This paper demonstrates quite convincingly, in my view, a case of classical conditioning in Paramecium caudatum (although I find that it is not a very easy read). Paramecia are mechanically stimulated by a speaker below the slide for 4 s (playing a 300-500 Hz tone). In the last 2 s, an electrical shock is applied, which induces an avoiding reaction. This is repeated for 10 trials, and after a few trials paramecia start to exhibit an avoiding reaction before the electrical shock. This does not happen if the two stimuli are not paired. The behavior is erased by repeated presentation of the vibratory stimulus alone (extinction). The authors also show evidence of long-term memory and specificity in learning, namely, that paramecia can be trained specifically to 300 or 500 Hz tones. Overall, the phenomenon seems genuine.

One issue is the vibratory stimulus. Can paramecia hear? Of course, they are very sensitive to mechanical stimuli. But in water, a 500 Hz tone is a pressure wave of wavelength about 6 meters. It seems surprising that a 250 µm organism can sense it. It is plausible that the stimulation apparatus (speaker against the slide) produces nonlinearities (i.e., percussion of the slide), or that Paramecium is sensitive mainly to the onset of the sound. In any case, it would be worth examining the sensitivity of Paramecium to vibrations.