Recall a memory under certain circumstances, and the brain might erase it, recent rodent research suggests. If that possibility seems like science fiction, consider other weird tricks played by the mind's memory machinery. False recollections, for example, can occur during a déjà vu experience or after hypnosis. And true recollections which can reconstruct experiences from decades earlier, often seem almost supernatural, even to those fully aware of the brain's complexity.

Because of its surpassing strangeness, memory has long frustrated philosophical and literary attempts to rationalize and describe it. Now, neuroscientists hope to explain how memory works, and they are deploying some of molecular biology's newest tools and experimental psychology's oldest methods in their quest. But the ultimate focus of recollection research, the memory trace, still resists intricate measurement and manipulation. As a result, the field progresses slowly, igniting intense, seemingly irresolvable controversies as it edges forward. A case in point...



© 2004 Annual Reviews

According to the standard model of synaptic consolidation (and perhaps reconsolidation), an animal's experiences produce information (left arrow) that triggers signal-transduction cascades within neurons. These cascades occur during a limited time period and involve various transcription factors, such as CREB, and so-called "early" genes. If protein-synthesis inhibitors and other agents do not disrupt this process, it ultimately leads to changes at the synapses that separate one neuron from another (bottom). (Reprinted from Annu Rev Psych, 55:51–86, 2004.)

From 1968 through the mid-1970s, several publications on rats and humans reported on whether electroconvulsive shock could cause amnesia for memories after they were retrieved. Nader contends that some of the findings supported reconsolidation; McGaugh disagrees, citing his own 1969 paper3 and later work by Larry R. Squire at UC, San Diego. After a dormant period, the field began to reawaken in 1999 when Susan J. Sara, at the National Center for Scientific Research in Paris, reported that rat memories, upon retrieval, deteriorate after administration of propranolol, a β-noradrenergic receptor blocker.4

Norepinephrine and Memory

Some neuroscientists heatedly debate whether reconsolidation exists as a discrete memory process. Now another controversy might be brewing over consolidation and retrieval, processes that precede reconsolidation (assuming it exists) in a memory's development.

The molecule at issue is norepinephrine (NE), a neuromodulator made in brainstem nuclei and released by axons into various forebrain structures, including the hippocampus and amygdala. Modulators influence a postsynaptic neuron's firing rates in response to a presynaptic neuron's release of neurotransmitters such as glutamate and GABA.

As a postdoc, Steven A. Thomas generated mice lacking an enzyme needed to produce NE. Now an assistant pharmacology professor at the University of Pennsylvania, Thomas recently investigated how these knockout animals respond to fear conditioning.1 Members of his lab looked at two types of recall: contextual memory, based in the hippocampus and amygdala, which causes rodents to become immobile, or freeze, when they're placed in an apparatus that formerly delivered a shock; and cued memory, amygdala-based, which prompts animals to freeze when they hear a shock-associated tone.

From the knockouts' lack of freezing, Thomas and his group deduced that NE is critical for retrieving contextual memories during a period lasting from two hours to four days after conditioning. But NE did not appear crucial either for the consolidation of contextual memories or for the retrieval of cued memories. Pharmacological experiments on nonknockout rodents bolstered the contextual-memory findings and implicated the involvement of the β1 adrenergic receptor, to which NE binds.

These results puzzle James L. McGaugh, director of the Center for the Neurobiology of Learning and Memory at the University of California, Irvine. Over the years, he says, many studies have demonstrated NE's role in enhancing memory consolidation. As for Thomas' finding that NE is needed to retrieve contextual memories, a new paper by a team including McGaugh and UC, Irvine colleague Benno Roozendaal suggests otherwise.

The researchers injected rats with corticosterone, which belongs to the family of glucocorticoid hormones whose levels rise in stressful situations. Corticosterone impaired retrieval, but concurrent administration of the drug propranolol, which blocks β-adrenergic receptors, prevented this effect.2 That an NE-blocking agent aided memory retrieval indicates that NE itself plays an inhibitory role.

Roozendaal proposes a possible reconciliation of the studies. "Under normal conditions, norepinephine might enhance memory retrieval," he says. "But if you already have elevated levels of glucocorticoids so it extremely increases the sensitivity of the noradrenergic system, then norepinephrine might become impairing." And McGaugh suggests a possible reason behind conflicting findings on NE and memory consolidation. "In all of our experiments, the conditions are phasic," he explains. "It is a turnoff or blocking of the [NE] receptors for a short time period." Knockouts, in contrast, involve a chronic, often lifelong condition that "is probably quite different."

According to Thomas, his study predicts that beta-blocker drugs, prescribed for various conditions including stage fright and cardiovascular disease, might weaken some human memories, an assertion that some clinicians might find questionable. His paper states that beta-blockers that don't enter the brain might be preferable for treating certain "peripheral" diseases.

Thomas hesitates, however, to speculate about whether exogenous NE could clinically strengthen memory retrieval. "In people, we really don't know what [norepinephrine's] role is," he stresses. "We're extrapolating from mice. And if we do extrapolate, we're not talking about all memories necessarily. We're talking about certain memories at certain times."

- Douglas Steinberg

"A distinct role for norepinephrine in memory retrieval," Murchison CF, Cell Vol 117, 131-43 April 2, 2004"A systematically administered β-adrenoreceptor antagonist blocks corticosterone-induced impairment of contextual memory retrieval in rats," Roozendaal B, Neurobiol Learn Mem , 2004 Vol 81, 150-4

Around this time, Nader was casting about for a more inspiring research project. He and fellow postdoc Glenn E. Schafe, then members of Joseph E. LeDoux's lab at New York University's Center for Neural Science, decided to test what would happen if they reactivated a rat's memory and administered a protein-synthesis inhibitor. They sounded a tone as they shocked rats' feet, causing the animals to become immobile, to freeze. Two weeks later, when the tone was sounded without the shock, 80% of the rats froze, signaling that the rodents had retrieved what's known as a "fear memory."

Some rats then received an infusion of the protein-synthesis inhibitor anisomycin into their amygdalae. When exposed to a tone a day later, only 40% of these animals froze, compared to 80% of the controls, which had been infused with artificial cerebrospinal fluid. Moreover, anisomycin did not decrease freezing unless it was preceded by memory retrieval. Nader, Schafe, and LeDoux proposed that retrieval destabilized consolidated fear memories. To endure, these memories required protein synthesis in the amygdala as part of a reconsolidation process.1

What physical changes occur when reconsolidation is blocked and a memory is lost? "We're not quite sure," Nader acknowledges. Supposing that new memories involve adding synapses between neurons, he speculates that protein synthesis during reconsolidation is needed to maintain those synapses. Nader also stresses that loss of memories can result from either a failure to store them or an inability to retrieve them. "I'm really agnostic," he says, about which of these two endlessly debated possibilities causes amnesia. He notes that the synapses that arguably vanish because of a reconsolidation blockade could have been contributing to either storage or retrieval.


After the LeDoux lab published its reconsolidation paper, related findings began popping up in papers on the hippocampus and on other organisms, including crabs, snails, sea slugs, chicks, and fish. And last October, Robert Stickgold and Matthew P. Walker, assistant psychiatry professors at Harvard Medical School, reported that reconsolidation occurs in humans, too.5

Human reconsolidation is difficult to detect because researchers cannot ethically dose their subjects with toxic protein-synthesis inhibitors. To avoid that taboo, Elizabeth A. Phelps, an associate psychology professor at NYU, says that she and LeDoux are conducting a human memory study in which they administer propranolol, normally prescribed for cardiovascular disorders. When rats in Sara's study received propranolol within two hours of memory retrieval, they showed far more amnesia in a maze task.4

Stickgold and Walker, however, skirted the drug issue. They trained their subjects to perform a finger-tapping task (#1) on day one and trained them to perform another such task (#2) the following day. On day three, subjects were retested in both tasks, and their performances improved markedly. But if on day two, subjects were retested on task #1 and immediately afterwards were trained to perform task #2, their performance of task #1 deteriorated sharply on the third day. Thus, training in the second task apparently interfered with reconsolidation of memory for the first task. Task #2 acted much as anisomycin infusions did in the LeDoux lab's experiments.

An unresolved issue is whether sleep, which seems to help consolidate new memories,6 plays a role in reconsolidating retrieved memories. To explore this idea, Stickgold says that he could ask subjects to retrieve task #1, learn task #2, and re-perform task #1 several hours later, all on day two. An alternative strategy would be to delay asking subjects to re-perform task #1 until day three and to prevent them from sleeping on the preceding night. "It would be so beautifully elegant," Stickgold muses, "to show that, if you sleep-deprive the night after reactivation and interference, that then you don't get the deterioration" in performance of task #1. But his small lab hasn't had the resources to carry out this experiment, he adds.


Some neuroscientists believe in a reconsolidation-like phenomenon but put their own spin on it. Cristina M. Alberini, an assistant professor in Mount Sinai School of Medicine's physiology and biophysics department in New York, says she's "not crazy about calling it reconsolidation because it's not exactly the same as consolidation," even though both processes require protein synthesis and the CREB transcription factor.

Alberini refers to the later process as reactivation. Using antisense, her lab discovered that consolidation requires expression of the transcription factor C/EBPβ, but reactivation does not. A newly published study also applied antisense techniques to show that consolidation depends on brain-derived neurotrophic factor but not the transcription factor Zif268, whereas reconsolidation requires Zif268 but not BDNF.7

Alberini and graduate student Maria H. Milekic learned that the age of retrieved memories affects their vulnerability to protein-synthesis inhibition. The task was inhibitory avoidance, which involves a more active rodent response than freezing. When memories were reactivated two days after their formation, they were far more sensitive to anisomycin than memories reactivated two weeks after formation.8 These findings are at odds with the LeDoux lab's data.1

Some findings pose a more fundamental challenge to reconsolidation theory. One recent study discovered that retrieved fear memories, though initially weakened by subcutaneous anisomycin injections, recover spontaneously after three weeks.9 (Nader says that an unpublished test of a slightly different kind of fear memory detected no recovery after 24 days.) Reconsolidation skeptics want to know why this temporary weakening occurs.


Experiments indicate that anisomycin itself is not the culprit, but memory extinction might be. Extinction occurs after an animal has learned to associate a tone with a shock. If researchers then continue to present the tone without a shock, the animal stops freezing when it hears the tone. A new memory has extinguished the original fear memory.

The problem with this explanation is that protein-synthesis inhibitors impair these new memories. Therefore, the drugs should boost, not weaken, freezing behavior. To investigate the interplay between extinction and reconsolidation, Silva and his former postdoc at UCLA, Satoshi Kida (now at Tokyo University of Agriculture), and colleagues recently sought out the factor that determines which of these processes predominates. The answer, to Silva's surprise, turned out to be how long a mouse was re-exposed to a chamber where it had once received foot shocks but was no longer being shocked.10

Three minutes of shock-free reexposure (SFR), preceded by an anisomycin injection, led to an interrupted-reconsolidation effect (less freezing than a control group exhibited); a 30-minute SFR led to more freezing, an interrupted-extinction effect. Silva hypothesizes that a short SFR prompts the animal's brain to try to "edit" its old fear memory by reconsolidating it; a long SFR, in contrast, creates a distinct memory whose novelty allows it to prevail over an older memory. The difference between extinction and reconsolidation, however, was not merely theoretical. The study also showed that extinction, but not reconsolidation, is blocked by pharmacological antagonism of a cannabinoid receptor or a particular type of calcium channel.


Courtesy of Lydia Kibiuk; Inset: Elizabeth A. Phelps, NYU

The amygdala and hippocampus are two of the various brain structures thought to underlie learning and memory, including fear conditioning. (Inset) Researchers presented human subjects with the image of a colored square (conditioned stimulus, or CS) while administering a mild shock. In a control experiment, the image of a square with a different color was not paired with a shock. When the CS alone (i.e., without a shock) was presented, the amygdala (circled) and the hippocampus (beneath it) became active, as shown by this functional magnetic resonance image. Differences between the CS-alone and control experiments appear in red, yellow, and orange.

Since 2001, the Weitzmann Institute's Dudai has published several groundbreaking papers on the anatomy and molecules underlying extinction. This work generally focuses on conditioned taste aversion (CTA), a situation in which a rodent learns to associate the taste of saccharine with an unpleasant, malaise-like feeling caused by a concurrent lithium chloride injection. Dudai and his colleagues recently reported that amygdalar areas needed to consolidate or extinguish CTA memories are not required to reconsolidate those memories.11

In another CTA study, members of Dudai's lab examined whether anisomycin would block a new extinction memory, so that rats' taste aversion would increase, or would block reconsolidation of the original memory, so that the aversion would weaken. The result hinged upon how intense the original conditioning was. After two training sessions, anisomycin prevented reconsolidation, and after one session, it blocked extinction.12 Dudai's theory is that stronger training renders the original memory dominant and active. "If it's active, it's plastic; if it's plastic, it's sensitive" to disruption, he explains.


© 2000 Nature Publishing Group

In a seminal study by Nader, Schafe, and LeDoux, rats learned to associate a tone (conditioned stimulus, or CS) with a foot shock (unconditioned stimulus, or US)(A). Two weeks later, the animals heard the tone alone and about 80% of them froze in fear (B). The researchers then infused artificial cerebrospinal fluid (ACSF) or the protein-synthesis inhibitor anisomycin into the rats' amygdalae. When the animals heard the tone a day later, those that had received anisomycin froze at a far lower rate than those that had received ACSF (C). The researchers hypothesized that anisomycin had disrupted reconsolidation of the rodents' fear memories. (Reprinted from Nature, 406:722–6, 2000.)

Dudai expects that reconsolidation proponents and skeptics will soon find common ground. "I feel that within two, three years, we'll be able to look back and say there is some consensus in the field," he says. LeDoux is less sanguine. Recalling that the field of memory consolidation "went through decades of controversy," he fears that "we may just hit a brick wall."

Douglas Steinberg dsteinberg@the-scientist.com

Possible PTSD Treatment

People with posttraumatic stress dis order persistently reexperience a trauma, avoid stimuli associated with it, and feel numb. As PTSD caseloads have risen because of the Sept. 11, 2001, attacks and the wars in Afghanistan and Iraq, researchers have grappled with how best to treat the disorder.1

Administering the β-adrenergic blocker propranolol just after a trauma can prevent or mitigate PTSD, according to studies by two groups that included psychiatrists Roger K. Pitman at Harvard Medical School and Charles R. Marmar at the University of California, San Francisco. But no one knows whether patients with longtime PTSD also might benefit from propranolol.

Reconsolidation theory provides a rationale for such a treatment. The theory postulates that memories, upon retrieval, become fragile and subject to disruption. Propranolol, moreover, can disrupt retrieved memories in rats.2 Some neurobiologists object, on various grounds, to reconsolidation theory, but "as far as human research is concerned, we're not too concerned about the fine points," says Pittman. What matters, he adds, is a therapeutic outcome, even if it's "only for a limited period of time."

Joseph E. LeDoux, at New York University's Center for Neural Science, is starting a study of reconsolidation and PTSD. Pitman notes that his group is "putting our toes in that water" and might soon launch a similar investigation.

Michael Davis, a psychiatry professor at Emory University School of Medicine in Atlanta, sounds a cautionary note, however. In experimental situations, he asserts, propranolol generally blocks hippocampus-based memories but not fear memories that reside in the amygdala. So Davis offers a hypothetical scenario of what might happen to raped women treated with the drug. "When they see a man, they actually may be very fearful, but they don't know why because they can't remember they were raped," he says. "And clinically that would be very bad, because it would make that memory inaccessible for psychotherapy."

- Douglas Steinberg

"Understanding PTSD takes on urgency," Steinberg D, The Scientist Vol 1522, 1 Nov. 12, 2001"Attenuation of emotional and nonemotional memoires after their reactivation: role of β-adrenergic receptors," Przybyslawski J, J Neurosci , 1999 Vol 19, 6623-8

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?