Event:

How does intelligence emerge within the brain? How does the brain differ from a computer in terms of the computation principle? To address these fundamental questions, our particular interest is a rich repertoire of spatio-temporal activity patterns that must underlie the computation principle in the brain. In the cerebral cortex, the theory of neural Darwinism predicts that variation and selection within neural populations are crucial to adaptive computation. Using cutting-edge neural interfaces (1)-(4), we attempted to identify empirical traits of the Darwinian principle in both in vivo and in vitro experiments. First, we demonstrated that the degree of response variance is closely correlated with the size of its representational area in the auditory cortex of rats (5). Importantly, both the response variance and representational area increased during the early stage of learning and decreased after the completion of learning, supporting the theory that learning-induced map plasticity is a sign of Darwinian computation. On the other hand, the phase locking of neural responses better predicted our observations than response magnitude, suggesting that neuronal synchronization plays a critical role in selection of appropriate neural activities (6). Second, in a dissociate primary culture of neuronal cells grown over a petri dish (1), we showed that reproducible spatiotemporal patterns emerge in spontaneous activities and the variety of patterns increases as the neural network develops (2). Furthermore, we demonstrated that feedback signals based on ongoing activity are able to extract a coherent output from a chaotic neural network, endowing the network with problem-solving ability. Thus, our results provide additional insights into how the Darwinian principle is implemented in neural networks and serves as an origin of intelligence and problem-solving ability.