Working with Harvey Fishman and other colleagues from Stanford University, Bent prepared a patterned array of electrodes with a neuron growth-promoting protein called laminin, then seeded it with retinal ganglion cells from rats. After 48 hours, tentacle-like growths from the cells were found to have followed the patterns of the laminin.

The group then applied signals to these electrodes, to show that the cells could be stimulated at power levels an order of magnitude less than previously required. That's important because in a device it reduces the chance of a signal leaking out and affecting neighboring cells, says Bent.

Not everyone is convinced that such controlled and directed growth of bipolar cells will be possible in a live implant, however. While the idea of using dendrite growth is a good way to bridge the gap between electrode and neuron, Peter Fromherz, head of the Department of Membrane and Neurophysics at the Max Planck Institute for Biochemistry in Germany, believes the complex patterns needed to keep the electrodes separate would quickly get swamped.

Furthermore, there'd be the issue of keeping the cells alive under such conditions, says Eberhart Zrenner, director of the Department for Pathophysiology of Vision and Neuro-Ophthalmology at the University Eye Hospital of Tübingen, in Germany. "They need a proper oxygen supply," he says, as well as nutrients.

Now Bent's group wants to apply this approach to stimulate cells using chemicals instead of electricity. This would involve using microfluidic channels instead of electrodes to squirt neurotransmitter chemicals at nerves or their dendrites to trigger them. Several research groups are now turning to this approach because it's a natural progression, says Bent. "It is one step closer to what happens in nature," she explains.