Conference Agenda

For predictive processing, perception is tied to the upshot of inference—the posterior, which makes perception internal and indirect. Tobias Schlicht has argued that this is equivalent to a commitment to Kantian transcendental idealism and that this commitment in turn undermines predictive processing’s claims to completeness, realism and naturalism. But, I shall argue, this argument requires an immodest epistemological stance, it over-emphasises the hallucination part of the notion that perception is controlled hallucination and ignores the control part, it relies on unsatisfactory explanatory strategies, and sets aside fairly modest versions of completeness, realism and naturalism. The debate is useful because addressing Schlicht’s argument moves beyond predictive coding, bringing to the fore the self-evidencing aspects of predictive processing.

The integration of continuous self-reports of subjective experience with physiological data is essential for advancing both cognitive and affective science. Here, we present CAMPEONES, a novel public database designed to capture the temporal evolution of emotional experiences in immersive virtual reality (VR) environments.

In this study, participants engaged in controlled VR tasks specifically designed to elicit a range of emotional responses. Emotional states were continuously recorded using a joystick that tracked dynamic rating trajectories, while central and peripheral physiological signals were simultaneously measured. Synchronization protocols ensured high-quality, multimodal data collection, providing a robust framework for subsequent predictive modeling.

Preliminary analyses reveal associations between continuous emotional annotations and their corresponding physiological markers. The VR paradigm successfully elicited realistic emotional responses that correlated with the temporal dynamics of the continuous annotations.

CAMPEONES makes a significant contribution to affective science by integrating subjective emotional experience with objective physiological correlates. This framework not only deepens our understanding of emotional dynamics in naturalistic settings, but also lays the foundation for AI-driven models capable of predicting the continuous experience of affective states.

The rise in performance and availability of large language models (LLMs) has prompted excitement about their potential to support research in psychology (Demszky et al., 2023). However, research on their application and performance as interviewers in cognitive science is limited. In this study, we explore LLMs’ potential to conduct semi-structured interviews as part of experiments in cognitive neuroscience. To approach this objective, we designed an LLM-based conversational agent and conducted a within-subject experiment whereby 28 participants engaged with our agent after taking part in a robotic sensorimotor task designed to induce presence hallucinations (PHs). PHs are the "sensation that somebody is nearby when no one is actually present" (Blanke et al., 2014) and are a frequent symptom of neurodegenerative diseases such as Parkinson’s disease but can also be experienced by healthy individuals (Critchley, 1955). Consistent with previous results, ratings collected by our agent showed that the asynchronous condition of the task led to stronger PH sensations (p=0.02), more passivity experiences (p<0.001), and a lower sense of self-touch (p=0.04). Furthermore, sentiment analysis revealed that the polarity of responses in the asynchronous condition was significantly lower, suggesting the use of more negative vocabulary (p=0.01). Finally, we obtained nuanced insights into the phenomenological experience of robot-induced PHs, with participants describing various temporal and emotional qualities of the felt presence. These results shed light on how LLMs can be harnessed to conduct semi-structured interviews in experimental cognitive science and consciousness studies, thus providing a scalable, accessible, and consistent method to acquire rich qualitative data.

The hypothesis that the brain operates near a critical point has been a longstanding topic of debate in neuroscience. Systems poised at criticality exhibit enhanced information processing, long-range spatiotemporal correlations, and scale-free dynamics—properties that may be essential for the emergence of human consciousness. In this study, we quantify these signatures of criticality on fMRI datasets from subjects under different psychedelic drugs, patients with disorders of consciousness, and a healthy control population. We combine information-theoretic metrics quantifying synergy, redundancy and integrated information with statistical physics measures used to study critical phenomena such as divergence of the specific heat, susceptibility, correlation length and avalanche statistics. These complementary metrics of brain dynamics enable us to construct a phase space of human consciousness. Our findings reveal that signatures of critical behaviour display distinct and non-trivial variations with respect to the control group depending on the condition (psychedelic drug or disorder). Given the brain’s intrinsic functional heterogeneity, we discuss how different markers of criticality respond to dynamics in different brain regions and how they might diverge in their sensitivity to altered states of consciousness. This work provides insights into the mechanistic role of criticality in cognition and its disruptions in altered states, advancing our understanding of consciousness through the lens of complex systems and critical phenomena.

Block's distinction between access and phenomenal consciousness concerns the difference between the availability of information for executive functions and the way an experience feels to the subject. Over the past three decades, this distinction has exerted a profound influence on the science of consciousness even though it has originally been introduced as resting on a conceptual rather than an empirical difference (Block 1995). In this paper, I explore the empirical status and significance of Block’s distinction by analysing it using Imre Lakatos’ methodology of scientific research programmes (Lakatos 1979; Negro 2024).

I begin by observing that the access-phenomenal distinction came to be empirically relevant only with the introduction of two further auxiliary assumptions—the dual NCC hypothesis (Block 2005) and the phenomenal overflow hypothesis (Block 2007, 2011)—both of which postulate that the core conceptual distinction maps onto real differences in neural information processing. I then argue that the empirical significance of Block’s distinction rests wholly on these two hypotheses, as they offer the only way to operationalize the conceptual difference that is being postulated. Finally, I proceed to review recent evidence for and against Block’s auxiliary hypotheses, pointing out that both of these hypotheses rest on post hoc reasoning, might not offer the best way of accounting for the available data, and, in some cases, lead to contradictions. The paper concludes with an answer to the question whether Block’s distinction bears the marks of a degenerative research programme and, if so, what are those marks.

Contemporary Western accounts of consciousness often treat the mind and environment as separate processes. In contrast, evidence from non-Western traditions—such as Kabbalah—illustrates more interactive models of mind–environment engagement (Varela et al. 1991; Sendor 1994). The extensive cross-cultural research on consciousness, coupled with the lack of definitive scientific outcomes, underscores the need for a culturally pluralistic approach to its study.

Drawing on principles from biological systems to model continuous feedback between an organism and its environment (Hutchins, 2010), we employ both closed-loop and open-loop paradigms to highlight the differences between Western and non-Western approaches. In other words, continuous feedback mechanisms—akin to those observed in biological and physical systems (Bettinger & Friston, 2023)—can be represented through a coupling scheme that characterizes the dynamics between mind and environment. Notably, the convergence of insights from the natural sciences and traditional consciousness studies reveals a critical epistemological intersection.

Inspired by this transdisciplinary intersection, we propose a theoretical framework termed the Ecological Approach to Consciousness (EAC). This framework addresses the limitations of contemporary paradigms by integrating non-dualistic, multicultural, physical, biological, and ecological perspectives. The EAC posits that the mind and environment engage in mutually constitutive relationships, forming an interdependent whole. By incorporating diverse cultural perspectives, alternative models of mind–environment relationships emerge, and modeling these dynamics within a coupling scheme provides a robust framework for capturing the closed-loop, reciprocal processes that characterize natural phenomena.