Given humans' habitual use of screens, they rarely consider potential differences when viewing two dimensional (2D) stimuli and real-world versions of dimensional stimuli. Dogs also have access to many forms of screens and touch pads, with owners even subscribing to dog-directed content. Humans understand that 2D stimuli are representations of real-world objects, but do dogs? In canine cognition studies, 2D stimuli are almost always used to study what is normally 3D, like faces, and may assume that both 2D and 3D stimuli are represented in the brain the same way. Here, we used awake fMRI of 15 dogs to examine the neural mechanisms underlying dogs' perception of two-and three-dimensional objects after the dogs were trained on either a two-or three-dimensional version of the objects. Activation within reward processing regions and parietal cortex of the dog brain to 2D and 3D versions of objects was determined by their training experience, as dogs trained on one dimensionality showed greater activation to the dimension on which they were trained. These results show that dogs do not automatically generalize between two-and three-dimensional stimuli and caution against implicit assumptions when using pictures or videos with dogs.