Fly, Snail, Fly!

Sea angel. Photo © Alexander Semenov, Belomorsk Biological Station named after N. A. Pertsov

Sea angels are snails that have learned to fly. The shell falls off in infancy, and the angel flaps its former crawling leg like a butterfly flaps its wings to soar in the water column or rush for prey.

Pablo Varona, a colleague of Mikhail Rabinovich in the development of concepts of stable heteroclinic channels and winnerless competition, chose, together with his colleagues, an outlandish mollusk as a model to study the transition from sensation to movement in the nervous system due to the simplicity of its structure and behavior.

Although it is hard to call a simple device gravimetric organs, the statocysts, with the help of which an angel determines the position of its body. Inside the angel, there are two round bubbles, on the surface of which a ring of six statocyst receptors is located. Something like a stone — the statolith — dangles inside the bubble. When the stone touches the surface of the bubble, the receptor closest to the place of contact is triggered.

The signal from the statocyst receptor ring goes to central pattern generators of wings and tail. A central pattern generator is a ring of neurons, which fire chaotically one by one but together produce a stable periodic signal due to rigid hierarchical coupling of their phase orbits according to the “winner takes all” principle.

The signal from the statocyst receptor ring makes central pattern generators change the pattern of movements of wings and tail so that the mollusk maintains an upright position and slowly hovers in the ocean water.

The situation changes when the chemical receptors of the angel detect the presence of prey nearby. The mollusk has no vision, and its chemical receptors cannot determine direction. However, the mollusk knows how to hunt. Upon receiving a signal of the presence of prey the statocyst ring turns central pattern generators of wings and tail to a state of rapid movements along a complex unpredictable trajectory similar to a random walk or to its chaotic sibling — a Levy flight. The angel’s roaming entropy increases dramatically and such a hunt often ends in success.

What happens in the control system of the movement of a mollusk during a hunt?

The coupling of the phase orbits of neurons firing in the rings of central pattern generators becomes less rigid. The dynamics of central pattern generators’ interaction with the statocyst ring go into a winnerless competition mode.

Central pattern generators continue to generate rhythmic movements, but the pace of switching between different rhythms increases. Furthermore, the new regime of movements overrides the coupling between the statocyst ring and the central pattern generators, which maintains the angel’s upright position under normal circumstances.

The hunting angel isn’t chasing the prey. It has no information about the movements of the prey. It simply moves along a chaotic trajectory that maximizes the chances of the intersection of his path with the path of the invisible prey irrespectively of the way the prey is choosing.

I describe in such detail the twenty years old work of Varona because, as it seems to me, it contains the answer to the question of how the autopilot (the reptilian brain) and GPS (the mammalian brain) interact in the human brain.

It turns out that by default under normal conditions the autopilot completely takes over the control of the movement. When a signal about a change in the environment arrives, the GPS takes control over the piloting by broadening dramatically the repertoire of available patterns of movement and the pace of changing them.

That, in fact, is all. If you overtrain the autopilot using Reinforcement Learning, the GPS function simply won’t turn on. And we will not be able to catch the prey for the mind, new knowledge, the presence of which we can only subtly feel, but we can neither see nor touch it.

References:

1. Latorre R, Levi R, Varona P (2013) Transformation of Context-dependent Sensory Dynamics into Motor Behavior. PLoS Comput Biol 9(2): e1002908. https://doi.org/10.1371/journal.pcbi.1002908
2. David W Sims, Nicolas E Humphries, Nan Hu, Violeta Medan, Jimena Berni. Optimal searching behaviour generated intrinsically by the central pattern generator for locomotion. eLife 2019; 8:e50316 DOI: 10.7554/eLife.50316
3. Joseph Klafter, Michael F. Shlesinger, Gert Zumofen (1996) Beyond Brownian Motion. Fractal generalizations of Brownian motion have proven to be a rich field in probability theory, statistical physics and chaotic dynamics. Physics Today 49, 2, 33 (1996); https://doi.org/10.1063/1.881487
4. V. Zaburdaev, S. Denisov, and J. Klafter (2015) Lévy walks. Rev. Mod. Phys. 87, 483 — Published 9 June 2015, DOI: https://doi.org/10.1103/RevModPhys.87.483
5. Karl Friston. Competitive dynamics in the brain: Comment on “Information flow dynamics in the brain” by M.I. Rabinovich et al. Physics of Life Reviews Volume 9, Issue 1, March 2012, Pages 76–77. https://doi.org/10.1016/j.plrev.2011.12.006