Schedule of Events | Search Abstracts | Symposia | Invited Symposia | Poster Sessions | Data Blitz Sessions

Invited Symposia

Advancing knowledge through open science adversarial collaboration

Invited Symposium 1: Sunday, April 14, 2024, 10:00 am – 12:00 pm EDT, Ballroom East

Chair: Lucia Melloni^1,2; ¹Max Planck Institute for Empirical Aesthetics, ²NYU Grossman School of Medicine
Presenters: Lucia Melloni, Aya Khalaf, Ole Jensen, Melanie Boly, Fanis Panagioraropoulos

Consciousness is all there is for us. It is there where we exist, we feel pain, love and also reason. Yet, a unified explanation of the mechanism(s) that mediate our subjective experience is still lacking. Following the decline of behaviorism, which once sidelined consciousness in scientific discourse, the past 30 years have seen a resurgence of research in the field of consciousness. Several neuroscientific theories of consciousness have been proposed. However, theories have evolved in parallel, without cross-talk between, and oftentimes offer contradictory explanations. To break those theoretical siloes, the Cogitate consortium is experimenting with open science adversarial collaboration. Competing predictions from two theories of consciousness are tested: Global Neural Workspace (GNW) and Integrated Information Theory (IIT), across two experiments, designed and endorsed by the theories' proponents. Preregistered experiments were conducted on a large number of subjects (250) combining three techniques (fMRI, M-EEG and iEEG) using similar analytical approaches. This symposium aims to showcase a new model of collaborative science and a systematic approach for theory development. First, Lucia Melloni will outline the promises and challenges to advance science and theory development through adversarial collaborations, presenting the case study of the Cogitate consortium. Next, Aya Khalaf will showcase results of two predictions concerning decoding and maintenance of the content of consciousness in which GNWT and IIT are tested. Ole Jensen will present results from the third prediction, concerning network connectivity. We conclude with a debate (Boly, Panagioraropoulos, Jensen, Melloni) on the role of adversarial collaborations in advancing theoretical understanding.

Presentations

Advancing science and theory development through open science adversarial collaboration

Lucia Melloni^1,2, Cogitate Consortium; ¹Max Planck Institute for Empirical Aesthetics, ²NYU Grossman School of Medicine

This presentation aims to accomplish three objectives: First, it offers a concise overview of the current state of consciousness science, where theories have developed in parallel, often showing signs of confirmation bias. Second, it introduces a model of science based on adversarial collaboration, designed to bridge the divide between theoretical camps, facilitating robust theory testing and development. The discussion centers on two prominent theories of consciousness: the Global Neuronal Workspace (GNW) Theory and the Integrated Information Theory (IIT), both evaluated through adversarial collaboration. This approach allows for a direct comparison of the divergent perspectives on neural mechanisms underlying consciousness, employing uniform experimental and analytical methods. Third, the presentation will demonstrate the efforts of the COGITATE consortium in realizing this objective. By conducting two experiments that employ MEG, fMRI, and iEEG techniques across six independent laboratories with a large sample, including cross-lab replication, the consortium aims to rigorously test pre-registered predictions. Specifically, we will focus on Experiment 1, where stimuli of varying visibility and duration are presented in different task contexts to assess critical aspects of GNW and IIT. We will describe the path from divergent claims of GNW and IIT to the core of their testable predictions to reveal how each of the findings justifies or challenges the theories. Further, we will discuss how robust theory testing requires testing across multiple predictions which, together, provide convergent or divergent evidence for a given theoretical position.

Adversarial collaboration to evaluate Global Neuronal Workspace Theory and Integrated Information Theory: results and challenges part 1

Aya Khalaf¹, Cogitate Consortium; ¹Department of Neurology, Yale School of Medicine

In this talk, we will present results of two key predictions of GNW and IIT evaluated in the adversarial collaboration. Prediction #1 focused on where and when do cortical areas contain information about clearly consciously perceived stimuli. IIT predicts that such information should be present primarily in posterior cortex, while GNWT predicts the involvement of PFC. Prediction #2 focused on the temporal dynamics predicted by each theory for the maintenance of content in consciousness. GNW and IIT make specific, yet distinct, predictions about the spatial and temporal dimensions of such information in the brain: GNW predicts phasic responses in PFC supporting the maintenance of conscious content. More specifically, brief content specific ignition in PFC ~0.3-0.5s both after stimulus onset and offset indicating the update of the workspace. Conversely, IIT proposes sustained responses in the posterior cortex, aligned with the duration of consciousness, reflecting the theory's view that conscious experience equates to persistent cause-effect structures. We utilized multivariate decoding techniques on MEG, fMRI, and iEEG data to identify information about the category and orientation of visible stimuli, thereby decoding content of consciousness. Additionally, we analyzed the temporal activation profiles and conducted representational similarity analysis to examine the dynamics of sustained consciousness content. Our findings shed light on the spatio-temporal dynamics underpinning consciousness and contrast the theoretical predictions of GNW and IIT. By applying a Lakatosian approach, we emphasize how the challenges to these theories’ predictions provide deeper insights than mere confirmations, offering a nuanced understanding of the neural mechanisms of consciousness.

Adversarial collaboration to evaluate Global Neuronal Workspace Theory and Integrated Information Theory: results and challenges part 2

Ole Jensen¹, Cogitate Consortium; ¹Centre for Human Brain Health, University of Birmingham

In this presentation, we discuss the evaluation of the third prediction by Global Neuronal Workspace Theory (GNWT) and Integrated Information Theory (IIT) within this adversarial collaboration. Both theories offer distinct views on interareal communication in the brain to support conscious perception. GNWT suggests that consciousness involves the broadcasting of information from a fronto-parietal network to specialized modules for further processing, predicting transient and specific communication between the prefrontal cortex and high-level visual areas. In contrast, IIT posits sustained, content-specific communication between early sensory cortices and high-level visual areas, reflecting the theory's emphasis on experience-specific substructures.These are further bound by relations – overlaps between causes and effects - which in the brain should typically be accompanied by firing synchrony. Our research utilized MEG, iEEG, and fMRI to measure connectivity through phase-locked synchrony, dynamic functional connectivity, and psychophysiological interactions, aiming to test these theories' predictions regarding interareal communication. Findings from fMRI and iEEG showed significant convergence, highlighting the robustness of these methods in capturing functional connectivity, despite MEG's less consistent results, indicating the need for further exploration of connectivity metrics. We conclude by addressing the challenges faced by both theories across three tested predictions. While for a ‘science-as-usual’ approach these results might be deemed sufficient to arbitrate among theories, we argue that theory testing requires more. Specifically, our project seeks convergence across several predictions, and experiments, which as a whole tests the theories’ prediction in the context of a research program.

Open science adversarial collaboration to advance theory: A debate

Melanie Boly¹, Fanis Panagioraropoulos, Ole Jensen², Lucia Melloni^3,4; ¹Department of Neurology and Department of Psychiatry, University of Wisconsin, Madison, ²Centre for Human Brain Health, University of Birmingham, Birmingham, ³Max Planck Institute for Empirical Aesthetics, Frankfurt, ⁴NYU Grossman School of Medicine, New York

We will conclude the symposium with a debate addressing the challenges encountered while testing three predictions using multiple methodologies on a substantial number of subjects (over 250). The primary goal is to engage in an open discussion about what these challenges reveal concerning the theories' predictive capabilities. Subsequently, we will expand the conversation to consider the significance of adversarial collaborations in theory testing and as a means to reinforce theoretical foundations. We advocate that both consciousness science and cognitive science at large demand more robust theories, more rigorous testing, and systematic strategies to counteract confirmation bias, coupled with a substantial degree of intellectual humility. However, it's important to acknowledge that science operates within a system of incentives that often does not support thorough theory evaluation.