Department of Physics, Chemistry and Biology (IFM)

Motivated by past experiences, current life conditions, personality or state of mind, we create expectations and accommodate our actions. Humans develop judgement patterns for particular situations, sometimes leading to distorted perception and illogical interpretation of events – cognitive bias that may involve attention, memory, judgment and risk assessment for example. In non-human animals, the term judgement bias is commonly used to describe how an individual reacts to an ambiguous stimulus compared to a determined positive or negative stimulus. Cognitive bias have also been defined as all cognition affected by an emotional state and named “affect-modulated cognition” by Crump et al. (2018). Mendl et al (2009) suggested the term “affect induced cognitive bias” to describe an area of study that investigates the influence of affective state on information processing in animals. As it has been observed in human and non-human animals, the emotional state can affect cognition. A state can be defined within a scale of how much an individual has to do in order to cope with the environment, and how well coping attempts succeed. Under this definition, a poor state would be that in which a lot of effort (e.g. amount of energy spent) has to come from the individual in order to cope with a situation, or a case in which the individuals’ attempts to cope repeatedly fail. Emotion is hard to define and tricky to analyze under any definition, but here I think of it as discussed by Paul & Mendl (2018): an internal state of an individuals’ central nervous system, elicited by instrumental reinforcers, that gives rise to physiological, behavioral and cognitive responses. Instrumental reinforcers that can be something an animal will work for (positive) or something an animal will work to avoid. Indeed, emotions set a mood that regulates incoming stimuli, focusing the attention on the mood-related aspects of it. Consequently, they are in control of what the individual regains in its memory, shaping the processing of information and the accuracy of the conclusions. Mood has been defined as a relatively enduring affective state that arises when an experience in one context modifies the individual’s reaction to future events. Additionally, it is argued that some of these effects may be the result of trait-related conditions, as well as or instead of, state differences in affect, in human cases. A trait is here defined as a component that, in order to be favored by natural selection, must improve an animal’s fitness. Affect refers to an experienced emotion, which can be defined as a stimulus-directed affective state, consists of behavioral, physiological, and cognitive components, and may occur outside 3 awareness. Humans reporting negative affective states, e.g. depression or anxiety, tend to make negative judgements about the future, and are commonly referred to as “pessimistic”, while people reporting positive states tend to judge the same ambiguous information with an “optimistic” outcome.

 

Early studies on cognitive bias, in non-human animals, focused on the correlation between different environments (e.g. enriched vs barren housing) and the animal’s response to ambiguous stimuli. Harding et al. (2004), the first published study regarding cognitive bias as an animal welfare indicator, concluded that rats living in unpredictable environments were more pessimistic, displaying behaviors that suggest negative bias, such as fewer and slower responses to ambiguous stimuli. Likewise, Burman et al. (2008) found that rats held in barren environments responded with less excitement to an ambiguous stimulus, when their responses were compared to those of rats being held with enriched surroundings. Analogous studies mentioned ahead had similar results, suggesting that negative affective states are associated with negative cognitive bias in non-human animals too. However, in those cases, behaviors are harder to interpret, there is a lack of positive affective state measures and many of the criterion lack a priori hypotheses for how they should change according to the individual’s emotional state. Nonetheless, cognitive bias tests have been set as a very reliable and non-evasive indicator of animal welfare, for several mammal, avian and invertebrate species. 

Studies have found that negative affective states are not only associated with shifts in judgement but also other cognitive processes such as attention and memory, both in human and non-human animals. The understanding from human studies is that cognitive function is likely enhanced by short-term exposure to low levels of corticosterone, and impaired by short-term or sustained exposure to higher levels of this stress hormone. Receptors for the latter, glucocorticoid receptors, are abundant in two important brain regions for the modulation of learning and memory - the hippocampus and the amygdala. Mendl et al. (2009) summarized the various reports asserting that people in a negative affective state (e.g. anxiety) show enhanced attention to threatening stimuli, when compared to people in a more positive state. Crump et al. (2018), confirmed this evidence and reviewed the inconclusive 4 results for animal species, stating that attention biases are observed, related to welfare and very informative across animal taxa. There are numerous indications that affective state influences memory retrieval, with happier people being more likely to recall positive memories and unhappy or depressed people more likely to recall negative ones. This question was analyzed in Paul et al. (2005), where researchers discuss the indisputable role of the amygdala in storing emotional memories. Elevations of the hormones catecholamines and/or glucocorticoids in the hippocampus are associated with enhanced memory and learning any kind of event: positive, negative or neutral. However, if those increases are too big, they will have disruptive effects on memory, which makes it hard to interpret results. 

Furthermore, it has been noted that the term cognitive bias suggests an irrational decisionmaking process, and although it is often associated with detrimental emotional states, a biased decision may prove itself adaptive if taken into consideration with the information on which the decision is based. There is a possibility that emotions have developed to meet different environmental demands, leading to links with different cognitive methods. Anxiety facilitates vigilance and preparation for action, while depression triggers a sharper reflection of events that have led to failure or loss - both presumably adaptive qualities. Ultimately, attending to arousing stimuli, regardless of whether they are positive (e.g. sexual, food related) or negative (e.g. threat) is more likely to contribute to survival and reproduction than attending to neutral stimuli.

 

In the hope of finding more about the function of cognitive bias, emerging research is looking at its possible link with personality. Is there a personality component to cognitive bias? At what level are individuals bound to be optimistic or pessimistic, independently of their current life circumstances? Questions that are yet to be answered both in human and non-human animals, and the motivated the present work. Réale et al. (2007) defined animal personality as the differences between individuals’ average level of behavior that are repeatable across time and contexts, thus the same definition was adopted in this study. Personality has been associated with cognition in several occasions, setting a precedent for possible outcomes and interpretations of the results in this study. Amongst black-capped chickadees, exploration of novel environments has been proven to be correlated with learning speed and slow-exploring individuals may be generally more able to modify their behavior to adapt to different stimuli. Guenther et al. (2014) found strong positive relationships between all personality traits and learning speed in cavies, whereas flexibility was negatively associated with aggressiveness. Researchers working with zebrafish proposed that personality-related cognitive traits partake in mediating the trade-off between current and future reproduction. Experiments undertaken with mynas show that individuals that solved a foraging task were quicker in a discrimination learning challenge, but slower to shift their behavior when the cues changed. In mice, imposed social subordination impairs exploratory behavior and general cognitive abilities including learning. Nonetheless, within the same group, naturally subordinate individuals do not perform below average in learning tasks. In 2016, Asher et al. found a positive correlation between proactive personality traits in pigs and optimism and later, a similar link was found in dogs.

 

Such questions were only recently investigated in equids and very briefly in donkeys. Horses are able of complex cognitive processes such as relational concept learning, discrimination, categorization, quantification and long-term memory, and have shown variable judgement bias, for example, when housed in either restricted or naturalistic situations and when housed either in stalls or herds. Additionally, fearfulness and stress in this species have been associated to shorter working memory.

Although from different lineages and having different ecologies, donkeys and horses belong to the same genus and share many similarities, probably including cognitive strategies. Donkeys have been able to solve a detour task, shown discrimination learning abilities and were able to recall the location of a hidden object after a delay. The only reported results for judgment bias tasks in donkeys comes from McGuire et al. (2018), who looked at the performance of horses and donkeys, to conclude that rescued individuals were more optimistic than non-rescued individuals.

Miniature donkeys have been evaluated differently than other donkey breeds in physiological tests but have not been reported to be different in a way that could influence the results for the kind of behavioral tasks performed here. However, there is a strong possibility that selective breeding has led to temperamental and cognitive differences, as manifested in horses.

 

Whereas most cognitive bias studies compare results for different treatments (e.g. different housing quality, different levels of arousal, training style), here only individual differences were analyzed. Since all study subjects were under the same conditions during and at least for three years preceding the study period, any differences in their response to the trials should be related to early life experiences, personality, or short-term stress. Thus, it was hypothesized that individuals should vary in their judgement bias upon an ambiguous stimulus, and that those bias would be associated both to personality factors and other cognitive processes. Particularly, nervous individuals were expected to be more pessimistic, and consequently to perform worse in the cognitive tasks. The capacity to, based on aspects of personality, predict that an individual might more often than others experience negative emotions, and for that reason have impaired cognitive abilities, is of extreme importance for animal welfare studies and care takers who interact with animals on a daily basis. For the same reasons, it is relevant to understand at what level individuals living under good conditions and showing low stress levels might experience negative mental states and low expectations towards new stimuli.