Make science more collegial: why the time for ‘adversarial collaboration’ has come
Bringing together proponents of rival theories to test their ideas against each other can advance science — but only if all sides can accept that they might be wrong
Adversarial collaborations enable proponents of opposing theories to work together.Credit: Clerkenwell/Getty
Controversy is a “terrible way to advance science”, psychologist Daniel Kahneman said in a 2022 lecture for Edge, a science website. The context was a concept called adversarial collaborations, which Kahneman, best known for his studies in behavioural economics and decision-making psychology, had developed with his colleague and wife Anne Treisman to combat what he called “angry science”. By stimulating competing groups to work together on difficult problems, the aim of these collaborations is to avoid, he said, “the nasty world of critiques, replies and rejoinders” and allow rivals to arrive at shared truths.
Kahneman, who received the economics Nobel prize in 2002, was not the only researcher to think along these lines. In 1988, the Journal of Applied Psychology published a study between competing researchers, although it did not use the term adversarial collaboration1. In 2022, Nature Human Behaviour published an adversarial collaboration testing a theory that says individuals’ subjective experiences of emotion are influenced by their facial expressions2. Now, Nature has published a study that tests two theories of consciousness in which proponents of both worked together3.
The findings were first released as a preprint4 in 2023. In addition to the study, we are also publishing the peer reviewers’ reports and author responses (see go.nature.com/4afvahp), so that readers can judge for themselves how things went. The results were mixed. In this instance, an adversarial collaboration could not completely avoid some scientists using strong language to defend their theory — not the authors of the study, but others in the field. The best way for an adversarial collaboration to truly advance science is for all concerned to be willing to accept that there is a chance that they could be wrong.
Consciousness describes all of the subjective experiences, internal thoughts and feelings that give humans a sense of themselves as real, existing beings. It’s been 35 years since molecular biologist Francis Crick and cognitive scientist Christof Koch proposed that the study of this phenomenon, long considered mostly the realm of philosophers, could get a boost from neuroscience5. The advent of brain-imaging technologies, such as functional magnetic resonance imaging, opened one door, and neuroscientists tumbled through enthusiastically.
Anyone expecting swift clarity was disappointed. There are now at least 20 theories proposing a physical explanation for subjective experiences — for example, when one ‘senses’ something without conscious cues. Researchers often have strong opinions on these theories, meaning that the study of consciousness is ripe for an adversarial collaboration.
Two of the leading theories are integrated information theory (IIT) and global neuronal workspace theory (GNWT). According to IIT, conscious sensations reside in the posterior cortex — at the back of the head. Neural activity associated with consciousness is thought to be transient; it happens for just the time a conscious experience takes. By contrast, GNWT says that consciousness mainly involves the prefrontal cortex, at the front, which also processes sensory information from various brain regions.
An agreed study design to test both theories did not come easily, nor quickly. After months of deliberation, principal investigators representing each theory, plus an independent researcher acting as a mediator, signed off on a design that involved 6 laboratories running tests on 256 participants. Study participants were presented with various images, including faces, clocks and letters of the alphabet, and their brain activity was assessed in accordance with each theory.
Neither theory was validated fully by the collaborative study. Each theory’s principal investigators and the other authors accepted this finding: “These results align with some predictions of IIT and GNWT, while substantially challenging key tenets of both theories.”
Daniel Kahneman helped to pioneer adversarial collaborations and was involved in several such studies.Credit: Roger Parkes/Alamy
However, many scientists in the wider ecosystem of consciousness did not continue the conversation in the same collegial spirit — quite the opposite, in fact. In an open letter after the initial release of the results, supporters of GNWT called IIT a pseudoscience6. Such language has no place in a process designed to establish working relationships between competing groups.
Should we be surprised by this development? Not in some respects. As the mediator, neuroscientist Lucia Melloni at the Max Planck Institute for Empirical Aesthetics in Frankfurt, Germany, said in a News Feature in Nature last year (see Nature 625, 438–440; 2024), Kahneman had warned that people change their minds only when fresh evidence accumulates over time. Does this mean accepting that pockets of science will remain angry while disagreements persist? Kahneman showed that this does not need to be the case.
In an adversarial collaboration of his own7, he worked with psychologist Gary Klein to study expert intuition, a concept in psychology that describes a kind of intuitive judgement that arises from knowledge and experience. Klein supported this idea whereas Kahneman broadly didn’t. Over the course of working together, the two became friends — while still disagreeing.
Fittingly, Kahneman’s last paper before he died in March 2024 also involved an adversarial collaboration8. In 2010, together with fellow economics Nobel laureate Angus Deaton, Kahneman had proposed that happiness tracks with rising income up to a certain point, after which it plateaus9. Another group reported no such plateau — more money corresponded to more happiness10. Authors of both studies formed an adversarial collaboration that found that the curve did flatten, but the pattern held only for the least happy people7 — a result accepted by both sides.
Kahneman and his colleagues not only pioneered adversarial collaborations, but also showed how to conduct them. This collegial spirit must be revived in future collaborations in consciousness, and in other fields. We need more of them as a way to advance science.
Nature 641, 281-282 (2025)
doi: https://doi.org/10.1038/d41586-025-01379-3
This story originally appeared on: Nature - Author:furtherReadingSection
