Data and code for "Wild fish continue to discriminate human divers using subtle cues"

Wild animals make decisions in contexts where cue availability and reliability can change over short timescales. This problem becomes acute when animals interact with humans, for whom identity cues may vary with clothing, equipment, or hairstyle. We tested whether wild fish, Mediterranean saddled seabream (Oblada melanura) and black seabream (Spondyliosoma cantharus), could continue discriminating between human divers after overt visual differences in diving equipment were removed. Fitting a Bayesian kinked binomial mixed model to per-trial choices, we estimated the immediate step at cue removal and the rate of learning either side of it. Black seabream showed an immediate decrease in performance when divers switched from different gear to identical gear, indicating that they relied heavily on obvious cues during initial discrimination, but were able to update and use more subtle cues to discriminate among humansdressed identically. Saddled seabream showed a similar trend, but naturally low sample sizes in this wild experiment precluded strong inference. These results show that these fish can continue to discriminate between two divers even when external visual cues are removed, potentially due to inherent features like hair colour, and that trial-level trajectories can distinguish immediate disruption from subsequent behavioural recovery in wild cognition experiments.