Recent experiments in neuroscience suggest that cortical networks at many scales may be operating near a critical point, much like that seen in a second-order phase transition. In this talk I will review experimental data from several labs, including ours, which suggest this. I will also explain why information transmission, information storage and computational power are all expected to peak at the critical point in neural networks. Finally, I will discuss open questions raised by this "criticality hypothesis." These questions could be addressed now that we are on the verge of recording activity from hundreds of connected neurons.