Dimensions underlying the representational alignment of deep neural networks with humans