OREL Flash Talks @ Neuromatch 5
Neuromatch 5 begins this week (Tuesday the 27th and Wednesday the 28th) on Crowdcast. The Orthogonal Research and Education Lab has four (4) presentations covering topics from bio-inspired cognitive modeling to phenomenology, and from embodiment to intuitionism. Engage via YouTube, with our preprints for more information, or through the conference chat channels (Reddit).
Research Methods in Cognition Studies and Phenomenology: challenges and opportunities.
Jesse Parent, Amanda Nelson, Brian McCorkle, Avery Lim, Morgan Hough, and Bradly Alicea
[YouTube][Cognition Futures][Reddit chat]
The role of the interface between internal and external factors which shape an agent’s experience have been considered for millennia. In the 21st century, we have access to new technologies and ideas, but what are the breakthroughs we need that will afford new discoveries? Similarly, what disciplinary boundaries are appropriate as opposed to inhibitory; how much does neuroscience overlap with or have functional or explanatory distinctness from psychology, or other disciplines within the original umbrella of the Cognitive Science project? These questions drive our work in methods for cognition studies and phenomenology.
The Research Methods in CogSci project aims to investigate methodological limitations, opportunities, and questions within the broad array of disciplines contributing to our understanding and framing of what cognition is. To address such questions, we first seek to categorize and explore specific methodologies in use for exploring or modeling cognition, the paradigms which are supported or challenged by such methods’ findings, and the philosophical claims underpinning such investigations. We also seek to consider emerging technologies in brain-computer interface and virtual / augmented reality as research mediums. Future steps include conducting interviews with practitioners across different fields of study, and mapping the frontiers of what methodologies currently are capable of, which may designate fertile areas for future research.
Intuitionistic Mathematics and Neurophenomenology.
Brian McCorkle
[YouTube][Blogpost][Reddit chat]
What are the parallels between Intuitionistic Mathematics and Neurophenomenology, and their implications for how we conceptualize Computation? Inspired by AA Cavia’s recently published ‘Logiciel’ and work with the OREL Cognition Futures Reading Group, I hope to connect these foundational innovations in Mathematics and Cognitive Science to show a path toward an intentional, ethical, and suitably complex computational future.
First an historical survey of the development of the two fields from the perspective of their initiators, LEJ Brouwer and Francisco Varela. Then a folding in of advancements beyond their founders, Martin-Löf and Homotopy Type Theory in the case of Intuitionism and the many facets and implementations of the enactive approach at work today for Neurophenomenology. Finally, some musings on a combined research program and what fruits it may bring.
Layers, Folds, and Semi-Neuronal Information Processing.
Jesse Parent, Avery Lim, and Bradly Alicea
[YouTube][Preprint][Reddit chat]
What role does phenotypic complexity play in the systems-level function of an embodied agent? The organismal phenotype is a topologically complex structure that interacts with a genotype, developmental physics, and an informational environment. Using this observation as inspiration, we utilize a type of embodied agent that exhibits layered representational capacity: meta-brain models. Meta-brains are used to demonstrate how zonally organized phenotypes process information and exhibit self-regulation from development to maturity. We focus on two biological phenomena that explain this capacity: folding and layering. The formation of meta-brain zonal structure and resulting modular interactions can be demonstrated in the context of phenomena such as connectivity activation encoding, morphogenetic encodings, and developmental contingency.
As layering and folding can be observed in a host of biological contexts, they also form the basis for our hybrid computational representations. First, an innate starting point (genomic encoding) is described. The generative output of this encoding is a differentiation tree, which results in a layered phenotypic representation. A formal meta-brain model of the gut is shown to exhibit folding and layering in development along with different degrees of representation for processed information. This organ topology is retained in maturity, with additional folding and representational drift arising in response to inflammatory stimuli. Next, we consider topological remapping using the developmental Braitenberg Vehicle (dBV) as a toy model. During topological remapping, it is shown that folding of a layered neural network can introduce a number of distortions to the original model, some with functional implications. The paper concludes with a discussion on how the meta-brains method can assist us in the investigation of enactivism, holism, and cognitive processing in the context of biological simulation. Understanding semi-neuronal information processing more generally can help us work with the bodies of soft robots, particularly those that rely upon distributed information processing.
Intelligence Offloading for Developmental Neurosimulation.
Bradly Alicea, Angela Pang, and Jesse Parent
[YouTube][Preprint][Reddit chat]
Cognitive offloading occurs when environmental affordances expand cognitive capacity as well as facilitate spatial and social behaviors. Yet capacity-related constraints are also important, particularly as an embodied agent comes online during development. Vast differences in brain size and offloading capacity exist across the tree of life. Taking a developmental perspective, we utilize several concepts from cybernetics and the evolution of development to understand how we might determine the proportion of internal model (brain) to externalized processing (offloading) in a developing embodied computational agent. As developing animal nervous systems scale with body size and/or functional importance, offloading capacity is also driven by neural capacity. Thus, cognitive capacity is ultimately determined by various innate and environmental constraints. We propose that a similar model can be used to develop cognitive systems for computational agents. A regulatory model of cognition is proposed as a means to build a cognitive system that interfaces with a biologically-inspired substrate. Multiple tradeoffs resulting from energetic, innate, and informational constraints determine the proportion of internal to external information processing capacity. As growth of the biologically-inspired substrate accelerates or decelerates over developmental time, it changes the acquisition capacity of the cognitive system. Capacity limitations of our agent’s internal model determine the externalization potential, which is characterized by three parameters and two mathematical functions. This approach to simulating developmentally-inspired cognitive regulation can be applied to a broad range of agent-based models and Artificial Intelligence approaches.
