The mental rotation of visual features and feature conjunctions

When we mentally rotate a coloured object, that object’s colours can introspectively seem to remain ‘stuck’ to the object’s shape without any additional effort on the part of the subject. Here, we use an objective measure to provide evidence that this introspective impression is, in fact, correct. We compared participants’ ability to rotate complex shapes, whose parts are differently coloured, with their ability to rotate identical monochromatic shapes. Our results indicate that colour can remain 'glued' to shape parts for free, i.e., without implying any cost to mental rotation performance. These results are discussed in relation to ongoing debates over the format of visual imagery. Our results further suggest that the human brain can produce ‘inextricably bound’ representations of multi-feature objects, wherein the manipulation of one feature implies the manipulation of others. We consider some possible ramifications for the study of conscious experience as well as for deep-learning models of human cognition.