Although highly suspected, long-term effects of postpartum experience on behavior and emotions are rarely demonstrated for different reasons. One is the duration of long-term developmental studies; another is the interference of uncontrolled factors during development and their potential resilience effects. Thus, while the impact of early care of babies and their associated experience (being separated from their mother, handled, having suckled, or not during the first hour after birth) is questioned (Klaus and Kennell, 2001) with observable short-term effects (Long et al., 1980; Jansson et al., 1995; Bystrova et al., 2003), some scarce studies on potential long-term effects indicate impacts of postpartum experience still visible in 3- or 5-year-old children (Ringler et al., 1978; Wiberg et al., 1989).

Most evidence of long-term effects of early experience comes from laboratory rats and monkeys and involves maternal separation (see review in Loman and Gunnar, 2010). Even short and temporary separations induce durable effects on emotional reactivity and social competencies (Rhesus monkeys: Spencer-Booth and Hinde, 1971; rats: Weinberg et al., 1978). The famous monkey models of deprived animals are well represented by hunched postures and reduced mobility (Harlow et al., 1965; Kraemer, 1992). Increased immobility when facing a stressor (forced swim test) was also observed in rats that experienced regular long separations from their mothers during their two first weeks of life (Lehmann and Feldon, 2000).

Immobility or locomotor inhibition are observed regularly in stressful situations and are traits included in concepts such as “conservation-withdrawal” or “learned helplessness” (Engel and Schmale, 1972; Maier and Seligman, 1976). Both correspond to situations when inputs become excessive and beyond the organism’s capacity to cope actively with the stressors, e.g., removal of mother (Kaufman and Rosenblum, 1967). In the second case, the animal has learned that trying actively to escape the situation is useless. Separation from the attachment object leads first to protest (active seeking of contact) and then to despair, characterized by activity decline (Bowlby, 1969; Hofer, 1994).

Postpartum handling may require separation from the mother at a very early stage, and intensive handling (bathing, weighting, eye ointment…) that persists even if the infant protests by crying or agitating its arms and legs (Bystrova et al., 2003). This is a typical situation when the individual has no control over the situation and this questions the possibility of its effects on later reaction modalities to stressful situations.

To test this hypothesis, “naturalistic” animal models with individuals that experience impaired maternal care and social living are required. Previous studies show that horses, a species with a strong dam-foal bond, are highly sensitive to postpartum experience. Thus, “routine” handling of foals at birth has short-, medium-, and long-term effects on their attachment to their dam and their social development. Indeed, horses that have been handled at birth spend longer time close to their mothers and interact less often with peers than unhandled individuals, from early stages of development to “adolescence” (Henry et al., 2009). Even a slight interference during first suckling has similar consequences on attachment (Hausberger et al., 2007). The way they are handled at birth influences their emotional reactivity at later stages (de Boyer des Roches et al., 2011). When handling was performed only on one side of the body, the reactions of 10-day-old foals to a human approach differed according to the side stimulated. Foals handled on the right were fewer to accept contact with humans (long delay, contact avoidance). Long-term effects of a single early experience on adult behaviour in this species could be demonstrated despite their “normal” ecological living conditions in the meanwhile.

In the present study, we hypothesized that postpartum handling, which consists in maintaining the foal immobile and rubbing its body all over, would induce a form of learned helplessness (Simpson, 2002) that would have consequences on its later ability, as an adult, to react actively to stressful situations. During this neonatal handling, foals react first by struggling (“protest”) and then remain motionless (“despair”; Simpson, 2002; Henry et al., 2009). They may thus have learned that motor behavior is useless to escape a stressful situation (inescapable stress). Therefore, we compared the reactions of early handled foals and of control (unhandled) foals to stressful situations when one (early adolescence) and 2 (pre-adult stage) years old. Previous studies showed that horses’ reactions to stress include postural, vocal, and locomotor components (Wolff et al., 1997; Visser et al., 2001). Following social separation, horses tend to seek restoration of contact actively through whinnies and active locomotion (trot, canter). Whinnies are commonly used as a marker of stress, above all in social separation situations (Harewood and McGowan, 2005; Pond et al., 2010; Yeon, 2012). We expected the locomotor pattern to be affected by their early experience of inescapable stress while the vocal component should not be affected.

At birth, foals are routinely handled by a human following a standard procedure consisting in maintaining the foals in a recumbent position on the floor and stroking it all over its body for about 1 h. Early handled and unhandled (control) foals were observed in stressful situations: social isolation in familiar and unfamiliar settings and presence of a novel object, when they were one and 2 years old.

One-year-old early handled and control foals reacted strongly to social isolation in their familiar stall by whinnying. There was no difference in the number of whinnies between the two groups (Mann–Whitney U test: N_{early handled} = 9; N_control = 8; z = 0.73, p = 0.48; Figure 1A). Level of locomotor activity did not differ significantly between the two groups (Mann–Whitney U tests: N_{early handled} = 9; N_control = 8; immobility: z = 1.08, p = 0.28; active locomotion: z = −1.17, p = 0.29; Figure 1B). When foals were confronted to a novel object, latencies to approach the object did not differ significantly between groups (Mann–Whitney U test: N_{early handled} = 9; N_control = 8; z = −1.06; p = 0.28; Figure 2A). Nevertheless, the control group’s level of locomotor activity was significantly higher when a novel object was presented (Mann–Whitney U test: N_{early handled} = 9; N_control = 8; immobility: z = 0.48, p = 0.6; slow walk: z = −0.1, p = 0.89; active walk: z = −2.66, p = 0.005; Figure 2B).

Both 2-year-old groups of horses expressed similar high levels of whinnies after social separation (unfamiliar settings: Mann–Whitney U test: N_{early handled} = 9; N_control = 8; z = −0.1, p = 0.89; familiar settings: Mann–Whitney U test: N_{early handled} = 9; N_control = 8; z = −1.36, p = 0.17; Figure 3A). However, in unfamiliar settings, the control group’s level of locomotor activity was significantly higher than that of the early handled group (Mann–Whitney U test: N_{early handled} = 9; N_control = 8; immobility: z = −0.24, p = 0.81; slow walk: z = −1.5, p = 0.14; active walk: z = 2.04, p = 0.03; Figure 3B). On the second day of test, when the settings were less unfamiliar, levels of locomotor activity of the two groups did not differ significantly even though control foals still tended to be more active (Mann–Whitney U test: N_{early handled} = 9; N_control = 8; immobility: z = 0, p = 0.96; slow walk: z = −0.34, p = 0.74; active walk: z = 1.74, p = 0.07; Figure 3B).

As stallions underwent castration when they were 1-year-old, we tested a potential effect of sex on the active locomotion. No difference could be evidenced in the level of active walk between males and females whatever the day of test (Mann–Whitney U test: N_females = 10; N_males = 7; unfamiliar settings: z = −0.34, p = 0.74; familiar settings: z = 0.98, p = 0.31). The castration did not seem to interfere in the way horses react to social isolation.

One single prolonged bout of handling just after birth clearly affected later adult expression of stress. Interestingly, in very stressful situations, such as social separation associated with a novel object or a novel environment, both experimental groups of horses produced the same quantities of whinnies, showing that they were similarly distressed by social separation. On the other hand, early handled foals differed from control (unhandled) foals as their occurrences of active forms of locomotion such as trot or gallop were less frequent. These behaviors reflect contact or escape seeking and therefore are considered as the expression of a distress (Wolff et al., 1997; Visser et al., 2001; Yeon, 2012).

This is an especially major effect as in the meanwhile both groups lived in the same conditions, in social groups with little contact with humans, apart from feeding. Similarly, rats that experienced prolonged repeated maternal separation (several hours per day for 2 weeks) reacted to stress (e.g., forced swim test) by an increased immobility rate (Lehmann and Feldon, 2000). Locomotor inhibition has been considered as depression-like behavior in the forced swim test, the Porsolt, or tail suspension tests (Porsolt et al., 1977; Steru et al., 1985). It is considered to be the expression of behavioral despair (Porsolt et al., 1978). Individual differences in expressing active/passive strategies when facing challenging situations are generally interpreted as variations in personality traits. Thus, in tonic immobility tests, some pigs struggle while others stay immobile but tensed, but all reacted vocally (Erhard et al., 1999). These authors interpreted these responses as indications that all pigs found the situation challenging and that they reacted differently according to their personality traits. In view of our study, one can wonder whether these traits could be the issue of differences in early experience. Emotional reactivity can be modulated, for instance, by maternal behavioral characteristics to which individuals have been exposed in their early life (Francis et al., 2003; Richard-Yris et al., 2005). So, even though our experiment did not allow evaluating the personality traits of the subjects, it suggests that a single experience, very close to birth, may be enough to influence later behavioral reactions to stress.

Indeed, to our knowledge, the present data for horses are the first to demonstrate the effects of postpartum handling experience not so much on adult stress reactivity per seas on the way reactions to stress are expressed. Horses’ reactions to motion-inducing situations include postural/vocal components as well as locomotor components (Wolff et al., 1997; Visser et al., 2001). Long-term effects of early handling obviously affect the locomotor component, but not the other expressions of stress reactivity. The fact that, in the social separation situation, both groups of horses produce similar quantities of whinnies indicates a high level of stress. Whinnies are typical contact calls, mostly produced during separation and aiming to restore contact (Waring, 2003; Lemasson et al., 2009; Pond et al., 2010). It is a typical expression of stress in social separation tests (LeScolan et al., 1997; Wolff et al., 1997; Moons et al., 2005). Interestingly, tranquilization by acepromazine maleate induces a decrease of calling but no changes in locomotor components in separation tests (McCall et al., 2006). These locomotor components are not particular to emotional reactions but reveal increasing arousal (Wolff et al., 1997). Learned helplessness typically decreases an individual’s attempts to escape a situation, as in previous experiences, movements were of no use (Maier and Seligman, 1976). Some authors suggest that active postpartum handling created a situation of learned helplessness (Simpson, 2002). Indeed, foals struggle for a while (“protest”) before becoming motionless (“despair”) during this phase when they are lying on the ground and physically separated from their mother while, at that stage, they are supposed to be up, and suckling (Waring, 2003; Henry et al., 2009). Reports suggest that the young individual must play an active part in bonding with its mother in order to develop a secure attachment (Hausberger et al., 2007). Our results could be interpreted as foals may have learned that movements (struggling with legs) were useless for restoring contact during the handling procedure. Thus, when facing again social separation and novelty, when sub-adult, they may “renounce” seeking actively contact or escape and therefore reduce their active locomotion. In particularly stressful contexts, only the vocal component was expressed, reflecting real stress, while the locomotor component, previously affected by unsuccessful attempts to escape, was inhibited.

These findings are of primary importance as they encourage further reflections on long-term effects of routine handlings of humans and animals.

The experiment was conducted at the “Station expérimentale de Chamberet” (French National Stud Farms, France). Both procedure and testing were conducted in accordance with the French regulations governing the care and use of research animals and have been approved by the scientific committee of the French National Studs. The experiment was performed in accordance with the European Communities Council Directive of 24th November 1986 (86/609/EEC).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.