The Ego Machine: a Computational Phenomenal Model

The self is fundamental to human experience. Every conscious moment necessarily includes both the perceived external world and an internal embodied perceiver. Despite this pervasiveness, philosophical and neuro-scientific models of the self are wide ranging and grossly underdeveloped showing more influence from folk wisdom than scientific experimentation. Recent neuro-scientific advances especially in brain imaging technologies are making possible for the first time a truly scientific and objective investigation of the phenomenon of the self. These changes are leading to a “disenchantment of the self” [2] analogous to Weber’s “disenchantment of the world” which followed the enlightenment [1]. The functional models of consciousness resulting from these advances, especially Thomas Metzinger’s phenomenal self model (PSM) promise to provide a coherent scientifically verifiable description of consciousness and self-hood. Surprisingly, many of the functional roles of the PSM are analogous to practical concerns in the construction of modern computational systems. These include the coordination of heterogeneous parallel components, managing system-wide global state and the offloading of computationally expensive functions into specialized sub-components. This work proposes to further investigate Metzinger’s PSM through computational experimentation. A computational implementation of a PSM which aligns PSM functionality with real computational problems such as those mentioned above can be used to test the logical consistency and completeness of Metzinger’s PSM, to gain experiential knowledge of the workings of a PSM, and to perform experimentation on a PSM which would be impossible, impractical or immoral to perform on a living system. In this proposal Metzinger’s PSM is described with a special focus on those aspects relevant to the proposed computational implementation (Section 2). The computational implementation is described and a detailed work plan for its construction is presented (Section 3). Metzinger addresses the morality of implementing artificial consciousness [2], expressing deep concern that such an act would be unethical. This concern is addressed in section 4 in which we present a moral justification for this work. The proposal concludes with a discussion of the implications of such a computational model on our understanding of both consciousness and computational systems (Section 5).