This is the first study to apply a social-pair training and testing approach in a judgment bias task. We inferred the affective state of slow-growing broiler chickens housed in low- or high-complexity environments from behavioral responses in the JBT. In addition, we evaluated the effects of personality traits (fearfulness and anxiety) and chronic stress (feather corticosterone concentration) on JBT training and testing performance. During 13 days of training across three weeks, 83% of chickens (20 out of 24) were successfully trained to discriminate between multimodal reinforced cues. Chickens showed a generalization gradient response in the JBT, demonstrating that chickens successfully learned the discrimination task3,13,20. All chickens approached and pecked ambiguous cues close to the reward cue more quickly or more often than cues close to the neutral cue, without showing a color or side bias. This suggests a greater expectation of a reward for ambiguous cues near the reward cue. Based on these findings, a social-pair JBT approach can be used as a tool to infer affective states in slow-growing broiler chickens.
Complexity treatment, personality traits, and chronic stress did not impact the number of rounds chickens needed to meet the learning criterium for each training phase. Training chickens in pairs may have attenuated the effects of social isolation and novelty inherent to the JBT environment, especially for chickens with negatively valenced personality traits or affective states. Studies using individual training approaches showed that stress and fearfulness negatively impacted laying hens’ and fast-growing broilers’ training performance in cognitive tests27,34,43. Broilers that were stressed by social isolation showed impaired spatial memory learning compared to a control group27. The authors argued that chronically stressed broilers were less capable of coping with negative conditions associated with the test (isolation and novelty)44,45, sensitizing them to respond to future stressful events (more testing) and provoking a shift in cognitive functions away from spatial learning and toward a stress response27,46. Similarly, rats and mice chronically stressed by social isolation showed impaired spatial learning in a water maze test compared to a control group47,48. Fearful rats made more side errors in Y-maze49 and water maze tasks50, while anxious rats showed poor learning performance in a water maze task51–54 compared to their control group. Furthermore, fearful, stress-sensitive laying hens developed side biases during JBT training34,43. The authors argued that fearful hens use a rigid response strategy during early learning phases by choosing a specific side repeatedly irrespective of success, indicating cognitive inflexibility34,43. In the current study, no such effects of personality or chronic stress were observed on learning success or test responses.
The social-pair testing approach may have attenueated the negative effects of negatively valenced personality traits and chronic stress. During testing, broilers experienced social support from a flock mate, which could increase their motivation to perform tasks. Broilers have a strong motivation for social reinstatement and chickens in natural settings live in relatively small, highly social groups55–57. In line, laying hens exhibited less fear-related behaviors when undergoing an open field test58 and fast-growing broilers performed better in an attention bias test when tested with two conspecifics compared to being tested alone59. These results suggest that chickens benefit from social support in testing environments that require learning or attention23. Our social-pair testing approach could have been especially beneficial for fearful, anxious, or chronically stressed animals, reflected in their similar learning performance compared to birds that were considered fearless, calm, or experiencing less chronic stress.
The learning success rate (20/24 chickens) in this study was greater than reported in earlier studies using an individual approach for fast-growing broilers and laying hens13,15,17,20,21,60, but lower than reported in14. Days needed to train birds were comparable to or faster than most other JBT studies. Training took between 10–30 days for fast-growing broilers, with low learning success rates (between 25 and 51%)13,17, and training took between 13 days and 8 weeks for laying hens, with better learning success rates (between 62 and 100%)14,15,20,21. However, all genetic strains differed from the strain used in the current study, which could influence the result.
We theorize that social facilitation improved chickens’ learning ability. Social learning helps chickens to decide what to eat and avoid23. Aversive behavior of one chicken towards a food item will result in consistent avoidance of that food item in an observing chicken61. As training phases 1A and 1B required birds to peck a reward cue with a food item, social learning (one bird observing another) could have contributed to chickens learning to peck the reward cue containing attractive food items. Furthermore, the paired approach could have facilitated spatial memory development and cue discrimination ability in training phase 2, as young chickens can locate hidden objects due to their developed spatial memory62, which allows them to learn from conspecifics through observation63.
The benefit of a social training approach could differ depending on genetic strain, yet in the current study only a slow-growing broiler strain was tested. Inherent stressors associated with JBT training and testing are the frequent handling by and close proximity with humans, plus repeated removal from home environments and flock mates, which could result in chronic stress64. These JBT procedures may be more distressing to slow-growing broilers than fast-growing broilers, as slow-growing broilers are more reactive to human interaction65–69. The improved training success compared to previous studies could suggest that the presence of a conspecific alleviated some of these negative experiences, thus a social approach may be especially beneficial for slow-growing broiler chickens. As chickens are a social species, this benefit is expected for other genetic strains too. Further research on social approaches in other genetic strains can confirm this.
Chickens from the low-complexity treatment were faster to approach all cues and the middle cue compared to chickens from the high-complexity treatment. Furthermore, more chickens from the low-complexity treatment tended to peck cues than those from the high-complexity treatment. These differences indicate that chickens from simple environments were more optimistic than chickens from enriched environments, in contrast with our hypothesis. One explanation could be related to the asset-protection principle, where high-asset individuals are less willing to risk their lives for a given benefit, making them more cautious, and low-asset individuals more willing to take risks36. Chickens from the high-complexity treatment received eight types of enrichments, including nutritional enrichments such as oats, alfalfa, cabbage, and seeds. Chickens from the low-complexity treatment experienced a simple environment without those enrichments. Broilers in high-complexity pens therefore could be considered high-asset individuals compared to broilers in low-complexity pens, as they had access to more attractive resources. Their willingness to ‘risk’ approaching an ambiguous cue with an expectation to get a food reward might therefore be lower compared to birds from low-complexity pens with little to lose. In line, laying hens housed in high-complexity pens containing nest boxes with straw, perches, peat, wood shavings, hay, and nutritional enrichments (apples and sunflower seeds), thus high-asset individuals, tended to be slower to approach the middle cue in a JBT compared to hens housed in a low-complexity environment (low-asset individuals)15. In contrast, fast-growing broilers housed in high-complexity pens with wood shavings, sand, perches, occupational enrichments, and nutritional enrichments (high-asset individuals) were quicker to approach ambiguous cues than broilers housed in low-complexity pens (low-asset individuals)13. As the complexity treatments in that study were similar to the treatments in the current study, the contrast is possibly due to genetic strain effects. However, the asset-protection principle was argued to not be reflective of long-term behavioral decision-making as tested in the JBT and rather reflects a more short-term state70. Furthermore, the behavioral mechanism has not been confirmed in poultry.
Broilers in the current study were less optimistic when housed in complex environments compared to simple environments, opposite to expectation. The environmental enrichment used could have been inappropriate for slow-growing broiler chickens and negatively impacted their affective state. Our enrichment strategy was to provide a complex and varied environment, maintaining environmental novelty and providing resources to fulfill highly-motivated behavioral needs. This highly complex environment effectively improved emotions and affective states in fast-growing broiler chickens13,59. However, slow-growing broiler chicken strains are more active and interact more with conspecifics and the environment than fast-growing strains71,72, which could have negated the potential benefit of the chosen enrichment items. If the enrichments were unsuited for slow-growing broilers, they might have elicited frustration or other negative emotions, resulting in negative affective states. In line, a highly-complex environment increased chronic stress parameters in mice and corvids compared to animals kept in barren environments73–75. Further supporting this theory and previous research findings, the chickens from the high-complexity treatment tended to show an increased chronic stress response compared to chickens in the low-complexity treatment. The increased chronic stress response in high-complexity chickens could in part be due to the increased human-animal interactions associated with providing temporary enrichments, since slow-growing broilers are more reactive to human interaction than fast-growing strains65–69. Alternatively, the novelty of these enrichments, assuming that novelty was maintained, might have increased birds’ arousal and thus increased corticosterone deposition in feathers. Increased arousal can increase circulating corticosterone concentrations, also when animals experience a positively valenced emotion such as pleasure, excitement, and winning76–78. As the majority of research on environmental enrichments for broiler chickens is focused on fast-growing strains79, we recommend further research assessing slow-growing broilers’ preferences for environmental enrichments.
In the current study, complexity treatments did not affect gait score, with gait being perfect or slightly deficient for all birds assessed. In line, slow-growing broilers generally have good walking ability66,80. Poor gait could impact training responses and reduce the number of successfully trained chickens since the task required chickens to walk toward a cue. Latency to approach cues and proportion of chickens pecking cues were not impacted by gait, in line with fast-growing broilers in a JBT13. These results suggest that observed differences in latencies and proportion of chickens approaching were reflecting a cognitive bias instead of physical limitations to approach the cues.
In order to avoid the effects of personality traits tests on JBT responses, we performed the personality tests ten days after the JBT testing phase. As life experiences shape personality and individual preferences over time35, the JBT procedure could have impacted the chickens’ personality traits. Repeated training and testing could have reduced fear towards humans as chickens habituated to repeated handling81. This study design did not allow us to assess a balanced sample of fearful, fearless, anxious, and calm chickens in both complexity treatments. In addition, our limited sample size may have reduced the statistical power to assess the effects of personality traits and chronic stress on training performance. We recommend further research assessing personality traits (fear and anxiety) and chronic stress with a larger sample size to confirm the lack of impact found in the current study. Furthermore, this study did not incorporate a control group to directly compare a social-pair JBT approach with an individual JBT approach due to time constraints. Yet, even without a direct comparison, the social approach seems to result in improved learning ability compared to training success when birds are tested individually13,15,17,20,21,60.
To conclude, this study is the first to show that a social-pair judgment bias training and testing approach can be used to successfully assess affective states in slow-growing broiler chickens, with no effects of personality traits (fearfulness and anxiety) or relative chronic stress (based on feather corticosterone concentrations) on the chickens’ learning ability during training or testing. The judgment bias test in this study showed that slow-growing broilers housed in unenriched, low-complexity environments were more optimistic to receive a reward in an ambiguous situation than broilers from an enriched, high-complexity environment. Chickens from the high-complexity treatment tended to show an increased chronic stress response and a more negative affective state compared to the chickens in the low-complexity treatment, possibly related to the suitability of provided environmental enrichments.