Escaping from imminent danger is an instinctive behaviour that is fundamental for survival, and requires the classification of sensory stimuli as harmless or threatening. The absence of threat enables animals to forage for essential resources, but as the level of threat and potential for harm increases, they have to decide whether or not to seek safety . Despite previous work on instinctive defensive behaviours in rodents, little is known about how the brain computes the threat level for initiating  escape. Here we show that the probability and vigour of escape in mice scale with the saliency of innate threats, and are well described by a model that computes the distance between the threat level and an escape threshold. Calcium imaging and optogenetics in the midbrain of freely behaving mice show that the activity of excitatory neurons in the deep layers of the medial superior colliculus (mSC) represents the saliency of the threat stimulus and is predictive of escape, whereas glutamatergic neurons of the dorsal periaqueductal grey (dPAG) encode exclusively the choice to escape and control escape vigour. We demonstrate a feed-forward monosynaptic excitatory connection from mSC to dPAG neurons, which is weak and unreliable-yet required for escape behaviour-and provides a synaptic threshold for dPAG activation and the initiation of escape. This threshold can be overcome by high mSC network activity because of short-term synaptic facilitation and recurrent excitation within the mSC, which amplifies and sustains synaptic drive to the dPAG. Therefore, dPAG glutamatergic neurons compute escape decisions and escape vigour using a synaptic mechanism to  threshold threat information received from the mSC, and provide a biophysical model of how the brain performs a critical behavioural computation.

8Escaping from imminent danger is an instinctive behaviour fundamental for survival that requires classifying 9 sensory stimuli as harmless or threatening. The absence of threat allows animals to forage for essential resources, 10 but as the level of threat and potential for harm increases, they have to decide whether or not to seek safety 1 . 11Despite previous work on instinctive defensive behaviours in rodents 2-13 , little is known about how the brain 12 computes the threat level for initiating escape. Here we show that the probability and vigour of escape in mice 13 scale with the intensity of innate threats, and are well described by a theoretical model that computes the distance 14 between threat level and an escape threshold. Calcium imaging and optogenetics in the midbrain of freely behaving 15 mice show that the activity of excitatory VGluT2 + neurons in the deep layers of the medial superior colliculus 16 (mSC) represents the threat stimulus intensity and is predictive of escape, whereas dorsal periaqueductal gray 17 (dPAG) VGluT2 + neurons encode exclusively the escape choice and control escape vigour. We demonstrate a feed-18 forward monosynaptic excitatory connection from mSC to dPAG neurons that is weak and unreliable, yet 19 necessary for escape behaviour, and which we suggest provides a synaptic threshold for dPAG activation and the 20 initiation of escape. This threshold can be overcome by high mSC network activity because of short-term synaptic 21 facilitation and recurrent excitation within the mSC, which amplifies and sustains synaptic drive to the dPAG. 22

When faced with potential predators, animals instinctively decide whether there is a threat they should escape from, and also when, how, and where to take evasive action. While escape is often viewed in classical ethology as an action that is released upon presentation of specific stimuli, successful and adaptive escape behaviour relies on integrating information from sensory systems, stored knowledge, and internal states. From a neuroscience perspective, escape is an incredibly rich model that provides opportunities for investigating processes such as perceptual and value-based decision-making, or action selection, in an ethological setting. We review recent research from laboratory and field studies that explore, at the behavioural and mechanistic levels, how elements from multiple information streams are integrated to generate flexible escape behaviour.

SummaryInstinctive defensive behaviors are essential for animal survival. Across the animal kingdom, there are sensory stimuli that innately represent threat and trigger stereotyped behaviors such as escape or freezing [1, 2, 3, 4]. While innate behaviors are considered to be hard-wired stimulus-responses [5], they act within dynamic environments, and factors such as the properties of the threat [6, 7, 8, 9] and its perceived intensity [1, 10, 11], access to food sources [12, 13, 14], and expectations from past experience [15, 16] have been shown to influence defensive behaviors, suggesting that their expression can be modulated. However, despite recent work [2, 4, 17, 18, 19, 20, 21], little is known about how flexible mouse innate defensive behaviors are and how quickly they can be modified by experience. To address this, we have investigated the dependence of escape behavior on learned knowledge about the spatial environment and how the behavior is updated when the environment changes acutely. Using behavioral assays with innately threatening visual and auditory stimuli, we show that the primary goal of escape in mice is to reach a previously memorized shelter location. Memory of the escape target can be formed in a single shelter visit lasting less than 20 s, and changes in the spatial environment lead to a rapid update of the defensive action, including changing the defensive strategy from escape to freezing. Our results show that although there are innate links between specific sensory features and defensive behavior, instinctive defensive actions are surprisingly flexible and can be rapidly updated by experience to adapt to changing spatial environments.

SummaryInstinctive defensive behaviors are essential for animal survival. Across the animal kingdom there are sensory stimuli that innately represent threat and trigger stereotyped behaviors such as escape or freezing [1][2][3][4]. While innate behaviors are considered to be hard-wired stimulus-responses [5], they act within dynamic environments, and factors such as the properties of the threat [6][7][8][9] and its perceived intensity [1,10,11], access to food sources [12][13][14] or expectations from past experience [15,16], have been shown to influence defensive behaviors, suggesting that their expression can be modulated. However, despite recent work [2,4,[17][18][19][20][21], little is known about how flexible mouse innate defensive behaviors are, and how quickly they can be modified by experience. To address this, we have investigated the dependence of escape behavior on learned knowledge about the spatial environment, and how the behavior is updated when the environment changes acutely. Using behavioral assays with innately threatening visual and auditory stimuli, we show that the primary goal of escape in mice is to reach a previously memorized shelter location. Memory of the escape target can be formed in a single shelter visit lasting less than 20 seconds, and changes in the spatial environment lead to a rapid update of the defensive action, including changing the defensive strategy from escape to freezing. Our results show that while there are innate links between specific sensory features and defensive behavior, instinctive defensive actions are surprisingly flexible and can be rapidly updated by experience to adapt to changing spatial environments. KeywordsInnate behavior, defensive behavior, escape, freezing, mouse, spatial learning, spatial memory, shelter not peer-reviewed)