I want to build a neural network. I want to know how young children categorize animals. What features (colour, amount of legs, etc) are they using to identify what the animal is. For example, if they were exposed to a white dog previously, how can they tell that a white cat is not a dog. what do they use as their indicators? I want to have different images such as a different colour dog, a polar bear, white cat, white bunny, etc. I want my model to be a multi-layer perceptron trained using backpropagation, with just 1 input and 1 output layer. I would like the output layer to be "dog".How would I go about doing this given my question and also how would I code the pictures as an input. Is this even possible. What would be my steps in explanation form