Is brain consciousness not always online? Chinese scientists discover neural activity in both conscious and unconscious states.

Is brain consciousness not always online? Chinese scientists discover neural activity in “conscious” and “unconscious”…

Rubin’s vase is a classic optical‑illusion image; when you stare at it, the picture alternates between a white vase and two opposing black profiles. It raises a fundamental neuroscience question: how does the brain enable us to perceive these two scenes alternately? Assuming the image itself remains unchanged, where does this selection happen, and how does it occur?

 

In the 1990s, Nobel laureate and co-discoverer of the DNA double helix, Francis Crick, encouraged neuroscientists to search for neural activity that directly corresponds to conscious experience.

 

Subsequent researchers have mostly used binocular rivalry experiments: presenting different images to each eye, with the subject consciously perceiving only one at a time, thereby isolating the signal of consciousness.

 

Around the location of this signal, academia has formed two major camps: one side believes consciousness depends on a broadcast network centered on the prefrontal cortex, while the other side argues that the sensory cortex in the back of the brain is sufficient.

To determine which side was superior, scholars from both camps jointly designed a large‑scale experiment, publicly stating each theory’s predictions and failure conditions in advance. In 2025, the conclusion of this showdown was published in Nature: the posterior cortex did not exhibit the expected sustained synchronized activity, and the prefrontal cortex could only decode partial contents of consciousness, leaving the core predictions of both sides insufficiently supported.

 

On October 16 local time, the team of neuroscientist Doris Tsao at the University of California, Berkeley (UC Berkeley) posted a paper on the preprint server bioRxiv , providing a new answer from a temporal perspective.

 

In their macaque experiments, they found that even when the visual scene and subjective perception remain unchanged, the same group of neurons in the brain switches back and forth between “conscious” and “unconscious” states at a rate of about 15 times per second. The pure consciousness signal that researchers have been searching for over many years is hidden within half of that time.

 

Cao Ying later introduced this work on social media, saying it overturns decades of understanding about the biological basis of consciousness. Previously, consciousness research has tried to find answers on the brain’s map; this work suggests that the answer may also be written on the timeline.

 

(Source: X@doristsao)

 

From Face to Consciousness

 

Cao Ying’s series of previous work laid the foundation for this breakthrough. She earned her undergraduate degree in biology and mathematics at the California Institute of Technology (Caltech), and received a Ph.D. in neuroscience from Harvard University in 2002.

 

In 2024, she was awarded the Kavli Prize in Neuroscience, together with German neuroscientist Winrich Freiwald and others, for discovering the face patch system in the macaque brain and decoding facial neural representations (reconstructing face images from neuronal firing).

 

In recent years, Cao Ying has gradually shifted her focus to the study of consciousness. In 2020, she and Janis Hesse, who was then pursuing a Ph.D. at Caltech, published a no‑report paradigm binocular rivalry study that inferred the images seen by monkeys from their eye movements and pupil changes, eliminating the need for training the animals to make button reports and thereby removing the interference of that action on neural signals.

 

The research also found that face cells in the inferior temporal cortex (the brain region responsible for high-level visual recognition) indeed change with conscious perception, but in single trials both the image that the animal is aware of and the suppressed image can be decoded.

 

This leaves a paradox that has troubled the two for nearly a decade: classic domain narratives claim that there are pure conscious representations in the front of the brain, yet the signals they actually recorded show a weak association with consciousness in every brain region.

 

Cao Ying described the two as “trapped by their own expectations.” In fact, since the 1990s, researchers have begun recording neurons involved in binocular rivalry in monkeys, and people have repeatedly found that neuronal discrimination of conscious perception under competitive conditions is markedly weaker than when the two images are presented separately.

 

Currently, Cao Ying serves as the chief scientist of the neuroscience program at the non-profit research organization Astera Institute, dedicated to understanding the specific mechanisms by which the brain generates conscious experience, cognition, and intelligent behavior.

 

(Source: UC Berkeley)

 

Capture the switch between “conscious” and “unconscious”

 

The new study used the classic binocular flash suppression method: first present an image to only one eye, then suddenly present a second image to the other eye. The newly appearing image almost always “captures” consciousness, causing the previously shown image, although still in view, to “disappear” from perception.

 

Subsequently, the team had two rhesus monkeys view the same stimuli while inserting Neuropixels probes into their two inferior temporal face patches and two prefrontal cortex regions (responsible for decision‑making, planning, and other higher‑order cognitive functions). These probes can capture the spiking activity of hundreds of neurons simultaneously, allowing researchers to track, millisecond by millisecond, the real‑time representation of the entire neuronal population within a single trial.

 

According to traditional methods, if you average neural activity over a period of time, the discrimination strength of the inferior temporal lobe neurons for conscious perception is still only about one quarter of that when the image is presented alone.

 

But this time, the team divided each trial into 10‑millisecond segments and evaluated, segment by segment, whether the current population activity was closer to the pattern associated with “seeing face A” or “seeing face B.”

 

The results surprised the team: when the image and perception are stable, the neuronal population switches back and forth between representations of a seen face and an unseen face approximately every 60 milliseconds. The study labeled these two intervals as “micro‑consciousness periods” and “micro‑suppression periods”.

 

When calculated separately by state, the neuronal discrimination intensity for perception during the conscious micro‑period almost returns to the level when the image is presented alone, whereas the inhibited micro‑period shows the opposite. This reveals that the previously observed “weak signal” was actually the mutual cancellation of two strong signals with opposite directions.

 

More importantly, the study did not find two groups of cells dedicated to “conscious” and “unconscious” processing; the same face‑responsive cells fire sequentially according to the features of two faces. Neural activity during the conscious micro‑period can reconstruct the face the monkey saw; activity during the inhibited micro‑period can reconstruct the face it did not see. After removing statistical artifacts, the conclusion still holds.

 

(Source: bioRxiv)

 

Timing is key

 

The research not only solved a mystery that has persisted for over thirty years, but also provides a new dimension for consciousness studies: time.

 

The introduction mentioned that the two camps locate consciousness signals based on brain regions, but Cao Ying’s team recorded both the prefrontal cortex and the inferior temporal lobe, finding that neither location contained stable consciousness signals; neurons continuously switched between the two faces.

 

This directly undermines the default assumption that “consciousness signals remain stable in a particular brain region.” Previous experiments mostly averaged long‑duration activity, and the temporal resolution of human brain imaging is relatively coarse, possibly failing to detect rapid signal switches occurring every 60 ms.

 

Since the same group of neurons alternately represent two faces, how does the brain distinguish which one to perceive? Logically, one would expect the face that is ultimately perceived to elicit stronger neuronal firing, but the data show that during the non‑perceptual period, the overall firing is actually slightly stronger.

 

The difference between the two states actually lies in the neurons’ sensitivity to facial changes: during the conscious micro‑interval, neuronal firing responses are more pronounced; whereas during the inhibitory micro‑interval, the same group of neurons responds more smoothly.

 

As for what controls this switching and decides which period’s signal is clearer, the author speculates it may arise from interactions among neurons within the visual cortex, or from regulation by other brain regions such as the parietal lobe, thalamus, and so on.

 

In the research, another set of data reveals the complexity of the consciousness mechanism: The two temporal‑lobe face patches recorded simultaneously by the team switch rhythms as if each were keeping its own beat, with very weak correlation. Yet a person can only see one image at any moment; if each brain region operates independently, how this unified perception is formed remains a mystery.

 

Can unconscious images also alter behavior?

 

Cao Ying emphasized in the post that the most exciting part of the entire work for her is that, during the suppressed half of the time, the brain still retains complete and clear facial information.

 

This echoes some classic findings in psychology: a fearful face flashes on the screen, the participant does not notice it, yet the amygdala, which processes emotions, has already responded. Even when people are unaware of what they have seen, it still influences judgment, attention, and emotion. If the brain continuously retains the full content of these “unseen” items, it is not surprising that they can shape behavior.

 

Why does the brain expend effort to retain an image that was never truly “seen”? The author speculates that during the period when the image is not seen, the brain may be silently checking alternative interpretations of the current visual scene without interfering with behavior; during the period when it is seen, the brain carries the interpretation it has adopted at that moment, passing it to the regions responsible for language, reflection, and action.

 

The mechanism whereby the same group of neurons processes different information at different times is not uncommon in the brain. Previous research from Cao Ying’s lab also found that the same group of neurons in the posterior temporal lobe handles the faces currently seen and the memory of familiar faces at different moments.

 

The alternation between consciousness and unconsciousness may simply be another application of the brain’s universal strategy.

 

This work is still at the preprint stage without peer review; the data come from two macaques, and the generalizability of the conclusions requires further experimental validation. The team says they will release the raw data and analysis code when the paper is formally published, for other labs to verify.

 

To further demonstrate a causal relationship between consciousness and this rapid switching, active intervention is required; Cao Ying’s research program at the Astra Institute includes large‑scale manipulation of neural circuits, among other things.

 

In the tweet, Cao Ying exclaimed, finding a signal that richly represents “things you are not aware of” finally provides a foothold for uncovering how neural activity enters consciousness.

 

Reference content:

https://www.biorxiv.org/content/10.64898/2026.10.15.759299v1

https://x.com/doristsao/status/2111318739698561330

https://x.com/HesseJanis/status/2111349628247708127

https://elifesciences.org/articles/58360

https://doi.org/10.1038/s41586-025-08888-1

https://www.nature.com/articles/s41586-024-07349-5

https://www.kavliprize.org/prizes/neuroscience/2024

Note: Cover/featured image generated with AI assistance