Deleuze talks about the struggle for primacy in terms of the aspiration of various discipines to become “the official language of a Pure State” (DIALOGUES, 13). Thus in the French context the ancient supremacy of philosophy has been superseded by new contenders: psychoanalysis, linguistics, and cognitive science (Deleuze doesn’t mention physicalist reductionism because this has far less impact in the visible battle for hegemony in France). In each case we have a new “interiority”, ie the very opposite of the transversality that combines various disciplines on an equal footing to produce new ideas and new explanations. Instead we have the same form of an autonomous discipline (interiority) staking a claim for explanatory primacy over its fellow disciplines. But this self-promotion to primacy and universality is accompanied by a desperate endeavour to formalise and unify the hegemonic discipline, to give it explicit and coherent structure, canonical method, apodictic force. From a fruitful heuristic messiness and multipiplicity (diverse partially incompatible hypotheses producing only partially overlapping results) the hegemonic will imposes a unified paradigm or system of judgement. This unificatory interiorisation then functions as what Deleuze calls a “represser of thought”, a rulebook for producing (politically) correct ideas. No wonder mathematicians and physicists resent the incursions of cognitive science – they are too comfortable with their supposed intellectual primacy and academic prestige. Behind the epistemological babble about disinterested research and objectivity as a bulwark against “irrationalism” and “relativism” there is a combat for power, posts, resources, status and funding.
Personally I do not like to talk in terms of “correlationism” as I believe it to be an ill-formed concept (allying maximum extension with minimum intension). However, if one is going to talk about correlationism in the case of cognitive science’s pretentions to account for science itself, its nature and legitimacy, this can only be the case under an extended acception that I would call “structural correlation”. This nicely captures a little-noticed fact: whenever a physicist leaves off his equations and experiments to declare that everything is made of and reducible to sub-atomic particles (or strings, or spacetime, or whatever) and that all disciplines are ultimately explained by his own he is no longer making a scientific claim inside his discipline, but is himself guilty of structural correlationism.
I think that Andrew Gibson is right to emphasise that the concept of hermeneutics is crucial. Harman is in denial of hermeneutics, and as with his denegation of epistemology, ends up doing bad hermeneutics. His hermeneutics of Eddington’s text is quite inadequate and erroneous as is his hermeneutics of the history of philosophy. “Correlationism” and “epistemology of access” are ill-formed hermeneutical concepts, giving a grotesque simplification and deformation of the history of philosophy and of contemporary rival philosophies. Feyerabend argues that the sciences are not abstract cognition only, but have a constitutive, and thus necessary, hermeneutic dimension. This is why even the sciences provide some resistance against neo-liberal neo-leibnizian abstraction and speculative modeling and manipulation. The model, as you say, is not enough to account for knowledge, and it is Harman who is being reductionist with his real objects and their supposed sensual instanciations. This philosophy splits hermeneutic, ie participative, exploration into objective speculation (an absolutised and thus “withdrawn” context of justification and sensual or subjective encounter (an absolutised, and thus “sham”, context of discovery). This splitting demotes the subject to the world of shams, which leads to a “reurn of the repressed”, in the form of an implied subjectivity, but one that he is either unaware of or unwilling to endorse explicitly, adapted to the neo-liberal order. Thus far from eliminating subjectivity from the world of objects Harman’s OOO is subtended by an all-pervasive degraded subjectivity masquerading as its opposite. Harman then proceeds to re-subjectify his philosophical vision with expressions connoting a subjectivity that is ruled out by the strict application of that philosophy. He talks of how we must love the object: “The real is something that cannot be known, only loved” (THE THIRD TABLE, 12); thinking must be indirect, “its approach to objects can only be oblique” (12), and “allude to objects that cannot quite be made present” (14). All this talk of loving and hunting and approaching and alluding to, all these expressions are strictly ill-formed. A sensual subject cannot love, hunt, approach, or even allude to a real object. It’s not that objects cannot quite be made present, they cannot be made present at all. Withdrawal is all or none, it does not admit of degrees. Yet to give appeal to the theory Harman has need of descriptors of the subjective attitude of those who endorse it. Hence the constant talk of objects that redounds in unthematised subjective participation in the theory as vision of the world. The objectal conversion as the passage to the constructed “naiveté” that sees objects everywhere is thus a subjective conversion to a hard-headed noetic asceticism of intelligible objects coupled with a soft-hearted sensual exoticism of the aesthetic play of the simulacra. You can be a geek and an esthete at the same time, with the contradiction being covered up by the medial subjectivity of loving indirectness, hunterly obliquity, and diaphonous allusion.