Modulating factors of emotional contagion in rats. A behavioral study on the effects of familiarity and repeated testing on the socially triggered freezing response.

(1)

Modulating factors of

emotional contagion in

rats.

A behavioral study on the effects of familiarity and repeated

testing on the socially triggered freezing response.

08

Fall

Mirjam Heinemans

10326111

Master Brain and Cognitive Sciences

University of Amsterdam (UvA)

Supervisor: dr. Maria Carrillo

Netherlands Institute for Neuroscience (NIN)

Social Brain Lab

(2)

2

1. Abstract

Multiple experimental studies provide evidence for the existence of emotional contagion in the social life of rodents. Based on this evidence rodent models for emotional contagion are being developed in both mice and rats. In this process multiple factors that modulate emotional contagion responses of these species have been identified, such as context, genetic make-‐up and familiarity. Studies with mice reported familiarity between animals to be a crucial factor for emotional contagion expression. In the present study the main aim was to investigate whether is also true for rats. This was done by using a socially triggered freezing paradigm, in which one rat (the observer) witnesses a conspecific freeze when receiving painful foot shocks (the demonstrator). The foot shocks trigger the

demonstrator to freeze, which in turn elicits socially induced freezing in the observer. The

level of familiarity between observers and demonstrators differed, ranging from being unfamiliar with each other to spending 13-‐weeks together as cage mates. The results showed that observer animals paired with an unfamiliar demonstrator exhibited a socially triggered freezing response during the shock period similar to observers paired with familiar

demonstrators. Thus, rats do not have to be familiar with one another in order to experience

emotional contagion. Interestingly, animals in the unfamiliar condition froze more during the pre-‐shock period compared to animals in the familiar conditions, which could indicate elevated stress levels in these animals. A second aim of the present study was to discover whether yawning in rats could be related to elevated stress levels. In a previous study performed in this lab, the effect of repeated testing was examined. In this experiment an emergence of yawning was observed. In the present study it was assessed whether this emergence of yawning upon repeated witnessing of a conspecific in distress was related to elevated stress levels. This was done by testing the rats on six consecutive days, while administering the stress-‐reducing drug Metyrapone on either day 5 or 6 and comparing the yawning frequencies on these days. Comparison of the behavior with and without Metyrapone administration showed a significant effect of Metyrapone on the number of yawns, thereby supporting the existence of a relationship between stress and yawning in repeatedly tested rats. In short, we found that familiarity between animals is not required for emotional contagion in rats. Furthermore, repeated testing changes the socially triggered freezing response while inducing yawning in observers, probably as a sign of heightened stress-‐levels.

(4)

4

2. Introduction

The concept of empathy

Empathy – the ability to share and understand the emotional state of others – is generally considered to be the product of a multi-‐level process that can be decomposed into several underlying behavioral, affective and cognitive component processes (Decety, 2011; Decety & Jackson, 2004; J. B. Panksepp & Lahvis, 2011; Preston & Waal, 2002). The most basic process underlying empathy is emotional contagion. Emotional contagion is a process in which the perception of an emotional state in another individual (the demonstrator) automatically activates the same emotional state in the observing individuals (observers), without distinguishing the origin of the emotion (Hatfield, Rapson, & Le, 2009; Singer & Lamm, 2009). A second, higher-‐level process underlying empathy is emotional empathy. This process is similar to emotional contagion in the sense that both individuals share the same affective state, with the crucial difference that an individual possesses the ability to distinguish itself from the other. In other words, the individual is able to discriminate whether his present emotional state originates from his own experience or through emotional contagion from another individual (J. B. Panksepp & Lahvis, 2011; Preston & Waal, 2002). The highest level of empathy is cognitive empathy, which requires the understanding that one’s own experience can be different from another one’s experience (Preston & Waal, 2002). This ability to distinguish between perspectives – better known as the phenomenon ‘Theory of Mind’ – enables individuals to infer the emotional state of another without necessarily sharing this same emotional state (J. B. Panksepp & Lahvis, 2011; Preston & Waal, 2002).

Perception-‐Action Model of empathy

A useful model proposed by Preston and De Waal (2002) to explain the neural mechanisms underlying the processes of empathy is the Perception-‐Action Model (PAM) of empathy. The PAM of empathy states that activation of specific brain areas upon both experiencing an emotion and observing another experiencing

(5)

an emotion enables individuals to understand and empathize with the emotional state of others. This model is based on the presumed existence of a mirror neuron system for emotion similar to the mirror neuron system for action.

Mirror neurons for action were originally discovered in the pre-‐motor area of macaques by Di Pellegrino and colleagues (Di Pellegrino, Fadiga, Fogassi, Gallese, & Rizollati, 1992). They discovered neurons that became active both during execution and observation of a goal-‐directed movement. The experimenters hypothesized that the activation of the neurons during observation of movements plays an important role in perception of other individuals’ actions (Di Pellegrino et al., 1992). After this discovery, many brain-‐imaging studies using PET, fMRI and EEG have been performed on both humans and primates to investigate which regions are involved in the mirror neuron system for action perception (e.g. Iacoboni, 2005; Molenberghs, Cunnington, & Mattingley, 2009; Oberman et al., 2005; Parsons et al., 1995).

The PAM of empathy proposes a similar mechanism for emotional contagion as for action perception: mirror neurons in specific brain regions are thought to become activated both upon experiencing an emotion and witnessing that same emotion (Preston & Waal, 2002). Activation of the same neural circuits while experiencing or observing an emotion would enable the understanding of the perceived emotion of another individual and at the same time it would enable us to empathize with that individual (Bastiaansen, Thioux, & Keysers, 2009; Gallese, Keysers, & Rizzolatti, 2004). There is indeed evidence suggesting that mirror-‐like mechanisms play a key-‐role in the perception of emotion in others. For instance, several imaging studies investigating disgust have shown increased activation in the anterior insular cortex during both observation and experience of disgust (Phillips et al., 1997; Small et al., 2003; Wicker et al., 2003). In addition, multiple studies on empathy for pain show involvement of the anterior insula and the anterior cingulate cortex during observation and experience of pain (for a review, see Lamm, Decety, & Singer, 2011).

However, although imaging studies using human subjects support the involvement of certain brain regions in both observation and experience of emotions, the methods in these studies are not accurate enough to provide direct

(6)

6 evidence for the existence of individual mirror neurons for emotion. To investigate whether the observed activation in these areas could indeed be the result of activation of single mirror neurons for emotion, research at a cellular level is essential.

Animal models of empathy

Augmenting our understanding of empathy at a cellular level requires invasive techniques such as single cell recordings and experimental manipulation of the involved regions. Due to ethical and practical reasons, it is extremely difficult to conduct these experiments on human subjects; thus animal models such as rodent models with rats or mice are required. This raises the question whether rodents would be suited for such a model for empathy. In order to develop a proper animal model for human empathy, first of all the animals have to have a similar capacity to experience emotions or ‘affective states’ themselves. As reviewed by J. Panksepp (2011), there is indeed strong empirical evidence for the existence of at least seven types of emotional arousal in animals. Furthermore, the brain areas involved in the expression of these emotions show a strong homology in all tested vertebrates, including humans (J. Panksepp, 2011). A second prerequisite for a proper animal model for empathy is that animals have the ability to experience some level of empathy. Although the higher levels of empathy are currently considered to occur exclusively in humans, an increasing body of evidence suggests that at least emotional contagion, a level of empathy that does not require self-‐other distinction, indeed occurs in animals other than humans (J. B. Panksepp & Lahvis, 2011). Several behavioral studies in both mice and rats, looking at social modulation, social priming and social buffering provide evidence that these animals do also express emotional contagion for pain and fear (Church, 1959; Gonzalez-‐Liencres, Juckel, Tas, Friebe, & Brüne, 2014; Guzmán et al., 2009; Jeon et al., 2010; Knapska, Mikosz, Werka, & Maren, 2010; Langford et al., 2006). The foundational experiments for emotional contagion research in rats were conducted almost sixty years ago by different researchers: Church and Rice (Church, 1959; Rice & Gainer, 1962). He showed that food deprived rats which were well-‐trained on a lever press task for food would

(7)

significantly reduce the pressing rate when they were concurrently exposed to a conspecific getting shocks (Church, 1959). His study showed two particularly interesting findings. First of all, it showed that perception of social stress would disrupt the behavior of the food-‐deprived animals, even though they would obtain food by expressing the behavior. Secondly, Church included an ‘emotional conditioning component’ in the experiment, during which the rat and its social partner would receive shocks simultaneously in presence of a conditioned stimulus (CS). When these animals were subjected to the lever press task again, they showed a suppression of lever presses that persisted for ten days. This effect was significantly more robust than the suppression of lever pressing of rats in the control groups, which had received shocks at different time points than their social partner, or were not conditioned at all. These results indicate that shared experience of pain is a strong modulator of behavior and has a bigger effect than consecutive experience of pain and perception of pain in others. In other words, rats can identify whether their experience of pain is temporally coordinated with a similar experience in a conspecific (Church, 1959). In a separate study, Rice and Gainer (1962) investigated whether rats could exhibit helping behavior to terminate the distress of a conspecific. Again, rats were trained to press a lever, this time to avoid a receiving a shock that was predicted by a visual cue. Afterwards the animals were tested in presence of a conspecific signaling distress cues in response to being lifted up by a hoist system. These cues were audible (squealing) and visible (wriggling) for the rat that had to press a lever. In this part of the experiment the lever pressing resulted in lowering the animal down and alleviating its distress. This experiment showed that the animals would press up to 10 times more to lower a conspecific down and alleviate stress when compared to a control group which would lower a Styrofoam block, thereby showing that the rats would actively work to reduce the distress of a conspecific (in other words to help the conspecific), which is also a phenomenon relevant to empathy (de Waal, 2008). More recently, Langford and colleagues (2006) continued the research on emotional contagion in rodents by looking at social modulation for pain. In this study they showed that pain expression in mice is modulated by witnessing pain in other mice. The researchers placed two same-‐sex mice in two cylinders facing each other and subsequently induced pain in either one

(8)

8 or both mice by administration of 0.9% acetic acid in the abdomen and subsequently evaluated the intensity of ‘writhing behavior’ when only one animal was in pain or when they were co-‐experiencing pain. They found that mice that were experiencing pain simultaneously showed a significantly higher amount of writhing behavior in comparison to mice that experienced pain individually while watching an animal that was not in pain. These findings support the presence of emotional contagion in mice, since the emotional state of one individual directly influenced the emotional state of the other (Langford et al., 2006).

Social influence on the emotional state of an animal has also been observed in several studies on social learning. These studies revealed that a brief exposure with a demonstrator animal modulates the performance of an animal in a subsequent learning paradigm. For example, Bredy & Barad (2009) had unexpected results in their fear conditioning experiment with mice. They let observer mice interact with either a fear-‐naïve demonstrator animal or a demonstrator animal that had just undergone a fear conditioning session. After this interaction the observer animals would be subjected to the same fear conditioning procedure. Whether the

demonstrator mouse was fearful or not had significant influence on the reaction of

the observer mouse on the conditioning: after interaction with a fearful mouse, the

observer would show a reduced acquisition of a fear response (i.e. freezing) in

comparison with observers that interacted with non-‐fearful animals, as well as a diminished ability to recall the CS-‐UCS association, and an increased extinction of the memory. These results came as a surprise, since the researchers had hypothesized that the effect would be the other way around (i.e. interaction with fearful animals would result in a facilitation of fear conditioning memory) (Bredy & Barad, 2009). Another study by Knapska and colleagues (2010), showed an opposing effect of social learning in rats: their results revealed that interaction between a fear conditioned rat and a non-‐fear conditioned rat would facilitate subsequent fear conditioning in the latter: they found that rats that had been in contact with the fear conditioned individuals right before learning a shock induced avoidance task were faster in learning the task. Furthermore, rats that interacted with conditioned conspecifics before being submitted to a fear conditioning session showed significantly increased conditioned freezing the next day. Interestingly, this type of

(9)

social modulation of fear appears to be bidirectional; non-‐fearful animals can also transfer their emotional state to observer animals. This is illustrated by a different experiment looking at social buffering in mice, in which it was shown that fear can be reduced by contact with non-‐fearful conspecifics (Guzmán et al., 2009). In this study the experimenters pre-‐exposed observers mice to either fearful mice (that had received shocks in that context before), or non-‐fearful mice (that had been habituated to the context without receiving any shocks), upon which they fear conditioned the observer animals in the same context. They found that the mice that had observed a non-‐fearful animal froze significantly less during the conditioning phase when compared to all other groups.

The abovementioned paradigms measure the amount of social modulation by looking at the amount of freezing after interaction with either a fearful or non-‐ fearful conspecific. Expression of freezing in both mice and rats (i.e. complete absence of movement of the body, except through breathing) is a direct behavioral response to fearful and stressful stimuli and therefore an accurate read-‐out in those experiments (LeDoux, 2000). In the previous social modulation studies, freezing was measured when the observer animals themselves underwent fear conditioning. However, it has been shown that freezing can also be triggered purely through emotional contagion upon observing another individual in pain or distress (Atsak et al., 2011; Church, 1959; Jeon et al., 2010). In these paradigms freezing was successfully induced by letting an observer rat witness the freezing response of a

demonstrator animal undergoing mild foot shocks. This reaction of the observing

animal to the situation of the demonstrator animal has recently been referred to as socially triggered freezing, since it is purely elicited by the observation of a conspecific in distress (Carrillo et al., 2015).

The effect of familiarity and repeated exposure on emotional contagion

The expression of socially triggered freezing of an observer animal witnessing a conspecific getting foot shocks is modulated by multiple factors, such as genetic background (Chen, Panksepp, & Lahvis, 2009), (social) context (Jeon et al., 2010), previous experience with foot shocks (Atsak et al., 2011) and repeated exposure to

(10)

10

demonstrator animals experiencing painful shocks (Carrillo et al., 2015). In the

present study we focused on two of such modulatory factors: the degree of familiarity between the two individuals (Gonzalez-‐Liencres et al., 2014) and repeated observation of a conspecific receiving foot shocks (Carrillo et al., 2015).

Familiarity between the demonstrator animal and the observer animal has been shown to be a modulatory factor for expression of emotional contagion in mice (Gonzalez-‐Liencres et al., 2014; Jeon et al., 2010; Langford et al., 2006; Martin et al., 2015). In the study performed by Jeon et al (2010), the amount of weeks an observer mouse was housed with the mouse undergoing the aversive stimulus had a significant increasing effect on the socially triggered freezing response (Jeon et al., 2010). A second study by Gonzalez-‐Liencres et al. (2014) showed concurrent results. In this study observers witnessed a cage mate or a non-‐cage mate during a pre-‐shock period and a shock period, during which the demonstrator received multiple shocks.

Observers that watched a cage mate getting shocked froze significantly more during

the shock period when compared to the observers that witnessed non-‐cage mates undergoing shocks. The latter group did not show any significant increase of freezing from pre-‐shock period to shock period, while the observers in the ‘cage mate’ condition did show a significant increase in freezing while watching the

demonstrator getting shocks when compared to the pre-‐shock period (Gonzalez-‐

Liencres et al., 2014). These findings are consistent with findings in the abovementioned pain expression modulation experiment performed by Langford et al (2006). In this experiment the familiarity between the animal pairs had a significant increasing effect on the level of pain-‐related writhing behavior. The animals experienced pain by abdominally injected acetic acid in one of three conditions: this would either happen in dyads where one or both mice were injected with acid, or in mice that were isolated. When mice experienced pain simultaneously with a familiar mouse, they increased writhing compared to being tested in isolation. There was no difference between writhing behavior in isolation and in presence of a familiar mouse that did not experience pain simultaneously. In contrast, mice that experienced pain together with an unfamiliar conspecific would not increase writhing compared to isolated mice, while writhing behavior in presence of an unfamiliar mouse without pain decreased compared to writhing of isolated mice.

(11)

This indicates that the presence of an unfamiliar conspecific has an analgesic effect, while a familiar conspecific can increase the level of experienced pain (Langford et al., 2006). These findings all show a clear effect of familiarity on emotional contagion in mice. In rats, however, results on the influence of familiarity on social modulation and emotional contagion are not conclusive. In one experiment on social buffering through olfactory cues, the experimenters found a significantly higher amount of social buffering (i.e. less freezing in the observer animals) if the olfactory information belonged to a familiar conspecific as opposed to an unfamiliar conspecific (Kiyokawa, Honda, Takeuchi, & Mori, 2014). Nevertheless, both conditions resulted in decreased freezing when compared to a control condition in which no olfactory cues were presented. In contrast, in the study performed by Knapska et al. (2010) social learning from either a familiar or unfamiliar conspecific resulted in a similar level of learned fear, indicating that in this study familiarity did not have a major influence on social learning. The abovementioned observations are from studies looking at social priming and buffering; to date there is nothing known about the role of familiarity on socially triggered freezing in rats. Therefore, with this present study, the first question we aim to answer is whether the degree of familiarity between rats modulates their display of socially triggered freezing, by using the same paradigm as Atsak et al. (2011). The animals in our present experiment are either unfamiliar with each other (i.e. the animals first interact with each other during test day), or they spent 1, 3, 5 or 13-‐weeks as cage mates before testing. These different degrees of familiarity were chosen based upon the results of socially triggered freezing response in mice (Jeon et al., 2010).

The second factor we look into is repeated exposure of the observer animal to a conspecific undergoing shocks. This factor is particularly important from a methodological point of view, since many experiments require repeated testing in order to obtain optimal results. It is therefore important to know how repeated testing modulates expression of emotional contagion, since accurate behavioral read-‐outs are needed to measure the amount of emotional contagion during multiple-‐consecutive tests. In a previous experiment from our group, originally focusing exclusively on the effect of repeated exposure on socially triggered freezing, it was shown that repeated exposure to a conspecific undergoing foot shocks

(12)

12 resulted in a statistically significant reduction in freezing already from the third test onwards (Carrillo et al., 2015). This experiment reports the emerging of another – unexpected – behavior upon repeated testing: yawning. Yawning is a relatively infrequent behavior that under normal circumstances is related to the circadian rhythm and expressed when animals just woke up or are about to fall asleep (Anias, Holmgren, Holmgren, & Eguibar, 1984; Provine, Hamernik, & Curchack, 2010). However, yawning also appears to be expressed under other situations and its role under all these circumstances is still poorly understood. One situation relevant to this study is the increase of yawning under highly stressful circumstances. Several studies indicate some relation between yawning and elevated stress levels in animals, but the exact role of yawning in this situation is unclear at present (Gallup & Gallup, 2008; Kubota, Amemiya, Yanagita, Nishijima, & Kita, 2014; Major et al., 2009; Tufik et al., 1995). For instance, Kubota and colleagues (2014) found that emotional stress evoked by classical fear conditioning resulted in a significant induction of yawning frequency in rats. Furthermore, in an experiment with rhesus monkeys, an increase in yawning was reported upon administration of a stress-‐inducing anxiogenic (Major et al., 2009). Therefore, the observations of Carrillo et al (2015) of both the emergence of yawning in the observers, as well as possibly the gradual reduction in freezing of observers across test days could be due to chronic high stress levels in the observers (caused by repeated testing). How increased stress levels could affect an emotional contagion response is illustrated by an experiment of Martin and his colleagues (2015), involving both humans and mice. The results of this study showed that high levels of stress reduced the expression of emotional contagion. Interestingly, the effect could be reversed by blocking glucocorticoid synthesis with a drug called Metyrapone (2-‐metyl-‐1,2-‐di-‐3-‐pyridyl-‐1-‐propane). Glucocorticoids are produced by the adrenal glands as a result of activation of the hypothalamic-‐pituitary-‐adrenal-‐axis (HPA-‐axis). The hypothalamus releases corticotropine-‐releasing hormone (CRH) in reaction to a stressor, which subsequently releases adrenocorticotropic hormone (ACTH) from the pituitary gland. ACTH promotes glucocorticoid production in the adrenal glands, which induce stress responses in the animal. The HPA-‐axis is considered to be the major neuroendocrine system for stress regulation (Tsigos & Chrousos, 2002). The findings of Martin and

(13)

his colleagues therefore strongly suggest that expression of emotional contagion is modulated by the activity of the HPA-‐axis and the production of glucocorticoids.

To conclude, two questions are addressed in this present study: 1) what is the effect of familiarity on socially triggered freezing, and 2) are the observed reduction in socially triggered freezing and emergence of yawning as observed in the first experiment by Carrillo et al. (2015) indeed related to increased glucocorticoid production? The relation between yawning and glucocorticoid production was tested by scoring socially triggered freezing and yawning during six consecutive tests and administering a stress-‐reducing drug (Metyrapone) to the observers on either day 5 or day 6 (when yawning frequency appeared to have reached its maximum in the first experiment by Carrillo et al. (2015)) to see whether the drug reduces the yawning frequency and restores the freezing response (results of both yawning experiments are published by Carrillo et al., 2015).

(14)

14

3. Materials & Methods

Subjects

Upon arrival, 70 male Long-‐Evan rats (arrival weight 200-‐250 g) were maintained on a reversed light cycle (lights off at 07 AM, lights on at 7 PM) and kept in a temperature-‐controlled room (22-‐24 °C, 55% relative humidity, SPF, type III or type IV cages with sawdust). The rats were socially housed as 2-‐6 rats per cage, with ad libitum access to food and water. All experiments were conducted in strict accordance with the European Community’s Council Directive (86/609/EEC) and all experimental procedures were approved by The Institutional Animal Care and Use Committee of the Netherlands Institute for Neuroscience (IACUC-‐NIN-‐1493).

Experimental procedures

Test set-‐up – The testing apparatus consisted of a transparent Plexiglas cage

with metal grids on the floor (L: 48cm x W: 25cm x H: 34cm; Med Associates, Inc.). The cage was separated into two equal compartments by a transparent Plexiglas grid that allowed the animals to see, smell, hear and touch each other through the bars. A schematic illustration of the set-‐up is shown in figure 1. One of the compartments was connected to a stimulus scrambler (ENV-‐414S, Med Associates, Inc.) in order to deliver foot shocks. Behavior during pre-‐exposure was video recorded from ca. 50 cm above the floor grid (JVC HD Everio, JVC Kenwood corporation). For the emotional contagion test the behavior was videotaped from the side in order to optimize detection of yawning (Ethovision XT 9.0, Noldus Information Technology). A schematic timeline of the testing procedures is shown in figure 2.

Figure 1. Schematic illustration of the experimental test set-‐up.

The observer and demonstrator animals were placed in adjacent chambers, divided by a perforated Plexiglas separation. Both animals stand on stainless steel rods, through which the demonstrator received a series of foot shocks during the tests.

Adopted with permission from Carrillo et al. (2015)

(15)

Pairing – The animals arrived in groups of 5/6 animals per cage and they

were housed in their arrival groups until they were paired. To establish different degrees of familiarity, observer animals were randomly paired with unfamiliar

demonstrator animals from a different arrival group for 1, 3, 5 or 13-‐weeks prior to

testing. Animals in the unfamiliar condition were paired with an animal with the same role (i.e. observers where paired with observers and demonstrators were paired with demonstrators) one week prior to testing. A total of 35 pairs were distributed over the conditions as follows; the 3-‐week condition comprised 7 pairs, the 13-‐week condition 10 pairs, and the unfamiliar, 1-‐week and 5-‐week conditions comprised 6 pairs each.

Handling – After one week of acclimation the animals were handled twice

per week for 3 minutes, in order to let them get habituated to the experimenters. In the week preceding the first test day, they were handled for three minutes on three separate days.

Habituation – Animals were habituated to the testing environment for

twenty minutes on three consecutive days preceding the pre-‐exposure day. Transportation and placement of the animal pairs in the testing apparatus was done

Figure 2. Schematic timeline of the experiment

In this figure the timeline of the entire experiment is represented. Upon arrival (day 1) the animals acclimated for seven days, after which they were paired for 1, 3, 5 or 13-‐weeks. During this time all animals were handled twice a week for 3 minutes. In the week prior to habituation they were handled 3 times for 3 minutes. ‘Day 0: End familiarization’ represents the last day of the familiarization period, after which the habituations started (3 days, green stripes). On the following day, the observers were pre-‐exposed to the shock (day 4, black stripe). From day 5 to 10 the animals underwent the emotional contagion tests once a day.

(16)

16 in the exact same way as during actual testing, as shown in Fig. 1 (no shocks were delivered). Between every habituation session the cage was cleaned with soap and 70% alcohol to establish a specific odor-‐context and the testing room was illuminated with red dim light.

Shock pre-‐exposure – A previous study by our lab showed that rats express a

significantly higher socially triggered freezing reaction when they are familiarized with foot shocks when compared to when they never experienced a foot shock before (Atsak et al., 2011). Therefore, to optimize the socially triggered freezing response, observer animals were pre-‐exposed to the foot shocks on the day prior to the test. To ensure that context-‐fear would not interfere in any way with the socially triggered freezing reaction during test days, a distinct visual, auditory and olfactory context was used during pre-‐exposure. The animals were transported to the testing chamber in a clean and neutral cage and placed in the pre-‐exposure environment for a total of 31-‐39 minutes. Upon a 10-‐minute pre-‐shock period, 4 foot-‐shocks (1 second each, 0.8mA) with a random inter-‐shock interval ranging from 240-‐360 seconds were administered. After the fourth shock there was a 5-‐minute post-‐shock period, after which animals were placed back in a clean cage and transported to their room. After pre-‐exposure the animals were kept individually for at least 60 minutes before they were put back in their home cage, as to avoid any transfer of fear cues from the observer to the demonstrator.

Empathy test – Preceding each test, the testing apparatus was cleaned with a

dishwashing soap with neutral smell followed by 70% alcohol solution. Animals were transported to the testing environment in their home-‐cage and placed in the testing apparatus for a total of 24 minutes, starting with a 12-‐minute pre-‐shock period. After the pre-‐shock period, the demonstrator rat received 5 foot-‐shocks (each: 1 second, 1.5mA), separated by an interval of either 2 or 3 minutes. After the 5th_shock

there was a post-‐shock interval of 2 or 3 minutes, adding up to a total of 12-‐minutes shock period. Pairs were tested on 6 consecutive days; the testing order of the pairs was randomized every day and all tests were conducted between 8:30 and 11:30 – in the dark phase of the circadian cycle. On test day 5 and 6 (these days correspond

(17)

with day 9 and 10 in Fig. 2) the animals were randomly assigned to two groups. The first group underwent a pre-‐treatment with the glucocorticoid synthesis inhibitor Metyrapone (2-‐metyl-‐1,2-‐di-‐3-‐pyridyl-‐1-‐propanone; Sigma-‐Aldrich Co., Oakville, ON, Canada) on day 5 and were tested under normal testing conditions on day 6, while the second group was tested under normal conditions on day 5 and received the Metyrapone pre-‐treatment on day 6. All observers received a 25mg/kg subcutaneous injection (prepared by dilution in 9% NaCl solution) 30 minutes prior to testing.

Behavioral scoring – During the empathy tests, freezing behavior and

yawning frequency exhibited by the observer rat were manually scored during pre-‐ shock period, inter-‐shock intervals and post-‐shock period by two experienced researchers. The inter-‐rater reliability was assessed with a Pearson’s r-‐correlation test that showed a correlation of >90%.

Freezing was defined as absence of any movement except for breathing for a period of at least one second. Yawning was defined as a wide opening of the mouth, usually accompanied by stretching of the entire body (see figure 3 for an example).

Figure 3. Consecutive frames of an empathy test clip showing an observer yawning during a shock period.

(18)

18

Statistical analysis

All behavioral data was analyzed using the statistics program SPSS 22. The

demonstrator and observer freezing data was analyzed with two independent

repeated measures ANOVA’s. The factors in the demonstrator’s ANOVA were condition (5 levels: unfamiliar versus 1, 3, 5 or 13-‐weeks of familiarity; between groups factor), test day (4 levels: days 1-‐4; within-‐subject factor) and shock period (2 levels: pre-‐shock vs. shock; within-‐subject factor). The observer data was analyzed in two separate parts; day 1 to 4 was considered to be regular repeated test days, whereas day 5 and 6 were considered as two counterbalanced test days (Metyrapone versus control). A repeated measures ANOVA was used to assess the effect of the following factors on the socially triggered freezing response of the observers: (1) condition (5 levels, between), (2) test day (4 levels, within), and (3) shock period (2 levels, within). In addition, the increase in freezing per test (i.e pre-‐shock period freezing versus shock period freezing) was analyzed for demonstrators and observers separately with a repeated ANOVA per test-‐day and per condition, in order to assess whether there was any difference in alteration of freezing patterns over the test days in different familiarity conditions. Furthermore, freezing of the pre-‐shock periods was compared per condition over test days, as well as freezing during shock periods (for both demonstrators and observers separately) to further assess whether any observed changes in patterns are due to changes in freezing during pre-‐shock periods, shock periods or both periods. Finally, the correlation of freezing time between observers and their demonstrators per shock period was analyzed, to assess to what extent familiarity had an effect on concurrence of freezing.

For the observer animals, the effect of Metyrapone on freezing was analyzed using a separate repeated measures ANOVA with the following factors: test day (2 levels: test with metyrapone and control test; within factor), shock period (2 levels: pre-‐shock and shock; within factor) and condition (5 levels: between factor). In case of a statistical significance of any factor in one of the ANOVA’s, post hoc tests were performed on this factor to assess which differences were driving the significant result. Similarly, the freezing behavior of the observer animals during the

(19)

metyrapone and control test was compared with a repeated measures ANOVA (test day, 2 levels, within; shock period, 2 levels, within; condition, 5 levels, between), in order to assess any possible influence of metyrapone on the socially triggered freezing response of the observer rats.

Yawning was normalized to baseline by subtracting yawning frequency during pre-‐shock period from yawning frequency during total shock period. Normalized yawning frequencies from days 1 to 4 were analyzed with the Friedman rank sum test (repeated measures). Post hoc analyses were conducted using a Wilcoxon signed ranks test to compare day 1 with day 2, 3 and 4. To examine the effect of Metyrapone on yawning, a repeated measures Friedman test was performed on the Metyrapone versus control condition. A planned post hoc Wilcoxon test was conducted to determine the direction of any effect.

p-‐values <0.05 were considered statistically significant in all analyses.

(20)

20

4. Results

Day 1-‐4: Effects of repeated exposure and familiarity

Freezing

The average freezing percentages of all animals (demonstrators and observers)

on all test days are depicted in figure 4. The pre-‐shock period of 12 minutes can be compared to the five different between-‐shock periods (2 or 3 minutes) to have a detailed look of the freezing pattern over the total test period. The total freezing of the 12 minutes pre-‐shock period versus the total freezing of the 12 minutes shock period on test day 1-‐4 is shown in figures 5 (demonstrators) and 6 (observers). All the freezing results presented and discussed in this paper derive from analysis of the data as shown figures 5 and 6. As shown in these figures, in all conditions the freezing increased from pre-‐shock to shock period during the first test day for both demonstrators (Fig 5 a-‐e) and observers (Fig 6 a-‐e). After test day 1, the freezing of the demonstrators remained high during both pre-‐ shock and shock period, but the freezing of the observers during shock periods declined over the subsequent days.

Overall freezing

A 5 condition (between factor) x 2 shock period (within factor) x 4 test days (within factor) ANOVA for the demonstrators revealed a significant main effect of test day (F(3,90)=248.29, p<0.001), a main effect of shock period (F(1,30)=75.81,

p<0.001), but no significant effect of condition (F(4,30)=1.54, p=0.23). There was

however a significant interaction between test day and condition (F(12,90)=2.79,

p<0.003), shock period and condition (F(4,30)=3.467, p<0.02) and test day and shock

period (F(3,90)=100.07, p<0.001). A post hoc test on test days was performed for the

demonstrators’ freezing (all conditions combined and the overall freezing: pre-‐shock

and shock period taken together), which revealed a statistically significant difference when day 1 was compared to days 2, 3 and 4 (p<0.001 in all cases), as well as a difference between test day 2 and 4 (p=0.035). The remaining days did not differ significantly (test day 2 vs. 3: p=0.48, day 3 vs. 4: p=0.19).

(21)

The same ANOVA was performed for the observers’ freezing data. Again, there was a main effect of test day (F(3,90)=15.23, p<0.001) and shock period (F(1,30)=9.47, p<0.005). For the observers also a main effect of condition was observed (F(1,30)=3.07, p=0.031). Furthermore, a significant interaction effect between shock period and condition (F(4,30)=4.22, p=0.008) and between test day and shock period (F(3,90)=19.30, p<0.001) was found. There was no significant interaction between test day and condition for the observers (F(12,90)=1.59,

p<0.11). Post hoc analyses of test day (observers, all conditions combined, pre-‐shock

and shock combined) showed that both test days 1 and 2 differed significantly from 3 and 4 (but not from each other), and 3 and 4 were also significantly different from one another (day 1 vs. 2: p=0.71, day 1 vs. 3: p=0.014, day 1 vs. 4: p<0.001; day 2 vs. 3: p<0.005, day 2 vs. 4: p<0.001; day 3 vs. 4: p=0.006). A post hoc test on total freezing per conditions (pre-‐shock and shock combined, all test days taken together) showed that the week 13 condition differed from all other conditions except the unfamiliar condition (unfamiliar vs. 1-‐week: p=0.16, unfamiliar vs. 3-‐week: p=0.18, unfamiliar vs. 5-‐week: p=0.054, unfamiliar vs. 13-‐week: p=0.49, 1-‐week vs. 3-‐week:

p=0.90, 1-‐week vs. 5-‐week: p=0.59, 1-‐week vs. 13-‐week: p=0.027, 3-‐week vs. 5-‐week: p=0.49, 3-‐week vs. 13-‐week: p=0.029, 5-‐week vs. 13-‐week: p=0.006).

Differences between conditions

To further investigate whether this effect of condition was due to differences in pre-‐shock or shock period freezing, a repeated measures ANOVA was performed on the observer data for pre-‐shock period and shock period separately (test day as 4 level within factor and condition as 5 level between factor). The freezing of observers during the pre-‐shock period showed a main effect of test day (F(3,90)=8.65, p<0.001) and condition (F(4,30)=6.61, p<0.001). A planned post-‐hoc on conditions showed that the freezing of observers during pre-‐shock period (freezing from all days combined) was significantly higher in the unfamiliar and 13-‐week condition than in all other conditions (unfamiliar vs. 1-‐week: p=0.03; unfamiliar vs. 3-‐week: p=0.04; unfamiliar vs. 5-‐week: p=0.01; unfamiliar vs. 13-‐week: p=0.28; 1-‐week vs. 3-‐week:

p=0.81; 1-‐week vs. 5-‐week: p=0.69; 1-‐week vs. 13-‐week: p<0.001; 3-‐week vs. 5-‐

Modulating factors of emotional contagion in rats. A behavioral study on the effects of familiarity and repeated testing on the socially triggered freezing response.