Refinement
Refinement: a rigorous description of autonomous adaptive agents; Kybernetes, vol.42 no.9/10, 2013
J. Augustus Bacigalupi
Principal Researcher - Institute for Augmenting Minds
This paper describes the theoretical aspects of SOARSE technology, which is summarized on the TECH page.  It describes, both informally and formally, the core concepts that are used to test and refine built implementations.  Success, or the Alt. Hypothesis, is achieved when Refinement is achieved, which is informally defined as follows:
Refinement: a process whereby a system increases its sentience (engagement, sensitivity and awareness) of its complex surroundings by increasing its own internal capacity to resolve internal noise, which is the result of increased sentience, into an increasingly varied and inter-dependent internal structure able to increase capacity to do adaptive work, which are the increasingly complex behaviors that tend, on average, towards Refinement.

click on image below for full published paper


from constraint to organization at the origin of life
From Constraint to Regulation at the Origin of Life; Frontiers in Bioscience, 19:945-957, June 2014
Terrence W. Deacon, Alok Srivastava, J. Augustus Bacigalupi
Researcher - Berkeley Consortium on Emergent Dynamics
Abstract: The origin of living dynamics required an evasion of second law degradation effects by maintaining critical dynamical and structural constraints, by synthesizing new components for replacement and reproduction, and regulating these interactions with respect to critical intrinsic needs and extrinsic conditions. Model systems for life’s origin that focus on molecular replication (e.g RNA-world), or co-production of components (e.g. autopoiesis), or physical containment of molecular interactions (e.g. protocells) fail to distinguish between constrained chemistry and regulated metabolism. For this reason they do not address the question of how living processes first emerge from simpler constraints on molecular interactions. We begin with a simple molecular model system consisting of coupled reciprocal catalysis and self-assembly in which one of the catalytic bi-products tends to spontaneously self-assemble into a containing shell (analogous to a viral capsule). We term this dynamical relationship autogenesis because it is self-reconstituting in response to degradation. Self-reconstitution (and reproduction) is made possible by the fact that each of these linked self-organizing processes generates boundary constraints that promote the perdurance of their combination by limiting each other. This synergy thereby becomes embodied as a persistent rate-independent constraint on dynamical constraint generation. It is proposed that this synergy of dynamical/structural constraint-generation is necessary and sufficient to constitute regulation as opposed to mere constraint. Two minor elaborations of this simple model system demonstrate that this simplest form of regulation can be the foundation for the evolution of two higher-order forms: cybernetic and template-based regulation. We offer this thought experiment as a plausibility proof that regulatory relationships can spontaneously emerge from physicochemical constraints.
click image below for published paper
human eeg responses
Human EEG Responses to a Conditioned Stimulus Reward Delay; poster presented at Neuroscience 2007
A. V. Popple, A. C. Provost, J. A. Bacigalupi, T. Carney, S. A. Klein, D. M. Levi
Researcher - Professor D.M. Levi's lab; Hellen Wills Neuroscience Institute, UC Berkeley
We studied in humans reports1 that visual information from a rat's eyes is modified by past experience before it is processed by the lowest levels of the rat's visual cortex.
Back to Top