in this case is a real object, for it retains a mysterious aura to it in a way that a sensual cypress would not. Insofar as the cypress is a real object, it follows that it withdraws from presence. As a result, since the sensual qualities always belong to an object, it follows, at least for Harman, that the only real object which can fill in for the cypress’ absence is the real me, the “aesthetic beholder” of the metaphor.^[40]

While this seems to be a legitimate account of metaphor, we must admit that we are not clear about how it fits with his earlier account of allure. How would this work as an account of causation? Most worrying, if there is only the real me performing the metaphor, does this imply a sort of subjective idealism which never makes contact with the real? Have Berkeley and Hegel won after all? It would be an understatement to call this unfortunate, for if it were the case that Harman needed metaphor to make sense of causality, and that simultaneously his account of metaphor was rebarbatively idealist, the entire edifice of Object-Oriented Ontology would collapse. In our concluding section, we will be able to show in greater detail why this is an issue and, in the process, illustrate the novelty and plausibility of Harman’s view. However, it will be clear from our discussion that much more work needs to be done. In particular, the more one develops Harman’s epistemology of metaphor in a non-idealist direction, the less clear it becomes how metaphor connects with allure. Again, all we hope to accomplish here in this respect at least is to raise the issue in such a way to be productive for future developments of the framework by us and others.

7 A Realist Theory of Metaphor

It is easy to see why Harman at least initially presents metaphor as being connected with allure. For, if literal language misleads us into thinking of the world as consisting of facts which make each true sentence true, metaphorical language allows us to non-literally articulate our sense of what the world is like without language. Part of what is constitutive of metaphor is that it involves sentences which are literally false (in the sense of propositional truth or falsity) yet still apt for assertion in an epistemically relevant way. For Odysseus is not lying in the sense that one lies when asked about their friend’s new bangs, when he says,

And what if one of the gods does wreck me out on the wine-dark sea? I have a heart that is inured to suffering and I shall steel it to endure that too.

Attempting the return home to his wife is worth being wrecked on the metaphorical wine-dark sea. He is rejecting Calypso’s offer of immortality, forcing the already metaphorical “being wrecked on the wine-dark sea” into double metaphorical duty as now “wine-dark sea” is functioning as part of a metaphor for human mortality. Whatever the gods doom us to after we die, Odysseus chose the infinite loss of control that is dying while having lived the kind of life of a person who attempted to return home to his wife. He prefers this to immortality.

One might think we understand metaphors such as “wine-dark sea” because we subconsciously translate them into literal language, something like “sea that looks blackish-red during storms that move the boat in a way that makes me feel unpleasantly intoxicated.” Such a view of metaphor is similar to philosophers’ “theory-theory” view of our knowledge of other minds. Under this view, we understand other people by observing their actions and then subconsciously figuring out what sets of beliefs and desires would plausibly give rise to their actions, deriving those actions from a subconscious linguistic description of the way the world is and the beliefs and desires we ultimately attribute to them.^[41] This then allows us to predict how the person would behave in novel situations because we subconsciously know what their beliefs and desires are. Just as the theory-theory of external minds has us understanding other people via our supposed ability to translate their behaviour into a set of literal beliefs and desires truly attributed to them, the theory-theory of metaphor has us understand metaphors via our supposed ability to translate sentences containing metaphors into sentences not containing them.

In cognitive science, the main competition to the theory-theory of understanding other minds is the simulation theory, which posits that in understanding other people we imagine ourselves in their situation and try to predict what we would do in those situations.^[42] Of course, this requires great imaginative feats in some cases, since putting oneself in another’s situation often means imagining that we are them. As a result, the simulation theory ultimately requires that we simulate the other person. The virtue of the simulation theory is that its defenders want to be able to understand this simulation as something other than subconsciously using language to predict behaviour via logical derivations from sets of beliefs and desires. The simulation theorist is absolutely correct that there is no independent reason to think that minds actually work the way theory-theorists allege. Among other things, nothing we know about localising language areas in the brain suggests that the brain processes that correlate with having a good sense of what other people will do are linguistic. And there is no evidence that the kind of psychological acuity demonstrated by empathic humans tracks linguistic acuity as one would expect from the theory-theory. We are not texts and human interaction is not reading.

To understand Harman’s theory of metaphor, one must understand the extent to which the view he is rejecting, metaphor as condensed literality, just is the theory-theory of other minds applied to things. Just as the theory-theory has us understand other minds in terms of a set of beliefs and desires that can be articulated in language, Harman’s opponent understands our understanding of metaphor as involving a paraphrase into literal language. Once we see this, we can appreciate that Harman offers a simulation theory of metaphor, where we understand the wine-dark sea by simulating it in ourselves. With respect to metaphorically attributing fire to a cypress tree, Harman writes,

For if the real cypress is just as absent from the metaphor as it is from thought and perception, there is nonetheless one real object that is never absent from our experience of art: namely, we ourselves. Yes, it is we ourselves who stand in for the absent cypress and support its freshly anointed flame-qualities.^[43]

Just as I simulate another person to understand them and discover what it is like to be them, and hence also what they are likely to do, I internalise the sea to understand its wine-darkness.

While this is an extremely innovative and intriguing idea that plays a central role in much of Harman’s productive thinking, for our current purposes we needn’t explicate all of the ways it fits into Harman’s broader system. First, a key premise in Harman’s argument for his theory of metaphor is that the real objects (e.g. the real sea) do not literally possess the properties metaphorically ascribed to them (being wine-dark). For Harman, what makes an object real is that it transcends any literal properties we attribute to it. Again, the world is a collection of objects, not facts. This is why Harman concludes that it must be the sensual sea, which is the sea I perceive or think about formed out of the one-way relation between the real me and the real sea when I perceive or think about the real sea.

However, the sensual sea is not literally wine-dark either! At no point in reading The Odyssey do I think that my representation of the fictional Odysseus could drink the sea water and accurately or not guess its alcohol content. If a reader of The Odyssey tried to ascertain which of the ancient Greek varietals (i.e. agiorgitiko, assyrtiko, athiri, moshofilero, roditis, limnio, or xinomavro) produced their sensual sea wine, they would be misunderstanding the narrative.

But we have to be very careful now about Harman’s suggestion that I am the bearer of the property in question. Does he mean the real me, inaccessible to literal description? But he has already told us that this is not possible!

Our initial reaction might be to turn from these third-person experiences to the first-person kind, seeking refuge in the direct truths of inner life. Yet this is not really possible, since in introspection we also reduce ourselves to shadows or outlines: after all, there is no direct access to the noumenal self any more than to a noumenal house, dog or horse.^[44]

But then might Harman mean the sensual me, that self which is visible to and describable by others? That can’t be the case either, as sensual properties are literal properties. When I internalise Homer’s metaphor, it is not the case that oenologists can taste me and discern the chalk content of the soil in which the grapes grew that were used in my vinification.

Harman’s initial argument that an entity’s not literally possessing the metaphorically attributed property entails that there must be another entity that possesses it would then seem to entail that there must be a third supersensual realm where the supersensual object can literally possess the supersensual quality in question, perhaps the supersensual object being the occasionalist’s third entity. While this is more than superficially similar to the superordinate objects that are part of the theory of sincerity, Harman never suggests such a role for those entities. And it’s not clear how that entity literalising a metaphorical quality would contribute to the success of a metaphor in any case. The problem then is that we seem to be out of options here.

Second, as we noted, the original philosophical context that moves Harman to write about metaphor involves trying to make sense of our ability to in some sense understand the withdrawn reality obscured by our conceptual schemes and literal experience. Unfortunately though, when developing his mature theory of metaphor, Harman is so quick to turn to issues in the philosophy of art proper that he never returns to the issue that motivates it.^[45] Again, what does our ability to create virtual versions of objects virtually possessing properties have to do with the pregnant notion of allure, which for Harman allows the Western tradition to replace theory and knowledge as the foundational epistemic category? Our understanding of metaphor is supposed to be a paradigm instance of how we have access to reality beyond that granted by literal language. But Harman’s theory of how metaphor works is internal, a person simulating, which on its own seems independent of the external world.

This leads to our third worry, one we take to be the most serious issue with Harman’s theory. That is, to the extent that metaphorical truth can do the epistemological and metaphysical work that Harman wants it to do, we must develop a notion of metaphorical falsity, or at the very least of a metaphor being unjustified. Harman does not mean to rule this out, e.g.

If I do not step in and attempt the electrifying work of becoming the cypress-substance of the flame-qualities, then no metaphor occurs. This might happen for any number of reasons, including the poor quality of the metaphor, the obtuseness of the reader or even the boredom or distractedness of the reader.^[46]

Harman accepts that metaphors can fail to illuminate the real due to their own poor quality, though he does not provide a theory of this failure. And if all there were to a metaphor were interior simulations, it’s not clear what would even constitute failure. This final problem is connected to the first and second problems. For if we understand the epistemic appropriateness of metaphor, we will solve the second problem, understanding how metaphor gets us in touch with the real without falsely portraying the real as linguiform.

As far as we can tell, the only consistent way to develop what Harman has in mind with respect to metaphor and allure (that is in terms of an intending/audience real object performing a virtual version of the real object emanating/peforming for it) is for us to consider how a sensual object can literally possess a property such that it (or perhaps its corresponding real property) is metaphorically possessed by a real object. The real sea’s metaphorical possession of wine darkness is in part a function of the literal properties possessed by the sensual sea. Remember that sensory objects only come into being when one object performs or emanates for another that intends or is the audience for. For Harman, metaphor requires then a second performance, where understanding the metaphor involves the audience themselves simulating/performing the real object for themselves. However, we still have no account of how this takes place.

Luckily, there are many ways a theory of sensual literality and real metaphoricity with respect to properties could be worked out. From the perspective of analytical philosophy, the most natural might be that our understanding of Homer’s metaphor is based on our ability to imagine non-actual universes as much as possible like the fictional universe of The Odyssey but where the sea is dark wine, and then imagining what we would do were we, like Odysseus, wrecked alone on that sea by the gods. It is grim stuff.

To be clear, this account does not in any way entail that the actual Odysseus (in Homer’s fictional world) treats the sea like literal wine. He clearly doesn’t. Nor does it mean that our imagined counter-counterfactual Odysseus, where the sea is wine, treats the sea like us non-shipwrecked moderns treat wine. The question concerns what Odysseus would do in a world as much as possible like the projected actual fictional one but where he ends up being eternally shipwrecked and the sea is made of dark wine. It would be disgusting with all sorts of rotten ocean effluvia sifting through it. He would drink it and drink it until he was eternally stupefied and miserable, the way many ancient Greeks in fact thought of our state after death (and this, not the way the sea actually looks, is surely the main point of the metaphor for Homer). But a hell world of wine darkness (or rather actual death as a miserable and confused ghost) would have been better than immortality with Calypso purchased at the cost of having abandoned his wife. And that this picture of counter-counterfactual imagination lets us so clearly articulate what Homer had in mind with the metaphor is some evidence that it is right.

We have begun to see the manner in which metaphor can get us closer to teleological reality, which requires a sensitivity to situations which call us to courage and open-eyed sacrifice for things worth sacrificing for. There is no algorithm or ethical theory which can describe all such situations and tell us how to proceed, but literary metaphor gives us a sensitivity to exactly these things. And this begins to answer our third criticism. Once we understand what the metaphor is doing, it is clear that most other concepts would fail. Falafel dark sea? No. Pacman dark sea? No. Imagining a counter-counterfactual world where the sea is a dark falafel does nothing in the context of Odysseus’ narrative. It gets us no closer to the existential truth Odysseus realises in his bones.

Harman’s critic might respond by reminding us of Harman’s claim that the world cannot be adequately described by literal language. But wouldn’t Harman be committed to just this if he accepted our suggested account of real objects’ metaphorical possessing of real qualities in terms of corresponding sensual objects (in the relevant counter-counterfactual worlds) literally possessing sensual qualities? And, to the extent that this is Harman’s theory, we don’t seem to have a simulation theory of metaphor after all, but just a more complicated theory-theory with a sensual intermediary. If the sensual counter-counterfactual objects literally possess the sensual stand ins for metaphorically possessed real properties, then literal descriptive language should be able to non-problematically describe this.

This would be a fair criticism if simulating a metaphorical object, becoming the wine-dark sea, were a matter of giving a literal definition description of a world where the sea is both dark and wine. But nowhere does Harman suggest this, and our discussion shows why it is not the case. Simulating the wine-dark sea in this context involves imagining a world, as much as possible like Odysseus’ but where the sea is dark and made of wine. It would only be if this reductive view of imagination were correct, where imagination involves the inner production of non-metaphorical text, that the criticism would land.^[47]

But it doesn’t. First, attempts to describe the content of our imaginations would invariably involve metaphor. But then the Harmanian process could not possibly reduce metaphor to literality. Second, our example from Homer shows that any such descriptions would involve recapitulating large chunks of the narrative of the Odyssey. In a different story, the wine darkness of the sea is quite a different metaphor. Imagine a story taking place during summertime in Alaska and in a language community where the phrase “spring darkness” is used to paradoxically denote the omnipresence of sunlight around the summer solstice at sufficiently northern latitudes. And imagine that in this story during the solstice, underground communities perform eldritch rites that help with their sunlight-caused insomnia and give them a vision of true light of Heaven in the dream world. As part of these ceremonies, they drink an adulterated wine with the same golden quality of the omnipresent northern solstice sunlight and refer to the dreamworld as the wine-dark sea. Here, the metaphor of “wine-dark sea” means something completely different than it does in Homer.

So, as a final note, for metaphors to do the work that Harman wants, that is give us a kind of understanding of the real that does not mislead us into thinking of the real as itself a mirror of language, they must be narratively situated. I don’t just simulate wine-dark sea. I simulate Odysseus’ wine-dark sea. But this means that Harman’s insight into the role metaphor plays in allowing us to access reality requires a longer story about how fictional narrative allows us to access reality. Imaginatively complicit readers of Homer have better insight into death and, all else being equal, are better prepared for the manner in which the world calls us to sometimes futile sacrifice.

8 Concluding Remarks

We have said more than enough to save Harman from the allegation that his recent work about art must entail a regression into correlationism/idealism. But in showing that we have raised as many questions as we have solved. How exactly do the counterfactual simulations fit into Harman’s fourfold ontology of object and quality, real and sensual? Is the connection made by successful metaphor an instance of allure or something else? And if the latter, are we replacing allure or supplementing it? On the other hand, to the extent that the more developed theory of metaphor is part of a theory of causation, might it replace sincerity instead of allure? In doing so might it obviate the need for the regress of superordinate objects that sincerity brings with it? How is the holistic nature of metaphor, where a metaphor’s meaning and success depend in part on its narrative context modelled in such a robustly anti-holist metaphysics such as Object-Oriented Ontology? Answers to all these questions will non-trivially impact the speculative extension of the epistemology of metaphor into the non-human world. In all such cases of speculation, such as Harman’s interventions with Heideggerrian and Husserlian epistemology while developing his fourfold division of reality into sensual and real objects and qualities, the epistemic relation is understood as a specialised version of general relations that hold independently of humans.

But do non-human objects really simulate one another? Do objects in fact situate themselves in a penumbra of possible versions of themselves in situations where metaphorically true properties are literally manifested? We find this to be an intoxicating idea, but at present, we have no settled intuitions about how plausible it is and what further philosophical work is done by it. Much more is needed. If the penumbra of possibilities is part of the epistemic picture that gets speculatively extended, then the Object-Oriented Ontologist would perhaps face revisions to the fourfold itself.