Validating reward-related behaviour for welfare assessment, and improving welfare through increased predictability of events
Lead Research Organisation:
University of Stirling
Department Name: Psychology
Abstract
There are many ways we can measure the welfare of animals in captivity. These include measures of biological functioning (such as health, longevity), whether the animals are expressing natural behaviours which they are adapted to perform, and how animals are feeling (emotional state, such as happy or fearful). New and practical methods of assessing the emotional state of animals are needed, and their value determined in relation to these other welfare measures. Ways to improve the welfare of animals housed in laboratories and used for research and testing are also required, within space and financial limitations.
The first aim of this project is to validate a new way to measure emotional state in laboratory-housed macaque monkeys and dogs, which are, with special justification to the Home Office, used in laboratory research and testing. There is good evidence that how sensitive an animal is to an announced reward, which can be measured in their anticipatory behaviour when they know the reward is coming, reflects underlying welfare state. For example, if the animal is living in an enriched environment, with opportunities to perform natural behaviours and with positive social contact with companions, it is predicted that it will be less excited about a positive reward because it already has many positive elements with its life. An animal that is living in a less enriched enclosure, with fewer positive and indeed some negative experiences, is predicted to be more aroused by an announced positive reward, and have increased sensitivity to it. The usefulness of reward sensitivity in measuring welfare state will be measured in group-housed animals in the laboratory, and the findings compared with other welfare measures collected routinely in industry.
The second aim of the research it to improve the welfare of laboratory animals by informing them of what is going to happen to them. There is considerable evidence that shows that animals prefer to know when events will occur - whether these events are positive or negative. We can inform animals when the events will occur, and what their nature is, by introducing a unique and reliable signal associated with a specific type of event, and always preceding that event with this specific signal. For example, a specific auditory tone might inform the animals that cage cleaning will occur, a different tone that they are to be captured for weighing, and another tone that food will arrive. These unique reliable signals allow the animals to prepare for the event, and in the absence of signals, they are also able to relax as they know they are "safe". Conversely, if a positive event is signalled the animal can experience the pleasurable state of anticipating this event (think of an excited dog when it knows it is going to go out for a walk). Using an experimental design with different groups of animals that received signalled positive events, signalled negative events, and a combination of these signals, the study will record the behaviour and other welfare-related measures collected in industry (such as heart rate) to determine the effects of the introduction of signals. If this simple method that does not require staff training, or time investment and can be introduced without changes to laboratory infrastructure, has clear welfare benefits, then it could be rolled out for all animals used in laboratories, and in other captive settings as well.
The first aim of this project is to validate a new way to measure emotional state in laboratory-housed macaque monkeys and dogs, which are, with special justification to the Home Office, used in laboratory research and testing. There is good evidence that how sensitive an animal is to an announced reward, which can be measured in their anticipatory behaviour when they know the reward is coming, reflects underlying welfare state. For example, if the animal is living in an enriched environment, with opportunities to perform natural behaviours and with positive social contact with companions, it is predicted that it will be less excited about a positive reward because it already has many positive elements with its life. An animal that is living in a less enriched enclosure, with fewer positive and indeed some negative experiences, is predicted to be more aroused by an announced positive reward, and have increased sensitivity to it. The usefulness of reward sensitivity in measuring welfare state will be measured in group-housed animals in the laboratory, and the findings compared with other welfare measures collected routinely in industry.
The second aim of the research it to improve the welfare of laboratory animals by informing them of what is going to happen to them. There is considerable evidence that shows that animals prefer to know when events will occur - whether these events are positive or negative. We can inform animals when the events will occur, and what their nature is, by introducing a unique and reliable signal associated with a specific type of event, and always preceding that event with this specific signal. For example, a specific auditory tone might inform the animals that cage cleaning will occur, a different tone that they are to be captured for weighing, and another tone that food will arrive. These unique reliable signals allow the animals to prepare for the event, and in the absence of signals, they are also able to relax as they know they are "safe". Conversely, if a positive event is signalled the animal can experience the pleasurable state of anticipating this event (think of an excited dog when it knows it is going to go out for a walk). Using an experimental design with different groups of animals that received signalled positive events, signalled negative events, and a combination of these signals, the study will record the behaviour and other welfare-related measures collected in industry (such as heart rate) to determine the effects of the introduction of signals. If this simple method that does not require staff training, or time investment and can be introduced without changes to laboratory infrastructure, has clear welfare benefits, then it could be rolled out for all animals used in laboratories, and in other captive settings as well.
Technical Summary
Animal welfare can be measured on a scale of poor to good, and will at any one point reflect the balance of their positive and negative affective states. Behaviour is the most accessible measure of welfare, being the result of all the animal's own individual decision making processes and also expressing their emotions. By measuring anticipatory behaviour to announced rewards, we can determine an animal's reward sensitivity, which reflects the brain's evaluation of its subjective state and the environment. Behavioural activation in anticipation of a reward involves mesolimbic dopaminergic systems; these systems are known to be sensitized by stressors suggesting the biological mechanism underlying the use of such behaviour in welfare assessment. The animal is adapted to select the most efficient (ie most rewarding) response, and those with more stress will show higher levels of anticipatory behaviours for a reward, whilst those with positive experiences with have less anticipatory behaviour. The first goal is to validate this measure, with a battery of other welfare measures in macaques and dogs in a working laboratory environment, to show practicality. The second goal is to manipulate signalled predictability to inform laboratory animals of what will happen to them. Evidence strongly suggests that animals prefer predictable events, be they positive (eg appetitive) or negative. Knowing, through a reliable signal which can be classically conditioned over paired trials, when events will happen is good for welfare (the preparatory hypothesis), and also informs animals when they are safe (the safety signal hypothesis). The project will systematically investigate and validate this new welfare assessment tool, and assess an associated Refinement that is cost effective, requires no changes to the physical infrastructure, and can be implemented widely to improve welfare and reduce the adverse impact of procedures, with potential to improve the quality of scientific output.
Planned Impact
The goals of the research are to validate a welfare assessment tool and to improve welfare through manipulations of predictability. As such they relate primarily to the Refinement R of the 3Rs.
There is considerable misunderstanding about the definition of Refinement (Buchanan-Smith et al 2005) but the NC3Rs and Buchanan-Smith et al define it in similar ways. "Refinement refers to improvements to scientific procedures and husbandry which minimise actual or potential pain, suffering, distress or lasting harm and/or improve animal welfare in situations where the use of animals is unavoidable." (NC3Rs website). Refinement has the potential to impact upon quality of scientific output; stressed, fearful animals will increase the unwanted variation in the scientific output and hence the number of animals required to reach statistical significance. The project will therefore, also impact on the Reduction R of the 3Rs.
Accurate assessment of animal welfare is fundamentally required to determine whether changes, planned or otherwise, to any aspect of the housing, husbandry or procedures performed on animals are having a positive impact (Refinement). Although there is a battery of welfare assessment measures, it is extremely hard to measure welfare (Mason & Mendl, 1993). An animal's welfare may lie on a scale from bad to good, and a combination of behavioural, physiological, biochemical, physical and subjective measures will produce a more complete picture of an animal's welfare. Studies on cognitive bias, as a potential measure of an animal's feelings (its emotions or subjective welfare, eg see Mendl et al 2009) are increasing in a range of species. It remains critical that these measures are validated against other more traditional welfare tools, and that ways to collect data on these subjective measures are streamlined, for wider uptake. If such convergent validity is shown other easily accessible signs of welfare can be used as a validated proxy. As the research will be conducted within a working laboratory (Charles River Laboratories, CRL), and with care staff, the findings should be easily implemented on site, globally within the company, and promoted through a workshop hosted by CRL to demonstrate positive impact on Refinement and Reduction.
Dogs and macaques, the chosen study animals, make up only 0.16% of all animals used in research in Great Britain in 2010. However, the benefits are likely to have widespread applicability to toher animals. These two study species have been selected for three main reasons. First there is particular society concern about the use of dogs and primates in research and testing given their potential capacity to suffer more than other laboratory animals due to difficulties in meeting their complex social and cognitive abilities within restricted laboratory environments (JWGR, 2004; 2009). Second, the two species differ in several ways, particularly in their responses to humans (who often, unintentionally, provide unreliable cues to the nature of events). Therefore comparison of the two species will be interesting and show breath of application. Third, Buchanan-Smith has experience of working with both these animals, and therefore a good understanding of their behaviour and welfare measures.
Buchanan-Smith HM et al 2005 Harmonising the Definition of Refinement. Anim. Welf. 14, 379-384.
Mason, GJ & Mendl M 1993 Why is there no simple way of measuring animal welfare. Anim. Welf. 2, 301-319.
Mendl M et al 2009 Cognitive bias as an indicator of animal emotion and welfare: emerging evidence and underlying mechanisms. Appl. Anim. Behav. Sci. 118, 161-181.
NC3Rs website (http://www.nc3rs.org.uk/ - accessed February, 2012)
There is considerable misunderstanding about the definition of Refinement (Buchanan-Smith et al 2005) but the NC3Rs and Buchanan-Smith et al define it in similar ways. "Refinement refers to improvements to scientific procedures and husbandry which minimise actual or potential pain, suffering, distress or lasting harm and/or improve animal welfare in situations where the use of animals is unavoidable." (NC3Rs website). Refinement has the potential to impact upon quality of scientific output; stressed, fearful animals will increase the unwanted variation in the scientific output and hence the number of animals required to reach statistical significance. The project will therefore, also impact on the Reduction R of the 3Rs.
Accurate assessment of animal welfare is fundamentally required to determine whether changes, planned or otherwise, to any aspect of the housing, husbandry or procedures performed on animals are having a positive impact (Refinement). Although there is a battery of welfare assessment measures, it is extremely hard to measure welfare (Mason & Mendl, 1993). An animal's welfare may lie on a scale from bad to good, and a combination of behavioural, physiological, biochemical, physical and subjective measures will produce a more complete picture of an animal's welfare. Studies on cognitive bias, as a potential measure of an animal's feelings (its emotions or subjective welfare, eg see Mendl et al 2009) are increasing in a range of species. It remains critical that these measures are validated against other more traditional welfare tools, and that ways to collect data on these subjective measures are streamlined, for wider uptake. If such convergent validity is shown other easily accessible signs of welfare can be used as a validated proxy. As the research will be conducted within a working laboratory (Charles River Laboratories, CRL), and with care staff, the findings should be easily implemented on site, globally within the company, and promoted through a workshop hosted by CRL to demonstrate positive impact on Refinement and Reduction.
Dogs and macaques, the chosen study animals, make up only 0.16% of all animals used in research in Great Britain in 2010. However, the benefits are likely to have widespread applicability to toher animals. These two study species have been selected for three main reasons. First there is particular society concern about the use of dogs and primates in research and testing given their potential capacity to suffer more than other laboratory animals due to difficulties in meeting their complex social and cognitive abilities within restricted laboratory environments (JWGR, 2004; 2009). Second, the two species differ in several ways, particularly in their responses to humans (who often, unintentionally, provide unreliable cues to the nature of events). Therefore comparison of the two species will be interesting and show breath of application. Third, Buchanan-Smith has experience of working with both these animals, and therefore a good understanding of their behaviour and welfare measures.
Buchanan-Smith HM et al 2005 Harmonising the Definition of Refinement. Anim. Welf. 14, 379-384.
Mason, GJ & Mendl M 1993 Why is there no simple way of measuring animal welfare. Anim. Welf. 2, 301-319.
Mendl M et al 2009 Cognitive bias as an indicator of animal emotion and welfare: emerging evidence and underlying mechanisms. Appl. Anim. Behav. Sci. 118, 161-181.
NC3Rs website (http://www.nc3rs.org.uk/ - accessed February, 2012)
Organisations
People |
ORCID iD |
Hannah Buchanan-Smith (Principal Investigator) |