On the Emergence of Epiphenomena

Soren Louv-Jansen has created an application that can decide when one of his friends on Facebook is asleep, by monitoring their online posting activity. Apparently, for some of his friends, to be awake is to be on Facebook.

Given this, I have a weird question, but one with a reason behind it, an analogy with current neuroscience and the correlates of consciousness. What can Facebook posting activity tell us about the underlying neuroscience of sleep?

If an alien who knows nothing else of humanity was provided with a stream of the average voltage of the WiFi network signals emitted by a Facebook user, fopr example one of Soren's Facebook friends, which we know corresponds to their Facebook posting activity over time, what could the alien deduce about the poster's brain?

It's my opinion that the alien could decide if the friend's network voltages were rhythmic and might be able to deduce a pattern that matched the Earth's day-night cycle. The alien might be able to tell if the friend's network voltages were typical or atypical compared to a database of many others like him. But the alien might also conclude there was little difference between a human's measured activity and that of a plant, which would have different metabolic phases and thus emit somewhat different electrochemical activity from night to day.

So, what about consciousness seems missing from the averaged Wifi signal amplitudes? Almost everything!

The alien's Wifi voltage amplitudes can be seen as a low level phenomenon (voltage versus time) which is incidentally derived from a very high level phenomenon (Facebook posting by an intelligent person). The electrochemical properties of the human body contibute a lower level basis for high level cognition, which then influences the electrical properties of the Wifi network, which the aliens measure. Certainly the connection between thought and Wifi voltage is highly indirect. Is the connection between thought and fMRI signal really any better?

What we have here is the emergence of epiphenomena, which is analogous mathematically to a "forgetful functor": a process which has a transformed output which loses most of the structure and other information present in the input. Whether measuring BOLD fMRI or correlating evoked potentials, what we get is a limited proxy for the higher properties of consciousness.

Emergent epiphenomena, if used to sieve our data, can be useful when we know that the emergent epiphenomena can function as a reliable filter to remove extraneous higher order phenomena such as anomalous behaviors from our dataset. For example, EEG can usefully distinguish seizure from conversion disorder behaviors. But much of the published neuroscience of consciousness takes highly filtered epiphenomenal data and then tries to extrapolate lost phenomena from limited shadows of the real, higher order data. It's no wonder many such studies are not reproducible, since they measure mostly poor correlates of what they attempt to measure.