Jan 26 2009
An fMRI Wrinkle
Functional magnetic resonance imaging, or fMRI, is an extremely useful new technology (created in 1993 and increasingly used over the last decade) to study brain activity. Like any new and complex tool, however, it is critical to understand how the tool works, especially when interpreting research. A new study published today in Nature may cause fMRI researchers to change how they conduct an interpret their research.
fMRI is an application of regular MRI, made possible by advances in computer processing power and researchers finding new ways to exploit how living tissue responds to a powerful magnetic field. In the case of fMRI it was discovered that blood reacts differently when it is highly oxygenated than when it is less oxygenated. This is because oxyhemoglobin (the form of hemoglobin with oxygen) is diamagnetic, while deoxyhemoglobin (the form of hemoglobin without oxygen) is paramagnetic. Most living tissue (and in fact most stuff) is diamagnetic – in response to a strong external magnetic field it creates a weak opposite magnetic field. In superconductors this property can be exploited to create levitation above a strong magnet. A paramagnetic substance, rather, in response to a strong external magnetic field creates an attractive magnetic field. Unlike ferromagnetic materials (like iron), diamagnetic and paramagnetic substances lose their magnetic field as soon as the external field ends – they are not permanent or long-lasting.
All of this means that oxygenated and less oxygenated blood behave differently in the magnetic field of an MRI and this differential response can be exploited to image relative arterial blood flow to different parts of the brain. An entire branch of neuroscientific research has sprung up over this fact – using fMRI to see which parts of the brain are active when subjects are asked to perform a specific task, when they are exposed to external stimuli, or in different disease states.
Of course, using fMRI in such research is based upon the assumption (reasonable, but an assumption none-the-less) that blood flow as imaged by fMRI correlates with brain activity. This premise is based upon sound physiological knowledge, and also experience from other technologies, such as PET and SPECT scanning, that also image blood flow.
However, it is also this exact premise that is being challenged, at least to a degree, with this new study.
Neuroscientists Yevgeniy Sirotin and Aniruddha Das at Columbia University performed studies on two monkeys. They monitored their brain activity with implanted electrodes and simultaneously monitored their cerebral blood volume and oxygenation with optical methods. They then exposed the monkeys in a dark room to a tiny light at regular intervals and rewarded the monkeys if they fixed their gaze on the light for a few seconds. What they found is that the electrical activity in the visual cortex remained stable (there was not enough light to cause significant activity in the visual cortex), however the blood flow did rise and fall and peaked a few seconds before the light was to go on.
It seemed in this study that blood flow and cortical activity were not closely correlated. Of course, interpretation is key. As reported in a Science news article:
Although the findings “by no means call into question the whole body of fMRI research,” Das says they should cause fMRI researchers to rethink how they design and interpret their experiments.
I agree this study does not contradict the whole of fMRI research. The usual caveats apply – this was just one study with two subjects, for example. Many studies with various study designs will be necessary to see which variables are important for determining blood flow and cortical activity. But taking these results at face value, what can they mean.
It does not seem as if there was no relationship with blood flow and brain activity. The changes in blood flow were not random – they correlated with the task. One interpretation is that the specific measures of blood flow in this study are more sensitive to changes in brain activity than the electrical recordings.
Another intriguing possibility is that the brain can anticipate its metabolic needs and increase blood flow to parts of the cortex that are about to become active. In this study the light was shown at a regular interval, and the monkeys were being rewarded for fixing their gaze upon it. Perhaps they were anticipating the appearance of the light and priming their visual cortex for the task, but because the light was so small it didn’t actually require that much visual cortical activity.
As is typical for scientific research, this new finding does not so much contradict prior research as add a new level of depth and complexity. There is copious evidence that blood flow tracks with brain activity, but this relationship is probably not as simple as was at first assumed. The good news is that if there in an anticipatory effect, as suggested by this research, then that is something that can be controlled for in the design and interpretation of future fMRI studies.
It also means there is a previously unknown physiological mechanism in the brain for anticipating immediate metabolic needs. This makes sense as the brain is a hungry organ, requiring incredible optimization of delivery of glucose and oxygen as well as a tightly controlled metabolic environment. Even a fleeting drop in blood pressure, or a drop in blood sugar, can cause someone to pass out. Therefore if there were a way to deliver much needed oxygen and glucose to brain tissue just prior to its increased activity, that would be a useful adaptation.
What this study does more than anything else is raise an interesting question, and will likely result in a slew of follow up research. There will likely be criticism and disagreement as well. But that’s the nature of science – it’s all good.
23 Responses to “An fMRI Wrinkle”
Leave a Reply
You must be logged in to post a comment.
interesting stuff… i would be surprised if Dr. Egnor didn’t wildy misinterpret and misconstrue the findings
Hemoglobin, not “hemaglobin”.
Levitation, not “levatation”.
Sorry, compulsive.
ADR150, Of course he will… it is quite clear that this PROVES the monkeys are psychically predicting when the light will shine. As interesting as this story is (and as important as the point Dr. Novella is making should be), I could not stop myself thinking the worst about how this study will be tortured by the woo-merchants. Urgh.
This result is very interesting. It does show that what the fMRI signal is showing is more complex than was first thought.
The brain is active tissue, in that it can support the active propagation of signals, changes in state. These include nerve impulses but also things such as spreading depression mentioned in the blog on migraines. Any active media requires internal energy to sustain the change in state and a propagating signal to trigger the change.
Blood flow and neural activity are regulated somewhat independently (and they have to be because the velocity of propagation of blood flow and of nerve signals is so different). Vasodilation of blood vessels is triggered by NO, the firing of a nerve cell is triggered by an action potential propagating down an axon.
The “ideal” regulation would increase blood flow before the neural activity, so the substrates (O2 and glucose) would be available before they were needed by that neural activity. The brain doesn’t have the metabolic capacity to support all neurons firing at maximum capacity all the time, so there has to be an allocation of blood flow. The more efficient that allocation, the better neural activity and blood flow will track each other.
Lots of physiology is regulated in a feed-forward manner. The heart pumps blood that will be used at a future time, when that blood reaches the target organ. The liver puts glucose into blood so it can be used when it reaches the target cells (which can be tens of minutes for plasma to reach extravascular regions). Feed-forward regulation is necessarily less precise than feedback regulation, but when you need very rapid response (as in brain activity), you must tolerate the inefficiencies of feed-forward regulation.
In hindsight it is pretty much what I would expect, the brain anticipates which brain volume needs to be activated (i.e. which brain volume is needed for a certain task), NO releasing neurons in that brain volume release a puff of NO to cause vasodilation to increase metabolic capacity in that volume (and perhaps do other things to facilitate neural firing such as shift triggering thresholds, triggering voltage of voltage gated channels and such), then either the action potentials propagate into that volume (perhaps facilitated by NO mediated regulation) and further computation is effected, or they do not. The increased oxyhemoglobin ties up the NO and everything goes back to baseline. Sending blood only to those parts of the brain that are needed in the next few seconds allows for greater efficiency and allows a larger brain to be metabolically supported.
If Egnor is smart, he won’t try and use this as evidence for dualism. First, this result was in monkeys, not humans. If this result was due to an immaterial mind, a “soul”, then monkeys have a soul. Second, the signal showed up because blood flow and neural activity did not track each other exactly. If the lack of a 1 to 1 correspondence is due to an immaterial mind, that immaterial mind is somehow interfering with and degrading the correspondence between two material systems (blood flow and neural activity). If an immaterial mind is interfering with the operation of the material brain, evolution would pretty quickly sever any connection to the immaterial mind that is causing the interference.
You realize, of course, that if this were my blog, I would correct the errors and then ruthlessly delete your comment? That’s my policy when it comes to comments that are nothing more than spelling or grammar flames.
Orac, the poster’s name is Nitpicking, which constitutes a clue, he apologizes for his compulsion, Dr. N has thanked commenters for corrections in the past, it wasn’t a flame post, and, in fact, this isn’t your blog.
If I may add, for the benefit of proponents of magnets in therapy- diamagnetic and paramagnetic effects are really, really small compared to ferromagnetism. So, such (dia- and para-) effects do not support any significant result from applying ordinary magnets to one’s body. I wish I had numbers to quote; but I don’t. I do know that it takes a sensitive instrument to measure those (dia- and para-) effects when using a strong magnet. I also know that an MRI magnet (or, in my case, NMR magnet) can snatch forceps from my pocket at a distance two feet; while exerting no, perceptible effect on my (diamagnetic) self.
Thanks for the interesting post.
I thought we it had already been discussed in the past that somewhere in the eye -> Brain circuitry that there is evidence that it “anticipates” things due so as to not lag behind real time, and that’s why we experience optical illusions. So couldn’t we then just be seeing (more) evidence of this anticipatory pathway?
I dont know though. That anticipation aspect of visual circuitry is always present, so I guess that in this test it is seen as further anticipation, something above and beyond whats normally seen?
I’ve never done an fMRI study, so I can’t accurately answer my question.
The monkey researchers were measuring what we call in learning theory a fractional anticipatory response. The McAllisters, a team of learning psychologists have extensively studied this learned response. I guess between the “pictures” and the monkeys, it must have been a two banana day.
Is a “two banana day” a joke, or is it jargon for a high reward day?
Orac, if you corrected the typo and deleted my message that would be fine. No flame intended, no offense meant, none taken.
Nice post Steve. Interesting. We shall see where this leads.
All living things have to anticipate the future in order to survive. Monkeys should anticipate the future.
daedalus2u- I agree with your analysis of the situation. I’m not sure why this would be a problem for a dualist unless said dualist did not b
Oops-
unless the dualist did not believe that monkeys have souls.
As I understand the history- Descartes thought animals were machines and people had souls. Up to that point people generally thought living things were possesed. (Some people apparently did not like the idea that there would be flys in heaven, however)
Certainly the vast majority of humans have believed that animals have the a life force- and still believe so today.
I don’t know of Egnor’s beliefs on this matter.
I will preface this statement by saying I think fMRI research is incredibly interesting and a huge boon to science. However, I image there are probably huge problems in overanalysis of fMRI results and exprapolation of conclusions, especially since we do not fully understand the relationship of blood flow to brain activity. The idea that the brain “anticipates” brain activity with blood flow seems to me there would necessarilly be a built-in error in correlation. Deadalus’s comments on feed-forward physiological mechanisms makes sense to me.
I see the role of fMRI as giving science a better idea of the general responsibilites of certain areas of the brain – tagging them for further research via other methods. Has anyone heard of fMRI being used as a diagnostic tool for psychological conditions, and what are your thoughts?
Interesting historical fact – 1959, Singer demostrated the use of
MRI in blood flow. In general, MRIs are used to detect tumors, lesions, subdural and epidural hemotomas, contusions, diffuse axonal injury,cerebral infarctions, cerebritis and AIDS patient granulomas and toxoplasma abscesses,leukoencephalopathy and encephalitis from HIV virus, and cerebral atrophy.
The spelling should be hematomas.
MM, BOLD (blood oxygen level dependent) fMRI is only one type of fMRI (functional MRI). It looks at the oxy/deoxy hemoglobin levels. There are other types of fMRI that look at pH (via phosphate signals), ATP levels, and other metabolite levels (often called MRS (magnetic resonance imaging spectroscopy)). These are more tricky and difficult because the concentrations of these compounds are lower and so the MRI signal is smaller requiring larger magnetic fields. There are some that use 9 Tesla fields now.
The advantage of BOLD fMRI is that it is really fast. Reliable and reproducible differences can be observed in time scales less than a second. The whole brain can be imaged simultaneously at very good resolution (both in time and space). The assumption has pretty much always been that the BOLD fMRI signal = brain activation. I think (unfortunately) there has been less attention played to the physiology behind the BOLD fMRI signal and how that is coupled to neuronal activation because many of those doing neuroimaging are more concerned with anatomy and structure and less with physiology (which is usually not measurable via imaging devices).
I am especially interested in BOLD fMRI because it is NO that does the coupling between blood flow and neural activity. Neurogenic NO release is what causes the increased blood flow observed in BOLD fMRI (via activating sGC). The basal NO level (the NO level before there is neurogenic NO release) has to be important in the onset time, range and duration of the vasodilation that causes the BOLD fMRI signal. There are hints in the literature that there are characteristic differences between individuals that may be due to their basal NO levels (e.g. people with autism have smaller “regions of interest”, that is their activated volumes are smaller, what would be expected if their basal NO levels were lower).
The levels of NO that are important in this respect (nM/L) are difficult to measure in vitro. There are no techniques to measure them in vivo on the length, time and concentration scales that we know are important. Using endogenous sGC as the “NO sensor” and looking at the increased vasodilation is one way, but deconvoluting the NO signal out is tricky because there are other things that affect sGC (such as ATP levels).
There are multiple techniques to measure blood flow with MRI, the earliest technique (reported by Singer JR) used the fact that blood moves, and that movement can be detected by measuring the time between when the protons are magnetized and when that magnetization is detected.
The BOLD fMRI technique is more measuring the amount of deoxyhemoglobin present in a certain volume rather than flow or movement per se.
Noting that this is a learning paradigm used within this examination of fMRI analyses, I would be most interested to see a series of experiments mapping out brain activity/blood flow results in several additional learning preparations. First and foremost, moving to conditional stimulus control (from the simple stimulus control in this study). This learning paradigm should allow for differential observation of acitivity change. Next, it would certainly be interesting to look at blood flow in a classical conditioning paradigm. Another move would be to produce differential schedule control.
Very interesting work even though I don’t feel the brain anticipates, it is the person. The brain is the vehicle rather than the cause. Viewing the brain changes following learning reveal the mechanism of physiologic change during learning but not the less reduced level of psychological change during learning.
The evidence is mounting that there are some serious problems with the standard protocols and tools in cognitive neuroscience… Things are always more complicated than they seem.
If I am to interpret this study along the lines of a visual fractional anticipatory response, I would state that this involves several types of conditioning, impacting the retina,visual area,visual association cortex and lateral geniculate body. These pathways to the brain are responsible for fixation of gaze, tracking of objects, and visual awareness including memory and discrimination of visual stimuli.
Dr. Egnor addresses you guys as “Darwinists” because there’s no better name for you. Darwinists are really atheists who justify their atheism by attributing everything but the kitchen sink to Darwin. What other explanation could there be for the manic pursuit of a dogma that is attributed to a mere human being. Darwin was not a god, yet he is revered as such by you guys. As a mere man, he erred. Yet you geniuses seem unwilling to acknowledge that as a possibility. You are unreasonable, and unreasoning. You are incapable of respectful dialogue with those who have differing views. There’s a saying that if you are a hammer, everything looks like a nail. Could that be the reason the evidence SEEMS to stack up in your favor?
I am not a scientist, but I have read a bit, and I am very much interested in evaluating all ideas according to their merit. I have uncovered questions and challenges to the scientist/naturalist/atheist/whateveryouwanttocallyourself dogma. I wrote a series on my blog, and I was challenged a few times. Yet each time I responded with calm reason, asking questions that seemed to challenge Darwinian dogma, my challengers simply disengaged. All they seemed interested in was yelling, cursing, deriding and name-calling. Once I challenged them to defend their positions rationally, they simply disappeared.
Drop on by and take a look. See if you can answer my questions in the spirit of truth-seeking. That means reason with me. If you think I don’t get it, explain it to me. http://andj4613.wordpress.com