Research Progresses on Neural-Silicon Hybrid Circuits

by Warren Grill, senior technical editor

Hybrid circuits, which combine the silicon circuitry traditionally used for computers with cultured networks of neurons, provide a new approach to sensing and computing. Rather than relying exclusively on the digital computing power of large numbers of silicon transistors, such hybrids seek to exploit the enormous computational power present in small neural networks.

Such biological networks are envisioned to have the capacities to adapt and self-repair. Further, the exquisite sensitivity of membrane-bound receptors has prompted the development of cell-based sensors. In an early example, Peter Fromhertz and colleagues from the Max Planck Institute of Biochemistry, Martinsried, Germany, demonstrated that neurons cultured over the gate of transistors can implement a switch controlled by neuronal activity.

Now Itay Baruchi and Eshel Ben-Jacob from Tel-Aviv University in Israel have demonstrated that memories can be imprinted in a network of cultured neurons. They cultured rat cortical neurons on commercial planar arrays containing 60 individual electrode contacts (MultiChannel Systems, Reutlingen, Germany). The cultured neuron system enabled repeatable and systematic recording of spontaneous burst firing activity that propagated across the distributed network of neurons. Baruchi and Ben-Jacob found that chemical stimulation produced new patterns of burst activity propagating in the cultured network, and that the new patterns of activity were persistent. Published in Physical Review E, their findings were recognized by Scientific American as one of the most significant scientific discoveries of 2007.

Chemical stimulation was achieved by delivery of 10 µl of picrotoxin using a micropipette positioned above one of the neurons in the culture that overlay an electrode. Picrotoxin is an antagonist of the inhibitory neurotransmitter GABA (gamma-aminobutyric acid), which renders the inhibitory synapses between the cultured neurons less effective. Calculations indicated that the concentration of picrotoxin dropped ten-fold within only 5 µm of the microelectrode, suggesting that the effects were localized to a very small region within the culture.

When activated by chemical suppression of inhibitory synaptic transmission, the cultured networks generated new and persistent patterns of activity. “The main achievement was the fact that we used the inhibition of the inhibitory neurons” to stimulate the memory patterns, said Ben-Jacob. In contrast, previous efforts to cause persistent changes in network activity in cultured networks used either electrical or chemical stimulation of excitatory neurons, which often resulted in an increase in the overall activity in the network, rather than the emergence of new, persistent, organized patterns of activity.

Delivery of chemical stimulation three times per minute for 20 minutes led to new patterns of activity that persisted for as long as 40 hours after stimulation. Remarkably, the network continued to generate the same patterns of spontaneous burst activity that were evoked prior to chemical stimulation. Thus, the new sequence of activity coexisted with the activity that was spontaneously generated in the culture. A day later, a third pattern could be introduced by chemical stimulation of a different part of the culture. Again, the third pattern was able to coexist with the other two patterns of bursting activity. “The surprising thing is it doesn’t affect the other patterns that the network had before,” Ben-Jacob said. These repeatable spatiotemporal patterns of neuronal activity could thus be viewed as primitive memories and represent the potential to program patterns of activity in networks of cultured neurons.