The path to colour discrimination is S-shaped: behaviour determines the interpretation of colour models.

Most of our current understanding on colour discrimination by animal observers is built on models. These typically set strict limits on the capacity of an animal to discriminate between colour stimuli imposed by physiological characteristics of the visual system and different assumptions about the underlying mechanisms of colour processing by the brain. Such physiologically driven models were not designed to accommodate sigmoidal-type discrimination functions as those observed in recent behavioural experiments. Unfortunately, many of the fundamental assumptions on which commonly used colour models are based have been tested against empirical data for very few species and many colour vision studies solely rely on physiological measurements of these species for predicting colour discrimination processes. Here, we test the assumption of a universal principle of colour discrimination only mediated by physiological parameters using behavioural data from four closely related hymenopteran species, considering two frequently used models. Results indicate that there is not a unique function describing colour discrimination by closely related bee species, and that this process is independent of specific model assumptions; in fact, different models produce comparable results for specific test species if calibrated against behavioural data.