Conference Agenda

Science requires objective measurements to be able to falsify predictions, whereas consciousness is thought to be intrinsically personal and subjective. Consequently, behavioral measures of consciousness typically either fall in the “objective” or “subjective” category. However, the exact meaning of the terms objective and subjective is often implicit, and typically not clearly defined. Moreover, applying such a label to the measure is somewhat misleading, as the same measure can often be analyzed within a “subjective” or “objective” framework. Thus, this designation is not only relevant to the measure but also to the task, and to the approach that is used to analyze and interpret the results. In this talk we provide a brief overview of dichotomies along which one might conceptualize the difference between objective and subjective approaches, such as (1) whether the mode of stimulus presention is forced-choice or nonforced-choice, (2) the degree to which the measure is criterion-free or criterion dependent, (3) whether the response is a Type 1 or a Type 2 response and (4) whether the response/outcome measure is intended to capture the performance of the participant in relation to a ground truth, or whether it is intended to capture how a stimulus appeared to the participant. Although some of these dichotomies clearly overlap, the correspondence is not always 1:1. We call for researchers to more clearly outline how stimulus presentation, task, response measure, analysis approach and outcome score on these dichotomies, and argue why they believe their empirical approach best captures “consciousness” as a construct.

Changes in conscious content are characterized by complex connectivity patterns within and between brain areas. We have recently argued that interactions between oscillations at the same frequency cannot fully explain this coordination due to the heterogeneity in cortical oscillation frequency and amplitude and rapid cognitive state changes linked to aperiodic transients (1,2). Instead, we have proposed that nonlinear interactions, which enable pattern recognition and integration beyond simple information relaying, are essential for understanding large-scale integration of conscious content. Using EEG and ECoG, I will show how nonlinear metrics (e.g. WSMI, Transfer Entropy) outperform traditional linear metrics (e.g. WPLI, coherence) when distinguishing visual (n=40), auditory (n=33), and olfactory (n=16) contents across perceptual tasks, and when combined with phenomenological reports (n=29). Secondly, I will show how nonlinear metrics (e.g. co-Information) distinguish the type of information encoded by large-scale interactions. Using ECoG, and computational modeling, I will show that auditory (n=22) and visual (n=3) contents are integrated through distributed, synergistic interactions (i.e. complementary information) rather than by purely redundant interactions (i.e. common information). Together, these new empirical and theoretical observations will considerably impact the understanding of the functional role of nonlinear dynamics in encoding the contents of consciousness.

1. Vinck, M., Uran, C., Spyropoulos, G., Onorato, I., Broggini, A.C., Schneider, M., Canales-Johnson, A., 2023. Principles of large-scale neural interactions. Neuron 111, 987–1002.

2. Vinck, M., Uran, C., Dowdall, J.R., Rummell, B., Canales-Johnson, A., 2024. Large-scale interactions in predictive processing: oscillatory versus transient dynamics. Trends in Cognitive Sciences.

Recent work in cellular neuroscience has shown that a subset of the corticothalamic loop connecting layer 5 (L5) of cortex, and matrix-thalamus cells, plays a causal role in the threshold for conscious perception [1]. Similarly, in human neuroscience non-invasive focused ultrasound stimulation of matrix-rich thalamic nuclei increases perceptual sensitivity for near-threshold stimuli [2], suggestive of a shared thalamic mechanism controlling perceptual thresholds. There is, however, a theoretical gap between the cellular and mesoscale mechanisms of conscious perception.

To bridge this gap we leverage a combination of neurobiologically constrained neural mass modelling, connectomics, and signal detection theory. We use a transfer function fit to L5 cell activity, and constrain cortical-cortical and corticothalamic connectivity using 7T diffusion imaging and RNA expression data allowing us to quantify and study thalamic contributions to cortical dynamics.

Across the thalamic parameter space trails classified as “seen” by an ideal observer were associated with a late metastable ignition-like state across the cortex. Bridging the results of cellular and non-invasive stimulation studies, we found that shifting the model into a matrix-thalamus dominant regime reduced the perceptual threshold of the ideal observer by controlling the bifurcation structure of each neural mass on the connectome. Increased input from matrix-thalamus widened the region of cortical bistability thereby reducing the stimulus strength necessary for inter-areal signal transmission.

By building cellular-level mechanisms into a mesoscale neural mass framework our model provides a computational bridge between cellular and mesoscale mechanisms of conscious perception.

1-Takahashi et al. (2020): https://pubmed.ncbi.nlm.nih.gov/32747790/

2-Jang et al. (2024): https://pmc.ncbi.nlm.nih.gov/articles/PMC11483030/

Leading theories of consciousness make contrasting predictions about the neural correlates of consciousness. For instance, while Global Neuronal Workspace Theory emphasizes the crucial involvement of prefrontal cortex (PFC), Integrated Information Theory predicts a key role for recurrent processing within the posterior cortex. Recent work on the role of the PFC has focused on dissociating neural processes underlying perceptual awareness from those linked with post-perceptual processing.

To advance this, human participants performed a visual task where a grating was presented to the left or right of fixation, sometimes followed by a backward mask. Subjective awareness of stimulus location was assessed with the Perceptual Awareness Scale (PAS), and EEG recordings were obtained. Trials were divided into subjectively seen and unseen based on PAS scores, enabling comparison of trials that resulted in conscious or unconscious processing despite identical physical stimulation.

Multivariate pattern analysis was used to decode the location of the grating from EEG data. Using a cross-decoding scheme, a classifier was trained on data from an independent task in which participants did not report the (non-masked and clearly visible) gratings. It was then tested on the backward masking task, thereby isolating neural representations of stimulus location from those associated with reporting. Decoding across all EEG channels revealed a 130-170 ms post-stimulus window associated with perceptual awareness, with significant contributions from occipital, temporal, and parietal regions, but not the prefrontal cortex. Our findings strengthen the claim that perceptual awareness is linked to early processing in posterior regions, while PFC involvement supports post-perceptual processing.

Electrophysiological activity encoding visual stimulus detection gradually emerges across the cortical hierarchy. Mechanistically, noisy sensory evidence is accumulated until a boundary is reached leading to detection. Here, we aim at modeling how neural correlates of consciousness could emerge from the sensory-to-decision transformation occurring during visual stimulus detection. Crucially, we consider how subjective experience unfolds over time to isolate its neural correlates from decisional activity. Our model has two populations of recurrently connected rate-based neurons. The sensory layer receives and integrates sensory inputs and projects to a decision layer with longer integration timescales. This decisional layer defines whether a near-threshold stimulus is detected and the associated response times. We tested how biological parameters such as top-down feedback, excitation-inhibition balance, and the presence of NMDA receptors affect simulated behavior and relate to mathematical models of leaky evidence accumulation. We found that simulated activity in the decisional layer closely resembles high-gamma activity recorded in the anterior insula of patients implanted with stereotaxic EEG electrodes while simulated activity in the sensory layer matches high-gamma activity in the inferior temporal cortex. Next, we devised testable hypotheses on how a neural correlate of consciousness could be found in subspaces of neuronal population activity during decision-making and test whether it could explain the temporal unfolding of subjective experience including the magnitude-duration illusion (longer perceived duration for higher intensity stimuli). We will discuss these findings considering the role of top-down feedback – which is central to many theories of consciousness – during decision-making and no-report paradigms.

Although spontaneous fluctuations in brain states and neuromodulation play key roles in some leading models of consciousness (e.g., Global Neuronal Workspace theory), in scientific practice these factors are often ignored. In a recent line of work, we instead put these internal factors center stage to further understand how they shape our conscious experience. As a starting point, this work is inspired by the Yerkes-Dodson law, stating that task performance is optimal at moderate levels of arousal. Although this law is standard textbook material for all Psychology and Neuroscience students, surprisingly little is known about its overall characteristics, its’ underlying neural implementation and it’s relation to conscious perception. I will present recent studies in which we pharmacologically enhanced the overall arousal state of human participants using cholinergic and noradrenergic pharmaceuticals, while we simultaneously measured pupil-linked arousal fluctuations and brain activity during various perceptual tasks. First, we show that perception is optimal at mid-levels of arousal and impoverished at too low and too high arousal states and that this holds across different tasks (discrimination, detection) and sensory modalities (visual, auditory). Second, catecholaminergic enhancement increased overall arousal and shifted the entire arousal-performance curve, illustrating that the arousal-perception relationship is adaptive, similarly to well-know normalization mechanisms in neuroscience. We can reproduce these findings in a neurobiologically plausible computational framework, showing how catecholaminergic modulation alters a disinhibitory neural circuit that encodes sensory evidence. Together, these findings underscore the flexibility and efficiency of neural circuits shaping the arousal-perception relationship, both within and across arousal states.