Self Reflected was designed to be a highly accurate representation of a slice of the brain and is informed by deep neuroscience research to allow it to function as a reliable educational tool as well as a work of art. In the future we will have educational kits available with images, video, and explanations suitable for instructional purposes. If you have interest, thoughts, or questions, please email me.
The Research:
After considerable discussion we settled on depicting an oblique sagittal slice of the human brain as this view provides a great deal of variety in structures, interesting circuitry, and avoids the ventricles (fluid filled cavities in the center of the brain) that would simply have appeared as holes in the brain in this view.
- Location of the slice (series of images on the left, image courtesy Dr. John Pyles, Carnegie Mellon University). The green line indicates the position of the slice.
- The oblique sagittal section used as a reference isolated from the Big Brain dataset from McGill University by Dr. Craig Stark, University of California at Irvine.
In order to guide the piece with the latest neuroscience research, two neuroscience undergraduate students at the University of Pennsylvania, Melissa Beswick and Carl Wittig, assisted in collecting data for each of the regions in the section from the primary literature. These data contained information about each region and its location, its neuron types, sizes, what other neurons and regions they are connected to, firing patterns, neurotransmitter type, etc. This research served as the scientific basis for the further development of the art and computation.
- All gray matter (regions that have cell bodies in them) in the slice simplified to a cartoon.
- A later iteration of this map with the cortex separated into its six functional layers.
We consulted with many expert neuroscientists and neurologists to ensure the accuracy of the neuroscience behind the work. There are several instances wherein structures were moved either into or out of the plane of the piece to clarify certain points or to complete a neural circuit.
Caudate
If you’ve ever restrained yourself from eating a second slice of delicious cake, you’ve exhibited inhibitory control and therefore engaged your caudate nucleus. The caudate nucleus is located deep in the frontal part of the brain and is also responsible for various types of learning, such as the type that allows you to play the piano or ride a bike. The substantia nigra has many dopamine neurons that connect to the caudate.
Cerebellum
The cerebellum can be found at the back of the skull, forming its own distinct structure. It contributes to coordination and balance, so someone with damage to the cerebellum might have trouble walking or dancing. Most of the cerebellum is comprised of 3 layers of different types of neurons, but the most distinctive cells are the Purkinje cells. These serve as the sole source of output from the cerebellum.
Cortex, Frontal
As its name suggests, the frontal cortex is found at the front of the brain. It is separated from the parietal lobe by a deep groove known as the “central sulcus”. The frontal cortex contains the motor cortex, which controls voluntary movements. The portion of the frontal cortex found at the very front of the brain is important for making decisions and acting in a way that is deemed acceptable by society. If you’ve ever stopped yourself from shouting at someone when angry, you’ve activated parts of your frontal cortex.
Cortex, Motor
Every time you move, you are activating your motor cortex. This region is located right next to the somatosensory cortex, but it is closer to the front of the brain. The motor cortex can be broken down into three regions which all work together to plan and execute voluntary movements.
Cortex, Parietal
The parietal cortex refers to the large lobe that comprises the upper back portion of the brain. This large region has many functions, such as integrating sensory information and determining the location of objects. The parietal lobe plays a key role in allowing you to read this paragraph; it is involved in language processing.
Cortex, Somatosensory
If you’ve ever jerked your hand away from a hot stove, your somatosensory cortex has saved you from burning yourself. This brain region processes a variety of types of sensory information, such as temperature, touch, pain, and the presence of various chemicals. It is found along the surface of the brain, forming a strip that goes from the left side to the right.
Cortex, Visual
Though our eyes typically get all the credit for vision, the visual cortex is what really allows us to see. Visual input from the eyes goes to the thalamus and ends up in the visual cortex, a small region found at the back of the brain. The visual cortex is characterized by six layers which help us pick apart visual information like color, motion, and shape.
Inferior Olive
The inferior olivary nucleus plays a role in coordination and motor control, so it makes sense that it’s located near the cerebellum. The ION projects to the cerebellum but it also receives input from the cerebellum, creating a connectivity loop. The ION also receives information from the somatosensory cortex. If you have damage to your ION, it might be hard for you to accurately throw a ball at a target.
Locus Coeruleus
The locus coeruleus is in the brainstem and facilitates stress and panic-related responses. If you’ve ever had an adrenaline rush, you’ve felt your LC at work. The norepinephrine (the adrenaline) that the LC produces has an excitatory effect on most of the brain, so the effects of the LC are far-reaching. The LC receives projections from the amygdala, which is also involved in the fear response.
Pons
The pons is a part of the brainstem that carries signals down from the cortex to the cerebellum and medulla, acting as a bridge (pons means “bridge” in Latin). It also carries signals to the thalamus from the brainstem. The pons is home to cranial nerves which play roles in hearing, maintaining equilibrium, feeling touch and pain in the face, chewing, swallowing, and many other sensations.
Inferior Colliculus
The inferior colliculi facilitate many aspects of hearing. This duo integrates a wide range of auditory information and can help you locate a sound in space. When you can’t see someone but you still know where they are because you can hear them, that’s evidence of your IC at work. There are four main pathways within the IC: an ascending pathway, a commissural pathway (it connects the right and left inferior colliculi), and two descending pathways that both end up at the cochlear nucleus.
Mammillary Bodies
What was the happiest day of your life? Your mammillary bodies are responsible for helping your remember it. These two clusters of neurons are found on the bottom part of the brain. They are part of a pathway from the amygdala and hippocampus to the thalamus.
Nucleus Accumbens
The nucleus accumbens can be found in the basal forebrain, which means it is at the bottom portion of the front of the brain. Different regions of the nucleus accumbens do different things; some parts are involved in maternal behavior, while others play important roles in rewarding and reinforcing stimuli as diverse as eating an ice cream cone to taking cocaine. Most of the cells that make up the nucleus accumbens are medium-sized spiny neurons.
Olfactory Bulb / Olfactory Tubercle
The olfactory bulb and olfactory tubercle work together to help you smell. Smell is important in processing rewards. The olfactory tubercle in particular is characterized by crescent-shaped cell structures known as the “islands of calleja”. The olfactory bulbs in animals like rats are much larger than in humans since smell is not as important to humans.
Raphe
The raphe nuclei are a collection of densely-packed clusters of neurons found in the brainstem. They play a key role in well-being; they are responsible for releasing serotonin, a neurotransmitter that influences the way we humans feel happiness. The nuclei are somewhat isolated from each other and project to a diverse range of brain regions. The raphe nuclei are one portion of the reticular formation.
Reticular Formation
The reticular formation is found in the brainstem and is divided into three main columns, one of which is composed of the raphe nuclei. The raphe nuclei plays an important role in the experience of happiness, while the other two columns take part in motor coordination and exhalation. More broadly, the RF is thought to be important to consciousness and awareness.
Substantia Nigra
The substantia nigra is found in the midbrain and is darker than nearby areas, which is why its name is Latin for “black substance.” The substantia nigra plays important roles in learning and addiction. If you’ve ever had a really good pastry and craved it later on, your substantia nigra probably had some say in that. There are two subsections of the substantia nigra; one has neurons that respond to GABA while the other section’s neurons respond to dopamine.
Superior Colliculus
The superior colliculus takes visual input, transports it through the brain, and transmits signals to muscles to respond with appropriate movements. There are two superior colliculi, both of which sit below the thalamus and above the inferior colliculus. The surface-level layers of this region receive input from the visual system, while the deeper layers tend to receive input from motor-related brain areas (to which they also project). You can thank your SC for allowing you to quickly catch that glass of juice you almost spilled.
Thalamus
The thalamus is the relay center of the brain; most sensory and motor signals go through the thalamus before they reach the cortex. The only sensory system that doesn’t interact with the thalamus is the olfactory system. The thalamus’ location in the center of the brain is unsurprising given its important role and connectivity to many brain regions. The thalamus is comprised of two bulb-shaped masses located on each side of the brain.
VTA
The ventral tegmental area is one of the most emotional regions of the brain. It’s responsible for intense emotions of love and for things like drug addiction. It is found in the midbrain and projects to the areas like the prefrontal cortex and the amygdala. The majority of neurons in the VTA respond to dopamine, which makes sense since dopamine is important in reward prediction and response.
To collect data to inform the layout and construction of the white matter, the complex, threadlike collections of axons that make up about half of the brain, we collaborated with scientists at Carnegie Mellon University’s Brain Science Institute. We used Diffusion Spectrum Imaging (DSI) data from Dr. John Pyles’ brain to render and filter a map of what the white matter looks like in our slice of interest. Each bundle of axons, or track, was rendered and filtered by Kevin Jarbo, a PhD student at CMU.
- Dr. John Pyles entering the MRI scanner that will use the DSI technique to image his brain’s white matter.
- A render of the white matter tracks that are within our slice of interest.
As this data could not directly be used as is for our purposes, it became a scaffold off of which we based hand drawn vector images that would function as the white matter in the final piece. As the algorithm that we used to compute all of our neural “choreographies,” or circuits, required very specific inputs we generated dozens of datasets like these below that functioned as the white matter:

The most challenging part of this project was taking the neuron and axon datasets and algorithmically combining them to choreograph the neural circuits. As microetching allows the display of a third dimension of information based on angle of reflectivity, we assigned this dimension as time so that the viewer could perceive how action potentials propagate through the brain. As the viewer walks from one side of the piece to the other, or as a light moves over the piece, these neural circuits animate. For detailed information on how the algorithm works and how it simulates neural activity, please see the Engineering section.
Based upon our research and consultations with experts, we programmed and directed the algorithm to build neural circuits that propagate through the brain in a simulation of how it actually functions.
- An example of how circuits are programmed. Each horizontal line is one step in a choreography, and all columns are variables controlling such things as action potential speed, whether a population of neurons fires as a burst or individually, how thick axons are and how much randomness there is in their appearance, etc.
- A simplified diagram that allows for the visualization of how information is flowing throughout the brain from region to region.
We are working on a video of the Self Reflected circuits here. Come back later for the update.
Cerebellum-thalamus-motor cortex-brainstem
VTA- nucleus accumbens
Visual system- LGN to V1, etc.
Frontal cortex to caudate, GPi, thalamus, STN
Feedback loops in somatosensory and parietal cortex
Olfactory bulb to tubercle and mammillary bodies
There were several characteristics about the brain’s functioning that became apparent through our analysis of its structure. First of all, the brain’s characteristic, wave-like neural firing patterns (think: brainwaves) emerge from its layout in space. As adjacent neurons are often wired together and connect out of their immediate region through long range axons, if a signal is propagated from a small number of neurons outward this activity will manifest as wave-like behavior. Local processing circuits with interneurons and recurrent connections add noise and fidelity to this system.
The three dimensional construction of the brain is paramount to its functionality. The brain has taken advantage of this naturally emerging wavelike activity to sequence processing steps efficiently. Of particular interest to us were the strategies that the brain has used to evolve to address the problems of how to arrange neurons and axons along three dimensional, strongly convoluted structures. There is a natural bunching of neurons and axons that occurs in a concave curvature as the space in which axons can travel becomes constricted. In contrast, in a convex curvature there is a reduction in axon bunching as a natural consequence of the geometry. These local minima and maxima are laid out in three dimensional space and encourage dramatic twisting of axon bundles to thread through areas where there is room made by convex curvatures. As every square micron of space in the brain is utilized, it is very densely packed and the pressure of developing regions undoubtedly crowds structures together in space. This tendency was most apparent in the cortex and cerebellum, structures characterized by a large amount of surface area relative to their volume. Indeed, these regions proved very challenging to paint and choreograph as flattening a highly three dimensional structure into two dimensions required much re-imagining and artistic license in order to realize.
The brain has evolved several strategies that it applies to multiple regions. For example, the cerebellum, olfactory bulb, superior colliculus, lateral geniculate nucleus, olfactory tubercle, and cerebral cortex have all adopted a layered structure that vertically organizes processing circuitry while also allowing horizontal connections to take place. This provides more symmetrical and systematized access to local information than do spherical or elliptical structures. As opposed to many midbrain structures whose organization is largely dispersed, the aforementioned structures (which tend toward the periphery of the brain) have adopted this layered structure for one reason or another, presumably in order to maximize processing efficiency.
Interestingly, this layered structure in the cortex can give rise to very different circuitry depending on the size of cells, their connectivity, the number and types of interneurons present, etc. The relatively simple organization of primary visual cortex, for example, propagates distinct wavelike activity throughout higher order visual cortex as the number of interneurons tends to be low relative to other areas of cortex. The frontal cortex, in contrast, has a very large number of interneurons that feed back onto the primary output neurons which cause this region of cortex to be “noisy” relative to the more wavelike activity of other cortical regions.
Finally, it was interesting to speculate that natural variations in the way that the brain develops and the consequent impacts on how neurons connect with one another has a degree of randomness to it that is outside the realm of genetic control and organization. As the structure of every neuron, every axon, and every region is affected by their times in development, what they are surrounded by, and what types of information they are given to refine their own behaviors, even genetically identical brains would contain a large amount of variation in how they wire up due to nothing more than random variables. This is a factor essentially outside the consideration of epigenetic or even environmental concerns. The brain’s stunning order amidst this randomness is one of its greatest marvels as it builds in natural variability and flexibility in function that can achieve a much higher degree of sophistication. The brain, like many natural systems, reaches its maximum potential when teetering on the edge of utter chaos.