The question of whether artificial intelligence can be conscious has moved well beyond science fiction. It now sits at the center of scientific debate and is increasingly shaping discussions about a range of contentious issues, from AI ethics to animal welfare, fetal development, and laboratory-grown brain tissue.

However, according to a new analysis published in Neuron, the science used to answer that question may not actually be measuring what researchers think it is. A research team led by Hakwan Lau at the Institute for Basic Science in South Korea, with collaborators from the Université de Montréal and New York University, argues that many common experimental methods in consciousness research do not separate subjective experience from general information processing.

In the paper, The Ethical Impasse of Current Consciousness Science, the researchers argue that current scientific tools may not be capable of reliably detecting consciousness.

The Measurement Problem

Consciousness research relies heavily on methods such as visual masking, binocular rivalry, and the detection of perceptual limits. These methods usually compare brain responses when a person is aware of something versus when they are not. The idea is that the difference between these two cases shows whether conscious experience is present or not.

Lau and his team challenge this assumption. When experiments make a stimulus invisible, they often reduce both conscious awareness and the brain’s ability to process information about that stimulus. This means that what appears to be a marker of consciousness in the brain may actually reflect general cognitive activity.

“Many current theories of consciousness appear to be supported by a range of experimental findings,” Lau said. “But those findings may actually reflect general information processing rather than consciousness itself — so it remains difficult to conclude that these theories truly explain consciousness.”

A Historical Warning

The authors compare the current situation to the late 19th and early 20th centuries, when strong claims about consciousness led to a crisis in psychology. The resulting backlash led to the rise of behaviorism, which focused only on observable behavior and halted consciousness research for many years.

Researchers caution that a similar situation could occur again. As AI systems become more advanced and public interest in machine consciousness increases, scientists are under pressure to provide answers. If researchers make strong claims about consciousness in AI, organoids, or fetuses that lack robust methods to support them, scientific credibility could be undermined.

Better Science Required

The authors suggest a different approach. Conditions like blindsight, in which people with brain damage can respond to stimuli they do not report seeing, offer a more controlled way to study consciousness. Another example is hemispatial neglect, where patients fail to notice one side of their visual field while still having basic perception. For researchers, these conditions provide a rare opportunity to separate awareness from information processing and investigate each process on its own.

These conditions show that subjective experience and information processing are distinct from one another. The team says that building experiments around this difference is needed to make reliable scientific claims about consciousness.

The implications of this study extend far beyond the academic world. Deciding whether non-human entities are conscious has direct legal and ethical concerns. The researchers say that the science behind these decisions must meet high standards.

“Questions about consciousness increasingly carry ethical and societal implications,” Lau said. “If scientific claims about consciousness are going to influence discussions about animal welfare, AI ethics, or bioethics, then the scientific foundations supporting those claims must be especially rigorous.”

The researchers conclude that the most urgent challenge is not deciding whether AI, animals, or organoids are conscious, but developing better tools to identify consciousness if it emerges.

Austin Burgess is a writer and researcher with a background in sales, marketing, and data analytics. He holds an MBA, a Bachelor of Science in Business Administration, and a data analytics certification. His work focuses on breaking scientific developments, with an emphasis on emerging biology, cognitive neuroscience, and archaeological discoveries.

Dreams can seem to occur at random, from everyday scenarios to unpredictable, surreal experiences. Now, a new study shows that our personal traits as well as real-life events and experiences actually shape what we dream about, creating patterns in our subconscious.

The study, published in Communications Psychology, analyzed thousands of dream and waking experience reports collected over four years. The researchers used natural language processing tools to quantify the structure of dreams. They found that personal traits like how often someone daydreams, their attitudes about dreams, and their sleep quality all influence dream content. Major shared life events, such as the COVID-19 pandemic, also impacted what people dreamed about.

“Our findings show that dreams are not just a reflection of past experiences, but a dynamic process shaped by who we are and what we live through,” said Valentina Elce, researcher at the IMT School for Advanced Studies Lucca and lead author of the study.

Four Years of Dream Reports

The main dataset included 207 adults aged 18 to 70 who kept a dream diary for two weeks. Each morning, they wrote down everything they remembered from the night’s sleep. Once a day, at a random time, they also recorded what they had been thinking about in the previous 15 minutes. This created a set of waking experience reports to compare with their dream reports.

In addition to the daily records, the researchers collected detailed information about each participant’s sleep habits, cognitive skills, personality, and psychological traits. By the end, they had gathered 1,687 dream reports and 2,843 waking reports from the main group, plus 351 dream reports from 80 people during the first COVID-19 lockdown in Italy in spring 2020.

Dreams Reorganize Reality

When researchers compared participants’ reported dream experiences with situations they reported experiencing while awake, they noticed that dreams don’t simply replay scenarios from our daily lives. Instead, dreams seem to mix familiar places like workplaces, hospitals, and schools into new scenes that blend memories with imagination. Compared to reported waking experiences, the reported dreams tended to focus more on visual details, feature more characters, and make less logical sense. They were also less self-focused and less driven by conscious thinking.

These dream transformations weren’t the same for everyone. Participants who spent more time daydreaming during the day tended to have dreams that jumped rapidly from one scene to another. Those who placed more importance on dreams described them as more vivid and immersive. Sleep quality also played a role: participants who slept poorly showed different patterns in dream content when compared with those who slept better.

Pandemic Influenced Dreams

The lockdown dataset gave researchers a unique opportunity to see how a major external stressor, such as a pandemic, could affect dreams across an entire population.

Dreams recorded during the strict lockdown period were more emotionally intense and mentioned restrictions and limitations more often than dreams from later years. As people adjusted to the new situation, these differences faded. The results suggest that dreams reflect both our personal psychology and the social conditions we share.

AI as a Tool for Studying Consciousness

The team used three large language models, LLaMA 3, ChatGPT-4, and ChatGPT-4 Turbo, to rate dream reports on 16 different features, such as mood, excitement, strangeness, social content, spatial details, and freedom of movement. They combined the scores from the three models and checked them against human ratings. The results showed that these language processing tools could analyze the structure of dream reports as reliably as trained human evaluators. This finding could have uses that extend far beyond this study.

“By combining large-scale data with computational methods, we were able to uncover patterns in dream content that were previously difficult to detect,” Elce said. “This opens new possibilities for studying consciousness, memory, and mental health in a scalable and reproducible way.”

Austin Burgess is a writer and researcher with a background in sales, marketing, and data analytics. He holds an MBA, a Bachelor of Science in Business Administration, and a data analytics certification. His work focuses on breaking scientific developments, with an emphasis on emerging biology, cognitive neuroscience, and archaeological discoveries.