Figure 4 - available via license: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Content may be subject to copyright.
Prediction #3: Interareal communication a. iEEG electrode coverage used to assess content-selective synchrony for IIT ROIs (top, Nsubjects=4) & GNWT ROIs (bottom, Nsubjects=21). Electrode coverage varied between ROIs as interareal connectivity was assessed between electrodes on a persubject basis. In addition, two example category-selective electrodes are shown: one face-selective, and one object-selective. b. iEEG Pairwise phase consistency (PPC) analysis of task irrelevant trials reveals significant content-selective synchrony (e.g. faces > objects for face-selective electrodes; objects > faces for object-selective electrodes) in V1/V2 ROIs (top row), but not in PFC ROIs (bottom row). c. MEG cortical time series were extracted per participant from cortical parcels in V1/V2 (blue), PFC (green) and in a fusiform (red) ROIs. Category-selective signals were obtained by creating a category-selective GED filter (i.e., contrasting face/object trials against any other stimulus category trials) on the activity extracted from the fusiform ROI. Face-(bottom left) and objectselective (bottom right) responses averaged across participants are shown at the bottom. d. MEG PPC analysis of task irrelevant trials (N=65) reveals significant category-selective synchrony below 25 Hz for the face-selective GED filter (i.e., faces > objects for face-selective electrodes) in both V1/V2 (top row) and PFC ROIs (bottom row) and for the object-selective synchrony (objects > faces for object-selective electrodes) in PFC only.
Source publication
Different theories explain how subjective experience arises from brain activity. These theories have independently accrued evidence, yet, confirmation bias and dependence on design choices hamper progress in the field. Here, we present an open science adversarial collaboration which directly juxtaposes Integrated Information Theory (IIT) and Global...
Contexts in source publication
Context 1
... decoding of category and orientation places a higher bar for theory testing as it requires the satisfaction of two constrains, making it less probable to pass the test 30 . Here, we found divergent results for the predictions of IIT and GNWT: decoding of face orientation (left/right/front views) was found in posterior cortex but not in prefrontal cortex, both in the iEEG theory-relevant ROIs decoding approach (Figure 2f, h; accuracy improved to ~95% with pseudotrial aggregation as shown in Extended Data Figure 5a) and in the fMRI searchlight approach (Figure 2g, ~45%). From the MEG cortical time series, decoding of face orientation was robust in posterior cortex (~75% with pseudotrial aggregation), and reached above chance levels, albeit weakly (35%) in prefrontal ROIs (Figure 2i). ...
Context 2
... iEEG, we restricted analyses to electrodes showing face and object selectivity, using a different subset of electrodes to test connectivity with V1/V2 and PFC (see methods, Figure 4a for examples of face and object selective electrodes). Due to the sparse coverage, the requirement to focus on 'activated' electrodes (see methods) was relaxed. ...
Context 3
... to the sparse coverage, the requirement to focus on 'activated' electrodes (see methods) was relaxed. We found increased category selective, e.g., faces>objects synchrony between category-selective and V1/V2 electrodes ( Figure 4b, top row). However, these effects were early and short-lived (e.g., <0.75 s), observed only at low frequencies, i.e., 2-25Hz, and mostly explained by the synchronous activity elicited by the stimulus evoked response (Extended Data Figure 8). ...
Context 4
... the findings did not match IIT predictions, as the activity was not found in the gamma frequency predicted by IIT, and was not sustained. No content-selective PPC was found between face-and objectselective electrodes and PFC in the relevant time window, in contrast to GNWT's prediction ( Figure 4b, bottom row). ...
Context 5
... MEG, we used Generalized Eigenvalue Decomposition (GED) 36 to extract faceand object-selective components from ventral temporal areas ( Figure 4c) and then computed PPC. We found selective synchronization between face-selective areas and both V1/V2 and PFC. ...
Context 6
... found selective synchronization between face-selective areas and both V1/V2 and PFC. However, these effects were early and restricted to low frequencies (2-25 Hz), which was inconsistent with both IIT and GNWT ( Figure 4d) and mostly explained by stimulus evoked responses (Extended Data Figure 8). ...
Context 7
... iEEG, we observed significant connectivity between object selective electrodes and V1/V2 (Figure 4e). Connectivity was evident in several frequency bands, most predominantly the gamma band. ...
Context 8
... connectivity was observed between PFC and both the face and the object-selective areas, in the frequency (gamma) and time range predicted by GNWT. For MEG, brief DFC in the alpha-beta frequency bands was found only between face-selective nodes and both PFC and V1/V2 (Figure 4f). ...
Context 9
... computed generalized psychophysiological interaction (gPPI), defining Fusiform Face Area (FFA) and Lateral Occipital Complex (LOC) as seed regions per subject based on an anatomically constrained functional contrast (see methods) and combining task relevant and irrelevant trials. FFA showed content selective (face>object stimuli) connectivity with V1/V2, Inferior Frontal Gyrus (IFG) and Intraparietal Sulcus (IPS), consistent with the predictions of both IIT and GNWT (Figure 4g). No selective increases in interareal connectivity between object selective nodes and PFC or V1/V2 was found in fMRI, also when separating task relevant and irrelevant trials (Extended Data Figure 8). ...
Context 10
... determine whether connectivity to PFC and V1/V2 might be driven by the task in gPPI, we explored the iEEG data separating trials by the task. We found task independent, selective DFC connectivity (face>objects) for face selective electrodes with both IFG and V1/V2 (Figure 4h). ...
Context 11
... unprecedented data collection effort brings several new insights relevant to our theory. Most importantly, the results confirm that PFC exhibits a metastable bout of activity ("ignition") for about ~200 ms, in a content-specific manner, even for task irrelevant stimuli, irrespective of stimulus duration (Figures 2b, 3f, Supplementary Figure 23), and with a concomitant transient increase in long-distance dynamic functional connectivity with face-and object-selective posterior areas (Figure 4e-h). Those findings, unpredicted by IIT but predicted by GNWT, support previous findings that PFC contains a detailed code for conscious visual contents 28,52-56 . ...
Context 12
... Generalized psychophysiological interactions (gPPI) task-related connectivity analysis of task irrelevant (left) and task relevant (right) conditions revealed weak clusters of content-selective connectivity when FFA is used as the analysis seed (p < 0.01, uncorrected). Common significant regions showing task related connectivity in task irrelevant, task relevant, and combined conditions (Figure 4) include V1/V2, right intraparietal sulcus (IPS), and right inferior frontal gyrus (IFG). f. gPPI task-related connectivity analysis of task irrelevant (left), task relevant (middle), and combined conditions revealed weak clusters of content-selective connectivity when lateral occipital complex (LOC) is used as the analysis seed (p < 0.01, uncorrected). ...
Context 13
... the left, the original predictions made by the IIT (top) and GNWT (bottom), preregistered here (see also 12 ; Figure 1). The table describes the key hypotheses (second column, 'Key hypotheses') made by the theories (see also Figure 1a), and probed in three different test analyses (third column, 'Test'; decoding (prediction #1; Figure 2), activation & RSA (prediction #2; Figure 3) and synchrony (prediction #3; Figure 4). Next, we describe the possible outcomes of each of these analyses, and how they would inform the theoretical predictions (fourth column, 'Possible outcome and interpretation'). ...
Similar publications
Across the animal kingdom, neural responses in the auditory cortex are suppressed during vocalization, and humans are no exception. A common hypothesis is that suppression increases sensitivity to auditory feedback, enabling the detection of vocalization errors. This hypothesis has been previously confirmed in non-human primates, however a direct l...
Different theories explain how subjective experience arises from brain activity1,2. These theories have independently accrued evidence, but have not been directly compared³. Here we present an open science adversarial collaboration directly juxtaposing integrated information theory (IIT)4,5 and global neuronal workspace theory (GNWT)6, 7, 8, 9–10 v...
Goal-directed reaches give rise to dynamic neural activity across the brain as we move our eyes and arms, and process outcomes. High spatiotemporal resolution mapping of multiple cortical areas will improve our understanding of how these neural computations are spatially and temporally distributed across the brain. In this study, we used micro-elec...
Objective. Detection of the epileptogenic zone is critical, especially for patients with drug-resistant epilepsy. Accurately mapping cortical regions exhibiting high activity during spontaneous seizure events while detecting neural activity up to 500 Hz can assist clinicians’ surgical decisions and improve patient outcomes. Approach. We designed, f...
Across the animal kingdom, neural responses in the auditory cortex are suppressed during vocalization, and humans are no exception. A common hypothesis is that suppression increases sensitivity to auditory feedback, enabling the detection of vocalization errors. This hypothesis has been previously confirmed in non-human primates, however a direct l...
Citations
... (1) collaboration between proponents of different theories to develop paradigms that test their respective predictions (ARC; e.g., Consortium et al., 2023;Melloni et al., 2023); and (2) the establishment of a set of criteria to assess the scope and explanatory power of each theory regarding conscious phenomena, largely independent of empirical data (CRIT; Doerig et al., 2021). Building on these two approaches, the author introduced the "quantification to the best explanation" (QBE) method, based on Bayesian confirmation theory, to complement and address the shortcomings of the existing approaches. ...
... Having opted to equate performance across manipulations in our study, behavioral performance was above the chance level for both consciousness manipulations. Follow-up research investigating perceptual integration of fully unconscious stimuli could address ongoing debates between influential theories of consciousness (Ferrante et al., 2023;Mudrik et al., 2014). The global neuronal workspace theory suggests a durable, yet unconscious processing stage (referred to as 'preconscious'), where the input is not globally available, and amplification through top-down attention is required for conscious access and report (Dehaene et al., 2006). ...
This study investigates failures in conscious access resulting from either weak sensory input (perceptual impairments) or unattended input (attentional impairments). Participants viewed a Kanizsa stimulus with or without an illusory triangle within a rapid serial visual presentation of distractor stimuli. We designed a novel Kanizsa stimulus that contained additional ancillary features of different complexity (local contrast and collinearity) that were independently manipulated. Perceptual performance on the Kanizsa stimulus (presence vs. absence of an illusion) was equated between the perceptual (masking) and attentional (attentional blink) manipulation to circumvent common confounds related to conditional differences in task performance. We trained and tested classifiers on electroencephalogram (EEG) data to reflect the processing of specific stimulus features, with increasing levels of complexity. We show that late stages of processing (~200–250 ms), reflecting the integration of complex stimulus features (collinearity, illusory triangle), were impaired by masking but spared by the attentional blink. In contrast, decoding of local contrast (the spatial arrangement of stimulus features) was observed early in time (~80 ms) and was left largely unaffected by either manipulation. These results replicate previous work showing that feedforward processing is largely preserved under both perceptual and attentional impairments. Crucially, however, under matched levels of performance, only attentional impairments left the processing of more complex visual features relatively intact, likely related to spared lateral and local feedback processes during inattention. These findings reveal distinct neural mechanisms associated with perceptual and attentional impairments and thus contribute to a comprehensive understanding of distinct neural stages leading to conscious access.
... But it is worth keeping in mind that adversarial collaborations do not have a strong track record of delivering convergence elsewhere in science (Kahneman & Klein, 2009;Latham et al., 1988;Mellers et al., 2001). And, when we take a look at the state of theory testing in consciousness science today, divergence, rather than convergence, appears to be currently well under way (Yaron et al., 2022;Cleeremans 2022;Cogitate, 2023). ...
Philosophers of mind and philosophers of science have markedly different views on the relationship between explanation and understanding. Reflecting on these differences highlights two ways in which explaining consciousness might be uniquely difficult. First, scientific theories may fail to provide a psychologically satisfying sense of understanding—consciousness might still seem mysterious even after we develop a scientific theory of it. Second, our limited epistemic access to consciousness may make it difficult to adjudicate between competing theories. Of course, both challenges may apply. While the first has received extensive philosophical attention, in this paper I aim to draw greater attention to the second. In consciousness science, the two standard methods for advancing understanding—theory testing and refining measurement procedures through epistemic iteration—face serious challenges.
... The first 'IIT is pseudoscience' letter -a precursor to the one in this issue of Nature Neuroscience -was prompted by the announcement of initial results from an adversarial collaboration sponsored by the Templeton World Charity Foundation (TWCF) in June 2023, which addressed whether the prefrontal cortex is necessarily involved in visual consciousness 1 . Displeased by the announcement, the authors attacked IIT, claiming that it lacks empirical support. ...
... Because each model appears to target only a very specific aspect of experience, there exists the need for a larger integrative framework to better understand their link to phenomenal consciousness: individual processes need to be related globally for their (collective) outputs to contribute to any specific (unified) experience. Somewhat counter the trend of "adversarial collaborations" (Cogitate Consortium et al., 2023) between different models, integrative approaches have also been proposed 22 Note that in all subsections the relevant formalizations should be seen as first steps rather than as conclusive analyses. Also, as pointed out earlier, the concrete mathematical model to be used is up to grabs at this point. ...
... In the future, it could be made more precise by closely scrutinizing known models in the neuroscience of consciousness, the respective empirical material, and its contextualization in the framework of mathematized phenomenology. This work is in principle theory-neutral concerning neuroscience and should be data-driven, thus complementing the current strategy of collaboration beyond models (Cogitate Consortium et al., 2023). ...
In this contribution, we emphasize the relevance of mathematical approaches to the study of the first-person perspective. We briefly discuss why this is necessary and why mathematics plays a special role when it comes to integrating scientific (empirical) and philosophical (conceptual) projects. We also sketch some potential ways that mathematics could be used in phenomenology, based on a selection of four “phenomenological primitives” that characterize constitution: intentionality, minimal subjectivity, lifeworld, and time-consciousness. We also spell out a minimal topological toy model and refer to some contemporary approaches using modern mathematics (e.g. category theory). We then discuss how this could be usefully applied to the (neuro)scientific study of consciousness and throughout stress the need to highlight (and formalize) phenomenology’s transcendental character.
... Currently, most data either supporting or challenging theories of consciousness comes from a combination of sensory stimulation paradigms, behavioral tasks and neural recordings [3]. All this has greatly advanced our understanding of which brain areas and patterns of activity correlate with consciousness -and in particular with perceptual awareness [4,5]. However, the inability to perturb neural activity in a precise fashion in human experiments precludes the possibility to go from the level of correlation between ...
A major debate in the field of consciousness pertains to whether neuronal activity or rather the causal structure of neural circuits underlie the generation of conscious experience. The former position is held by theoretical accounts of consciousness based on the predictive processing framework (such as neurorepresentationalism and active inference), while the latter is posited by the integrated information theory. This protocol describes an experiment, part of a larger adversarial collaboration, that was designed to address this question through a combination of behavioral tests in mice, functional imaging, patterned optogenetics and electrophysiology. The experiment will directly test if optogenetic inactivation of a portion of the visual cortex not responding to behaviorally relevant stimuli will affect the perception of the spatial distribution of these stimuli, even when the neurons being inactivated display no or very low spiking activity, so low that it does not induce a significant effect on other cortical areas. The results of the experiment will be compared against theoretical predictions, and will provide a major contribution towards understanding what the neuronal substrate of consciousness is.
... Along the same lines, some sensory theories of consciousness also predict temporal generalization of decoding, albeit only in posterior cortical regions and only persisting for the duration of the conscious experience [48,49]. Future research using more spatially precise methods of decoding (e.g., fMRI or intracranial EEG) and including manipulations of stimulus duration coupled with manipulations of perceptual awareness will be necessary to fully evaluate whether these meta-stable patterns of decoding are more consistent with cognitive or sensory theories of consciousness [50]. ...
... Az IIT ebben a kontextusban 4 pontot kap, mivel a szintén ismert Global Workspace Theory tudatteóriával holtversenyben tudott predikciós eredményeket felmutatni [6]. Ha a teória még nem mutatott fel kísérleti eredményt, akkor ebben a kontextusban 1-es pontszámot adunk. ...
Korunk mesterséges intelligencia- (MI) felhasználása egyre jobban kiterjed, ezért nem kérdés, hogy új MI-koncepciókra igény van. Kutatási módszertanunkat a kognitív tudományokból érkező új elméletek befogadására dolgoztuk ki, amelyet egy konkrét példán keresztül be is mutatunk: hogyan született, és mire jó az integráltinformáció-teória (IIT), amely a kognitív idegtudományok és a fenomenológia egyik fő elmélete a tudatról. Módszertanunk alappillére a tudományfilozófiai gondolkodásmód, amely kontextusba helyezi az IIT-t, megvizsgálja annak eredetét és felhasználhatóságát. Módszertanunk azt a célt szolgálja, hogy e keretrendszerrel a legkülönbözőbb területekről érkező ötleteket is képesek legyünk értelmezni és azokat rangsorolni mérnöki felhasználhatóságuk szerint, hogy ezekből új, hasznos MI-eszköztárakat készíthessünk.
... IIT scores 4 points in this context because it has produced predictive results comparable to the Global Workspace Theory of consciousness [6]. Additional achievements in brain research include. ...
The use of artificial intelligence (AI) in our time is becoming more and more extensive, so it is no question that there is a need for new AI concepts. We have developed our research methodology to accommodate new cognitive science theories, which we present through a concrete example: how Integrated Information Theory (IIT) was born and what it is good for. IIT is one of the main theories of consciousness in cognitive neuroscience and phenomenology. The main pillar of our methodology is the philosophy of science, which puts IIT into context, examines its origin and usability. This serves the purpose of a framework to be able to interpret ideas from the most diverse fields and to rank them according to their engineering applicability, so that we can create new, useful AI toolkits from them.
... Demonstrating the timeliness and heat of the topic is a recent controversy. Several groups with different theories of consciousness were able to find funding to plan and carry out an adversarial collaboration (Cogitate Consortium, 2023) in which some predictions of several different theories of consciousness were tested with several major brain imaging techniques. The authors concluded that of two major theories examined, neither the Information Integration Theory nor the Global Neuronal Workspace Theory accounted for all of the results; both theories would have to be modified. ...
Considerable recent research in neurosciences has dealt with the topic of consciousness, even though there is still disagreement about how to identify and classify conscious states. Recent behavioral work on the topic also exists. We survey recent behavioral and neuroscientific literature with the aims of commenting on strengths and weaknesses of the literature and mapping new directions and recommendations for experimental psychologists. We reconcile this literature with a view of human information processing (Cowan, 1988; Cowan et al., 2024) in which a capacity-limited focus of attention is embedded within the activated portion of long-term memory, with dual bottom-up and top-down control of the focus of attention. None of the many extant theories fully captures what we propose as the organization of conscious thought at cognitive and neural levels. It seems clear that information from various cognitive functions, based on signals from various brain areas, is integrated into a conscious whole. In our new proposal, the integration involves funneling information to a hub or focus of attention neurally centered in the parietal lobes and functionally connected to areas representing the currently attended information. This funneling process (bringing information from diverse sensory and frontal sources to contact a small parietal area where attended information is coordinated and combined) may be the converse of global broadcasting, from other proposals (Baars et al., 2021; Baars & Franklin, 2003; Dehaene & Changeux, 2011). The proposed system incorporates many principles from previous research and theorization and strives toward a resolution of the relation between consciousness and attention.
