Much of life for every individual animal — a human, a rabbit, or a minnow — involves taking decisions in order to cope with the world around that individual by minimising poor welfare and maximising good welfare. The timescale may necessitate short-term or long-term trade-offs. The opportunities will be constrained by general environmental aspects, so options for a free-living human, rabbit, or minnow will be somewhat different from those for a captive human, rabbit, or minnow. Hence, scientific measures of the welfare of individuals assess degree of negative and degree of positive welfare, and an understanding of motivation is of great importance in welfare assessment. The motivational systems that function in humans and other animals have evolved and exist now because they are adaptive (Broom, 1981; Mendl and Paul, 2020; Broom, 2022a). Most of the strong positive preferences of animals are for resources or actions that benefit them: that is, that help them to survive, breed successfully, and have good welfare. During development, individuals will have acquired further information that helps them to take decisions that lead to benefits. Although motivational systems are complex, it is helpful to think of motivation as including incentives or goals (Toates, 2002). Kringelbach and Berridge (2015) suggest that actions aimed at increasing the likelihood of feeling pleasure are ubiquitous in sentient animals. Incentives encourage the occurrence of much behaviour (Dickinson and Balleine, 1995; Dickinson and Balleine, 2002; Berridge, 2018; Perez and Dickinson, 2020). One consequence of this for welfare research (Duncan, 1978; Dawkins, 1983; Dawkins, 1990; Duncan, 1992; Broom and Johnson, 2019) is that the assessment of motivational strength during tests of preference is useful in any attempt by humans to ensure that poor welfare is avoided and good welfare is maximised. Although most strong positive preferences relate to something wholly positive, there can be preferences for drugs that result in some short-term pleasure but longer-term negativity, or for high-risk activities in humans and in non-human species. Similarly, the extreme level of avoidance shown towards a running insect or small mammal may not be justified by the average degree of danger from such animals. Extreme preferences should be investigated by studying the consequences when the preferred resource is obtained or the preferred action is taken. For the assessor of welfare, as for the individual studied, the balancing of positive and negative is necessary, either continuously or at intervals. Some examples of this are now considered.

Many measures of good or poor welfare have to be considered when planning how to assess the welfare of an individual, and single measures sometimes do not indicate the extent of how good or how poor welfare is. For example, facial expression can indicate pain or pleasure (Lansade et al., 2018; McLennan et al., 2019b), but can be simulated by individuals and can convey false information in humans and other species (Barrett et al., 2019). Collections of measures and further information about the state of the individual can improve accuracy of evaluation. However, individuals may be deceived about their own welfare. A diseased individual may be unaware of pathology that limits its ability to cope with environmental impacts (Fraser 2008; Broom and Johnson, 2019; Broom, 2022a). In a largely philosophical discussion of whether suffering dominates enjoyment, Groff and Ng (2019) point out that welfare measures may sometimes indicate what the individual studied perceives to be their welfare at that time, rather than their actual welfare.

Some components of coping systems may act independently, but there will also be summating effects and integration of component effects. For example, rats may suppress pain responses when predators are present, and a sheep with pain from foot-rot may continue to show behavioural evidence of pain if it is with a familiar flock feeding calmly, but may suppress any such behaviour if potential predators are present (Sorge et al., 2014; McLennan and Mahmoud, 2019a; McLennan et al., 2019b, Broom personal observation). Such summation by an individual must often involve compensation between negative and positive in the determination of the outcome. Negative experiences often lead to avoidance behaviour, while positive experiences tend to lead to approach. A positive balance may be demonstrated using simple behavioural measures. However, preference measures are most valuable if the strength of preference is assessed. Measurement of the amount of work that an individual will invest in the approach or avoidance often, but not always, gives good information about the extent of net negative or positive welfare.

Welfare measures for an individual can indicate good welfare, poor welfare, or net welfare, i.e., the balance between good and poor. At any particular time, there may be net positive or net negative welfare, so resolving the negative and the positive is important for the assessor. Biological mechanisms result in individuals having needs, and the fulfilment of these is associated with good welfare. When all needs are met, the individual’s behaviour and physiology are more likely to indicate normality and positive welfare to animal welfare scientists. As a consequence, normal behaviour can be used as an indicator of good welfare and lack of normal behaviour can indicate poor welfare. However, care is needed when using readily visible indicators of welfare. Human and non-human individuals often attempt to appear normal when they are disturbed in some way. Displaying signs of pain or fear to a predator or revealing to another individual that a communication is perceived in a very negative way could have fatal or other negative consequences, so such signs are often concealed. Behaviour judged to be abnormal behaviour by human observers is often associated with a welfare problem, both when it is adaptive and when it is maladaptive (Broom, 2006; Broom, 2014). Hence, one of the ways of identifying and quantifying good welfare is to observe a wide range of normal behaviour. Especially important is behaviour that can be demonstrated to be positively preferred, taking account of the strategies and inefficiencies of functioning that may be evident when normal behaviour is occurring (Broom, 2022a). Welfare is usually better if all positively preferred behaviours can be shown than if some are prevented. Both what is liked, in that the consequences are greater numbers of happy actions and fewer unhappy actions, and what is wanted, in terms of what is often chosen when the opportunity arises, are important to individuals (Yeates and Main, 2008). If positive preferences have been fulfilled and negative preferences avoided, the welfare of the animal should be good, but if that which elicits negative preference predominates, net welfare is poor. Much information about the welfare of humans and non-humans is obtained in this way (Broom, 2014). Several motivational tests, such as state-specific conditioning and judgement bias, can provide useful information about the extent to which welfare is positive or negative (Mendl and Paul, 2004; Lagisz et al., 2020; Mendl and Paul, 2020). Judgement bias is the influence of affect on judgement (Broom, 2022a). Typically, subjects are trained to react differently to two stimuli that elicit relatively positive- and negative-valence outcomes. Responses to subsequent presentations of intermediate “probe” stimuli indicate whether subjects judge them more positively (optimistic responses) or negatively (pessimistic responses). The result of a judgement bias test can indicate a net positive or net negative state in an individual (Crump et al., 2018; Broom, 2022a).

Physiological measures can directly indicate good or poor welfare. A wide range of measures examining body fluids (measuring adrenal hormones, proteins, and other indicators of tissue damage), emergency response, and other negative impacts on individuals can enable assessment of how poor welfare is (Broom and Johnson, 1993; Fraser, 2008; Broom and Johnson, 2019; Broom, 2022a). Pleasure is often associated with certain kinds of measurable brain activity and behaviour (Broom and Zanella, 2004) that may be reflected by elevated oxytocin levels in blood. Seeing positive images can increase magnetic resonance imaging (fMRI) activity on one side of the frontal area of the cerebral cortex. Phillips et al. (1998) describe a set of brain regions in which they observed activity when pictures of happy faces were seen but not during neutral or sad situations. Rewarding experiences, and positive feelings associated with these when they occur over long periods, can affect the hippocampus with consequences that are measurable via either MRI measures or post-mortem measures of previous brain function (Poirier et al., 2019). Oxytocin is produced in circumstances where there are clear signs of positive feelings, for example when a female mammal is nursing her young. Oxytocin is associated with the let-down of milk, but additionally leads to a feeling of pleasure (Chen and Sato, 2017). It is synthesised in the paraventricular nucleus (PVN) of the hypothalamus and in the supraoptic nucleus and binds to receptors that regulate HPA axis activity. Oxytocin is associated with parental care, emotional understanding of others, social support, inter-individual trust (Feldman et al., 2007; Ross and Young, 2009; Meyer-Lindenberg et al., 2011), and positive judgement bias in both dogs and humans (Kis et al., 2015). Oxytocin increase is associated with ACTH (adrenocorticotrophic hormone) and glucocorticoid decrease, lymphocyte proliferation, brain GABA (gamma-amino butyric acid) increase, and cardiac vagal tone increase (Carter and Altemus, 1997; Parker et al., 2005; Smith and Wang, 2012; Cardoso et al., 2014). Oxytocin can modulate emotional responses controlled by the amygdala, anterior cingulate cortex, lateral septum, ventral tegmentum, and nucleus accumbens (Tost et al., 2015). The effects of oxytocin in suppressing glucocorticoid production lead to experimental possibilities for physiological evaluation of the balance of affect when there is good welfare for one reason and poor welfare for another. However, it is not justifiable to state that oxytocin is the basis for all social behaviour (Leng et al., 2022). For example, maternal defence (Bosch et al., 2005; Bosch, 2013) may be associated with anxiety, so the effects of oxytocin may not all be positive for welfare. Change in cardiac vagal tone is another physiological measure that has been used as an indicator of good welfare. Gygax et al. (2013) described the relationships between pre-frontal cortex activity, sympatho-vagal reaction, locomotion, and behaviour indicating anticipation in rewarded and frustrated goats. Some of these physiological and behavioural measures can last for a short time and indicate brief periods of poor or good welfare. Anticipatory physiology or behaviour may indicate good or poor welfare according to what is anticipated, and the positive and negative are normally easy to distinguish.

Several behavioural measures provide some information about the balance between good and poor welfare. Prolonged stereotypies, self-mutilation, and other depressed behaviour indicate long-term difficulties in coping and hence a high magnitude of poor net welfare (Broom, 2001; Broom, 2022a). Positive experiences may not have sufficient impact to counteract the negativity. In other circumstances, there may be positive, measurable effects that compensate for anything that is negative (Lawrence et al., 2018). A net positive balance between factors with a negative impact and reward systems can be indicated by anticipatory behaviour when a reward is imminent (Spruijt et al., 2001). For example, positive-oriented anticipatory behaviour has been found to occur when rats knew from previous experience that they were about to be transferred from a barren cage to an enriched cage or to one where there would be sexual contact (van der Harst et al., 2003). Some forms of anticipatory behaviour are definable as play. Play occurs in a wide range of vertebrate and invertebrate animals (Held and Špinka, 2011; Pruitt et al., 2012; Kuba et al., 2014) and usually indicates good welfare, as it occurs in situations where welfare is good and is generally suppressed when welfare is poor (Boissy et al., 2007; Ahloy-Dallaire et al., 2018). While most of the behaviour that might be categorised as play is likely to indicate that welfare is good, some intellectual play is not readily identified by observers and some behaviour is not unambiguously play (Broom, 2014).

People who are very familiar with a domestic animal species sometimes have considerable ability to evaluate welfare. They are often making observations that are similar to those used in quantitative studies by welfare scientists. Wemelsfelder (2007) advocates the use of such human abilities in ‘qualitative behavioural assessment’. This can be a useful way of assessing the balance between good and poor welfare, but can be subject to error or observer bias (Keeling, 2009; Bokkers et al., 2012; Andreasen et al., 2013), so should never be used in the absence of other welfare measurements (Muri et al., 2019; Broom, 2022a).

Many experimental studies have assessed the extent to which subjects will tolerate a negative experience in order to obtain a positive experience. Every prey species that ventures into a dangerous open space where food might be found is doing this. The assessment of risks and benefits is a key ability in life for all sentient animals (Broom, 2014). Laboratory rats have been found to be willing to move down a freezing cold pathway in order to access a preferred food source, even with other palatable food available (Cabanac and Johnson, 1983). It would seem that, in the pursuit of pleasure, considerable discomfort is tolerated (Broom and Johnson, 2019). However, having to take such decisions can be disturbing and can have sufficiently negative effects to be called stressful. Human children may risk parental reprimand in order to take a food item. In some situations, and in many species, a quantifiable risk of pain may be incurred in order to obtain a food reward. A risk of potential pain can be easier to tolerate than a risk of fear. The pain risk is more likely to be predictably limited, whereas the potential cost of some fear is death. Fish have high levels of cognitive ability (Salwiczek et al., 2012; Broom, 2014). Studies of the responses of fish to having been caught on a hook and then released (Thorstad et al., 2003) show that further capture on a hook is avoided to a greater extent by salmon in cases of increased playing time, increased number of runs during the angling event, hooking in the throat, bleeding at the hook wound, increased handling time, or increased air exposure. Some fish are less likely to be caught on a hook if they have observed other fish being caught (Lovén Wallerius et al., 2020), but other fish allow themselves to be caught more than once (Beukema, 1970), perhaps because food is difficult to find. There can be trade-offs between predicted pain and predicted positive experience. Proximity to an individual conspecific, allogrooming, tactile stimulation, and heightened shoaling behaviour have been used as indicators of positive experience in fish (Fife-Cook and Franks, 2019). Context is important in interpretation of heightened shoaling behaviour because similar responses can be exhibited when fish are fearful.

Positive experience is normally associated with approach, but may be avoided because of highly predictable risks of negative events. A pig may refrain from approaching food because of the risk of attack by another stronger and hence dangerous pig (Mendl et al., 2010). The welfare of the fearful pig would be relatively poor but not easy to assess unless the negative state were reflected in behavioural or physiological changes.