Why Mechanism Is Not Proof
A personal essay, not an evidence assessment. Our graded assessments of individual ingredients — written to a published standard, from full texts — are in the evidence section. Nothing here is a statement about what any product does.
I love mechanisms.
This is probably why I distrust them so much.
A beautiful mechanism can make an ingredient almost impossible not to believe in.
It increases this neurotransmitter.
Activates that receptor.
Supports this metabolic pathway.
Crosses the blood-brain barrier.
Participates in membrane synthesis.
A convincing chain begins to form.
A leads to B.
B affects C.
C is associated with cognition.
Therefore:
A improves cognition.
Except that the final arrow may never have been demonstrated.
Biological plausibility is valuable
I do not want to minimise it.
Mechanisms help us generate hypotheses.
They help explain observed effects.
They guide dose selection.
They help identify interactions and safety concerns.
They can tell us why a compound deserves human research.
But:
deserves research is not the same as:
deserves a consumer claim.
Complex systems resist simple arrows
Human biology is full of feedback loops.
Increase one substrate and the system may compensate elsewhere.
Alter one receptor and downstream effects may depend on baseline status.
Change a biomarker and the organism may preserve function almost perfectly.
A pathway can be real.
The predicted outcome can still fail.
“Crosses the blood-brain barrier” is not a cognitive
endpoint
I see this phrase constantly.
It sounds impressive because entry into the brain feels like proof of relevance.
But reaching the brain tells us location.
Not benefit.
The next questions remain:
What does it do there?
At the studied concentration?
For how long?
And does any measurable human outcome improve?
Mechanism can explain an effect after the fact
This is where it becomes useful.
Suppose rigorous human trials repeatedly show an effect.
Mechanistic research can then help us understand why.
The combination of:
observed outcome plus:
credible biological mechanism
is much stronger than either one alone.
But the direction matters.
I do not want to convert every credible mechanism into an assumed outcome.
Marketing loves mechanistic certainty
A molecular diagram is persuasive.
Arrows look causal.
Receptors look precise.
A pathway image can give a weak evidence base the appearance of engineering.
This is why I think evidence pages should label mechanistic evidence visibly.
Not bury it.
Label it:
MECHANISTIC / PRECLINICAL Then separately:
DIRECT HUMAN OUTCOME EVIDENCE
The question I now ask
Whenever I read:
“Compound X increases Y, which is important for Z…”
I pause at the comma.
Everything after that comma may be an inference.
Sometimes a very reasonable inference.
But still an inference.
And the supplement industry has spent too many years selling the final arrow in a mechanistic diagram as if somebody measured it in a human being.
A mechanism tells me why something might work.
A good human outcome study tells me whether it actually did.
I want both.
But I do not confuse them.