The harmonics-in-connectome approach to modeling brain activity is a fascinating paradigm. I am privileged to have been at this talk in the 2017 Psychedelic Science conference. I’m extremely happy find out that MAPS already uploaded the talks. Dive in!
Below is a partial transcript of the talk. I figured that I should get it in written form in order to be able to reference it in future articles. Enjoy!
[After a brief introduction about harmonic waves in many different kinds of systems… at 7:04, Selen Atasoy]:
We applied the [principle of harmonic decomposition] to the anatomy of the brain. We made them connectome-specific. So first of all, what do I mean by the human connectome? Today thanks to the recent developments in structural neuroimaging techniques such as diffusion tensor imaging, we can trace the long-distance white matter connections in the brain. These long-distance white matter fibers (as you see in the image) connect distant parts of the brain, distant parts of the cortex. And the set of all of the different connections is called the connectome.
Now, because we know the equation governing these harmonic waves, we can extend this principle to the human brain by simply solving the same equation on the human connectome instead of a metal plate ( Chladni plates) or the anatomy of the zebra. And if you do that, we get a set of harmonic patterns, this time emerging in the cortex. And we decided to call these harmonic patterns connectome harmincs. And each of these connectome harmonic patterns are associated with a different frequency. And because they correspond to different frequencies they are all independent, and together they give you a new language, so to speak, to describe neural activity. So in the same way the harmonic patterns are building blocks of these complex patterns we see on animal coats, these connectome harmonics are the building blocks of the complex spatio-temporal patterns of neural activity.
Describing and explaining neural activity by using these connectome harmonics as brain states is really not very different than decomposing a complex musical pieces into its musical notes. It’s simply a new way of representing your data, or a new language to express it.
What is the advantage of using this new language? So why not use the state-of-the-art conventional neurimaging analysis methods? Because these connectome harmonics, by definition are the vibration modes, but applied to the anatomy of the human brain, and if you use them as brain states to express neural activity we can compute certain fundamental principles very easily such as the energy or the power.
The power would be the strength of activation of each of these states in neural activity. So how strongly that particular state contributes to neural activity. And the energy would be a combination of this strength of activation with the intrinsic energy of that particular brain state, and the intrinsic energy comes from the frequency of its vibration (in the analogy of vibration).
So in this study we looked at the power and the energy of these connectome harmonic brain states in order to explore the neural correlates of the LSD experience.
We looked at 12 healthy participants who received either 75µg of LSD (IV) or a placebo, over two sessions. These two sessions were 14 days apart in counter-balanced order. And the fMRI scans consisted of 3 eyes-closed resting states scans, each lasting 7 minutes, in the first and the third scan the participants were simply resting, eyes closed, but in the second scan they were also listening to music. And after each scan, the participants rated the intensity of certain experiences.
So if you look at, firstly, at the total power and the total energy of each of these scans under LSD and placebo, what we see is that under LSD both the power as well as the energy of brain activity increases significantly.
And if we compute the probability of observing a certain energy value on LSD or placebo, what we see is that the peak of this probability distribution clearly shoots towards high energy values under LSD.
And that peak is even slightly higher in terms of probability when the subjects were listening to music. So if we interpret that peak as, in a way, the characteristic energy of a state, you can see that it shifts towards higher energy under LSD, and that this effect is intensified when listening to music.
And then we asked, which of these brain states, which of these frequencies, were actually contributing to this energy increase. So we partitioned the spectrum of all of these harmonic brain states into different parts and computed the energy of each of these partitions individually. So in total we have around 20,000 brain states. And if you look at the energy differences in LSD and placebo, what we find is that for a very narrow range of low frequencies actually these brain states were decreasing their energy on LSD. But for a very broad range of high frequencies, LSD was inducing an energy increase. So this says that LSD alters brain dynamics in a very frequency-selective manner. And it was causing high frequencies to increase their energy.
So next we looked at whether these changes we are observing in brain activity are correlated with any of the experiences that the participants themselves were having in that moment. If you look at the energy changes within the narrow range of low frequencies, we found that the energy changes in that range significantly correlated with the intensity of the experience of ego dissolution. The loss of subjective self.
And very interestingly, the same range of energy change within the same frequency range also significantly correlated with the intensity of emotional arousal, whether the experience was positive or negative. This could be quite relevant for studies looking into potential therapeutic applications of LSD.
Next, when we look at a slightly higher range of frequencies, what we found was that the energy changes within that range significantly correlated with the positive mood.
In brief, this suggests that it’s rather the low frequency brain states which correlated with ego dissolution or with emotional arousal, and it’s the activity of higher frequencies that is correlated with the positive experiences.
Next, we wanted to check the size of the repertoire of active brain states. And if you look at the probability of activation for any brain state (so we are not distinguishing for any frequency brain states), what we observe is that the probability of a brain state being silent (zero contribution), actually decreased under LSD. And the probability of a brain state contributing very strongly, which corresponds to the tails of these distributions, were increased under LSD. So this suggests that LSD was activating more brain states simultaneously.
And if we go back to the music analogy that we used in the beginning, that would correspond to playing more musical notes at the same time. And it’s very interesting, because studies that have looked at improvising, those who have looked at jazz improvisation, show that improvising jazz musicians play significantly more musical notes compared to memorized play. And this is what we seem to be finding under the effect of LSD. That your brain is actually activating more of these brain states simultaneously.
And it does so in a very non-random fashion. So if you look at the correlation across different frequencies. Like at the co-activation patterns, and their activation over time. You may interpret it as the “communication across various frequencies”. What we found is that for a very broad range of the spectrum, there was a higher correlation across different frequencies in their activation patterns under LSD compared to placebo.
So this really says that LSD is actually causing a reorganization, rather than a random activation of brain states. It’s expanding the repertoire of active brain states, while maintaining -or maybe better said- recreating a complex but spontaneous order. And in the musical analogy it’s really very similar to jazz improvisation, to think about it in an intuitive way.
Now, there is actually one particular situation when dynamical systems such as the brain, and systems that change their activity over time, show this type of emergence of complex order, or enhanced improvisation, enhanced repertoire of active states. And this is when they approach what is called criticality. Now, criticality is this special type of behavior, special type of dynamics, that emerges right at the transition between order and chaos. When these two (extreme) types of dynamics are in balance. And criticality is said to be “the constantly shifting battle zone between stagnation and anarchy. The one place where a complex system can be spontaneous, adaptive, and alive” (Waldrop 1992). So if a system is approaching criticality, there are very characteristic signatures that you would observed in the data, in the relationships that you plot in your data.
And one of them is -and probably the most characteristic of them- is the emergence of power laws. So what does that mean? If you plot one observable in our data, which for example, in our case would be the maximum power of a brain state, in relationship to another observable, for example, the wavenumber, or the frequency of that brain state, and you plot them in logarithmic coordinates, that would mean that if they follow power laws, they would approximate a line. And this is exactly what we observe in our data, and surprisingly for both LSD as well as for placebo, but with one very significant and remarkable difference: because the high frequencies increase their power on LSD, this distribution follows this power law, this line, way more accurately under LSD compared to placebo. And here you see the error of the fit, which is decreasing.
This suggests that LSD shoots brain dynamics further towards criticality. The signature of criticality that we find in LSD and in placebo is way more enhanced, way more pronounced, under the effect of LSD. And we found the same effect, not only for the maximum power, but also for the mean power, as well as for the power of fluctuations.
So this suggests that the criticality actually may be the principle that is underlying this emergence of complex order, and this reorganization of brain dynamics, and which leads to enhanced improvisation in brain activity.
So, to summarize briefly, what we found was that LSD increases the total power as well as total energy of brain activity. It selectively activates high frequency brain states, and it expands the repertoire or active brain states in a very non-random fashion. And the principle underlying all of these changes seems to be a reorganization of brain dynamics, right at criticality, right at the edge of chaos, or just as the balance between order and chaos. And very interestingly, the “edge of chaos”, or the edge of criticality, is said to be where “life has enough stability to sustain itself, and enough creativity to deserve the name of life” (Waldrop 1992). So I leave you with that, and thank you for your attention.