Anil Seth

I was delighted, challenged, and somewhat overwhelmed by the breadth and depth of the commentaries elicited by the target article. I am very grateful to every one of my commentators for offering such thoughtful perspectives. Here, I attempt to isolate some common themes and offer some thoughts in response. One conclusion is immediately evident: it is increasingly urgent to think clearly about the prospects for conscious AI and about the consequences of seemingly conscious AI. The commentaries also flesh out the potential for an exciting research programme investigating the substrate dependence of both consciousness and cognition.

How can we investigate non-human experiences scientifically? Given substantial differences in sensory abilities and the private nature of consciousness, this remains an open question. In this review, we propose a way to gain empirical traction on one key feature of experience: its temporal structure, or 'timescape'. Perceptual contents follow systematic temporal principles-synchronisation, revision, and persistence-and are sampled across attentional windows and vary in stability. These principles can be explored through temporal illusions and experimental paradigms. We conceptualise an animal's timescape in terms of five key windows, all of which are experimentally testable, and analyse evidence across animal species to highlight varying timescapes. Together, these ideas lay the foundations for a research programme comparing the temporal Umwelts of non-human animals.

How can we investigate non-human experiences scientifically? Given substantial differences in sensory abilities and the private nature of consciousness, this remains an open question. In this review, we propose a way to gain empirical traction on one key feature of experience: its temporal structure, or 'timescape'. Perceptual contents follow systematic temporal principles-synchronisation, revision, and persistence-and are sampled across attentional windows and vary in stability. These principles can be explored through temporal illusions and experimental paradigms. We conceptualise an animal's timescape in terms of five key windows, all of which are experimentally testable, and analyse evidence across animal species to highlight varying timescapes. Together, these ideas lay the foundations for a research programme comparing the temporal Umwelts of non-human animals.

The integrated information theory of consciousness (IIT) is divisive: while some believe it provides an unprecedentedly powerful approach to address the ’hard problem’, others dismiss it on grounds that it is untestable. We argue that the appeal and applicability of IIT can be greatly widened if we distinguish two flavours of the theory: strong IIT, which identifies consciousness with specific properties associated with maxima of integrated information; and weak IIT, which tests pragmatic hypotheses relating aspects of consciousness to broader measures of information dynamics. We review challenges for strong IIT, explain how existing empirical findings are well explained by weak IIT without needing to commit to the entirety of strong IIT, and discuss the outlook for both flavours of IIT.

This paper presents a version of neurophenomenology based on generative modelling techniques developed in computational neuroscience and biology. Our approach can be described as _computational phenomenology_ because it applies methods originally developed in computational modelling to provide a formal model of the descriptions of lived experience in the phenomenological tradition of philosophy (e.g., the work of Edmund Husserl, Maurice Merleau-Ponty, etc.). The first section presents a brief review of the overall project to naturalize phenomenology. The second section presents and evaluates philosophical objections to that project and situates our version of computational phenomenology with respect to these projects. The third section reviews the generative modelling framework. The final section presents our approach in detail. We conclude by discussing how our approach differs from previous attempts to use generative modelling to help understand consciousness. In summary, we describe a version of computational phenomenology which uses generative modelling to construct a computational model of the inferential or interpretive processes that best explain this or that kind of lived experience.

Recent years have seen a blossoming of theories about the biological and physical basis of consciousness. Good theories guide empirical research, allowing us to interpret data, develop new experimental techniques and expand our capacity to manipulate the phenomenon of interest. Indeed, it is only when couched in terms of a theory that empirical discoveries can ultimately deliver a satisfying understanding of a phenomenon. However, in the case of consciousness, it is unclear how current theories relate to each other, or whether they can be empirically distinguished. To clarify this complicated landscape, we review four prominent theoretical approaches to consciousness: higher-order theories, global workspace theories, re-entry and predictive processing theories and integrated information theory. We describe the key characteristics of each approach by identifying which aspects of consciousness they propose to explain, what their neurobiological commitments are and what empirical data are adduced in their support. We consider how some prominent empirical debates might distinguish among these theories, and we outline three ways in which theories need to be developed to deliver a mature regimen of theory-testing in the neuroscience of consciousness. There are good reasons to think that the iterative development, testing and comparison of theories of consciousness will lead to a deeper understanding of this most profound of mysteries.

Throughout his career Andy Clark has shaped how scientists and philosophers think about the role of representation in action, perception, and cognition. In the latest iteration of this debate he has foregrounded the influential perspective of predictive processing, which sees perception as a process of action-oriented “best guessing” (inference) about the causes of noisy and ambiguous sensory signals and which involves the brain-inducing “generative” models of how hidden causes mediate the effects of actions on sensory signals. This chapter develops this position in the context of interoception (the sense of the body from within) and physiological regulation. A key idea here, which recalls twentieth-century cybernetic theory, is that interoceptive inference is targeted towards maintaining physiological homeostasis rather than inducing complete and accurate internal models of an external state of affairs. The chapter explores how this perspective helps connect control-oriented interoceptive inference to phenomenological properties of embodied selfhood and subjectivity. The upshot echoes (or perhaps subverts) a classic philosophical trope of the Enlightenment philosopher Julien de La Mettrie: to find the origins of our conscious selves in our nature as beast machines.

Brain–computer interfaces (BCIs) enable direct communication between the brain and external computers, allowing processing of brain activity and the ability to control external devices. While often used for medical purposes, BCIs may also hold great promise for nonmedical purposes to unlock human neurocognitive potential. In this Essay, we discuss the prospects and challenges of using BCIs for cognitive enhancement, focusing specifically on invasive enhancement BCIs (eBCIs). We discuss the ethical, legal, and scientific implications of eBCIs, including issues related to privacy, autonomy, inequality, and the broader societal impact of cognitive enhancement technologies. We conclude that the development of eBCIs raises challenges far beyond practical pros and cons, prompting fundamental questions regarding the nature of conscious selfhood and about who—and what—we are, and ought, to be.

As artificial intelligence (AI) continues to advance, it is natural to ask whether AI systems can be not only intelligent, but also conscious. I consider why people might think AI could develop consciousness, identifying some biases that lead us astray. I ask what it would take for conscious AI to be a realistic prospect, challenging the assumption that computation provides a sufficient basis for consciousness. I'll instead make the case that consciousness depends on our nature as living organisms – a form of biological naturalism. I lay out a range of scenarios for conscious AI, concluding that real artificial consciousness is unlikely along current trajectories, but becomes more plausible as AI becomes more brain-like and/or life-like. I finish by exploring ethical considerations arising from AI that either is, or convincingly appears to be, conscious. If we sell our minds too cheaply to our machine creations, we not only overestimate them – we underestimate ourselves.

Grapheme-colour synaesthesia is a perceptual experience where graphemes, letters or words evoke a specific colour, which are experienced either as spatially coincident with the grapheme inducer (projector sub-type) or elsewhere, perhaps without a definite spatial location (associator sub-type). Here, we address the question of whether synaesthesia can be rapidly produced using a hypnotic colour suggestion to examine the possibility of ‘hypnotic synaesthesia’, i.e. subjectively experienced colour hallucinations similar to those experienced by projector synaesthetes. We assess the efficacy of this intervention using an “embedded figures” test, in which participants are required to detect a shape (e.g., a square) composed of local graphemic elements. For grapheme-colour synaesthetes, better performance on the task has been linked to a higher proportion of graphemes perceived as coloured. We found no performance benefits on this test when using a hypnotic suggestion, as compared to a no-suggestion control condition. The same result was found when participants were separated according to the degree to which they were susceptible to the suggestion (number of coloured trials perceived). However, we found a relationship between accuracy and subjective reports of colour in those participants who reported a large proportion of coloured trials: trials in which the embedded figure was accurately recognised (relative to trials in which it was not) were associated with reports of more intense colours occupying a greater spatial extent. Collectively, this implies that hypnotic colour was only perceived after shape detection rather than aiding in shape detection via colour-based perceptual grouping. The results suggest that hypnotically induced colours are not directly comparable to synaesthetic ones.