Insect senses suggest novel neural networks

Insect senses suggest novel neural networks. Animals gather information about their environments as sensory neurons generate minute electrical signals in response to chemicals, light, sounds, and other stimuli. A new model of neural networks, based on recent studies of insect olfactory systems, suggests that neurons can be linked in a way that allows them to identify many more stimuli than was previously possible. Researchers from the Institute for Nonlinear Science at the University of California, San Diego, propose that one neuron is able to delay the firing of another neuron. This inhibitory capability means that a given stimulus leads to a specific, robust, and reproducible time sequence of neural activity. The researchers used observed features of a locust’s antenna lobe, exposed to fragrances such as cherry and mint, to guide development of the model, which they call competitive networks, or winnerless competition (WLC). They found that, with a network built of N neurons, their model could identify roughly (N − 1)! different stimuli. That is, an 11-neuron WLC network should be able to identify more than 3 million items. The WLC model helps explain why a rose, by any other name, would smell as sweet—but wouldn’t smell like an onion. Ultimately, the WLC model may lead to high-capacity, potent computing networks that resemble an insect antenna or a human nose more than a desktop PC. (M. Rabinovich et al.., Phys. Rev. Lett. 87, 068102, 2001 http://dx.doi.org/10.1103/PhysRevLett.87.068102 .)