Ludwig Wittgenstein claimed, “The world of a happy man is a different one from that of the unhappy man.” This is definitely true at least of the world of a man in danger. Here I argue that part of what we classically believe about perception no longer applies when we are under threat.
According to a unified theory of perception, the function and the rules that govern perception do not change depending on what is perceived, whether it is clouds in the sky, birds in the trees, flowers in the grass, or snakes nearby. On this model, if danger modifies perception, it is only indirectly and superficially, through top-down influences. For instance, the fact that you feel afraid of the snake that you see in your garden can make it more salient in its surroundings, but it does not fundamentally alter how perception works.
I reject this unified account. I claim that there can be different operating modes of vision, depending on the circumstances. In computer science, the notion of an operating mode is defined as a method that determines how an action is carried out to accomplish a given task, such as inputting, processing, outputting, storing, and controlling. A computer can switch from one mode to another depending on the context. What I propose is that when we are in danger, perception switches to a different operating mode, let us call it the Alert mode. It is characterized by a distinctive function, a distinctive affective phenomenology, and a distinctive temporal profile. There may be other special operating modes of perception, possibly a social mode when perceiving other agents, but I leave this question open here.
The specificities of the Alert mode follow from its unique function. It is typically assumed that the function of perception is to provide a veridical rendering of the perceived object. For instance, when you see a tree, visual processing typically results in a relatively accurate representation of its shape, its color, and so forth. Compare this with the perception of a snake. In this case, its function is not to afford an accurate representation of the properties of the animal for the sake of accuracy. It is primarily to enable the organism to survive the threat. This means that the processing of danger must take priority. Indeed, if there is a threat within your perceptual environment, then it is typically relatively close by, and it is thus urgent to discriminate it as a source of potential danger. Perception in the Alert mode is more under time pressure than in other modes of perception. In brief, you can afford to take time to find mushrooms under the trees, but you cannot wait to spot a snake; perceptual processing must then be quick and efficient. It thus comes as no surprise that it has been repeatedly confirmed at the experimental level that the visual processing of threatening stimuli has a distinctive behavioral and neural signature. We are better and faster at detecting, tracking, and discriminating snakes, spiders, angry faces, and other objects associated with an evolutionary negative value than other more ‘innocent’ objects. Even the earliest stages of activity in the visual cortex have been shown to be then modified (for review, see Vuilleumier, 2015; Maratos and Pessoa, 2019; Carretie et al., 2022).
I thus propose that the visual attitude itself differs when one is under threat. As a first approximation, one may say that one looks rather than merely sees threats (looking being a more active mode of vision), but this would not fully capture what is special about the Alert mode. Indeed, prioritization of threats should not be reduced to simple attentional boosting. Instead, I suggest that one becomes affectively engaged with what one sees: visual experiences in the Alert mode are intrinsically affective. The affective phenomenology is relatively minimal, often dim and elusive, possibly close to a feeling of unease, but it is a constitutive part of the perceptual experiences. Imagine seeing a large hairy spider on your pillow. It clearly does not feel good. It is not only that you feel afraid; it is the visual experience itself that is disturbing.
The immediate benefit of this unpleasant visual feeling is that one is immediately motivated to act. There is no need to wait to be afraid; seeing the threat suffices for motivation. One might reply that defensive behaviors are automatic, based on direct sensorimotor coupling independently of any visual or emotional experiences. However, it has been shown that even apparently basic escape responses are actually more cognitively sophisticated than generally assumed. They depend on a prior cost/benefit analysis, which computes, for instance, the energy required, the loss of opportunities (such as eating and mating), and the predation risk. In these computations, motivation becomes a key element.
The most efficient way to avoid harm is then to be one step ahead of the threat and thus, to anticipate what is soon awaiting you. The very notion of perceptual threat involves an anticipatory component. It is only because you expect that something bad will happen that you need to perceptually process the object in a special way. It is typically assumed that perception represents only what is currently happening now, but the Alert mode of perception is characterized by a distinctive temporal profile, oriented both towards the present and the close future, indeed a future in which you may be harmed.
However, there is collateral damage to being in the Alert mode. Indeed, it has been shown that it takes an extreme form in individuals suffering from anxiety disorders. This means that they are even faster at detecting threats, but this also means that their perceptual experiences are more disturbing, so disturbing that sometimes they feel the urge to look away or close their eyes. Their visual experience is only a sensory messenger carrying information about the bad news, and suppressing it is not going to change the situation. If they act in such a strange way, it is for the same reason that one takes painkillers, namely to suppress the unpleasant experience.
Over the last thirty years, the cognitive science community has heard again and again about 4E cognition. Our cognitive abilities are said to be Embodied, Enactive, Extended, and Embedded. It may now be time to explore the triple-A mode of perception: Affective, Action-oriented, and Anticipatory.
Frédérique de Vignemont
Frédérique de Vignemont is a CNRS research director, deputy director of the Jean Nicod Institute in Paris. Her research is at the intersection of philosophy of mind and cognitive science. Her major works focus on bodily awareness, self-consciousness, peripersonal space, and more recently, on danger perception. She has published two monographs, Mind the body (Oxford University Press, 2018) and Affective bodily awareness (Cambridge University Press, 2023). She has also edited two multidisciplinary volumes, The Subject’s Matter: Self-Consciousness and the Body with A. Alsmith (MIT Press, 2017) and The World at Our Fingertips: A Multidisciplinary Exploration of Peripersonal Space, with A. Serino, H.Y. Wong, A. Farnè (Oxford University Press, 2021).
