Search

The target article draws a sharp contrast between intelligence and consciousness. We believe this contrast is premature and rests on an overly narrow characterisation of intelligence as abilities that can be exhaustively characterised as mappings from sensory inputs to behavioural outputs. Instead, we argue that the fortunes of intelligence and consciousness are interwoven more tightly than Seth’s framing implies.

It is a pleasure to comment upon, which covers everything that could – and should – be said about the matter. It left me wondering whether artificial consciousness is an oxymoron. I take this opportunity to unpack the thermodynamic issues that attend mortal computation and their implications for conscious artefacts.

I have recently suggested not to ask what science can do for consciousness but what consciousness can do for science. Here, I extend such a recommendation in the context of naturalism, the meta-ontology shared by most contrasting approaches to the problem of “conscious artificial intelligence”: do not ask what naturalism can do for AI but what AI can do for naturalism.

Seth argues in favour of strongly substrate-dependent biological naturalism. I suggest we need more reason to believe the substrates in question are consciousness-specific. Instead, the case for mortal computational biological naturalism is more promising. A self-evidencing framing helps elucidate what the consciousness-specific functional characteristics of mortal computation might be, leaving conscious AI waiting in the wings.

Seth’s claim that only biological or biological-like systems can be conscious leaves open the question of what renders only some organisms conscious. We address this question by pointing to the material, ontogenetic, memory, and evaluative affordances of living organisms and arguing that evolved, minimal consciousness requires complex cognitive-affective architecture for which these biological affordances are obligatory. Conscious thinking of humans cannot be separated from minimal consciousness, and non-biological AI robots’ computations do not render these operations conscious.

Biological naturalism, as explicated by Seth, is indispensable for a balanced and metaphysically neutral science of AI consciousness. However, most of the properties that Seth explores might not be unique to living systems. Therefore, I argue, a biological naturalist research programme in consciousness science requires an explicit definition of biological naturalism. I discuss how such a definition might be obtained.

Seth provides a nuanced but compelling case for consciousness being substrate dependent: a product of living processes rather than computation. Here, I offer a biological underpinning for these arguments grounded in the wetware of metabolism in single-celled organisms. Fluctuations in membrane potential give an integrated real-time readout of metabolic state in relation to the environment – the simplest meaningful ‘feeling’.

We agree that consciousness, as we know it, is inextricably biological – but not every biological detail is essential. Identifying which features are necessary, across scales, is vital for assessing the limits of computationalism. This process will reveal a richer, more biologically informed computationalism – one that future machines could, in principle, instantiate, but only after much deeper biological understanding.

Computation is a description sandwiched by meaning that is supplied by an external observer. This observer-dependence limits computation to a mere simulation of mind, not a realisation. I argue that this makes computational functionalism logically untenable. Consciousness, as a real phenomenon, must instead be grounded in a system with intrinsic meaning – a living, teleodynamic organism.

Seth’s insightful discussion must be extended in light of the field of Diverse Intelligence. A broader understanding of embodiment in unconventional spaces, of hybrids between evolved and engineered materials, and of unexpected competencies in very minimal systems makes it currently impossible place convincing limits on the possible substrates of consciousness. Doing so risks underestimating not only “machines” but also ourselves.

Distinctions can be made between information-processing, different kinds of knowledge, and understanding (intelligence). AI may be incapable of the type of understanding/intelligence known to humans. On the origins of consciousness, panpsychism may be a more rationally satisfying explanation than emergence. Some brief reflections are included on the relation between matter and consciousness, and on hemisphere differences and AI.

Biological naturalism may collapse into eliminative materialism, and the ethics of synthetic phenomenology will remain.

To be conscious is to be an experiencing subject. This can be defined not in terms of computational functions or particular biological substrates, but rather in terms of relations: between subject and world, between parts of the subject, and through time. These kinds of relations – comprising a conscious mode of being – may well be implementable in artificial systems.

Unity is intrinsic to being both an experiencing and an embodied subject. Computational accounts of consciousness, being observer-relative, cannot account for this unity. The thermodynamics of life can, but, contra Seth’s suggestion, the free energy principle fails to capture the distinctive metabolic interdependencies by virtue of which a living system constitutes itself as a unified perspective on the world.

Seth presents several arguments for why multiple realization is not necessarily true. However, this is not a common opinion in consciousness research. He fails to provide arguments why we should believe in biological naturalism over multiple realization, which is the actual challenge to create progress.

This commentary picks up on two related themes in Seth’s article – which have roots in embodiment. The first is that perceptual inference has a metabolic cost, and it is important to consider the relationship between thermodynamic and informational energies. The second is that embodied systems actively select their data over time – which may be a core feature of biological consciousness.

I argue that creature consciousness depends on specific neurobiophysical properties that constitute the medium for phenomenally conscious states. Such properties are not medium flexible in the way required by computational functionalism. Therefore, computational functionalism fails, and one of the following holds: biological naturalism, a nonbiological yet noncomputational version of functionalism, or a view according to which consciousness requires specific (macro)physical qualities.

This commentary concerns the distinction between computational modelling and computational metaphysics (i.e., the Computational Theory of Mind) in the context of consciousness studies. I argue that the ability to simulate the dynamics of a phenomenon using computational models does not entail that that phenomenon is literally performing the computations that are prescribed by the model.

We agree with Seth that consciousness is grounded in processes related to life, namely prediction, homeostasis, and autopoiesis. However, we believe he is wrong to frame these processes as non-computational. In fact, prediction, homeostasis, and autopoiesis are inherently computational, as shown by von Neumann’s Universal Constructor theory of general computation. Given this, we believe that AI consciousness is likely.

Seth’s biological naturalism claims consciousness requires living systems. We propose experiments operationalizing this debate through testable predictions about allostasis-predictive processing, autopoiesis-recurrent processing, and causal closure-global workspace relationships. Using a probabilistic framework, we show how experimental results could reduce metaphysical disagreement about computational functionalism, motivating empirical research that informs both consciousness theory and governance of potentially conscious technologies.