In May 2026, a team at Caltech did something no lab had done before: they pressed an ultrasound probe against a transparent window built into a living patient’s skull and watched, in real time, as layered brain rhythms flickered through the cortex during a simple video-game task. The images they captured are the sharpest direct view yet of the oscillations that many neuroscientists now believe bind sensory input to internal thought. Combined with separate MIT experiments showing that the anesthetic propofol severs exactly those cross-layer signals, the findings are narrowing the search for the physical machinery of consciousness to a specific, testable set of cortical rhythms organized by depth.
What the Caltech ultrasound actually showed
The technique is called functional ultrasound imaging, or fUS. Unlike fMRI, which tracks blood oxygenation across the whole brain at relatively coarse resolution, fUS measures rapid changes in blood volume within small cortical regions, producing maps that sit between the pinpoint detail of single-neuron electrodes and the broad strokes of a standard brain scan.
Sumner Norman, a neuroscience researcher at Caltech, and colleagues used the method on a patient who already had an acoustically transparent cranial implant. While the patient played a connect-the-dots game on screen, the probe tracked which cortical zones activated and when, resolving activity patterns at a spatial and temporal scale that conventional noninvasive imaging cannot match.
The foundational work that made this possible stretches back more than a decade. In 2011, a team led by Emilie Macé and Mickael Tanter published a landmark paper in Nature Methods demonstrating that ultrasound could detect transient cerebral blood-volume changes in rodents with enough sensitivity to capture evoked neural responses and even seizure spread. Subsequent preclinical studies, including work archived on PubMed, validated the method against established electrophysiological recordings and provided the safety data that regulators require before human use. The Caltech results represent the payoff of that pipeline: the same physics, now applied inside a conscious human brain during a cognitive task.
The layered rhythm that keeps appearing
What makes the ultrasound data significant is the pattern it can now help interrogate. In a 2023 study published in Nature Neuroscience, André Bastos, Earl Miller, and colleagues at MIT’s Picower Institute for Learning and Memory reported a remarkably consistent finding across 14 cortical areas in multiple primate species: fast gamma-band oscillations (roughly 50 to 150 Hz) dominate the superficial layers of the cortex, while slower alpha-beta oscillations (10 to 30 Hz) dominate the deep layers.
The Picower team interprets this split as a two-lane highway. Superficial gamma carries feedforward signals, shuttling raw sensory data upward from primary sensory areas toward association cortex. Deep-layer alpha-beta carries feedback signals, sending predictions and contextual information back down from higher-order regions. The pattern held across visual, motor, and association cortices, suggesting it is a fundamental organizing principle of cortical computation rather than a quirk of any single brain area.
To extend these findings beyond the handful of sites where depth electrodes had been placed, the Picower team developed a computational tool called FLIP (frequency-based layer identification procedure) that infers laminar sources from surface recordings. That tool opens the door to mapping the layered rhythm across the entire cortex without requiring invasive electrodes at every location.
What anesthesia reveals about the pattern
The sharpest clue about whether these rhythms matter for consciousness comes from propofol. In experiments summarized by the Picower Institute, Miller’s group showed that when subjects go under anesthesia, sensory signals still register in primary sensory cortex. The early feedforward response does not vanish. What collapses is the communication bridge: the long-range synchronization between cortical layers and regions that normally lets a sensory signal propagate, get interpreted, and enter awareness.
Put plainly, the brain under propofol is not silent. It still “sees” and “hears” at the earliest processing stages. But the information never gets stitched into the broader network that makes you aware you are seeing or hearing. If the spectrolaminar motif is the wiring diagram of conscious binding, propofol appears to cut the cables selectively, leaving local circuits running while the integrative loops go dark.
This finding matters because it moves the conversation from correlation to something closer to causation. Plenty of brain-imaging studies have identified neural “correlates” of consciousness, patterns that appear when someone is aware and disappear when they are not. The propofol data goes further by showing a specific mechanism of disruption: it is the cross-layer, cross-region communication that fails, not the raw sensory processing.
The push to image through intact skull bone
One major limitation of the Caltech work is that it required a cranial window. The patient already had one implanted for clinical reasons; the researchers did not drill a hole just to run an experiment. But if fUS is ever going to become a standard bedside tool in intensive-care units or operating rooms, it needs to work through intact bone.
A 2026 paper in IEEE Transactions on Biomedical Engineering (DOI: 10.1109/TBME.2026.3680008) tackles that barrier directly. The authors characterize how skull thickness and angle distort ultrasound beams and test adaptive focusing strategies designed to compensate. The results are promising but preliminary: no team has yet published transcranial fUS images of an intact adult skull that match the resolution achieved through a cranial window. For now, the highest-quality human data still depends on surgical access, and most attempts without it either target neonatal fontanelles (where bone is thin and pliable) or accept substantially lower image quality.
Stimulation: reading the brain vs. writing to it
Imaging is only half the equation. To truly test whether these layered rhythms are necessary for consciousness, rather than just present during it, researchers need to perturb them and observe what happens to a person’s awareness.
Transcranial focused ultrasound (tFUS) offers one path. Separate Caltech experiments have shown that brief ultrasound pulses can measurably shift cortical oscillatory power recorded with EEG, nudging specific frequency bands up or down. A roadmap paper affiliated with MIT outlines how tFUS could be used to causally test theories of consciousness by targeting specific deep brain structures at millimeter scale while monitoring behavioral reports and network responses.
But a critical gap remains: no published study has simultaneously imaged with fUS and stimulated with tFUS in the same conscious human subject during the same session. The two capabilities exist in parallel. Merging them into a closed-loop system that reads cortical rhythms, identifies a target pattern, and then writes a precise perturbation to that pattern remains a stated goal of multiple labs, not a demonstrated achievement. Steering exact phase relationships between superficial gamma and deep alpha-beta across several cortical areas is a far harder engineering problem than nudging overall power in one region, and current neuromodulation results show substantial variability between individuals.
What the competing theories predict
These findings do not land in a vacuum. Two major frameworks have dominated consciousness research for decades, and the new data bears on both.
Global Workspace Theory, developed by Bernard Baars and later formalized by Stanislas Dehaene and Jean-Pierre Changeux, holds that consciousness arises when information is broadcast widely across cortical networks through long-range connections. The propofol data fits neatly here: anesthesia spares local processing but kills the broadcast, which is exactly what the theory would predict.
Integrated Information Theory (IIT), developed by Giulio Tononi, focuses instead on the intrinsic causal structure of a system, quantified by a measure called phi. IIT predicts that consciousness resides in the “posterior hot zone” of the cortex and depends on the degree to which a system’s parts are both differentiated and integrated. The spectrolaminar motif, with its distinct but interacting superficial and deep rhythms, could be read as consistent with IIT’s emphasis on structured integration, though IIT proponents would argue that oscillatory patterns alone do not capture the full picture.
Neither theory has been definitively confirmed or ruled out. But the convergence of depth-resolved imaging, laminar electrophysiology, and anesthesia pharmacology is generating the kind of data that could, in principle, distinguish between them. If disrupting specific cross-layer rhythms reliably toggles awareness while leaving other measures of integration intact, that would favor workspace-style theories. If awareness tracks more closely with a global measure of integrated information regardless of which specific oscillations are involved, IIT gains ground.
Where the evidence stands in June 2026
Three tiers of evidence support the central claim, and keeping them distinct matters for anyone following this field.
The first tier is direct experimental data: the Caltech fUS-through-cranial-window recordings, the Picower Institute’s spectrolaminar motif across primate cortex, and the MIT propofol experiments. Each involves controlled conditions, peer review, and quantified results. These are the load-bearing pillars. They establish that cortical activity can be imaged at high resolution in humans, that oscillations are systematically organized by depth, and that anesthesia disrupts long-range communication while sparing early sensory responses.
The second tier is engineering progress. The 2026 IEEE transcranial paper and the Caltech tFUS-EEG work represent real technical advances, but they document capability, not consciousness findings. They show the tools are approaching the precision needed to test binding theories directly. They do not show those tests have been run.
The third tier is interpretive framing. Press releases from MIT and Caltech describe what these tools could eventually prove about consciousness, drawing on roadmap papers that outline testable questions. When an institutional release says a tool “could tell us how consciousness works,” the operative word is “could.” The underlying peer-reviewed articles typically make more modest claims about signal quality, reproducibility, and specific mechanistic hypotheses.
The practical upshot: the field has moved from asking whether consciousness has measurable neural signatures to asking which specific patterns of layered oscillations and cross-region communication are necessary for conscious perception. Functional ultrasound through cranial windows, laminar electrophysiology, and anesthesia pharmacology now converge on a picture in which superficial gamma and deep alpha-beta rhythms coordinate the flow of information between sensory input and higher-order interpretation. The next phase will test whether deliberately perturbing those rhythms with tools like tFUS can reliably toggle awareness of the same stimulus on and off in human subjects. Until those causal experiments are published, the most defensible position is cautious optimism: the physical basis of consciousness appears increasingly constrained, but not yet conclusively pinned down.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
