What does a horseshoe crab look for in a mate? The right moves, according to a computer model that simulates how neurons in the crab eye respond to different objects. These neural images, reported in the current issue of the Proceedings of the National Academy of Sciences, reveal what a crab sees in its mind's eye as it scans for mates, and they offer a model of how ensembles of neurons encode visual stimuli for the brain.
Each spring, male horseshoe crabs patrol shallow reaches of the ocean looking for receptive females. Like all animals, the crab's eye does not passively record the world, but emphasizes information important for survival. For example, studies of single optic neurons have revealed that inhibition of certain neurons in the retina allows many animals to better detect contrast, which heightens their ability to see prey against a background. But, up to now, no one has been able to look at the activity in a complete network of optic neurons, says Robert Barlow, a neurobiologist at the Marine Biological Laboratory in Woods Hole, Massachusetts, and the State University of New York Health Science Center in Syracuse, New York.
Barlow and his colleagues simulated the neural network by creating a computer model of the horseshoe crab's eye. The model mimics nerve signals as they travel from light receptors in the eye through the 1000 optic nerves to the crab's brain. The researchers tested their model for the first time in a natural setting by using a mini-video camera, called the CrabCam, mounted on a crab's back as it explored an underwater habitat. They also wired up a single neuron from the crab's eye and recorded its activity as the crab moved around. Researchers had placed black and gray cylinders, which mimicked female crabs' size and contrast, in the area.
Back in the lab, they digitized the CrabCam footage and "fed" it to the computer eye, which then predicted the activity of hundreds of neurons, resulting in a series of snapshots of the neural images that scientists believe show what the crab actually sees. When Barlow compared the single-neuron recording taken in the field with the computer's predicted activity of that particular neuron over the series, he found a 75% or better match; a similar test in a lab setting produced a greater than 95% match. What's more, "the crab's eyes are tuned to best detect objects about the size of crabs, moving at crab speeds, while ignoring stationary ones," Barlow says.
The research has "taken a step forward to show that we can model more complex and transient" responses, says John Dowling, a neurobiologist at Harvard University. However, model neural networks for a system as complex as the human eye are still far off, he adds.