Threat history controls flexible escape behavior in mice

Abstract

In many instances, external sensory-evoked neuronal activity is used by the brain to select the most appropriate behavioral response. Predator-avoidance behaviors such as freezing and escape^1,2 are of particular interest since these stimulus-evoked responses are behavioral manifestations of a decision-making process that is fundamental to survival.^3,4 Over the lifespan of an individual, however, the threat value of agents in the environment is believed to undergo constant revision,⁵ and in some cases, repeated avoidance of certain stimuli may no longer be an optimal behavioral strategy.⁶ To begin to study this type of adaptive control of decision-making, we devised an experimental paradigm to probe the properties of threat escape in the laboratory mouse Mus musculus. First, we found that while robust escape to visual looming stimuli can be observed after 2 days of social isolation, mice can also rapidly learn that such stimuli are non-threatening. This learned suppression of escape (LSE) is extremely robust and can persist for weeks and is not a generalized adaptation, since flight responses to novel live prey and auditory threat stimuli in the same environmental context were maintained. We also show that LSE cannot be explained by trial number or a simple form of stimulus desensitization since it is dependent on threat-escape history. We propose that the action selection process mediating escape behavior is constantly updated by recent threat history and that LSE can be used as a robust model system to understand the neurophysiological mechanisms underlying experience-dependent decision-making.

Keywords: adaptation; decision-making; flexible; looming stimulus; mouse; suppression of escape.

Figure 1

Robust escape behavior in individually housed mice (A) Schematic showing the experimental arena containing a circular-shaped shelter and the threat zone (TZ), above which looming stimuli were presented. Looming stimuli were composed of 5 sequential expanding spots. (B) Image frames from a single trial in chronological order showing the mouse entering the threat zone (B1), turning towards the shelter after stimulus onset (B2), fleeing (B3), and finding refuge inside the shelter (B4). (C) A single example trace of the position of the mouse from the trial shown in (B). (D) Population data showing positional traces of all trials from all mice housed in IVCs for between 2–7 days (traces are classified as escape [solid line], freezing [dashed line], or no reaction [solid grey line]). (E) Population data showing positional traces of all trials from all mice housed in IVCs for at least 28 days. (F) Overall percentage of trials classified as escape according to IVC housing period. (G) Peak speed of escape plotted for all trials where escape responses were observed in mice singly housed for either 2–7 days (black, n = 44 mice) or more than 28 days (purple, n = 20 mice). Horizontal lines indicate the median values for each group. (H) Left: percentage of escape trials plotted against the number of looming spots presented prior to escape for all mice isolated for between 2–7 days versus all mice group housed in the enriched pen. Only trials that were classified as escapes are shown. Right: percentage of all trials that were classified as no escape and includes freezing trials.

Figure 2

Rapid and prolonged escape suppression in loom-naive mice (A) Schematic of the stimulus protocol used to elicit learned suppression of escape (LSE). (B) Population data showing positional traces for test trials performed after the LSE protocol. (C) Top: average heading direction (20 frames prior to stimulus onset, grey dashed line) and mouse position (circles) at the onset of stimulus presentation for each mouse and each trial. Minimum heading angle relative to shelter recorded following stimulus onset (STAR Methods) is represented by the “arrow.” Solid lines with an arrowhead indicate escape trials. Filled circles indicate a freezing response. Bottom: post-stimulus trajectories (from above) for all mice and all trials plotted relative to shelter location. (D) Population average velocity traces (solid lines) and standard deviation (shaded area) for naive and LSE mice. (E) The percentage of trials classified as escape, grouped by most recent looming spot preceding escape initiation (left) and the percentage of responses classified as freezing (middle) and no reaction (right) for each experimental group. (F) The probability of escape suppression in mice at various time points following the LSE protocol (brown circles) with a line of best fit (dashed brown line). Each mouse (n = 53) was tested at a single time point only. The dashed green line indicates non-escape probability of naive control mice.

Figure 3

LSE is stimulus specific and dependent on recent threat history (A) Left: schematic of the experimental paradigm in which a mouse was free to interact with a novel cricket tethered to the center of the threat zone (grey circle indicates the upper limit of the range of cricket at retreat onset). An example trace of a retreat to shelter is represented by the orange line, starting from the moment the mouse enters the regional limit of the tether point (orange circle). Right: positional traces for all mice and all bouts exhibiting returns into or near to shelter for each experimental group. Small grey circles indicated the position of the cricket at the onset of retreat for each bout. (B) Top: raster plots of the onset times of all retreats for naive (green, n = 5) and LSE (brown, n = 6) mice. Middle: histograms of retreat onset times scaled to the first 4 min time bin. Bottom: line plots of the fraction of all bouts that result in a retreat, over time. (C) Bar plot of escape probability to auditory stimuli in loom-naive (green) and LSE (brown) mice. (D) Schematic showing the experimental paradigm for assessing the effect of prior exposure to a pre-test at either <0.2 h (orange) or 24–26 h (blue) before LSE and the subsequent total number of stimulus exposures (right). (E) Pre-test escape probability of mice tested <0.2 h before or 24–26 h before LSE. (F) Post-LSE escape probability for the <0.2 h and 24–26 h, and no pre-test experimental groups.∗∗, ∗∗∗ indicate p values of less than 0.01 and 0.001, respectively.