Brain patterns of haunting memories

Altered brain activity patterns mark the difference between forgetting and being haunted

The zombie that haunts my dreams. I blame my brain activity patterns.


Please welcome this month’s Scicurious Guest Writer, Emilie Reas!   —Bethany

My first trip to a haunted house is as vivid today as when I was 5 years old. As I made my way past a taunting witch and a rattling skeleton, my eyes fell upon a blood-soaked zombie. My heart raced, my throat swelled, and the tears began to flow. Even now, as a mature (ahem) adult, the ghosts and goblins don’t faze me. But those vacant zombie eyes and pale skin? Oh, the horror! My rational brain knows how irrational my fear is, yet still I shudder, gripped by the same terror that first overwhelmed me decades ago.

Unsettling experiences occur daily that we easily brush off – a creepy movie, a turbulent plane ride, or a nip at your ankle by the neighbor’s dog. But occasionally, the fear sticks, establishing a permanent memory that can haunt us for years. At their mildest, such fear memories cause discomfort or embarrassment, but at their worst, they can be downright debilitating. Do spiders make you scream? Are you unable to speak in public without a trembling voice and hands? Maybe you suffered a traumatic accident that’s made you terrified to get back behind the wheel of a car. We’ve all experienced the disruptive effects of a fearful experience we just can’t shake.

Yet scientists don’t fully understand why some traumatic events are fleeting, while others are stored as lasting memories. Past research has shown that particular areas of the brain, such as the anterior cingulate and insula, are active during fearful experiences, but also during many other situations, including while monitoring surroundings and emotions or paying attention to important information. Other regions, including the amygdala, hippocampus and prefrontal cortex, are more specialized to support memory for emotional experiences, as they play important roles in emotional processing, memory and attention. While it’s clear that establishing fear memories relies on cross-talk between these regions, it’s not known how they solidify fears into memory and determine which particular ones will endure for the long-term.

The similarity of brain activity patterns for fearful images (dark lines) was higher for participants who later remembered (left) than those who forgot (right) the fearful associations. R. Visser et al/Nature Neuroscience 2013

Recently, scientists from the University of Amsterdam looked for signals in the brain that predict whether a fear memory will be stored. Thirty-eight participants viewed pictures of faces and houses during functional magnetic resonance imaging (fMRI) of their brains. The researchers paired some of the faces and houses with a (scary but not painful) electric shock to the leg, others with a neutral sound and others with nothing. As expected, the participants described the images paired with the shocks as more unpleasant than the neutral images. To confirm these reports, the researchers estimated their fear level by measuring how much their pupils dilated — a proxy for fear — in response to the pictures. Sure enough, the participants’ pupils dilated more when viewing the fearful images than when viewing the neutral images. The participants returned for a second fMRI session two to six weeks later, during which they viewed the same images as during the first session, but this time no shocks were delivered. During this session, only 16 of the 38 participants’ pupils dilated in response to the pictures that were associated with shocks, indicating that only some of them stored the fear associations into long-term memory.

Next, the scientists looked at activity when the participants learned the fear memories (first session) and when the researchers later re-exposed them to the images (second session), focusing on a set of brain regions thought to be important for fear learning. During both learning and memory retrieval, the anterior cingulate and insula were more active when participants viewed the frightening images than the neutral ones. This wasn’t all that surprising, as earlier studies found that these areas become active during emotional experiences.

Close up you see the difference, but far away, candy is candy. Brain activity patterns are the same way. E. Raes
But the average activity level across a brain region can only suggest so much about what’s going on inside the region. At a given moment, some neurons are active, while others are quiet. Most fMRI studies average activity across an area — for example, the amygdala — to measure how the overall activity level changes with the experiment. But this blurring can obscure other valuable information about the pattern of activity within the area. For example, imagine there are two jars of purple candy on a distant table. As you approach, you see that one jar contains all purple candy, while the other contains half red and half blue. The difference was indistinguishable from afar, but closer inspection revealed a critical difference (c’mon, watermelon and blueberry trump grape flavor any day!) in the underlying pattern, generating the average purple color.

Likewise, the researchers looked beyond the large-scale average activity and into the fine-scale pattern of brain activity within each region, using a technique known as multi-voxel pattern analysis. This method relies on the premise that a subset of active neurons encodes a specific representation – here a fear memory – by increasing their activity while the others remain less active. Although fMRI doesn’t directly measure neuronal activity, this specific group of active neurons should generate a distinct fMRI activity pattern. The more similar the two mental states, the more overlapping their sets of active neurons will be — and the more similar their fMRI patterns are. Multi-voxel pattern analysis can then be used to test the similarity of patterns, or how well they correspond to a particular mental state. For instance, other studies have shown that activity patterns can predict whether someone is in pain, or if they’re remembering Madonna or the Taj Mahal. The researchers applied this tool to look for spatial patterns of brain activity that inform about whether a fearful experience will be stored into long-term memory. They compared activity patterns when participants viewed either the fear-inducing or the neutral images and measured how related the patterns were. Strongly correlated patterns would suggest that the brain processes generating them were more similar than those giving rise to less correlated patterns.

Brain activity patterns were highly similar whenever a fearful picture was viewed, even when the pictures were of different types (i.e., face versus house). Notably, these matching patterns occurred during both fear learning and retrieval, and appeared in several brain regions implicated in fear memory. The areas included some (anterior cingulate and insula) that were active on average when learning or remembering the fearful images (compared to the non-fearful images). But they also included new areas (amygdala, hippocampus, frontal and prefrontal cortex) that weren’t activated on average during fear memory. Thus, it appeared that the small-scale activity patterns within an extensive brain network signaled the expression of a fearful memory, regardless of the content of that memory. What’s more, the average signal in several of the regions didn’t change, suggesting that, despite no overall activation, they still carried an underlying neural representation of fear memory.

Of greater interest to the researchers was identifying neural patterns that signaled whether the fears would develop into lasting memories. To test this, they divided participants into 22 individuals who did and 16 who did not recall the fear associations during the second session, based on whether their pupils dilated upon seeing the fearful images.  To identify signals that might predict subsequent memory formation, the researchers looked for differences between the groups in brain activity patterns during learning. In both groups, activity patterns were more similar for fearful images than neutral ones in a few brain areas thought to be involved in processing emotional experiences (anterior cingulate, insula, frontal cortex). But the “learners” showed greater similarity for fearful images than neutral ones in several additional areas (amygdala, hippocampus, prefrontal cortex). These areas are thought to be specifically important for remembering emotional events, not just experiencing emotions. Neither the pupil response nor the average activity in any of these regions during learning predicted who would later remember the fearful associations – only the brain activity patterns did. Beneath the behavioral expression of fear and the large-scale fluctuations in brain activity, lurked fine activity patterns that represented the birth of fear memories.

Learning to fear is complex. The brain attends to salient features of its environment, sparks a strong emotional reaction, and encodes the experience into memory. Many of the brain areas identified in this study are already known to support attention, emotion and memory functions that are necessary for fear learning. These new findings add an important piece to the puzzle by identifying a higher-order representation of fear memory acquisition — detectable as a unique brain activity pattern — distributed across these regions. Rather than showing that a given area is simply active during fear expression, as past studies have done, it highlights a fine-tuned activity pattern that broadly predicts whether a fear will be remembered.

It is tempting to interpret this remarkably consistent activity pattern during fear learning as a general “fear memory” signal. But neither brains nor fears are one-size-fits all, begging a question this study hasn’t answered of whether such a marker of fear learning universally applies to all fears or all people. The study tested memory for a very specific type of fear, that of getting shocked. There are clear subjective differences between the fright you feel as a spider crosses your path, the anxiety from speaking to a large audience, and the terror of ascending Half Dome. It may be the case that an overlapping set of neurons are active whenever we associate an image with an electric shock, but were these same neurons also active when we were first spooked by spiders? The study also found that only some participants learned the fear memories. It’s not clear whether the brain patterns predicting fear memory retention were specific to the learners, or would have also appeared in the nonlearners, had they remembered the fear memories. Your brain isn’t identical to my brain, but they do share many common features. Do we all share a similar method of storing fear memories, or might they differ across individuals? More research is needed to know if these findings generalize to both you and your grandma, and hold up during other fearful experiences, besides getting shocked inside an MRI scanner.

If a neural signature of fear memory does indeed exist, future studies will help to better understand exactly what it represents, and how it’s communicated at the neuronal level. Such work might explain why select situations — like that first encounter with the zombie — trigger the brain to enter “fear memory storage” mode. This understanding could be the first step in developing interventions to prevent the acquisition of permanent, pathological fears, like those persistent in anxiety or post-traumatic stress disorders. For example, by understanding how the signal to store a fear memory is triggered and coordinated across a widespread brain network, neuroscientists could develop behavioral or pharmacological treatments that disrupt such activity. In the meantime, we can dream of an idyllic future in which our own kids could explore their first haunted house without fear of being eternally spooked by witches, ghosts and zombies.

Emilie Reas is a neuroscience Ph.D. student at the University of California, San Diego. She works with James Brewer using functional MRI to study how people encode and retrieve memories. Outside of the lab, she loves to write and is passionate about science communication. You can follow her on Twitter at @etreas, and read her blogs at and

More Stories from Science News on Neuroscience