• View
• Edit
• History
• Print

Modeling Introspection

Sources to consider

• Introspection
• Stanford Encyclopedia. Introspection.
• Elizabeth Irvine. Subjective Measures of Consciousness.
• Tasha Eurich. The right way to be introspective (yes, there’s a wrong way).
• David Ruhe. C2C 1: Complex and Quaternion Neural Networks
• David Ruhe. C2C 3: Geometric Clifford Algebra Networks

Models of Consciousness Conference

Videos

• 2023 Programme
  • Maja Spener (Invited), Department of Philosophy, University of Birmingham, UK. Are subjective measures of consciousness reliable? The Reliability of Subjective Measures of Consciousness, Maja Spener. Introspection: First-Person Access in Science and Agency.
    • Subjective measures of consciousness face a well-known, seemingly intractable challenge: they use introspection to gain access to conscious data, but introspection is unreliable. In this talk, I outline a framework for evaluating subjective measures of consciousness, based on important but neglected distinctions between different types of introspection. I then show how to apply this framework to evaluate a mainstream subjective measure of consciousness, the Perceptual Awareness Scale.
    • In her book she seems to focus on the distinction between inner attention (deliberately attend to your ongoing thinking and experiencing), inner apprehension (perhipherally aware of your ongoing experiencing) and retrospection (recall an experience after it has passed), all of which have to do with sensory experience of our life in the world rather than our life in the mind. Whereas I focus on the possibility of perspective, what perspectives I can take up, what perspectives I can choose from.
  • Kenneth Williford. The Projective Consciousness Model: Phenomenological Prolegomena.
    • Projective Consciousness Model Research Group. According to the Projective Consciousness Model (PCM), consciousness is structured by (3D) Projective geometry; and perception, action, and imagination entail perspectival transformations governed by the action of the Projective Group, driven by a value-optimization dynamics (e.g., variational Free Energy minimization). To set the stage for a more formal presentation of the PCM in a companion presentation, we here offer some of the phenomenological data that motivate the theory and briefly indicate how the theory captures them. These data include: the sense of perspective or point of view (in vision but in other sensory modalities and imagination as well), the elusiveness of the point of view (or "subject"), the sense of a vanishing point or limit of experiential space, the anticipatory dynamics involved in the perspectival navigation of ambient Euclidean space, reciprocity (and what we call "proto-intersubjectivity"), perspectival anomalies (e.g., OBEs, autoscopy, heautoscopy), some common illusions (esp. the Moon Illusion and the Ames Room Illusion), and key features of pre-reflective self-consciousness. We close by illustrating how the PCM generates novel predictions that can be tested using fairly straightforward psycho-physical paradigms, provides guidance for simulation strategies geared toward artificial intelligence, and raises some intriguing theoretico-mathematical questions.
  • David Rudrauf. The Projective Consciousness Model: Formal developments from projective geometry to epistemic drives. Grégoire Sergeant-Perthuis, David Rudrauf, & Kenneth Williford.
    • The Projective Consciousness Model (PCM) yields a testable framework for understanding the relationships between consciousness, cognition, and behaviours. See companion presentation of Williford et al. for motivation of the model. In the PCM, the space of conscious experience acts as a homogeneous workspace for representation and action planning. 3D projective geometry models the subjective perspective of agents, and the 3D projective group structures their internal representations for active inference or more generally (stochastic) optimal control. The action of the projective group directly contributes to maximising expectation satisfaction and information gain, resulting in different adaptive and maladaptive strategies of exploration and exploitation, including in agents capable of Theory of Mind. We proved mathematically how changing the group (Euclidean versus Projective) that structures the agents’ internal space strongly influences how « epistemic value » is quantified and maximised, and ensuing agents’ exploratory behaviours. We discuss how this extends to general groups and the stability of group structured representations. The PCM offers a framework for designing experiments with human participants, to falsify or validate the predictions of the model. Our goal is both to build a theory of consciousness, and derive consciousness-inspired principles for the control of social robots.
  • Zhouwanyue Nata Yang, MCMP, LMU, Germany. What justifies the application of the Active Inference Theory of consciousness?
    • The Active Bayesian Inference Theory (AIT) posits a process-based explanation for the emergence of consciousness. Despite its extensive explanatory power, the underlying conceptual foundation of AIT remains underexplored. This talk introduces this theory within an explanation framework consisting of the explanatory account and the to-be-explained subject. In term of category theory, we investigate the implicit presupposition that justifies AIT’s application in explaining consciousness emergence. Within this framework, we consider two types of prediction-perception processes: one employed by AIT to study consciousness, and the other that deviates from the AIT’s prerequisites. Both types of prediction-perception processes are embedded in a category equipped with an evaluation morphism. In light of Lawvere's Fixed Point theorem, we explicate criteria for AIT’s application in consciousness studies. Building upon this, we employ categorical pullback to examine the formal condition of AIT's inherent presupposition. Finally, we interpret this presupposition as cognitive capacity, which not only retains AIT's deflationary nature but, crucially, also provides a justification for AIT’s application
  • Joscha Bach, Thistledown Foundation, United States. Consciousness as a training mechanism for self organizing modelling systems
    • A theory of consciousness should capture its phenomenology, characterize its ontological status and extent, explain its causal structure and genesis, and describe its function. Here, I advance the notion that consciousness is best understood as an operator, in the sense of a physically implemented transition function that is acting on a representational substrate and controls its temporal evolution, and as such has no identity as an object or thing, but (like software running on a digital computer) it can be characterized as a law. Starting from the observation that biological information processing in multicellular substrates is based on self organization, I explore the conjecture that the functionality of consciousness represents the simplest algorithm that is discoverable by such substrates, and can impose function approximation via increasing representational coherence. I describe some properties of this operator, both with the goal of recovering the phenomenology of consciousness, and to get closer to a specification that would allow recreating it in computational simulations.
  • Robert Prentner - Mathematizing phenomenology within a process framework
    • Phenomenology is one of our most comprehensive resources for systematically analysing conscious (first-personal) experiences. For example, phenomenologists typically emphasise the correlation between subjective and objective aspects: a fundamental “self/world” structure that is constitutive of any conscious experience. In addition, we demonstrate that such ideas can be made more precise using mathematics, thus contributing to mathematical consciousness science. Yet, the question of where to conceptually locate phenomenology is still open. Is it a form of (transcendental) idealism? Or is it a high-level account of physical (brain) processes? We will argue instead that a non-dual framework, which presupposes neither mental nor material primitives, is best suited to ground phenomenological studies. A candidate is provided by process philosophy with its concept of global relatedness. We briefly review the process framework and establish the connection with phenomenology.
  • Alex Kostova, Sofia University "St. Kliment Ohridski", Bulgaria. Formalizing panpsychism through Integrated Information Theory.
    • Panpsychism posits the perplexing primacy of consciousness considering it a pervasive feature that permeates the fabric of the universe and extends beyond human consciousness. Panpsychism has undergone a notable resurgence in the past few decades that can be attributed to the persistently insurrmountable challenge that despite the advancements in comprehending the intricacies of the brain and its interplay with conscious states, the fundamental question of how consciousness arises from ostensibly non-conscious elements of the physical realm remains unresolved. For instance, the solution of Sautrantika focuses on the experiential complexity of mental phenomena prioritizing their phenomenological accessability, rather than delving into their essence. In this paper I will investigate the possibility of formalizing panpsychism within the IIT framework that could capture the idea of the panpsychist systems that emergence of consciousness within the physical realm would have been implausible, if not inconceivable, without the preexisting presence of the mental fabric from the very outset of existence.
  • Ouri Wolfson, Computer Science, University of Illinois, Chicago, United States. Consciousness as a form of coordination.
    • We present a hypothesis that consciousness is related to traffic of signals in the brain; specifically, that consciousness is produced by a type of coordination that is required to move a traffic system from user equilibrium to system optimum. The hypothesis has been verified by datamining of connectomes. Our datamining approach considers the brain regions and the tracts that connect them as a road network, and the signals traveling between them as traffic. We analyze travel patterns by a process called traffic assignment. The results are unexpected in the sense that the movement of signals in the brain seems to follow some global optimization patterns as opposed to the anarchical system that would be favored by evolution. The hypothesis can be further examined by datamining the connectomes of human and animal subjects. Thus, our results can be viewed as a neural correlate of consciousness. However, the results may be deeper in the sense that coordination may be the mechanism that actually produces the subjective experience. In other words, is subjective experience the way the ability to coordinate brain activities manifests itself? This is supported by the fact that coordination is a form of control or agency.
  • Serena Doria, Department of Engineering and Geology, University of Chieti-Pescara, Italy. Conditional probability models based on fractal measures to represent the awareness process
    • In the subjective approach to probability, conditional probability and its extensions represent the level of partial knowledge we have about a phenomenon based on partial information. A new model of coherent upper and lower conditional probabilities defined by Hausdorff outer and inner measures has been proposed to represent respectively the unconscious and conscious activities of the human brain. In the model uncertainty measures are defined according to the complexity of the conditioning event that represents a piece of information. According to the model the partial knowledge is updated when theconditioning event has positive and finite Hausdorff measure in its Hausdorff dimension. In that case, if the Hausdorff dimension s of the conditioning event is less than the Hausdorff dimension t of Omega, the conditioning event is an event with zero probability with respect to the prior probability; the new Hausdorff outer measure is considered to define conditional probability and to update partial knowledge. According to the model the unexpected events, represented by sets with zero probability, are those which really update the knowledge. The awareness process of human beings is described by a class of coherent conditional defined by different fractal measures.
• 2022 Programme
  • Moritz Kriegleder, Faculty of Philosophy and Education, University of Vienna, Austria. The role of information in the Free Energy Principle
    • The Free Energy Principle is a computational model of self-organisation that aims for a unified explanation of living and conscious systems. While deriving free energy from physicist von Helmholtz, it has become clear that explaining consciousness with free energy involves different assumptions and novel interpretations in contrast to physics. Recently, the authors focus on an informational interpretation of free energy, where minimising free energy corresponds to minimising uncertainty about future states of the environment. But, as I argue in my talk, a general model of consciousness has to do more ontological work. Framing cognition as minimising uncertainty or equivalently maximising mutual information of the agents model and the environment is ignoring recent important insights in cognitive science and philosophy of mind, such as embodied cognition and enactivism. I review the main points of these approaches to explain consciousness and subjective experience and evaluate the possibilities of the free energy principle to incorporate embodied and enacted inference. A comparison with other models of consciousness highlights the difference in the approaches to explaining the phenomenological qualities of the mind. I conclude that the free energy principle leaves open its ontological commitments and that clarifying the philosophical foundations can aid its future development.
  • Tudor Baetu, Arts et philosophie, Université du Québec à Trois-Rivières, Canada. Informational models of the phenomenon of consciousness and the mechanistic project in neuroscience
    • I argue that the main contribution of informational models of consciousness, including the popular Integrated Information Theory (IIT), is methodological in nature, amounting to a quantitative recharacterization of the phenomenon of consciousness. This recharacterization offers a more detailed and complete description of the phenomenon; provides the basis for assessment methods of accrued sensitivity, specificity, and content validity; and is expected to guide mechanistic research. However, the issue of the physical interpretation of informational relationships underpinning quantitative models of the phenomenon of consciousness is open to debate. In particular, the panpsychist interpretation proposed by IIT is insufficiently well justified and plagued by internal contradictions.
  • Robert Prentner, Munich Center for Mathematical Philosophy, LMU Munich, Germany. Mathematized phenomenology and the scientific study of consciousness
    • Many consciousness researchers share the idea that a science of consciousness ultimately needs to target the phenomenological properties of experience. However, when looking at most current theories in the scientific study of consciousness, it is unclear what exactly those properties are supposed to be, how they are operationalized within these theories, and whether this could be systematized. The scientific study of consciousness should therefore pay closer attention to the phenomenology of consciousness beyond encoding top-level intuitions or cherry-picking ideas. A sometimes-neglected philosophical discipline, phenomenology, has tried to more specifically outline the invariant structures that underlie conscious, first-personal experience. What is most urgently needed, we claim, is a formalization of these structures that is precise and amenable to empirical practices in the field, in short: to integrate mathematized phenomenology with the scientific study of consciousness. We will briefly review conceptual, theoretical, and methodological issues and sketch some entry points for mathematical consciousness science. In particular, we focus on the project of transcendental phenomenology, namely to specify the conditions of possibility of having an experience (vis-à-vis a description of these experiences “as such”.
  • Camilo Miguel Signorelli, Department of Computer Science, University of Oxford, United Kingdom. Mathematical phenomenology, from subjectivity to objectivity through mathematical invariants.
    • Mathematical phenomenology is part of the growing field of mathematics of consciousness. It has its roots in the work of the German mathematician and philosopher Edmund Husserl and evolved into the program “Naturalizing Phenomenology” during the late nineties. Currently, a few models of consciousness claim a mathematical phenomenology approach, however, most of them fail to acknowledge the deep scientific consequences of the original project. In this talk, I will contextualize what is mathematical phenomenology by quickly reviewing the progress in the mathematics of consciousness. Then, I introduce a mathematical and phenomenological perspective based on the foundations of mathematics, particularly category theory and mathematical invariants. This framework allows us to sketch how objectivity naturally arises from subjectivity, making the former secondary and the latter primary. Finally, this move seems to be free of philosophical contradictions and acknowledges rigorous and methodological requirements to study subjective experience.
• 2021 Programme