Interview Questions from Richard Bright (Interalia Magazine) and Julia Buntaine (SciArt Magazine)
Your work Self Reflected has been described as “the most complex artistic rendition of the human brain in the world.” How did this work come about?
Greg Dunn: Self Reflected exists because I feel a strong need to help enrich the average person’s understanding of the brain. The brain’s complexity is astonishing beyond description, and amidst that complexity is both an aesthetic and conceptual beauty that unfortunately is underappreciated by most folks because there is little opportunity to do so. I wanted to give a glimpse into this reality not by boiling down the complexity into simple illustrations, but by making a work of art so large and detailed that it would be possible to communicate through emotions and direct perceptions rather than academically. Stating that the brain contains 86 billion neurons is attempting to communicate through an abstract intellectual method. You can scream this incredible fact in the classroom until you are hoarse, but it will have little emotional and lasting impact. Art, on the other hand, is truly powerful because it can communicate more directly. Self Reflected shows the collective activity of 500,000 neurons at once, still over 100 million times less complicated than the brain actually is, engaging your brain on a different level.
Self Reflected was made possible by a grant from the National Science Foundation. My work had come to their attention through some other exhibits that I’d had in scientific venues, and they had asked me and my collaborator Dr. Brian Edwards, an applied physicist and artist who is my collaborator on the microetchins, to consider applying for a grant to continue our work together. I think that they saw the potential that I had been talking about, that creating a work of art designed to shift the paradigm of neuroscience illustration away from the typical computer 3D renderings of several neurons or the brain as a whole would be helpful in educating the public as to the important work being done on the Human Connectome Project. We proposed making a huge, 8’ X 12’ multipanel microetching that would animate the neural activity in a slice of the brain at as close to full complexity as possible. I think that they saw not only saw how impactful this final image would be but that through our ambitious plans of researching and creating the technology required to make it that we may also contribute to the scientific understanding of the brain as well. It is an attempt to bridge the gap between the scientific and artistic worlds, a method to inspire a new generation of neuroscientists, and a way to give the average person an opportunity to deepen their understanding of their own brain. As this piece is designed to be as close to what is happening in your own mind as you are looking at a piece of art like it, Self Reflected is your brain directly understanding itself.
Your previous ‘microetchings’ have involved a great amount of complexity. The Self Reflected project has taken this complexity to a higher level. What were the new challenges and new techniques you were able to explore that ended up making this project so complex?
GD: Yes, Self Reflected was probably ten times more challenging in every regard than anything we had attempted before. The challenges were so numerous, tenacious, and at times bordering on the edge of what was even possible that I still sort of can’t believe that we actually pulled it off.
For starters, there was a depth of realism in this piece that we had not yet attempted. Self Reflected is meant to be so highly accurate that even a professional neurosurgeon, neurologist, or neuroscientist would recognize the tiniest details yet still be surprised by things that they may not have considered before. We employed the help of several undergraduate students at the University of Pennsylvania to help us research all of the brain regions in our slice of interest. All of the neural cell types, their sizes, their shapes, images describing them, their connectivities, their action potential firing patterns, etc. All of this data was assembled and catalogued to inform the creation of the final piece.
Where this project really took things into uncharted territory is through depicting the brain’s activity. Microetching allows us to animate a brief window in time by precisely engineering how light reflects off of a gilded surface. In our case, the animations in Self Reflected represent about 500 millionths of a second in actual brain time. Within this fraction of a second, however, exists a tremendous amount of neural activity that is responsible for everything we typically take for granted- our thoughts, perceptions, emotions, consciousness, etc. In order to communicate this point, simply showing the anatomy wasn’t nearly enough. We needed to figure out how to model the brain’s communication patterns so that we could enrich the viewer with this seldom depicted dimension of its function.
The first step was to use all of our assembled data to accurately paint the anatomy of the brain using a combination of physical and digital techniques. Once this was completed, we essentially simplified all of this work into mathematics- the neurons and fragments of axons were turned into a series of points. Using this simplified information, we then painstakingly fed in our researched circuitry information, i.e. which regions are connected to which regions, how fast the action potential velocities are, whether neurons in a region fire as a burst or individually. We worked out a set of custom algorithms that allowed us to impart causality into the relationships between neurons.
This means, for example, that if we were designing a neural “choreography” that would create an animation between the thalamus (a central routing nucleus of the brain) and the motor cortex (a region responsible for organizing and executing how we move), that we would tell the algorithm to select some number of neurons within the starting region, connect that neuron to the designated set of axons that would be connecting the thalamus to the cortex, and then to select the destination neurons. We could control how chaotic the connections appeared to be, how many originating neurons connected to how many destination neurons, etc. Depending on all of the variables we input to describe this step in the choreography, the computer would then assign each neuron and each part of the axon a “timing value”. This step assigns each neuron one of 255 different colors depending on where we want the viewer to see that neuron. For example, if the neuron is red, it is our convention to have that neuron be visible in the microetching when the viewer is standing all the way to the left. As they walk to the right, they would see orange, yellow, green, blue, and finally violet colored neurons.
After hundreds of steps of this choreography, we were left with an image of the brain in bright multicolored hues that represented an 8’ X 12’ map of how information is traveling. In effect, our algorithm helped us to both accurately simulate the brain’s natural connectivity but also to impart a degree of randomness into the process that would have been very difficult to achieve otherwise. We invented this entire process just for Self Reflected.
Once we had these maps, it was a similar process to that we have used before to translate the color of each pixel into tiny lines whose angle would later dictate how light reflects off of that point in the microetching.
Because this piece was so enormous and the dimensions of the materials we use for actually creating the physical artwork are limited in size, we had to tile 25 individually microetched plates together in order to create the final piece. This presented an array of much more challenging problems than we had anticipated.
The difficulty of doing this lies in the repeatability of manual and chemical processes. We create microetchings using photolithography, the process used to make silicon microchips wherein ultraviolet light burns an image through a printed transparency into a UV sensitive polymer called photoresist. Repeating this process identically 25 times for the purposes of Self Reflected was akin to taking an analog film camera, shooting 25 adjacent pictures while manually adjusting the settings on the camera for each picture, taking that film and developing each shot in a traditional dark room environment using wet chemistry while ensuring that the concentrations of everything are precisely controlled from image to image, and then cutting and tiling those 25 shots into one seamless picture.
The line between success and failure at this point is maddeningly thin. If the etches across plates don’t line up or one plate is made under different conditions from the others, this discrepancy will be immediately apparent and the effect of the microetching being one seamless piece will be lost. On top of this, when we are gilding the plates the application of the gold leaf sizing (glue) needs to be precise to within several micron accuracy in order for the microetching to be seen from all angles, and to be the same across 25 plates. We are talking about a handmade process whose tolerances are the merest wisp of a shaving off of a hair. This process was so difficult in fact that we had to completely remake the piece after having completed 23 of 25 plates after discovering that the visibility of the microetching from acute angles was unacceptably blotchy.
We then had to figure out how to seamlessly tile 25 finished, gilded plates of steel next to one another. After having attempted several methods we learned how to cut down the plates to within several thousands of an inch accuracy, and then to use powerful magnets to pin them all in place so that the whole array looks like a seamless piece of art.
Finally, lighting the piece was a totally new process because we had decided that we wanted a moving light source (microetching animations can be activated either by the moving viewer or a moving light source). Brian ended up designing and building a massive, 24 foot long array of 144 ultra bright LEDs along with banks of colored lights that could be programmed to fade into and out of one another. When the viewer is standing at the mathematical “sweet spot” where all of the reflections off of the microetching are aimed, this light source animates the piece by sweeping a dart of light across the LEDs again and again. This repeats the 500 microseconds of brain time indefinitely so that the viewer can take their time to absorb the massive amount of complexity and detail going on within that brief time frame.
You are trained in neuroscience. Did you find yourself encountering things you didn’t know about the brain as you created this work? How has the work changed your own conception of the brain?
All images courtesy and copyright of Greg Dunn
Get the Full Experience
Read the rest of this article, and view all articles in full from just £10 for 3 months.