Accelerating Research on Consciousness
Using a variety of funding, networking and outreach mechanisms, we will support targeted scientific experiments to empirically investigate scientific theories of consciousness.
Consciousness is a foundational concept for many big questions, but one which has been plagued with practical and conceptual challenges. Most scholars agree that we need our brains to be conscious – but there is no comprehensive explanation for, much less agreement on, how the brain gives rise to consciousness. As is the nature of science, eminent scholars across disciplines can look at the same data and draw different conclusions about consciousness – what it is and how it arises.
In order to accelerate the progress of research, support must be designed to catalyze cross-disciplinary cutting-edge practices, such as adversarial collaboration and the filing of registered reports. These practices promote open and rigorous engagement among leaders of opposing theories. We hope that demonstrating success in this area will lead to groundbreaking experiments and rekindle the interest of other philanthropies and government organizations in supporting research on consciousness. It may be a century or more before humans fully understand consciousness, so what can be achieved in five years must be measured with realistic expectations. However, research can be accelerated tangibly in the near term as detailed below.
If successful, we expect:
- A tangible reduction in the number of plausible scientifically testable theories of consciousness (and increase in legitimacy of the remaining theories).
- A series of experiments comparing 2-4 theories at a time.
- A grant-development mechanism that might be replicated, improved, or scaled through an iterative process.
- To compare theories empirically, by focusing on testable predictions.
We do not expect:
- Experiments that will, or will attempt, to solve all the mysteries of consciousness.
- To compare theories only conceptually or metaphysically.
- A single experiment to compare all theories at once.
- To solve all challenges upfront, but will instead learn from each round of grant development to improve the next one.
To develop maximise the potential impact of our experiments we invite leading scientists and philosophers to design collaborative experiments through a six-stage process of grant development and execution. Throughout this process, we also promote and support efforts to maximise scientific rigor and replicability.
Our Six-Stage Process
To maximise the likelihood of success, our main grant-development process has been structured in six stages, each of which involves interactions with different stakeholders and a different role for TWCF.
1. Planning a Meeting:
A host will be identified by surveying the literature and interviewing scholars in the field. TWCF will reach out to the host, pay for the meeting, and work with the host to identify participants. In order to identify a date that will maximize attendance, schedules will be surveyed six to nine months in advance of the meeting. To facilitate discussion at the meeting, participants will be interviewed in advance to discern the theories to which they are committed, and how they are likely to respond to their adversaries.
2. Meeting with a Small Group of Experts:
During the meeting, we present the opportunity and conditions of adversarial collaboration. We will promote creative suggestions for adversarial collaborations, but emphasize that resources are not being offered beyond the scope of the meeting. Based on the planning interviews, we work with the host to focus the conversation on testing core predictions that are incompatible between theories. Participants are encouraged to identify a discrete experiment and draft a design that is as detailed as possible. The group will be expected to agree on clearly articulated next steps, including assigning responsibility for drafting and managing the registered report.
3. Proposal for an Experiment:
Participants will write a registered report collaboratively over a period of about six to twelve months. To maintain momentum, TWCF will host webinars, keep in frequent contact with key stakeholders, and provide general coordination. The whole group will submit the report together as co-authors. Detail will include but is not limited to:
- The names and affiliations of investigators;
- The methods they will use to collect and store data;
- The experimental protocol;
- Methods for data analysis and interpretation;
- Predictions of each theory, and the findings required to eliminate it.
Once the registered report recieves an In Principle Acceptance (IPA), the study design cannot easily be changed, but leading figures in the community will be invited to include their predictions if they wish, allowing broader adversarial collaboration. This step provides an opportunity to test theories not discussed at the meeting.
4. Decision on the Report and Grant Application:
The registered report is submitted to a leading journal which sends it out for peer review.
If approved by the journal, the report and reviews will become the basis for a proposal to be considered for funding by TWCF. If approved for funding, the project will be launched. When basing a funding decision on registered reports is not practicable, TWCF may require applicants to file a preregistration with an organization such as the Open Science Framework, including letters of support from the leaders of the theories under investigation.
5. Conducting the Experiment:
The research may require multiple methodological approaches and investigators. Where possible, data will be collected by a contract research organization and/or with simultaneous replication. Given the complexity of such an approach, funds will be provided for researchers to visit collaborators, and for a project manager to ensure proper standards for data collection and sharing. The data collection and storage must, wherever feasible, be open to other researchers.
6. Publication of Findings:
The registered report process allows for an article to be published even if the experiment yields null results or disproves one or more of the theories being tested. A media campaign will be enlisted to raise the profile of the publication. We also plan to provide recognition and support for those who submit to disproof as a result of legitimate experimental testing.
Best Practices and Tools
The following practices will be incorporated into the main grant-development mechanism wherever possible. Though it may not be feasible to always include all seven practices, priority will be given to adversarial collaboration, registered reports, open data sharing, and simultaneous replication.
Adversarial collaboration involves two or more scientists, representing contradictory theories, designing an experiment together with the aim of discovering legitimate evidence in support of one theory over another. Our adversarial collaborations focus on testing incompatible predictions between theories, which requires a priori agreement that there is sufficient overlap between the theories for an experiment to be meaningful. By its nature, we do not expect this approach to be applicable to metaphysical commitments or conceptual frameworks, though we remain open to such collaborations. There is always the risk of failure in adversarial collaboration when the theories involved are so radically distinct that participants are unable to identify testable and mutually preclusive predictions between them. Though such an outcome would not lead to a research project, it could provide useful information about whether a theory of consciousness is mature enough for testing.
Registered Report is a publication format which involves two steps of expert review – before and following data collection. Currently being offered by at least 140 journals, Registered Reports have already been used to some extent in the grant application process to increase both the quality of experiments and the likelihood of publication (even if the experimental findings are not as expected).
This format can be used to ensure collaborating scientists specify the experimental design in advance, and in sufficient detail for publication. When making use of registered reports is not practicable, applicants can file a preregistration with the Open Science Framework as an alternative. Preregistration allows for the experiment design to be thoroughly documented, but it does not involve commitment from a journal.
Open peer review is the practice of making reviews visible and accessible to anyone, not just the study author(s), and requires that the reviewer is identified. This approach has the advantage that those who read the published article gain a broader perspective on the strengths and limitations of the study. Moreover, it helps to improve the quality of review and limit the potential for bias by holding reviewers accountable for their claims. It also encourages scholarly debate and stimulates discourse in a broader scientific community, rather than limiting scholarly debate to author and reviewer. This may help those not completely immersed in the field to take in the findings from a more balanced perspective. A potential hurdle of open review is that some reviewers may be reluctant to take part before such practices are normalized.
Crowdsourcing external reviews creates an opportunity to source comments from a broad audience. (This is a practice rarely applied to the review of single experiments.) Crowdsourcing encourages critique from the broad scientific community, thus enabling researchers to improve their study design before the experiments begin. This in turn reduces the likelihood of proceeding with experimentation based on a flawed design. The downside is that crowdsourcing introduces the risk of distraction caused by biased or uninformed reviews. If it is to be used, crowdsourcing will be spearhead by TWCF to avoid imposing a potential extra burden on the researchers and/or the journal.
Inviting Other Theorists to Make Predictions
Generally, this grant-development mechanism will involve a relatively small group of experts to compare two to four theories. If it seems possible that the proposed experiments can report on other theories of consciousness not included in the original discussion, then we may create an opportunity for such theories to be involved by:
- Completing the experiment design;
- Inviting leaders of other theories to add their own predictions to the experimental design, provided the protocol doesn’t have to be significantly altered;
- Including comments either in the registered report or as a separate publication.
To support the process of adversarial collaboration, experiments must be executed impartially and rigorously. Where possible, we intend that experiments will be either conducted by a contract research organization, or that they will be replicated in parallel. This could take the form of two (adversarial) collaborators agreeing on an experiment that they both conduct, or of the collaborators jointly nominating one or two other researchers to conduct the research.
Open Data Sharing
To maintain transparency in the interpretation of experimental data, researchers can be encouraged to make their data and analysis pipelines openly available on a shared portal. This approach will allow multiple scholars to explore the same data and identify limitations in the interpretation of the data and/or alternative explanations. We will promote data sharing according to the FAIR Principles.