The DOI matches its Crossref record; 155 of the 155 references it cites were found in a registry or an index; 19 of the 51 sources its claims cite were checked for support.
Hard criteria for empirical theories of consciousness
Open accessAbstract
Consciousness is now a well-established field of empirical research. A large body of experimental results has been accumulated and is steadily growing. In parallel, many Theories of Consciousness (ToCs) have been proposed. These theories are diverse in nature, ranging from computational to neurophysiological and quantum theoretical approaches. This contrasts with other fields of natural science, which host a smaller number of competing theories. We suggest that one reason for this abundance of extremely different theories may be the lack of stringent criteria specifying how empirical data constrains ToCs. First, we argue that consciousness is a well-defined topic from an empirical point of view and motivate a purely empirical stance on the quest for consciousness. Second, we present a checklist of criteria that, we propose, empirical ToCs need to cope with. Third, we review 13 of the most influential ToCs and subject them to the criteria. Our analysis helps to situate these different ToCs in the theoretical landscapeand sheds light on their strengths and weaknesses from a strictly empirical point of view.
Overview
CoChat Summary
TL;DR
The paper argues that consciousness can be studied empirically and proposes four hard criteria that empirical theories of consciousness (ToCs) must address: paradigm cases with an unconscious alternative, the unfolding argument, the small/large network arguments, and the other systems argument. Applying these criteria to 13 influential ToCs, it sorts them into four classes and argues that each class faces characteristic challenges rather than being discarded outright.
Key findings
- The paper proposes four criteria for empirical ToCs: (1) paradigm cases of consciousness with an unconscious alternative, (2) the unfolding argument, (3) the small and large network arguments, and (4) the other systems argument.
- It reviews 13 influential ToCs and classifies them into four classes: Class I does not address paradigm cases (e.g., NDT, Orch OR); Class II is subject to the unfolding argument (e.g., IIT, RPT); Class III is subject to the small network argument (e.g., GWT, HOTT, PPT, ART); Class IV is subject to the other systems argument (e.g., TLT, NMDA, AST, SMT, SCMT).
- The small network argument states that many ToCs imply small networks with fewer than ten neurons are conscious; the large network argument notes the human brain contains more than ten billion neurons, so such theories may imply billions of conscious subsystems.
- The unfolding argument holds that recurrent and feedforward networks can approximate any input-output function, so causal structure theories such as IIT and RPT imply functionally identical systems can differ in consciousness, making them difficult to address empirically.
Methods
- Conceptual and theoretical analysis; no new experiments, participants, or primary empirical data are reported.
- The authors review 13 influential ToCs, subject them to the proposed criteria, and summarize the results in Table 1, with additional theories described in an appendix.
- The analysis focuses on prototypical argumentation rather than exhaustive coverage or the latest version of each theory.
Limitations
- The authors state the review cannot be exhaustive, that it is impossible to perfectly portray complex, multi-faceted ToCs in a few lines, and that they are not interested in minute details of specific ToCs.
- The authors state the criteria are not exhaustive or unique, are a starting point/steppingstone, and that ToCs challenged by the criteria should not necessarily be discarded because fixes may exist.
- Evident from the text: the classification partly reflects the authors’ own interpretation, and some ToCs have vaguely defined mechanisms, making it difficult to determine exactly how the criteria apply.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
12 Valid · 16 Not valid · 21 Not checked · 145 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“GWT is not subject to the unfolding argument because broadcasting is a general concept, which can be implemented in many di ff erent ways (even though GWT is usually framed in recurrent terms, recurrence is not essential, Doerig, Schurger, et al., 2019; but see Kleiner, 2019; Kleiner & Hoel, 2020).”
What the sources say
- Could not verifycites Doerig, Schurger, et al., 2019 sourceExperimental — accuracy not yet measured
The indexed copy returns no body-text passage addressing whether GWT is exempt from the unfolding argument or whether broadcasting can be implemented in different ways. Searches for those points returned only landing-page material and an abstract, so the claim cannot be verified from the available evidence. (LLM, conf=0.70)
- Not checked — no full text of the sourcecites Kleiner, 2019 source
- Disputedcites Kleiner & Hoel, 2020 sourceExperimental — accuracy not yet measured
The Introduction (page unavailable) calls the unfolding argument against IIT a subset of a broader problem affecting theories that predict experience from internal observables; the paper treats Global Neuronal Workspace as an example. It also says that interpretations of Global Workspace Theory based on report accessibility face a different problem, empirical unfalsifiability. It does not present flexible implementations of broadcasting as an exemption. (LLM, conf=0.86)
Checked by a reasoning model — may be wrong.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
Some cited sources support this claim and some don’t — read each source’s verdict below.
In the paper
“The unfolding argument does not apply, because predictive processing can be unfolded (Doerig, Schurger, et al. (2019); but see Kleiner (2019) and Kleiner and Hoel (2020)).”
What the sources say
- Supportedcites Doerig, Schurger, et al. (2019) source
The indexed Abstract (no page number available) distinguishes predictive coding, which it treats as a function realizable by different patterns of brain connectivity, from the causal-structure theories targeted by the unfolding argument. That supports the claim that predictive processing can be unfolded and is not a target of the argument. (LLM, conf=0.84)
- Not checked — no full text of the sourcecites Kleiner (2019) source
- Disputedcites Kleiner and Hoel (2020) sourceExperimental — accuracy not yet measured
In Formal Description of Testing Theories (page unavailable), the paper explicitly includes predictive processing among the theories it considers. In The Substitution Argument (page unavailable), it says the flaw illustrated by unfolding is not limited to IIT but affects almost all contemporary theories; it does not present predictive processing as exempt. The citing sentence’s “but see” therefore fits this paper as a counterpoint, not as support for the stated claim. (LLM, conf=0.79)
Checked by a reasoning model — may be wrong.
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Relatedly, Sergent et al. (2013) showed that a cue presented after a target stimulus that would otherwise remain unconscious can retrospectively render the stimulus conscious.”
What the sources say
- Supportedcites Sergent et al. (2013) source
The Summary reports that directing attention to a near-threshold stimulus 100–400 ms after presentation increased subjective visibility. The Discussion explicitly concludes that postcueing can trigger conscious perception of a stimulus that would otherwise have remained unconscious. (LLM, conf=0.95)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“For example, words can be semantically processed without being consciously perceived (Gaillard et al., 2006)”
What the sources say
- Supportedcites Gaillard et al., 2006 source
The abstract (p.1) states that the threshold for conscious access is lower for emotional than neutral words, 'even for words that have not been previously consciously perceived, thus implying that written words can receive nonconscious semantic processing.' The Discussion reinforces this: 'genuine nonconscious semantic access to the emotional content of words can occur even for words that were never seen before.' The citing sentence is a general paraphrase of exactly this conclusion; the paper's own demonstration is confined to a mask/visual-threshold paradigm contrasting emotional vs. neutral words, which the source itself generalizes to 'written words'. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“However, for many de fi nitions of complexity, even very small networks with less than 10 neurons exhibit high complexity (Herzog et al., 2007; Oizumi et al., 2014).”
What the sources say
- Not checked — no full text of the sourcecites Herzog et al., 2007 source
- Partly supportedcites Oizumi et al., 2014 sourceExperimental — accuracy not yet measured
In “Consciousness and connectivity: Modular, homogeneous, and specialized networks” (Figure 17C; page unavailable), the paper describes a five-gate network with high values of its complexity-related measures. The elements are logic gates, not neurons, and the passage does not establish that this holds for many definitions of complexity; Figure 17B also shows a five-gate network with low values. (LLM, conf=0.86)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
71 more claims · 5 with their cited sources checked
Create a free account to see all 194 claims and every source verdict.
- Probably not a substantial claim. 92
- It may state two separate claims as one. 56
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“The problem with very liberal formulations is that they can be satis fi ed by very simple arti fi cial systems that are not plausibly conscious; many computational functionalist theories allegedly fail by giving such liberal conditions [54 ,55].”
From Identifying indicators of consciousness in AI systems (2026) · cites this paper as [55]
The cited source says what the claim attributes to it.
In the citing paper
“Based on how F is calculated to ''quantify'' PC, artificial systems with purely feedforward architecture have F = 0, and hence lack consciousness (PC), even when their overt behaviors (input-output functions) are indistinguishable from that of conscious (F > 0) systems with feedback 94,95 (but see Doerig et al. 155 ).”
From An integrative, multiscale view on neural theories of consciousness (2024) · cites this paper as [155]
The cited source says what the claim attributes to it.
In the citing paper
“Indeed, there are now so many theories of consciousness that an increasing number of researchers are turning their attention to the meta-theoretical question of how they might be compared and evaluated (Chis-Ciure et al., 2024; Del Pin et al., 2021; Doerig et al., 2021; Kirkeby-Hinrup, 2024; Negro et al., 2024; Signorelli et al., 2021), with a view to advancing the field via the unification, integ”
From Integrated information and predictive processing theories of consciousness: An adversarial collaborative review (2026) · cites this paper as Doerig et al., 2021
The cited source says what the claim attributes to it.
In the citing paper
“The key idea here is to exploit the resources availed by a domain-general modelling framework such as Active Inference to construct computational models of paradigmatic phenomena in consciousness science (e.g., binocular rivalry; Doerig et al., 2021), and to compare the properties of these models such that salient commonalities and particularities may be identified (Vilas et al., 2022; Whyte et al”
From Integrated information and predictive processing theories of consciousness: An adversarial collaborative review (2026) · cites this paper as Doerig et al., 2021
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“(1) L5 pyramidal neurons associate information arriving at separate dendritic compartments through a non-linear apical dendritic process leading to burst firing.120-122”
From An integrative, multiscale view on neural theories of consciousness (2024) · cites this paper as [120]
Experimental — accuracy not yet measured
1 more claim about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
James, W. (1913) · Henry Holt and Company eBooks · doi:10.1037/11059-000 · Cited by 25,226
Multilayer feedforward networks are universal approximators
Hornik, K., Stinchcombe, M., & White, H. (1989) · Neural Networks · doi:10.1016/0893-6080(89)90020-8 · Cited by 21,728 · cited 1×
“It is a mathematical fact that both recurrent and feedforward networks (along with many other kinds of universal function approximators) can approximate any input-output function to any degree of accuracy (Hornik et al., 1989; Schäfer & Zimmermann, 2006).”
BackgroundMultilayer feedforward networks are universal approximators
HornikK., StinchcombeM., & WhiteH. (1989) · Neural Networks · doi:10.5555/70405.70408 · Cited by 9,081
The free-energy principle: a unified brain theory?
Friston, K. (2010) · Nature Reviews Neuroscience · doi:10.1038/nrn2787 · Cited by 8,120 · cited 1×
“The brain is often seen as a predictive processing machine (Clark, 2013; Friston, 2010; Rao & Ballard, 1999).”
BackgroundKripke, S. A. (1972) · Semantics of Natural Language · doi:10.1007/978-94-010-2557-7_9 · Cited by 7,028 · cited 3×
“More precisely, although the software and hardware used in that implementation are sufficient to instantiate Microsoft 2Multiple realization has been a fruitful topic in philosophy (Fodor, 1974; Kripke, 1972; Putnam, 1967, 1988).”
BackgroundMethodJournal of Consciousness Studies
(2022) · Journal of Consciousness Studies · doi:10.53765/20512201 · Cited by 6,832
Neuronal Oscillations in Cortical Networks
Buzsáki, G., & Draguhn, A. (2004) · Science · doi:10.1126/science.1099745 · Cited by 6,576 · cited 1×
“Examples include NMDA synapses (Flohr, 1992), processing in layer 5 of V1 (Crick & Koch, 2003), the claustrum (Crick & Koch, 2005), gamma-band oscillations (Buzsáki, 2004), thalamocortical loops (Llinas et al., 1998), or the microgenesis of consciousness in the thalamus (Bachmann, 2000).”
BackgroundWhatever next? Predictive brains, situated agents, and the future of cognitive science
Clark, A. (2013) · Behavioral and Brain Sciences · doi:10.1017/s0140525x12000477 · Cited by 5,924 · cited 1×
“The brain is often seen as a predictive processing machine (Clark, 2013; Friston, 2010; Rao & Ballard, 1999).”
BackgroundRao, R. P. N., & Ballard, D. H. (1999) · Nature Neuroscience · doi:10.1038/4580 · Cited by 5,483 · cited 2×
“The brain is often seen as a predictive processing machine (Clark, 2013; Friston, 2010; Rao & Ballard, 1999).”
BackgroundRubber hands ‘feel’ touch that eyes see
Botvinick, M., & Cohen, J. (1998) · Nature · doi:10.1038/35784 · Cited by 4,410 · cited 1×
“In the rubber hand illusion, participants can be tricked into attributing ownership of their real hand to a fake rubber hand (Botvinick & Cohen, 1998).”
Background
Research works printed but not linked: 10
- Baars B. J., 2013, Frontiers in Psychology, 4, 200
- Chalmers D. J., 2017, Panpsychism: Contemporary Perspectives, 179
- Descartes R., 1996, Discourse on the method: And, meditations on first philosophy
- Flohr H., 1992, Emergence or Reduction, 220
- Gazzaniga M. S., 1970, The bisected brain, 2
- Lamme V., 2015, Open MIND
- Noë A., 2004, Action in perception
- Park H.-D., 2014, Soc., 369, 20130208
- Searle J. R., 2000, Journal of Consciousness Studies, 7, 3
- Tononi G., 2015, Soc., 370, 20140167
Cited by
Predictive processing as a systematic basis for identifying the neural correlates of consciousness
Hohwy, J., & Seth, A. (2020) · Philosophy and the Mind Sciences · doi:10.33735/phimisci.2020.ii.64 · Cited by 170
The ConTraSt database for analysing and comparing empirical studies of consciousness theories
Yaron, I., Melloni, L., Pitts, M., & Mudrik, L. (2022) · Nature Human Behaviour · doi:10.1038/s41562-021-01284-5 · Cited by 147
“The parameters whose average importance was higher than 95% of the distribution of importance scores of a random parameter were (in order of importance): (1) studying state versus content consciousness, (2) using report versus no-report paradigms, (3) using connectivity measures and (4) using subjective measures of co…”
BackgroundMethodInfluentialAn integrative, multiscale view on neural theories of consciousness
Storm, J. F., Klink, P. C., Aru, J., Senn, W., Goebel, R., Pigorini, A., Avanzini, P., Vanduffel, W., Roelfsema, P. R., Massimini, M., et al. (2024) · Neuron · doi:10.1016/j.neuron.2024.02.004 · Cited by 114
Francken, J. C., Beerendonk, L., Molenaar, D., Fahrenfort, J. J., Kiverstein, J. D., Seth, A. K., & van Gaal, S. (2022) · Neuroscience of Consciousness · doi:10.1093/nc/niac011 · Cited by 136
“how it should be operationalized), what would constitute an explanation, and how valid experiments should be designed (Block 1995; Chalmers 1995, 2000; Dehaene and Naccache 2001; Dennett 2001; Lamme 2006; Tononi 2008; Dehaene and Changeux 2011; Doerig et al. 2020; Hohwy and Seth 2020); see Seth and Bayne (2022) for a…”
Background'Orch OR' is the most complete, and most easily falsifiable theory of consciousness
Hameroff, S. (2021) · Cognitive Neuroscience · doi:10.1080/17588928.2020.1839037 · Cited by 84
“Reduced brain volume and normal function Figure 3 in Doerig et al. (2020) shows the brain of a person with markedly reduced brain volume, but normal cognition and consciousness.”
Background
Metadata
- Authors
- Doerig, A., Schurger, A., & Herzog, M. H.
- Affiliations
- École Polytechnique Fédérale de Lausanne (CH)
- Inserm (FR)
- Corresponding author
- Doerig, Adrien
- Venue
- Cognitive Neuroscience
- Publisher
- Informa UK Limited
- Published
- 2020-07-14
- Volume
- 12
- Issue
- 2
- Pages
- 41-62
- ISSN
- 1758-8928 (ISSN-L)
- 1758-8936
- OpenAlex
- W3043429318
- Semantic Scholar
- 6a2fd985d926673863670e7a43cb52d63de95773
- Cited by
- 194
- References
- 153
- Citations by year
- 2026: 41 · 2025: 33 · 2024: 30 · 2023: 33 · 2022: 32 · 2021: 18 · 2020: 7
- FWCI
- 21.77
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Cognitive Neuroscience (publishedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- Radboud Repository (Radboud University) (submittedVersion, other-oa)
- Chapman University Digital Commons (Chapman University) (submittedVersion, cc-by-nc-nd)
- Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
- Radboud Repository (Radboud University) (submittedVersion)
- Funders
- Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (176153)
- Topics
- Social Sciences › Arts and Humanities › Philosophy › Philosophy and Theoretical Science
- Social Sciences › Arts and Humanities › Philosophy › Free Will and Agency
- Social Sciences › Arts and Humanities › History and Philosophy of Science › Philosophy of Biology and Scientific Explanation
- Keywords
- theory of consciousness
- MeSH terms
- Consciousness, Humans, Empirical Research
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-28, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Hard criteria for empirical theories of consciousness. https://app.cochat.ai/papers/10.1080/17588928.2020.1772214
- The paper
- Doerig, A., Schurger, A., & Herzog, M. H. (2021). Hard criteria for empirical theories of consciousness. Cognitive Neuroscience. https://doi.org/10.1080/17588928.2020.1772214
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.