NEW YORK CITY—Amidst rock music, a rap about consciousness, and the calling in of a 25-year-old drunken bet, camps backing two leading theories of how consciousness arises from the brain waited anxiously in a Greenwich village theater on Friday to hear who had won the first round of an ambitious “adversarial collaboration.” On balance, the initial results presented appeared to favor the idea that consciousness is a feature of networks of neurons found at the back of the brain.
But the opposing camp is far from ready to concede. It still contends that consciousness emerges within the brain’s “executive” center, the prefrontal cortex. “The results ended up challenging both [groups], with key predictions of the two theories being disconfirmed by the data,” says Liad Mudrik, a cognitive neuroscientist at Tel Aviv University who helped design and assess the scientific showdown.
The unusual evening event, part of the annual meeting of the Association for the Scientific Study of Consciousness (ASSC), also served as the denouement of a wager placed in 1998 at the second such conference. There, cognitive neuroscientist Christof Koch bet philosopher David Chalmers that the neural correlates of consciousness would be nailed down in 25 years. Drawing on the new experimental results, Koch yesterday conceded that those correlates remain unclear and on a stage gallantly offered up a bottle of 1978 Madeira to Chalmers, with five more fine reds in the wings.
For the collaboration, funded by the Templeton World Charity Foundation (TWCF), both sides of the consciousness debate agreed on experiments to be conducted by “theory-neutral” labs with no stake in the outcome. It pits Integrated Information Theory (IIT), the sensory network hypothesis that proposes a posterior “hot zone” as the site of consciousness, against the Global Neuronal Workspace Theory (GNWT), which likens networks of neurons in the front of the brain to a clipboard where sensory signals, thoughts and memories combine before being broadcast across the brain.
On Friday night, IIT’s advocates were ready to declare victory. “The results corroborate IIT’s overall claim that posterior cortical areas are sufficient for consciousness, and neither the involvement of [the prefrontal cortex] nor global broadcasting are necessary,” said Melanie Boly, a neurologist and neuroscientist at the University of Wisconsin and a leading proponent of IIT.
But GNWT’s stoic chief architect, Stanislas Dehaene, Director of the INSERM-CEA Cognitive Neuroimaging Unit in Orsay, France, believes this experimental round had limitations and the results of other tests in the adversarial collaboration–still to be announced–will support the role of the prefrontal cortex. He adds that the new findings locating conscious perception to the back of the brain are predicted by lots of theories, and don’t confirm the specifics of IIT.
Consciousness has compelled philosophers since Plato, but over the last three decades, neuroscientists have entered the fray. Both disciplines seek a working theory of consciousness as the first step towards measuring the phenomenon–whether to make life and death decisions about brain-damaged patients, ascribe rights to animals or determine whether AI may have it.
Among dozens of theories of consciousness, GNWT and IIT are among the most widely discussed. GNWT gained initial support from experiments that asked participants to report the moment they became aware of a stimulus, such as an image flashing on a screen. In those studies, many of them led by Dehaene, brain scans showed that the prefrontal cortex lit up at the moment of perception.
But philosophers and experimentalists questioned whether these studies captured the neural markers of conscious perception, or simply the task of reporting it. Cognitive processes such as paying attention and storing information in memory, both of which enable participants to respond that they’ve seen an image, are known to take place in the prefrontal cortex.
“No-report” studies, where participants passively view images, seemed to offer a workaround. A popular one involves binocular rivalry: if different images are shown to a person’s left and right eye, their conscious perception flips between them. These flips can be monitored—independent of participants’ report—by tracking eye movements. And lo, these experiments found signals of conscious perception at the back of the brain, the area predicted by IIT.
The front-of-brain camp fought back, arguing that these studies were themselves rife with confounders. For example, participants might be so wearied by staring at onscreen images that they stop paying attention to them and let their mind wander to other tasks, a phenomenon that New York University philosopher Ned Block dubbed the “bored monkey problem.”
It was this cauldron of contested evidence that fueled the adversarial collaboration. The project, launched in 2019, was the brainchild of Koch, then chief scientist at the Allen Institute and a proponent of IIT, and Dawid Potgieter, director of Discovery Science programs at TWCF, which committed 20 million dollars to a series of grants for adversarial collaborations testing theories of consciousness.
For the GNWT-versus-IIT phase of the project,Mudrik and the two other independent project leaders, psychologists Lucia Melloni at Max Planck Institute and Michael Pitts at Reed College spent a year working closely with Dehaene and Giulio Tononi, a psychiatrist and neuroscientist at the University of Wisconsin and chief architect of IIT, to design two experiments for which each theory offered clearly distinct predictions. Dehaene and Tononi would have no role in performing the experiments or writing up the results.
The team preregistered the experimental design on an open science website and published the details last February. Six theory-neutral labs would scan the brains of 250 total participants using three techniques: functional magnetic resonance imaging, magnetoencephalography, and electrocorticography, in which electrodes are placed on the brain’s surface prior to a surgery.
The first of the two planned experiments showed participants images with and without an accompanying task—pressing a button in response to either of two target pictures—so researchers could look for differences in the resulting brain signals. IIT predicts that passive perception will activate the back of the brain, but perception while performing tasks will spark the front. GNWT predicts similar brain activation in the two situations.
Key to the experiment were algorithms called multivariate pattern decoders, which could predict which image a participant was viewing at a given time based on their brain signals. Researchers initially “trained” these decoders by feeding them examples of that participant’s brain activity data along with the corresponding image.
GNWT predicts that the frontal networks supporting both active and passive perception should be similar enough to allow the decoder to cross train. That is, if it’s been trained only on signals related to the task of passively observing a face, it should still be able to decode data from the task of pressing a button in response to a face. IIT predicts that cross training will only work well with brain signals from the posterior regions, the proposed site of conscious perception.
The results were surprisingly mixed. When it came to decoding different categories of objects, the data provided strong support for GNWT to the eyes of many. But when it came to decoding the orientation of faces, IIT was the better fit.
In another analysis, the tables were turned. During conscious perception, IIT predicts neural communication within posterior areas, while GNWT predicts it should be between visual and frontal zones. And in the study, “the expected communication patterns were in line with GNWT,” says Mudrik.
The timing of the recorded signals, meanwhile, offered stronger support for IIT. In the posterior region, activity persisted as long as the image was presented onscreen, as IIT predicts. GNWT instead predicts an initial spike of activity—the “ignition” of the frontal workspace—and another spike when the stimulus disappears. That theory scored a partial win: there was evidence for an initial spike, but not the “off” spike.
Dehaene says the design of the experiment compromised the sensitivity of signal decoding from the front of the brain that would have supported GNWT. It was, he says, a design that Tononi was keen on. In a trade-off, Dahaene scored his preferred design for the subsequent TWCF-funded experiment, which the research team hopes to present at next year’s ASSC meeting. Using a customized video game to distract participants, this experiment will isolate neural signals of conscious perception by comparing brain signals when subjects are aware of seeing an image and when they’re not.
Although Koch’s favored theory now has a leg up on GNWT, he says the continuing doubts around the new results were enough to pay off the bet to Chalmers. “I’ve lost the battle,” he declared onstage, “but won the war for science.”
Disclaimer: The views, suggestions, and opinions expressed here are the sole responsibility of the experts. No Marketwise Analytics journalist was involved in the writing and production of this article.
