Welfare metrics have been established for octopuses in the laboratory, but not for octopuses living in the wild. Wild octopuses are constantly exposed to potentially harmful situations, and the ability to assess the welfare status of wild octopuses could provide pertinent information about individuals’ health and species’ resilience to stressors. Here, we used underwater photos and videos to identify injuries and stress-related behaviors in wild Octopus insularis in a variety of contexts, including interacting with fishermen, interacting with other octopuses and fish, proximity to predators, in den, foraging, and in senescence. We adapted established metrics of octopus welfare from the laboratory to these wild octopuses. In addition to observing all of the stress measures, we also identified two previously unknown measures associated with decreased welfare: (1) a half white eye flash and (2) a half-and-half blotch body pattern. More than half of the individuals analyzed had arm loss, and almost half of the individuals had skin injuries. We also observed that irregular chromatophore expression and abnormal motor coordination were associated with interactions with fishermen. This is the first study to apply measures of welfare from the laboratory to wild octopuses. Our results may also aid in the identification of welfare measures for other wild invertebrates.
Understanding the evolution of nervous and sensory systems in animals is key to understanding the distribution of animal sentience. However, the use of model organisms – mostly vertebrates, and especially mammals – often biases comparative analyses. Sneddon et al. (2018a,b) point out that using nonhuman animals as models helps us better understand the multifaceted aspects of animal pain and sentience. Several concerns need to be considered in dealing with model organisms. Here we discuss how models that are unrepresentative phylogenetically influence hypotheses about the evolution of the myelination in animals. Greater effort is needed to escape “vertebrate-centrism” in evolutionary research.
Studies suggest that dogs exhibit prosocial behaviors toward humans and can rescue their guardians trapped in a box. We investigated whether dogs could communicate with an experimenter to gain access to a crying actor (the crying condition) who was visible but inaccessible, as she remained behind a fence throughout the test, or when she was simply speaking (the neutral condition). This experimenter was positioned to allow access to the actor, and the dog was free to move within the environment. Our findings indicate that dogs alternated their gazes more frequently between the actor and the experimenter, as well as between the actor and the dog’s guardian when the inaccessible actor was crying. Furthermore, the number of gazes at the guardian and the actor, as well as the duration of gazing at the actor, were greater in the crying condition. Dogs also remained closer to the actor during the crying condition. Our results suggest the dog made more communicative signals in the crying condition, bringing to light the possibility that this communication was being used as a tool to exhibit prosocial behaviors toward the actor.
Recently, a paper published in a Brazilian malacology online journal argued against the existence of sentience in octopuses based on disputable arguments – the presence of cannibalistic behavior, absence of sociality and parental care, short lives, size and complexity of the nervous system and intelligence. This response discusses a different perspective on the issue of octopus sentience, rebutting each of the arguments above.
Analyzing the behavior and body patterns of octopuses in nature is challenging due to their camouflage capabilities, changes in skin texture and color, body deformations, and frequent occlusions. Underwater lighting and turbidity further complicate these challenges. Addressing the lack of large-scale annotated datasets, we introduce HideAndSeg, a minimally supervised, AI-based tool for segmenting octopus videos. HideAndSeg integrates SAM2 with a custom-trained YOLOv11 object detector. First, the user provides point coordinates to generate the initial segmentation masks with SAM2. These masks serve as training data for the YOLO model. Our approach automates the pipeline by providing a bounding-box prompt to SAM2, eliminating manual intervention. We introduce two unsupervised metrics – temporal consistency and new component count – to quantitatively evaluate segmentation quality and guide mask refinement in the absence of real-world information used to train, validate, and test AI models. HideAndSeg reduces segmentation noise compared to the manually prompted approach. Our method can re-identify and segment the octopus even after complete occlusion, a scenario in which the manually prompted model fails. By reducing manual analysis in real-world scenarios, this work provides a practical tool that enables more efficient behavioral studies of wild octopuses.