The setup confounds chemistry and physics on side (transport of molecules by «normal» air flow because a door/window, sash of a hood is open; convection by a radiator, diffusion by concentration gradient) on one side, with psychology on the other. The former can be monitored by organisms, and machines.
The later may be more complex. Over time, we are primed for instance by the taste of a lemon, can develop a response expectancy about the scent associated by other stimuli e.g., sound and sight. Say you were a sparring partner for the uneven bars; after some training together, don't you equally move while the other performs the exercise -- not only for safety to promptly catch him/her if necessary? If you usually play an instrument and go to a concert, knowing some of the passages by heart, isn't your brain equally similarly engaged except for the actual play of the instrument, too?
Placebos play a role in medical research -- even to the degree that there are active placebos intentionally containing ingredients e.g., to cause dizziness in order to prevent de-blinding medical studies because patients can be conditioned to the presence of typical side effects the verum generates -- by the talk with the physician, the leaflet in the package, hearsay. Why else running blinded and even double-blinded studies where neither patient, nor physician know the content of the box until the end? Their role is even more prominent (or better, more important) in treatments which are less (or not) evidence based.
And group dynamics like «well if my dear neighbor feels sickened...» You equally might repeat the experiment, unveil a beautiful container of tap water «I'm so glad to present you a new, exquisite parfum; the first to sense it, gets it for free».