As I explained in the previous post, CSR’s account of scientific representation is based on the neuroscientific account of the brain-world relationship. The neuroscientific account is presented in terms of the Predictive Processing Theory (PPT) and the Free Energy Principle (FEP) as being developed by Karl Friston and others. PPT and FEP have found their way to the centre of some fascinating debates in the philosophy of perception and action. In a considerable part of this book, I take great pains to use PPT and FEP to consolidate CSR’s account of the representation.
According to PPT, the brain is capable of minimising its prediction error. The brain uses variational Bayesian inferences to decrease the discrepancy between its internal models and causal structure of the world. FEP bestows upon PPT some biological viability because it indicates that organisms that want to stay in the state of the equilibrium with their environment (and thus maximize their survival by minimizing their free energy) must be able to decrease the discrepancy between their models of the world and the world itself.
There are alternative philosophical interpretations of PPT. One interpretation that is mainly advocated by Jakob Hohwy interprets PPT along the lines of inferentialism. I rely on this construal to provide a new solution to the problem of scientific representation. First I submit that cognitive models could be specified as embodied (variational) Bayesian structures in the context of CSR. Then I built upon an inferentialist construal of PPT to account for the representational relationship between models and reality. PPT explains how the brain latches onto the world by decreasing its prediction error. Scientific theories are made in brains, and scientific representation could be explicated on the basis of PPT’s account of inferential links between the brain’s internal models and the causal structure of reality. Given FEP, PPT-based account of scientific representation comes with some inherent biological viability. I defend the objectivity and context sensitivity of PPT to conclude that PPT’s inferential links are reliable enough to underpin a realist account of the brain-reality relationship.
A form of scepticism about the objectivity of PPT-based account of representation might survive this solution. This is because the inferentialist construal indicates that the brain’s internal models are secluded from the real world by an inferential veil. The brain could not actually go beyond its inferential veil to ensure that the features of the mind-independent reality resemble the brain’s internal models. And it is possible to harbour scepticism about the veracity of inferential links that represent the world to the brain from beyond the inferential veil. I draw on an enactivist-embodied construal of PPT to overcome this radical form of scepticism (I unfolded this embodied approach in chapter 7, sections 7.2-7.5). According to this reading, agents invoke active sensorimotor predictions to engage the world’s windows of affordance directly and without invoking representations or any form of amodal cognition. The radical embodied construal of PPT dispenses with the inferential veil and inferential relations, to the effect that it does not leave room for any form of radical scepticism. Thus, the reply to radical scepticism comes with a direct realist as well as radically pragmatic (viz-a-viz semantic) tendency. This radical solution is inlaid by the elements of embodied theory and ecological psychology. I realise that this solution to the radical form of scepticism about the issue of representation might be too naïve and austere. Representationalism or moderate forms of embodied theories (as being advocated by Andy Clark) seem to be philosophically more sophisticated. But I invoke radical embodied construal to address a radical form of scepticism that survived my inferentialist reply to the problem of representation, and radical problems demand radical solutions. Direct realism leaves no room for the problem of representation. I remain neutral between inferentialist and embodied replies to the problem of representation. The choice between them depends on which (weak or strong) version of scepticism we want to address. As far as I am concerned they both provide naturalistically plausible replies to the problem of scientific representation.
CSR makes ontological commitments to embodied informational structures. These informational structures could be identified in terms of information processing in the biological, cognitive systems. Owing to the embodied nature of these mechanisms and their reciprocal dynamical interactions with the environment, they could be assumed to be entwined with the causal structure of the world.
Thanks for posting the summary of your interesting-sounding book. A couple of questions pop up.
First (and this applies to previous posts too) I am not convinced that a theory of individual brain function will be all that helpful in shedding light on theories of scientific theories, which are social, language-dependent, abstract constructs that seem fairly far removed from the kinds of low-level structures you find in individual brains. That’s why only humans do science: only we trade in the the appropriate kinds of complex representational structures that are uncoupled from the exigencies of current sensorimotor demands, and this is precisely what allows for scientific theorizing.
Second, I don’t understand the stuff on representation. Coming down against representations seems this is to throw out the baby with the bathwater: what remains of predictive processing when you don’t have these rich loops of internal feedback generating predictions, and what are these predictions if not a genus of representation? This uncoupling of world from internal state is not something to run away from, but what we observe all the time in neuroscience (consider dreams, hallucinations, optical illusions, but also just garden-variety sensory, memory, and motor processes as in my paper with Piccinini Neural representations observed). Such phenomena are handled naturally within a representational framework, including a predictive processing framework as a special case. It seems there has to be a better solution to the purported problem than jettisoning representations. If I understand what you are doing, it seems analogous to solving the problem of other minds by rejecting the existence of other minds. Please tell me I’ve misunderstood. 🙂
Anyway thanks for the summary, it seems interesting and very ambitious!
Dear Eric,
Thanks for the comments, which are quite insightful.
Concerning the first comment, I do not think that *a theory of individual brain function will be all that helpful in shedding light on theories of scientific theories* either. Of course, I had to gloss over a lot of details in the posts, but in the book, I acknowledge that “the embodiment of the brains in biological environments and social contexts enhances the complexity of the situation” (p.121). Actually, I also point out that the embodied construal of FEP considers “environmental and social aspects of perceptions and emotions, attention deployments, embodied strategies of learning, language comprehension, memory, visual categorization tasks” (p.159) (page numbers refer to pages of the book). And there are works by Friston, Gallagher, and others that explore the cultural and social aspects, in terms of social nitches and situations. I aim to develop this theme still further, and this book only takes the first steps. But in principle, predictive coding and FEP can provide a framework for including social and cultural factors.
Concerning the second point. I agree with this point too. In chapter six I defend a form of Kantian CSR which is agreeable to representation and inferences (see “6.10 Dissolving the Problem of Representation, a Kantian Approach”). My sympathies with this version of CSR notwithstanding, a form of radical scepticism about the veracity of representations survives (see section 6.12). And since radical problems demand radical solutions, I appeal to a research stream that defends radical embodied theory to uproot the radical form of scepticism.
To be clear, I do not say that radical embodiment is correct what come may. My point is that radical scepticism and radical embodiment are a match.
Thanks for comments again,
Majid