Are you a glass half full or a glass half empty kind of person? In other words are you an optimist or a pessimist? What you might actually have noticed is that your underlying moods and emotions influence how you interpret the world around you. Having a good day with everything going well and in a positive mood then you might well be viewing the world through a glass half full lens. On the flip side if you’ve had one of those bad days, things might well look more pessimistic overall. This in fact represents an adaptive form of cognitive bias whereby our underlying emotional state influences our interpretation of ambiguous information. Importantly, in recent years we have demonstrated that non-human animals, including a range of farm animal species also demonstrate these cognitive biases. This matters for two important reasons. First it provides evidence that animals experience emotions and second it has implications for animal welfare.

 

An influential model of animal welfare assessment is the Five Domains Model that considers the domains of nutrition, physical environment, health and behavioural interactions, all of which feed into a fifth domain of mental state. This holistic approach helps emphasise that animal welfare is more than health. An animal can be healthy but have poor welfare. Thus, it highlights the importance of mental wellbeing. However, this presents a major challenge as how do we assess mental wellbeing in farm animals that can’t communicate directly with us? This has led to exciting, pioneering research that aims to better understand the emotional lives of non-human animals. Rather than using the word ‘emotion’, the more technical term used in this area is affective state. So now back to cognitive bias!

 

This approach has been developed by Professor Mike Mendl, Dr Liz Paul and others from the University of Bristol. Back in 2004, they published an influential Nature paper that has ultimately led to a new area of animal welfare science. In the 2004 study they trained lab rats so that when they heard one sound frequency they could receive a food reward by pressing a lever. Thus that sound frequency predicted a positive reward. They were also trained that when they heard a second different sound frequency they needed to avoid pressing the lever or something negative would happen (e.g. a blast of white noise). Once the rats learnt this so called ‘Go / No-go’ task, half of them were maintained in standard housing conditions. However, the other half were placed under housing conditions known to induce a degree of stress (e.g. damp bedding and unpredictable husbandry regimes). This experimental treatment was the manipulation of emotional state, the prediction being that those housed in unpredictable conditions would be in a more negative affective state than those maintained in the standard housing conditions. Following the housing period, the rats were then re-tested in the previous learning task. However, this included an important clever additional element! In addition to giving the rats trials where they heard the previously learnt sound tones, one frequency predicting the positive reward and the other something negative, in some trials they also played a sound frequency that was intermediate between the two previously learnt cues. That is, they exposed the rats to an ambiguous stimulus, essentially asking does the animal perceive it as predicting a reward or predicting a punishment. The prediction was that if those housed in the predictable housing were in a more positive emotional state than those in the unpredictable housing the former would be more likely to respond to the ambiguous sound tone as if it predicted a food reward. This was indeed what they found. This pioneering study then set the scene for a suite of so called judgement bias tasks than now span from invertebrates to a range of farm animal species, and was the subject of a recent meta-analysis of the topic. This includes some of my own research which has used the approach to demonstrate that giving dairy cows access to pasture had benefits for their affective state. Also, in work as part of a collaboration with Professor Simon Turner at SRUC in Edinburgh, we’re using the approach in pigs to assess if those animals that have lost an aggressive encounter are in a more negative affective state compared to winners. This matters because regrouping aggression is a welfare issue that needs to be managed in pig production.

 

Using judgement bias to infer animal emotion also has a number of limitations. For example, the approach of first training an animal on an associative learning task means it is only feasible in a research setting and not practical for on farm welfare assessment. This has sparked interesting research in alternative approaches. One such approach is termed attention bias. This is another form of cognitive bias whereby animals (including humans) in a negative affective state will pay more attention to potentially threatening stimuli compared with those in a more positive affective state. For example, this approach has been pharmacologically validated in sheep. In the study, individual sheep were allocated to one of three treatments, receiving either an anxiogenic drug to increase anxiety, an anxiolytic to reduce anxiety or a saline control. They were tested individually in an arena during which they were briefly exposed to a threatening stimulus (a dog seen through a window which was then covered after a short period of time). Subsequently those given the anxiogenic drug spent more time being vigilant and looking towards were the threat had been, while also being less likely to feed. This approach has the advantage of not requiring the animals to have been previously trained on a task. It has been used in a number of practical settings. For example, we’ve recently used it in a study involving dogs from a licenced breeding establishment and demonstrating that those that had been given additional environmental enrichment were less vigilant in an attention bias test, consistent with a more positive affective state. We’ve also recently used it as part of a study comparing the welfare of sheep managed using either virtual fencing or physical fencing and if you’re interested you can hear the results of that study at the upcoming BSAS conference!

 

While attention bias avoids the need for animals to learn a task, it is still not practical for inclusion in on farm assessments of animal welfare. For that an approach called Qualitative Behaviour Assessment (QBA) is proving very useful. QBA was pioneered and developed by Professor Francoise Wemelsfelder based at SRUC in Edinburgh. The approach involves a holistic assessment of the behavioural expression of an individual animal in relation to a list of descriptive terms, with the animal scored on a scale for each term. The descriptive terms could include words such as ‘relaxed’, ‘alert’, and ‘playful’. This sounds highly subjective and anthropomorphic. However, there is convincing evidence that how animals score using QBA is related to other validated measures of welfare. Indeed, the utility and success of this non-invasive observational approach is also evidenced by the fact it has now been incorporated into a number of UK assurance schemes used by retailers.

 

Previously research on animal emotion and sentience was deemed somewhat off limits and anthropomorphic. Now, thanks to pioneering researchers and advances in animal welfare science we are beginning to shed light on the mental wellbeing of non-human animals. However, there is still much that remains to be explored. In biology and science in general we talk about the ‘hard problem’ of consciousness. Research into animal mental states is within this challenging area. Rather than shy away from this we should embrace it and use it to attract talented researchers who like a challenge! Animal welfare science brings together a range of disciplines and perspectives with the goal of generating and using fundamental knowledge and understanding to have applied relevance to improve the lives of animals. Animal sentience, the capacity to experience emotions both positive and negative, is now enshrined in UK law with the Animal Welfare (Sentience) Act 2022. Importantly this includes all vertebrates, as well as invertebrate decapod crustaceans and cephalopods. This is also raising the question regarding other species not currently included in the Act. For example, in the cognitive bias studies discussed above, honey bees have been found to perform as well as many other vertebrate species! With this in mind I see there is an interesting session on insect welfare at the upcoming conference. The UK also has the recent Genetic Technology (Precision Breeding) Act 2023 with the aim of harnessing advances in animal breeding technology to enhance performance, health and welfare. While technologies in this area can have great potential the legislation also requires approaches to follow up with welfare assessments of the animals involved. It will be important that this includes practical approaches to assess mental state and wellbeing. This is an exciting time to be involved in animal welfare science. For those interested in finding out more about the opportunities for research in this multidisciplinary area then take a look at the Animal Welfare Research Network (https://awrn.co.uk/). This is funded by BBSRC and Defra and is free to join.      

 

Written by Gareth Arnott, Queens University Belfast