> Interesting. I haven't read the paper, but I took the "theory" to be: "when presented with a visualization of one's body, augmented with a signal pulsing synchronously with one's heartbeat, one's perception of self extends to include the visualization.
But that's not a theory, that's an observation. A theory would explain what had been observed, generalize it, make predictions about things not observed, and be falsifiable by new evidence. You know, like in science.
> Can this not be empirically tested?
Observations can only be collected, the term "test" isn't meaningful. Theories can be tested because they presume to explain what's been observed, and there can be any number of possible explanations for a given observation.
Here's an example. If I say, "the night sky is filled with little points of light," I have made an observation, for which the term "test" doesn't apply -- everyone will make the same observation about the night sky. But if I say, "those points are actually thermonuclear furnaces like our sun but at much greater distances," that theory can be tested and potentially falsified -- it meets the requirements for science.
> Can it not be falsified or generate consensus?
Not unless the observation leads to a testable theory, and won't produce consensus unless that theory is confirmed. For example, in some neuroscientific work I was involved in many years ago, we established that people could tell which direction a sound was coming from even if the volume in each ear was artificially changed, meaning the sense of direction of the sound wasn't dependent on it being louder in one ear than the other -- something else was involved.
That's an observation, not very important and certainly not falsifiable. But then neuroscientists crafted a theory that people identified the direction of a sound based on its arrival time, not its volume. Now we had a testable theory, one that we confirmed using signal generators that presented sounds that had been time-delayed but that were exactly the same volume, by way of headphones, and asked subjects to identify the apparent direction of the sound. That study in turn led to a reassessment of how the human hearing system works, and its ability to distinguish arrival times in the millisecond regime, something that at the time was rather controversial.
Put simply, observing things can only lead to science (is not itself science), because (among other things) science requires a basis for falsification, and observations cannot be falsified -- everyone who goes out after dark sees a sky filled with points of light. But the claim that people directionalized sound based on its arrival time was simultaneously testable and initially very questionable because of the time discrimination ability it implied. It turned out to be true, and it was certainly science.
Psychologists famously do not craft theories or establish general principles that might unite their field or produce consensus among different workers, as a result of which psychology drifts from one fad to another and breaks apart into more and more splinter groups over time (psychiatry and clinical psychology are only the best-known of 54 subfields presently recognized within the field).
Symptomatic of psychology's way of doing "science" was the fact that all the new mental illnesses included in the new DSM, recently published, were included based on -- not medical research -- but secret votes. Some conditions that had been voted into the prior version were voted out (including Asperger Syndrome), but not one new condition was included based on medical and clinical evidence that identified a cause, as is required in mainstream medicine. All decisions were carried out in secret, by voting experts. No microscopes, no brain scans, and not one identified cause.
This is all rather academic, because as we speak, emphasis is shifting away from psychiatry and psychology to neuroscience, because the public has gradually become aware of the importance of evidence-based practice in medicine and other fields, a move reinforced by the fact that psychiatry and psychology have such poor evidence to support its clinical activities.
> Still, how is the statistical nature of a sufficiently large sample of humans providing qualial data different from that of a sufficiently large number of H20 molecules providing neutrino collision data?
The difference is that physicists don't bother publishing until they have a theory about why the neutrinos behave the way they do. Also physicists don't have clinics, and they don't try to tell people how sick they are, based on diseases they voted into existence.