Interestingly, we observed significantly different time requirements to complete the same training paradigm between cynomolgus and rhesus macaques suggests there are inherent learning differences and motivational preferences between species [76,77], although the basis for this has yet to be defined. However, this difference of 3 h versus 5 h average total training time in rhesus and cynomolgus macaques, respectively, has little clinical relevance; this is a modest difference in investment in time to train behaviors that will be used in the majority of animals for years. Notwithstanding, understanding the average time commitment required for training a certain animal could have practical implications for study planning related to time management and budgeting. Knowledge of these behavioral differences could prove to be valuable in guiding future model selection, in addition to shaping effective behavioral management practices. From the scientific perspective, deeper understanding of differences in learning behavior influences preclinical model selection to ensure members of the selected species are able to demonstrate the capacity to cooperate, cope, and comply with study procedures in ways that more closely represent the clinical situation. Overall, advancements in our understanding of species-specific differences in training ability can help improve accuracy of study timeline planning, and improve animal welfare by increasing understanding of what is necessary in order to prepare animals to acquire the coping skills necessary to flourish in the research environment.

We also focused on the interaction between temperament and skills acquisition since behavioral inhibition is associated with an increased risk for stress, and subsequently, anxiety and depression [2,8,9,41,42]. Behaviorally inhibited primates were significantly less likely to accept treats from novel handlers, consistent with vigilance, neophobia, and avoidant tendencies during unfamiliar situations. These results agree with previous research showing that inhibited primates are less likely to directly take treats, and further support the use of the human intruder test to accurately assess temperament [20]. Training can support cognitive processes capable of moderating negative reactivity or modulate fearful reactions to allow instead for engagement with frequent, rewarding, and successful interactions, an adaptation consistent with positive coping that decreases stress. In this study, behaviorally inhibited NHPs successfully performed target behaviors at the same rate as those that were more exploratory, suggesting the early-phase mixed reinforcement used in our program is useful to support skill acquisition and positive coping mechanisms in animals that display high levels of reactivity and distress to novel or unfamiliar objects or people.

Temperament-associated differences in willingness to accept highly palatable treats or other appetitive rewards has important implications related to motivational preferences, which can help inform the design of the most suitable training paradigm for a given animal (e.g., PRT only vs. mixed reinforcement). Previous studies have shown that inhibited primates are significantly less successful in training with a PRT-only based approach, which might be explained by lower motivation to accept positive reinforcers, such as high-valued treats, by behaviorally inhibited animals relative to their more exploratory counterparts [20]. For these inhibited animals that exhibit more fearful, avoidant, hesitant, and relatively neophobic behaviors, the appeal of removing (NR) something unwanted may initially be more motivational than appetitive (PR) rewards [75]. When an unwanted stimulus or condition is subtracted (NR), these animals experience relief, a powerful reinforcer, and trust can be built on the understanding that the trained interaction is safe. When combined sparingly with PRT, NRT can be used to improve training efficacy to reinforce that all interactions with trainers end in a favorable outcome for an animal [68,78,79]. We show that although inhibited animals were less likely to take treats at the initiation of training, they were accepting treats comparably to exploratory animals by the end of the P-Phase of training. Using P-phase as an easy learning phase to modulate reactive tendencies and instead engage animals with frequent rewarding appeared to effectively diminish the effects of temperament in future phases, evident by the absence of significant difference in the time required to complete subsequent training phases. It should be highlighted that during Phase-1 and -2 training, primates were offered a high valued “jackpot” at the end of each training session to offset potential training-related anxiety, foster a positive association with training. Affective state-based cognitive biases can occur when there is no way to buffer or counter-condition a stressful event, and post-training high value rewards can help provide this buffer [80]. It is important to note that even the more inhibited animals were able to benefit from the “jackpot” reward strategy owing to the coping skills and willingness to accept appetitive reinforcers developed during the P-Phase of training and the mixed reinforcement-based training paradigm. These findings suggest that the mixed reinforcement training model, with the incorporation of appropriate familiarization and desensitization steps (in the form of the P-Phase here), supports the learning and development of improved coping strategies required for training success in both exploratory animals, and notably, inhibited animals. The combination of push–pull motivators utilized by mixed-reinforcement may be more effective than either push or pull motivators used in isolation, and should be considered in the design of training programs, especially for use in animals with inhibited temperaments who are less likely to have high success with appetitive rewards alone [79].

We did not explore the role of social learning in this analysis. Animals have demonstrated the ability to learn at a faster rate after observing conspecifics performing a task, and animals in this study were housed among animals with varying experience with cooperative tasks [81]. Others have shown the role observational learning has in positively shaping animal perceptions of handlers/trainers [82]. We generally find that animals who observe conspecifics having positive interactions with trainers have reduced overall anxiety behaviors related to close human contact. As such, future studies could investigate the effects of the incorporation of social learning on similarly designed training paradigms. As macaques are highly social animals with hierarchical societies, social rank should also be considered as a variable that may have a potential impact on training efficiency during the learning of complex tasks; rank was not analyzed in this study. Such differences have previously been described in rhesus macaques during simple task training [83,84,85].

Despite individual differences in time to complete training, it is important to note that all animals in our study, regardless of temperament, age, sex, or other demographic characteristics, were able to successfully complete the entire training paradigm and acquire the coping skills necessary for appropriately dealing with exposure to potentially adverse medical situations. In biomedical research, ideal sample choice not only includes selecting the best species for a specific question of interest, but also requires the use of a set of subjects that are as representative of the target population as possible. This has most notably led to a recent push for sex-balancing in scientific studies, but also supports the inclusion of other demographic characteristics, such as temperament, to improve generalizability. Temperament-related selection bias can lead to favoring exploratory animals when candidates are chosen based on preconceived perceptions of their trainability, willingness to cooperate, lack of aggression, and other indicators of affective state [86]. Considering temperament is an interaction between individual biological factors and contextual factors, excluding more inhibited animals in favor of those that are more exploratory has the potential to bias results [7,8,9,10,11]. Despite preconceived notions, the present findings show that inhibited animals can be equally successful at learning a complex training task, and should not be excluded from studies based on perceived study suitability. Using behavioral management as a great equalizer, behaviorally inhibited animals can develop equivalent coping skills to protect their welfare and limit stress bias, so that study populations are more generalizable, improving validity and translatability [36].

Welfare is largely affected by the ways that an animal perceives both its environment and its ability to exert control over it, both of which are closely intertwined with temperament, coping skills, and training. This study shows that well-designed cooperative training has the ability to enable animals of any temperament to develop the coping abilities required to adequately deal with routinely experienced medically necessary aversive situations and reduce consequent stress, demonstrating cooperative training’s critical role in fostering overall welfare in captive animals [68]. Further, training often serves as a form of enrichment, giving animals the chance to perform novel physical and cognitive activities, in addition to providing the opportunity to bond with familiar trainers. Captive animals benefit from the opportunity to learn and perform novel behaviors, and training provides this opportunity [21,87].

An important benefit of successful training is the opportunity for animals to express choice and control over their own care, a key component of welfare. Without proper training, animals often develop a learned helplessness to deal with potentially aversive experiences. Learned helplessness, associated with the inability to escape a stressful situation, can induce uncontrollable, unmediated stress and anxiety [88]. Beyond its ethical implications, the welfare benefits and potential reduction in stress introduced by training can help improve the validity of scientific results and the translational value of the research model. Excess stress has been shown to affect a range of research-related physiologic outcome measures such as heart rate, blood pressure, cortisol, and functional immune state in humans [89,90,91,92], and its reduction through training may lead to a more representative model with less stress-related confounding of results. We have demonstrated previously that the ability to make choices to cooperate with a task, learned during training, fosters a sense of control that is protective psychologically and physically in NHPs [68].

Beyond its significance for animals in captivity, this research has implications for the management of clinical patients who experience potentially fear-inducing medical procedures for the treatment of a variety of acute and chronic medical conditions. Coping skills training in patients can effectively reduce anxiety and increase compliance with treatment to improve outcomes [93,94,95], but there are gaps in understanding of conditions necessary to support practical treatment options. Primates have become increasingly prominent in understanding biological mechanisms underlying neuropsychiatric disorders, e.g., autism spectrum disorder, emphasizing their relevance in biobehavioral research towards successful interventions [96,97,98]. Behavioral management techniques that foster coping in animals undergoing intensive medical management have strong potential to similarly benefit clinical patients. For example, certain adjustments in management technique for behaviorally inhibited patients, such as reducing the ability to perform avoidance-type behaviors and increasing the opportunity to experience reward while in a controlled setting, could more successfully address the needs of these patients in order to improve coping, foster higher levels of cooperation with medical procedures, and reduce overall stress.