Empirically Assessing The Emotions Of Non-Human Animals
When you hear the words “animal welfare,” what do you think of? Do you focus on an animal’s physical health, emotional experiences, or whether they live in a natural environment? Scientists who study animal welfare typically focus on one (or more) of these aspects, sometimes referred to as the Three Circles of Animal Welfare (see Figure 1). But these concepts don’t encompass everyone’s view of animal welfare. For example, the public may also value factors like duty of care (the responsibility of farmers to protect animals from harm) and the type of relationship between a farmer and an animal, which don’t necessarily follow the three domains of animal welfare. Additionally, these different views of animal welfare can sometimes come into conflict with each other when their implications diverge.
Picture this: a laboratory mouse lives in a cage that’s the size of a shoebox (far from their natural average home range size of just over half an acre) with two other mice (mice are social animals). All the mice are physically healthy (they drink, eat, and have a good body condition), yet they repeatedly chew on the cage bars out of frustration. What would you conclude about their welfare? Is good physical health and living with other animals enough for good welfare? Or do the feelings of frustration trump everything else? Like many other animal welfare scientists, I typically focus on emotions when discussing and assessing animal welfare, and only consider natural living and physical health in relation to an animal’s emotional state.
I entered the field of animal welfare science because I cared deeply about animal suffering and wanted to do something about it. I believed that studying the ways that animals can suffer would identify opportunities for improvement. It wasn’t until later that I realized that the scientific discipline of animal welfare is full of contradictions. This field emerged out of societal concerns for animal suffering in factory farming, yet rather than calling for an end to factory farming, scientists instead investigated (and continue to investigate) how to reduce suffering in incremental ways.
This is the classic animal rights versus animal welfare dichotomy that I faced during my PhD, and I wasn’t even vegan or an animal rights activist when I first entered the animal welfare field — although that changed a couple of years ago. Briefly, the animal rights or abolitionist perspective views animals as having the right to not be used or killed for human purposes (e.g., for food, research, clothing, etc). The concept and scientific discipline of animal welfare, on the other hand, assumes that using animals for any reason is justified as long as it is done “humanely,” including breeding and killing them for food. At face value, such a discipline does not seem valuable for animal advocates due to the ethical dilemmas it brings. But the truth is, animal welfare research has been used by animal advocates, and for good reasons too. On a broad level, animal advocates can and often do use empirical evidence that animals suffer (or experience specific emotions like boredom) to create more convincing arguments that they should not be farmed or bred for captivity. To be effective animal advocates, we need to use all the available data to advocate for animals’ behalf, even if it comes from a source that we may morally disagree with. The data exists, so why not use it to advocate against animal exploitation?
Figure 1: The Three Circles of Animal Welfare (adapted from Fraser, 2008).
How To Assess Emotions in Animals
In humans, the gold standard measurement of emotions is to simply ask people how they are feeling. Of course, this isn’t applicable to non-verbal individuals such as babies or those with cognitive disabilities. Emotions are also accompanied by facial cues, behaviors, and physiological and cognitive markers – all of which can be measured in non-verbal individuals, including non-human animals. While we can’t measure emotions in non-human animals per se, we can infer them using measurable indicators. However, these indicators need to be valid – i.e., they need to show that they actually reflect an emotional state.
There are different ways to achieve this. One way is to use human data and investigate whether the markers (behavioral, physiological, or cognitive) vary with self-reported emotions in humans. If they do, we can use those markers to assess emotions in other animals. For instance, human data on mental health conditions such as depression are inspiring research into whether other species can display markers of depressive-like states.
A second way to validate indicators of animal welfare is to identify situations or factors that evoke negative or positive emotions in humans, and then monitor what changes (behaviorally, physiologically, or cognitively) when other species are put in similar situations. Such changes can then be used to infer emotional states in animals. For example, plenty of pharmaceutical drugs increase or decrease certain emotions in humans (e.g., anxiety), so these drugs can be used to alter emotions in other species to then observe changes in behavior, physiology, or cognition.
A third way to validate animal welfare indicators is to monitor changes in situations that animals would naturally avoid (for negative emotions) or approach (for positive emotions) since emotions are thought to have evolved to help animals avoid dangerous or threatening situations, and to approach situations that are rewarding. For example, being restrained is a negative experience for many different species. Researchers have recently compared behavioral and physiological responses when cats were fully restrained during veterinary examinations, which can now be used as negative welfare indicators in cats. In general, these different methods have been successful at inferring animal emotions in the short term.
Validating Animal Welfare Indicators: An Example In Laboratory Monkeys
Although animal welfare scientists have traditionally assessed short-term emotions, much less is known on how to assess the cumulative emotional impact of different events across an individual’s lifespan, both negative and positive. This is called lifetime welfare (or cumulative experience) and is the subject of my PhD dissertation. I used non-invasive methods: I mainly reviewed pre-existing studies, accessed data from primate centers, and at most, collected behavioral data of monkeys who were already living in laboratories. As someone who wanted to improve the welfare of laboratory monkeys, I did not want to cause more suffering by putting these animals through additional experiments. The main objective of my PhD was to assess the validity of different abnormal behaviors as markers of lifetime welfare in laboratory monkeys. These animals are typically re-used in experiments and consequently, live long lives in captivity and are therefore exposed to repeated stressors related to research procedures (e.g., restraint) or captive life (e.g., indoor housing). Laboratory animals are also at-risk of developing abnormal behaviors- behaviors that are known or suspected to result from underlying distress or brain pathology. In laboratory monkeys, these include pacing back and forth (the most common), hair-pulling, and hunched postures (to name a few). If you are interested in seeing what these look like, you can watch a video here (but warning, this content may be upsetting).
For abnormal behaviors to be valid lifetime welfare indicators, I came up with a list of life events (e.g., location moves, being hospitalized, being socially housed, being outdoors, etc.) and pre-determined whether they were aversive or rewarding for monkeys through a literature review using the framework above. That is, events were classified as aversive (bad for welfare) if previous work found these situations to be avoided by monkeys, or identified as aversive by humans. In contrast, events were classified as rewarding (good for welfare) if previous work found them to be preferred by monkeys, or identified as pleasurable by humans. Events ambiguous for welfare were not included.
This resulted in 13 different life events across the lifespan (anywhere from separation from the mother in infancy to whether monkeys were currently single housed). Primate centers typically log every event that happens to a monkey (whether that be a cage change, blood draw, or medication administered). As such, I was able to obtain life event records from over a thousand monkeys from two different primate centers in the U.S., in addition to being able to access or collect my own behavioral data. For abnormal behaviors to be valid indicators of lifetime welfare, they should be predicted by an aggregated score of lifetime stress (i.e., just the sum of the frequency of negative events). So as lifetime stress increases, the frequency of abnormal behaviors should increase, too.
Out of all the abnormal behaviors I tested (including pacing, hunched postures, and hair-pulling, to name a few), only hair-pulling was associated with the score of lifetime stress in the expected direction (they both increased together). Interestingly, in humans, the number of traumatic life events is higher in individuals diagnosed with hair-pulling disorder than individuals without hair-pulling disorder, suggesting that monkeys may hair-pull as a means to cope with their negative emotions resulting from life stressors. I also conducted additional analyses to understand the relationship of each individual life event and the frequency of hair-pulling.
To my surprise, I found that not every aversive life event increased the odds of hair-pulling. This suggests that, at best, I can only recommend hair pulling as a potential candidate to infer lifetime welfare. Future validation work is now needed, specifically looking at why some negative life events fail to predict an increase in hair-pulling. One potential explanation could be the existence of individual differences in responding to stressors, which in turn, may be influenced by coping style (such as personality). In humans, high levels of neuroticism result in more stressful appraisals of life events, while higher levels of extroversion increase the likelihood of viewing life events as more positive. It is then likely that such personality factors play a role in how severe a life event is perceived in non-human animals too, and thus how much it affects them emotionally. In other words, some laboratory monkeys may be hyper-sensitive to stressors (and hair-pull as a result), while other monkeys may be more resilient, both of which may be influenced by personality differences.
Animal Welfare Research As An Avenue For Effective Animal Advocacy
The results of my PhD show that lifetime suffering may be assessed by the presence of hair-pulling (although future work is needed to confirm this), which will be useful for laboratory technicians to assess welfare since this behavior is practical to observe. For animal advocates, these results could be used to demonstrate how the stress of animal experimentation increases the risk for monkeys to pull out their own hair (a behavior not seen in the wild), and to help advocate for an end to animal testing. This is just one example of the value that animal welfare data can serve for advocates who want to reduce or eliminate animal suffering.
As mentioned before, on a broad level, animal welfare science proves that animals can experience diverse emotions, which can be used to create more convincing arguments that animals should not be farmed or bred for captivity. Again, some veg*n advocates may find this ironic – i.e., having our animal advocacy influenced by a scientific discipline that has bred animals into captivity and put them into aversive situations to assess emotional states. This is especially true if you are against animal experimentation since animal welfare research uses animals (even if it’s to understand their emotional lives). However, being an effective animal advocate means being open to using all available evidence to advocate for animals, even if these studies came from a discipline that has values we disagree with.
This leads to my second point: Animal welfare recommendations and campaigns are typically supported by empirical evidence. Some examples include the Codes of Practice for farmed animals and guidelines for research animals in Canada. These guidelines and recommendations for best practices were formed, thanks in part to the findings of animal welfare research. However, neither the Codes of Practice nor the research guidelines are legally required, so it is expected that farmers and researchers voluntarily adhere to them. This isn’t the case everywhere though. The European Union does take into account animal welfare research when it comes to their farmed animal legislation. Accordingly, advocates can campaign for the best practices of animal welfare, based on empirical research.
Overall, there are two broad kinds of animal welfare research: fundamental (data showing that animals simply can experience emotions) and applicable (for example, research comparing different housing conditions to identify the best ones for animal welfare). Both branches can be used as instruments in our activist toolkit, to bolster campaigns or legislation for improved welfare, or even abolitionist campaigns. It’s an advocacy paradox, but it’s one we need to accept in the shorter term as we work towards animal liberation: by pointing out how animals are capable of experiencing rich emotional lives, often coming from research with captive subjects, we can make a case that they deserve to live freely versus living a life full of exploitation.