The Virtual Dogfish

The Virtual Dogfish is a 3D reconstruction of a spiny dogfish (Squalus acanthias)
There are three versions of the virtual dogfish. Two are used in teaching and one is used as a research tool in computational neuroscience
Virtual Dogfish One demonstrates the sensory ecology of the spiny dogfish - the cues it uses to hunt and navigate in the ocean.
Virtual Dogfish Two illustrates the anatomy of the sense organs and nervous system relevant to hunting and navigation
Virtual Dogfish R1 is used for research into brain mechanisms for target tracking and navigation


virtual dogfish screen shot
Virtual Dogfish One
This version has a simple "AI" brain. It swims aimlessly, but does not like the dark, so keeps returning to the lit region.
It has no sense organs, so it does not detect prey or respond to it.

See a video of  Virtual Dogfish One here

Download the interactive 3D application Virtual Dogfish One  for Windows here.
You can follow or steer the dogfish.

 visible dogfish screen shot
Virtual Dogfish Two
This version contains a detailed anatomical reconstruction of the special sense organs, nerves and brain of the spiny dogfish.
It has a "vestibulo-ocular reflex" that stabilizes its eyes when it swims, but this is implemented using computer animation techniques, not a computational neural model.
It is used to teach vertebrate neuroanatomy, and to introduce students to the dogfish as a  model system in sensory-motor systems neurobiology and neuroethology.

See a video of  Virtual Dogfish Two here

Download the interactive 3D application Virtual Dogfish Two for Windows here.
You can explore the neuroanatomy of the dogfish.

virtual dogfish R1 screen shot
Virtual Dogfish R1.
This version contains computational models of the physics of the environment, sense organs, sensory neurons and the brain.
It is used in research to develop and test theories about how the brain makes inferences from sense data.
Inference about prey location from electic field and water movement is thought to have been a key driver for brain evolution in the first animals,
during the Cambrian explosion 540 million years ago (ie that's what I think).  So we are asking fundamental questions about how brains work,
with implications for understanding the human mind and for building autonomous robots.

This screenshot shows neural action potentials being recorded by an electrode in the main electrosensory nerve

See a video demonstration of  Virtual Dogfish R1 electrosensory prey detection here

See a video demonstration of  Virtual Dogfish R1 magneto-electric navigation here

R1 is not available for public download (it is complicated, undocumented and unstable). But if you are a developer or a scientist interested in using it, email me (Mike Paulin).


HOME