Researchers at the University of Michigan are studying feline brain cells as a model for a new generation of supercomputers that can process and recognize information in a similar way as humans. Such brainy computers will hopefully be able to accomplish more simultaneous processing and complex decision making. Microchips in conventional computers usually rely on transistors that switch on and off to represent data in binary code. The new technology that is being developed at the Univeristy of Michigan instead use "memristors," which are circuit elements that can actually remember information. For example, when you turn the voltage off to the device, memristors retain information about how much voltage had been applied and for how long. A parallel can be drawn between memristors and the synapses beween brain cells (neurons) because they too 'remember' information about the strength and timing of electrical signals from the neurons.
In a conventional computer, logic and memory are located in different parts of the circuit and each element is only connected to a few neighboring elements, resulting in a linear form of operation. Such linear processing allows computers to perform simple tasks, but makes multitasking difficult. Brains obviously can perform many different computations at once, and the goal for the memristor devices is to mimic the interconnected nature of the brain. The researchers devised a paradigm similar to how neurons are connected and connect two circuits through a memristor. With this technology they hope to make a computer brain that is about as smart as a cat (sorry dog lovers). From then they hope to build an even bigger system containing hundreds of artificial neurons connected by memristor synapses.