Two cups of coffee after learning may cure your forgetful streak

Like most junkies, I struggle to come up with excuses to justify my addiction. Lucky for me, increasing evidence is supporting my semi-hourly coffee habit: caffeine, the world’s favourite drug, not only keeps you awake and alert, but may also boost your memory.

Screen Shot 2014-01-15 at 4.36.04 PM

Here, have a memory-jolting Latte. Source:

Perhaps in an effort to excuse their own coffee addiction, many research groups have studied whether caffeine enhances memory. The results, unfortunately, are highly mixed. One study, for example, found that 200mg of caffeine – roughly the amount in two cups of drip coffee – consumed before a memory-taxing game enhanced working memory, the ability to flexibly maintain and manipulate information in your head to solve a problem. The catch? Only if you’re an extrovert. In a separate cohort of volunteers, 75mg of caffeine (roughly that in a cup of espresso) taken together with glucose on an empty stomach helped stabilize a new verbal memory. This is called “memory consolidation”, whereby new and unstable memories are moved into semi-permanent storage. However, in that study caffeine by itself had little effect on memory.

One problem with these previous studies is that caffeine was always given prior to learning or testing. This makes interpreting any improvements in performance difficult: is caffeine directly boosting memory or is it enhancing performance indirectly through increasing attention, vigilance and/or processing speed, thus giving the appearance of memory gain?

D Borota et al. Post-study caffeine administration enhances memory consolidation in humans. Nature Neuroscience, published online Jan 12, 2014. doi:10.1038/nn.3623 

To get to the bottom of this, researchers from University of California, Irvine* decided to see how caffeine consumed after learning affects memory consolidation. They recruited 160 uncaffeinated adults, a rare breed that drank less than 5 cups of coffee per week and showed no traces of caffeine or its metabolites in their saliva prior to the experiment. In fact, average caffeine intake of most of these “caffeine naïve” people lingered around 70mg a week, coming mostly from chocolate and soda rather than coffee per se. (*The research described in this post was done at Johns Hopkins before the lead author moved to UC Irvine)

The volunteers first looked at a series of images of various objects, such as a saxophone, a sea horse or a basket, and categorized them as either an indoor or outdoor object. Upon completing the task, they immediately popped a pill containing either 200mg of caffeine or a placebo and left the lab.

A day later, the volunteers returned. By now all traces of caffeine and its metabolites had washed out of their system; they were stone-cold sober. The researchers then showed them a new series of pictures, instructing them to identify whether they had previously seen the picture (“old”) or if it was new. To make things harder, researchers sneaked in several pictures extremely similar those shown before. For example, instead of the old picture of a svelte sea horse arching its back, they now presented a “lure” picture of the animal hunched over. This type of “pattern separation” task is considered to reflect memory consolidation to a deeper degree than simple recognition.

Regardless of caffeine intake, both groups had no trouble identifying the old and new pictures. However, as shown below, the caffeinated group outperformed their peers in picking out the lure, with a higher propensity of calling them out as “similar” rather than “old” (though the effect was small and barely reached significance, more on that later). In other words, caffeine seemed to help them retain minute details present in the original pictures. A similar boost in performance was seen when researchers repeated the experiment with 300mg of caffeine (~1 cup more than before), but the advantage disappeared when they dropped the dose down to 100mg. Remember that caffeine was administered after viewing the photos, hence the drug was not increasing attention to detail during the learning process.

Screen Shot 2014-01-15 at 4.45.12 PM

White bar is caffeine and grey bar is placebo. Notice the shorter white bar in the “Old” group (fewer lure images identified as old) and taller white bar in the “Similar” group (more images correctly identified as lure).

However, not everyone metabolizes the same 200mg caffeine pill to the same degree. When researchers accounted for individual differences in caffeine absorption and metabolism, they found that participants who broke down the largest amount of caffeine performed worse than those who metabolized slightly less. In other words, there is a sweet spot for caffeine’s memory enhancing powers – go either under or over and you loose that edge.

Finally, what if you waited too long after learning, only to remember to chug that Starbucks mocha the day after? In a separate study, researchers allowed 24hrs for volunteers to consolidate the memory of the initial picture stack before giving them the same caffeine pill, just one hour before the test. This time it didn’t work – these volunteers mixed up similar and old pictures just like the placebo control group. Whatever caffeine is doing, it has to be done during consolidation.

Researchers aren’t quite sure how caffeine induces memory gain, but they have a few ideas. The image discrimination task used here engages the hippocampus, a key brain area involved in learning and memory. It expresses high amounts of the caffeine receptor (adenosine A1 receptor) in its CA2 subregion, thus allowing caffeine to tweak (strengthen?) its function in memory consolidation. Caffeine can also indirectly boost the level of norepinephrine, a neurotransmitter that helps you lay down a memory for good.

While exciting, this study cannot end the debate on whether caffeine improves memory. The effect sizes were small, with some only scraping significance – that is, researchers were only barely able to say with some confidence that the effect is real. This doesn’t reflect the quality of the research, but most likely represents individual variance among the volunteers: different gene variants for faster caffeine metabolism, BMI, basal metabolic rate, oral contraceptives and so on. It would also be interesting to see if caffeine boosts memory reconsolidation: when you retrieve a memory, it temporarily becomes labile. Can coffee help the memory restablize?

Unfortunately, we don’t known if caffeine-induced memory gain applies to caffeine junkies like me. But to quote the lead author: one needs to do the experiment with habitual drinkers to find out, but my guess is that it’s why we’re so awesome!

Many thanks to the principal investigator @mike_yassa for patiently answering my questions over Twitter. You can check out our full conversation in my timeline.
Borota D, Murray E, Keceli G, Chang A, Watabe JM, Ly M, Toscano JP, & Yassa MA (2014). Post-study caffeine administration enhances memory consolidation in humans. Nature neuroscience PMID: 24413697


Top 10 science stories I wish I’d blogged about in 2013

As 2013 grinds to an end, the internet fills with reminiscence of the year’s top stories and moments. I, for one, especially can’t resist ruminating about the past, especially when packaged in a brain-tickling, “top n”  list form. Without further ado, here is my Top 10 list of the year: Science Stories I Wish I’d Blogged About.

Bonus. A List of Reasons Why Our Brains Love Lists. By Maria Konnikova.

To start off, why are we drawn to lists anyway? Is it due to the clean, structured organization that helps us navigate the material efficiently? Or is it more a product of our current “bite-sized” information culture? Maria has the answers.

10. Can we lessen the effects of fearful memories while we sleep?

Sleep therapy can change bad memories. By Helen Shen (original paper here)

Spontaneous activation of memories during sleep is generally thought to strengthen them. However, when researchers in Northwestern University repeatedly brought up a recently learned fear memory in their sleeping participants by presenting a fear-associated odour, the participants showed a smaller fear reaction to the odour after they awoke. According to the researchers, this is the first time emotional memories have been successfully manipulated in humans.

Similar: A gene for forgetting. MIT researchers identified a gene Tet1 that is critical for memory extinction in mice. Original paper in Neuron.

9. Men and women’s brains are wired differently. Is THAT why men can read maps better (or so the cliché goes)?

Here‘s one cover of the study that would let you believe that (gasp) it is indeed so!

Here are a few level-headed analyses that tackle the nitty-gritty of the study and how its conclusions got blown out of proportion. The bottom line? Brain scans don’t tell us anything about behaviour. Here’s the original paper for reference.

Are men better wired to read maps or is it a tired cliché? By Tom Strafford.

Men, Women and Big PNAS Papers. By Neuroskeptic.

Getting in a Tangle Over Men’s and Women’s Brain Wiring. By Christian Jarrett.

8. Mice inherent fears of their fathers. (And update) By Virginia Hughes.

You know how you are what your grandpa ate? Epigenetics offers an answer to how our interactions with the environment can influence the expression of our and our offspring’s DNA. However there is little evidence that stress and fear can directly change the germline, so that offsprings inherit the fear memory (or something akin to it) of their parents. (There was this interesting report earlier in the year on how cocaine-addicted sires lead to cocaine-resistant male pups through a purely epigentic means, though I remain skeptical.)

Virginia Hughes broke this story at the 2013 Society for Neuroscience conference. Since then, it has garnered plenty of attention from media and neuroscientists alike, with opinions from “deep scepticism” to “awe-inspiring”. Here’s the original paper if you’d like all the juicy details.

7. From Club to Clinic: Physicians Push Off-Label Ketamine as Rapid Depression Treatment. By Gary Stix

Ketamine, the clubbing sweatheart and horse tranquillizer, is now being repurposed as a fast-acting antidepressant; this is perhaps THE most breakthrough new treatment for depression in the last 50 years. In this 3-part series, Gary Stix explains the uprising of grassroots ketamine prescriptions, big pharma interest in the drug and how ketamine is directly the development of next-generation antidepressants.

6. Computer Game-Playing Shown to Improve Multitasking Skills. By Allison Abbott.

Rejoice, gamers of 2013! Not only has the year given us PS4 and Xbox One, this study from Nature has also given us an excuse to game (uh, or not): in subjects aged 60-85, playing a 3-D race car-driving video game reduced cognitive decline compared to those who didn’t.

Commercial companies have claimed for years that brain-training games help improve cognition; yet whether their games actually work is hotly debated (I’m looking at you, Luminosity!). In this new study, researchers from UCSF show that a game carefully tailored to a specific cognitive deficit can be useful, even months later. Unfortunately this doesn’t mean any ole’ video game will do. Shame.

5. 23andme versus the FDA.

I’m sure by now you’ve heard about the fight of the year.

David Dobbs has a full page of links over the 23andme and FDA food fight. What’s the big deal? Why did the FDA issue a cease and desist order? Is it simply a clash of cultures between the company and government department? Or are we selling out our own genetic data to the next-generation Google, and should we fear the services the company offers?

4. Sleep: The Ultimate Brainwasher? By Emily Underwood (Here’s another cover by Ian Sample).

Why do we sleep? Reasons range from learning and memory, metabolism and body-weight regulation, physiology, digestion, everything. A study this year proposes that sleep has another function: nightly cleaning, in which the cerebral spinal fluid washes a day’s worth of brain waste down the sewers. That is, if you’re a rat.

3. Death by sugar? by Scicurious.

With fat making a come-back, sugar and/or carbs are the devil this year. This study in Nature Communications says yes: when mice consumed a diet that has an equivalent amount of sugar to that of many people in the US, the animals’ health and reproductive ability declines.

However, as Scicurious astutely asked, can we really directly translate conclusions derived from mice to humans? Is sugar really that evil?

Here is another article on the topic by Ferris Jabr that’s well worth a read.

2. False memories implanted in mouse’s brain by linking portions of two real memories together. Wow. Just, wow.

False memory planted in mouse’s brain. By Alok Jha

Scientists Plant False Memories in Mice–and Mice Buy It. By Joel N. Shurkin

This is one paper I REALLY wish I had the time to cover when it first came out. An MIT group artificially connected the memory of a safe box and the memory of a footshock in another box to generate a new hybrid memory. This is not “implanting” a de novo memory – that is, researchers didn’t use electrical stimulation (or something similar) to generate a memory from scratch. The study also can’t tell us how false memories are generated biologically in our brains (ie linking imagined material to actual memories), but the study is genuinely fascinating all the same.

Here is a link to the paper, and here is the lead author doing an “ask me anything” interview on Reddit.

1. Knockout blow for PKMzeta, the long-term memory molecule. 

Single protein can strengthen old faded memories, Exposing the memory engine: the story of PKMzeta, and Todd Sacktor talks about the memory engine by Ed Yong

In a nutshell, previous studies have identified a single protein called PKMzeta that helps maintain long-term memory. Unlike other kinases (a type of protein involved in many cellular processes, including memory) PKMzeta is always active, and seems to help sustain the strengthening of connections between neurons during memory formation. Inhibit PKMzeta, and the memory’s gone.

These results spurred HUGH interest in the “memory molecule”, often with references to Eternal Sunshine of the Spotless Mind. However, things went for a downward spiral at the 2012 Society for Neuroscience conference, when researchers presented the first evidence that mice without PKMzeta had no impairments in LTP (long-term potentiation, widely considered a cellular mechanism for learning and memory) could still form memories. The two groups published their findings in early January 2013 in Nature (here and here).

These observations don’t necessarily mean that PKMzeta is not a memory molecule – it very well could be one of the MANY memory-associated proteins. Given the redundancy that often comes with evolution, it’s hard to believe that one particular molecule would be the sole guardian of our memories. The question remains whether PKMzeta is a MAJOR player, but overall, the debate is a cautionary tale against putting one molecule on the pedestal. So if (or when) you see another article with the headline “erasing a bad memory”, remember there’re plenty of other players in memory that you haven’t been told about.

OTC painkiller may blunt memory loss from puffing pot

Pot’s not the best thing for your memory. Yes, I know there are functional potheads who enjoy their greens and get also their work done. Still, it’s hard to ignore the legions of studies that show Δ9-THC consumption impairs spatial learning and working memory – that is, the ability to hold several pieces of information in mind and manipulate them to reach a mental goal.


Welcome to downtown BC and BC Bud! Source:

Yet paradoxically, THC may benefit those with Alzheimer’s disease. Previous research in rats show that the compound breaks down clumps of disease-causing proteins (called β-amyloid plagues) by upregulating a “scissor” enzyme that chops them up. Sweeping out these junk protein plagues decreased the number of dying neurons in the hippocampus, a brain area crucial for learning and memory. THC also has powerful anti-oxidant effects and may protect the integrity of mitochondria – the “power plants” of our cells.

So here’s the dilemma: THC may potentially battle dementia, yet it also naturally impairs memory. In an unexpected turn of events, scientists from Louisiana State University discovered a key protein that mediates THC-caused memory loss, and show in mice that you can have your edibles and eat it too.

The protein in question is COX-2, a crucial player in inflammatory pain – think headaches, muscle pains and fever. Sound familiar? That’s because COX-2 is one of the targets of OTC painkillers such as Asprin and Tylenol (the other one is COX-1). Scientists have previously linked 2-AG, a THC-like substance produced endogenously in the brain, to inhibiting COX-2 signaling. Blocking COX-2 led to problems with memory retention. So naturally, they wondered whether THC impaired memory in the same way.

Screen Shot 2013-11-27 at 1.12.56 PMScreen Shot 2013-11-27 at 1.13.13 PM

They found the opposite. As you can see on the left (blue bars), a single injection of THC boosted the level of COX-2 in both neurons and astroglias (“structural” non-neurons that play a role in memory and inflammation) in the hippocampus; the more THC, the more COX-2. This effect went away by 48hrs after the injection, but when the mice went on a weeklong THC binge (1 dose/day), their COX-2 levels remained chronically high cough unregulated (right graph, red bar compared to control black bar). When researchers blocked the THC/endocannabinoid receptor CB1R by either genetically deleting it or using a selective pharmaceutical blocker, the effect went away, showing that THC administration is indeed the cause of COX-2 increase.

Why would endogenous cannabinoids (2-AG) and THC have polar effects? Further molecular sleuthing revealed that it’s all in the messenger: although both 2-AG and THC activated the same receptor, 2-AG recruited Gα as courier, while THC opted for Gβγ. It’s like slapping a different address sticker on two boxes shipped to the same sorting facility; they’re now going different places. Indeed, Gβγ triggered a molecular cascade that activated several proteins previously shown to impair memory.

Naturally, researchers went on to block COX-2. After a week of THC, neurons begin to loose their spines – that is, little protrusions along the dendrite that house proteins necessary for forming and maintaining synapses (compare red bar/THC to black bar/control below). The breakdown of spines caused a decrease in the many proteins and receptors needed for normal excitatory signal transmission. Unsurprisingly, eliminating these channels of communication blunted the response of a cohort of neurons in the hippocampus after electrical stimulation. However, giving a COX-2 selective blocker concurrently with THC rescued all these deficits – structural, molecular and electrical (green bar – the spines are back!).

Screen Shot 2013-11-27 at 2.02.09 PM

Spines come in all shapes and sizes. Grey bar: COX-2 inhibitor alone; Green bar: THC+COX-2 inhibitor

As for mutant mice that lack COX-2 at birth? They didn’t suffer any of these problems associated with THC. In the case of spines, as you can see above, THC (burgundy bar) had no effects compared to control (blue).

Do any of these “under-the-hood” changes lead to observable behaviour? In a fear-conditioning experiment, researchers trained mice to associate a box with electrical shocks. They then gave some of the mice 7 days of THC with or without a COX-2 inhibitor. When tested 24hrs later – presumably to weed out THC’s effect on anxiety* – stoner mice showed little fear when put back into the box. Those on the multi-drug regime, however, froze in fear. Like their sober peers, they retained and retrieved the fear memory. (The half-life of THC is ~20.1 hrs in mice, so they might have still been high at the time of testing.)

In a spatial memory task, researchers trained mice to find a hidden platform in a big tub of water. After 5 days of training, they then gave a subgroup a single injection of THC 30min before the test, which resulted in these mice taking roughly twice as long to find the platform as the controls. Once again, concurrent COX-2 administration “saved” the memory of the platform location. 24hrs later, after the mice had sobered up, they were tested again – same results.

Amazingly, inhibiting COX-2 did not destroy THC’s ability to wipeout Alzheimer’s-related protein plagues in a mice model of the disease. Treatment with THC once daily for a month, with or without the OTC COX-2 inhibitor Celebrex, significantly decreased the number of protein clumps (green below) and protected hippocampal neurons (blue).

Screen Shot 2013-11-27 at 2.23.23 PM

Top row: control, middle: THC, bottom: THC+COX-2. Last lane is a magnified look.

Before you reach for the bottle of aspirin, joint in hand, maybe hold back on the self-medication just yet. For one, it’s hard to extrapolate these findings to humans, there are some interspecies differences in THC metabolism. Second, chronic COX-2 inhibition is linked to serious side effects such as ulcers and heart problems (think Tylenol is safe? Think again). Third, mice with inhibited COX-2 showed didn’t seem as couch-locked as they normally would; so if you’re after that body high, an aspirin would be rather counter-productive.

As a molecular neuroscientist, I love the detailed characterization of THC-CB1R signalling pathway, but the behaviour data could use some strengthening. Although researchers claimed that the water maze task assessed working memory, the protocol they used looks at normal spatial memory. To specifically probe working memory, they would’ve needed to move the platform to different locations and see how well the mice updated their memory. The results also directly counter those of a previous study, which showed that once the mice learn the location of the platform, THC did not impair the memory. They also didn’t report whether THC mice were simply too stoned to swim (or motivated enough) – tracking total swimming distance and speed at the time of testing would’ve helped .

This study focuses mostly on neurons*; a previous study published in March 2012 showed that THC impairs memory through a type of glia called astrocytes (the non-neuron brain cells); in fact, marijuana impaired working memory only when it was able to bind to the CB1Rs on astrocytes. That study pointed to deregulation of excitatory neurotransmitters as the cause of memory impairment; could COX-2, which is expressed in glia, also have a role?

*Edit: HT to reddit/u/superkuh. The text suggests that the authors of this paper did not consider the role of astroglia; in fact they explicitly did, when they showed that COX-2 upregulation occurred greater in astrocytes than neurons. The authors also showed that the reduction of glutamate (excitatory) receptors was due to COX-2-induced increase in glutamate release from both neurons and glia.
Rongqing Chen et al (2013). Δ9-THC-Caused Synaptic and Memory Impairments Are Mediated through COX-2 Signaling Cell, 155 (5), 1154-1165

People with superhuman memories still mistake fantasy for reality

When I was young, the one superpower I craved above all was perfect memory. I’d picture my eyes as camcorder lenses, recording everything that I read, saw and experienced into the Kodak film that was my brain. Anytime I wanted to re-experience something, I’d simply hit a mental “play” button and BAM! The video of my life would play before my eyes, as clear and detailed as the day it was created.


Reality or fantasy? Source:

Little did I know that for those with Highly Superior Autobiographical Memory (HSAM), my fantasy was their reality. Researchers from University of California reported the first case of this phenomenal ability in 2007 when a lady dubbed AJ reached out to them – with a desperate plea for help – with the following email (experts):

Dear Dr. McGaugh,
As I sit here trying to figure out where to begin explaining why I am writing you and your colleague (LC) I just hope somehow you can help me.
I am thirty-four years old and since I was eleven I have had this un-believable ability to recall my past...I can take a date, between 1974 and today, and tell you what day it falls on, what I was doing thatday and if anything of great importance (i.e.: The Challenger Explosion, Tuesday, January 28, 1986) occurred on that day I can describe that to you as well...Whenever I see a date flash on the television (oranywhere else for that matter) I automatically go back to that day and remember where I was, what I was doing, what day it fell on and on and on and on and on.
... Most (people) have called it a gift but I call it a burden. I run my entire life through my head every day and it drives me crazy!!!…

AJ was different than any “memory experts” that the researchers had previously encountered. Her ability seemed completely innate; she did not use little tricks (“mnemonics”), such as mental imagery or story telling to help her remember. Her superior memory only pertained to specific aspects of her own life story and related events; she had trouble learning historical dates and reciting poems. Like your average Joe, she couldn’t remember what each of the 5 keys on her keychain was for. Although specific dates triggered recall in an automatic, “non-stop, uncontrollable and totally exhausting” manner, when asked if she’d talked about a particular date with the researchers in previous interviews, she said she couldn’t remember.

AJ, and the handful of people whom had come forward with HSAM since then, pose a troubling dilemma to memory researchers. Recalling an event is nothing like watching a video recording. Instead, it is an active reconstructive process prone to distortions from misinformation. For example, in labs, researchers have been able to manipulate people into remembering events they’d only previously imagined or trick them into recalling that they’d watched a news video that did not exist. In courts, “corrupt” memory plague eyewitness reports. Elizabeth Loftus, an expert in memory research, had even suggested that planting fake memories in children may be a successful way to modify delinquent behaviour.

Contamination seems like an inevitable part of our memory process, yet common sense suggests that those with HSAM would be spared. Their memories, instead of malleable and fluid, should be etched in stone. But is this really the case?

Screen Shot 2013-11-23 at 11.38.28 AMResearchers from UC Irvine recruited 20 HSAM participants and compared them to 38 age- and sex- matched controls. In the first test, they asked the volunteers to remember a string of related words, such as “light”, “shade” and ”table”. What was missing from the list was the highly similar lure word “lamp”. During the test, when volunteers were asked whether they had seen a particular word in the list, around 70% of the control group (light grey, left graph) falsely remembered seeing the lure. Incredibly, HSAM volunteers (dark grey, left graph) did not fare any better.

Verbal memory is hardly autobiographical. Next, researchers showed the volunteers photos of two separate crimes – a man stealing a wallet and putting it into his jacket pocket, and a man jacking a car with a credit card. Forty minutes after the picture show, volunteers read 50 sentences describing each crime. Unbeknownst to them, 3 sentences included misinformation about minute details, such as “he put the wallet in his pants pocket” or “he used a clothes hanger to break in”. When tested an hour later, the volunteers not only misremembered details but also attributed them to the incorrect source. Astonishingly, those with HSAM were far from immune, generating significantly more false memories than controls.

Again, you may say, the above events did not happen to HSAM individuals themselves. Nor were they particularly disturbing – perhaps volunteers simply didn’t care enough to pay attention to detail? Emotionally charged memories tend to demand attention and stick around for longer than neutral ones; like most people, I have no trouble recalling what I was doing when I first heard of the plane crashes on September 11, 2001 (whether my recall is accurate, however, is another question).

Researchers tapped into this powerful memory. Specifically, they asked what the volunteers remembered about the crash of United Airline Flight 93 with a questionnaire. Hidden in the questions were sprinkles of misinformation and flat-out lies -for example, “a witness had filmed the crash on the ground and the film was subsequently aired”, when no such footage actually existed.

Screen Shot 2013-11-23 at 11.41.13 AM

Fig 3D and 3E from paper. Dark grey=HSAM, light grey=control. High PEQ=better memory on initial screening test.

As you can see in the graph above, later in an interview, roughly ~20% of HSAM participants (D left, dark grey) said they had seen the footage compared to 29% of controls (D left, light grey). The percentages were roughly the same if the fabricated video was mentioned off-hand by researchers during the interview (not shown). Even more mind-boggling is this: in the HSAM participants, those tested at top 50% for autobiographical memory accuracy performed worst; in fact, they were just as bad as the controls (D right, high PEQ=better autobiographical memory on initial test). The participants weren’t reporting vague feelings of “having seen the video” either; when probed further, they gave elaborate details:

Interviewer: OK, Can you tell me what you remember about the footage?
HSAM: Uh, I saw it going down. I didn’t see all of it. I saw a lot of it going down uh, on air.
Interviewer: Ok, do you remember how long the video is?
HSAM: Just a few seconds. It wasn’t long. It just seemed like some-  thing was falling out of the sky. It was probably was really fast,   but I was just, you know, kind of stunned by watching it you know, go down.
Interviewer: Ok, so now is the last question, I would like for you   tell me how well you can remember having seen the video on the scale from 1 to 10, where 1 means no memory at all and 10 means a very     clear memory?
HSAM: I’d say about 7.

So here’s the conundrum: we have a cohort of people with astonishingly accurate and detailed autobiographical memory, whom nevertheless are susceptible to misinformation and false memories. Perhaps, you may have already pointed out, it’s due to the small sample size. While certainly a problem in most studies, here researchers have consistently found false memories in HSAM individuals, thus proving the point that fallible memory is (somewhat) inescapable. In other words, increasing sample size would not reduce misremembering to zero percent and thus would not disprove their conclusion (it may help tease apart differences between low- and high- performing HSAMs though). The results would have been more difficult to interpret if they had not found false memories in those with HSAM.

The second problem is that the crash of United 93 is semiautobiographical in nature. The volunteers witnessed the event vicariously through multimedia rather than experienced them first-hand. Previous experiments with AJ, our first identified HSAM individual, showed that while “she could quickly, reliably and accurately tell what she was doing on a given date, she couldn’t recall specific events from a videotape the month before”. In other words, her –and presumably, other HSAMs- extraordinary memory are selective. Here the researchers show they could “implant” a fake factoid 11 years after the fact, and some HSAM individuals would incorporate the memory into their original set of memories of the crash. What I would love to know is whether researchers could distort the volunteers’ own experience of the event. That is, whether they can alter an existing autobiographical memory rather than introduce a new one.

If HSAM individual have these common flaws, how are they still capable of remembering trivial details from a decade ago down to a tee? Researchers think it’s because very little misinformation is generally introduced in their daily lives. “No one comes up to them and says March 2nd, 2001 was a Monday not a Friday,” said one researcher to National Geographic. I’m not so sure that’s true – I know every time I go down memory lane with my friends we recall things differently; misinformation abounds. Our personal accounts end up influencing each other’s memory of the event.


Regardless, HSAM is in and of itself a fascinating phenomenon. Along with many other studies that investigated memory distortion in people with typical memory, this paper suggest that fallible memory reconstruction (reactivation and reconsolidation) is a basic and widespread part of human memory, a global phenomenon deeply routed in our neurobiology. No one, not even those with superhuman memory, is spared. But who knows, maybe discoveries of future “memory experts” will prove this wrong.

For those interested in HSAM (how could you not?!), here’s a 60 Minute episode on the topic. Pleading to @radiolab to do a show on it.
Patihis L, Frenda SJ, Leport AK, Petersen N, Nichols RM, Stark CE, McGaugh JL, & Loftus EF (2013). False memories in highly superior autobiographical memory individuals. Proceedings of the National Academy of Sciences of the United States of America PMID: 24248358

#SfN13 Running boosts cognition by increasing aerobic capacity

Poster: 671.Learning and Memory: Genes, Signalling and Neurogenesis II.


Running for health. Source:

There’s no doubt that aerobic exercise benefits the brain. Running, for example, reduces anxiety, improves sleep quality, boosts learning of a new task and maintains spatial memory*. Many of these mental perks stem from an increase in adult neurogenesis; that is, the birth of new neurons in the hippocampus and the olfactory bulb. (*That is, if rats run before new learning. See here for more.)

Yet perhaps the most apparent health benefit of running is increased cardiovascular and lung function. As any runner can attest to, an initially exhausting 10k soon becomes a breeze – you’ve increased your aerobic capacity. This led researchers from Duke University to wonder: is improving exercise capacity –by whatever means – necessary and sufficient to boost neurocognitive function?

Better bodies, better minds

Just like us humans, rats have an innate sensitivity to the effects of exercise. After the same 8-week running regime, high-response rats drastically increased their maximal capable running distance (~75%), while low-response rats barely improved (~22%). Surprisingly, compared to their sedentary peers, only high-response rats showed elevated neurogenesis in the dendate gyrus, a subregion of the hippocampus, as compared to their sedentary peers.

One hypothesized function of the dentate gyrus is pattern separation, or VERY simply put the discrimination between two very similar spatial contexts or things (Jason Snyder of Functional Neurogenesis fame has a great blog post on the matter). Researchers decided to challenge these rats with two Lego pyramids that only differed in the colour of their tops – imagine two Christmas trees with either a yellow or orange star. After the rats familiarized themselves with the yellow-topped Lego, researchers waited a minute before presenting them with both. High-response runners (but not their sedentary controls) instantly realized something was up – they approached and sniffed the new construct in earnest, ignoring the old familiar one.

Low-response runners, on the other hand, behaved just like their sitting peers, spending a similar amount of time with both objects. Low-responders had no problem with their memory; when faced with a mug and a can, they could easily discriminate between the two. They just couldn’t pick out minute differences in the Lego pieces, a skill often attributed to enhanced neurogenesis.

These data, perhaps somewhat dishearteningly, suggest that running doesn’t always boost brainpower – neurocognitive benefits only occur in tandem with improvements in aerobic fitness, as measured by total running distance until exhaustion. These results parallel that of a human study, in which increased lung capacity after training correlated with better performance on a modified pattern separation task (although understandably they did not show enhanced adult neurogenesis, so it’s hard to attribute behavioural output to increased new neurons per se).

Running-improved aerobic capacity seems to be the crux to exercise-induced brain benefits. But is running really needed? To explore this idea further, researchers decided to take treadmills out of the equation and focus on genetic differences in aerobic fitness.

Innate aerobic capacity accounts for cognitive benefits


Rats on treadmills. Source:

Allow me to introduce to you the low and high capacity runners. Selectively bred for their capability (or not) to “go the distance”, these rats differ up to 3 times in a long-distance standard fitness test, without ever setting foot on a treadmill. At 10 months old, they also had a two-fold difference in the total number of newborn neurons in the dentate gyrus as a result of increased neuron survival, which increased to three-fold at 18 months old.

Researchers took sedentary rats from both groups and challenged them to the Lego task described above. High capacity runners significantly outperformed their low capacity peers, expertly telling apart the Lego constructs. Similarly, in an object placement task in which researchers minutely moved one of two objects, low capacity runners could not identify the moved one after an hour’s delay, though they managed if the wait was only a minute. High capacity runners, on the other hand, excelled in both cases.

These results argue that high aerobic capacity in and of itself promotes pattern separation. But what if, unbeknownst to researchers, high capacity runners were maniacally jumping around everyday in their home cages? A few days of stealthy observation proved this wrong; paradoxically, low – compared to high- capacity runners were much more hyperactive. They also seemed more outgoing in a social interaction test, and exhibited a lower tendency to generalize trained fear from one context to another.

Running-induced neurogenesis is generally considered to ease anxiety. So why do high capacity runners (with higher rates of neurogenesis) seem more neurotic?

Born to laze, born to run


Sitting on a couch is really not that stressful. Don’t make me run!

Running is physiologically stressful in that it increases the level of corticosterone (CORT), a stress-response hormone. Unlike chronic stress that continuously elevates CORT, running only induces a transient, benign increase that quickly returns to baseline after recovery.

Researchers trained low- and high- capacity rats on treadmill running 5 days a week for a month. By the end, both groups showed increased running capacity, though trained low-capacity rats were only as good as untrained high-capacity ones (life’s unfair!). However, their acute stress responses drastically differed in a running-stress test.

Untrained low-capacity rats remained calm throughout the test, as measured by unchanging CORT levels. “They waddled on the treadmill for a bit, got tired and gave up.” said the researcher, “so they really weren’t that stressed out.” Trained low-capacity rats however hated the treadmill – their CORT shot through the roof. “You’re chronically forcing them to do something they’re terrible at, of course they’re going to be stressed out” explained the researcher, “and once they’re done, their CORT goes back to normal.” (I’m paraphrasing.) While this scenario is certainly possible, an alternative explanation is that only trained low-capacity rats were able to exercise to the point to induce a normal elevation in CORT levels; untrained rats simply don’t workout hard enough.

Intriguingly, untrained high-capacity rats had elevated levels of CORT during the running test, while previous training eliminated this response. Why? Researchers believe that chronic running habituated them to the stressor: “You know when you have this itch to run? You get stressed out when you can’t, and feel relieved when you finally do exercise.” In other words, these rats were “born to run”.

On the cellular level, running did not significantly increase neurogenesis in the ventral hippocampus in either low- or high-capacity rats, which I find rather surprising. Finally, high-capacity rats (compared to low) had less Mmneralocorticoid receptor (MR) and glucocorticoid receptor (GR) in the amygdala and hypothalamus, but not in the hippocampus. This is also surprising, as MRs and GRs in the hippocampus are crucial for negative feedback to the stress response axis (below).

Screen Shot 2013-11-16 at 7.50.14 PM

Hippocampal GR negatively regulates the stress response. Source:

Taken together, these data point to increased aerobic fitness– through genetic means or exercise- as the key to enhancing neurocognitive function in rats. Inbred differences in aerobic fitness may alter how one responds to exercise (and perhaps other types of) stress.

These studies beg the question: what if we could artificially mimic the effects of exercise (pharmaceutically or otherwise) and reap its benefits? While “exercise pills” may not necessarily benefit healthy individuals, they could potentially improve both physical and hippocampal health of the elderly or the disabled.

Such research is under way, though as of now the results are not yet convincing.

PS. This is the end of #SfN13 blogging. It’s been hectic, a bit overwhelming and a LOT of fun!! Thank you to all the presenters for your patience & feedback and the PIs who let me write about your work. Thank YOU for reading!

Regular research blogging will resume soon. Stay tuned!

671.01. CL Williams et al. Rats selectively bred for high running capacity have elevated hippocampal neurogenesis that is accompanied by enhanced pattern separation ability. 

671.02. KM Andrejko et al. Rats selectively bred for high running capacity have elevated hippocampal neurogenesis that is accompanied by a greater expression of hippocampal glucocorticoid receptors and altered contextual fear conditioning. 

671.04. JM Saikia et al. Treadmill exercise training only enhances neurocognitive function if it is accompanied by significantly increases in aerobic capacity. Duke Univ., Durham, NC; Univ. of Michigan Med. Ctr., Ann Arbor, MI

#SfN13 Getting rid of an unwanted memory for good

Poster 99.06/JJJ40 – Gradual extinction prevents the return of fear. SJ Gershman, CE Jones, KA Norman, MH Monfils, Y NIV. Brain and Cognitive Sci., MIT, Cambridge, MA; Psychology, The Univ. of Texas at Austin, Austin, TX;Neurosci. Inst. & Dept. of Psychology, Princeton Univ., Princeton, NJ

Poster 99.07/JJJ41 Gradual extinction prevents the return of fear in humans. JW Kanen, SJ Gershman, MH Monfils, EA Phelps, Y NIV. Dept. of Psychology, Ctr. for Neural Sci., New York Univ., New York, NY;Dept. of Brain and Cognitive Sci., MIT, Cambridge, MA; Univ. of Texas, Austin, TX; Nathan S. Kline Inst. for Psychiatric Res., Orangeburg, NY;Princeton Neurosci. Inst. and Psychology Dept., Princeton Univ., Princeton, NJ

Psychiatrists have a problem. Memories, especially fearful memories, are exceedingly hard to erase. Say you’ve learned that every time you touch a doorknob in the winter you get a painful electrostatic shock; fairly soon you might form an irrational fear of the doorknob. What can the good doctor do?


Ouch! Source:

The go-to therapy is extinction training. Here, you’ll repeatedly touch a doorknob that’s been treated to eliminate static – hence, no shock. After several sessions you loose your phobia. Great. Yet a few weeks later, you once again feel butterflies fluttering in your stomach at the thought of touching a doorknob. Somehow, the fear has returned.

The above scenario may seem ridiculous; yet for those suffering from post-traumatic stress disorder or debilitating phobias, the spontaneous recovery of a fear memory is nothing to laugh at. Scientists aren’t quite sure why this happens. Erasing a memory, or memory extinction, in theory “updates” the original memory trace, such that a fearful stimulus (eg doorknob) is now encoded as safe. Yet in practice, when the object of fear suddenly dissociates from harm (eg no more shocks!), the mPFC generates a large prediction error signal, such that the new information is treated as something entirely new and encoded in a separate memory trace.

Herein lies the problem. The original fear memory, alive and well, competes with the opposing new one for expression. Behaviourally, this often results in the return of fear. But here’s the silver lining: if you keep the prediction error signal small, the brain may opt to modify the old trace rather than encode an entirely new one, thus mitigating or erasing fear in actuality.

Researchers decided to test this theory out. First, they taught a cohort of rats to fear a tone by associating it with a shock. The rats were subsequently divided into three groups: the first received gradual extinction, in which the frequency of shock delivery declined gradually. In other words, as the trials progressed, rats more often than not experienced the tone without the shock. In a sense, researchers “weaned” these rats off the tone-shock association, thus triggering a small prediction error. The second group received the opposite treatment, with the frequency of shock delivery steadily increasing until the last 9 trials, in which all shocks were omitted to facilitate extinction. This is necessary as researchers wanted to observe the return of fear. The third group went through normal extinction; that it, only tones, no shocks. By the end, all rats lost their fear of the tone.

Fast-forward a month. Researchers returned the rats to the testing chamber and played the same tone. Rats that underwent gradual extinction showed significantly less fearful behaviour than the other two groups. In another experiment, researchers re-established the fear memory by giving the rats another shock. Once again, those that had undergone gradual extinction showed less fear. These results strongly suggest that gradual extinction is especially effective at persistently decreasing fear. But how do these results transfer to humans?

Using a similar fear-conditioning paradigm, researchers presented images of snakes to a group of volunteers. This was followed by an uncomfortable electrical shock to the wrist. Fear, as well as other states of arousal, increases sweating and skin conductance, the latter of which was carefully monitored to objectively measure the volunteers’ level of fear. Within a day, volunteers learned to fear the innocuous snake image.

One day later, researchers divided volunteers into groups and eliminated their fear memory through one of the three extinction processes as above. When tested on the third day, those who went through gradual extinction showed a trend towards less spontaneous recovery; that is, they sweated less at the sight of the snake images. However, the results are preliminary and due to the small number size (10-14 per group), the effect was not yet statistically significant.

Clarifying the conditions that facilitate persistent fear extinction may help clinical psychiatrists optimize extinction-based exposure therapies for the treatment of anxiety disorders and phobias. The evidence presented here – from rat to human – strongly suggest that minimizing prediction error through gradual extinction is a more effective way to modify and erase a memory, maybe for good.

Sometimes slow is a better way to go.

#SfN13 Adult neurogenesis and the fluidity of memory flow

Poster III32. Adult neurogenesis protects against proactive interference.

JR Epp, R Silva Mera, LCP Botly, AC Gianlorenco, S Kohler, SA Josselyn, PW Frankland. Hospital For Sick Children, Toronto, ON, Canada; Federal Univ. of Sao Carlos, Sao Carlos; Univ. of Western Ontario, London, ON, Canada

Think about the last time you started up iTunes in search of a song. Every flick of your finger brought a new, dazzling piece of cover art into view. With one goal in mind, you fixed your gaze steadily on the centre of the screen, barely noticing as previous covers gradually drifted from your sight and disappeared from your mind.


Need a reminder? Cover Flow, iTunes. Source:

This is how I picture memory. As we go through our daily lives, new memories silently replace older ones that are similar in context and scheme. The neurobiology behind this “refreshing” of memory is mind boggling in complexity and not well understood. Now, researchers from the Hospital for Sick Children in Toronto, Canada have uncovered a potential, if somewhat surprising, mechanism – adult neurogenesis.

Throughout our adult lives, the olfactory bulb and the dentate gyrus constantly produce new neurons; rapidly in infancy then gradually slowing as we age. Increasing adult neurogenesis in the dentate gyrus, either through drugs or exercise, helps the brain encode and differentiate between two or more similar memories. As such, high rates of neurogenesis have always been considered a “good” thing: more computational power, better memory.

Yet as new neurons reach out and connect into an existing neural network, it inevitably disrupts old information stored within. Following this line of thought, could adult neurogenesis paradoxically deteriorate existing memories?

Researchers first trained adult mice on a spatial recognition task. Picture a large pool filled with murky water with a platform hidden just beneath the surface. Mice are efficient swimmers, but they prefer to relax on the platform (who doesn’t?) if given the chance. A few training sessions later, all the mice managed to find and remember the location of their resting spot. Researchers then separated them into two groups: the “couch potato” group was housed in a standard cage; the “runner” group was given a running wheel. Mice are like your average pet gerbils – give them a wheel and they’ll happily go at it for hours.

4 weeks later, compared to the couch potatoes, the runners had significantly more new neurons in their dentate gyrus. When challenged with the same water maze, a clear difference in performance emerged. Compared to sedentary controls, the runners had a hard time homing in on the right spot. They spent much less of their time circling waters close to the platform, suggesting that their memory of the location had deteriorated by a greater extent. On the contrary, when researchers used a chemical-genetic method to eliminate new neurons in another cohort of water maze-trained mice, they remembered the platform location even better than sedentary controls.

Before you go and throw your running shoes out the window, pause and consider this: far from disadvantaged, the runners’ memories were more “flexible”. When researchers covertly moved the platform to a new location, runners learned much faster than their sedentary peers. Conversely, mice with disrupted neurogenesis stubbornly clung onto their old outdated memory, taking longer than controls to find the new platform location in every single training session. These changes in memory flexibility weren’t a generic effect of running but a specific result of neurogenesis: when runners had their numerous new neurons ablated, they behaved just like sedentary controls – better retention, slower relearning.

Incredibly, neurogenesis only seems to help with learning new information that is highly similar to that previously remembered. Researchers trained mice to discriminate between two boxes: one was shaded on one side and smelled of coffee; the other was striped with a hint of cinnamon. A month later, researchers swapped the odours between the two boxes. Once again, mice that ran during the interim learned the reversed odour-box pairings faster than their sitting peers. However, when researchers trained these mice in a similar task with two completely new odour (ginger, thyme)-box pairings, both groups learned at the same rate.

Taken together, it seems that adult neurogenesis after learning weakens old memories, which in turn facilitates learning of new but similar information. This is not to say that adult neurogenesis induces a state of tabula rasa; the data clearly shows that existing memories are weakened, not completely wiped clean. In a sense, adult neurogenesis tips the static-fluid memory scale just enough so that new information about the environment can be incorporated, either through altering the original memory trace or the formation of a new one. Hence, we live and thrive in the present.

There are cases where we resist: when learning a new language, you often automatically refer to a more familiar language for guidance – a headache dubbed “proactive interference” by psychologists. This study suggests that maybe you should close your books and go for a run. Come back later and who knows? You just might learn that second language faster.

My previous posts on adult neurogenesis can be found here.