Saturday, October 26, 2013

Scanning for consciousness

Can technology tell us whether someone is conscious or not? Just barely.

How does the brain work? What causes or is that most basic phenomenon- consciousness? Many theists and philosophers dispair of ever finding an answer, or indeed of being able to properly pose the question, calling it the "hard" problem. Our intuitions are perhaps too strong to overcome this sense of magical awe, yet materialists plug along, going with the logical indications from evolution and biology that something very physical is going on in there to mount the drama that flits across our inner stage.

Functional MRI is regular MRI abetted by analysis of blood flow, which responds on a few-second time scale to changes in local brain activity, the brain being a big gas hog, as it were. One would think that with such technology in hand, it would a snap to detect the physical correlates of consciousness and describe all the patterns surrounding it. But no- the brain runs all the time, and the differences in blood flow under activity are very small. Also, the time scale of the key brain activities, like most brain waves, are far faster and spatially far smaller than what fMRI can detect, so it remains, sadly, an extremely blunt instrument.

A recent study looked at twelve volunteers as they went under with the anaesthetic propofol, of Michael Jackson fame. I doubt that propofol-induced unconsciousness resembles sleep very much, so while it may knock you out, it can hardly be the way to a refreshing wake-up the next day. Another study in 2011 , incidentally, did very similar work and came to the same conclusions, and also provides the rationale for using propofol in particular: "The reason why propofol was chosen for this study is that this particular anesthetic has been shown not to interfere with regional cerebral blood flow response at sedative concentrations, and does not modify flow-metabolism coupling in humans".

The researchers tried to measure brain activity in the broadest possible way, tracking correlations among far-flung areas. The upshot is that as sedation becomes deeper, even though over-all blood flow does not change as noted above, correlations among brain activities become increasingly local, losing their long-range character. Which is certainly in line with the general ideas in the scientific community about what consciousness is in physical term: large, wide-ranging, and constantly varying coalitions or patterns of neuronal activity, which are coherent in some sense. This coherence would represent thought to the experiencer, and detectable statistical correlations to the onlooker (inlooker?).

A map of the parcels used by the experimenters to divide up the brains of their subjects into regions of interest (ROI), in order to draw inter-regional activity correlations.

How can these correlations be drawn? "In our analysis the connection is the Pearson correlation  statistic between each pair of nodes." So, despite the crude time scale, they assumed that time-coincident activitions in different locations of the brain reflect functional connection, i.e. communication. They parcelled their brains out into 194 small regions, (using someone else's scheme from prior work), and then computed the average time course of activity within each parcel. Then using statistical methods, one can make a matrix of the correlations among all these time courses and parcels, into the figure below:

Region-to-region matrix of correlations under various conditions: W, waking; S, sedated; LOC, loss of consciousness, and R, recovery of consciousness (to Ramsay level 2).

Clearly, the condition of complete anesthesia (LOC) can be picked out as having sharply reduced connections between different regions, while even just after recovery, connections remain significantly impaired. "As expected, we found a significant effect of condition (... ), indicating that correlation strength systematically varied across conditions. Specifically, W consistently exhibited the strongest average correlation level, across all bins, followed by S and R, while LOC consistently exhibited the weakest average correlation across all bins." 

This result is stated more simply in a graph of correlation to distance apart:


The conclusion they  draw from this is that  the correlation at long distances are not specially impaired relative to that at medium distances. Connections at most distances are impaired, which would, however, naturally decimate long-range communication.

Meanwhile, within the individual regions, some showed increased activity (yellow) and some decreased (blue), consistent with the idea that the long-range effects are dominant overall.

Activity within nodes (also called regions, or regions of interest, ROI) at different levels of anesthesia. Yellow denotes higher activity in the sedated or unconscious states, while blue denotes higher activity in waking or recovery.

Let me wrap up with a couple more quotes from the paper:
"... we find that loss of consciousness is marked by an increase in normalized clustering (), which measures the ‘cliquishness’ of brain regions, potentially indicating an increase in localized processing and thus a decrease of information integration across the brain." 
"... our graph theoretic analysis further indicates that, in terms of network information processing, propofol-induced loss of consciousness is marked by a specific change in the quality of information exchange (i.e., decreased efficiency) ..."

So it remains extremely difficult to differentiate consciousness from living unconsciousness. This is very early days in the  decipherment of brain patterns, and we are far from having tricorders. But there really is something in there to peek at, and one gets the sense that yet more philosophical conundra will eventually be dissolving in this pool of data. Next week, another post on brain science, from a far loftier perspective- that of Douglas Hofstadter.


"The inconvenient facts that the senior officers of JPMorgan, Bear Stearns (Bear), and Washington Mutual’s (WaMu) grew wealthy through the frauds that drove the financial crisis and that JPMorgan’s senior officers will not be prosecuted and will not even have to repay the proceeds of their crimes never appear in the article."
...
"The CEOs’ paramount strategic objective is to prevent real investigations staffed by vigorous financial regulators working with FBI agents that lead to hundreds of grand jury investigations of elite bankers and civil suits, enforcement actions, and prosecutions that make public the facts about the elite frauds that drove the crisis. ... Fraud is criminal even if Holder is too spineless to prosecute it."