Markov Blanket

With the publication of Wilfred Bion’s text 'Learning from Experience', psychoanalysis was afforded a new schema for understanding the processes and implications involved in an infant’s contact with their caregivers. As a result, our conception of some of the most fundamental phenomena of psychic life was significantly enriched. By proposing his theory of alpha-functioning, Bion mapped out how meaningful connections to the internal and external worlds become established in the mind. In contrast, and through working clinically with psychotic patients, Bion revealed how these ties can catastrophically come undone. It is with these ideas, as well as their links to a corresponding set of neuroscientific constructs relating to the Markov blanket and principally developed by Karl Friston, that this paper is concerned. Through an investigation of the psychic functioning originally dubbed ‘dream-work-alpha’, the paper’s first section focuses on how Bion conceived of the creation of a ‘contact-barrier’ that allows for the differentiation of consciousness from an unconscious mind. Casting the ramifications of this organisation in sharp relief, the psychotic disorganisation of the contact-barrier is then explored. The discussion subsequently broadens to incorporate contemporary theories from free energy neuroscience that bear significant and illuminating relations to the psychoanalytic ideas espoused by Bion over half a century ago. Finally, through posing a series of three questions with accompanying discussions, a superimposition of these theoretical schemas is attempted. These suggestions directly address 1. whether there is an intimate connection between the interoceptive contact-barrier and the exteroceptive Markov blanket, 2. whether a disobjectalising of the contact-barrier may be reflected as a tear in the functional fabric of the Markov blanket, and 3. what the clinical implications are of working at the level of the projected surface. Ultimately, the aim of the paper is to expose relevant points of contact within and between the varying conceptual frameworks; frameworks that ultimately derive from disciplines that are both concerned with examining the underlying mechanisms of the mind-brain.

UPDATED (15/10/2021). this is still an unsubmitted WIP[feel free to read (and cite!), but keep in mind it might change!]

Predictive processing and extended consciousness: why the machinery of consciousness is (probably) still in the head and the DEUTS argument won't let it leak outside Abstract Consciousness vehicle externalism (CVE) is the claim that the material machinery of a subject's phenomenology partially leaks outside a subject's brain, encompassing bodily and environmental structures. The DEUTS argument is the most prominent argument for CVE in the sensorimotor enactivists' arsenal. In a recent series of publications, Kirchhoff and Kiverstein have deployed such an argument to claim that a prominent view of neural processing, namely predictive processing, is fully compatible with CVE. Indeed, in Kirchhoff and Kiverstein's view, a proper understanding of predictive processing mandates CVE. In this essay, we critically examine Kirchhoff and Kiverstein's argument. Our aim is to argue in favor of the following three points. First, that Kirchhoff and Kiverstein's emphasis on cultural practices lends no support to CVE: at best, it vindicates some form of content externalism about phenomenal content. Secondly, the criteria Kirchhoff and Kiverstein propose to identify a subject's phenomenal machinery greatly overgeneralize, leaving them open to a "consciousness bloat" objection, which is an analog of the cognitive bloat objection against the extended mind. Lastly, we will argue that the "consciousness bloat" problem is inbuilt in the very argumentative structure of the DEUTS argument. We will thus conclude that, contrary to the philosophical mainstream, DEUTS is not the best argument for CVE in the sensorimotor enactivists' argumentative arsenal.

Abstract. Extracting sentiments from unstructured text has emerged as an important problem in many disciplines. An accurate method would enable us, for example, to mine on-line opinions from the Internet and learn customers ’ preferences for economic or marketing research, or for leveraging a strategic advantage. In this paper, we propose a two-stage Bayesian algorithm that is able to capture the dependencies among words, and, at the same time, finds a vocabulary that is efficient for the purpose of extracting sentiments. Experimental results on the Movie Reviews data set show that our algorithm is able to select a parsimonious feature set with substantially fewer predictor variables than in the full data set and leads to better predictions about sentiment orientations than several state-of-the-art machine learning methods. Our findings suggest that sentiments are captured by conditional dependence relations

This is a poster presented at the RU Bochum at the Predictive Processing, Consciousness and Self workshop. In it, I argue that Predictive Processing (PP) when understood in terms of hierarchically layered Markov blankets packed with representation requirements is internally inconsistent. In short, the argument is as follows. The states external to the blanket require representation for the internal states. The states within a Markov blanket (A) that is itself embedded in a further Markov blanket (B) will be required to represent the states of that further Markov blanket as those states are external to the smaller system. However, when the system is viewed as AB, both the states of A and the states of B are internal to the system and the states of B do not require representation for the states of A. The states of A thus need to both represent and not represent the states of B depending on where we look. I propose to remedy this by going non-representational. In this proposal I rely heavily on the work of Michael Kirchhoff and Nico Orlandi to make my case.