Dynamics in brain activation and behaviour in acute and repeated social defensive motivated behaviour

In nature, confrontations between conspeciﬁcs are recurrent and related, in general, to the lack of resources such as food and territory. In this sense, adequate defence against a conspeciﬁc aggressor is essential for the individual’s survival and the group integrity. However, repeated social defeat is a signiﬁcant stressor, promoting several behavioural changes, including on social defence per se. But what would be the neural basis of these behavioural changes? To explore some hypotheses about this, we investigated the effects of repeated social stress on neural circuits underlying the motivated behaviour social defence in male mice. The hypothalamus is an essential centre of these circuits. Different hypothalamic structures receive information about the conspeciﬁc from the medial amygdala and the bed nucleus of the terminal stria. Furthermore, the hypothalamus can receive environmental information via the septo-hippocampal-hypothalamic circuit. Both information is processed by the dorsal premammillary nucleus (PMD) and the ventrolateral portion of the ventromedial nucleus of the hypothalamus, which communicate with the periaqueductal grey, an important downstream site for behavioural emission. During our analysis, we observed that animals re-exposed three times to the aggressor spent more time in passive defence during their last exposure than in their ﬁrst one. These animals also present a smaller mobilization of areas related to the processing of conspeciﬁc cues. In contrast, we did not observe changes in the PMD mobilization. Therefore, our data indicate that the balance between the activity of circuits related to conspeciﬁc processing and the PMD determines the pattern of social defence behaviour. Changes in this balance may be the basis of the adaptations in social defence after repeated social defeat. show the participation of these regions as an integrator centre for regulation of the hypothalamic-pituitary- This regulates corticosterone production, a that mediates the endocrine response to in activity to increase in aversive social


Introduction
Conflicts between animals of the same species are frequent during the life of social animals [1][2][3] . In general, these conflicts 2 occur due to a lack of resources (water, food, territory) or sexual partners, and their occurrence divides animals between 3 dominants -those who were often the winners -and subordinates [1][2][3] . Submission is a relevant stressor in groups of social 4 animals. Subordinate animals have a shorter life expectancy, a lower success rate in mating and greater weight loss than their 5 dominant ones 1,2 . Furthermore, these animals show a series of neuroendocrine and autonomic responses associated with 6 stress, such as changes in blood corticosterone levels and in heart rate 1,2,4 . In the laboratory, the exposure -acute or chronic - 7 of an animal to a conspecific aggressor is one of the most used protocols for the study of coping and psychopathology-like 8 behaviours 4-7 . 9 Chronic exposure to social stress promotes a series of changes in the subordinate animal behaviour 6 . In short, animals begin 10 to express what is known as anxious-and depressive-like behaviours, which is related to changes in the pattern of locomotion 11 and exploration in paradigms such as the open field and the elevated plus-maze 4,6,8 . Coping strategies are also changed. Butler as predatory and social stress, which would be expected since the environmental context is relevant to coping with general 48 threats 15,23,25,26 . 49 To study how reexposure to social stress impacts social defence as a goal-oriented behaviour, and to evaluate the possible 50 neural bases of these changes, we have submitted C57Bl/6 male mice to one or three sessions of social stress. We also have 51 investigated if, like rats, a septo-hippocampal-hypothalamic circuit is responsive to two different stressful contexts: social and 52 restraint stress. Our analysis suggests that animals submitted to three social defeat sessions spent more time in passive defence 53 during the last exposure. This is followed by a decrease in the mobilization of brain areas that process social cues. Also, there 54 is no evidence of a difference in the mobilization of areas related to the septo-hippocampal-hypothalamic circuit during the first 55 and last exposure. Our study suggests a broad mobilization of this circuit in aversive contexts since it is also mobilized during 56 restraint stress. Our work provides insightful data for future studies about the neural bases of social defence and changes in 57 defensive behaviour under repeated defeated male mice. 58

59
Behavioural effects of reexposure to the resident-intruder paradigm in social defence 60 Intruders and residents behavioural pattern during the resident-intruder paradigm 61 Two groups of C57Bl/6 mice were submitted to the resident-intruder paradigm, one group to only one exposure (n = 7; 1 62 Exp. group) and another group to three exposures (n = 11; 3 Exp. group), for three days, once a day, and with different 63 residents. During the sessions, the resident's attacks began almost immediately after the intruder was placed and, after the 64 first attack, intruders were left for 5 min with the dominant male. Most of the time, the intruder animals expressed defensive 65 behaviours, which were classified as active defence -when the animal is under attack by the resident -or passive defence - 66 when the resident leaves the intruder alone ( Fig. 1; Supplementary Information Table S1). In this case, the most common 67 posture performed, classified as passive defence, was freezing, followed by checking (turning to maintain intruder's orientation 68 toward the resident) (Supplementary Information Table S1). As regards the active defence, intruders generally performed a 69 defensive upright position (standing under the hind limbs and towards the resident) (Supplementary Information Table S1). 70 Flight and jumping were also very common during the paradigm (Supplementary Information Table S1). In general, intruders 71 spent almost 20% of the time in active defence, which in all cases, was lower than the time spent in passive defence ( Fig. 1; 72 Supplementary Information Table S1). During the first exposures to the paradigm, mice intruders spent nearly 25% of the time 73 in exploring the resident's home cage ( Fig. 1; Supplementary Information Table S1). As to resident's behaviours, lateral threat 74 (moving laterally to the intruders, via an arc-like path) was the most common performed behaviour (Supplementary Information   75   Table S3). Move towards (moving in a straight line toward the intruder) and clinch attack (direct attacks against the intruder) 76 was also very frequent (Supplementary Information Table S3). 77

2/15
Intruder animals spend more time in passive defence and less time in exploration during last exposure 78 The behavioural pattern of the intruder animal changes dramatically between the first and third exposure. On Table S2). The short confidence intervals (CIs), and mostly restricted 83 to high standardized effect size values, suggest that there is a large and robust effect of reexposure under the behavioural 84 pattern of the intruder (Supplementary Information Table S2). During the first exposure, the intruder starts the session with 85 a pattern of exploratory behaviour interspersed with confrontational events between intruder and resident, whose frequency 86 decreases throughout the session, while in the last exposure intruder behaviour already starts predominantly defensive (Fig. 2). 87 On the other hand, the behavioural pattern of residents has not substantially changed between the first and third sessions of 88 the paradigm, which has also occurred with the intruder's time spent on active defence (Fig. 1 Fig. S1 and Tables S1, S3). 92 Overall, despite variations in the behaviour pattern of intruders, exploratory analyses do not provide clear evidence of 93 a variation in the spatio-temporal measurements between the first and third exposure to the paradigm, which also occurred 94 with residents ( Fig. 1; Supplementary Information Table S2, S4). The CIs are long and, for the most part, are consistent with 95 anywhere from large reduction, no change up to a large increase in times or distances evaluated. The only exception is when 96 comparing the distance covered by the intruders submitted to 3 exposures of the paradigm during the first and last exposures of   Table S2). In this 99 case, the standardized effect sizes suggest a moderate difference between these groups, although the CIs are consistent with no 100 difference up to a large reduction of the covered distance in the third day of exposure. In short, intruder and resident animals 101 spent most of the time at the border of the resident's home cage and, in general, the resident had a greater distance travelled 102 than the intruder during the sessions (Fig. 1, Supplementary Information Fig. S1 and Tables S1, S3).

103
Effects of reexposure to the resident-intruder paradigm in brain activation pattern 104 We analysed the activation of some areas of the brain in control animals (n = 12), animals exposed once to the resident-intruder 105 paradigm (n = 7) and animals exposed to this paradigm three times (n = 11). It was quantified the density of Fos-labelled cells 106 in 31 brain regions, which includes septal, amygdalar, hypothalamic and periaqueductal grey sites. Only the caudodorsal part 107 of the lateral septum (LSc) and the dorsomedial portion of the ventromedial hypothalamic nucleus (rVMHdm) did not show 108 differences in the Fos-labelled cell density between the three groups (Fig. 5; Supplementary Information Table S5, S6). In 109 general, these CIs are large and include moderate positive and negative values of the standardized effect size and does not 110 provide clear evidence of differences between these groups. In all other analysed sites, exploratory analyses provide clear 111 evidence that animals exposed one or three times to social stress showed considerably greater activation compared to the control 112 group (Fig. 5; Supplementary Information Table S5, S6).

113
Circuits that process social information are less mobilized during last exposure 114 When comparing animals that have gone through 1 or 3 exposures of the resident-intruder paradigm, the more robust and 115 clear differences in the brain activation pattern between these groups are in regions related to the processing of conspecific  About the rVMHvl, cVMHvl and PMV, the CIs are large but restricted to moderate up to high standardized effect size values, 125 indicating that the effect of reexposure on activation of these areas is probably no less than moderate in size. The standardized 126 effect size of the difference in TU is moderate, but the CI is long and consistent with a small negative effect size value up to a 127 large positive effect size value, suggesting that the effect is, most likely, no more than moderate in size. Regarding the MPN,

3/15
The mobilization of septo-hippocampal-hypothalamic circuit between first and third exposure is not different 132 Concerning the analysed nuclei of the septo-hippocampal-hypothalamic circuit (rostroventral part of the lateral septum -LSr, 133 the juxtaparaventricular part of the lateral hypothalamic area -LHAjp, and LHAjd), effect size analyses do not suggest clear 134 evidence of a difference in the mobilization of these nuclei between animals submitted to a one or three exposure of social  Table S5, S6). Finally, the small standardized effect size and the large CI suggest that there is no difference in the Fos expression 140 of the rVMHdm between these two groups (Fig. 3, 5; Supplementary Information Table S5, S6). It is important to note that the 141 rVMHdm does not participate in the response to the social cues, but the predatory cues 27 . Thus, the absence of difference in 142 neural activation between these two groups suggests that there is no baseline reduction in Fos expression due to reexposure to 143 the resident-intruder paradigm.  To explore the relationship between different evaluated regions and analysed behaviours, we calculated Pearson's r and its 95% 156 CI among these variables (Supplementary Data 1 and 2). We performed these analyses with the data from animals submitted to 157 one or three exposures of the resident-intruder paradigm, separately. Interestingly, the correlation pattern changes between the 158 two groups.  Correlations between brain activation and behaviour 181 Regarding the portions of the MeA, when we analysed data from animals submitted to only one exposure to the resident-intruder 182 paradigm, we were not able to observe a clear correlation with any behavioural variable or other brain areas mobilization  We also observed a negative correlation between passive defence and cPAGvl activity, however with a moderate coefficient 220 and a large 95% CI consistent with a lack of correlation with a strong negative correlation (r = -0.62, 95% CI [-0.89, -0.03]).

221
About the cPAGl, there is a moderate positive correlation between this column and the rVMHvl and LHAjd mobilization. In  To investigate if the neural circuits mobilized during social defeat are also mobilized by other aversive situations, we compared 226 the activation pattern between a group of animals submitted to one exposure of the resident-intruder paradigm (n = 7) and one 227 group of animals submitted to restraint stress (n = 7). As the restraint stress is characterized by the limitation of the animal 228 movement, the comparison between these types of stress can suggest the circuits that would be behind the signal of entrapment 229 imposed by the aggressor during the social defeat. As before, most of the time, intruder animals expressed defensive behaviours

5/15
The septo-hippocampal-hypothalamic circuit is also mobilized during restraint stress 239 About the nuclei of the septo-hippocampal-hypothalamic circuit, exploratory analyses do not suggest a clear difference in the 240 mobilization of LHAjp and LHAjd between animals submitted to social stress or restraint stress (Fig 4, 6; Supplementary 241 Information Table S7, S8). In these cases, the standardized effect sizes are small and the CIs are consistent with anywhere from 242 negative difference, no difference up to positive difference between the groups. The same was observed with the PMD, the 243 rVMHdm and hypothalamic nuclei that participate in neuroendocrine and autonomic responses to stress (Fig 4, 6; Supplementary 244 Information Table S7, S8). The only exception is the anterior part of the dorsomedial nucleus of the hypothalamus (DMHa), 245 whose standardized effect size is high and the CI is restricted to high up to small negative effect size values, which suggest 246 that the neural activation of this area is greater in restrained animals than social subjugated animals (Fig 4, 6 Table S7, S8). The effect size related to rPAGdl is small and the CI is consistent with anywhere from negative difference, no 256 difference up to positive difference between these groups. Concerning cPAGdl, the effect size is moderate and suggests that this 257 area is slightly more mobilized in social stress than restraint stress. However, the CI is large and consistent with small negative  During acute social stress, the mobilized structures in mice are similar to those observed in rats, although the behavioural 271 pattern is different between these animals 9, 15, 23 . While rats rarely perform exploratory behaviours, mice, during their first 272 exposure, spend a substantial part of their time exploring the environment 9,15 . After reexposure to the paradigm, the intruder's 273 time spent in exploration decreases and the relation between exploratory and defensive behaviour becomes more similar to the 274 rat's relation 9,15 . Interestingly, during acute social stress, the PMD is relatively more mobilized than the VMHvl in rats 9, 15 , 275 while in mice it is the opposite. Since the PMD is essential in passive defence and the VMHvl is associated with active defence 276 and social exploration 9,28,29 , this brought us to the idea that, in mice, the behavioural variation after reexposure to social stress 277 (balance between exploration and passive and active defence) would be related to the balance in the mobilization of these neural 278 structures. Our data corroborate this idea since, during the last exposure to social stress, the intensity of mobilization of the 279 circuits underlying the social defence becomes closer to what is observed in rats.

280
As expected, the nuclei of the sexually dimorphic circuit are mobilized during a social confrontation. Different works 281 with rats and mice suggest the participation of this circuit in many types of social behaviour as fighting, parenting and mating 282 behaviour 9, 20, 21, 30-36 . These nuclei are also very important for social memory and social recognition 37,38 . In the present work, 283 VMHvl and PMV were the most mobilized regions during social stress. Concerning VMHvl, some works indicate its role in 284 defensive behaviour, especially active defence 28 . This region is also related to social fear and the VMHvl inhibition impairs 285 fear response to a social aversive stimulus 39,40 . The PMV is activated by male and female olfactory cues and their inhibition 286 impairs both maternal and inter-male attacks 20,21,41 . Their activity during the social confrontation can be due mainly to the 287 social representation of the resident and the execution of active defence behaviours 20,28,38,42 . Our data suggest a reduction in 288 the mobilization of these areas after the reexposure to the resident-intruder paradigm. This reduction in the sexually dimorphic

292
While some neurons are related to social exploration, other neurons participate mainly in active defence behaviour 3, 29, 42 . 293 Therefore, the mobilization of VMHvl may depend not only on active defence. Another possibility -which may be occurring 294 concurrently -is that the active defence can be defined not only on the number of neurons mobilized in the sexually dimorphic 295 circuit but also on the variation in the frequency or intensity of neural activation of these areas.

296
One relevant region upstream of the sexually dimorphic circuit is the MeA 12, 43 . Its nuclei receive many direct and indirect 297 projections from main and accessory olfactory systems and process the social information from olfactory cues 16,44 . The MeA 298 participates in the neural representation of social conspecific and its activity is related to sexual, defensive and aggressive 299 behaviours both in mice and rats 9,16,17,45,46 . The activation of gabaergic MeA neurons promotes aggressive and sexual 300 behaviours, depending on the intensity of activation 47 . As the activity of the sexually dimorphic circuit is also necessary for 301 these behaviours, it has been proposed that the role of MeA occurs by a disinhibitory circuit 42,47 . This idea agrees with our 302 data since we observed an increase in Fos-labelled neuron density in MeA and sexually dimorphic circuit during social stress.

303
After reexposure to social stress, there was a reduction of mobilization of MeA nuclei, which may be a factor that leads to a 304 reduction of activity of the sexually dimorphic circuit. However, there is not a clear correlation between the mobilization of 305 this circuit and MeA mobilization. These data suggest a more complex relationship between these regions, which also may be 306 related to direct and indirect pathways between MeA and sexually dimorphic circuit 12,43 .

307
Another region that can participate in the processing of social information is the BST 19,20,45,48,49,52 . Its activity is 308 related to different social behaviours such as social defence, inter-male aggression, maternal aggression and parental be-309 haviours 19,20,45,[48][49][50]52 . Neuroanatomic evidence suggests that the BST participates in the disinhibitory circuit between MeA 310 and sexually dimorphic circuit 12,43,48 . If the BST activity was just dependent on this disinhibitory circuit, it would be expected 311 an increase in BST mobilization in animals of the reexposure group, since in these animals there was a decrease in MeA 312 activation. However, our data do not suggest a clear difference in BST mobilization between animals submitted to one or 313 three sessions of the resident-intruder paradigm. As the MeA, BST neurons also receive direct and indirect projections from 314 main and accessory olfactory systems, which also can influence the BST activity during social defeat 19,51 . The BST activity is 315 important to the expression of conditioned defeat, which is related to defensive behaviours 52 . So, the BST may be relevant to 316 the increase in time spent in passive defence during the last exposure.

317
In addition to social information, spatial information is essential to the organization of defensive strategy not only for social 318 confrontation but also in many aversive situations 22,53 . The distance of the threat and the existence of an escape route or safe 319 place can impact the behaviour to be developed 22,53 . In this sense, we believe that the LHAjd is important for the processing of 320 environmental cues. The LHAjd is mobilized during the first and the last exposure to social stress and is also mobilized during 321 restraint stress, both in mice, as we observed in the present work, and rats 15 . Since the restraint stress is characterized only 322 by the presence of a physical limit that prevents animal movement and, in social stress, the conspecific represents a physical 323 and psychological limit, the LHAjd may be processing the entrapment component associated with both contexts 15,23 . Some 324 functional studies also highlight the role of the LHAjd in defensive behaviour. Rats with lesioned LHAjd perform less risk 325 assessment during post-encounter context associated to a conspecific 23 . Also, the inhibition of LHA promotes a deficit in escape 326 behaviour in response to predatory and physical threats [54][55][56][57] . Thus, a septo-hippocampal-hypothalamic circuit, that comprises 327 the LHAjd, is expected to participate in the processing of spatial information during many aversive contexts 15,23,25,26,58 . This 328 circuit also comprises the two major sources of afferent to LHAjd: the subiculum and the lateral septum 59 . The subiculum 329 detains many border-vector cells, which participate in the processing of environmental limits 60 . Neuroanatomic evidence also 330 suggests that this structure participates as an intermediate of information present in the hippocampal system to cortical and 331 subcortical structures 61,62 . The lateral septum is the structure that most receives projections from CA1 and CA3 hippocampal 332 regions 63 . Recent works show the role of the lateral septum as a decoder of the cognitive map from the hippocampus to 333 downstream regions 64 . These data suggest a broader role for LHAjd in defence by participating in the spatial information 334 process.

335
The integration of spatial and social information is relevant to social defence. The PMD seems to be an important region 336 for the integration of threat cues (social, predatory and physical) and spatial information 9, 15, 25, 26, 65-68 . As LHAjd, the PMD 337 is mobilized during social and restraint stress. This nucleus receives strong projections from LHAjd and, therefore, may be 338 receiving environmental-related information from the septo-hippocampal-hypothalamic circuit during aversive situations 15, 59 . 339 Furthermore, the PMD activity is essential to the passive defence behaviour 9 . However, there is not a clear difference between decreased during the last exposure to social stress, which agrees with a presumed role of LHAsf in the communication between 347 sexually dimorphic nuclei and PMD 59, 65, 69 . 348 For the development of this defensive strategy, the PAG is an important effector region that receives projections directly from 349 hypothalamic nuclei, like PMD, VMHvl and LHA 9,28,55,68 . The PAG location suggests its role in mediating the information 350 output from prosencephalic structures to the brainstem and spinal cord regions 70 . Also, PAG participates in the regulation of 351 primary emotional responses by projecting information to prosencephalic structures 70 . PAG columns (especially rPAGdm 352 and rPAGl) are mobilized during both social and restraint stress. These columns receive projections from PMD and LHAjd 353 and may be the effector sites from the septo-hippocampal-hypothalamic circuit since they are also mobilized during other 354 aversive situations as predatory encounter and foot shock 15,23,25,26,58 . Also, the mobilization of PAGdm,l was similar between 355 the first and last exposure to social stress. These columns also receive projections from VMHvl and have subpopulations 356 associated with escape or freezing behaviours 55,68,[71][72][73] . Therefore, these PAG columns may be important to the variation in 357 social defence strategy, which could be related to the relationship between the activity of PMD and VMHvl. PAG columns also 358 receive projections from the medial prefrontal cortex (mPFC) and the inhibition of mPFC-PAGd projections mimics behavioural 359 alterations after a social defeat protocol 74 . So, this circuit may also be relevant for the variation in the social defence strategy 360 during reexposure to the resident intruder paradigm.

361
The behavioural response to an aversive context is accompanied by neuroendocrine and autonomic responses. The

364
Previous works show the participation of these regions as an integrator centre for regulation of the hypothalamic-pituitary-365 adrenal axis 75,79 . This pathway regulates corticosterone production, a hormone that mediates the endocrine response to stress 79 .

366
The increase in DMH activity is also related to an increase in cardiac frequency, which is common in aversive situations, 367 including social defeat and restraint stress 78, 80-82 . 368 Taken together, our data indicate a change in the mobilization pattern of neural circuits underlying social defence during 369 the last exposure. This change is characterized by a decrease in the mobilization of structures that process social information, 370 namely MeA, VMHvl and PMV. We speculate that some mechanism, which involves memory recovery about the social stress  84 . In this sense, at the beginning of the last 379 exposure to the resident-intruder paradigm, the BST could be participating in an anticipation system for the social stress that 380 will be suffered by the animal, based on past experiences. Thus, BST could interfere with the balance between the conspecific 381 responsive circuit and the PMD activity, which promotes a pattern of social defence predominantly passive. Indeed, more 382 studies are needed to elucidate this anticipation system that can act on neural circuits underlying social defence and orchestrate 383 this behaviour.

384
In summary, the reexposure to social stress promotes changes in the social defence behavioural pattern of C57Bl/6 male 385 mice, which is accompanied by changes in the neural mobilization of circuits related to the expression of this goal-oriented 386 behaviour. Our data are following previous works that have shown that social defence behaviour depends on the mobilization 387 of neural circuits that process conspecific cues and environmental cues. Furthermore, our data suggest that the pattern of 388 social defence (e.g. more passive or more active) is determined by the balance between the mobilization of circuits related 389 to the processing of the conspecific information (in particular the VMHvl nucleus) and the PMD activity (which can receive 390 environmental information from the septo-hippocampal-hypothalamic circuit). Further studies are necessary to investigate the 391 mechanisms by which repeated social stress affects the mobilization of these circuits and impacts social defence behaviour.  405 Animals submitted to the resident-intruder paradigm were separated into two groups. One group was submitted to only 406 one session of the resident-intruder paradigm (n = 16). The other group was submitted to three sessions of the resident-407 intruder paradigm, performed for three days, once a day, with different residents (n = 11). During each session, as previously 408 described 9, 15 , the Swiss female was removed and a C57Bl/6 mouse was placed in the dominant home cage. Animals were 409 separated 5 min after the first resident attack and the intruders were returned to their home cage. For a control group (n = 12),

462
We quantified the density of Fos-labeled cells in 31 brain regions that followed the Allen Mouse Brain Atlas and The  The objective of the analysis was to explore differences in neural activation and behavioural pattern between mice submitted 480 once or three times to the resident-intruder paradigm and to compare neural activation between mice submitted to social and 481 restraint stress. We expected to generate new hypotheses of how different neural circuits are involved with the expression of 482 social defence and how these circuits are mobilized in response to different types of stress.

483
All data are expressed as the mean ± sample standard deviation ("s"), and "n" is the sample size. We also estimated the 484 95% confidence interval (95% CI) of the mean. To compare different experimental groups, we calculated the difference 485 between means (M diff ) and estimated the 95% CI for each M diff . We also calculated the unbiased Cohen's d (d unb ) and its 486 respective 95% CI for each comparison 98,99 . To measure the correlation between variables, we calculated Pearson's r and its 487 95% CI 98 . These analyses were conducted using the esci v0.9.1 library from jamovi v1. 2 Figure 1. Behavioural analysis of intruder animals submitted to the resident-intruder paradigm. A) Behavioural pattern of intruder animals submitted to one exposure of the resident-intruder paradigm or three exposures, during the first and third days. B) Time spent by intruder mice on the border and centre of the resident's home cage during paradigm sessions. C) Total distance covered and distance covered in the centre and the border by the intruders during the paradigm sessions. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method 100 . The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the first and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the first and third quartiles, respectively.    Rostral portion of the periaqueductal grey matter and H) caudal portion. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method 100 . The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the first and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the first and third quartiles, respectively. Figure 6. Pattern of neural activation in animals submitted to social or restraint stress. A) Septum; B) Medial amygdala; C) Paraventricular nucleus; D) Sexually dimorphic circuit of the hypothalamus; E) Lateral hypothalamic area; F) Dorsomedial hypothalamic nucleus and dorsal premammillary nucleus; G) Rostral portion of the periaqueductal grey matter and H) caudal portion. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method 100 . The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the first and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the first and third quartiles, respectively. Figure 1 Behavioural analysis of intruder animals submitted to the resident-intruder paradigm. A) Behavioural pattern of intruder animals submitted to one exposure of the resident-intruder paradigm or three exposures, during the rst and third days. B) Time spent by intruder mice on the border and centre of the resident's home cage during paradigm sessions. C) Total distance covered and distance covered in the centre and the border by the intruders during the paradigm sessions. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method100. The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the rst and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the rst and third quartiles, respectively.    The kernel is based on the Gaussian function. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method100. The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the rst and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the rst and third quartiles, respectively.

Figure 6
Pattern of neural activation in animals submitted to social or restraint stress. A) Septum; B) Medial amygdala; C) Paraventricular nucleus; D) Sexually dimorphic circuit of the hypothalamus; E) Lateral hypothalamic area; F) Dorsomedial hypothalamic nucleus and dorsal premammillary nucleus; G) Rostral portion of the periaqueductal grey matter and H) caudal portion. The data is represented as a combination of the kernel density estimate and the boxplot. The kernel is based on the Gaussian function. The scale factor to use when computing the kernel bandwidth was determined by the Scott method100. The actual kernel size was determined by multiplying the scale factor by the standard deviation of the data within each bin. In the end, we chose to preserve an equal width between each violin. The white circle in the centre of the boxplots represents the median, the black rectangle involves values within the rst and third quartiles and whiskers go up to 1.5 times the interquartile range below or above the rst and third quartiles, respectively.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryInformation.pdf