Neuroimaging, morality, and other fallacies
Epistimology should never dictate metaphysics.
1. Introduction
A few years ago a happily married school teacher with an exemplary record suddenly began to sexually molest children [51]. He was arrested, found guilty, and remanded for sentencing. While awaiting final arraignment, he complained of an excruciating headache and was sent to hospital for examination. An MRI revealed a brain tumour in the right lobe of the orbifrontal cortex. This is an area thought to affect judgement, impulse control and social behaviour [50][52]. When the tumour was removed, all paedophilia urges vanished and the man returned to completely normal behaviour. In their paper on the issue [1], the neurologists involved speculated as to the obvious broader legal implications. Perhaps as others at the time suggested, the courts should consider extreme physiologic underpinnings (such as a tumour) to apparent moral aberration before reaching decisions.
But it turns out that behavioural change can occur from even mild brain trauma: A cleric experienced TBI (traumatic brain injury) when he fell against a stair railing [53]. The fall left no visible marks, not even a scratch. Yet a few days later and for the first time in his life, he began to swear at people. After a few weeks, he could not stop swearing - every word was foul. This is not unusual [59]. Those with even apparently mild TBI are known to sometimes express strong behavioural changes. During a TBI there can be neurological bruising which does not show in an MRI but none the less damages neural tissue. Sometimes this is also accompanied by one or more brief strokes (TIA - transient ischemic attacks) which cause scaring (which may or may not show in an MRI). However when the cleric sang, the content of what he expressed was normal. Since the right frontal lobe is thought to mediate a host of behaviours, amongst them singing and swearing, one may assume that when he fell he suffered injury to this area either directly through bruising or additionally through mild TIA. Needless to say, he was unable to pursue his profession and his personal life was shattered.
There are many available examples in both the literature and apocrypha (see [54][55] for a brief summary). The conclusion is that even mild brain damage may lead to radically different behaviours. The extrapolation therefore is that behaviour is brain site and moreover specific locus, mediated.
2. Experiments
Now a few experiments before getting into the neuroimaging debate:
2.1 Illusions
In a well known experiment, subjects were asked to choose one pair of pantyhose in a row of several pairs . Asked to elaborate their choices, subjects quite reasonably explained that they chose depending upon preference for colour, elasticity, knit, smoothness, looks, etc. As you may have guessed however, all the pantyhose were identical. But there was a surprising consistency to the choices - most subjects selected pantyhose from the right hand side of the row [8]. That is to say, their choices were really based upon spatial cues - not the ones they stated and believed. Since then, many similar experiments have been done, all confirming preference based on introspectively unknown cues.
This finding is illustrative of ’introspective illusion’. It is said to occur when subjects unintentionally confabulate mental processes to justify choice.
Introspective illusions have applicability to studies in game theory, risk taking, and ethical decision making (see [49]). They can be extended to describe acculturation in which choice is based upon absorbing messages completely, whilst automatically blinding the the chooser to her real reasons for future choices. For example a racist may kill someone with different skin pigmentation justifying his actions as ’preserving our way of life’, whilst the real reason is that he has been acculturated into a particular belief system and Weltanschuung (world view).
There are also experiments in the related area of ’choice blindness’. One such has subjects choose between one of two photographs. They are then asked to justify their choice. Subjects do so without difficulty. And justify their choices on rational grounds exactly as one would expect. However, they fail to notice - are blind to - the fact that the photograph selected, the one they are justifying was not the one they originally chose. In other words subjects happily state that they prefer the photograph of the blond in the picture in front of them, failing completely to notice that their original choice was a red-head and the experimenter had swapped pictures after the choice had been made [9]. Again, there are many experiments in support of the existence of choice blindness.
There is sufficient experimental literature in this area to say quite categorically that introspection illusion and choice bias are well established psychological phenomena.
2.2 Morality
Guided transcranial magnetic stimulation (TMS) is an fMRI technique which uses focused magnetic impulses. The focused beam is guided by a feedback coordinate system for spatial resolution. In theory this allows more precise targeting of specific neural areas. But more interestingly, TMS can also be used to target and disrupt location specific neural activity.
Researchers have used TMS to target the right temporo-parietal junction (RTPJ), and area just above the right ear. Like the ACC (anterior cingulate cortex) the RTPJ shows quite a lot of fMRI activity when subjects are asked to think about the intentions, thoughts, beliefs of others. The researchers [10] passed a magnetic field through the scalp (which in turn caused electrical current in the brain) targeted at the RTPJ in order to disrupt its normal functioning. Both long term (25 minutes) and short term bursts of 500 msec were used. In both cases subjects where then presented with various scenarios which the concerned possible harm to others.
After RTPJ disruption, there was a statistically small albeit significant likelihood that subjects found it more difficult to focus upon intent and so found some types of potential harm morally permissible, on the researchers’ seven point scale. In other words, subjects concentrated more upon potential outcomes in artificial moral scenarios than upon the supposed intentions of those in the scenario. This affect was diminished with time until subjects responded ’normally’ again Here the claim was made that applying TMS in this way reduced the role of beliefs in moral scenarios. Maybe. But a magnetic pulse and the resulting current disrupting some normal connections and normal chemical processes would be expected to produce some behavioural change regardless of the area probed. EM fields are bound to affect cognitive dynamics. That targetting the RTPJ produces some change is a QED triviality. To assume said change is linked to moral judgement is in my opinion, highly problematic. But this experiment and others like it do illustrate that choice illusion can be induced by altering physical state.
Consider now experiments concerning the so-called crying baby dilemma: You are with Anne Frank and other people hiding in the attic. The Nazis have entered the house. We can hear them searching from below. Everyone is collectively holding their breath, fearful that any sound would alert the Nazis to their presence. When suddenly a baby belonging to one of the young women in the attic begins to cry and cannot be stopped. The dilemma facing you is this: If nothing is done the Nazis will soon hear the child, and kill you all. If you try to cover the baby’s mouth it will be smothered and die - but everyone else will be saved. Is it morally permissible to kill the baby (note: you are not Cartman of Southpark fame)? This is a version of the philosophically famous ’trolley problem’ (see [2] and [3]). In the trolley problem a speeding trolley car will kill several people if you do not throw a switch to divert it. If you do trow the switch, it will only kill one person. Another version of the problem has you personally push a person in front of the trolley resulting in the lives of several others being saved. Notice the difference - a remote control switch versus personal contact (or a military drone remotely controlled from 5000 miles away, versus you piloting a helicopter. Death to many or to one in both cases is the result. (Startrek’s Spock - the needs of the many out-way the needs of the few - is the western, U.S.-centric amoral solution.)
Both sorts of moral dilemmas tends to entice the ACC into considerable activity. The ACC is an area generally thought to be linked to conflict response. In this case, a moral conflict in which choice is a factor. The solution ’kill the baby’ (shades of Eric Cartman not withstanding) resolves the conflict in a utilitarian way. Such utilitarian judgement tends to cause increased activity in anterior regions of the dorsolateral prefrontal cortex.
The dorsolateral prefrontal cortex (DLPC) is believed to be heavily involved in memory, particularly regarding working and short term memory. It is believed to perform in a manner similar to a buffer, temporarily caching data in a FIFO (first in, first out) manner. Because of this caching ability it is believed to be directly related to attention, for if data is lost in the cache then it is unavailable for the further processing required to attention and focused behavior. However, much of this is speculation based upon research which is very much in its infancy. I feel that because of this the extrapolations made by several researchers [71] concerning the roles of the ACC and DLPC in mediating the perception of errors may be premature.
Of more interest perhaps is the fact that the ACC is strongly connected (hardwired) to lateral frontal and parietal areas of the brain. These are areas thought to be involved in strategizing. That is to say, in mediating choice. Additionally the dorsal ACC is may be involved in reward based decisions [4]. For example, there is some tentitive indication that the ACC may be involved in mediating social drinking [5] (although this is a conclusion is I feel, premature given the problems listed below). Finally, there is some research [6] that the ACC has some association with reward/action behaviour and in conflict monitoring [7]. All of this is in the very tentitive early days of research. Despite this it may be that researchers such as [12][13][14] are accurate in saying that dilemmas such as the crying baby scenario are mediated by the ACC. If only because the ACC mediates many neural activities.
To conclude that it mediates specific behaviours may however, be very premature, perhaps even incorrect. I will show why below. But for the moment suffice it to say that a number of researchers notably [11][12][13][3][14] , believe that during trolley-like problems one neural area is specific to conflict-like decision making, and another to decisions involving utilitarian resolution. Another way to say this is that when subjects think about personal involvement in a trolley-like problem, ’emotional’ areas light up; whereas when thinking about non-personal involvement the areas believed to be prototypical ’rational’ centres, light up.
And so the proposal has been made that consequentialist decisions (viz. the remote switch case) correlate to activity in the ’rational’ areas, whereas deontological decisions (viz. direct physical involvement) correlate with activity in ’emotional’ areas. A few of those who have proposed this distinction have gone a significant step further however. They have proposed that utilitarian moral philosophy is a result of neurological rather than philosophical constraint.
Or more concisely, the tail wags the dog.
2.3 Freedom of choice
Various papers [11] have argued that illusions of choice, judgement conflict, etc. are driven by emotional and/or other non-rational factors. The rational justification of choice whether from conflict or illusion, is hypothesised to be mere rationalised obfuscation. The subject is fooling herself. For the real reasons for choice are usually unknown to the subject, as demonstrated by the panyhose experiment I described previously. Both choice and rationalization are said by these researchers, to be neurologically based and moreover, location specific.
In other words, it is suggested that reasoning - including moral reasoning - is largely post hoc. That is after the real neurological process has already made the ’decision’ prior to conscious realization or rationalization.
Some neurologists, psychiatrists, and philosophers have gone even further. They have postulated that all, even moral reasoning is usually derived from post hoc justification. And so emotion, introspective illusion, choice bias, neurological constraints... all these play a greater role in decision making whether moral or otherwise, than hitherto given credit. That is, the most fundamental philosophical conclusions are based upon the manner of neural wiring and functioning rather than upon the clever arguments of philosophers.
Do psychotics, criminals, paedophiles, arms dealers, those who wage genocide for oil, and other nefarious deviants have stunted response in the anterior cingulate and increased response in the anterior dorsolateral prefrontal areas? Do poets, saints, and gurus have this site-specific activation reversed? Is rational decision making, moral choice, freedom of thought constrained by a the wiring of a handful of neurons?
My answer to this is a simple ’no’. Conclusions concerning ’morality’ in neuroimaging/brain studies are highly questionable. There are several issues here:
For example there is the issue of cultural relativism (almost all published studies are from a particular westernesqe-scientific-medical vantage). There is also the obvious question of moral relativism, since the assumption inherent in most papers is that there is a sliding scale upon which moral judgement/decision making can be studied. And there is the issue that in the absence of direct observable outcome - the homunculus driven "theory of mind" may not be very objective. Further, "moral judgement" and ’illusion’ are artifacts of a belief system. That artifacts of any kind can be damaged (viz. TMS to the RTPJ) seems self-evident. None the less, there are more and more papers being published concerning the study of moral decisions, decision illusions, choice in general, legal implications of fMRI findings, and the assumed relation to site specific neural activation.
A number of papers have presented the hypothesis that there are two subsystems at work in the neural processes underlying choice, particularly moral choice [12][13][14][15]. These are:
1. The ’emotional neural process’ which generates judgements from deontological positions. Hence deontology would issue from moral rationalisation, rather than moral reasoning.
2. The second suggestion in the literature is that there are neural processes which generate utilitarian/consequentialist judgements. The strongest subsystem for any given decision wins. (This sounds to me like a reworking of Hebb’s work with neural clusters [16] or some basic borrowing from AI neural net work [58].)
A number of researchers argue that their empirically based hypothesis should convince us to discount all deontological intuitions while keeping utilitarian/consequentialist intuitions [17][18][19]. But consider the neurological data on which these hypotheses concerning decision making and morality are drawn. fMRI studies are a vogue - an average of eight peer reviewed papers concerning fMRI data are published per day [20]. This is in part simple fadism, in part lazy science, and in part a belief that high computational power, massive amounts (gigabytes to terabytes) of voxels , complex algorithms, plus a few magnets in a big machine, adds up to explanation of complex human behaviours. It does not.
3. False Positives
To see why, let us begin with the problem of false positives. A false positive is an experimental result which seems to confirm a hypothesis but actually does not. For example, finding a fingerprint match at a crime scene to a particular person does not scientifically implicate that person. Because as pointed out in [54] there is no scientifically sound corroborative statistical evidence for the uniqueness of, the permanence of, or the unique identification of, dermatoglyphes (fingerprints).
Sadly this has not stopped the courts from imprisoning people, nor police forces everywhere arresting people based upon this ’evidence’ from pseudo-science. Sadly dermatoglyphes are not the only pseudo-science tool used on accused persons. Stylometry, polygraphs, forensic DNA which I discussed in [46], and many other similar unscientific tests are used. Yet despite the overwhelming scientific evidence of the questionable status of these tools, courts, government, police, and military regularly make use of them in assigning motive and indeed, blame. All in a manner reminiscent of the use of phrenology in an earlier time to similarly impose constraint upon the innocent. Very sad.
One particular experiment highlights this difficulty with false positives in MRI work. Researchers presented pictures of people in different social situations with different emotional valences (defined below) to a subject in an fMRI. The subject was asked to determine what emotion individuals in the pictures were experiencing [21]. As always in such experiments, certain areas of the brain lit up with the different images, indicating increased neural activity according to which picture was presented. Each picture was present for 10 seconds, with a 12 second rest period between pictures. A total of fifteen photos were presented to the subject. Voxel statistics were calculated by commonly used least-squares estimation of the general linear model (GLM). Hemodynamic responses were modeled. A temporal high pass filter was used to account for any frequency drift in the fMRI. Commonly used algorithms were applied to account for overall error rates. In other words - standard procedures, standard controls, standard testing protocols. Yet as [22] recounts, despite appropriate credentials held by the experimenters, proper fMRI analysis, good source universities, and the like, not one journal was willing to publish this research. This despite the very high acceptance rate of almost identical fMRI studies.
Almost identical - there was perhaps one slight difference in the experiment just described and those typically carried out. For this time the subject in the fMRI who was doing the assigned task of differentiating emotion valences in photographs, was a dead Atlantic salmon which had been purchased from a local fish market the morning of the experiment.
The experimenters were of course making a point: Given the vast amount of data collected in these sorts of neuroimaging studies (~150,000 voxels), the probability of a false positive rapidly approaches one. Yet common statistical techniques (multiple comparisons in datasets, autocorrelative corrections, Markov correction, etc.) are still lacking in many papers which draw their conclusions from neuroimaging. In a study of all papers from six major neuroimaging journals, it was found that between 25% and 30% of fMRI analysis came from uncorrected thresholds [23].
In fact, many fMRI papers use arbitrary and incorrect statistical thresholds. For example, using P < 0.001 for voxel clustering of 10 voxels is obviously inappropriate for datasets of 60,000 voxels. And certainly not for 150,000 voxels. Yet these are used to argue everything from DSM diagnoses to moral specificity in the ACC!
Of course small and large datasets would have very different false positives propabilities. One would expect that a statistical methodology which resulted in the same expected probability of false positives for any number of voxels must be used. But it seldom is. Even though appropriate tests and methodologies are easily available from [24][25][26][27][28] and many others. False positives are inevitable. They are to be found in epidemiological studies [58], DNA forensics [48], police pronouncements concerning internet behaviour [64], no-fly lists [71], and any large-data-set induction in general. But this fact is often ignored in neuroimaging. Not by skilled researchers of course, but rather by the many others who add to the publication load of eight papers per day cited above.
Additionally since BOLD (Blood Oxygen Level Dependent contrast mechanism) fMRI signals from individuals can be ambiguous, results are generally averaged across subjects. The brightly lit brain areas of published fMRI images often do not represent neuronally active areas. Rather they are often statistical representations obtained by contrasting subject results (relatively small N) from those of controls [56]. Again I repeat my comments from [47] that proper and easy to do statistical correction is often ignored by researchers, particularly in medicine [72]. Whilst this is well known and accepted by good researchers in the area.
The sad reality is that given the number of papers published and the conclusions drawn, there may be room for improvement.
4. Neuroimaging interpretation
There are at least three misinterpretations common in this field:
-
Much of the research in functional brain mapping is based upon the (usually implicit) assumption that the brain consists of separate parts, each of which has a specific function. This has been shown time again to be incorrect (Friston et al’s well known paper for example, as well as [75] and [76], and particularly Vigneau et al and Cabeza et al). Functional brain mapping does not show complex mental functions is localized.
-
Another assumption in many papers, again implicit, is that cognitive processing is linear. Hence there is the feeling that new cognitive components can be inserted sequentially without being affected by previous ones during cognitive testing. This is false, as indicated by studies already mentioned as well as those described below. Yet neurologists, clinical psychologists, and other health workers persist in using such testing to diagnose, or even perform invasive surgery.
-
Finally there is the assumption BOLD signals are valid measure of neuronal activity at specific sites. More on this similarly false assumption, later. BOLD arises from susceptibility effects of deoxyhemoglobin in venous blood and reflects, thereby, the blood oxygen level [8]. Research has shown, however, that BOLD may reflect neuronal activity only under certain, as yet undefined, conditions [9;10].
All of these rest in turn upon the assumption that there is a one-to-one relationship between region and mental task. Yet very few (if any) regions within the brain completely align in a one-to-one manner with mental tasks. The example of pain is an obvious one - pain which a subject reports as occurring in only the left toe can and does in fact light up many regions of the brain. Almost every aspect of cognition (conscious or unconscious, asleep or awake, resting or intellectually engaged) involves widespread neural activity. Consider the the recognition of a musical phrase - much of the brain alights with even the simplest evocation of music.
The fact that some areas “light up” and other areas are more quiescent with selected stimuli is very misleading.
But what does it mean to say an area ’lights up’ during an fMRI? It has been found [40] that chemically quieted neurons (for our purposes, dead neurons) were none the less capable of generating fMRI signal data which made them appear to be active. That is to say, quiescent neurons also result in signals which ’light up’. Magnetic resonance imagery is a statistical interpretation taken from indirect measurements of one small aspect of brain activity - change in blood flow. Neuronal electrical communication, neurotransmitter chemicals, sodium-ion activity at the synaptic junction, and so on are invisible to an fMRI. The idea that blood flow is a measure of these is problematic. There is some correlate, but how much and what a correlation means is yet to be determined.
Using BOLD functional MRI several studies [55] looked at blood flow and electrical activity, pointing out that neurotransmitter release and presynaptic and postsynaptic currents do not necessarily correlate to site specificity [31]. In other words, activation areas in an fMRI may or may not relate to underlying neural activity, although there is some correlate between oxygen-concentration changes and local field potentials which are concurrent in time and space [32]. It has further been shown that the overall amount of blood available in an area strongly affects fMRI results. Low vasculature density can show areas rich in neural activity to be quiescent [33].
“...when input into a particular area plays what is primarily a modulatory role, fMRI experiments may reveal activation in areas in which physiological experiments find no single-unit activity.” [34]
Finally, it should be noted (but often is left out of discussions) that fMRI scans are more highly correlated with activity which provides input to a neural area... but not output [66]. Hence the resultant image or scan does not imply the area in question is generating activity in response.
All of this may lead one to believe that fMRI interpretation somewhat akin to dowsing. Perhaps it is. There is generally there is a widespread recognition amongst psychophysicists and other competent researchers that neuroimaging provides at best a loose coupling between brain activity and functional specificity. Subject age, gender, type of fMRI, vascularity differences, disease, detecting neural excitation or inhibition, ... and a long list of other variables, not the least of which are the statistical issues already mentioned, interfere with interpretation of BOLD hemodynamics. To conclude (as [12][13][14] and others seem to be doing) that functional and especially moral or choice specificity can be tracked by fMRIs is I feel, questionable:
“The limitations of fMRI are not related to physics or poor engineering, and are unlikely to be resolved by increasing the sophistication and power of the scanners; they are instead due to the circuitry and functional organisation of the brain, as well as to inappropriate experimental protocols that ignore this organisation.” [35]
fMRI images are pretty. A multitude of colours and flowing shapes. Some areas light up when the subjects are presented with certain information. And so it is tempting to take the attractive pictures and make up pretty stories around to describe what they represent, much as is done with abstract art. This Rorschach game is a current fashion upon which grants, reputations, and most of all commercialisation ride. From neuromarketing research to military applications there is money and status to be won. It can be compared to the use of quantum theory in computer science - put forward some harebrained marketing scheme proposing ’instantaneous computation’, wrap it up in quantum mechanical terminology, and voila - publications, grants, television interviews, and commercial products. And maybe tenure. The science of applications of quantum mechanics is diminished by this rush toward big science and exploitation. In a similar fashion the science of neurological study is diminished by every one with access to an fMRI jumping on the interpretation-to-functional-specificity bandwagon. But association is not causation, correlation is not proof (particularly with the small N available)... something which we would all do well to remember.
Or consider this: A voxel takes up on average a few millimetres of 3-space. Hence each voxel could easily contain at least six million neurons, as well as many more connections to other areas of the brain. Further, some neurons dampen, some activate. Neurotransmitters, differing axon bandwidths, on-the-fly neural plasticity, electrical excitation, quantum gravitational collapse in nanotubules and reduction of quantum coherence, and so on - all processes which occur constantly, constantly shifting, constantly remapping and redirecting.
It may be important to remember too that the brain always operates in 4-space - the processes and encoding [38][39] are not just spatially but temporally mapped. Nor should be forget the informational complexities involved within this 4-space area (here I am refering specifically to chaotic sets and complexity theory). The combinations and permutations available in even a single voxel are large - let alone in an fMRI dataset of say, 150,000 voxels. Add in temporal dimension and informational complexity - and we have an astronomically large amount of data.
Using the Hubble telescope to look at the earth from hundreds of millions of light years away would not really tell one much about what is going on on regarding Yo Yo Ma’s latest foray into jazz. So too with fMRI interpretation of neural specificity regarding behaviour.
There is another point here too - electricity moves faster, and can distribute further, than liquid. The implications for fMRI scanning are obvious - by the time the scanner recognizes and records blood (liquid) activity, the electrical potentials have scattered far and wide. That is, neurons all over the brain may have been involved during an experiment but the fMRI simply cannot detect them. And so looking at this data alone may lead one to incorrectly conclude functional specificity in regards to say, decision making. Further, it is almost impossible with current technology to accurately track electrical activation in the brain. But even if this were not the case, the complexity of the interactions would be very difficult to understand. Even the relative simplicity of a perceptron-like neural network of more than trivial size is so complex as to challenge the commonly available mathematical tools to the utmost. (My personal bias is that the patterns observed are best understood by modelling with the mathematics of chaotic swarm behaviour. For example path integration and horizontal projection have some applicability here.)
Some regions are extremely plastic. Others not so much - binocular vision is an example, where plasticity is lost once a critical period in childhood is passed. On the other hand some general locus specific functions can be activated after decades without there being any critical period (reading is an example). Others such as figure-ground discrimination do not seem to have much plasticity at all (ie. if damaged do not appear to regenerate or be taken over by other neural processes). Studies of brain damage help in mapping general location function and plasticity. Yet these too can suffer from false positives, small sample size, and the like. Hence the tools of mapping - operation with awake patients, data from torture victims, data from functional damage due to TBI etc. - show some specificity of function, some plasticity, some age dependency, and so on. Neural activity is chaotic (in the sense of complexity theory) - deterministic models such as locus-specific-behavior are unlikely. Concluding behavioural and even moral/philosophical location specificity in such and environment is highly problematic and indeed damaging. Why damaging? Because non-scientists, and even not-very-good scientists, particularly in medicine and psychiatry, have come to use fMRI finding as if they were accurate and true. They are using them in the same manner that pseudo-science (the polygraph for example - see [41][42]) is used in the courts.
Again: Association is not causation; induction is not proof. Reduction of morality, choice, or cultural predilection to a sole organ or worse a specific locus is the poor science of Cartesianism.
5. Neuro-quackery
There have recently been stories in various major media to the effect that it is possible to directly infer preference (such as choice of political party) by watching activity in specific locus areas such as the amygdala. No real cognitive neuroscientist [73], indeed no real scientist, would make such a claim. As I pointed out above, a one-to-one mapping between locus and psychological state simply cannot be done. Especially when speaking of higher functions.
Similarly there are those who have cashed in, using thinly disguised infomercials, talk show venues, public television, or self-promoting books to allegedly push their very expensive ’brain clinics’ and/or treatment regimes. These neuropundits go so far as to allegedly claim that by using SPECT (single photon emission computed tomography, they can reverse Alzheimer’s disease or dementia, and that they can diagnose neurological problems from ADD to autistic spectrum disorder or even substance abuse [74]. However as I indicated above, behavioral problems such as substance abuse cannot be diagnosed using neuroimaging.
Snake oil salesmen have always used the aura of science to lure the gullible or, sadly, the desperate to part with their money. Whether the salesman is wearing L.E.D. glass and making factually inaccurate claims about their effect, or whether he is flogging his ’expertise’ in analysing Alzheimer-images, the results are always the same: the gullable pay. This is no different in pseudo-medicine than in pseudo-spirituality. But falsely raising the hopes of those who are very ill while smilingly taking their money, is appalling. And a very sad comment upon the misuse of science in our society.
6. Conclusion
"... The most absurd piece of pseudo-scientific nonsense is the dogma that consciousness is a by-product of matter." — Douglas Harding, 2002
Hypothesising brain biology and neuronal circuitry as correlates of consciousness does not encompass the entirety of the phenomenology of experience. Physical theories based on reductionist approaches that take into account only brain biology and neuronal circuitry are by their nature, limited and limiting ontologies.
