In the present study, using a Pavlovian to instrumental transfer (PIT) design, we demonstrated that external cues can trigger specific behaviors in horses. In addition, we found that the magnitude of PIT was significantly modulated by individual temperament. These findings have important implications, which are addressed in detail below.
The first aim of this study was to demonstrate the existence of PIT in horses. A detailed analysis of the instrumental responses measured over the three CS presentations during the PIT session indicates that the first CS was associated with the strongest transfer effect and quadrupled the number of correct responses compared with baseline. The results also show that the number of correct responses temporally decreased throughout the three CS presentations, suggesting extinction of the capacity of the CS to elicit the instrumental response. Nevertheless, throughout the entire PIT session, the horses responded significantly more frequently during the CS than the ITI periods. Therefore, to our knowledge, this study is the first to show that a Pavlovian CS can restore an extinguished instrumental response in horses and thus demonstrate PIT in this species.
The present results also revealed that instrumental responses can be modulated by specific temperamental dimensions. During the acquisition phase of instrumental conditioning, the high performers were those that neighed the least in the novel arena and had lower-magnitude flight reactions in the suddenness test; these were indicative of low levels of gregariousness and fearfulness, respectively 
. Negative influences of fearfulness, anxiety, shyness/boldness, or emotional reactivity on acquisition performance have been reported in the literature frequently across many species (e.g., rats: 
, dogs: 
, horses: 
). The present observation of negative relationships between acquisition performance on the one hand and the dimensions of fearfulness and gregariousness on the other may be explained in terms of individual differences in stress levels between horses, which in turn may depend on temperament. Indeed, animals high in gregariousness and fearfulness are more prone to becoming stressed during the instrumental learning procedure, which represents a novel situation and involves separation from familiar conspecifics. The fact that stress is often reported to be detrimental to learning and memory (for reviews, see 
) may explain the negative relationships between these temperamental dimensions and instrumental acquisition performance.
High responders during the PIT test session were those that neighed the most during the novel arena test, had greater flight distances during the suddenness test, touched the novel object less often, and reacted the most to von Frey filaments. Thus, they were more gregarious, more fearful, and more sensitive to touch, according to Lansade et al. 
. This relationship between a specific pattern of temperamental dimensions and PIT efficacy has never been shown before. Because PIT involves an interaction between Pavlovian and instrumental processes 
, the effects of these specific temperamental dimensions on PIT efficacy might be explained by their possible influences on each of these processes.
First, the influence of these behavioral dimensions on the magnitude of PIT may be due to their effects on Pavlovian processes. Dimensions such as gregariousness and fearfulness might have enhanced the motivational value of the CS during Pavlovian conditioning. Animals that are more prone to becoming stressed under social isolation or when facing novelty might have established a stronger CS-reward association than other horses. In line with this view, previous studies reported a strengthening impact of stress on Pavlovian conditioning 
. Consequently, the capacity of the CS to trigger an instrumental response would result in a higher PIT magnitude in these horses. Particularly, previous studies indicated dendritic hypertrophy in the basolateral complex of the amygdala or increased reactivity of amygdala neurons that potentiate emotionality or emotional learning in stressed animals 
. The present results indicate that high trait fearfulness and gregariousness enhance Pavlovian conditioning but disrupt instrumental learning. These findings (and previous studies that indicated modulation of stress by memory systems 
) suggest that given temperamental profiles might favor or disrupt learning and memory, depending on the cognitive processes engaged in the task.
The relationship between reaction to von Frey filaments and PIT magnitude can be explained along the same line. Indeed, the response to von Frey filaments reflects a dimension of temperament called sensory sensitivity 
. According to Dunn 
, high–sensory-sensitivity humans notice sensory stimuli quite readily and perceive more sensory events than others; such people are easily distracted by movements, sounds, or smells. Horses with high sensory sensitivity may have noticed the CS more easily, facilitating the incentivization of this CS. Accordingly, Talmi et al. 
described that PIT was larger when participants were aware of the presence of the Pavlovian CS.
Several studies have reported individual variations among rats in the propensity to attribute incentive motivational properties to reward cues. Such variability may partially explain the present observation of a specific pattern of behavioral dimensions on PIT performance. These studies showed that when a cue was paired with a food or cocaine reward, rats bred for high reactivity to novel environments learned to approach the cue, whereas rats bred for low reactivity to novelty learned to approach the location of food delivery. These results indicated that only rats selected for high reactivity to novelty attributed incentive value to the rewarded cue 
. Animals that attribute greater incentive value to rewarded cues also work harder to obtain rewards 
and are more prone to drug-cue–controlled behaviors 
. Together, these findings suggest that individual variation in the propensity to attribute incentive value to reward cues or form Pavlovian CS-UCS associations may explain the influence of specific behavioral dimensions, such as novelty seeking or sensitivity seeking, in the modulation of PIT efficacy.
The second explanation for the influence of this pattern of temperament on PIT efficacy may be that these dimensions of temperament affect instrumental processes. Recent research in humans and rats suggests that stress favors habitual process during instrumental conditioning. This was demonstrated by Schwabe & Wolf 
and Dias-Ferreira et al. 
, who have subjected individuals to acute or chronic stress before instrumental learning. When tested in a devaluation experiment, stressed individuals were less sensitive to change in the value of the outcome compared with non-stressed individuals, suggesting habitual rather than goal-directed behavior (for a review: 
). Thus, in the present study, because of the higher level of stress imposed during instrumental training, the most fearful and gregarious animals may have preferentially formed habitual responses. Furthermore, it seems that PIT is stronger when the behavior is controlled by habitual (rather than goal-directed) processes. This was demonstrated by Holland 
, who found that extended training of an instrumental response known to favor habitual processes enhances susceptibility to the facilitatory effects of Pavlovian CS during PIT (for a review: 
). Following this reasoning, we suspect that the most fearful and gregarious individuals would be more sensitive to the facilitating effect of CS during PIT, because they would preferentially form habitual responses.
Finally, we postulate that individuals with a fearful, gregarious, and sensitive temperament profile exhibited stronger PIT response because this profile confers specific constitutive cognitive abilities to horses independently of the emotion felt during the learning processes. Indeed, previous studies in rodents showed an impact of anxiety trait on cognitive performance 
. As hypothesized for these lines of rodents, good PIT performance in fearful, gregarious, and sensitive horses could be mainly cognitively driven. Indeed, this temperament profile could be associated with enhanced or more-accurate processing of environmental cues. This cognitive ability would result in both enhanced PIT performance when a CS is presented during the transfer test and strengthened emotional responses during temperament tests. Continuing research may provide a better understanding of the relationship between temperamental traits and cognitive performances.
The demonstration of the existence of a Pavlovian to Instrumental Transfer in a domesticated animal (i.e., horses) could optimize its use by humans through improvement in training methods. Especially, PIT could be useful to elicit previously established behavior when the delivery of reinforcement is impractical. It could also provide a better understanding of existing training methods in which PIT is probably involved, such as clicker training 
. Moreover, this experiment shows that the influence of external cues on individual behavior depends on temperament. The study of the influence of temperament on sensitivity to external cues could be further developed in several species and in different contexts involving PIT, such as addiction, shopping behavior, and animal training, in order to identify predisposing factors to the presently described effects.