The visual cortex and convolutional neural networks both have local receptive fields and connectivity, but the visual cortex does not have translational invariance in network weights. The brain has its own unique processing principles, but engineering tricks can still be applied to recreate its function.