What the evidence shows
Conditioning experiments demonstrate a capacity the nervous system has after a real physiological challenge; they do not establish the origin of MCS. In a 1999 experiment, healthy participants inhaled an odor paired with CO2-enriched air that produced respiratory symptoms. Later, the odor alone produced more symptoms than a comparison odor. Subsequent work showed that learned symptoms could generalize to new odors and that stronger respiratory challenges could make extinction more difficult.
These experiments are scientifically useful because they show a real physiological learning process. They do not require a participant to consciously decide that an odor is dangerous. Expectancy and awareness can matter, but associative learning can operate automatically.
The limitation is equally important: producing conditioned symptoms in healthy volunteers is not the same as demonstrating the origin of MCS. The experiments begin by creating an unpleasant physiological event and pairing it with an odor. In that sense they can also be read as a model of what might happen after a genuine initiating event: a biological reaction occurs first, and environmental cues become linked to it.
A 1999 review by Siegel made this point explicitly by proposing that conditioning may contribute to some MCS cases when environmental cues present during a toxicant exposure later evoke conditional responses. That model is compatible with environmental initiation followed by learned amplification; it does not require choosing between "biological" and "conditioned" as mutually exclusive categories.
This is also why retraining testimonials cannot settle causation. If conditioning became one layer of a larger illness, reducing that layer could improve symptoms or function without proving that conditioning created the underlying susceptibility. Conversely, failure of retraining does not prove conditioning plays no role at all.