Chicks were trained to discriminate between two identical boxes on the basis of their position. Subsequently, the colour of parts of the positive (reinforced) box was changed and chicks were retrained. Results showed that chicks were more or less impaired during retraining depending on the spatial distribution of the changed stimuli. Chicks behaved as if a figure (a disc or a spot of dots) painted on a box was irrelevant to them, whereas they did respond to changes in the colour of a uniformly coloured box or of scattered dots painted on a box. Similar results were obtained in simultaneous discrimination learning tasks involving addition of cues (e.g. colour plus position). Addition of cues facilitated learning using boxes the same colour all over or with painted scattered dots, but not using boxes with a disc or a spot of dots. Furthermore, addition of shape and position information had different outcomes depending on the use of three-dimensional objects or of painted figures: learning facilitation occurred only using three-dimensional objects. Results are interpreted in terms of an “object hypothesis”, and the validity and usefulness of traditional terms such as cues is questioned.