Visual laterality in pigs: monocular viewing influences emotional reactions in pigs

Understanding animal emotions is an important scientific and ethical question but assessing emotional valence is still considered challenging. As the observation of lateralization (hemispheric asymmetries in structure and/or function) can provide insight into the underlying processes of the cognitive, physiological and behavioural components of emotions, it is a promising approach for studying them. The emotional valence hypothesis states that positive emotions are mostly processed by the left hemisphere, while negative emotions are mostly processed by the right hemisphere. Support for this hypothesis is still not conclusive; therefore, our study tested it in the context of visual laterality for viewing positive or negative emotionally conditioned stimuli. Ninety male piglets were either positively (food-reward) or negatively (mild punishment) conditioned to an object. Afterwards, the object was presented without the reinforcer under three different treatments: patch on the left or right eye (reducing input to the contralateral hemisphere) or patch between the eyes (the control). Monocular viewing had no clear effects on the negatively conditioned subjects. In contrast, in the positively conditioned group, covering the right eye caused a longer interruption of vocalization, a longer latency to touch the object, a shorter duration of exploring the arena and an increased vagal activity compared to the control. This suggests that reduced processing in the left hemisphere leads to heightened attention that is accompanied by a general orienting response, possibly resulting from a reduced positive appraisal. These findings therefore partially support the emotional valence hypothesis and suggest an important role of the left hemisphere in the quick recognition of a positive stimulus. This study demonstrated that investigating the lateralized processing of emotions can provide insight into the mechanisms of positive appraisal in animals.

Assessing emotions addresses important scientific and ethical issues that could permit a better understanding of (proximate) behavioural control of animals (Gygax, 2017) and that have implications for animal welfare (Boissy et al., 2007). Emotions can be defined as intense and short-lived affective responses to an event and are mostly accompanied by neurophysiological, behavioural, cognitive and subjective changes (D esir e, Boissy, & Veissier, 2002). To objectively assess animal emotions, a two-dimensional framework with arousal and valence as axes has been introduced (Mendl, Burman, & Paul, 2010). While arousal can be assessed from physiological reactions (e.g. heart rate increases), it remains difficult to measure and distinguish emotional reactions along the valence axis (Paul, Harding, & Mendl, 2005). To address this issue, cognitive approaches have been established (e.g. cognitive bias: Harding, Paul, & Mendl, 2004) and physiological measurements of heart rate and blood pressure variability have been proposed (Krause, Puppe, & Langbein, 2017;von Borell et al., 2007). Lastly, in particular contexts the least arduous way to assess emotional valence remains the use of the behavioural indicators of approach and avoidance (Cabanac, 1992;Dawkins, 1990). Ideally, combining multiple (cognitive, physiological and behavioural) indicators into a componential view would help to increase comprehension of animal emotions (Paul et al., 2005).
The study of lateralization can, arguably, utilize this componential approach to the study of emotions because it has been shown that the cognitive, physiological and behavioural components of emotions are underlain by lateralized cerebral processes in many vertebrate taxa (Leliveld, Langbein, & Puppe, 2013;Rogers, 2010;Rogers, Vallortigara, & Andrew, 2013). Lateralization refers to the fact that the brain hemispheres can play different roles in many cerebral processes (MacNeilage, Rogers, & Vallortigara, 2009;Rogers & Vallortigara, 2015;Vallortigara & Versace, 2017), such as in emotional processing. The emotional valence hypothesis states that positive emotions are processed predominantly by the left (L) hemisphere while negative emotions are processed predominantly by the right (R) hemisphere (Demaree, Everhart, Youngstrom, & Harrison, 2005;Quaranta, Siniscalchi, & Vallortigara, 2007;Siniscalchi, Lusito, Vallortigara, & Quaranta, 2013) and seems to prevail over alternative hypotheses, for example the approach/ withdrawal hypothesis (Leliveld et al., 2013). Indeed, many animal taxa show opposite hemispheric dominances for perceiving food or a predator. However, this may not be explained only by differences in emotional valence since such naturally emotionally charged stimuli often trigger highly aroused states and involve other cognitive functions. Emotional conditioning standardizes the emotional contexts because it permits the use of the same stimulus (for instance an artificial object, unlikely to elicit species-specific behavioural responses) for testing negative and positive emotional processing (Bisazza, Cantalupo, Capocchiano, & Vallortigara, 2000). Therefore, this approach allows a primary focus on emotional valence (Mendl, Burman, Parker, & Paul, 2009) without interference from arousal or from other lateralized cerebral processes (e.g. processes involved in flight and feeding; Gupta, Raymond, & Vuilleumier, 2018;Rogers et al., 2013). However, few studies on emotional lateralization have used emotionally conditioned stimuli (e.g. Bisazza et al., 2000;Reddon & Hurd, 2009). Also, most previous studies have analysed side preferences, while manipulation of lateralized processing by reducing sensory input to one hemisphere (e.g. by covering one eye to induce monocular viewing) allows researchers to investigate the role of each hemisphere in a cerebral process (Vallortigara, 2000). Therefore, our aim was to test the emotional valence hypothesis using emotional conditioning and manipulated monocular viewing.
As pigs, Sus scrofa, are acknowledged to be a suitable animal model in physiology and neuroscience (Lind et al., 2007;Swindle & Smith, 2008), they may offer new insights into common patterns of emotional lateralization in vertebrates. The established methods for assessing pigs' emotions often combine behavioural indicators with physiological parameters such as heart rate variability measurements (Düpjan et al., 2011;Krause et al., 2017). In cognitive approaches where pigs were conditioned to stimuli, this combination of indicators was also used to study emotional valence (Stracke, Otten, Tuchscherer, Puppe, & Düpjan, 2017;Zebunke, Langbein, Manteuffel, & Puppe, 2011). Moreover, in pigs lateralization has been recently found in motor functions and during aggressive encounters (Camerlink, Menneson, Turner, Farish, & Arnott, 2018;Goursot, Düpjan, Tuchscherer, Puppe, & Leliveld, 2017;Goursot et al., 2018), making them a good candidate species to test the emotional valence hypothesis.
In this study, we decided to focus on vision as one of the most investigated lateralized modalities (Andrew, Mench, & Rainey, 1982;Vallortigara, 2000). Pigs have a good visual system (Broom, Sena, & Moynihan, 2009;McLeman, Mendl, Jones, White, & Wathes, 2005) and a high degree of decussation in their optical fibres (87.8%; Herron, Martin, & Joyce, 1978). This means that visual stimuli are predominantly processed in the contralateral hemisphere (Jones, 1989). Therefore, monocular presentation (which can be achieved by covering the other eye) in this species would guarantee a reduced input to the ipsilateral hemisphere and permit insight into the role of each hemisphere. To test the emotional valence hypothesis in the context of visual laterality in pigs, we studied their behavioural and physiological responses during monocular, compared to binocular, presentation of an emotional conditioned stimulus. We expected that seeing a positive conditioned object with the R eye covered (reduced input to the L hemisphere) would lead to a less positive appraisal than seeing it either with both eyes or with the L eye covered. In addition, we expected that seeing a negative conditioned object with the L eye covered (reduced input to the R hemisphere) would lead to a less negative appraisal than seeing it with both eyes or with the R eye covered.

Ethical Note
The experimental procedure was approved by the ethics committee of the federal state of Mecklenburg-Western Pomerania, Germany (LALLF M-V/TSD/7221.3-2-046/14) and adhered to the legal requirements of the European Union (directive 2010/63/EU) and also the ASAB/ABS guidelines for the use of animals in research. Reduction of animals used was guaranteed by performing a sample size determination using the SAS Power and Sample Size application, a desktop application that provides power and sample size computations for a variety of statistical analyses (e.g. analyses of variance and linear models). Refinement was achieved by providing environmental enrichment exceeding the usual chewing toys (see section Subjects, housing and husbandry), and by reducing social isolation to a minimum (allowing acoustic contact where possible).

Subjects, Housing and Husbandry
The study was conducted at the Experimental Facility for Pigs (EAS) of the Leibniz Institute for Farm Animal Biology (FBN) in Dummerstorf, Germany. The subjects were 90 prepubertal, uncastrated male German Landrace piglets 5e6 weeks old. Experiments were performed with five consecutive replicates between August and November 2015. At 4 weeks of age, 20 piglets per replicate were preselected from a greater pool based on their health status, absence of injuries and weight (> 5 kg). When possible, the number of full siblings was set at two per sow or (if not enough sows were available) at four per sow (always even numbers for a pairwise design). The preselected piglets were weaned and grouped in a pen (2.50 Â 3.95 m) with fully slatted plastic floors and two solid concrete sections. They had access to food and water ad libitum. Straw and some other physical enrichment (buckets, rags, etc.) were provided twice a day during the entire experimental period, except on days 5 and 6, when no experiments took place. For the experiment, 18 subjects consisting of nine pairs of full siblings per replicate (90 subjects in 45 pairs in total) were randomly selected among the preselected group and divided into two groups: the positive and the negative conditioning group (nine subjects per replicate, 45 subjects in total per conditioning group, balanced for kinship). Each subject was randomly given an ID number, which determined the individual test order throughout the experiment. The experimental procedure is summarized in Fig. 1. Three days after weaning (day 1), the experiment began and lasted for 12 days. The pigs were habituated to being handled by the experimenters (two sessions of 1 h per day on days 1e3), to the food reward and to the experimental set-up (see section Habituation below).

Experimental Set-up
The conditioning took place in a conditioning arena (1.50 Â 1.50 m) that was connected to the home pen through a corridor (0.51 Â 1.85 m). A sliding guillotine door (0.39 Â 0.45 m) that could be operated from the corridor was used to provide access to the arena. The tests took place in a test arena (3 Â 3 m) located in a sound-attenuated room, with a microphone (Sennheiser ME64/K6) placed centrally above and connected to a digital audio recorder (Marantz PMD 670; sampling rate: 44.1 kHz; accuracy: 16 bit; mono). Two cameras, connected to a digital video recorder, were positioned centrally above both arenas. The object was an orange ball hanging from a metal cord that was lowered into the arena from outside. A preparation box (94.5 Â 40 cm and 73 cm high) was used for fixing the heart rate measurement belt.

Conditioning
Conditioning started on day 1 (one session on days 1 and 3, two sessions on day 2) in the conditioning arena. To optimize learning, the conditioning occurred in groups of three subjects that were randomly chosen for each session. During the first and fourth conditioning sessions (days 1 and 3), a heart rate measurement belt connected to a heart rate monitor via Bluetooth (Polar system; measurement belt: WearLink þ H3 sensor; monitor: RS800CX; Polar Electro Oy, Kempele, Finland) was fitted on the pigs in the preparation box so that they became accustomed to the procedure. The measurement belt was fixed directly behind the front legs with the transmitter positioned in the left 'armpit' as described in Düpjan et al. (2011). Ultrasound transmission gel (Henry Schein, Melville, NY, U.S.A.) was used to improve contact with the skin. A conditioning session consisted of four trials (except on day 4 in the test arena, only two trials) of 1 min each and with a break of 1 min between trials: in total, a session lasted 8 min (4 min on day 4). During a trial, the object (an orange ball, Zeus Bomber dog toy, 18 cm diameter) was lowered into the middle of the conditioning (or test) arena by the experimenter. For positive conditioning, 2 tablespoons of food reward were spread on the object (positive reinforcer: a viscous mixture of dry food (Porcistart, Trede und von Pein, Itzehoe, Germany) with apple juice), from which the pigs could eat directly. For the negative conditioning, we used the same methods validated in Düpjan, Stracke, Tuchscherer, and Puppe (2017): a plastic bag was waved by the experimenter above or in front of the subjects until they showed a clear avoidance response. This negative reinforcer was given as soon as one pig touched the object, except during the last trial of the first conditioning session, the first trial of the conditioning session on day 4 and one pseudorandomly chosen trial during all other conditioning sessions (with the same trial for a maximum of two sessions) in which it was given at a randomized time point. This procedure allowed us to give the reinforcer at least once per session. Before and after each session, the subjects received a small portion of food reward in the corridor to maintain their motivation to participate in the conditioning sessions. To prevent extinction of the emotional response during the tests, the pigs received further conditioning systematically 1 day before each test (days 7, 9 and 11; one session per day).

Habituation
On day 4, the pigs were habituated to the test procedure in the test arena and underwent one conditioning session there. Each pig was guided from its home pen into the sound-attenuated room. In the entrance area, in the preparation box, a heart rate measurement belt was fitted on the subject, using the same procedure as during the conditioning sessions, and an eye patch (6.5 x 4.5 cm cotton textile fixed with adhesive tape (Fixomull stretch, BSN medical GmbH, Germany)) was fixed on its forehead so that the pigs became accustomed to the test procedure. After this, the pig was guided into the test arena and left alone for 4 min, while the experimenter stayed in the room, outside the test arena, hidden from view.

Tests
On days 8, 10 and 12, the subjects were individually tested with three different treatments (Fig. 2). Before the test sessions, subjects were fitted with the heart rate measurement belt and, according to the treatment, an eye patch either covering the left eye (left eye covered, LEC), right eye (right eye covered, REC) or on the forehead as a control (binocular viewing, BIN). After this, the pig was guided into the test arena and left alone for 4 min, while the experimenter stayed in the room, outside the test arena, hidden from view (same as in the habituation procedure). The object, without the reinforcer, was lowered in the middle of the test arena after 2 min and removed again after 1 min by the experimenter. The order of the three treatments was pseudorandomly chosen for each individual and was balanced across subjects, as far as possible. In our statistical analyses, we included the minute before ('Pre'), during ('Object') and after ('Post') the presentation of the object. Owing to illness, one subject could not participate during the last test day (Test 3). Additionally, for technical reasons, six entire test sessions and the Post minute from two sessions were excluded from the analyses.

Behavioural Analyses
For behavioural analyses from video, we used The Observer (The Observer XT 12, Noldus Information Technology bv, Wagenigen, The Netherlands). Based on the behavioural observations, the following parameters were analysed: first, parameters concerning behaviours that occurred during the whole test, which were duration of locomotion and of exploration of the arena, as well as number of vocalizations; second, parameters concerning behaviours that exclusively occurred during the object presentation, which were duration, frequency and latency of touching the object and latency to vocalize after the introduction of the object. If a subject did not vocalize or touch the object during its presentation, the latency was set at 60 s. Since each video observation represented one subject during one test, we obtained 269 video observations (see Results) which were analysed by two different observers (blind for the conditioning group). Before the beginning of the analyses, an interobserver reliability analysis was performed: three observations were analysed by both observers as well as a third observer and were compared. All combined comparisons (observer 1eobserver 2; observer 1eobserver 3; observer 2eobserver 3) resulted in a kappa of 0.83, which indicates a very good agreement between the three observers.

Heart Rate Measurements
When a measurement was started, the heart rate monitor made a sound. Later, in the analysis, this sound was used to synchronize the heart rate measurements with the video and acoustic recordings. Using the Polar system, we measured the interbeat (ReR) intervals. For the corrections, we used the same methods described in Leliveld, Düpjan, Tuchscherer, and Puppe (2016). In short, using 1 min intervals, data were corrected for artefacts (Software: Polar Precision Performance SW, version 4.03.040; settings: very low sensitivity, peak detection on, minimal protection zone of 20) or excluded if artefacts comprised more than 10% of the data or gaps of more than 3 s occurred. After corrections, sections with a linear development for five or more consecutive ReR intervals (indicating that a reliable correction was not possible) were excluded as well. Owing to these artefacts in the HR measurements, 3873 (from 6480) 10 s intervals were excluded from the HR analyses. The mean heart rate and heart rate variability parameters were calculated in 10 s intervals. The heart rate variability parameters that were derived in the time domain were the standard deviation of the interbeat intervals (SDNN, which indicates both sympathetic and parasympathetic activity), the root mean square of successive differences between interbeat intervals (RMSSD, which indicates parasympathetic activity) and the ratio of these two (RMSSD:SDNN, which reflects the balance of the autonomic nervous system). Means of the available 10 s values were then calculated for the Pre, Object and Post minutes.

Statistical Analyses
For statistical analyses, we used the SAS System for Windows, version 9.4 (SAS Institute Inc., Cary, NC, U.S.A.). We tested the effects of conditioning (positive or negative), treatment (BIN, LEC or REC) and minute (Pre, Object or Post) and their interactions. The behavioural parameters and heart rate variability parameters were analysed by repeated measurement analyses of variance (ANOVA) with the MIXED procedure of SAS/STAT software using a model with replicate, conditioning, treatment, minute and the interaction of conditioning*treatment*minute as fixed factors. Additionally, the duration of locomotion was included as a covariate for the heart rate parameters. Object-related parameters were analysed for the minute of the object presentation only (PROC MIXED for durations and PROC GLIMMIX for the frequency for touching the object), with replicate, conditioning, treatment and the interaction of conditioning*treatment as fixed factors. Test order and sow (effect of kin) were set as random effects. Repeated measures on the same animal were taken into account by using the REPEATED statement in the MIXED procedure or the RANDOM _residual_ statement of the GLIMMIX procedure. Pairwise comparisons were made with TukeyeKramer tests (i.e. adjusting P values to correct for multiple testing) and by using the SLICE option for performing a partitioned analysis of the least square means for an interaction that permits the selection of only the relevant comparisons (e.g. within a conditioning group, within the same treatment or within the same minute). Effects and differences were considered significant if P < 0.05.

RESULTS
The sample size consisted of 90 animals (N) with varying repeated measurements (N i ) according to the type of parameters (N i ¼ 3 for object-related parameters e 3 test sessions Â 1 minute; N i ¼ 9 for the other behavioural parameters and for the HR variability parameters e 3 test sessions Â 3 minutes). This resulted in different total numbers of observations (N Obs ) according to the type of parameters: N Obs ¼ 263 for the object-related parameters, N Obs ¼ 787 for the other behavioural parameters and N Obs ¼ 372 for the heart rate variability parameters. Comparisons between conditioning groups are presented in the Appendix.

Effect of Eye Covering after Positive Conditioning
Concerning the object-related parameters within the positively conditioned group (Fig. 3), REC led to a longer latency to touch the object (t 169 ¼ À2.96, P Tukey ¼ 0.010) and to vocalize after its introduction (t 118 ¼ À3.24, P Tukey ¼ 0.004) than BIN.
The interactions between minute and treatment within the positively conditioned group for all parameters are shown in  Figure 2. Summary of the treatments with an overview of the visual contralateral input to each hemisphere. When the left eye was covered with a patch (LEC) the contralateral input to the right hemisphere (RH) was reduced while the input to the left hemisphere (LH) was unaffected. The opposite was true when the right eye was covered (REC). In the control treatment (BIN, binocular viewing) the patch was fixed on the subject's forehead and the input to both hemispheres was unaffected.

Effect of Eye Covering after Negative Conditioning
Within the negatively conditioned group, we found no significant differences between the treatments among the object-related parameters (Fig. 4). The interactions between minute and treatment within the negatively conditioned group for all parameters are shown in Table 2. During the Pre minute, REC led to the lowest RMSSD compared to the other treatments (BINeREC: t 190 ¼ 6.17, P Tukey < 0.001; RECeLEC: t 349 ¼ À3.76, P Tukey < 0.001), and LEC led to significantly more vocalizations (t 205 ¼ À2.94, P Tukey ¼ 0.010) than BIN. The latter effect was also present during the Object (t 231 ¼ À3.24, P Tukey ¼ 0.004) and Post (t 297 ¼ À3.04, P Tukey ¼ 0.008) minutes.
Within BIN, the number of vocalizations increased significantly (t 236 ¼ 2.92, P Tukey ¼ 0.011) from the Pre to the Post minute. Within REC, the number of vocalizations also increased significantly from the Pre to the Post minute (t 235 ¼ 2.44, P Tukey ¼ 0.041), and the duration spent exploring the arena decreased significantly (t 243 ¼ 2.74, P Tukey ¼ 0.018) from the Pre to the Object minute. Within LEC, the subjects significantly increased their locomotion (t 250 ¼ À2.41, P Tukey ¼ 0.045) from the Pre to the Object minute. Additionally, the time spent exploring decreased significantly from the Pre to the Object minute (t 240 ¼ 2.68, P Tukey ¼ 0.022), as well as from the Pre to the Post minute (t 198 ¼ À2.68, P Tukey ¼ 0.022).

DISCUSSION
Our results show that within each conditioning group, the subjects differed in their behavioural and physiological reactions  Least square means are shown ± SEs. Minute: before (Pre), during (Object) and after (Post) the object was presented. Treatments: either the right (REC) or left (LEC) eye covered, or binocular (BIN) viewing. RMSSD: root mean square of successive differences between interbeat intervals; SDNN: standard deviation of the interbeat intervals. The lowercase letters indicate differences between the treatments within a minute. The uppercase letters indicate differences between the minutes within a treatment. The significant differences (P Tukey < 0.05) are in bold.
during the tests depending on the monocular treatment, which indicates that monocular viewing influences emotional reactions in pigs: covering the R eye (REC; leading to reduced input to the L hemisphere) influenced the positively conditioned subjects more than the other treatments (seeing with both eyes or with the L eye covered), while covering the L eye (LEC; leading to reduced input to the R hemisphere) influenced the negatively conditioned subjects more than the other treatments (seeing with both eyes or with the R eye covered). Within the binocular (BIN; control) treatment, the positively conditioned subjects showed a shorter latency and longer duration of touching the object and touched the object more frequently than the negatively conditioned subjects (see Appendix Fig. A1). This validates our conditioning paradigm as has already been applied in previous studies Leliveld et al., 2016). REC influenced the reaction of the positively conditioned subjects compared to BIN. Indeed, REC led to a longer latency to touch the object in comparison to BIN which suggests that the object was perceived as less rewarding (De Boyer Des Roches, Richardris, Henry, Ezzaouia, & Hausberger, 2008). Additionally, REC led to a longer latency to vocalize compared to BIN. An interruption in vocalizations has been suggested to reflect a heightened state of attention (Düpjan et al., 2011). Taken together, these longer latencies to vocalize and to touch the object more resembled the responses after negative conditioning than after positive conditioning in the control treatment (see Appendix Fig. A1), which may therefore indicate a less positive reaction. Additionally, REC induced the shortest duration of exploring the arena during object presentation. This was due to a significant decrease in exploration compared to the Pre minute, which was stronger than for BIN. Reduced exploration in pigs may indicate either decreased arousal or increased anxiety (Donald, Healy, Lawrence, & Rutherford, 2011). However, this behavioural response, combined with the increased latencies to touch the object and vocalize, could also reflect an increased attention directed to the object. Concerning heart rate variability, REC systematically led to the highest values in RMSSD compared to the other treatments, reflecting a higher vagal activity of the autonomic nervous system (von Borell et al., 2007). This effect, in the absence of changes in heart rate, can be assumed to be accompanied by a mild sympathetic activation and has been  Least square means are shown ± SEs. Minute: before (Pre), during (Object) and after (Post) the object was presented. Treatments: either the right (REC) or left (LEC) eye covered, or binocular (BIN) viewing. RMSSD: root mean square of successive differences between interbeat intervals; SDNN: standard deviation of the interbeat intervals. The lowercase letters indicate differences between the treatments within a minute. The uppercase letters indicate differences between the minutes within a treatment. The significant differences (P Tukey < 0.05) are in bold.
interpreted as a general orienting response (D esir e, Veissier, Despr es, & Boissy, 2004). Interestingly, REC led to an increase in RMSSD over time, from the Pre minute to the Post minute, without any changes in the heart rate, while the opposite trend occurred for BIN. However, since the same was found for LEC this indicates an effect of monocular testing. It may reflect a general orienting response (D esir e et al., 2004) caused by disorientation in the arena resulting from impaired depth perception (Hughes, 1977). REC showed fewer effects on the negatively conditioned subjects. It led to a significant decrease in exploration during the object presentation. Since LEC had the same effect this might reflect an increased anxiety (Donald et al., 2011) due to impaired depth perception. Apart from that, we only found that in the Pre minute, the RMSSD was lower for REC compared to BIN. This may indicate that REC influenced the reaction to the general situation of social isolation or to the anticipation of the stimulus. However, this result is difficult to interpret because it was not accompanied by any significant effects on the behaviour.
Altogether, the findings of REC were more prevalent for the positively conditioned than for the negatively conditioned subjects and showed that reducing the input to the L hemisphere when seeing a positive stimulus entailed an increased state of attention (based on behaviour) during the object presentation, which was accompanied by a general orientation response (based on physiology). This suggests a reduced positive appraisal which would be in accordance with the emotional valence hypothesis and with previous findings showing R eye preferences for observing positive stimuli (e.g. De Latude, Demange, Bec, & Blois-Heulin, 2009;Racca, Guo, Meints, & Mills, 2012;Rogers, Ward, & Stafford, 1994). However, most of these previous studies have shown this mainly in response to seeing food rewards or during the act of feeding, while the results in other putatively positive contexts are less clear (see Leliveld et al., 2013). Therefore, our findings in a standardized positive emotional context (achieved through conditioning) suggest that the processing of positive emotions, rather than other processes involved in responses to food, indeed predominantly take place in the L hemisphere. Note that the behavioural reaction alone would also support the approachewithdrawal hypothesis (Davidson, 1992;Rogers & Vallortigara, 2019), since a reduced input to the L hemisphere attenuated the approach behaviour directed to the object. However, combined with the physiological reaction, our results suggest that the response is an affective, rather than a purely behavioural, response and reflects a reduced positive appraisal of the object. Monocular testing permitted us to study the role of the L hemisphere during the appraisal of a positive stimulus. Indeed, these results suggest that the L hemisphere may be important for quick recognition and evaluation of positive stimuli. Although REC first and foremost can be assumed to lead to reduced processing by the L hemisphere, we cannot fully exclude the possibility that this could also lead to increased activation of the R hemisphere as compensation. However, the investigation of this issue (by using brain imaging in human research) is still at an early stage (Wang et al., 2018). Rather, we assume that the observed reaction is more likely to be the result of reduced L hemisphere processing than of increased R hemisphere processing. These findings shed new light on positive emotions, which are still considered difficult to assess since authors often emphasize the lack of measurable reactions (e.g. Bellegarde et al., 2017;Smith, Proops, Grounds, Wathan, & McComb, 2016). Note, however, that the significant differences between the treatments in behavioural and physiological reactions in this study were numerically small compared to other studies that investigated emotional appraisal (e.g. D esir e et Krause et al., 2017). However, the differences found within the conditioning group were expected to be subtle, since these differences were supposed to depend not on the nature of the stimulus but on how the stimulus was perceived (Pourtois, Schettino, & Vuilleumier, 2013).
LEC (reduced input to the R hemisphere) after positive conditioning did not lead to significant differences from BIN (control) during the object presentation. Instead, LEC influenced some parameters during the Pre minute: the RMSSD was the lowest compared to the other treatments, which may reflect, as mentioned before for the REC treatment, a reaction to social isolation or to the anticipation of the stimulus. Additionally, the duration of locomotion increased from the Pre minute to the Post minute, suggesting increased arousal and/or avoidance-related emotions during the test (Murphy, Nordquist, & van der Staay, 2014). However, without further results, it is difficult to make a clear conclusion about the effect of reducing the input to the R hemisphere after positive conditioning.
In contrast, LEC had more effects on the reaction of the negatively conditioned subjects: it led to an increased duration of locomotion during the presentation of the object compared to the minute before, also suggesting increased arousal and/or avoidancerelated emotions (Murphy et al., 2014). However, while BIN and REC led to a significant increase in the number of vocalizations during the test, which probably indicates an increase in arousal and/or anxiety (Manteuffel, Puppe, & Sch€ on, 2004;Murphy et al., 2014), LEC systematically led to more vocalizations than BIN, indicating higher anxiety and/or arousal independent of the object presentation. The results do not suggest an attenuation of the negative perception of the object when the input to the R hemisphere is reduced. This does not indicate a specialization of the R hemisphere in processing negative emotions as predicted by the emotional valence hypothesis. Therefore, our findings seem to contradict this hypothesis. However, without further evidence it is difficult to draw a clear conclusion. Our results thereby are not in line with previous findings of a R hemisphere specialization for processing intense negative emotions (Hook-Costigan & Rogers, 1998;Siniscalchi, Quaranta, & Rogers, 2008) and cannot provide clear support for the emotional valence hypothesis with regard to negative emotions.
Overall, covering the R eye caused more effects in the positively conditioned subjects than covering the L eye in the negatively conditioned subjects. This may be explained either by the fact that domestic pigs show a stronger lateralized processing of positive emotions or that the negative emotions were not as strong as the positive emotions. Because domestic pigs show particular resilience to acute stressors (Foury et al., 2007;Sutherland, Niekamp, Rodriguez-Zas, & Salak-Johnson, 2006), the latter explanation seems to be more probable: the negative reinforcer may not have been equivalent to the positive reinforcer. Although the punishment used was found to be reliable (Douglas, Bateson, Walsh, B edu e, & Edwards, 2012;Düpjan et al., 2017), it was probably not equivalent to life-threatening stimuli used in other laterality studies (Koboroff, Kaplan, & Rogers, 2008;Siniscalchi, Sasso, Pepe, Vallortigara, & Quaranta, 2010). Another explanation could be that the negative conditioning was less effective because the appearance of the object may have had little additional effect in an already somewhat negative situation caused by inevitable testing in isolation. Since laterality is suggested to improve brain efficiency (Güntürkün & Ocklenburg, 2017;Rogers, Zucca, & Vallortigara, 2004;Vallortigara & Rogers, 2005), it is possible that the more subtle emotional reactions of the negatively conditioned subjects were less in need of brain efficiency than the more intense emotional reactions of the positively conditioned subjects.

Conclusions
In this study, we have provided evidence of differential involvement of the two hemispheres in the visual processing of objects of opposing emotional valence. Our findings provide partial support for the emotional valence hypothesis because reducing the input to the left hemisphere seemed to attenuate the positive appraisal of a positively conditioned object. Monocular testing permitted us to provide new insight by suggesting that the left hemisphere plays a crucial role in the quick recognition and evaluation of positive stimuli. These findings demonstrate that investigating the lateralized processing of emotions can provide insight into the mechanisms of positive appraisal in animals. In contrast to positive conditioning, the results concerning negative conditioning were less clear, which was probably due to the mild nature of the stimulus. The results of this study have shown the importance of the left hemisphere in the initial recognition of positive stimuli. Therefore, an appropriate next step would be to investigate the role of each hemisphere in the long-term moods of animals to improve their welfare.