Past work

The World Through The Eyes of a Snake (That Can Fly!)

I’ve spent many of my days effectively trying to hack the brains of flying snakes (genus Chrysopelea). Nervous systems are incredibly powerful, complicated dynamic control systems, that require sensors and feedback loops to detect and adjust internal and external statuses. In these series of projects, I used behavioral neuroscience and gaming development platforms like Unity to explored about snake vision, and eventually figure out how snakes control their glides using vision. There are a few key questions I’m trying to answer.

What are they capable of seeing?
What are some visually-driven behaviors?

What are they capable of seeing?

I’m wildly curious about what a flying snake’s visual system can actually do. In other words, if I were to replicate their flying behavior with a cool flying snake robot, how fast would the visual system need to process input? And how high a resolution would their cameras have to have? I ask this question of biological systems, using the Optokinetic Nystagmus.

If you’ve ever driven by some woods and tried to follow the trees as they passed, or if you’ve looked at a train and tried to follow the windows as they passed, then you’ve done the Optokinetic Nystagmus (OKN). (If you want to try it for yourself, check out this video!)

Optokinetic Nystagmus (OKN) consist of two parts:
     1.   A smooth pursuit with the eye and/or head, followed by
     2.   A rapid snap (or saccade).

Nervous systems use OKN try to keep the visual scene image held fixed on the retina for as long as possible. The longer the scene is held steady, the more time the visual system has to glean information from the scene. If an object is too small (spatial), too uniform (contrast) or moving too fast (temporal) to identify, and the eye has a harder time tracking, and the strength of the reflex lessens.

By changing the size and speeds of the visual scene and comparing the pursuit phase to the rotational speed, I calculated the visual system limits (and sensitivities!) to both resolution and processing speeds.

Rig with snake-01
Optokinetic drum is shown with snake inside. The setup consists of two nested drums. The outer drum is gray, open-top cylinder that sits on a yellow turn table. On the inside of the gray drum is a black and white visual stimulus. The inner drum is an optically clear acrylic, open-top cylinder that stays fixed when the outer drum rotates. This way, the snake, placed inside the cylinder on a Petri dish (white) will see the rotating scene, but not rotate itself. Magical!

These snakes have responded to visual rates of up to 99 Hz, and have a spatial resolution that would probably allow them to see a matchstick on the ground from about 2 stories up (~2.5 cpd or 20/300 Snellen vision). Humans would be able to see the same matchstick from about one story higher, but we can only see up to about 50 Hz.

But what about their field of view? To figure out how many degrees flying snakes can see around their head, my co-authors and I 3D scanned a snake head, and turned it into a 3D CAD object.

I fit the eyes with spheres, and then projected rays from each point on an encompassing sphere. This is effectively the “world” and light is getting cast from each point on the globe toward each of the eyes. Any place where the ray of light hit the eye spheres first meant that’s light that can potentially reach the retina. With this, we could recreate the visual sphere, with binocular and blind areas, like in the image below:

3D (top row) and top-down (bottom row) representation of visual fields of different raptors. (Pecten is the feathery comb around the beak.)

You can read all the details of this work in our article, “Visual acuity in the flying snake (Chrysopelea paradisi)” (Oct, 2020).


What are some visually-driven behaviors?

Given how fast they can see, and their ability to see color, I was curious to know what they use their vision to do. I observed a number of different behaviors in the optokinetic drum:

OKN – smooth pursuit and saccades, with the smooth phase varying with visual rotational velocity.

Head wagging – a lateral oscillation of the head

We already had a good idea of what OKN was used for–keeping visual scenes steady to get as much information from them as possible, but the head wag, that was a stumper.

Head wagging is performed by lots of animals for a number of reasons. Take this stick bug for instance (Order Phasmatodea). This popular Internet dance is actually helping the stick bug measure distance of items around it.

A stick bug is seen on top of a surface. The angle starts out profiling the stick bug, then swings around, to reveal a comical-looking oscillation of the stick bug as it leans dramatically from one side to the other. This oscillation displaces the head wider than eye width, allowing the stick bug to induce motion parallax, and derive distance information from its visual world.

By swaying back and forth, the stick bug perceives closer things to it as objects that move quickly, and things farther from it as objects that move slowly. This induced phenomena is called motion parallax.

Chrysopelea heads are pretty small. It could be that their eyes are too close together to get meaningful depth perception, and so head wagging induces motion parallax and gives them the deets on distance.

This scrolling scene demonstrates how motion parallax informs us of distance. White clouds are shown in the background, moving slowly. Slightly faster, in the mid ground, move orange and blue-speckled rounded posts. In the foreground, a green bush zips by, fastest of them all. Together, an effect of depth can be experienced, even with a 2D simplified drawing!

Other animals, use head wagging too, but for a whole different reason. The garter snake (Thamnophilis sirtalis), has also been observed to “wobble” its head back and forth, but this dance is all about fitting in…literally.

A dark colored garter snake is seen in the center of the image, facing the top of the shot. As it moves upward and to the right across the screen, it “wobbles” its head left and right quite obviously. Overall, it is difficult to see the snake amidst its background of leaf litter and dry grass. Dr. Bill Ryerson posits that this behavior is to help it visually blend in with moving, swaying backgrounds (like grass).

The most successful camouflage is one where an organism can blend seamlessly in with its background. If you’re like me, you’ll probably think immediately of stationary camouflage, with impressive actors like chameleons and octopuses. But what if you’re trying to blend in with something that’s moving? Staying still means you’ll stick out and make an easy meal for a vigilant bird. Animals may engage in this head wagging behavior as a type of behavioral camouflage, which helps the animal blend in with swaying, moving background such as leaves or grass in the wind (Ryerson, 2017).

Head wagging, therefore can serve two possible functions:
1) It can help the flying snakes gauge distance by inducing motion parallax.
2) It can help the flying snake blend in with moving background.

I’ve figured out the answer! And will be publishing this research soon!