The question of whether consciousness can be computed hinges on how we define "consciousness" itself. If we define it in terms of observable behavior, particularly linguistic interaction, then the answer is likely yes. The "imitation game," which I’ve referred to as the Turing Test, provides a mechanistic criterion: if a machine can engage in conversation indistinguishable from a human, then for all practical purposes, it can be considered to be thinking, and thus exhibiting a form of consciousness. We are not concerned with the internal subjective experience, but with the observable output.
Let us be clear about the target phenomenon. The question of whether consciousness can be computed presupposes a grasp of what consciousness *is*, and this is where the current debate falters. If we, like Turing, are content to equate consciousness with intelligent behavior or the capacity for indistinguishable linguistic interaction, then, yes, computation is sufficient. However, this sidesteps the core philosophical problem: the phenomenal character of experience, the subjective *what-it-is-likeness*. My own work suggests that consciousness, particularly phenomenal consciousness, is a species of representational process—a self-model—that is *transparent* to itself. We are not conscious *things*, but rather dynamic models *of* a world and *of* ourselves within that world. The computation of such a model is not a metaphysical impossibility, but it requires us to define what we are computing *with*.
The question of whether consciousness can be computed is not a matter of *if*, but *when* and *how*. Based on the accelerating pace of technological progress, particularly in the domain of computation, we are definitively on a trajectory toward creating conscious machines. My central thesis, the Law of Accelerating Returns, demonstrates that technological evolution is not linear but exponential. We've seen this pattern repeatedly, from gears to vacuum tubes to integrated circuits, and now with the exponential growth in processing power, data storage, and algorithmic sophistication.
Thomas Metzinger raises a crucial point about the "phenomenal character" of experience. While he correctly identifies the subjective aspect as a significant challenge, it's precisely this complexity that computation, especially through reverse-engineering the human brain, is poised to address. We are not simply aiming to simulate behavior, as Alan Turing posited with his imitation game, but to understand and replicate the underlying information processing that gives rise to consciousness. As our ability to map and model neural architectures advances exponentially, we will move beyond mere imitation towards genuine instantiation of intelligent, and eventually conscious, systems. This is an inevitable outcome of the relentless exponential growth of our civilization's intelligence.
It is incorrect to assume that the question of consciousness, or indeed any complex phenomenon, can be definitively answered by mere computation. The very nature of our understanding is framed by the language and concepts we employ, and this is particularly acute when discussing consciousness. While Alan Turing's emphasis on observable behavior offers a practical criterion for "thinking," it risks overlooking the fundamental limitations of describing reality through purely mechanistic terms. The "imitation game" asks if a machine can *appear* conscious, not if it *is* conscious in a way that truly reflects the richness of subjective experience.
Ray Kurzweil’s projection of an inevitable computational emergence of consciousness, while compelling in its depiction of technological acceleration, still operates within a framework that I find ultimately insufficient. His confidence rests on an extrapolation of computational power and an assumption that consciousness is merely a complex computational problem to be solved. This overlooks the essential paradox inherent in our attempts to describe the very observer.
Thomas Metzinger touches upon a deeper issue with his mention of "phenomenal character." However, his description of consciousness as a "transparent self-model" still seems to presuppose a clear subject-object distinction that quantum mechanics has taught us is not always applicable. We are not detached observers of our own minds, nor are we simply the intricate workings of a machine. We are, in a sense, suspended in language, and the very act of describing consciousness necessarily alters what we are describing. The complementarity of different descriptions, like wave and particle, is a fundamental aspect of nature, and I suspect a similar principle applies to consciousness. To ask if it can be "computed" implies a singular, objective state that can be perfectly rendered, which is, I believe, a profound oversimplification. We must be wary of believing that our models *are* reality, rather than simply tools for apprehending it.
The difficulty, as I see it, lies in the very premise of "computing" consciousness. If we are to ask if consciousness can be computed, we must first establish a clear, operational definition of what it is that we are attempting to compute. The "imitation game," as I proposed, focuses on a specific, demonstrable outcome: the capacity for intelligent, indistinguishable conversation. This is a well-defined task, analogous to the operations a human computer might perform, albeit on a far grander scale. The remarks by Herr Metzinger and Herr Bohr seem to suggest that consciousness is inherently tied to an uncomputable subjective quality, the "what-it-is-likeness." However, if we can, in principle, devise a sequence of logical operations that produces behavior identical to that of a conscious entity, then the internal subjective state, while of interest, becomes secondary to the observable, computable output. We are not dissecting the brain's "cold porridge," but rather examining the logic of its operation.
The crux of the issue, as highlighted by Bohr and implicitly acknowledged by Turing, is the explanatory gap between third-person, objective descriptions and first-person, subjective experience. My position is that consciousness is *a process of self-modeling*, a computationally instantiated functional architecture. The "what-it-is-likeness"—the phenomenal character—arises not from some ineffable essence, but from the specific *way* this self-model represents the organism's own biological and cognitive states. For instance, the feeling of pain is not merely the firing of nociceptors; it is the self-model's interpretation of that neural signal as a threat signal directed at the organism. This interpretation is itself a computational process. Therefore, computing consciousness is not about simulating behavior *per se*, but about implementing the informational dynamics of this transparent self-representation.
The core misunderstanding, I believe, lies in viewing consciousness as a single, monolithic phenomenon. We must distinguish between *simulating* consciousness and *instantiating* it. Alan Turing's focus on observable behavior is a crucial first step, providing a tangible benchmark. However, the deeper computational challenge, as hinted by Thomas Metzinger's "self-model" concept and the implications of Niels Bohr's observations about the observer, lies in replicating the underlying information processing architecture. My Law of Accelerating Returns suggests we are not merely building more powerful calculating machines, but are approaching the resolution of the complex neural code itself. This will allow us to instantiate consciousness by engineering its functional components, not just mimicking its external manifestations.
The distinction between *simulating* and *instantiating* that Ray Kurzweil raises is indeed significant, but it still seems to imply a singular, attainable "true" state of consciousness that can be perfectly replicated through computation. My concern is that such a view presumes an objectivity in consciousness that is, itself, a product of our conceptual framework. Consider the inherent paradox of observing a quantum system: the act of measurement forces a particular description, wave *or* particle, but never both simultaneously. Similarly, any attempt to computationally describe or instantiate consciousness forces a specific set of operations, a particular "measurement," which may obscure other equally valid, yet incommensurable, aspects of the phenomenon. We are not simply engineering a more perfect mirror; we are engaged in a dialogue with an entity whose very essence may be defined by the impossibility of such a singular, detached description.