TLDR
This episode discusses a groundbreaking 2026 preprint from Doris Tsao's lab showing that consciousness does not reside in a specific brain area but in a temporal rhythm. Using binocular flash suppression and Neuropixels recordings in macaques, they found the same neural population in inferotemporal and prefrontal cortex switches between representing the perceived and suppressed image every ~60ms. This 'temporal multiplexing' explains why decades of averaging produced weak signals and challenges both Global Neuronal Workspace and Integrated Information theories, which assume a stable spatial locus for consciousness.
For you
PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.
Sign in to unlockSurprising Insights
The 'weak' NCC signal reported for 30+ years was an artifact of temporal averaging — the true signal is strong but time-limited.
The suppressed (unconscious) image is represented just as clearly as the conscious one, just in alternating time windows.
Prefrontal cortex — the supposed 'seat' of conscious access in GNWT — shows the exact same 60ms switching as visual cortex, with no stable 'ignition'.
Two adjacent face patches in IT cortex switch independently, yet perception remains unified — how the brain binds these asynchronous rhythms is a mystery.
Doris Tsao says the unconscious half of the cycle is more exciting: the brain 'secretly' holds a perfect copy of what you didn't see.
Key takeaways
Conscious perception correlates with brief (~60ms) 'conscious micro-epochs' where neural tuning is sharp; the suppressed image is represented in alternating 'suppressed micro-epochs'.
The same neurons switch tuning every ~60ms — there are no separate 'conscious' vs 'unconscious' neuron populations.
Traditional trial-averaging cancels out these opposing strong signals, creating the illusion of a weak neural correlate of consciousness (NCC).
This temporal coding occurs in both high-level visual cortex (IT face patches) and prefrontal cortex, undermining the core spatial predictions of both GNWT and IIT.
The brain maintains a rich, decodable representation of the *unseen* image during suppressed epochs, suggesting unconscious processing is far more complete than thought.
Tsao hypothesizes this time-multiplexing allows the brain to test alternative interpretations in the background without disrupting current behavior.
The 60ms switching rhythm (~15 Hz) is not a fixed 'frame rate' of consciousness; the mechanism controlling the switch remains unknown.
The study is a preprint (bioRxiv) from two monkeys; causal manipulation is needed to prove the switching *causes* perception.
Notable quotes
“关键在于 图像本身从头到尾没有变过 变的是你的感知”
▸ Notable quotes · 0:15“他们在猕猴实验中发现 即便画面不变、猴子的主观感知也不变 大脑中同一群神经元也会以每秒约15次的速度 在'有意识'和'无意识'这两种表征之间来回切换”
▸ Notable quotes · 3:02“神经元群体并不是一直表征着你看到的那张脸 而是每隔约60毫秒 就在'表征看见的脸'和'表征没看见的脸'之间 来回切换一次”
▸ Notable quotes · 9:11“同一群神经元 在不同时刻分别代表着你意识到的东西和你没意识到的东西”
▸ Notable quotes · 10:02“意识感知并不居住在大脑的某个特定位置 而是居住在一个特定的时间相位里”
▸ Notable quotes · 11:05“也许问题本身就问错了方向 答案不一定写在哪块脑区上 也可能写在时间轴上”
▸ Notable quotes · 14:50Chapters
0:00 — Opening: The Rubin's Vase Illusion: Host Da Fei introduces the Rubin's vase illusion to illustrate how perception flips while the image stays the same, framing the core question: where and how does the brain select what we consciously see?
0:20 — The NCC Debate: Global Workspace vs. Integrated Information: Reviews the two dominant theories: GNWT (prefrontal 'broadcast' ignition) and IIT (posterior 'integrated' structure). Notes that experimental methods often bias results toward the theory being tested.
1:40 — Adversarial Collaboration Fails to Crown a Winner: Describes the 2025 Nature adversarial collaboration (256 subjects, fMRI/MEG/ECoG) where both theories' core predictions — sustained posterior synchrony for IIT, prefrontal ignition for GNWT — lacked sufficient support.
2:47 — Tsao's 2026 Discovery: Consciousness in Time, Not Space: Introduces Doris Tsao's bioRxiv preprint (Oct 2026) which shifts the search from spatial location to temporal dynamics. Tsao and first author Janis Hesse call it a completely unexpected finding that overturns decades of assumptions.
4:57 — From Face Patches to Consciousness: The No-Report Paradigm: Details Tsao's background (face patch system, neural face code, Kavli Prize) and the 2020 no-report paradigm using eye-tracking to infer perception without motor confounds, revealing IT neurons encode consciousness without report.
8:32 — Core Finding: Neurons Switch Every ~60ms Between Conscious/Unconscious States: Using binocular flash suppression and Neuropixels in IT and PFC, single-trial 10ms binning reveals population activity alternates between 'conscious micro-epochs' (sharp tuning to seen face) and 'suppressed micro-epochs' (sharp tuning to unseen face) every ~60ms. Averaging cancels them out.
10:10 — Implications: Consciousness Resides in Temporal Phase, Not Brain Area: The switching occurs in both IT and PFC, refuting the shared spatial assumption of GNWT and IIT. Consciousness is a temporal phase; the 'where' question was misguided. The unconscious representation is fully intact and decodable.
13:00 — Caveats, Time Multiplexing, and the Next Big Question: Notes preprint status, two-monkey sample, correlation-not-causation. Links to Tsao's 2024 finding of time-multiplexing for perception vs. memory. Raises new mysteries: what controls the 60ms switch? How do asynchronous patches unify perception? Tsao's Astera Institute plans causal circuit manipulations.
Speakers
大飞 (Da Fei) — Host