In his UCLA talk, “Real Objects and Pseudo-Objects: Remarks on Method“, Harman presented a number of ways in which to identify the real existence of an object. Here my aim is simply to summarize Graham’s talk so I can fix it in my own mind. Graham begins his talk by referring to DeLanda’s account of assemblages. If this is relevant to the question of when an object is, then this is because, as Harman argued in Guerrilla Metaphysics, every object is an assemblage. As Graham there puts it, every object is wrapped in objects, wrapped in objects, wrapped in objects. Yet how do we distinguish between mere aggregates or collections that do not themselves form an object and genuine assemblages?

DeLanda proposes five criteria that an assemblage must have in order to count as a genuine assemblage. Graham will contest these criteria while nonetheless using them as a launching point for his own account of objects.

First, in order for an assemblage to count as an assemblage it must have emergent properties. This is to say that in order for an assemblage to account as an assemblage, the assemblage must have properties, qualities, or powers that cannot be found in the parts of that object. Thus, for example, if UCLA is an object it will be capable of doing things and will have properties not found among any of its parts (staff, faculty, students, buildings, etc).

Second, in order for an assemblage to count as an assemblage it must be characterized by redundant causation. Redundant causation characterizes that feature of objects such that the parts of the object are replaceable while the object remains that object. Returning to the example of UCLA, if UCLA is an object then this is because it doesn’t cease to be when students graduate, when faculty retire, or when it gets a new president.

Third, DeLanda contends that assemblages are entities in their own right insofar as they are able to affect other entities at their own level of scale. UCLA, for example, is able to affect other universities which are themselves entities that exist at their level of scale.

Fourth, DeLanda argues that assemblages are characterized by retroactive causality, which is to say that assemblages are able to affect their own parts. Thus, for example, a government affects its citizens, while a body affects its cells.

Finally, fifth, assemblages are entities that produce their own parts. UCLA produces students, governments produce citizens and criminals, and bodies produce cells.

Graham argues that DeLanda’s list lacks systematicity and is therefore in need of a deeper philosophical account. He begins by arguing that features three, four, and five are non-essential features of objects. While some objects can indeed affect other objects at their own level of scale, it does not follow that an object must affect objects at the same level of scale in order to count as an object. In short, objects are not to be defined by their effects on other things. This, of course, marks a fundamental difference between Harman and Latour, insofar as Latour defines an actor as the sum of its effects on other actors. This observation leads Harman to introduce the concept of submergent qualities. In addition to emergent qualities, every genuine object necessarily has submergent qualities. Submergent qualities are those qualities of an object that an object possesses without manifesting those qualities. In this connection, Graham gives the example of human talents. A human being might have a particular set of talents that never manifest themselves because environmental conditions are not ripe for the manifestation of those qualities. These submergent qualities are real features of the object, yet are never or not yet actual within the world.

read on!

If criteria four and five are not essential qualities of an object, then this is because while retroactive causation and the genesis of parts is common to biological systems and social systems (and, I would add, entities like hurricanes and tornadoes), they are not necessary features of all objects. A rock, for example, neither produces its parts nor has a retroactive effect on its objects.

Graham also introduces the concept of obtuse causation as a necessary feature of how objects, in many instances, relate to other objects. I find this concept particularly interesting with respect to my forays into Jakob von Uexküll and Luhmannian social systems theory. Obtuse causation consists in relating to other objects indifferently as the same, “ignoring” or being blind to differences among these objects. Thus, fire relates to cotton, paper, wood, etc., indifferently as something to be burnt. Bureaucracies relate to everything else bureaucratically, social systems relate to their elements (persons) as being the same, and so on. This concept strikes me as ripe ground for ideological analysis and as being of great significance for race and gender studies. What is racism if not a form of obtuse causation where one group treats all members of itself and another group as being interchangeable?

What remains, then, are criteria one and two. We are before an object when that object is characterized by emergent properties and redundant causation. Graham is thus led to outline four necessary features of objects:

First, in order for an object to count as an object it must be independent of its parts. In other words, objects are such that they contain properties that cannot be found among the parts. As such, objects are irreducible to their parts. This would be the feature of emergent qualities characteristic of objects.

Second, it follows that the parts of an object are independent of the larger scale object of which they are a part. Citizens cannot be reduced to the nation of which they are a part, nor is a bureaucrat reducible to the bureaucracy of which they are a part (here we might think of Kafka’s literature where we perpetually encounter a sort of fissure between obtuse causation between how, say, the castle relates to its elements and the excess of these parts over the roles that have been assigned to them).

Third, objects are independent of other objects, such that they cannot be reduced to their effects on other objects. Metaphysically, in the limit case, an object can exist without having any effects on another object, while nonetheless being entirely real. These objects are what Harman refers to as dormant objects. In my language, they would be objects that have no local manifestations in the world.

Finally, fourth, other objects are independent of any particular object.

To drive these points home, Graham contrasts real objects with pseudo-objects. Like real objects, pseudo-objects are characterized by four features.

First, a pseudo-object is an object that has no emergent properties over and above its parts. Collections of this sort are referred to as an aggregate. Graham cites many of the collections produced by Bogost’s Latour Litanizer as an example of mere aggregates.

Second, we encounter pseudo-objects when the alleged object has no submergent properties. We’ll recall that submergent properties are real properties of an object which nonetheless are not actual in the world. Submergent qualities are the hidden or withdrawn depths of an object. Graham refers to pseudo-objects that have no submergent qualities as events. Events are purely exhausted in their happening, such that they contain no excess within themselves.

Third, pseudo-objects that have no redundant causation such that their parts can be replaced while the object still remains that object are what Graham refers to as sets. It will be recalled that within the framework of Z-F set theory sets are defined purely by their extension. In other words, a set is nothing more than the elements that belong to it. Consequently, if you subtract an element from the set and replace it with another you no longer have the same set. As a consequence, sets do not have anything like redundant causation. Here, for example, a family might be a pseudo-object rather than a real object because the elements of such an “object” cannot be replaced (I’ll need to think on this particular answer some more). In short, families seem to be defined extensionally.

Finally, fourth, pseudo-objects are objects that lack any obtuse causation. Rather, their causation is exhausted by their interaction. Graham refers to these sorts of relations as mere impacts.

Graham presents four methods for identifying real objects and distinguishing them from pseudo-objects. The first of these methods is the counterfactual method. Here we imagine the object existing in different situations, striving to discover how the object would exist in that situation. What would Lincoln, for example, be like during the reign of Pericles in ancient Greece? The counterfactual method allows us to allude to the withdrawn style of an object or its inner structural reality. In close connection to this method, I have also been pushing experimentalism as a method of physically situating objects in different contexts to discover how they behave. Second, Graham proposes the hyperbolic method. I’m not entirely sure I understand the hyberbolic method, so I look forward to seeing how Graham develops it. In his talk he says that he employed the hyperbolic method in his books on Latour and Meillassoux. Their the aim is to imagine if these philosophers had been completely successful and come to dominate thought in the future. Given such a total success, Graham asks, we are to ask what would still be missing. Third, Graham proposes the method of simulation. The more an entity can be simulated the more real it is. Thus, the fact that it is easier to simulate Dostoyevsky or Kafka entails that they are more real than Dean R. Koontz. Finally, there is the “accidental method”. The accidental method reveals the work of parts beneath wholes. When we encounter unexpected accidents we discover that other entities are at work beyond our model.

Each of these methods relates to one of the four features of objects. Thus the counterfactual method reveals the existence of an object over and above its parts. The hyperbolic method reveals the independent existence of parts over and above larger scale objects. Simulation reveals the independence of an object from other objects. And the accidental method reveals the existence of other objects independent of an object.