Moving together is
the beginning of a case.
Imagine two dancers who always move together. That tells us something real about their timing, but it does not make them the same dancer. The studies below test related movement in signals; the cartoon shows why that is a beginning rather than the whole claim.
Do the studies prove Temple accurately measures brain blood flow?
The two September 2026 preprints examined here report correlations between Temple’s index and reference signals during specific physiological challenges. They support studying whether the signals track related changes, but do not establish absolute cerebral blood-flow accuracy, diagnosis or longer life. The studies are unreviewed versions; correlation, agreement and clinical usefulness require different evidence. Mira’s two dancers make that distinction visible before the equation.[5][6]
While Mira waits, consider the evidence behind the index rather than the fictional reading on her wrist or temple. A September preprint by Kumar and colleagues compared Temple’s Brain Flow with middle cerebral artery velocity measured using transcranial Doppler. Twenty-three adults completed the programme, with 20 analysed per protocol. Median within-person correlations were 0.795 for exercise and 0.799 for posture transitions. The standardised comparisons concern relative responses, rather than agreement in absolute flow units.[5]
Two controlled challenges
Median within-person r
Median within-person r
A second preprint, by Gulati and colleagues, compared Temple with brain-assigned and scalp-assigned changes in oxygenated haemoglobin, recovered using a two-layer time-domain NIRS model. It analysed 60 sessions from 44 healthy young men. The chart below preserves all three conditions, including the one where the paired brain-versus-scalp difference did not reach statistical significance. This is a comparison with model-derived haemodynamics, rather than a direct whole-brain flow measurement.[6]
Deeper and superficial signals
Both studies offer preliminary evidence of covariation under specific challenges. Their scope leaves cerebral specificity, generalisation to everyday conditions and clinical benefit unresolved. The first discloses Temple funding and a paid consultant; the second discloses company conduct and employee authorship. Those relationships matter when weighing evidence, without making the results disappear. Independent replication would add something these comparisons cannot supply on their own.[5][6]
“We’re in sync. Why aren’t we equal?”
Unequal values
Correlation rewards moving together. Agreement asks whether the readings match in the relevant units and within acceptable error.
Open the equation · What does correlation compare?
Correlation describes linear covariation when both variables have nonzero variance. For B = aD + b and a > 0, r = 1 even when the scale and offset differ. Agreement requires additional questions about units, bias and error.
Perfect correlation.
Different readings.
Temple’s website also describes Entropy as a measure of the body’s effort to shed heat, and Span as a combination of resting and maximum Entropy. These are product interpretations requiring their own validation; neither follows automatically from Brain Flow correlations. The shared name does not, by itself, establish equivalence between a proprietary score and thermodynamic entropy in physical units.[7]
Mira’s menu offers the same warning in a less technical form. Two restaurants can increase their prices in perfect proportion while serving very different food. The relation between the numbers is real, but its existence does not settle the question the diner cares about. A strong correlation helps build a case for a measurement. The case still has to name the claim it is building.