The influence of innate and learned
freezing on observational freezing
responses in rats.
Mirjam Heinemans
20-10-2017
Student number 10326111 University of Amsterdam Master Brain and Cognitive Sciences Supervisor Dr. Marta Moita UvA-representative Dr. Harm KrugersAbstract
Rats can use cues from conspecifics to detect potential threats. Upon detection of such social cues, the rats will display defensive behaviours like freezing. Previous studies have shown that in order to exhibit ‘observational freezing,’ the animals have to have experience with foot shocks. Unpublished data from our lab demonstrated that it was essential for the rat to have the opportunity to freeze while experiencing shocks: without this opportunity rats did not display observational freezing. This led us to hypothesize that if freezing occurs while the animal receives shocks, freezing could become a conditioned stimulus that predicts the event of shocks. With the present study we aimed to further elucidate the role of freezing experience and learning for subsequent observational freezing. The question we aimed to answer was whether stimuli other than shocks could also induce observational freezing, and whether this was related to the amount of learning these stimuli induced. To test this, two stimuli besides foot shocks were used to trigger freezing: visual looming and 2-MT. Looming is a visual cue that signals danger to rats and previous experiments in our lab showed that presenting a looming stimulus induced reliable freezing. 2-MT is an odorant developed by Kobayakawa, K., and Kobayakawa, R. (2011) that induces the same level of freezing as foot shocks. Our results show that only the combination of freezing with shocks led to contextual learning and display of observational freezing. Animals that experienced freezing through looming or 2-MT did not freeze to context, nor did they exhibit observational freezing. These results support the hypothesis that experiencing freezing while being shocked induces a memory connection between the two events. As a result freezing of a conspecific can be used as a conditioned cue to predict shocks.
Abstract 2 Introduction: threat detection and defensive responses in social animals 4 Social context of defensive behaviour 5 Olfactory cues 12 Visual cues 14 Conditioned defensive responses 15 Can visual looming or 2-MT induce observational freezing? 17 Materials and Methods 19 Subjects 20 Behavioural Apparatus 20 Optimization of visual looming and 2-MT protocol 23 Experimental procedures 23 Animal inclusion criteria 25 Statistical Analysis 26 Results 27 Summary of experimental procedures. 27 Observer and demonstrator freezing during conditioning and training 30 Freezing during social interaction 35 Correlations between training and social interaction 42 Contextual freezing 44 Discussion 47 Observational freezing 48 Conditioning plays a role in observational freezing 49 Involvement of the amygdala in shock, looming and 2-MT processing 50 Shocks: conditioned pathway 51 Looming: visual pathway 52 2-MT: olfactory pathway 53 Intensity of stimuli modulates strength of learning 55 Other behavioural measures for learning 57 Social buffering 58 Future analyses 60 Conclusion 62 Acknowledgements 63 References 64 Supplementary figures 68
Introduction: threat detection and defensive responses
in social animals
The ability to detect and avoid threats such as predators, aggressive conspecifics, or physically harmful stimuli is important for survival of all animals. The first way to detect these threats is by identifying cues that derive directly from the threat (i.e. predators or aggressive conspecifics; reviewed by (Gross & Canteras, 2012; Silva, Gross, & Graff, 2016)). These cues trigger innate defence responses: no prior learning is required for the response (Bolles, 1970). A second manner to detect threats is by learning which environmental cues predict the presence of impending danger, upon which they will display learned or conditioned defensive responses. The process of how animals connect the memory of an aversive stimulus (unconditioned stimulus) to a previously neutral stimulus (conditioned stimulus) and learn to express defensive behaviour to the conditioned stimulus has been studied extensively and is commonly referred to as 'fear conditioning' (reviewed by Maren, 2001; LeDoux, 2000). Both threat-derived and conditioned cues can be of olfactory, tactile, visual or auditory nature (Pereira & Moita, 2016; Silva et al., 2016). Defensive responses to such stimuli can be grossly subdivided into flight, freeze or fight responses (R J Blanchard & Blanchard, 1969; Robert J Blanchard & Blanchard, 1969; Misslin, 2003). The exact behaviour displayed depends on the threat and the environment. While flight and avoidance are thought to be the most commonly used strategies when possible (R J Blanchard & Blanchard, 1969; Misslin, 2003), freezing is displayed when animals are confronted with an inescapable threat (Robert J. Blanchard & Blanchard, 1971). Freezing has been studied extensively in laboratory rodent models and is defined as
adopting a crouching posture while showing absolutely no movement except for breathing (Robert J Blanchard & Blanchard, 1969).
Social context of defensive behaviour
In their natural environment, animals rely on cues deriving from conspecifics to increase survival chances (e.g. (Howell, 1979; Pulliam, 1973; Zuberb, 2001). These social cues can provide useful information regarding potential mates, presence of food, and safety of the surroundings (Beach, 1976; Howell, 1979; Townsend, Rasmussen, Clutton-brock, & Manser, 2012). The social environment also plays an important role in threat detection and avoidance. For instance, animals that live in groups can increase their survival chances by diluting the risk of being caught, better known as the phenomenon of the selfish herd (Hamilton, 1971; Lehtonen & Jaatinen, 2016). A second advantage of operating in a social environment is that the animals can combine forces to detect potential predators: two pairs of eyes see more than one (Pulliam, 1973). Upon detection of predators, individuals can communicate this information to others, for example through pheromones or alarm calls; this has been described in the wild in several species, including jays, squirrels, monkeys and meerkats (Griesser, 2009; Mateo, 1996; Townsend et al., 2012; Zuberb, 2001).
The use of social cues to detect threats has also been studied in rats and mice in the lab (e.g. Brechbühl et al., 2013; Kavaliers, Colwell, & Choleris, 2003; Kikusui, Takigami, Takeuchi, & Mori, 2001; E. J. Kim, Kim, Covey, & Kim, 2010a; Knapska, Mikosz, Werka, & Maren, 2010; Parsana, Moran, & Brown, 2012). In most of these studies, the animals react to social cues with freezing – a phenomenon that will be referred to as
‘observational freezing’ in this report. The existence of this phenomenon has been demonstrated first by Church as early as 1959. He trained a food deprived rat to press a lever for food and subsequently, while the trained rat pressed for food, he presented it with a conspecific receiving foot shocks. Upon witnessing the distress of a conspecific receiving shocks, the lever-pressing rats would stop pressing and become immobile. Further research on which social cues trigger defensive responses in rodents showed that these cues can be of olfactory, auditory, or visual nature (Kikusui et al., 2001; E. J. Kim et al., 2010a; Pereira, Cruz, Lima, & Moita, 2012). Olfactory cues that signal danger are known as alarm pheromones and are secreted by stressful individuals (Inagaki et al., 2014; Kikusui et al., 2001). These alarm pheromones are detected by others, which in turn react to this information with increased anxiety and defensive responses such as freezing and avoidance (Brechbühl et al., 2013; Inagaki et al., 2014; Kikusui et al., 2001). In one example study, Kikusui and colleagues (2001) shocked one rat in a confined box, and put another rat in that environment directly afterwards. Upon being exposed to the alarm pheromones the previous animal had released, the second rat reacted with increased freezing display. Recently, Brechbühl et al. (2013) isolated and identified one of the alarm pheromones in these odours that mediate the freezing response: the molecule 2-sec-butyl-4,5-dihydrothiazole or SBT. Furthermore, they discovered that the detection of alarm pheromones happens through a specialized olfactory organ, the Grueneberg Ganglion system.
Social alarm cues can also be of auditory nature, usually trough emission of alarm calls that warn conspecifics about potential danger. This phenomenon has been observed in various species such as chickens, jays, ground squirrels, and monkeys (Evans, Evans, & Marler, 1993; Griesser, 2009; E. J. Kim, Kim, Covey, & Kim, 2010; Mateo, 1996;
Parsana et al., 2012; Townsend et al., 2012; Zuberb, 2001). Experimental studies with rats showed that they react with freezing to 22kHz vocalizations of conspecifics, indicating that rats too can use vocalizations to infer potential threats (R J Blanchard, Blanchard, Agullana, & Weiss, 1991; E. J. Kim et al., 2010; Parsana et al., 2012). Finally, Pereira and colleagues (2012) discovered another type of auditory cue that can signal impending danger, namely the sudden cessation of movement-evoked sounds from conspecifics, as observed when animals suddenly freeze. In this study, the rats underwent a social interaction test. In this test an observer rat and a tone-conditioned demonstrator rat were put in a social environment (separated by a grid that allowed them to see, hear, smell and touch each other), after which the conditioned tone was played. The demonstrator would freeze to the tone, and the level of freezing of the (tone-naïve) observer in reaction to the demonstrator was the measure for observational freezing of this couple.
So far, there is little evidence for the use of visual social cues in rats with regard to threat detection, but different studies with mice showed that these animals react to visual cues of movement in others, indicating that, in theory, visual social cues could play a role in threat detection in rats too (Jeon et al., 2010; Langford et al., 2006). However, from the study by Pereira et al. (2012) with rats, it appears that social visual cues are not essential to evoke a response in rats.
Interestingly, in all the above-mentioned experiments the observing rats had to have experienced foot shocks in order to react to social cues with freezing. The rats in Church´s study only showed a response in the form of depression of lever pressing upon observing a conspecific being shocked if they had received shocks themselves before the test (Church, 1959). More recent studies showed this effect as well: observer
rats would respond with freezing to demonstrators only when they had experienced shocks themselves (Atsak et al, 2011; E. J. Kim et al (2010a); Pereira et al, 2012). Building on these observations, the aim of the current project was to better understand the relationship between different aspects of the aversive experience of foot shocks and expression of observational freezing responses. Previous work from our lab and others suggests that stress sensitization as a result of experiencing an aversive stimulus is not sufficient to explain occurrence of for observational freezing (Parsana et al., 2012). On the contrary, unpublished data from our lab suggests that stressful stimuli other than shocks, such as a forced swim test, do not trigger observational freezing in a subsequent social interaction test as described by Pereira et al., 2012 (Cruz et al., unpublished). Moreover, experience of shocks does not seem sufficient to produce an observational freezing response either. For instance, the studies by Parsana (2012) and E. J. Kim (2010a) and their colleagues both suggest that rats require experience with their own vocalizations during the foot shocks in order to respond with freezing to vocalizations emitted by a conspecific. The combined evidence of these experiments indicate that rats require experience of an aversive event, the opportunity to respond to that event, and to form a connection between the event and their response in order to subsequently react to behaviour of conspecifics. This is in line with the other studies looking into observational freezing: all observer rats had time to freeze during the experience with shocks (Atsak et al., 2012; Pereira et al., 2012). To test whether the opportunity to freeze while experiencing shocks was indeed necessary for subsequent expression of observational freezing, a series of experiments was performed in our lab with different shock paradigms. In these experiments, all observer rats received three foot shocks, although under three different conditions. In the first condition animals were shocked immediately after being placed in a novel box – without any time in
between the shocks – and removed from the box immediately after receiving the shocks (immediate shock condition). In the second condition the animals were given a five-minute baseline period before receiving shocks to explore the environment. Then, they received three consecutive shocks and taken out of the box immediately (delayed shock condition). In the last condition the animals had a five minutes baseline period to explore the box, and then received the shocks with a three-minute interval, thus providing them with an opportunity to freeze (spaced shock condition).
The animals in the immediate shock condition did not have any chance to explore the environment or freeze before or after receiving shocks. As a result, they did not get the opportunity to associate the shocks with any contextual cues. The rats in the delayed shock condition did have time to explore the box before, which meant they could link the context to the shocks, but they did not get the chance to freeze in between shocks. In contrast, the spaced shock animals had six minutes to relate the shocks to any present environmental cues, including their own responses. The animals in the delayed shock and the spaced shock showed similar amounts of context freezing, while the animals in the immediate shock did not freeze to the context. Similar results have been shown before by Fanselow et al. (1980), in an experiment where animals were either shocked immediately upon being put in a box, or after a 2-minute baseline, demonstrating that rats only froze to context if they had the opportunity to explore the environment before and therefore had the possibility to associate the contextual information to the shocks. The intriguing results from the experiments by our lab were that despite the fact that delayed shock observers froze to the context, they did not show observational freezing. In contrast, the spaced shock animals showed both observational freezing and contextual freezing. The results from these experiments suggest that it is essential for the rats to have had the opportunity to freeze upon
experiencing shocks; without this opportunity, shock-experienced rats will not exhibit an observational freezing response (Cruz et al., unpublished, see supplementary figures 3 and 4).
These results led to the question which components of the spaced shock paradigm caused the subsequent observational freezing response. It could either be the case that experience with freezing in general is sufficient to trigger this response, or it could be that the shocks drive learning the association between experienced freezing and an aversive event. The two possibilities are shown in figure 1. As illustrated in the figure, when a rat is shocked, it starts exhibiting conditioned freezing to the context. When the animal is shocked again while it is freezing, it could subsequently connect its own freezing response to the shocks. In other words, the freezing would become a conditioned stimulus and the next time the rat encounters freezing – either its own or from a conspecific – it would respond with a conditioned defence behaviour, i.e. freezing.
The aim of the current project was to further elucidate which scenario is the correct one: is experience with freezing sufficient to drive observational freezing, or is learning about the combination of shocks and freezing necessary? To do so, an experiment was designed in which rats experienced freezing either through the described spaced shock paradigm, or through one of two other stimuli. These other stimuli relied on different sensory modalities and were mediated by distinct freezing. The first stimulus we used was 2-MT, a synthetic derivative of TMT, which is a chemical found in fox faeces and known to trigger innate freezing through activation of an olfactory pathway. The second stimulus was visual looming, which drives innate freezing through the visual pathway. After the animals had experienced freezing
through one of these stimuli (shocks, 2-MT, or looming), they underwent a social interaction test as described by Pereira et al. (2012). In the next paragraphs the stimuli will be discussed in more detail.
Unconditioned stimulus
Context
Conditioned stimulus Freezing Conditioned responseContext
Conditioned stimulus Freezing Conditioned stimulus Freezing Conditioned response Freezing Conditioned response3
FreezingInnate or conditioned response
Freezing
Observational response
2
1
OR
Figure 1. Schematic representation of how experience with shock and freezing
could turn freezing into a conditioned stimulus.
This figure describes how freezing combined with shocks could lead to observational freezing. Figure 1.a represents the mechanism through which freezing could turn into a conditioned response. When a rat receives a shock, it will learn to associate that with the context it which the shock occurred (arrow 1). As a result rats will show a conditioned freezing response to this context (arrow 2). If the animal is freezing while receiving a next shock, it could link its freezing to the shock, just as the context is linked to shocks (arrow 3). This would turn freezing into a conditioned stimulus. If this occurs, the next time the animal encounters either the context or freezing again (figure 1.b), this will activate the memory of the shocks and the rat will display a conditioned freezing response. Alternatively, the experience with freezing in general (1.c) could be sufficient to trigger subsequent observational freezing responses (1.d).
b) a)
d) c)
Olfactory cues
Olfaction is a particularly important sense in nocturnal species like rats, which are known to respond with defensive behaviour to chemicals excreted by predators (Ferrero et al., 2011; Kobayakawa et al., 2007; Papes, Logan, & Stowers, 2010; Pérez-Gómez et al., 2015; Rosen, Asok, & Chakraborty, 2015). Research looking into the mechanism underlying this process shows that different olfactory subsystems are responsible for processing specific types of chemical cues (reviewed by (Gross & Canteras, 2012)). The vomeronasal organ (VNO) is thought to detect non-volatile chemicals such as major urinary proteins (MUPs) (Papes et al., 2010), while the main olfactory epithelium (MOE) detects different volatile molecules that are excreted by predators, such as 2-phenylethylamine (2-PEA) and trimethyl-thiazoline (TMT) (Ferrero et al., 2011; Kobayakawa et al., 2007). TMT is secreted from the anal gland of foxes and used in the lab to elicit defensive responses such as avoidance and freezing in rodents (Vernet-Maurt, Polak, & Demael, 1984; Wallace & Rosen, 2000). Neurons from the MOE that detect TMT are thought to project to the anterior cortical amygdala, a structure involved in defensive behaviours (Miyamichi et al., 2011; Root, Denny, Hen, & Axel, 2014). Root and colleagues (2014) showed that optogenetic inhibition of this area reduced TMT-induced defensive responses significantly, while selective activation of TMT-responsive neurons in the cortical amygdala triggered defensive behaviour. Furthermore, exposure to TMT is known to increase c-Fos expression in multiple areas in the brain, including the central amygdala, which is known to be an important output centre for defensive behaviours (Day, Masini, & Campeau, 2004). However, muscimol-inactivation of the central or lateral amygdala did not block freezing upon TMT exposure, while inactivation of the bed nucleus of the stria
important role in TMT-mediated defensive responses (Fendt, Endres, & Apfelbach, 2003).
The main defensive response of rats to TMT appears to be avoidance, while freezing responses seem less reliable, although there have been studies that reported freezing in reaction to the chemical (Endres, Apfelbach, & Fendt, 2005; Vernet-Maurt et al., 1984). However, pilots in our lab with TMT did not result in sufficient freezing, while presentation of 2-MT – a synthetic derivative of TMT – induced levels of freezing comparable to shocks in other studies and our pilot (Isosaka et al., 2015). Therefore, in the current experiment 2-MT was used to trigger innate freezing in rats through an olfactory pathway.
Visual cues
Visual stimuli that signal the rapid approach of a potential threat are known trigger innate defence responses in different species (Dunn et al., 2016; Schiff, Caviness, & Gibson, 1962; Yilmaz & Meister, 2013). In particular the effect of visual looming stimuli (a shadow rapidly approaching from above) on display of defensive behaviours has been investigated extensively (e.g. Yilmaz & Meister, 2013; Zhao, Liu, & Cang, 2014). Converging evidence from these studies shows that the superior colliculus (SC), which receives input from both the retina and the visual cortex, plays an important role in the detection of looming stimuli in mice (Shang et al., 2015; Wei et al., 2015; Westby, Keay, Redgrave, Dean, & Bannister, 1990). Multiple experiments have demonstrated that the SC projects to the amygdala via two different pathways: through connections with the parabigeminal nucleus (PBGN) (Usunoff, Schmitt, Itzev, Rolfs,
& Wree, 2007); and through the lateral posterior nucleus of the thalamus (LP) (Wei et al., 2015). Wei and colleagues (2015) showed that optogenetically silencing the SC neurons projecting to the LP attenuated the freezing response to looming stimuli in mice, while optogenetic activation of the same neurons elicited freezing. They furthermore demonstrated that the freezing was dependent on activation of the lateral amygdala occurring after stimulation of the SC neurons (Wei et al., 2015). Shang and colleagues (2015) looked at the role of a different population of SC neurons, projecting to the PBGN, in defensive responses towards looming stimuli. They found that activation of these neurons triggered a stereotyped defence response in mice: they would show an escape response followed by freezing (Shang et al., 2015). Both the PGBN and the lateral amygdala project to the central amygdala (Sah, Faber, De Armentia, & Power, 2003; Shang et al., 2015), a region that has been shown to drive freezing through projections to the PAG and is thought to be an important output centre for defensive responses (Ciocchi et al., 2010; Keifer, Hurt, Ressler, & Marvar, 2015).
Conditioned defensive responses
The discussed visual and olfactory stimuli elicit so-called innate freezing responses, i.e. no prior learning was required for the response (Bolles, 1970). Besides these cues, animals can also rely on learned cues to detect danger. In that case the animal learned through experience that a certain cue (the conditioned stimulus) predicts the occurrence of an aversive or painful event (the unconditioned stimulus, under experimental conditions usually a foot shock). Conditioned cues can be olfactory, visual, auditory or
conditioning’, explained in (LeDoux, 2000; Tovote, Fadok, & Lüthi, 2015). Pairing the tone with an aversive shock will create a link between the tone and the memory of the shock, and the next time a tone-conditioned animal hears the tone, it will exhibit a conditioned defensive response like freezing (reveiwed by LeDoux, 2000). Another form of conditioning is known as ‘contextual conditioning’ (Fanselow, 1980). In this paradigm no specific cue is presented before the aversive event, so the animals can only use the environment in which the shocks occurred as a conditioned stimulus. An animal that was shocked without a cue will learn to express conditioned defensive behaviour in that context, which is again usually freezing, since the animal cannot escape the environment (Robert J Blanchard & Blanchard, 1969; Fanselow, 1980). The neuronal circuits of conditioned defence behaviours have been investigated in great detail over the years (reviewed by e.g. (Duvarci & Pare, 2014; LeDoux, 2000; Tovote et al., 2015). From this research it has become apparent that tone conditioning relies on activation and plasticity in the lateral amygdala (e.g. Herry & Johansen, 2014; Pape & Pare, 2010). Certain cells in this brain region receive input from neurons originating in thalamic and cortical regions that carry information about either the aversive stimulus such as shocks or neutral stimulus like a tone (Romanski, Clugnet, Bordi, & Ledoux, 1993). While neutral stimuli only elicit weak activation responses in these cells, the aversive stimuli evoke larger responses (Duvarci & Pare, 2014). However, upon receiving simultaneous inputs from an aversive and neutral stimulus, the activation pattern in response to the neutral stimulus changes: from now on the cell will react to the previously neutral stimulus with a strongly increased activation, caused by a process known as ‘long term potentiation’ (Bliss & Colligridge, 1993; Rogan, Staubli, & LeDoux, 1997). In other words, the previously neutral stimulus has become a conditioned stimulus.
The lateral amygdala in turn projects to the central amygdala, both directly and indirectly via the basolateral amygdala (reviewed by Duvarci & Pare, 2014; LeDoux, 2000). The central nucleus of the amygdala is also thought to play an important role in both acquisition and expression of conditioned fear (Wilensky, Schafe, Kristensen, & LeDoux, 2006; Zimmerman, Rabinak, McLachlan, & Maren, 2007). Several studies showed that inactivation of the central amygdala during either acquisition or activation of conditioned responses resulted in a decrease of these responses. The efferent projections from the central amygdala to the ventral PAG are crucial for the expression of freezing in rodents (J. J. Kim, Rison, & Fanselow, 1993).
Could visual looming or 2-MT induce observational freezing?
In summary, freezing responses in rats can be triggered by cues deriving either directly from potential threats, from cues related to threats through learning, or from conspecifics (Atsak et al., 2011; Blanchard & Blanchard, 1969; Fanselow, 1980; Kikusui, Takigami, Takeuchi, & Mori, 2001; LeDoux, 2000; Parsana, Moran, & Brown, 2012; Rosen, Asok, & Chakraborty, 2015; Yilmaz & Meister, 2013). Threat-derived cues of different modalities have been investigated extensively (reviewed by e.g. Pereira & Moita, 2016; Silva et al., 2016). In contrast, only little is known about which mechanisms drive the expression of observational freezing. Converging evidence suggests that rats have to have the opportunity to freeze upon experiencing shocks in order to display observational freezing (Cruz et al., unpublished, supplementary fig. 3 and 4). There are different potential explanations for this observation. Firstly, it could be that experience with freezing in general is sufficient to
between freezing and shocks has to be learned in order to be capable of displaying observational freezing. To test this, the animals experienced freezing through shocks or one of two other freezing-inducing stimuli: looming and 2-MT before undergoing a social interaction test. As explained above, these stimuli trigger innate freezing and they differ from each other and from shocks in sensory modality and in the anatomical pathway through which they trigger freezing.
We expected that stimuli could only induce observational freezing if they were capable of driving learning, as shown for shocks in figure 1. Therefore, we looked at both the contextual fear and the observational fear in rats exposed to all three stimuli. As 2-MT does not appear to induce contextual freezing (Wallace & Rosen, 2000), we expected it would not trigger observational freezing either. To our knowledge, there have been no studies looking at looming as an unconditioned stimulus for aversive conditioning. In theory looming could drive learned defence responses, since it is processed in the lateral and central amygdala, both implicated in conditioning (reviewed by e.g. Ledoux, 2000; Tovote, Fadok, & Lüthi, 2015). Again, we expected that if looming can drive contextual freezing, the animals would also exhibit observational freezing. If experience of looming does not result in contextual freezing, we expect to see no observational freezing in the looming condition either. If however experience with freezing in general is sufficient to drive observational freezing, there should be no difference between the conditions with regard to freezing during the social interaction test. A summary of the stimuli characteristics is shown in table 1.
Stimulus Experienced
freezing
Learning
Observational
freezing
Central/Lateral
amygdala involved
Shock
Conditioned
✔
✔
✔
Looming
Innate
?
?
✔
2-MT
Innate
✖
?
✖
Table 1. Summary of the characteristics of the stimuli used to induce
freezing.
The differences between the three stimuli used in the experiment are shown here. As explained, experience of 2-MT triggered freezing was not expected to induce observational freezing. In contrast, it was unknown what experience with looming-triggered freezing will have on the subsequent tests. Furthermore, while the central and lateral amygdala are necessary for the processing of looming stimuli and shocks, 2-MT is processed through other brain regions.
Materials and Methods
Subjects
Naïve males Sprague Dawley rats, weighing 225g to 250g were obtained from Charles Rivers Laboratories (France). Upon arrival the animals were pair-housed in Plexiglas top filtered ventilated cages (GR900 for rats, Tecniplast S.p.A, Italy) with ad libitum access to water and food. They were maintained on a 12h light/dark cycle (lights off at 8 p.m.), a temperature of 20-22C and a humidity degree of 40-70%. After a minimum of one-week acclimatization, the experimenter handled all animals on three consecutive days in the week preceding experimental procedures. All animal procedures were performed under the guidelines of the Animal Welfare Body of the Champalimaud Research (Portugal) and in strict accordance with the European Community’s Council Directive (86/609/EEC).
Behavioural Apparatus
Three distinct environments were used for both demonstrator and observer rats: a conditioning box (where rats received one of three stimuli: shocks, visual looming stimuli, or 2-MT); a neutral box; and a social interaction box. Demonstrator and Shock condition: The conditioning box (model H10-11RTC, Coulbourn Instruments) for Demonstrators and Observers in the shock condition was equipped with a metal grid floor to deliver foot shocks (model H10-11RTC-SF, Coulbourn Instruments) and
placed inside a high sound isolation chamber (Action, automation and controls, Inc) with white walls. The side-walls of the conditioning box were made of clear Plexiglas and cleaned with rose scented detergent after every conditioning round. A precision programmable shocker (model H13-16, Coulbourn Instruments) was used to deliver foot shocks. For the tone conditioning of the Demonstrators a sound generator (RM1, Tucker-Davis Technologies) produced the tones for conditioning, which were delivered through a horn tweeter (model TL16H80HM, VISATON). The sound was calibrated using a Brel and Kjaer microphone (type 4189) and a sound analyser (hand held analyser type 2250). The rats’ behaviour was tracked by a video camera mounted on the ceiling of each attenuating cubicle. An infrared surveillance video acquisition system was used to record and store all videos on hard disk and freezing behaviour was posteriorly scored manually.
Visual looming exposure box: the behavioural box was made of black Plexiglas floor with dark red sides (30cm wide x 50cm height x 55cm depth) and was cleaned with a 70% ethanol solution. This box was placed in a room with ceiling lights on. Stimuli were projected with an LED projector (ML750e, Optoma Europe ltd, United Kingdom) onto an opaque white Plexiglas screen placed on top of the behavioural box. The behaviour was captured with an infrared camera (PointGrey Integrated Imaging Solutions GmbH, Germany) controlled by a custom workflow using the Bonsai visual programming language (Lopes et al., 2015), and stored on hard disk for posterior manual scoring.
2-MT exposure box: this box was made of clear Plexiglas walls (60cm wide x 34cm height x 27cm depth) (Gravoplot), and was cleaned with a 70% ethanol solution. Because the nature of the used stimulus, the 2-MT experiments were performed inside
a fume hood. The behaviour was recorded with a hand-held video camera, and the videos stored on hard disk for posterior manual scoring of freezing.
Neutral box: To reduce the levels of fear generalization in all animals, the shock conditioning chambers were used as neutral a environment, this time placed in a high sound isolation chamber (Action, Automation and Controls, Inc) with black walls. In order to neutralize the box more, the rod floor was covered with a black-and-red-striped acrylic plate, and a house light was on. Furthermore, the sidewalls of the box were made of polished sheet metal and cleaned with a natural soap scented detergent. Social interaction boxes: the boxes consisted of a two-partition chamber made of clear Plexiglas walls (60cm wide x 34cm height x 27cm depth) (Gravoplot), and were divided in two equal halves by a clear Plexiglas wall with 0.7cm wide vertical slits separated by 1.5cm, that allowed the animals to see, hear, smell and touch each other. Each side of the box floor contained a removable tray with bedding (the same used in the animal’s home cages). These boxes and trays were cleaned with water and ethanol 70%. The social interaction boxes were placed inside sound attenuation chambers (80cm wide x 52.5cm height x 56.5cm depth) made of MDF lined with high-density sound attenuation foam (MGO Borrachas Tcnicas) and a layer of rubber. The behaviour of the animals was tracked by infrared video cameras mounted on the walls of the sound attenuating chambers, one on each partition. A surveillance video acquisition system was used to record and store all videos on hard disk and freezing behaviour was automatically scored using FreezeScan from Clever Sys. In all tests, the rats were considered to be freezing if they did not show any movement except breathing for at least one second.
Optimization of visual looming and 2-MT protocol
Before running the experiment, the new freezing paradigms had to be optimized, to ensure the animals did experience similar amounts of freezing prior to the social interaction (SI) test. To achieve this, two separate pilots were run for the looming and the 2-MT exposure. The pilot with the 2-MT was done with four animals that were exposed to the 2-MT for three to seven minutes after a baseline around two minutes. The median freezing of the four animals was 33%, with a range between 14% and 68%. Baseline freezing was below 7% for all animals. The eight animals in the looming pilot were presented with different looming frequencies (5 presentations of 10 looms, 5 presentations of 20 looms and 8 presentations of 20 looms). To ensure freezing would not decline in between presentations the 8 times 20 looms paradigm was used in the experiment, with 30 seconds in between each presentation as to ensure high freezing percentages throughout the six minutes of training. The median freezing percentage during training in the pilot for this paradigm (n=4) was 87%, with freezing percentages between 48% and 96%.
Experimental procedures
The experiments were performed with pairs of cage-mate rats, each one assigned randomly to be the demonstrator or the observer. On the first three experimental days each rat was exposed for 15 minutes to each of the boxes; conditioning, social and neutral for demonstrators, and conditioning/looming/2-MT, social and neutral for observers, with time between exposures ranging from 20 to 28 hours. On the fourth day, demonstrator rats were placed in the conditioning chamber where, after an initial
baseline period of 5 minutes, received 5 tone shock pairings (tone: 15s, 5kHz, 70dB; shock 1mA, 0.5s), with the tone and shock co-terminating and an average inter trial interval of 180 seconds (ranging from 170 to 190s). After the last tone-shock pairing animals are returned to their home cage.
All observers were placed in the training environment for a period of five minute before stimuli were presented to establish a baseline. Observers in the shock condition received 3 unpredicted shocks with a three-minute interval in between (with the same shock intensity and schedule as demonstrators). Observers receiving looming stimuli were presented with eight looming sessions over a period of 380 seconds. Each session consisted of 20 one-second ‘looms’, with a 30 second interval between looming sessions. The visual looming stimuli were generated with PsychoPy (Peirce, 2009). The stimuli consisted of an expanding black dot (0cm to 30cm diameter in 0.5 seconds), on a grey background. Observers subjected to the 2-MT presentation were taken in single boxes to the room with the fume hood. After a five-minute baseline, 3 small filter papers inside a small plastic container embedded with 6 µl of 2-MT (98% 2-Methyl-2-Thiazoline, Sigma-Aldrich Co., Oakville, ON, Canada) were presented to the animal, with an inter-stimulus interval of 2 minutes, for a total duration of six minutes. After the stimulus presentation all observers were returned to their home-cage.
On the fifth day, the pairs of rats were tested in the social interaction box. Each animal was placed on one side of the two-partition box, and after a five-minute baseline period 3 tones (same tone as described above) were presented, with a three-minute inter-tone interval. On day six, observers were placed back in their conditioning chamber and
their behaviour was recorded for posterior assessment of context fear by manual scoring the time they spent freezing over a period of five minutes.
Animal inclusion criteria
To be able to compare the amount of observational freezing accurately between conditions, it was necessary to only include the pairs where there was a clear demonstration of freezing by the demonstrator. Three inclusion criteria were used for this. First, all pairs with demonstrators that showed baseline freezing percentages higher than 66% were excluded, since those animals would not demonstrate the transition from non-freezing to freezing well. Previous work from the lab demonstrated this transition important (Pereira et al., 2012). Secondly, only pairs with demonstrators that robustly demonstrated freezing after the first tone were included (defined as a freezing percentage higher than 33% during the first three minutes after playing the conditioned tone). Finally, to ensure the freezing of the observer was due to observational freezing and not generalized freezing to the context, all pairs of which the observer froze over 20% more than its demonstrator during baseline were excluded. Furthermore, one pair in the looming condition and three pairs in the 2-MT condition were excluded due to the loss of their videos. This resulted in the following numbers of pairs per group for the social interaction: 10 pairs in the shock condition (4 pairs excluded due to no demonstration and 2 pairs excluded due to generalization of the observer), 12 pairs in the looming condition (7 pairs excluded due to the lack of demonstration after the tone, one due to loss of videos), and 8 pairs in the 2-MT condition (again 7 pairs excluded due to the lack of demonstrating, 2 pairs due to loss
Statistical Analysis
The statistical analyses and graphs were made with GraphPad Prism version 7.00 for
Windows (GraphPad Software, La Jolla California USA, www.graphpad.com).
Differences across conditions in freezing percentages during training were analysed by using a non-parametric Kruskal-Wallis test comparing the amount of freezing of the conditions during baseline and stimulus presentation separately. Furthermore, it was checked whether animals in all conditions showed a comparable amount of increase in freezing upon stimulus presentation with a related samples Wilcoxon test. This was done for both observers and demonstrators.
The freezing during the three minutes directly after the first conditioned tone was chosen for the comparison of the social interaction, since demonstrators show the highest and most reliable freezing response during these minutes. A Kruskal-Wallis comparison was done on the normalised freezing of the pairs, by determining the contribution of freezing of the two individuals to their combined amount of freezing. This was calculated by subtracting the amount of freezing of the observer from the demonstrator and dividing it by the amount of freezing of the sum of observer and demonstrator (i.e. (demonstrators-observers) / (demonstrators+observers)). To investigate any correlation between freezing during training, social interaction and context within individuals, a Spearman's rank correlation coefficient test was used. All comparisons were two-tailed and when multiple comparisons were made the reported p-values were corrected using a sequential Bonferroni method. The grouped data is presented as medians with interquartile ranges. Significance was set at p<0.05 for all tests.
Results
Summary of experimental procedures.
A schematic summary of all experimental procedures is shown in figure 2. It shows that before the start of the experiment the pairs were given a week to acclimate to their environment, after which they were habituated to handling by the experimenter 3 times for 3 minutes. The animals were housed in pairs and randomly assigned to be either observer or demonstrator. At the start of the experiment both observers and demonstrators were exposed to three different environments on three consecutive days; they were exposed to the social box and the training box for 15 minutes to habituate them to these, and to a neutral box as to reduce fear generalization after training. On the fourth day, the demonstrators underwent a tone-conditioning paradigm, while the observers experienced freezing through either shocks, looming or 2-MT exposure. The amount of freezing during the training was scored manually afterwards. The subsequent day the animals underwent a social interaction test (SI). During this test demonstrators and observers from the same cage were put together in the social box, and after a baseline of five minutes the conditioned tone was played, upon which the demonstrator would start freezing. The freezing behaviour of both animals was scored and analysed. The final day of the experiment entailed a contextual fear test, where the observers were put back in the environment in which they experienced freezing in on day 4, and again the amount of freezing was scored for each animal.
All results presented here were obtained from analysis of the animal pairs that were considered fit for analysis as described in the methods (see supplementary figure 1 for
details). After excluding all pairs of which either animal showed too much generalization, or that had a non-demonstrating demonstrator, the following number of pairs were analyzed: shock: n=10; looming: n=12; 2-MT: n=6.
Figure 2. Timeline of experimental procedures.
All experimental procedures of the experiment are schematically shown in this figure. Pair-housed rats had ten days of acclimation during which the experimenter handled them three times. After acclimation the animals were exposed to three distinct environments on three consecutive days. On day four of the experiment, cage mates were randomly assigned to be demonstrator (DEM) or observer (OBS). Demonstrators were tone conditioned, while observers were exposed to one of three freezing-inducing stimuli (shock, looming or 2-MT). The next day, the pairs were tested together in a social interaction test where the conditioned tone was played three times after a baseline of five minutes. On the last day of the experiment, the observers were placed back in the same context as in which they had received the stimuli. Freezing behaviour was scored and analysed during all tests from day 4 until day 6.
Acclima'on/
handling (3x)
Day 1-3
Exposure to boxes
(Neutral, SI, Training)Day 4
Training
(shock or looming or 2-MT)Day 5
Social Interac'on
Day 6
Context
DEM ? DEM OBSNeutral Training Social Social
Day 1 Day 2 Day 3
OR OR OBS OR OR OBS Homecage Day -10 to 0
Observer and demonstrator freezing during conditioning and training
In order to assure that the results of the SI test could not be due to differences between the conditioning of the demonstrators, first the freezing of the demonstrators during conditioning was compared across groups. Individual freezing percentages during the five minutes before tone conditioning (baseline) and the seven minutes of the tone conditioning (tone) were calculated and the results are summarized in figure 2.a-2.c (shock: n=10; looming: n=12; 2-MT: n=6). The demonstrators in all conditions showed a significant increase in freezing during the tone conditioning (increase in all conditions had a p-value <0.0001). No statistical differences between groups were found in the amount of freezing during baseline (H(2)=4.74, p=0.094) or conditioning (H(2)=0.019, p=0.99), nor in the increase of freezing (H(2)=0.12, p=0.94; see fig. 2.d). These results indicate that the conditioning worked well in all three conditions and there were no differences in tone conditioning between the demonstrators.
Next, to be able to conclude anything about the influence of prior experience with freezing on observational freezing, we had to make sure that the observers all had experienced comparable levels of freezing during the training. To do so, freezing of all observers was scored during the five minutes of baseline and the six minutes of stimulus presentation and compared across conditions (see figure 4). The results of figure 4.a-4.c show that freezing percentages during baseline were low in all conditions. There is however a statistical difference between shock and looming (p=0.040). This difference in baseline freezing is probably due to the different environments. All three environments have a different level of illumination and differently sized testing boxes, with the shock environment being the smallest and the darkest, the looming being the largest size and the 2-MT having the strongest
illumination. These environmental factors are known to induce anxiety in rodents (Ramos, Berton, Mormede, & Chaouloff, 1997; Tovote et al., 2015).
When comparing the amount of freezing during baseline and stimulus presentation, it becomes apparent that freezing increases significantly in all three conditions (shock: p=0.0020; looming: p=0.0005; 2-MT: p=0.031). The graphs 4.a-4.c show that the freezing upon stimulus presentation is very variable in the shock condition, while observers in the looming condition show a more bimodal pattern, with most animals showing high percentages of freezing. The observers from the 2-MT condition generally showed lower freezing percentages than the other two conditions. Comparing the freezing levels after stimulus presentation across groups showed that the animals in the looming condition froze significantly more than the 2-MT animals (the average freezing during stimulus presentation was 74% for in the looming condition, 32% in the 2-MT), but neither differed significantly from the shock condition, indicating that both treatments elicited freezing levels similar to animals that experienced shocks, which froze on average 57% of the time (H(2)=8.07, p=0.018; multiple comparisons: shock vs. loom: p=0.31; shock vs. 2-MT: p=0.55; loom vs. 2-MT: p=0.017).
In figure 4.d, the increase in freezing per individual is depicted by subtracting freezing percentage during baseline from freezing during stimulus. Comparing the increase across the conditions gave results similar to the comparison of freezing during the stimulus presentation: although the increase in freezing differed between the looming and 2-MT condition (average freezing levels were 68% and 35% respectively), neither was different from the shock condition, which had an average increase of 56% (H(2)=7.52, p=0.023; shock vs. loom: p=0.48; shock vs. MT: p=0.43; loom vs. 2-MT: p=0.020). In conclusion, all observers in all conditions have low freezing in the
baseline, and show a significant increase in freezing upon stimulus. Furthermore, the increase in freezing and overall freezing percentages upon both looming and 2-MT presentation did not differ from the shock condition. These results confirm that all three stimuli induced freezing in the rats, and observers from the looming and 2-MT condition experienced freezing levels comparable to freezing upon shocks before undergoing the SI.
B a s e lin e T o n e 0 2 0 4 0 6 0 8 0 1 0 0 S h o c k F re e z in g % B a s e lin e T o n e 0 2 0 4 0 6 0 8 0 1 0 0 L o o m in g F re e z in g % B a s e lin e T o n e 0 2 0 4 0 6 0 8 0 1 0 0 2 - M T F re e z in g % S h o c k L o o m in g 2 -M T 0 2 0 4 0 6 0 8 0 1 0 0 In c r e a s e F r e e z in g F re e z in g %
a)
b)
d)
c)
*** *** ***Figure 3. Conditioning of the demonstrators.
In figures 3.a-3.c, individual freezing percentages during the 5-minute baseline (Baseline) and during the 12-minute tone conditioning (Tone) are presented per condition (Shock: n=10, Looming: n=12, 2-MT: n=6). Individuals of all groups showed a significant increase in the percentages spent freezing. Figure 3.d illustrates the increase of individual freezing percentages is per condition, with the bars indicating the median and interquartile range. No differences were observed across conditions. The symbol ‘***’ indicates a statistical increase in freezing of p<0.0001.
B a s e lin e S tim u lu s 0 2 0 4 0 6 0 8 0 1 0 0
S h o c k
F re e z in g % B a s e lin e S tim u lu s 0 2 0 4 0 6 0 8 0 1 0 0L o o m in g
F re e z in g % B a s e lin e S tim u lu s 0 2 0 4 0 6 0 8 0 1 0 02 - M T
F re e z in g % S h o c k L o o m in g 2 -M T 0 2 0 4 0 6 0 8 0 1 0 0In c r e a s e F r e e z in g
F re e z in g %Figure 4. Training of the observers.
In figures 4.a-4.c, the individual freezing percentages per different condition are presented during the 5-minute baseline (Baseline) and the 6-minute stimulus presentation (Stimulus) (Shock: n=10, Looming: n=12, 2-MT: n=6). Individuals of all conditions showed a significant increase in freezing percentage upon stimulus presentation. In figure 3.d the increase of each individual’s freezing percentage is shown for the three conditions. The bars indicate the median and interquartile range. Figure 4.d illustrates that freezing of the shock condition was not statistically different from freezing percentages in the other conditions. The freezing percentages of the looming and 2-MT condition did differ however. The symbols ‘*’, ‘**’ and ‘***’ correspond with a statistical significance of p<0.05, p<0.001 and p<0.0001 respectively.
* ** *
a)
b)
d)
c)
* *** ** *Freezing during social interaction
In figure 5.a-5.c, the average freezing behaviour of the demonstrators and observers during the twelve minutes of social interaction (SI) is illustrated (number of included pairs: shock: n=10; looming: n=13; 2-MT: n=6). The line graphs depict the average percentage of time spent freezing in fifteen-second bins over the course of the test session, with the shaded area representing the standard error of the mean. The grey vertical bars indicate the timing of playback of the conditioned tone. As illustrated by the graphs, demonstrators in all three conditions show a clear increase in freezing upon presentation of the first tone, and the freezing persists until the end of the social interaction test. The observers in the shock condition show a similar increase as the demonstrators, as was expected, since prior experience with freezing through shocks is known to induce observational freezing (Atsak et al., 2011; Pereira et al., 2012). Looking at the last minutes of the graphs it can be noticed that the freezing of the observers decreases again over time, suggesting that the observational freezing is a reaction that is subject to extinction or habituation. The observers with prior experience of freezing through 2-MT do not show any increase in freezing, thus no observational freezing was present. The observers in the looming condition show a slight freezing response to the freezing of the demonstrators, but it is significantly lower than the increase of freezing in the shock condition.
To be able to compare amount of observational freezing between groups, a closer look was taken at the three minutes directly after the first conditioned tone, when the demonstrators in all groups exhibited the freezing most reliably. The figures 5.a-5.c reveal that in the shock condition all observers except for one showed freezing percentages similar to the to the freezing of the demonstrators (p=0.13), thereby
indicating an observational freezing response in these observers. In contrast, the freezing of the observers in the other two conditions was significantly lower than that of their demonstrators (looming: p=0.0024; 2-MT: p=0.0078). In case of the looming, graph 6.c also illustrates that the freezing of the observers is mainly driven by two animals, while the remaining observers show considerably lower amounts of freezing. Next, a comparison across groups was made of the freezing during the first three post-tone minutes. Since observational freezing is directly dependent on the demonstrated freezing, first the variability of the demonstrators was filtered out by normalising the freezing for each pair to be able to compare accurately across pairs. Normalisation was achieved by dividing the difference between demonstrators and observers by the total freezing per pair (i.e. (demonstrators-observers)/(demonstrators+observers)). The calculated value represents the demonstrator´s contribution to the total amount of freezing per pair. In other words, if the value was 1, only the demonstrators froze; if it was 0, both animals contributed equally; if the outcome was negative, the observers contributed more to the total amount of freezing than the demonstrators. The results are depicted in figure 6.d. First, statistical analyses were performed to check whether the conditions deviated from zero (indicating there is a equal contribution to freezing). The results show that the shock condition did not differ from zero (p=0.13), but the remaining conditions did differ significantly (looming: p=0.0012; 2-MT: p=0.0078), showing that demonstrators freeze substantially more than observers in these conditions. To verify whether the observers had any contribution to the total amount of freezing, it was subsequently tested whether the normalised values of the looming and 2-MT conditions differed from 1 (which indicated only the demonstrators contributed to freezing during SI). According to the outcome, the animals in the looming condition differed significantly from one (p=0.0005), suggesting that although they did not
contribute equally to the freezing, the observers did contribute to some extent. The normalised freezing values from the 2-MT pairs were not different from one (p=0.063), indicating these observers did not contribute to the freezing during the SI test. Furthermore, the normalised freezing values showed significant differences between the conditions (H(2)=17.5, p=0.0002). Post-hoc analyses to check which conditions drive this outcome shows that shock and 2-MT differ significantly from each other (p<0.0001), while neither differs significantly from the looming condition, although there appears to be a trend in both comparisons (shock vs. loom: p=0.080 and 2-MT vs. loom: p=0.058).
The observational freezing as described above only entails the three minutes post-tone. As can be seen from the figures 5.a-5.c though, there are already freezing bouts before the tone, as well a freezing after the analysed three minutes. To include the information about the overall freezing during the SI, the percentage of time spent freezing was scored in 15s time bins throughout the test session. Next, to compare the fraction of time spent freezing by observers and demonstrators across conditions, we looked at the cumulative distribution of time bins with different levels of freezing, from 0-10% to 90-100%. The results of the cumulative freezing show that the demonstrators of all conditions freeze significantly more during the SI test than their observers, with the period of time they do not freeze being between 33-35% of the time, and the amount of time they were freezing over 95% being around 40% of the test; in all conditions comparing freezing of demonstrators versus that of observers gave a p-value <0.0001 (see figure 7.a-7.c). There were no differences in the overall freezing of the demonstrators (H(2)=2.38, p=0.30, see figure 7.d). The freezing of the observers on the other hand did differ across groups, with the animals from the shock condition freezing the highest amount of time; in 41% of the time bins there was no freezing, while in
27% of the bins observers froze over 95% in. The looming-observers did not freeze during 80% of the time, and the observers from the 2-MT condition did not freeze in over 90% of the time bins. Interestingly, comparing the total amount of freezing of the observers across groups showed that they all differed significantly from one another (H(2)=354.5, p<0.0001, figure 7.e). Post-hoc analyses revealed that the observers from the shock condition froze significantly more than the two other conditions (shock vs. loom: p<0.0001; shock vs. 2-MT: p<0.0001), and the observers in the looming condition in turn froze more than those in the 2-MT condition (p=0.0004).
In summary, the results of the social interaction test show that the observers in the shock condition showed robust observational freezing, while the observers in the looming condition showed very low levels of observational freezing, and the 2-MT-observers expressed no freezing at all. Furthermore, while the normalised freezing of observers from the looming did not differ from the remaining conditions, all three conditions turned out to have statistically significant differences in freezing patterns when looking at the cumulative data, with shocked animals still freezing the most, and 2-MT observers freezing the least, with looming observers being in between. Finally, since the demonstrators of the included pairs did not differ in their cumulative freezing patterns, it can be assumed that the different patterns in cumulative freezing of the observers is indeed due to different amounts of observational freezing across conditions.
Figure 5. Freezing during social interaction test.
Figures 5 depicts the average freezing percentages per 15-seconds of demonstrators (DEM) and observers (OBS) per condition over the course of the social interaction test ( a) Shock: n=10; b) Looming: n=12; c) 2-MT: n=6). The shaded areas represent the standard error of the mean. The social interaction consisted of a 5-minute baseline, after which 3 conditioned tones were played for 15 seconds, with a 3-minute inter-tone interval. The grey vertical bars indicate the timing of
-3 0 0 -2 4 0 -1 8 0 -1 2 0 -6 0 0 6 0 1 2 0 1 8 0 2 4 0 3 0 0 3 6 0 4 2 0 2 0 4 0 6 0 8 0 1 0 0 S h o c k S e c o n d s F re e z in g % D E M O B S -3 0 0 -2 4 0 -1 8 0 -1 2 0 -6 0 0 6 0 1 2 0 1 8 0 2 4 0 3 0 0 3 6 0 4 2 0 2 0 4 0 6 0 8 0 1 0 0 S e c o n d s F re e z in g % D E M O B S L o o m in g -3 0 0 -2 4 0 -1 8 0 -1 2 0 -6 0 0 6 0 1 2 0 1 8 0 2 4 0 3 0 0 3 6 0 4 2 0 2 0 4 0 6 0 8 0 1 0 0 2 - M T S e c o n d s F re e z in g % D E M O B S
a)
b)
c)
D E M O B S 0 2 0 4 0 6 0 8 0 1 0 0 S h o c k F re e z in g % D E M O B S 0 2 0 4 0 6 0 8 0 1 0 0 L o o m in g F re e z in g % D E M O B S 0 2 0 4 0 6 0 8 0 1 0 0 F re e z in g % 2 - M T S h o c k L o o m in g 2 - M T 0 .0 0 .5 1 .0 N o r m a liz e d F r e e z in g (D -O )/ (D + O )
Figure 6. Freezing during three minutes post-tone.
Figures 6.a-6.c illustrate the freezing of individual demonstrators (DEM) and observers (OBS) per condition (Shock: n=10, Looming: n=12, 2-MT: n=6) in the three minutes after the first tone. The dotted lines connect each pair of DEM and OBS. In figure 6.d the normalised freezing per pair is shown. Normalisation was achieved by dividing the difference in freezing by the total freezing per pair: (D–S) /(D+O). The normalised freezing of the shock condition did not differ from 0 (indicated with ‘a’), the freezing of looming differed from both 0 and 1 (‘b’), and the freezing of the 2-MT did not differ from 1 (‘c’). The symbols ‘**’ and ‘***’ correspond with a statistical significance of p<0.001 and p<0.0001 respectively.
a)
b)
d)
c)
** ** a b c ***Figure 7. Cumulative freezing percentages of social interaction test.
This figure represents the cumulative distribution of freezing percentages from demonstrators (DEM) and observers (OBS) in each condition (Shock: n=10, Looming: n=12, 2-MT: n=6). The relative frequency of freezing percentages per condition was calculated by categorizing all freezing percentages per 15-second time window in bins ranging from 0-10%, to 90-100% (Bin Center). In graph 7.a-7.c, the cumulative freezing percentages of the demonstrators are compared with those of the observers. Figure 7.d and 7.e compare the cumulative freezing percentages of demonstrators and observers across the different conditions. The freezing distribution of observers (7.e) differed between the shock condition and the other two (indicated with ‘a’, p<0.0001) and between looming and 2-MT (‘b’, p=0.004). The symbol ‘***’ corresponds with a statistical significance of p<0.0001.
