First, note that footnote 1 of the Ashihara paper quotes another study that found much lower levels of jitter to be audible.
Second, Meridian's Bob Stuart, in an interview in the current issue of The Absolute Sound, described some hazards of ABX testing that may be relevant to the Ashihara study. It is especially fascinating that Bob Stuart faults the traditional scientific ABX method because he is a scientist with many published papers.
Bob Stuart said he finds ABX comparisons unreliable because: (1) the brain responds to music in subtle ways that are disturbed by the pressure of an ABX test; (2) long listening sessions are required for the listener to appreciate differences because it takes time for the brain to acclimate to a new listening "environment". (The Ashihara paper said the subjects controlled when and how often they switched, but the paper did not report the typical time duration chosen by the subjects. They may have switched too fast knowing they had to undergo several iterations.)
Conversely, Bob Stuart also discussed hazards of long-term comparisons, i.e., the alternative to ABX tests. In long-term comparisons, successive playings of the same musical passage are not equivalent because your brain interprets each repetition based on its memory of the previous repetitions. You typically hear things the second time you didn't notice the first time. Importantly, he says that if you hear a low resolution component, then a higher resolution component, then return to the low res component, the experience of hearing details on the high res component will enable you to perceive those details on the low res component that you missed the first time. This can lead you to erroneously conclude that the low res component is just as good.
Bob Stuart also said that after listening to a passage a certain number times (he didn't suggest how many), it becomes useless and you need to move on to other music.
