Predictability of bubbles versus feeding as cognitive enrichment in rainbow trout: Contrasting effects on behavior, brain plasticity, and welfare

Animal welfare in aquaculture depends on both physical health and the cognitive and emotional states of fish. Predictability of positive events, such as feeding, has been proposed as a form of cognitive enrichment, but its effects remain debated. Rainbow trout ( Oncorhynchus mykiss ) are a suitable model due to their strong associative learning skills and anticipatory behaviors. Previously, we found that bubble diffusion was a salient cue when associated with feeding. Here, we tested whether bubbles themselves, when predictable, could act as a reinforcing stimulus and improve welfare compared to feeding predictability. Fish were exposed for 26 days to either: (1) predictable bubbles (fixed times, signalled by light), (2) predictable feeding (fixed times, signalled by light), or (3) unpredictable control (random bubbles and feeding). We measured activity, aggression, sudden accelerations, jumps, emotional reactivity (novel-tank test), growth, fin condition, immunity, plasma cortisol, and neurobiological markers in forebrain and myelencephalon. Fish developed anticipatory activity to the light cue associated with feeding, whereas no anticipatory response was detected for the light cue associated with bubble diffusion, even after 120 cue-stimulus pairings. Nevertheless, predictable bubbles increased activity during diffusion and were associated with differences in dopamine-related metabolites (notably increased telencephalic L-DOPA). Together, the behavioural data suggest increased engagement with the stimulus, while neurochemical interpretation remains limited to metabolite-level measures. Predictable bubbles modulated aggression over time, but without clear effects on fear responses, fin erosion, cortisol or immunity. Gene expression analyses indicated modest, region-dependent changes in neuronal markers under predictable conditions. Overall, bubbles did not generate conditioned anticipatory responses comparable to a feeding and may be limited as cognitive enrichment. However, bubbles remain valuable for physical enrichment or as a non-essential cue predicting feeding in aquaculture. • Fish succeeded in anticipating feeding with a light cue after one month, but not bubble diffusions based on swimming activity. • Dopamine-related metabolite measures quantified at the end of the trial (telencephalic L-DOPA and the HVA/L-DOPA ratio) differed between treatments. • Fish emotional reactivity, growth, fin condition, plasma cortisol levels, and innate immune responses remained unchanged by the predictability condition. • Bubbles predictability does not appear to be an effective cognitive enrichment strategy for improving the welfare of rainbow trout.