The neurons of the mammalian visual cortex perform a remarkable transformation on the visual input they receive from the lateral geniculate nucleus (LGN). The neurons of the LGN have circularly symmetric receptive fields and are unselective for the orientation and direction of motion of visual stimuli. Neurons in the cortex, on the other hand, are exquisitely sensitive to these stimulus properties and will only fire in response to a narrow range of stimuli. A long-standing debate centers on the synaptic circuitry that brings about this dramatic change in the representation of the visual image that takes place at the geniculocortical synapse. Hubel and Wiesel originally proposed in their compellingly simple model that the arrangement of the geniculate input determines orientation selectivity. Yet there are several properties of cortical neurons that Hubel and Weisel's model cannot explain, and that have prompted the development of more complex models that require nonlinear interactions among cortical cells. We have examined the orientation selectivity of the aggregate synaptic input to cortical cells from the LGN by selectively inactivating the cortical circuit, either by cooling or by electrical stimulation of inhibitory circuits. These experiments indicate that the LGN provides a well-tuned excitatory input to cortical simple cells, an input that accounts for approximately 30% of the total synaptic drive. The remainder presumably arises from other cortical neurons and might help to establish the contrast invariance of stimulus selectivity. How this is accomplished remains to be determined.