Allostasis Machines: a model for understanding internal states and technological environments

Abstract

In the present paper we will approach enactivism from the perspective of internal regulation: while the environment shapes the organism, it is also true that organisms have complex internal states with regulatory machinery with a set of continuous phenotype-environment interactions. The aim of the present paper is to provide a visual means to analyze these interactions in individuals and computational agents alike. An essential component of our approach is the representation of continuous internal states through the usage of the single continuous indicator we call an Allostasis Machine (AM). Consequently, we consider potential perturbation regimes for both naturalistic and virtual environments: within the naturalistic cases, it is possible to observe the effects of perturbations in isolation, or as overlapping, multiplicative events. In virtual cases, we can observe perturbations as the outcome of both realistic and fantastical environments. To conclude, we discuss how AMs can be utilized to improve our understanding of both the theoretical basis of embodied interaction and the dynamic regulation of complex psychophysiological states.

Allostasis Machines as Continuous Cognitive Modeling

“…We also introduce three ways in which isolated, serial, and new state perturbation can affect the regulation and achievement of allostasis. In [8], we discuss the role of AM in enactive and autopoietic processes related to cognition, which can add to the interpretation of AM output dynamics. Another way to analyze the relationship between input, output, and state variables is to think of AMs as ergodic dynamical systems [24].…”

“…States in an AM are defined as stable portions of the output trajectory over time. AM stability is similar to low-dimensional continuous attractor dynamics often used to characterize collective behaviors in neuronal (brain) networks [8].…”