Representations, direct perception and scientific realism. In defence of conservative predictive processing

Many researchers accuse the Predictive Processing (PP) framework of returning to nineteenth-century speculations regarding the knowledge of reality, the difference between phenomena and things in themselves, or questions about idealism. Dan Zahavi’s (2018) harsh criticism follows in this tradition. He argues that the supporters of PP are not able to justify realism or the common sense belief that the world of objects given in experience exists objectively, i.e. regardless of our cognitive capacities. In his opinion, adopting PP assumptions, we must abandon "our naïve realism", i.e. our conviction about the objective existence of everyday objects of experience, (Zahavi 2018, 48). In these considerations I will argue against the criticism made by this author. I will show that it can be reduced to three main objections: (1) representationalism objection; (2) indirect perception objection and (3) anti-realism objection. In response to these three objections, I will argue that Zahavi's criticism is based on a very selective and simplified reading of PP. Next, I will defend the thesis according to which perception in PP can be understood as indirect only in the psychological sense, not in metaphysical and epistemic. In response to the last charge, I will show that the representationalism postulated by conservative PP allows to justify the thesis that PP is the position of scientific realism. To this end, I will refer to the analysis of the concept of structural representations (S-representations), and then I will argue for the ontic nature of explanations using S-representations, based on the mechanistic model of scientific explanations