A change of mind: from bitter recollections to sweet memories

Image credits: http://www.pxleyes.com/

Image credits: http://www.pxleyes.com/

There are a LOT of articles on this study already, but most don't go into the technical details – the part that, in my opinion, really makes it shine. Here's my take on the study, in which I try to answer "so HOW did they get those results again?" 

I have a love-hate relationship with Benny’s Café. It’s not just my favourite writing spot, it’s also where I first received notice that my manuscript was rejected, where I turned down a dream job offer for love, where I went and cried when that romance ended.

Throughout these ups-and-downs, my memory of Benny’s remained the same; what changed were the emotions that I associate with the cafe. In a remarkable feat of genetic engineering, scientists from MIT have now shined light on how sweet memories turn bitter (and vice-versa) as we experience new events in life.

The where and what of memory

In essence, our memories are stored in the firing of a network of neurons in the brain. Activate these neurons – together or in sequence – and we recall that memory. Once brought to the surface of our minds, the memory trace becomes temporarily malleable so that it can incorporate new information before settling back into a stored, stable state.

Like stories, memories of life events contain information about the wheres and the whats. Representations of places– “where” memories – are encoded in the hippocampus, a sea horse-shaped structure buried deep within the brain. The emotional components – the “whats” – are processed in a different brain area called the amygdala. The hippocampus and amygdala talk extensively with each other, and the two memory components are thought to linkup during learning. The scientists wanted to see if this association between “where” and “what” could be artificially changed.

So how do you go about manipulating a single component of a memory?

Recording and replaying memories with light

The team engineered a group of male mice with several genetic quirks. When an antibiotic known as doxycycline is removed from their daily diet, these mice would express a protein called channelrhodopsin 2 (ChR2) in recently activated neurons in their hippocampus or amygdala. ChR2 is responsive to blue light: by shining light through an implanted optic fibre cable, researchers can re-activate neurons labeled with these proteins.

ChR2-labelled neurons in the hippocampus (DG) or amygdala (BLA) are in red.

ChR2-labelled neurons in the hippocampus (DG) or amygdala (BLA) are in red.

In a way, the system functions as a recorder with playback abilities. Removing doxycycline is like pressing “record”: any neurons active during this period gets tagged with ChR2. The memory is at least partially stored in these active neurons. Adding the antibiotic back stops the recording, and shining blue light replays the recorded memory.

Turning a painful memory sweet

Researchers removed doxycycline – beginning the recording process – and trained these mice to fear a certain box by repeatedly zapping their paws with electrical shocks. Two groups of neurons are activated to encode this painful experience: hippocampal neurons, to remember the place, and amygdala neurons to encode the negative emotional association. After this induction phase, researchers brought doxycycline back into the mice’s diets, and thus stopped the recording. This ensures that only neurons active during induction – and thus labeled with ChR2 – can be re-activated with light.

Two days later, researchers placed the mice in a box and allowed them to freely explore the place. Whenever the mice entered a designated area, the researchers would turn on blue light and activate ChR2-labelled neurons. Unsurprisingly, the mice dully fled the area, suggesting that their painful past memory was reactivated with light. In contrast, control mice that were not previously shocked were oblivious to light, as were mice that were shocked but couldn’t express ChR2.

The authors then went on to test their central idea: that a bitter memory could be turned sweet. They fear-conditioned a new group of male mice as before, and then reactivated their fear memory a few days later. This time, however, the scientists gave them something sweet to remember: the company of female mice (see graph below for timeline).

Mice were shocked on Day 3, and active neurons were labelled with ChR2 (red balls). The memory was reactivated with light on Day 5, and again on Day 7 when the mice met their female friends. OptoPA tests for avoidance, while OptoPP tests for preference (how much mice lingered)

Mice were shocked on Day 3, and active neurons were labelled with ChR2 (red balls). The memory was reactivated with light on Day 5, and again on Day 7 when the mice met their female friends. OptoPA tests for avoidance, while OptoPP tests for preference (how much mice lingered)

Two days later, when researchers reactivated the same hippocampal neurons with light – ones that had previously encoded the painful memory – the males lingered around the spot instead of running away, as if recalling the brief but sweet encounter. In contrast, mice with light-activated amygdala neurons quickly fled the scene, suggesting that female company didn’t alter the emotional component of the fear memory. In other words, the amygdala seems to be encoding a fundamental emotional response, regardless of context or circumstance.

Rewriting the past

Why did the male mice seem less afraid after their romantic encounter? Did the new happy memory fundamentally alter the old fear one? To test this idea, the authors placed the mice back into the box where they were originally shocked – the box acts as a strong trigger that brings back the fear memory.

DG: hippocampus; BLA: amygdala. "mCherry-only" are control mice that can't express ChR2. Black bars are fear-conditioned mice that weren't exposed to a female on Day 7. Day 3 is testing for light-reactivated fear memory. Day 11 is the final day, testing for box-reactivated memory (ie the original fear memory).

DG: hippocampus; BLA: amygdala. "mCherry-only" are control mice that can't express ChR2. Black bars are fear-conditioned mice that weren't exposed to a female on Day 7. Day 3 is testing for light-reactivated fear memory. Day 11 is the final day, testing for box-reactivated memory (ie the original fear memory).

Mice that had their hippocampal neurons activated during their romantic tryst spent far less time frozen in fear (graph d; blue bars, compare D11 to D3). Instead, they wandered around, sniffing for their female companions (graph e; see how the blue bars are higher than all the others?). The joyous encounter had seemingly sweetened the old memory, shifting the emotional association for the box from negative to positive. Those that had their amygdala neurons activated weren’t so lucky: they were even more afraid of the box (graph d; pink bars, D11 is higher than D3), perhaps as a result of being repeatedly reminded of the experience due to light-reactivation of the fear memory.

Additional experiments showed that a sweet memory could also be turned bitter if its associated hippocampal neurons were reactivated when the mice were shocked.

Researchers aren’t quite sure how new experiences change old hippocampal-amygdala connections. But they have some ideas. After the fear-conditioned mice met their lady friends, their hippocampal neurons were less able to activate fear-associated (i.e. ChR2-labelled) amygdala neurons. The connection had weakened. However, the total number of active amygdala neurons didn’t change, suggesting that the same hippocampal network formed connections with a new population of amygdala neurons – perhaps those that encode happiness.

This study elegantly illustrates the power of genetic engineering. The results are perhaps not that surprising to seasoned neuroscientists, who have known for quite some time that a memory can be broken down into components encoded by different brain areas. Given that light activates all ChR2-labelled neurons simultaneously, it is unlikely that light-triggered recall can fully reconstitute the richness of a memory, in particular those dependent on the sequential activation of neurons – for example, the memory of your favourite musical piece or the complex route we take to work each morning.

Nevertheless, the study is a beautiful – if perhaps a bit creepy – piece of research. And it brings me hope that I’ll eventually completely love Benny’s Cafe again; I just need more positive associations.


Redondo RL, Kim J, Arons AL, Ramirez S, Liu X, & Tonegawa S (2014). Bidirectional switch of the valence associated with a hippocampal contextual memory engram. Nature PMID: 25162525

PS I've quiet on the blogging front because I've been busy writing my doctoral dissertation (YIKES!). Now that's sort of out of the way, I'm hoping to write more often :) Here's to graduating in November!