Conference Agenda

Recent advances in consciousness science have exposed a critical impasse: competing theoretical frameworks—structuralist vs. functionalist (Fleming & Shea 2024, 2024b; Song 2024, Ellia & Tsuchiya 2024), universal vs. local (Albantakis et al. 2023; Kanai & Fujisawa 2024; Fleming 2024); intrinsic vs. extrinsic (Cohen and Dennett 2011; Doerig et al. 2019; Tsuchiya et al 2020; Negro 2020; Ellia et al. 2021)—appear to be inducing philosophical deadlocks and conceptual standstills. While these debates have generated valuable insights, they have proceeded in parallel, without a systematic framework for understanding their relationships and implications. In this theoretical contribution, we point out that these parallel disputes reflect deeper, unresolved tensions in conceptualizing consciousness. We argue that these debates can only be resolved by recognizing three fundamental dimensions that encompass all models of consciousness at a meta-theoretical level: (1) intrinsic (first-person) vs. extrinsic (third-person) perspectives, (2) universal (substrate-independent) vs. local (human-specific) scope, and (3) structure (formal organization) vs. function (cognitive roles) as explanatory priorities.

These dimensions interact in revealing ways. For example, intrinsic approaches often tend toward universal scope, as they define consciousness through internal properties that could exist in various systems. Understanding these interactions helps reveal potential blind spots in current frameworks. Our analysis shows clustering into extrinsic-local-functional and intrinsic-universal-structural approaches to consciousness, suggesting that theoretical conflicts reflect deeper methodological commitments rather than empirical disagreements, offering new pathways for investigation and transforming apparent deadlocks into tractable disagreements about dimensional priorities. Thus, its impact extends beyond theory, bridging the gap between foundational debates and translational applications.

Despite decades of research, the neural correlates of subjective emotional experience remain poorly understood. Multiple competing theories posit different neural substrates for feeling-states such as fear, ranging from localized regions such as the amygdala, periaqueductal gray, dorsolateral and ventromedial prefrontal cortex, or non-local distribution throughout the entire brain. Neuroimaging studies attempting to answer these questions are limited by their purely correlational nature, making them unable to discern between regions critically involved in subjective fear, and regions that show incidental co-activation. Adapting a previous experimental paradigm used with macaques (Ponce, 2019), we develop a method of causally probing patterns of fMRI activation. In a real-time fMRI paradigm, we target candidate fear patterns by evolving synthetic images that maximize their expression, and measure the subjective fear experienced as a result. Currently, we are in the process of collecting data from 30 healthy subjects with subclinical animal fears. In a preliminary scanning session, we collect fMRI responses to videos of frightening and non-frightening animals, which we use to create candidate fear patterns. In a second real-time fMRI session, we use this pattern as the target of the evolutionary image generation algorithm. By assessing the degree to which the resulting images are judged as frightening by participants, and by comparing the effectiveness of different fear patterns in increasing those ratings, we can isolate the set of regions which show evidence of a causal role in the generation of subjective fear.

Conscious access is suggested to involve ’ignition,’ an all-or-none activation across cortical areas. To elucidate this phenomenon, we carry out computer simulations of a detection task using a mesoscale connectome-based model for the multiregional macaque cortex. The model uncovers a dynamic bifurcation mechanism that gives rise to ignition in a network of associative regions. A hierarchical NMDA/AMPA receptor gradient plays a critical role: fast AMPA re-ceptors drive feedforward signal propagation, while slow NMDA receptors in feedback pathways shape and sustain the ignited network. Intriguingly, the model suggests higher NMDA-to-AMPA receptor ratios in sensory areas compared to association areas, a prediction supported by in vitro autoradiography data. Furthermore, the model accounts for diverse behavioral and physiological phenomena linked to consciousness. This work shed insights into how receptor gradients along the cortical hierarchy enable distributed cognitive functions, and provides a biologically-constrained computational framework for investigating the neurophysiological basis of conscious access.

The construction of a physically realised world model in the brain seems to not just be mediated by axonal action potential transmission, resulting in post-synaptic modulations, but also by electrical field effects of depolarizing neurons, i.e. ephaptic coupling. The significant role of ephaptic coupling in synchronizing neural ensembles has been conjectured to indirectly affect ontogenetic neural circuit development. Formalizing this conjecture, we expected synchronously firing ensembles to emerge at spatial distances where ephaptic and axonal signals were most temporally correlated. To test our model of ephaptic-axonal interactions during neural self-organization, we compared its predictions to developmental outcomes of cortical rat tissue on high-density multielectrode arrays in vitro. We observed a cosinusoidal variation of synchronous activity over radial distances that can be understood under our model to result from cumulative effects of ephaptic and axonal signals during gamma-band bursts. The recurrent amplification of this interaction effect during ontogenetic differentiation appears to bind the spectral profile of neural activity to radial biases in the spatial distribution of neural ensembles. While long-term plasticity has been conventionally attributed to synaptic action alone, ephaptic wave propagation represents a complementary mechanism. As current theories of consciousness almost exclusively focus on synaptic signal propagation for their explanations, our evidence for non-synaptic, wave-like propagation of signals exposes a crucial blind spot of current theories. We discuss how our results point towards a wider push needed towards exploring the role of analogue computation in the brain for understanding the neurobiological basis of consciousness.

The neural correlates of consciousness (NCC) remain central to neuroscience, with thalamocortical loops recognized for their role in sensory processing, cognition, and conscious experience. The thalamus integrates sensory and cortical inputs through dynamic, reciprocal interactions with the cortex and has been proposed as a potential constituent of consciousness. The basal ganglia (BG) are deeply embedded within these circuits but traditionally viewed as modulators of motor control and decision-making. Drawing on recent neurocomputational modeling results and an extensive literature survey, we argue that the BG are increasingly implicated in cognitive and affective processes related to consciousness.

Our contribution highlights studies demonstrating that BG activity reflects conscious perception and affective states even in the absence of explicit reports, suggesting a direct link to conscious content. This decodability, coupled with the BG’s regulatory role regarding attention, perception and predictive computations, points toward a more central role in shaping conscious experience. Furthermore, the synchronous dynamics between the BG and thalamocortical loops—a hallmark of constitutive neural processes—support the notion that the BG influence both the temporal coherence and informational structure of consciousness.

Disentangling causal from constitutive roles remains a methodological challenge. We propose a revised NCC framework that highlights the contribution of the BG and discuss its advantages in the light of existing theories. This reconceptualization offers potential insights into neurological and psychiatric conditions marked by BG dysfunction, such as Parkinson’s disease and obsessive-compulsive disorder. We aim to encourage subcortical-inclusive research perspectives and support underrepresented empirical directions in consciousness science.

The anterior insula (AIns) has long been implicated in conscious awareness, particularly through its role in interoception and the integration of sensory signals into subjective experience. More recently, model-based fMRI studies grounded in predictive coding frameworks have identified the AIns as a key player in inferential processes, suggesting its function extends beyond interoceptive awareness to a broader role in inference. At the same time, predictive coding has been proposed as a computational model of consciousness, yet a mechanistic account linking these frameworks remains incomplete.

Here, I propose that predictive coding provides a missing link between the anterior insula and its putative role in consciousness. As predictive coding is hierarchical—beginning with unconscious, low-level inferences that propagate errors up the cortical hierarchy—I suggest that unresolved inferential errors requiring conscious resolution are explicitly registered in the AIns. This framing positions the AIns as a critical (but not exclusive) neural gateway for the emergence of consciousness, where errors that resist lower-level resolution are tagged for awareness. While regions such as the anterior cingulate cortex (ACC) and prefrontal cortex (PFC) are also implicated in inference, the AIns may uniquely serve as a functional bottleneck between unconscious inference and awareness.

Empirical support for this hypothesis comes from studies linking AIns activity to explicit confidence, conscious Bayesian priors, and decision-making under uncertainty. By integrating predictive processing into theories of insular function, this work offers a computational perspective on why the anterior insula remains central in consciousness research.