Fig 1.
Emotional contagion and social buffering paradigms.
(A) A schematic representation of typical paradigms used to investigate emotional contagion. An observer rat witnesses a demonstrator rat receive an electric foot shock. The shock induces fear and pain responses in the demonstrator, which in turn is unidirectionally transferred to the observer, which shows increased freezing thought to indicate an increase in distress. In these paradigms, the variable of interest is the amount of freezing of the observer. (B) A schematic representation of the social buffering paradigm. A demonstrator rat receives an electric foot shock. The fear response of the demonstrator, freezing in particular, is reduced in the presence of an observer rat. The variable of interest is the amount of freezing of the demonstrator.
Fig 2.
The procedure started with a familiarization phase (left) in which a DEM (in orange) was housed together with an OBS (in blue) for different periods of time depending on the experimental group (rightmost table). After the familiarization phase, the OBSs were preexposed to foot shocks alone (middle). The preexposure procedure consisted of a 12-minute baseline and a 12-minute shock period in which the observer received four foot shocks (0.8 mA, 1 second each, ISI: 4–6 minutes). This was followed by the interaction test (right), consisting of a 12-minute baseline and a 12-minute shock period. During the shock period, the observer witnessed the demonstrator in an adjacent chamber receive five foot shocks (1.5 mA, 1 second each, ISI: 2–3 minutes). In the individual familiarity experiment, all animals were LE (hooded rats in the right table) and varied in how familiar they were with the particular individual they observe in the interaction test. In the strain familiarity experiment, the observer–demonstrator dyads were either from the same strain (i.e., both hooded LE or both albino SD) or from different strains (i.e., one hooded LE and one albino SD), and thus varied in their familiarity with the strain they observe. DEM, demonstrator rat; ISI, intershock interval; LE, Long-Evans; OBS, observer rat; SD, Sprague Dawley.
Fig 3.
Observer and demonstrator freezing.
For both the individual familiarity and the strain familiarity experiments, the scatter plots indicate the prop. of time spent freezing by the demonstrator (x-axis) and observer (y-axis) during both baseline (black dots) and the shock period (red pluses). The marginal histograms indicate the distribution of freezing behavior during baseline (black lines) and the shock period (red lines) using a kernel with 0.05 bandwidth. Red arrow: possible influence of the demonstrator freezing on the observer freezing (akin to emotional contagion). Green arrow: possible influence of the observer freezing on the demonstrator freezing (akin to social buffering if the level of the observer freezing is lower than that of the demonstrator). Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. prop., proportion.
Fig 4.
Variables included in the Bayesian models.
Several models were built, separately for the individual and strain familiarity experiments, based on the factors that could describe the observer and demonstrator freezing. Here, the full models for the individual familiarity experiment (top) and strain familiarity experiment (bottom) are shown separately for epochs in which shocks are delivered (shock = 1) and those in which no shocks are delivered (shock = 0). The target variables that the models explain are shown in a gray box. These full models were then compared against simplified models, and the variables included in the winning model are shown in red. The modulator “Weeks together” captures whether the effect across animals depends on the number of weeks the observer and demonstrator spent together before testing (i.e., 0, 1, 3, 5 weeks). This modulator was implemented in two different ways (see Table 1A and 1B, and Fig 5): (1) in a way that models a linear increase of interindividual influence with number of weeks spent together, with the impact thus five times stronger after 5 compared with 1 week spent together, or (2) in a way that simply models a different connection weight for each group. Strain OBS and strain DEM capture the effect of a particular strain on the average freezing level of that strain. Strain (same/diff.) is a binary variable indicating whether the observer and demonstrator dyad were of the same or different strain. Finally, the variable preexposure indicates the amount of freezing of the observer during preexposure. Unfortunately, we only collected movies during preexposure in the strain familiarity experiment and thus cannot retrospectively include that variable in the models of the individual familiarity experiment. Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. DEM, demonstrator rat; diff., different; Exp, experiment; OBS, observer rat.
Table 1.
Model comparisons for the individual and strain familiarity experiments.
Fig 5.
Parameter estimates and model-free analysis.
(A) Parameter estimates of the influence from OBS → DEM from model 7 in Table 1A as a function of weeks spent together. Note the considerable overlap and shift away from 0, illustrating the lack of a familiarity effect and the consistent feedback from the observer, respectively. (B) Model-free comparison across the familiarity groups. (C, D) Same as panels A and B, but using observer freezing as the dependent variable. (E, F, G, H) Same for the strain familiarity experiment. (I) Long-Evans rats froze more than Sprague Dawleys both in a social context during the shock period of the strain familiarity experiment and when tested alone during shock preexposure. For all pairwise comparisons, t test, *p < 0.05, **p < 0.01, ***p < 0.001. n.s. refers to the absence of a significant group × epoch interaction in an ANOVA (see text for details). Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. DEM, demonstrator rat; OBS, observer rat.
Fig 6.
Same-strain recognition experiment.
For this experiment, eight observers (OBS: all Long-Evans, four of which also served as demonstrators) and eight demonstrators (DEM; four Long-Evans and four Sprague Dawley rats) were used. (A) The test was conducted in a three-chamber testing box consisting of one large central chamber (L:72 cm × W:33 cm) and two small side chambers (each: L:27 cm × W:33 cm). The central chamber was separated from the side chambers by transparent perforated walls. The day prior to testing, all animals were habituated to the testing box. The test consisted of a 5-minute baseline period, in which observers were individually placed in the central compartment, followed by a 10-minute CPP, in which two unfamiliar demonstrators (DEM1 and DEM2: one Long-Evans and one Sprague Dawley rat) were simultaneously placed in one of the side compartments (placement was randomized). To avoid bias, the placement of the demonstrators occurred when the observer was in the center zone of the central compartment. Each observer had two tests, separated by an interval of 30 minutes to 1 hour, in which the location of the Sprague Dawley and Long-Evans rats was changed. The amount of time that the observers spent in the proximal zone during the initial 90 seconds of the baseline and CPP was scored, and a ratio score was estimated (difference in the time spent in the proximal zone of the Long-Evans rat and the time spent in the proximal zone of the Sprague Dawley rat divided by the sum of the time the observer spent in the proximal zone of the Long-Evans and Sprague Dawley rat. (B) Results show that, in test 1 and 2, observers spent more time in the proximal zone of the same-strain (Long-Evans) than of the different-strain (Sprague Dawley) demonstrator rats compared with baseline. This was significant for the first test of each trio (paired samples t test, test 1: t(6) = 2.58, p2tail = 0.04; test 2: t(6) = 1.64, p2tail = 0.15, *p < 0.05). Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. CCP, choice preference period; DEM, demonstrator rat; L, length; OBS, observer rat; W, width.
Fig 7.
Directionality of information transfer.
(A) G-causality results. Mean ± SEM for G-causality values for the demonstrator-to-observer direction (DEM–OBS, in red) and the observer-to-demonstrator direction (OBS–DEM, in green) during the shock period, for the individual familiarity experiment. (B) G-causality results for the strain familiarity experiment. Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. DEM, demonstrator rat; different, different strain; Exp, experiment; G-causality, Granger causality; OBS, observer rat; same, same strain; SEM, standard error of the mean; W, week.
Fig 8.
(A) Freezing levels for nine preexposed dyads taken from the 3-week group of the individual familiarity experiment for the baseline period and following each of the five shocks. Mean and SEM are shown separately for the demonstrators (orange) and observers (blue). (B) Same for nine other dyads in which the observer rat was not preexposed to shocks prior to the interaction test. For the observers, the significance of one-tailed paired-samples t test for freezing following each shock compared with baseline is shown in blue where significant (*:punc < 0.05, **:punc < 0.01, ***:punc < 0.001). Each animal is shown as a dot. (C) Schematic of the interaction test with the two rats color coded as in panels A and B. (D) Parameter estimates from the Bayesian model calculated over the shock period. The variables used for modeling the observer freezing are shown on the left, and those for modeling the demonstrator freezing are shown on the right. Variables that improved the model are shown in red. The effect of preexposure is shown underneath each model as the posterior distribution of the parameter estimates for the connection between DEM → OBS freezing (left) and OBS → DEM (right) separately for the preexposed and nonpreexposed pairs. Note that to improve the estimate for the preexposed animals, here we included all 32 preexposed rats from the individual familiarity experiment (because weeks of familiarity did not influence this parameter significantly), whereas for the nonpreexposed pairs, we have n = 9. (E) G-causality F values as a function of direction and preexposure. **: two-tailed t test, p < 0.01. Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. DEM, demonstrator rat; G-causality, Granger causality; OBS, observer rat; SEM, standard error of the mean.
Table 2.
Model comparisons for the preexposure experiment.
Fig 9.
(A) Effect of ACC deactivation on freezing. Percentage of time the observers (in orange) and demonstrators (in blue) spent freezing during baseline (light color) and the shock period (dark color) after ACC deactivation (muscimol) or after control treatment (saline). Freezing percent = 100*freezing time/total time of the corresponding period. (B) Effect of ACC deactivation on the flow of information. Mean ± SEM of the G-causality values in the demonstrator-to-observer direction (DEM–OBS, in red) and in the observer-to-demonstrator direction (OBS–DEM, in green) during the shock period, after ACC deactivation (muscimol), or after control treatment (saline). Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. ACC, anterior cingulate cortex; DEM, demonstrator rat; G-causality, Granger causality; OBS, observer rat; SEM, standard error of the mean.
Fig 10.
Localization of the cannula tips on coronal sections.
Coordinates estimated based on Nissl stains. A24 and A33 refer to area 24 and area 33, as in the atlas and the work of Vogt [46,50]. Data can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. cc, corpus callosum; cg, cingulate gyrus; LV, lateral ventricle.
Fig 11.
Computational modeling of danger detection.
(A) Internal danger signal simulated for an animal that is exposed to 100 time points of danger and 100 time points of no danger with noise added. The animal freezes when the danger signal surpassed a certain threshold (yellow line). (B) Time series of freezing for an animal by itself (individual freezing) or for one that is additionally taking the freezing of another animal into account (socially informed freezing). (C) Accuracy of danger detection shown as the area under the ROC curve (AUC) for different coupling factors (b = 0–1.5). A higher coupling factor increases the AUC. (D) Benefit of taking the freezing of others into account when both animals have the same access to danger signals, i.e., experience the same noise level. (E-H) Animals with twice (E) or thrice (G) as much noise as compared with another animal ([F] and [H], respectively) had stronger benefits from coupling. However, the low-noise animals (F, H) experience no disadvantages. The dotted lines indicate the coupling regime that our animals appeared to be in the individual and strain familiarity experiments. Codes can be found at http://dx.doi.org/10.17632/h8fkyr2z35.1. AUC, area under the curve; ROC, receiver operating characteristic.