Search

Animal affective computing is an emerging new field, which has so far mainly focused on pain, while other emotional states remain uncharted territories, especially in horses. This study is the first to develop AI models to automatically recognize horse emotional states from facial expressions using data collected in a controlled experiment. We explore two types of pipelines: a deep learning one which takes as input video footage, and a machine learning one which takes as input EquiFACS annotations. The former outperforms the latter, with 76% accuracy in separating between four emotional states: baseline, positive anticipation, disappointment and frustration. Anticipation and frustration were difficult to separate, with only 61% accuracy., Competing Interests: The authors have declared that no competing interests exist., (Copyright: © 2024 Feighelstein et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)

Comparative studies of human-dog cognition have grown exponentially since the 2000's, but the focus on how dogs look at us (as well as other dogs) as social partners is a more recent phenomenon despite its importance to human-dog interactions. Here, we briefly summarise the current state of research in visual perception of emotion cues in dogs and why this area is important; we then critically review its most commonly used methods, by discussing conceptual and methodological challenges and associated limitations in depth; finally, we suggest some possible solutions and recommend best practice for future research. Typically, most studies in this field have concentrated on facial emotional cues, with full body information rarely considered. There are many challenges in the way studies are conceptually designed (e.g., use of non-naturalistic stimuli) and the way researchers incorporate biases (e.g., anthropomorphism) into experimental designs, which may lead to problematic conclusions. However, technological and scientific advances offer the opportunity to gather much more valid, objective, and systematic data in this rapidly expanding field of study. Solving conceptual and methodological challenges in the field of emotion perception research in dogs will not only be beneficial in improving research in dog-human interactions, but also within the comparative psychology area, in which dogs are an important model species to study evolutionary processes., (© 2023. The Author(s).)

Domestic dogs (Canis familiaris) hold a unique position in human society, particularly in their role as social companions; as such, it is important to understand their emotional lives. There has been growing interest in studying behavioural biases in dogs as indirect markers (reflecting lateralized brain activity) of their emotional states. In this paper, we not only review the previous literature on emotion-related behavioural lateralization in dogs, but also propose and apply the concept of evidential weight to previous research. This allows us to examine different hypotheses about emotion-related brain asymmetries (i.e., Right-Hemisphere-, Valence-, Approach-Withdrawal-Hypothesis) on the basis of a "likelihood-ist" concept of evidence. We argue that previous studies have not been able to discriminate well between competing hypotheses and tended to focus on confirmation bias than critically assess different hypotheses; as such there is a strong case for more systematic investigation to pull these theories apart. We present the areas for future research and explain their importance for understanding the emotional lives of dogs., Competing Interests: Declarations of interest None., (Copyright © 2022 The Authors. Published by Elsevier Ltd.. All rights reserved.)

The ability to infer emotional states and their wider consequences requires the establishment of relationships between the emotional display and subsequent actions. These abilities, together with the use of emotional information from others in social decision making, are cognitively demanding and require inferential skills that extend beyond the immediate perception of the current behaviour of another individual. They may include predictions of the significance of the emotional states being expressed. These abilities were previously believed to be exclusive to primates. In this study, we presented adult domestic dogs with a social interaction between two unfamiliar people, which could be positive, negative or neutral. After passively witnessing the actors engaging silently with each other and with the environment, dogs were given the opportunity to approach a food resource that varied in accessibility. We found that the available emotional information was more relevant than the motivation of the actors (i.e. giving something or receiving something) in predicting the dogs' responses. Thus, dogs were able to access implicit information from the actors' emotional states and appropriately use the affective information to make context-dependent decisions. The findings demonstrate that a non-human animal can actively acquire information from emotional expressions, infer some form of emotional state and use this functionally to make decisions., (© 2021. The Author(s).)

Facial expressions potentially serve as indicators of animal emotions if they are consistently present across situations that (likely) elicit the same emotional state. In a previous study, we used the Dog Facial Action Coding System (DogFACS) to identify facial expressions in dogs associated with conditions presumably eliciting positive anticipation (expectation of a food reward) and frustration (prevention of access to the food). Our first aim here was to identify facial expressions of positive anticipation and frustration in dogs that are context-independent (and thus have potential as emotion indicators) and to distinguish them from expressions that are reward-specific (and thus might relate to a motivational state associated with the expected reward). Therefore, we tested a new sample of 28 dogs with a similar set-up designed to induce positive anticipation (positive condition) and frustration (negative condition) in two reward contexts: food and toys. The previous results were replicated: Ears adductor was associated with the positive condition and Ears flattener, Blink, Lips part, Jaw drop, and Nose lick with the negative condition. Four additional facial actions were also more common in the negative condition. All actions except the Upper lip raiser were independent of reward type. Our second aim was to assess basic measures of diagnostic accuracy for the potential emotion indicators. Ears flattener and Ears downward had relatively high sensitivity but low specificity, whereas the opposite was the case for the other negative correlates. Ears adductor had excellent specificity but low sensitivity. If the identified facial expressions were to be used individually as diagnostic indicators, none would allow consistent correct classifications of the associated emotion. Diagnostic accuracy measures are an essential feature for validity assessments of potential indicators of animal emotion., (© 2021. The Author(s).)

Facial expressions are a core component of the emotional response of social mammals. In contrast to Darwin's original proposition, expressive facial cues of emotion appear to have evolved to be species-specific. Faces trigger an automatic perceptual process, and so, inter-specific emotion perception is potentially a challenge; since observers should not try to "read" heterospecific facial expressions in the same way that they do conspecific ones. Using dynamic spontaneous facial expression stimuli, we report the first inter-species eye-tracking study on fully unrestrained participants and without pre-experiment training to maintain attention to stimuli, to compare how two different species living in the same ecological niche, humans and dogs, perceive each other's facial expressions of emotion. Humans and dogs showed different gaze distributions when viewing the same facial expressions of either humans or dogs. Humans modulated their gaze depending on the area of interest (AOI) being examined, emotion, and species observed, but dogs modulated their gaze depending on AOI only. We also analysed if the gaze distribution was random across AOIs in both species: in humans, eye movements were not correlated with the diagnostic facial movements occurring in the emotional expression, and in dogs, there was only a partial relationship. This suggests that the scanning of facial expressions is a relatively automatic process. Thus, to read other species' facial emotions successfully, individuals must overcome these automatic perceptual processes and employ learning strategies to appreciate the inter-species emotional repertoire.

Dogs are able to perceptually discriminate emotional displays of conspecifics and heterospecifics and possess the cognitive prototypes for emotional categorisation, however, it remains unclear whether dogs can respond appropriately to this information. One way to assess associations between specific behaviours and the perception of emotionally competent stimuli is to look at other reliable measures that are related to cognitive and physiological processing. Using a cross-modal preferential looking paradigm (Albuquerque et al., 2016), we presented dogs with pairs of facial expressions (positive and negative) combined with an emotionally charged vocalisation (positive or negative) or a control sound (neutral) and coded their mouth-licking behaviour. We found an effect of the valence of the face image dogs were seeing on the onset of the mouth-licking, with higher frequencies of this behaviour in response to the negative faces compared to images with positive valence. However, neither the sound being played nor the interaction between image valence and sound affected the behaviour. We also found an effect of species with mouth-licking occurring more often towards human stimuli. This spontaneous differential behavioural response, combined with previous evidence of cognitive emotional processing in these animals, suggests that dogs may have a functional understanding of emotional expressions., (Copyright © 2017 Elsevier B.V. All rights reserved.)

Leyhausen's (1979) work on cat behaviour and facial expressions associated with offensive and defensive behaviour is widely embraced as the standard for interpretation of agonistic behaviour in this species. However, it is a largely anecdotal description that can be easily misunderstood. Recently a facial action coding system has been developed for cats (CatFACS), similar to that used for objectively coding human facial expressions. This study reports on the use of this system to describe the relationship between behaviour and facial expressions of cats in confinement contexts without and with human interaction, in order to generate hypotheses about the relationship between these expressions and underlying emotional state. Video recordings taken of 29 cats resident in a Canadian animal shelter were analysed using 1-0 sampling of 275 4-s video clips. Observations under the two conditions were analysed descriptively using hierarchical cluster analysis for binomial data and indicated that in both situations, about half of the data clustered into three groups. An argument is presented that these largely reflect states based on varying degrees of relaxed engagement, fear and frustration. Facial actions associated with fear included blinking and half-blinking and a left head and gaze bias at lower intensities. Facial actions consistently associated with frustration included hissing, nose-licking, dropping of the jaw, the raising of the upper lip, nose wrinkling, lower lip depression, parting of the lips, mouth stretching, vocalisation and showing of the tongue. Relaxed engagement appeared to be associated with a right gaze and head turn bias. The results also indicate potential qualitative changes associated with differences in intensity in emotional expression following human intervention. The results were also compared to the classic description of "offensive and defensive moods" in cats (Leyhausen, 1979) and previous work by Gourkow et al. (2014a) on behavioural styles in cats in order to assess if these observations had replicable features noted by others. This revealed evidence of convergent validity between the methods However, the use of CatFACS revealed elements relating to vocalisation and response lateralisation, not previously reported in this literature., (Copyright © 2017 Elsevier B.V. All rights reserved.)