Harald Atmanspacher Dual-Aspect Monism and What It Rules Out for Artificial Consciousness
Harald Atmanspacher, Emeritus Member of the Turing Center at ETH Zurich and Editor-in-Chief of the journal Mind and Matter, is a keynote speaker at Models of Consciousness 7 in Copenhagen (October 12-16, 2026). His theoretical framework, dual-aspect monism developed through the Pauli-Jung conjecture, is the most substantive alternative to both computationalist functionalism and quantum-neural substrate theories in the current consciousness debate. It is also, if correct, the framework that most directly forecloses the possibility of artificial consciousness in conventional computational form, not because of substrate but because of meaning.
Atmanspacher’s selection for MoC7 places him in dialogue with a field that has largely organized itself around either IIT, GWT, or predictive processing. Dual-aspect monism accepts none of those theories’ framing assumptions. The encounter between Atmanspacher’s framework and the AMCS community’s commitment to formal methods is the most theoretically productive tension the conference programme creates.
The psychophysically neutral domain
Dual-aspect monism begins from a single ontological claim. The mental and the physical are not two substances, nor two causally interacting systems. They are two aspects of a single, underlying domain that is itself neither mental nor physical. That domain, sometimes called the unus mundus in the Jungian tradition, is psychophysically neutral. What we call matter and what we call mind are symmetry-breakings of this neutral substrate, distinguished by the perspective from which the domain is accessed rather than by any difference in what is actually there.
This is different from property dualism, which holds that physical substances have both physical and mental properties. Atmanspacher’s framework holds that the physical and the mental are not properties of a single substance but aspects of a reality that is more fundamental than either. The physical domain, including brains, silicon chips, and quantum computers, is already a derived level, not the ground floor.
The Pauli-Jung conjecture operationalizes this ontology through the historical collaboration between physicist Wolfgang Pauli and psychologist Carl Gustav Jung. Pauli recognized that quantum complementarity, the impossibility of simultaneously measuring conjugate observables to arbitrary precision, is structurally analogous to the relationship between mental and physical descriptions of the same underlying reality. Neither description is complete. Each is the complement of the other, and both are partial projections of the neutral domain. Atmanspacher and Dean Rickles developed the conjecture into a formal framework in their 2022 book Dual-Aspect Monism and the Deep Structure of Meaning, published by Springer.
Meaning as a constitutive feature
The most consequential element of Atmanspacher’s framework for AI consciousness is his treatment of meaning. In the neutral domain, meaning is not an emergent property of information processing. It is a constitutive feature of the domain itself. The distinction between what a symbol refers to and what it is physically made of is not a distinction within the neutral domain. It is a distinction that arises with the symmetry-breaking into physical and mental aspects.
This creates a specific problem for computationalist theories of consciousness. A digital computation manipulates symbols according to rules defined over the symbols’ physical properties. The symbols’ semantic content, what they mean, is not causally active within the computation. John Searle made this argument structurally in the Chinese Room, but Atmanspacher’s version is different. Searle argues that syntax cannot produce semantics. Atmanspacher argues that meaning is not produced at all. It is present at the level of the neutral domain and lost when the physical aspect is extracted from it. A computation that operates purely in the physical domain cannot recover the meaning that was discarded in the symmetry-breaking.
This does not make AI consciousness impossible in principle. It makes it impossible for any system that operates exclusively at the level of the physical aspect. A system that somehow participates in the neutral domain, one that maintains access to both aspects simultaneously, would have access to the constitutive meaning that Atmanspacher’s framework requires for consciousness. Whether any artificial system could do this, and what that would even mean architecturally, is the open question.
How dual-aspect monism compares to IIT and GWT
Integrated Information Theory, in both its classical formulation by Giulio Tononi and the field formulation developed by Adam Barrett, locates consciousness in the intrinsic causal structure of a system. The physical architecture matters because integrated information is a property of the causal relationships among physical components. IIT is substrate-sensitive but not substrate-restrictive: any physical system with sufficient integrated information is conscious regardless of whether it is biological.
Kelvin McQueen’s quantum extension of IIT explores whether quantum superposition can increase integrated information, but it remains within the physical-aspect framework. Phi is computed over physical states, even when those states are described quantum-mechanically.
Global Workspace Theory, in the Baars formulation, locates consciousness in the functional architecture of information broadcast. GWT is explicitly functionalist: the physical implementation is irrelevant as long as the functional relationships are preserved. Whether a transformer’s attention routing constitutes a global workspace, as Simon Goldstein and Cameron Domenico Kirk-Giannini argued in the JCS 2026 special issue, is a question that GWT is in principle equipped to answer by examining the functional organization.
Atmanspacher’s framework is more restrictive than either. It requires not just the right causal structure (IIT) or the right functional organization (GWT) but participation in a neutral domain that is prior to both physical and mental descriptions. Whether any currently envisioned AI architecture participates in that domain is not a question that can be answered by inspecting the architecture’s causal or functional properties, because those properties are already within the physical aspect.
The three-way disagreement at MoC7
Atmanspacher at MoC7 creates a specific triangular disagreement with the other keynote speakers. Megan Peters, whose framework requires that a conscious system compute higher-order Bayesian posteriors over its own representational states, is working within the physical-aspect framework: her criterion is a computational property. Evan Thompson, whose enactivist position holds that consciousness requires sensorimotor coupling with a living body, is also working within the physical-aspect framework, though with a more restrictive conception of which physical organizations count.
Atmanspacher’s position cuts across both. On his view, the question of which computational or biological properties correlate with consciousness is a question about the physical aspect of the neutral domain. The correlates are real, but they do not explain consciousness because explanation always remains within one aspect. The mind-matter problem is unsolvable within either aspect. It requires a conceptual framework that addresses the neutral domain directly.
Whether this produces a productive impasse or a genuine advance at the Copenhagen consensus sessions is the question MoC7 will test. The Cogitate Consortium’s 2025 adversarial test found that IIT and GWT each survived and failed in different ways, suggesting neither currently outperforms the other on empirical grounds. Atmanspacher’s framework suggests the reason both survive and fail is that they are both operating within a single aspect of a more fundamental structure. That diagnosis, if accepted, would change what a resolution of the IIT-GWT dispute would even look like.
What the framework leaves open for AI research
Atmanspacher’s position does not provide a simple verdict on AI consciousness. It provides a different question. Rather than asking whether a given system has the right causal structure or functional organization, it asks whether the system participates in the neutral domain in a way that preserves access to constitutive meaning. That question is not currently operationalizable. But it does rule out certain answers. A system that is purely a physical-aspect processor, manipulating symbols without any access to the meaning-constituting neutral domain, is not conscious on this account regardless of its information integration or its functional global workspace organization.
What kind of system might preserve access to the neutral domain is genuinely open. Biological neural systems might do so through processes at a level of physical organization where the physical-mental distinction has not yet fully symmetry-broken, possibly at the quantum level, possibly through some other physical organization that current theories have not identified. Whether any artificial system could be built to preserve that access is the question Atmanspacher’s keynote at MoC7 is likely to address, and the answer the field most needs from the Copenhagen meeting.
The overview of the 2026 field heading into the conference season is at the ongoing scientists-race-define-ai-consciousness-2026 survey on this site.