Prefrontal ensembles represent social defensive synchrony
Data files
May 18, 2026 version files 115.95 MB
-
f5_output.mat
32.21 MB
-
f7_output.mat
16.02 MB
-
m1_output.mat
18.06 MB
-
m2_output.mat
20.87 MB
-
m3_output.mat
13.94 MB
-
m4_output.mat
14.86 MB
-
README.md
2.69 KB
Abstract
Social buffering of fear is a process in which the presence of other members of the same species (conspecifics) decreases stress and fear reactions during threat exposure. In parallel, prior reports showed that groups of individuals show higher synchrony in their actions than expected by chance in non-threatening situations. However, it is unknown if individuals also synchronize defensive actions when facing threats. We first show social buffering of freezing in mice of both sexes during exposure to both innate and conditioned threats (a live predator and a shock grid, respectively). Furthermore, mice displayed synchronized freezing when exposed to a predator or a shock grid. This result suggests that freezing is not merely an individual behavior, but rather can be a strategy to coordinate defensive strategies across a group. The presence of freezing synchrony implies the existence of a neural representation of conspecific freezing. We reasoned that this neural correlate might be found in the medial prefrontal cortex (mPFC), as this region has been implicated in social behaviors, mirroring and freezing. Strikingly, mPFC neural activity represented the average level of freezing across conspecifics. Representation of self and conspecific freezing was orthogonal, indicating the presence of specialized circuits dedicated to processing information of social defensive cues. Indeed, prefrontal ensemble activity scaled linearly with the number of conspecific mice freezing. These data show, for the first time, neural representation of defensive actions of other individuals, and suggest that the mPFC may coordinate behavioral responses to threat across conspecifics.
Dataset DOI: 10.5061/dryad.wdbrv164n
Description of the data and file structure
Mice were exposed to a live rat during 12 minutes as shown in the scheme of figure 6a of this paper https://elifesciences.org/articles/77115
The difference from that scheme is that 4 cagemate mice were exposed to the rat. mPFC miniscope recordings were obtained from 1 mouse in each session. The data from each mouse is in a separate .mat file. m1,m2,m3 are male mice, and f5, f7 are female mice. For example, the file f5_output.mat has the data from female cage 5, along with 3 female cagemates during rat exposure. Inside every /mat file, the mouse with the implanted miniscope is mouse1. Mice 2, 3 and 4 in the same session are cage mates without miniscopes exposed to the same rat in the same session as mouse 1.
The sampling frequency is 15 Hz for all behavioral and neural data in each .mat file. Speeds are in cm/s, distances and x and y positions are in cm
%======descriptions of all variables in the .mat files below:
Each .mat contains variables x1-4, where xn is the position of mouse n (for example x1 is x position of mouse 1 in cm).
y1-4 are y positions for all 4 mice in cm
xrat and yrat are the x,y position of the rat
freeze1-4 are the freeze vectors for each mouse
z scored neural activity is in df(1 row per neuron, one column per time point)
%======GLM variables. For each cell, a GLM was performed using the predictor variables listed below.
GLM variable 1:rat speed
GLM variable 2: average x position of mice 2,3 and 4
GLM variable 3: x position of mouse 1
GLM variable 4: speed of mouse 1
GLM variable 5: average speed of mice 2 ,3 and 4
GLM variable 6: freeze of mouse 1
GLM variable 7: average freeze of mice 2, 3, and 4
glm_coefs contain the weights for each cell for the predictors (one row per cell, one column per GLM variable in the order).
glm_pval2 contain the GLM p values for each cell for the predictors (one row per cell, one column per GLM variable in the order).