Search

Pigs are commonly kept in intensive farming systems. Their use as model animals in biomedical research has increased. Both conditions may impact upon their welfare. Recent definitions of welfare emphasize the importance of emotion. Mood congruent biases in judgement have been proposed as proxy measures of emotion in animals; optimistic responses to ambiguous cues are said to reflect positive emotional states while negative biases reflect negative emotional states. We developed a novel active-choice task to measure judgement bias in conventional farm pigs (n = 7) and Gottingen minipigs (n = 8). Pigs were trained to distinguish a positive tone-cue, from a negative tone-cue, signalling the location (goal-box) of large or small rewards respectively. After learning this discrimination pigs were presented with a series of ambiguous tone-cues and the percentage of choices for the positive goal-box recorded as optimistic responses (phase I). After a 4-week break, pigs were retrained and challenged with a restraint treatment to induce a negative emotional state. Judgement bias was again measured (phase II). After another 2-week break, a third test was performed to assess the effect of repeated testing (phase III). Minipigs learned the initial discrimination faster than conventional pigs but there were no differences in subsequent relearning (Breed: F1,13 = 21.47, P = 0.0005; Phase: F2,26 = 121.56, P

The welfare of pigs both on farm and in laboratories can be influenced by the conditions associated with each environment. Modern definitions of welfare emphasize the importance of affective states as essential for understanding welfare needs, and welfare legislation refers to the avoidance of unnecessary ‘suffering’ and ‘distress’ of animals. To objectively measure affective states in animals is therefore a priority for assessing welfare. From human research we know that emotion and mood can influence cognitive processes, such as memory and judgement. In particular emotional valence (positive /negative) can lead to mood congruent judgements of ambiguous stimuli, e.g. people in a positive mood are more likely to make optimistic judgements. It has been proposed that these biases can be used as a proxy measure of the valence of emotion and mood in animals. In this thesis I develop and try to validate novel tasks for pigs which can be used to study emotion-biased cognitive processing. First, we developed a Judgement Bias Task (JBT) whereby pigs were trained to discriminate one tone-cue predicting a large reward (positive cue), from a second tone cue predicting a small reward (negative cue), by choosing the correct goal box. When pigs could sufficiently discriminate between these two cues, responses to three unfamiliar ambiguous tone cues, intermediate in frequency between the positive and negative cue, were measured. An optimistic choice was scored if the pig chose the goal box associated with the large reward. We demonstrated that both a common farm breed and laboratory breed of pig could relatively easily learn the discrimination, and showed a similar pattern of responding to the ambiguous cues. To validate the task we looked at responses of low birth weight (LBW) pigs. LBW in humans is associated with later emotional problems such as anxiety and depression. In pigs, LBW is associated with higher mortality, but also higher stress reactivity in surviving piglets, suggestive of altered emotionality. We hypothesized therefore that LBW pigs ought to make fewer optimistic choices in response to the ambiguous cues. We compared LBW pigs to their normal-birth-weight siblings and, in two studies found, that the LBW pigs did indeed make fewer optimistic choices. In a third study we failed to find a difference. We developed a second task, a probabilistic choice Pig Gambling Task (PGT), to look at decision making under risk, which is also susceptible to emotional influences. Pigs could freely choose between an option offering small but frequent rewards and a second option offering large but infrequent rewards. LBW pigs chose more for the small but frequent rewards suggesting higher levels of risk aversion. Responses on the PGT were independent from responses on the JBT suggesting that the two tasks are measuring different aspects of decision making under uncertainty (risk vs ambiguity). Further, responses on the JBT were independent of memory, learning, and motivation suggesting that our task is indeed targeting emotion-biased judgements. However, our tasks are not yet suitable for the on-farm testing of emotion in pigs due to the complex training involved.

The welfare of pigs both on farm and in laboratories can be influenced by the conditions associated with each environment. Modern definitions of welfare emphasize the importance of affective states as essential for understanding welfare needs, and welfare legislation refers to the avoidance of unnecessary ‘suffering’ and ‘distress’ of animals. To objectively measure affective states in animals is therefore a priority for assessing welfare. From human research we know that emotion and mood can influence cognitive processes, such as memory and judgement. In particular emotional valence (positive /negative) can lead to mood congruent judgements of ambiguous stimuli, e.g. people in a positive mood are more likely to make optimistic judgements. It has been proposed that these biases can be used as a proxy measure of the valence of emotion and mood in animals. In this thesis I develop and try to validate novel tasks for pigs which can be used to study emotion-biased cognitive processing. First, we developed a Judgement Bias Task (JBT) whereby pigs were trained to discriminate one tone-cue predicting a large reward (positive cue), from a second tone cue predicting a small reward (negative cue), by choosing the correct goal box. When pigs could sufficiently discriminate between these two cues, responses to three unfamiliar ambiguous tone cues, intermediate in frequency between the positive and negative cue, were measured. An optimistic choice was scored if the pig chose the goal box associated with the large reward. We demonstrated that both a common farm breed and laboratory breed of pig could relatively easily learn the discrimination, and showed a similar pattern of responding to the ambiguous cues. To validate the task we looked at responses of low birth weight (LBW) pigs. LBW in humans is associated with later emotional problems such as anxiety and depression. In pigs, LBW is associated with higher mortality, but also higher stress reactivity in surviving piglets, suggestive of altered emotionality. We hypothesized therefore that LBW pigs ought to make fewer optimistic choices in response to the ambiguous cues. We compared LBW pigs to their normal-birth-weight siblings and, in two studies found, that the LBW pigs did indeed make fewer optimistic choices. In a third study we failed to find a difference. We developed a second task, a probabilistic choice Pig Gambling Task (PGT), to look at decision making under risk, which is also susceptible to emotional influences. Pigs could freely choose between an option offering small but frequent rewards and a second option offering large but infrequent rewards. LBW pigs chose more for the small but frequent rewards suggesting higher levels of risk aversion. Responses on the PGT were independent from responses on the JBT suggesting that the two tasks are measuring different aspects of decision making under uncertainty (risk vs ambiguity). Further, responses on the JBT were independent of memory, learning, and motivation suggesting that our task is indeed targeting emotion-biased judgements. However, our tasks are not yet suitable for the on-farm testing of emotion in pigs due to the complex training involved.

We studied the ability of pigs to discriminate tone cues using successive and conditional discrimination tasks. Pigs (n = 8) were trained in a successive discrimination Go/No-Go task (Experiment 1) to associate a Go-cue with a reward at the end of a runway and a No-Go-cue with the absence of reward. Latency to reach the goal-box was recorded for each cue-type. Learning of a conditional discrimination task was compared between low-birthweight (LBW, n = 5) and normal-birthweight (NBW, n = 6) pigs (Experiment 2) and between conventional farm (n = 7) and Gottingen miniature (n = 8) pigs (Experiment 3). In this active-choice task, one cue signalled a response in a right goal-box was correct and a second cue signalled a response in a left goal-box was correct. Cues were differentially rewarded. The number of sessions to learn the discrimination and number of correct choices per cue-type were recorded. In Experiment 1, four out of eight pigs showed learning on the task, that is, a higher latency to respond to the No-Go-cue, within 25 sessions. In Experiment 2, eight out of 11 pigs learned the discrimination within 46 sessions. LBW learners learned faster than NBW learners. In Experiment 3, all 15 pigs learned the task within 16 sessions. Gottingen miniature pigs learned faster than conventional farm pigs. While some methodological issues may improve the Go/No-Go design, it is suggested that an active-choice task yields clearer and more consistent results than one relying on latency alone.