Conscious artificial intelligence and biological naturalism

"Predictive processing, as I understand it, labels a range of theories about perception, cognition and action ... These have in common the idea that perception is a form of probabilistic inference, approximated in neural systems via a process of prediction error minimisation."

4.1 The story from predictive processing

This passage explicitly characterizes perceptual contents as structured probabilistic representations updated by prediction-error minimization, linking representational geometry in brains to analogous representational structures in AI models that implement inference-like updates .

"According to IIT, a conscious experience is identical to the cause-effect structure unfolded from a physical substrate specifying a maximum of irreducible integrated information. Wherever there are such maxima, there will be consciousness. ... A purely feedforward neural network would have zero consciousness, whereas a neural network with the right kind of physically-instantiated recurrent connectivity will have non-zero consciousness, regardless of what it is made out of."

5.6 Other theories

This ties consciousness to integrated information and recurrent causal structure, aligning with biological integration markers (long-range coherence, hub engagement) and AI integration proxies (attention convergence, residual integration) .

Tables

Table 1 : Table 1. Scenarios for conscious AI. If the remit of ‘computation’ is broadened beyond Turing’s definition, the distinction between scenarios 3 and 4 may blur... Biological naturalism falls within scenarios 4 and 5 (and potentially 3, for ‘computational biological naturalism’).

"Recently, empirical measures have been developed which formalise notions of ‘informational closure’ and ‘causal closure’ which underwrite the separation of scales in complex systems... A macroscopic (higher-level) variable – or ‘coarse graining’ – is causally or informationally closed when interventions on it (causal closure), or observations of it (informational closure), are sufficient to determine or predict the higher-level outcome(s)..."

3.5 Emergence and the separation of scales

By emphasizing closure at higher scales, the text motivates emergent global-state dynamics relevant to consciousness signatures like criticality and complexity indices in brains and emergent strategies/in-context learning in AI .

"These could include continuous processes, field effects, stochastic effects, fine-grained timing relations, and other (non-computational) functional properties of neurobiological systems."

5.4 Substrate-dependent, weak

The explicit mention of 'fine-grained timing relations' links temporal coordination in biological systems (e.g., oscillations, phase-locking) to potential timing/routing dynamics in AI (e.g., attention synchronization) as relevant to conscious processing .

"Relevant theories here include global workspace theory, which proposes that consciousness depends on information sharing within a multimodal global neuronal workspace (Mashour et al., 2020) ... and attention schema theory, which proposes that consciousness emerges as a consequence of a model of the control of attention (Graziano, 2017)."

5.2 Theory-based computational

Global workspace and attention schema frameworks point to selective routing and gating of information as central to conscious access, paralleling AI routing mechanisms such as attention weights, masks, and mixture-of-experts selection .

Tables

Table 1 : Table 1. Scenarios for conscious AI. If the remit of ‘computation’ is broadened beyond Turing’s definition, the distinction between scenarios 3 and 4 may blur... Biological naturalism falls within scenarios 4 and 5 (and potentially 3, for ‘computational biological naturalism’).

"With consciousness comes the potential for negatively valenced conscious experiences: for suffering. Creating real artificial consciousness risks a mass inauguration of new forms of suffering."

6.1 Real artificial consciousness

The author explicitly connects consciousness to affective valence and welfare, emphasizing that creating conscious AI could instantiate persistent negative states akin to pain or aversion in biological systems, a core ethical concern for AI consciousness research .

"Various neuroscience-based theories of consciousness emphasise (or can be framed in terms of) dynamical rather than computational elements. These include classic works linking conscious perception to attractor dynamics in neural systems ... The ‘dynamic core’ hypothesis ... highlighted integration and differentiation as key neurodynamical properties underlying consciousness – and which later resurfaced as the integrated information theory of consciousness..."

3.6 Non-computational functionalism and dynamical approaches

By bringing together dynamic core, synchrony, and IIT, the text emphasizes information integration and long-range coordination as hallmarks of conscious processing, echoing global workspace-like access in biology and integration mechanisms in AI .