Yaron I.1, Melloni L.2,3, Pi�s M.4, and Mudrik L.1,5

Theories Studies (ConTraSt) database: analyzing and comparing empirical studies of consciousness theories

Introduc�on

References :1 Dehaene, S., & Naccache, L. (2001). Towards a cogni�ve neuroscience of consciousness: basic evidence and a workspace framework. Cogni�on, 79(1-2), 1-37.; 2 Lau, H., & Rosenthal, D. (2011). Empirical support for higher-order theories of conscious awareness. Trends in cogni�ve sciences, 15(8), 365-373.; 3

Lamme, V. A., Super, H., & Spekreijse, H. (1998). Feedforward, horizontal, and feedback processing in the visual cortex. Current opinion in neurobiology, 8(4), 529-535.; 4 Tononi, G. (2004). An informa�on integra�on theory of consciousness. BMC neuroscience, 5(1), 1-22.; 5 Breiman, L. (2001). Sta�s�cal modeling: The two cultures (with comments and a rejoinder by the author). Sta�s�cal science, 16(3), 199-231.

IIT4

Integrated Informa�on Theory

RPT3

Recurrent Processing Theory

HOT2

Higher-Order Theories

GNW1

Global NeuronalWorkspace

Informa�on Integra�onRecurrent ProcessingHigher Order

Representa�onSharing of Informa�onKey mechanism

PHI (Φ)Local synchroniza�onMetacogni�onIgni�on

Posterior Hot ZonePosterior areasPFCFronto-Parietal and anterior Temporal

areasSpa�al predic�ons

Early LatencyEarly LatencyLong LatencyLong LatencyTemporal predic�ons

Can we provide a neutral overview of the field?

Theories are not put to the test?

Why aren’t theories eliminated?

• Vast majority of studies confirming the theories rather than challenging them.

• Only 33% of studies are theory driven.

• Most studies post-hoc interpret their results as suppor�ng the theories.

Challenging at leastone theory

Not challengingany theory

Reviews are wri�en from the standpoint of specific theories

• The overall picture of findings in the field is highly heterogenous.• S�ll, the theories are backed by empirical data compa�ble with their predic�ons.

Spa�al findings (fMRI)

Temporal findings (EEG, iEEG, MEG)• Suppor�ve theory -driven experiments disambiguate some of the pa�erns.

Limited cross-talk between the theoriesIs the field frac�onated?

• Lack of cross-talk between the theories - Increase in support of one theory has no clear implica�ons for other theories.

Cumula�ve distribu�on of experiments through �meAll experiments Suppor�ng

each theoryChallengingeach theory

Diverse means to study consciousness

The outcome of studies can be predicted by methodological parameters• A random forest classifier5 learned the associa�ons between the parameters

and outcomes of N-1 experiments and predicted the outcome of an untrained experiment (leave one out strategy) with above chance accuracy.

State vs. Content

Measures

Popula�on

Report (All)

Measures

GNW RPT IIT HOT

GNW RPT IIT HOT

GNW RPT IIT HOT

GNW RPT IIT

Methods

The database includes 379 papers repor�ng 418 experiments interpre�ng their results in light of at least one of the theories. Each experiment was classified according to parameters of interest.

Distribu�on of papersthat underwent manual screening

1 Sagol School of Neuroscience, Tel Aviv University; 2 Department of Neurology, New York University School of Medicine.; 3Department of Neuroscience, Max Planck Ins�tute for Empirical Aesthe�cs.4 Department of Psychology, Reed College; 5 School of Psychological Sciences, Tel Aviv University

• Most experiments in the field yield suppor�ve results, and rela�vely few studies try to test opposing predic�ons in a theory-driven manner.

• A highly-variable pa�ern of temporal and spa�al findings, that cannot be easily explained by any of the suggested frameworks for consciousness.

• The interpreta�on of a study can be predicted based on the methodological choices made by the researchers, hin�ng at possible methodological biases.

Conclusions

Report (Content only)

GNW RPT IIT HOT