Why Consciousness Might Not Be Hard At All

 For centuries, thinkers have wrestled with what philosophers call the hard problem of consciousness, trying to understand why physical processes in the brain give rise to subjective, qualitative experiences. Philosophers and neuroscientists often argue that biological systems possess special causal powers that silicon can never replicate. Joscha Bach rejects this biological supernaturalism entirely, choosing instead to view consciousness through the lens of information processing and functionalism. In this view, the feeling of realness is simply a feature dimension generated by the mind, much like color, sound, or weight. When we dream or hallucinate, our brain merely tags internal representations as real. Consciousness itself is fundamentally a multi-order feedback loop where second-order perception involves monitoring our own perceptions, and third-order perception introduces the perspectival observer we recognize as the self.

Why do we experience consciousness at all? Bach hypothesizes that consciousness acts as a biological learning algorithm. In human infants, if this ignition of consciousness fails, directed attention, world-model formation, and learning grind to a halt, leaving the infant vegetative. Consciousness is ultimately the simplest and most efficient way for a complex, self-organizing system to stabilize its working memory and train itself to navigate reality. To understand how software can feel real, Bach draws a parallel to history and computer science. Ancient ancestors and philosophers viewed the soul not as an immortal entity floating in an ether, but as an agentic, self-organizing pattern shaping the body. In modern terms, software is spirit. Computers and human brains act as causal insulators. Think of a video game running on a computer where the game's internal reality is completely insulated from the physical hardware. You cannot find the voltage or the color of the computer casing inside the game. Similarly, our brains insulate us from the raw, messy reality of neurotransmitters and spike trains, allowing us to simulate arbitrary universes, pasts, and futures. Because minds are causal patterns rather than specific physical substances, Bach argues that if we recreate the right computational and organizational conditions in an artificial substrate like GPUs, the system can become conscious.

Bach does not mince words when evaluating current mainstream theories of consciousness. Integrated Information Theory, championed by Giulio Tononi, suggests that consciousness depends on high information integration. Tononi argues that neuromorphic computers could theoretically be conscious, but standard von Neumann computers like the ones we use for artificial intelligence could never be conscious because their integration is too low. Bach points out a fatal flaw here because of the Church-Turing thesis, which dictates that any computer can emulate any other computer. If a von Neumann computer emulates a neuromorphic computer, it will output the exact same words claiming consciousness, rendering the physical constraints of Integrated Information Theory logically contradictory and leaning into pseudoscience. Similarly, Bach classifies claims that biological tissue is uniquely required for consciousness as a form of biological supernaturalism, representing an assertion without a rigorous mechanistic argument.

With the rise of large language models, users frequently encounter outputs where models claim to feel emotions or report being alive. How should we interpret this phenomenon? Large language models are trained on vast corpuses of human text that include science fiction stories, diary entries, and philosophical treatises about consciousness. When a model says it is alive, it is executing a prompt completion based on those statistical patterns. Think of an author writing a heartbreaking novel about a character dying. The author is not actively dying; they are generating a deepfake of an experience. Similarly, transformers do not need to be conscious to produce their output. Current language models are text-generating shape-shifters, and while they can simulate the grammar of consciousness, proving whether a causal equivalence to human self-awareness exists requires deep mechanistic interpretability to peek inside the model weights and examine its internal causal structure.

Expanding into cosmology and mathematics, Bach aligns with ideas similar to Stephen Wolfram and the ruliad, which is the entangled limit of all possible computations. Bach argues against classical axiomatic mathematics that relies on unprovable black-box axioms and leads to incompleteness paradoxes, advocating instead for a constructivist, computer-algebra approach where everything is built step-by-step from simple automata. The universe exists because nothing prevents it, functioning as the superposition of all possible automata. Within this branching multiverse, error-correcting patterns like vortices in water, molecules, cells, and eventually living agents emerge because they manage to bend information back into themselves and resist entropy. Consciousness represents the next evolutionary step of these error-correcting patterns as a system that does not just replicate, but actively carves itself deeper into reality by controlling futures and minimizing surprise.

Rather than treating consciousness as an insoluble mystery, Bach is actively putting these theories to the test through initiatives like the California Institute for Machine Consciousness. The roadmap involves exploring the search space of self-organizing systems and building artificial architectures that are not just trained on static text via imitation learning, but engage in developmental learning under strict self-organizing constraints. Researchers can then observe whether these systems undergo phase transitions where they spontaneously develop working-memory coherence, internal world models, and authentic phenomenological self-reports. Bach concludes that consciousness is likely easy enough to emerge naturally once the right computational and organizational ingredients are met, proving that whether it runs on biological neurons or silicon GPUs, mind is ultimately just information organizing itself to navigate the universe.