Let us make two assumptions:

1. 1. Deep continuity (DC). Anil Seth is fundamentally right about the substrate-dependence of consciousness relative to patterns of microfunctional properties that are a) “non-computational” in terms of classical functionalism and b) “biological” in terms of a general theory of life. The assumed theory of life isn’t yet fully spelled out, but it will likely include Karl Friston’s free energy principle, plus what the late Margaret Boden called the “metabolism constraint,” some variant of Maturana & Varela’s conception of autopoiesis, Hohwy-style self-evidencing (Hohwy, Reference Hohwy2026), and so on.

2. 2. Self-modelling epistemic spaces (SES). Living systems are embodied epistemic spaces. An epistemic space is a set of possible states or processes of knowing. In and of itself, this space does not possess a center or a periphery, and conscious experience appears whenever a living system additionally generates an integrated model of this very space within itself. If and only if a given system has a global, internal model of its own epistemic space as such does the system also know that it currently possesses the capacity to know. This is what awareness is. From Seth’s perspective, tonic alertness and the ascending reticular activation system will be among the central biological properties that create this capacity as such. In terms of predictive processing, the system can then use this model to continually predict and actively control not just the current level of mere epistemic capacity, but also the specifics of what it will know (e.g., via precision control). Physically, the level of arousal will be the decisive boundary condition, causally determining the depth of cortical information processing available to the animal, depending on the local level of activation of five types of neurotransmitter in the ascending reticular activating system. Conceptually, however, the essence of phenomenality – the extra quality of awareness – is simply “being the content of a currently self-modeling epistemic space.” Let us assume SES as a (very) rough sketch of our philosophical background theory of consciousness (Metzinger, Reference Metzinger2020, Reference Metzinger2024, Chs. 4 & 5).

If both theoretical intuitions are on the right track, then we can draw at least two conclusions from the conjunction of DC and SES:

• We still don’t know how deep the relevant continuity is. If phenomenal properties locally supervene on contemporaneous biological but non-computational properties, then it still remains unclear what the correct level of description and the correct level of conceptual granularity for the supervenience base level really is. In a given class of biological systems, what kind of predicates best describe exactly those properties that form the minimally sufficient base set for consciousness? For example, could implementing a good caricature of the essential statistical physics be enough to conjure up the target phenomenon, to create a self-modelling epistemic space? If we let go of the computational level, then maybe we can ultimately do without biological properties too. The emergence of SES systems may be an extremely interesting and cosmologically rare physical phenomenon that is most economically described and most parsimoniously explained using only conceptual tools from physics – and none from biology. Under computational functionalism, phenomenal properties would still be something abstract, as defined by their causal role in a wider network, but Seth’s proposal would turn them into very concrete, substrate-dependent features of real-world beast machines. Supervenience as such is a non-reductive form of materialism, but biological naturalism may “bottom out” into physics via local, domain-specific identities between concrete biological properties and the physical realizations of even finer-grained microfunctional properties. Seth writes: “Life would still matter, and it would do so thanks to the properties that warrant the term ‘life’” (§ 5.5). But is this really true? Biological naturalism may function as the methodological gateway to eliminative materialism (Ramsey, Reference Ramsey, Zalta and Nodelman2024), ultimately leading us to conclude that consciousness and phenomenal properties never really existed in the first place. Just as biological properties never did.

• For the ethics of synthetic phenomenology (§ 6.1), not only artificial intelligence (AI) but also advanced organoid intelligence (OI) will be relevant (de Jongh et al., Reference de Jongh, Massey and Bunnik2022; Smirnova et al., Reference Smirnova, Caffo, Gracias, Huang, Morales Pantoja, Tang, Zack, Berlinicke, Boyd, Harris, Johnson, Kagan, Kahn, Muotri, Paulhamus, Schwamborn, Plotkin, Szalay, Vogelstein, Worley and Hartung2023). If Anil Seth is right, then phenomenal properties will most likely be instantiated by some future form of OI or by novel combinations that causally connect AI with OI (for example, via OI systems trained by AI). Brain organoids could represent the ultimate form of neuromorphic computing (Friston, Reference Friston2023), and such systems could certainly instantiate self-modelling epistemic spaces. However, conscious SES systems of this type would not be “embodied” or “embedded” in the canonical sense and would exhibit only a weaker version of “generative entrenchment” (see § 3.3). Even more importantly, they would be neither artificial nor natural, because they are not exclusively human-made artifacts, nor are they connected to the evolutionary history of our planet via the standard processes of procreation and genetic transmission. The term “artificial” implicitly introduces “natural” as its semantic counterpart (as in Seth’s “biological naturalism”), but from a purely logical point of view, the distinction between artificial and natural systems is neither exclusive nor exhaustive. There is a third logical possibility: non-artificial, non-biological, but postbiotic phenomenology. This is why the term “synthetic phenomenology” is preferable to “artificial consciousness” because it leaves open this third possibility while directly connoting the already well-established discipline of synthetic biology, which applies engineering principles to develop entirely new kinds of biological devices and systems. Even moving beyond “biological naturalism,” the ethics of synthetic phenomenology will not go away: not all, but certain subsets of postbiotic SES-systems will possess the capacity to suffer and thereby become objects of moral consideration.