The question of fundamental importance is whether a real causal structure can really be discovered by omitting the real properties of the entities standing in relations.
The intuitive answer is negative. Causal realism built up without entity realism means that we try to describe some causal connections underlying reality while conceiving the related things as having only secondary importance. According to the idea of structural realism, describing a structure is possible by effacing the real nature of the related things. In this context, Chakravartty (1998: 400–402) calls attention to the circumstance that knowledge of the structure involves knowledge of certain properties of the entities standing in relations—properties that are important in terms of facilitating such relations and the behaviour entities show there. Psillos (1995) draws similar conclusions when he suggests the notion of “structural properties”. Structures are underpinned by certain properties of the related entities; that is, knowing of the structures necessitates our knowing of those properties of the entities that are crucial with regard to relations. Relations contains information about entities. Such information describes what characteristics the related entities have in terms of relations and interaction. A given structure is compatible only with certain entities (those it conjoins), so a structure is not indifferent to the (nature of) entities. In other words, the description of a real causal structure cannot be built on entities the properties of which are not abstracted from the nature of real agents. As describing a causal structure requires us to specify some entity-level assumptions too, entity realism as to the relevant properties is the prerequisite for causal realism. The properties of entities are not transcendent relative to structure. In his critique on structural realism, Psillos (1995: 31–32) highlights that a description of a structure is nothing but a description of the way entities are related and the way they behave in the context of such relations. The causal roles and causal properties of entities are not over and above the details that can be described in terms of structure. As a consequence, if we intend to describe a real structure, we are in need of (approximately) true entity descriptions. However, if our entities are not like their real counterparts in terms of the relevant aspects, then our knowledge can be negative at best. The surrounding world evidently works in a way other than our ideas.
Chakravartty, Anjan (1998): Semirealism. Studies in the History and Philosophy of Science, 29(3): 391–408.
Psillos, Stathis (1995): Is Structural Realism the Best of Both Worlds? Dialectica, 49(1): 15–46.