From PAX Aus: The Psychology and Neuroscience of Jump Scares

-by Nathan Randall, Featured Author. The following article is based on Nathan’s portion of With a Terrible Fate’s horror panel at PAX Australia 2016.

Lately there has been a trend of games released that center on jump scares.[1] The moment-to-moment gameplay in these games is relatively minimal, and in some cases even rather dull. But then, apparently out of nowhere, the monster appears on screen, killing the protagonist and scaring the player in the process. Some of these games include Slenderman, the upcoming Resident Evil 7, and the Five Nights at Freddy’s series.

But what is it about these games that makes them so effective at scaring people? And why might it be that people actually enjoy the experience of being scared senseless? It turns out that the fields of behavioral psychology and neuroscience have some answers to these questions. In order to answer them I will discuss various types of learning and how they apply to jump scares, describe the effectiveness of jump scares when the player is trying to multitask, and wrap up with a discussion of how hormones create the positive feelings that lead players to keep playing.

Before diving into these academic fields, however, I’d like to summarize the game that I’ll be using as my paradigmatic example of a game that makes fantastic use of jump scares: Five Nights at Freddy’s. Feel free to skip to the following two paragraphs if you’re already familiar with the game.

In Five Nights at Freddy’s the player plays as a nighttime security guard who’s been hired to run five night shifts at the Chuck-E.-Cheese-type location “Freddy’s.” However, as quickly becomes clear to the player, the real security threat at Freddy’s is not a break-in, but rather the animatronics that come to life at night and try to eat the people in the building. So the goal of the game ends up being simply to keep the animatronics from killing you during your five-night employment.


A personification of the security guard from Five Nights with two of the deadly animatronics standing next to him.

You have to do this all from within the confines of the security room, but you do have a few tools at your disposal. You can check the security footage for any of the dozen or so cameras set up throughout the facility, and you can briefly lock the doors to the security room. If one of the animatronics successfully gets to the security room, a jump scare follows, and the player loses. You can see a video of the gameplay including a jump scare below.

Five Nights makes use of two different types of jump scares, which I term player-dependent and player-independent jump scares. The difference between these two types of jump scares is fairly intuitive. Player-dependent scares are contingent on the actions of the player. If the player sits still and does absolutely nothing, then the jump scare will not happen. However, if the player does some particular action, the jump scare will happen. Player-independent scares are exactly the opposite: they are not contingent on the actions of the player. The jump scare will happen even when the player does absolutely nothing.

However, there is one important complexity in this model. Jump scares that depend on player inaction function equivalently to jump scares that happen irrespective of player input. There are jump scares that only occur if the player fails to do certain things. The lack of occurrence of a jump scare is contingent on the actions of the player insofar as the player can prevent the jump scare through action. However, the occurrence of the jump scare is actually contingent upon player inaction. Thus, when the jump scare actually appears, it behaves as a player-independent scare rather than a player-dependent scare. More important than that, the jump scares in question make use of the same underlying psychology as the player-independent jump scares, and because of that it is useful to think of jump scares that occur only if the player fails to do certain things as player-independent.

Player-dependent and player-independent jump scares make use of different underlying psychology. Player-dependent scares are based on operant conditioning, whereas player-independent scares are based on classical conditioning. Operant conditioning occurs when an animal performs some behavior more frequently because it is rewarded (or performs it less if it’s punished). In contrast, classical conditioning is the process of associating certain stimuli with other stimuli. I’ll discuss each of these types of conditioning and the associated jump scare type in turn.

Operant conditioning was first described by B.F. Skinner (along with Edward Thorndike). Skinner was known for the “Skinner Box,” which was the primary experimental paradigm for operant conditioning studies for decades. The basic idea of the Skinner Box is to put an animal in a box rigged with various contraptions. These contraptions give the animal some reward or punishment in a fixed way to specific actions performed by the animal in the box (some of the rewards were food, juice, sex, or just freedom from the box; the usual punishment was an electric shock). Skinner and Thorndike’s crucial initial discovery was that the animals tended to perform the actions that gave them rewards more quickly and artfully as more trials were run. This idea that actions that are rewarded occur more frequently is the basis of operant conditioning.


Thorndike’s original experiment, in which a cat is placed in a box with a mechanism that opens the door.


A Skinner Box. The mouse can press the lever to receive a food pellet.

Creating an effective player-dependent jump scare, then, is a matter of playing with this tendency that people have to form action-response associations. The two ways of playing with this tendency that I’ll discuss in this article are: giving the player a false sense of security, and constantly changing the rules.

Creating a false sense of security is a fairly straightforward process. For a while, the game is very predictable. The player performs some action A in a specific context X, and then receives some reward R. This process repeats several times. Now whenever the player is in context X, they perform A without giving it much thought, and receive the reward R. To create the jump scare, all that need be done is make it so that at some point when the player is in context X, they perform action A, and instead of receiving R they receive a jump scare. This formula is very simple to execute, and when done properly is very effective, because it disrupts the operant conditioning process.

Another way that horror games play with operant conditioning is by never allowing associations to form in the first place. There are two ways in which this can happen:

  1. Nothing ever happens the same way given the same input.
  2. The player fails regardless of their input.

Both of these techniques have surprising consequences, however. Depending on how they’re used, games that incorporate these techniques can stray outside of the horror genre, or even create an emotional experience distressing enough that the player is more likely to stop playing then see the game through.

The tricky aspect about (1) is that this conditioning paradigm can easily stray out of horror and into absurdist comedy. One of the defining aspects of absurdist comedy is the inability for the audience to predict how events in the artwork will unfold. The two examples I’ll give are Jazzpunk and a very strange game, Japanese World Cup 3.

Rather than attempt to explain either of these games, I recommend watching the videos. The key takeaway from these examples is this: if the rules of the game are constantly changing and weird stuff keeps happening, then the game will likely induce laughter, or at least an “I don’t understand” response from the player.


A tourist in Jazzpunk talks to the player. The “incoherent nonsense” is the subtitle for what the tourist is saying.

The tricky aspect of (2) has to do with another idea within behavioral psychology called learned helplessness. To understand learned helplessness, I’m going to explain the experimental procedure that led to its discovery. The experimental setup is basically a specialized Skinner Box. There are two compartments in the box, each with a floor capable of delivering an electric shock to an animal. There is a hole in between the two sections through which the animal can pass.


A diagram of the experimental paradigm that was used to first discover learned helplessness.

The experiment was originally run with dogs. There were two different conditions for the dogs. In both conditions, a light would turn on preceding an electric shock from the floor. What differed between the conditions was how much of the floor was shocked. In one condition, only the compartment that the dog was in when the light turned on got shocked. In the other condition, both compartments delivered a shock.

The behavior of the dogs varied massively between the two conditions. In the condition where only one compartment was shocked at a time, the dogs learned to jump to the other compartment as soon as it saw the light. In the other condition, however, the dogs eventually stopped doing anything at all. They would just lie there and whimper as they were being shocked. As a matter of fact, this was still the behavior of the dogs even after switching to the other condition. These dogs were in a learned helpless state.

The conclusion of the experiment was that the dogs in the second condition had learned that there was nothing that they could do to prevent the shock, and this state persisted even after options became available for the dog to help itself. Learned helplessness is the state of hopelessness and despair when those feelings are at their most vivid.

Learned helplessness is an incredibly powerful emotional tool, and not something that game designers should overlook if they seek to make emotionally powerful games. But there is a huge problem with a game intentionally putting its player in a learned helpless state: the player is not actually trapped inside of the game in the way that the dogs were trapped in the cage. An average player is likely to quit long before they reach a state of despair, just out of frustration.


So in general, if a goal of game design is designing a game that people want to play, it’s probably better to avoid mechanics that make the player feel helpless.

However, some games are able to masterfully deploy learned helplessness without compelling players to give up as a result. One of those games is Undertale (warning: the following section has spoilers for the ending of Undertale). One of the final bosses of the game is Photoshop Flowey, Flowey’s form after he ascends to Godhood by absorbing the souls of six humans. He’s determined not only to defeat the player, but also to show them their powerlessness. To do so, he repeatedly kills the player and crashes their game, all the while telling the player that they can’t win and that they’re doomed to failure. The player learns one thing from Flowey: they can’t win. Personally speaking, the boss fight put me in a state of hopelessness unlike anything I’d felt in a game before.


Photoshop Flowey.

So why doesn’t the player just stop playing? Why aren’t there many rage quits during this boss fight? The answer has to do with a major tagline for the game: “You are filled with DETERMINATION.”


The player sees this line appear every time they save the game, and they are also told not to give up every time that they are killed. The player has been given hints throughout the game regarding what to do during the Photoshop Flowey boss fight: not give up. The learned helplessness induced by Photoshop Flowey is thus made palatable by giving the player an anchor so that they do not quit along the way, and eventually see the other side of the confrontation. Eventually the game does allow the player to win when the souls of the humans rebel against Flowey and help the player defeat him. The game takes the player through an experience of learned helplessness and then helps them come out of it into triumph.

Classical conditioning was discovered by Ivan Pavlov while working with dogs. The experimental paradigm worked as follows. Initially, when Pavlov rang a bell, his dogs would not salivate in response (there is nothing inherently salivation-inducing about the sound of a bell). But, after repeatedly pairing the sound of the bell with giving the dogs food, eventually simply ringing the bell would cause the dogs to salivate. The bell thus became predictive of food, and caused a response of food-expectation from the dogs.


A graphical description of classical conditioning.

Classical conditioning forms the basis of player-independent jump scares, especially in terms of suspense. By classically conditioning the player, a movie can create powerful feelings of suspense. While suspense is a powerful horror technique, I will not focus on it in this article other than to say that an effective player-independent jump scare tends be one that has little suspense beforehand, and thus is difficult to predict. One method of making a jump scare work well is to remove any predictive hints that it is about to happen. Thus player-independent jump scares depend on unpredictability to be effective.

But removing the predictive hints is actually harder to do than one may think. In our lives as consumers of media, we have been classically conditioned to consider many different things to be “suspenseful,” and thus predictive of a future jump scare. That’s part of the reason why watching a lot of horror makes jump scares in general less effective: the well-trained eye can see the scares coming. Modern culture has made many player-independent jump scares predictable. Their effectiveness has thus been undermined, and we as viewers are often not scared, or even find them laughable.

But video games are able to avoid the problem of the predictability of player-independent jump scares because of the potential for the use of randomness in games. The potential for video games to randomly generate content makes player-independent jump scares fundamentally less predictable than those of movies. A player-independent scare can just be set up on a random timer, and thus be less predictable than a movie, even in a second or third watching or play-through. Whereas in a movie you could pause at exactly the moment the jump scare occurs, look at the progress bar, and record the time that the bar reads, there is no plausible way to do this in games. An example of one of these random player-independent jump scares in a game comes from Five Nights at Freddy’s. The animatronics will at some point end up at the door to the security room and jump out to kill the player, but this event occurs on a roughly random timer.


A movie can be paused at a particular time. The same thing will be happening in a movie at that particular time every time it is watched. Games are not so consistent.

Thus it is easier in some sense to pinpoint exactly when a jump scare will happen in a movie than it is in a game.

At this point we have most of the tools we need to analyze why it is that Five Nights at Freddy’s will scare you. First, related to operant conditioning, the game-ending jump scares (which are the most potent ones) are player-independent. The player can take action to try to stop the scare from happening, but when the jump scare actually happens there are no player-dependent stimuli preceding it. So the main jump scares end up being player-independent. Second, since player-independent jump scares are more random in games than in movies, and since the game does a good job at hiding the cues for the jump scare, the jump scares are more likely to catch you off guard.

The third and final reason that this game is so effective at scaring its players is that the game induces in the player a state of cognitive overload. I will unpack this term by diving into some neural circuitry so that we can better understand just how Five Nights overload these circuits.

The model of neural circuitry that I will introduce makes use of an important hypothesis in neuroscience: the cellular connectionist hypothesis. The theory states that if we understand how a neuron (the primary communicative cell in the brain) functions, how it communicates to other neurons, and how systems of neurons are connected to each other, then we can understand the function of the brain, and how the brain creates human thought and behavior. One important corollary of this theory is that if a particular communication pathway in the brain is faster than another pathway, the cognitive or behavioral response associated with the former pathway will happen more quickly than the behavior associated with the latter pathway.

The following model will initially appear a bit confusing, but I will break it down piece by piece.


A diagram showing the communication pathways between various brain areas.

The chart shows the communication pathways in the brain that progress from sense to cognitive and/or behavioral responses. The four items in the middle are different brain areas that communicate with each other to progress from sense to response. The arrows simply represent communication pathways.

There are four brain areas to consider in this model. The first is the thalamus. I will not be discussing the function of the thalamus in this article, as it is complicated and not inherently related to fear response like the other brains areas I’ve included are. The only function the thalamus plays in the model I’ve presented is a time-waster: it takes longer to pass through the thalamus than it does to just traverse an arrow in the model.

The amygdala (to make an admittedly gross oversimplification) is the fear-center of the brain. When activated, it arouses the body, in a way that can either be positive or negative depending on context. Activation in the amygdala tends to correlate with a feeling of fear.

The prefrontal cortex is an area largely responsible for complex cognition and self-control. Thus most of its function is to suppress action in other areas of the brain, including the amygdala.

In a further top-down process, the dorsolateral prefrontal cortex manages the function of the prefrontal cortex. This process often displays as management of multitasking.

There three features of the diagram that I would like to emphasize in particular. The first is that the path from the senses, to the amygdala, to thoughts/feelings/responses is the shortest, and thus fastest, pathway. In contrast, the shortest pathway through the prefrontal cortex runs through the thalamus, and thus takes a little bit longer than the amygdalar pathways. Finally, the three brain areas that I’ve focused on can all communicate with each other.


The shortest communication pathway in the model. This one runs through the amygdala.


A slightly longer communication pathway that runs through the prefrontal cortex.


The three main brain areas in question communicate through the prefrontal cortex.

If we combine these three features of the diagram together with the mechanics of Five Nights at Freddy’s, we can start to get a clearer idea of the cognitive overload the game has the potential to put the player in, and the multiple levels of fear that a player is likely to experience. The mechanics of Five Nights at Freddy’s focus on multitasking. The player needs to keep track of multiple screens, multiple monsters, the battery levels on various devices, and even multiple doors to their room. Thus the dorsolateral prefrontal cortex is likely very active while playing Five Nights, as it is working to make sure that the prefrontal cortex is multitasking effectively and efficiently. Normally people are fairly decent at these sorts of multitasking games, but Five Nights adds in the complication of an impending jump scare.

The amygdalar pathway is faster than the more rational prefrontal cortex pathway, meaning that no matter what the player does, it is difficult not be scared for at least a fraction of a second in response to a good jump scare. But, if a person is expecting that a jump scare is coming, the prefrontal cortex can work to suppress the amygdala in order to keep the response from being as strong as it might otherwise be. But this takes work on the part of the prefrontal cortex, and prevents it from multitasking as effectively as it otherwise could. So, in the conditions of cognitive overload that Five Nights at Freddy’s imposes on the player, the player is likely to get scared by the jump scare, likely worried that a jump scare may happen at any moment, and likely anxious that they are not doing tasks well enough to prevent the jump scare. All in all, these are the proper conditions to leave a player a shivering mess (myself included).

So if Five Nights at Freddy’s is so effective at making people uncomfortable, why does anybody play it? One answer to this question relates to hormones in the body. After the jump scare occurs, there is a release of excitatory hormones throughout the body. These excitatory hormones are context-dependent: if you are in a safe place physically and/or mentally, you tend to feel good, and if you are in an unsafe place physically and/or mentally, you will be likely to feel terrible.

When the jump scare is over, hopefully the player detaches from the game a little bit, and realizes that they are in a safe space. So with the added hormone they feel good. So they decide to play another round. And then the hormone rush happens again so they play another round. This cycle could potentially repeat for a long time.[2]

But for two reasons the above cycle will not be infinite. First, players get better at games over time. As this happens, it does not take as much cognitive control to play the game, and the player can dedicate more cognitive effort toward suppressing the amygdala. Second, the player can also habituate to the jump scare, which means that there is less brain activation in response to the fear stimulus than there was when the player was first playing the game. These factors combine to cause less of a fear response upon seeing the jump scare.

In order to keep the players engaged from a neuroscientific and behavioral-psychological perspective is a scarier, more challenging game. In releasing sequels frequently that feature roughly the same gameplay but with more difficult challenges and scarier monsters, the developer of Five Nights at Freddy’s has accomplished exactly that. He’s given the players exactly what they want out of a sequel: a game way harder and scarier than the last one. One can see this progression by looking at the difference in monster art between Five Nights at Freddy’s (original) and Five Nights at Freddy’s 3.


An animatronic from the original Five Nights at Freddy’s.


An animatronic from Five Nights at Freddy’s 3.

We can use behavioral psychology to think about two different kinds of jump scares: player-dependent, and player-independent. Player-dependent jump scares make use of operant conditioning techniques to be effective, particularly by defying players expectations, or never allowing those expectations to form. Player-independent jump scares make use of classical conditioning, and are most effective when the player feels clueless about potential future jump scares.

Neuroanatomical pathways allow us to more precisely understand the jump scares at work in Five Nights at Freddy’s. Since the amygdalar pathway is shorter than the prefrontal cortex pathways, the only way to avoid being scared is to suppress the amygdala ahead of time, which is difficult to do in the cognitive overload situation that Five Nights puts the player in. So the player is highly likely to be scared. Even though the player eventually will habituate to these jump scares, or just get good enough at the game never to encounter one, since there is a new entry of the game every few months, there is always a harder, scarier challenge to take up.

Jump scares are a chance in games to systematically think about how the mechanics of a game emotionally impact the player. Jump scares do not need to be guess-and-check to create; they can be crafted to have precise emotional effects.

Nathan Randall is a featured author at With a Terrible Fate. Check out his bio to learn more.

[1] I’d like to thank my fellow With A Terrible Fate game analyst, Matt McGill, for sharing his thoughts about the place of classical and operant conditioning in the context of game design. In this article I both intend to advance my own ideas and to be a conduit for some of Matt’s.

[2] This cycle does not manifest for everyone. Personally, I get so shaken up after a good jump scare that I often end up never playing the game again.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s