You receive a paper to review. A critique of neural computation

Reconciling Mereological and Informational-Closure Emergence

Abstract:

In recent years, the impetus in research into emergence (broadly characterised as the relationship between microscopic and macroscopic levels of description of a complex system) has arguably shifted from the ontological "What is the nature of emergence?" to the epistemic "How shall we quantify emergence?" In particular, several information-theoretic approaches to measuring emergence have appeared in the past few years. In these studies, two apparently unrelated key principles have come to the fore: mereological principles [1], which focus on causal relationships between the whole and constituent parts of a system, and informational closure [2], which in the context of emergence is concerned with autonomy of macroscopic with respect to microscopic dynamics. Measures of emergence based on these two distinct principles are exemplified by Causal Emergence [3] and Dynamical Independence [4] respectively. I shall present an intriguing relationship between these measures, which may illuminate the different facets of emergence represented by mereological and informational-closure principles.

References

[1] Varzi, Mereology, in The Stanford Encyclopedia of Philosophy, 2016. 

[2] Bertschinger et al., Autonomy: An information-theoretic perspective. Biosystems 91(2), 2008. 

[3] Rosas et al., Reconciling emergences: An information-theoretic approach to identify causal emergence in multivariate data. PLoS Comput. Biol. 16(12), 2021. 

[4] Barnett & Seth, Dynamical independence: Discovering emergent macroscopic processes in complex dynamical systems. Phys. Rev. E 108, 2023. 

Agency, cognition and computation.

Abstract:

We present the idea that causal relations between system and environment can be explored through the lens of proto-cognitive properties. And that the notions of under-determination and self-determination can be attributed to autonomous dynamics, that may be measured through information metrics in the discrete case. 

From this, we argue that this causal influence of the system upon its future self is necessary, although not sufficient for agency. 

Furthermore, we argue that this is not, or not only because of the complexity required by the kind of systems we conceive as agents, but fundamentally, given by the purely mechanistic nature of the operations guiding the dynamics of proto-cognitive systems (i.e. void of a mind).

We believe that this reopens the discussion about the difference between cognition and computation, and the hypothetical role of emergence as a potential source of better understanding.

We'll focus on how complex systems across various domains generate emergent phenomena that are irreducible to, yet fundamentally dependent on, the organisation of their underlying materiality.

What are mental disorders, really?

Abstract:

What are mental disorders, really? This is a question that has long occupied philosophers and psychologists. The DSM, which classifies conditions such as depression or ADHD, defines disorders in terms of their symptoms. However, it does not offer an explanation of their underlying cause, which is rather surprising, as the very use of the word symptom implicitly suggests that such an underlying cause must exist.

Borsboom has introduced a new account, the network theory of psychopathology. According to this view, mental disorders do not require an underlying cause at all. Instead, symptoms cause and reinforce one another, such that a disorder can be understood as a self-sustaining network of symptoms.

But does this framework really offer an explanation of mental disorders? In this presentation, I will argue that by drawing on Kostic’s framework of topological explanation, if we can give a rich interpretation of symptom networks, the topological properties of networks can provide a causal explanation of mental disorders.

Mapping the emergent landscape of conscious states

Abstract:

Quantifying emergence – the operationalisation of identifying wholes that are ‘greater’ than the sum of their parts – has undergone a transformation with the advent of new tools, each capturing a distinct flavour of the phenomenon. Here, I focus on one such aspect of emergence: dynamical closure, operationalised through an information-theoretic measure of dynamical (in)dependence. The emphasis is on identifying and quantifying the closure of macroscopic processes across spatiotemporal scales. 

Applied to empirical brain data, primarily electroencephalography (EEG), this approach permits a mapping of conscious states in terms of their underlying emergent dynamical landscape. By unfolding emergent dynamics across scales, we can characterise conscious states not only by an overall degree of emergence, but also by the organisation of the emergent landscape itself. 

I illustrate this with results spanning wakefulness, anaesthetic-induced loss and alteration of consciousness, various sleep stages, and the fragmentation of selfhood under 5-MeO-DMT. These findings support a reframing of how emergence and integration are viewed in conscious processing. First, they reveal drug- and state-specific reconfigurations of emergent structure, dissociating the amount of emergence from the organisation of emergent dynamics, and cautioning against equating emergence with level of consciousness. Second, they suggest that wakeful consciousness depends not only on integration within a single scale, but also on integration across scales, broadening current assumptions about key biomarkers of consciousness.

Why Many Biological Agents are Conscious and A.I. Isn't 

Abstract:

A.I. isn’t conscious because consciousness requires properties biological agents have that A.I. doesn’t. What A.I. lacks that is relevant for consciousness is not predictive processing or the minimization of free energy. A.I. lacks embodiment, direct coupling with the environment, and multiscale complexity--plasticity, degeneracy, robustness. A.I. lacks the properties that enable living organisms to be highly adaptive to environmental changes.

The big 5 personality traits as emergence properties

Abstract: TBC

The Importance of Boundaries

Abstract:

Contextual emergence, in which both the local constituents and various contextual constraints are necessary for the emergent phenomenon, is conceptually distinct from supervenience. Some nonetheless wonder whether it is explicable in terms of broad supervenience, extending the supervenience base to the contextual surround. This talk will explore the importance of macro-level boundaries in contextual emergence, which are merely descriptive in supervenient systems, but functional in living ones. My claim will be that we can only make sense of a subject and its consciousness through the use of such a boundary.

Ben White

followed by Q&A

Bryce Huebner

Luiz Pessoa

follwed by Q&A

Bio

I am a musician and musical instrument designer, specialising in musician-computer interaction, physical computing, and machine learning. Currently my research focuses on feedback (or multistable) musicianship, machine learning and livecoding. I co-ran the AHRC Feedback Musicianship Network, and I was recently awarded an AHRC Fellowship grant, Musically Embodied Machine Learning. I'm also involved in projects on robot music theatre and ecoacoustics.

My main approach to research is to learn from building and performing with new musical instruments, both through my own experience, and through engaging in participatory design projects with others.

Recently instruments I have devloped include The Nalima, a membrane-based feedback instruments, and the Feedback Cello, a hacked cello based on the halldorophone. I'm also very interested in analogue/hybrid sound synthesis, and have been developing a eurorack module for livecoding, uSEQ.

Underpinning my research is a focus on complex and dynamical systems, and signal processing with machine learning.


Links:
Personal website: http://luuma.net
Code: https://github.com/chriskiefer