Does Temple Really Measure Brain Blood Flow? Its Studies Never Tested Heat, Where Scalp and Brain Part Ways

Press two fingers just in front of your ear and you will feel a pulse. That is the superficial temporal artery, and its blood is headed for your scalp, not your brain. It is a branch of the external carotid artery, which feeds the scalp and face. Your brain is supplied by a separate set of arteries running deeper, beneath the scalp, a layer of chewing muscle and the skull.

Temple, the ₹85,000 wearable from former Zomato chief executive Deepinder Goyal, sits on exactly that spot and reports a number called Brain Flow. Some doctors have called it a toy, a scam and a “Desi Theranos”; others have called it an encouraging concept to approach with caution. Goyal has answered that a scam would be found out within days [1][2][3].

Both sides are arguing past the evidence. Physiologists already know when blood flow to the scalp and to the brain go separate ways, which gives a simple test for any sensor on the temple. We read both of Temple’s validation studies in full to see whether it was run. It was not, and the condition it skipped is the one Temple’s new headline score is built on.

What is Temple, and what does it claim to measure?

Temple is a small optical sensor worn on the side of the forehead, made by Temple Private Limited in Gurugram and funded with $54 million raised in February 2026 [3][4]. Goyal says $20 million of the $30 million spent so far is his own money [1]. It is sold as a fitness device, with no FDA clearance and no approval from India’s drug regulator, CDSCO [3].

It combines reflectance photoplethysmography (PPG) at green, red and infrared wavelengths with heart rate, an accelerometer and temperature sensing [5], and a proprietary algorithm turns these into two numbers:

• Brain Flow, a relative index. Temple says it “does not claim to directly measure deep brain blood flow”: it is a “scalp proxy” that correlates with brain blood flow under the conditions studied [3][6].
• Entropy, “how hard your body is working to shed heat”, on a scale of 1 to 250. Temple says its resting value reflects load across nine systems, from immune to digestive to cognitive [2][7].

Goyal now calls Entropy the main feature, but no study of it has been published [2]. Both preprints test Brain Flow.

Can a sensor on your temple see blood flow to the brain?

In principle, partly. In practice, it is one of the hardest problems in optical sensing, and the obstacle is everything between the sensor and the brain.

Light from a sensor on the head has to pass through scalp and skull and come back out to reach a detector. Averaged around the adult head, where scalp is about 7 mm thick and skull about 6 mm, a simulation study found that a near-infrared sensor with its light source and detector 20 mm apart sent only about 6% of its light path through grey matter, rising to just over 19% at 50 mm [8]. At the temple there is also a layer of chewing muscle between scalp and skull. Temple has not disclosed its sensor geometry [6].

Even medical cerebral oximeters, built for exactly this job, pick up the scalp. When researchers cut off scalp blood flow with an inflatable cuff around the head, three clinical devices’ readings of brain oxygenation fell by 7% to 17%, although the cuff affected only the scalp [9].

So any head-worn optical sensor starts out mostly as a scalp sensor. Temple’s claim is that its multi-wavelength composite pulls out the deeper signal. That is possible: research systems separate the layers using the time light takes to travel through tissue. The question is how to check it.

When do scalp and brain blood flow disagree?

Most of the time, they move together. Lie down and gravity fills the whole head; stand up and it drains. Pedal at a moderate pace and flow rises to both. A sensor that read only the scalp would still correlate with the brain in all of these.

The way to tell the two apart is to test the conditions where they split, and physiologists have measured three of them with ultrasound in the neck.

• Harder exercise. In graded cycling, blood flow to the brain rose by about a fifth at moderate effort, then fell back almost to resting levels at 80% of peak oxygen uptake. Flow to the scalp and face kept climbing, to almost double its resting level, and forehead skin blood flow, measured as vascular conductance, almost tripled [10].
• Heat. When 12 people were heated at rest until their core temperature rose 1.4°C, blood flow to the brain fell by 13% to 17% while flow to the scalp and face rose 135% [11]. During exercise in the heat, the bigger the rise in scalp flow, the bigger the fall in brain flow [12].
• Breathing. Blood flow to the brain is highly sensitive to carbon dioxide, which is why hyperventilating makes you lightheaded. The external carotid’s response to CO2 is about a fifth of the brain’s [13].

That gives a one-line test for any sensor that claims to read the brain from the temple: does it follow the brain when the two split? So what did Temple test?

What did Temple’s two validation studies show?

Participants in Temple’s studies tilted head-down, squatted, moved from standing to lying down and cycled fast for four minutes. These are the conditions where scalp and brain blood flow move together. Neither study included a heat condition, and neither measured end-tidal CO2, the breathing signal that drives brain blood flow [5][6].

Study 1: against transcranial Doppler [6]. Researchers at Chaudhary Charan Singh University in Meerut and the Salk Institute compared the Brain Flow index with ultrasound-measured blood velocity in the middle cerebral artery. Twenty-three healthy adults aged 18 to 30 completed both tasks, with 20 analysed per task. The median within-person correlation was 0.80 for both cycling and the posture change.

• What it shows. The index moved with brain blood velocity. During the posture change heart rate went the opposite way and the index followed the brain, so it is not simply tracking heart rate.
• What it does not. Both signals were standardised within each session, so it compares the shape of the change, not agreement in any unit. The devices sat on opposite temples, and the cycling intensity was not reported. The authors write that the posture result “does not by itself establish a specifically cerebral contribution”. Temple funded the work, and one author is a paid consultant to the company.

Study 2: against time-domain NIRS [5]. This is the stronger study. Its co-authors include David Boas of Boston University, one of the best-known researchers in brain optical imaging, who serves on Temple’s scientific advisory committee and whose lab receives Temple funding. Its reference estimates scalp and brain blood volume separately, across 60 sessions from 44 healthy men aged 20 to 32.

The index tracked the modelled brain layer better than the scalp layer in all three tests, significantly so for head-down tilt and squatting but not for lying down. The scalp was never far away: correlations with the scalp alone ran as high as 0.77, just as the physiology predicts for tests where the two move together. After adjusting for a separate scalp measurement, the brain correlations fell to between 0.44 and 0.65. The study was designed, funded and run by Temple, its founder is an author, and its own conclusion lists absolute flow accuracy and “cerebral specificity at the temple” as still to be established [5].

That is a respectable early result for a new sensor, honestly reported, and posture tests are a normal place for a first validation to start. It answers whether Brain Flow moves with the brain when everything moves together. It does not answer whether it follows the brain when the scalp goes the other way.

Why does Entropy make heat the question that matters?

Picture a fast bowler in an afternoon net session at 38°C. Their body is shedding heat as fast as it can by pushing blood into the skin, and the forehead is one of the places it goes: in hard exercise, forehead skin blood flow almost triples [10]. Entropy, which Temple defines as how hard the body is working to shed heat, should be climbing, and a sensor on the temple is well placed to see it [7]. That may be a genuinely useful signal for athletes training in the heat.

Now look at the same moment through Brain Flow. The blood under the sensor is surging. The bowler’s brain supply may be flat or falling, since hard effort pulls it back towards resting levels and heat lowers it [10][11][12]. If Brain Flow rises in that moment, either the scalp is driving it or Temple has found something the ultrasound literature has not.

Nobody outside Temple knows which. It is the exact situation Temple’s two studies did not test, the one its headline score is about, and a standard experiment to run.

Is Temple a scam?

Nothing published supports that. Theranos concealed that its tests did not work. Temple has posted its study methods, its funding and its limitations in public, one study has a leading scientist in the field as a co-author, and buyers can return the device [1]. Goyal also has a fair point that wearables are often validated after launch, citing WHOOP’s years between shipping and its first published validation, and says he expects independent studies once the device is in people’s hands [1].

The fair criticism is narrower and harder to answer. What Temple has shown is a relative signal that moves with estimates of brain blood flow and volume, in young, healthy people, during posture changes and short exercise. Temple’s careful statements say about that much: a scalp proxy that correlates with brain flow in the conditions tested. But the product is named Brain Flow, sold by a founder who calls it an experimental device to calculate brain flow accurately, alongside a single score for the load on nine body systems [3]. The gap between the careful claim and the name is where the argument should be, and the studies to close it have not been published.

What would show whether Temple measures brain blood flow?

1. Heat stress, with duplex ultrasound of the internal and external carotid arteries running alongside.
2. CO2 challenges, breathing CO2 and hyperventilating, with end-tidal CO2 measured. Brain flow should swing sharply; the scalp should barely move.
3. Scalp occlusion, using the head-cuff method from the cerebral oximeter studies [9]. If Brain Flow drops when only scalp blood is cut off, the scalp is in the signal.
4. Independent replication by a lab with no funding from Temple, in women and older adults, with repeat visits to measure reliability.
5. A published Entropy study that shows what moves it and what a change means.

How should product teams judge a new wearable metric?

Every year brings new sensors and new proprietary scores. Before you build features, coaching or care decisions on one, ask five questions. Here is how Temple answers them today.

A sensible rule: until a new metric clears the third and fourth questions, treat it as a feature of the device’s own app, not an input to your scoring, triage or care logic. The same questions sort the evidence for cuffless blood pressure and smartwatch blood oxygen, and they apply to health scores too, as we covered in validating health scores.

Temple may turn out to be a good scalp perfusion and heat-strain sensor sold under a brain-flow name, or the first wearable that really separates the two. The studies needed to tell which use standard, well-established methods. Until they are published, the honest description of Brain Flow is the one Temple’s own researchers wrote: “a relative marker of cerebral hemodynamic change during postural perturbations”, tested so far in young, healthy people.