Modern brain imaging - a huge advance,
but still rife with ambiguity.
Written & Illustrated by Andrew Neff
fMRI stands for functional Magnetic Resonance Imaging.
Plain old MRI, without the f,
refers to a set of methods that rely on intrinsic differences
in magnetic properties between tissues.
These differences are exploited to generate pictures,
either detailing anatomy, or revealing physiological processes.
The key is the contrast,
different tissues have different magnetic properties,
or the same tissue in different conditions
may have altered magnetic properties
The functional part happens when you program the settings on the machine
to reveal the contrast between oxygenated and deoxygenated blood,
also known as the BOLD (Blood Oxygen Level Dependent) contrast.
Active brain cells are metabolically demanding.
To recharge after increased activity,
neurons must extract more oxygen from blood.
blood flow is increased in neighboring blood vessels.
However, incoming blood is highly enriched in oxygen,
more oxygen than even the most active neurons can handle.
This leads to a short lived spike in blood oxygenation,
that can be detected with fMRI.
fMRI researchers believe that the BOLD contrast indicates brain activity,
which indicates psychology.
Implanting electrodes deep into brain tissue
allows researchers to record “large scale” brain activity,
more or less a summary of all the electrical activity in a chunk of tissue.
With these direct measurements,
you can validate the source of the BOLD signal,
and you tend to see that electrically active tissue
has an increased BOLD signal (Logothetis, 2001).
However, when using more precise methodologies
to record from individual neurons,
rather than the summary of a large area,
many neurons didn’t correlate well with the BOLD signal (Park, 2017).
The brain is more complicated than what can be measured with fMRI,
nobody denies that.
But this is a fact that it’s far too easy to ignore,
to put off to future generations.
Understanding what fMRI measures
in the context of what the brain actually is
might be the only way forward.
What fMRI sees,
and what a brain is
Your brain has 100,000 neurons per mm3 (Azevedo, 2009),
that’s about the size of a poppy seed.
The volume of your brain is about 1,000,000 mm3 (Lüders, 2002.),
or one million poppy seeds.
Based on some highly irresponsible back of the envelope assumptions,
that’s 100,000,000,000, or one hundred billion neurons.
The best resolution fMRI can obtain is 1 mm3,
a chunk of tissue populated by a hundred thousand cells (Schuz, 1989).
That means from those hundred thousand cells,
you get one data point, a single number.
But that’s too complicated, too many neurons.
To help wrap our minds around it,
consider a chunk of tissue with only 25 neurons.
Some report on what scientists report.
We're here for the real thing.
Independent - design heavy - science journalism.
Support us today on Patreon.
fMRI works by comparing brains to each other.
A neuroscientist might compare brain activity
between groups of people,
maybe extroverts vs introverts,
or they might look within a person over time,
say while tapping a finger compared to not tapping a finger.
At any given moment,
a fraction of these 25 cells are experiencing
what’s called an action potential,
people say these cells are “firing”.
If 12 neurons are firing in extraverts,
but only 8 in introverts,
we might be able to see a difference in the BOLD signal.
In a system with 25 neurons,
where each of these neurons has only two possibilities,
firing or not firing,
at any given moment,
there could be 34 million possible combinations.
34 million combinations in the extroverts,
34 million in the introverts,
and all fMRI is willing to tell you is
whether one group has more neurons firing than the other group.
And, that’s only 25 neurons,
imagine if it was 100,000 neurons instead.
But the brain doesn’t function in a moment,
and psychological processes don’t happen in a moment,
what matters is the sequence of neuron firing over time.
It takes a little while for the wave of voltage to travel
through the dendrites, into the cell body,
down the axon, and into the axon terminals.
At any given point in the cell,
an action potential spikes and returns to baseline
within about 5 milliseconds.
In contrast to rapid neuronal activity,
changes in blood-flow is a slow process.
The diversion of oxygen rich blood into capillaries
depends on the physical rearrangement of blood vessels,
it just can't happen as fast as action potentials do.
When blood is delivered to active neuronal tissue,
that spike in oxygenation lasts for about 5 seconds.
From the beginning to the end of blood influx,
each neuron could experience 1,000 action potentials.
But who can understand a thousand.
Let’s say we look at a 25 milliseconds,
and divide that timeframe up into five 5 millisecond time-chunks,
about the length of a single action potential.
over 5 chunks of time,
each firing or not firing
that’s 2^125 possible combinations,
there's no name I'm aware of for a number this big.
2^125 is way more stars than in the Milky Way,
more than the number of grains of sand on earth,
actually, it’s the amount of sand
on 10 quintillion earths.
but fMRI has put it’s foot down
and decided that today,
all it’s willing to tell you
is whether one group is firing more neurons than the other.
And, that’s over 25 milliseconds,
binned into 5 second chunks.
Imagine 1,000 milliseconds,
that don’t fit neatly within artificially defined time chunks?
The BOLD signal has coarse resolution,
it only tracks large chunks of tissue,
over relatively long timescales.
Within this time there is an enormous
amount of opportunity for a diversity of brain activity.
It could be,
that all of this complexity just isn’t important.
all that matters is the coherent throbbing of hundreds of thousands
of neighboring neurons.
But it’s also possible that
this enormous amount of complexity is important.
Right now, we’re crossing our fingers
and hoping that a signal emerges from the coarse resolution
data that we have.
And sometimes, often even, trends emerge.
The engagement in a task,
or the existence of psychological characteristic
is often found to be correlated with
the location of the BOLD signal.
On the other hand,
studies often fail to replicate original findings,
and combining data across large numbers of experiments
has in some studies yielded no significant differences
between people with and without psychiatric disorders (Müller, 2017).
Azevedo, Frederico AC, et al. "Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled‐up primate brain." Journal of Comparative Neurology 513.5 (2009): 532-541.
DeFelipe, Javier, et al. "Estimation of the number of synapses in the cerebral cortex: methodological considerations." Cerebral Cortex 9.7 (1999): 722-732.
Lüders, Eileen, Helmuth Steinmetz, and Lutz Jäncke. "Brain size and grey matter volume in the healthy human brain." Neuroreport 13.17 (2002): 2371-2374.
Müller, Veronika I., et al. "Altered brain activity in unipolar depression revisited: meta-analyses of neuroimaging studies." JAMA psychiatry 74.1 (2017): 47-55.
Schüz, Almut, and Günther Palm. "Density of neurons and synapses in the cerebral cortex of the mouse." Journal of Comparative Neurology 286.4 (1989): 442-455.
Andrew Neff - 25 min: fMRI vs Phrenology: Is neuroscience bound to keep repeating old mistakes? Modern brain imaging vs phrenological weirdness (more)
Some report on what scientists report - we're here for the real thing.
Design-Oriented Science Media:
Skeptical Perspectives, Independently Produced.
Support us today on Patreon.
This article was an excerpt from \/
Andrew Neff - 15 min: The Gut-Brain-Axis: Scientists say gut bacteria are shaping human psychology, but the evidence? (more)
Andrew Neff - 15 min: Exosomes, biological treasure chests primed to revolutionize biomarker research? Why exosomes could, but probably won't, be about to change everything. (more)