Ethical challenges to patient autonomy in the era of artificial intelligence: a systematic review.

Animals in human-dominated environments face rapid and often unpredictable changes, with humans affecting access to resources and space, or expression of natural behaviour and social interaction. Domestic dogs () living in animal shelters provide a compelling model for studying animal behaviour in a human-created environment. Shelter dogs may experience abrupt environmental change upon relinquishment including loss of control, separation from previous social partners, and confinement in a highly structured setting where humans control access to resources, social interactions, and daily routines. These conditions can negatively affect welfare both acutely and over the long term (Raudies et al. ), limiting dogs’ ability to express species-specific behaviours such as exploration and social interactions, which are crucial for their mental and emotional well-being (Stephen and Ledger ). Confinement can also trigger stress-related behaviours, ranging from acute indicators such as panting, vocalization, and lip licking (Beerda et al. ) to chronic manifestations, including heightened vigilance, yawning, and repetitive behaviours (stereotypies), such as pacing and excessive self-grooming (Hubrecht et al. ; Beerda et al. , ; Hiby et al. ; Raudies et al. ). Stress responses in shelter dogs have a high degree of individual variability, since temperament and past experiences highly influence how dogs cope with stressors, including confinement (Beerda et al. ). Some dogs may show clear signs of distress, while others may exhibit a lack of positive behaviours, which has also been suggested as an indicator of reduced welfare (Broom and Johnson ). One strategy to improve the welfare of animals in captivity, including shelter dogs, is the implementation of effective forms of enrichment, consisting of adding new environmental conditions and temporary stimuli that give opportunities for species-specific behaviours and fulfil animals’ physiological and social requirement, promoting psychological and physiological well-being (Hunt et al. ). Enrichment has been shown to reduce stress, improve behavioural diversity, and positive social interactions (Markowitz ; Jensen and Toates ; Newberry ), and enhance cognitive function (Rosenzweig and Bennett ; van Praag et al. ; Wolfer et al. ) across multiple species and settings. These include for example lab rats, (Simpson and Kelly ) and mice, (Wolfer et al. ); farm pigs, (Godyń et al. ), broiler chickens, (Vas et al. ), and laying hens (Xu et al. ); house cats, (Herron and Buffington ); and captive geckos, (Zieliński ). In dogs, enrichment programs (ranging from inanimate objects such as toys and sensory stimuli, to animated forms including interactions with humans and conspecifics) have been linked to increased relaxation, reduced undesirable behaviours, and improved welfare outcomes (Wells ; Herron et al. ; Hunt et al. ). Dogs exposed to multi-faceted enrichment regimes display more relaxed body postures, are quieter when approached by people, and show less jumping activity; all these traits are associated with decreased stress levels and improved welfare states (Coppola et al. ; Herron et al. ). An essential step in implementing good environmental enrichment is a rigorous and efficient monitoring of its effects (Bloomsmith et al. ). The choice of data collection methods is crucial in ethological research as it can critically influence the accuracy of behavioural observations (Altmann ; Jauhiainen and Korhonen ; Ledgerwood et al. ; Quirke and O’Riordan ; Ceballos et al. ; Bateson and Martin ; Brereton et al. ). One important issue is the choice of whether recording behaviour continuously (continuous recording; CR) or at specific time intervals (instantaneous sampling IS). CR provides detailed data but is time-intensive, while IS at fixed intervals risks missing behaviours and increasing variability (Daigle and Siegford ; Chen et al. ; Bateson and Martin ; Brereton et al. ). Understanding how different sampling approaches affect behavioural data is therefore essential for reliably assessing interventions.