todd_c_sacktor's picture
Distinguished Professor of Physiology, Pharmacology, and Neurology, State University of New York Downstate Medical Center
Long-Term Memory Is Immutable

For over a century psychological theory held that once memories are consolidated from a short-term into long-term form, they remain stable and unchanging. Whether certain long-term memories are very slowly forgotten or are always present but cannot be retrieved was a matter of debate.

For the last 50 years, research on the neurobiological basis of memory seemed to support the psychological theory. Short-term memory was found to be mediated by biochemical changes at synapses, modifying their strength. Long-term memory was strongly correlated with long-term changes in the number of synapses, either increases or decreases. This intuitively made sense. Biochemical changes were rapid and could be quickly reversed, just like short-term memories. On the other hand, synapses although small were anatomical structures, visible under the microscope, and thus were thought to be stable for weeks, perhaps for years. Short-term memories could easily be prevented from consolidating into the long-term by dozens of inhibitors of different signaling molecules. In contrast, there was no known agent that erased a long-term memory.

Two recent lines of evidence have relegated this dominant theory of long-term memory ready for retirement. First is the discovery of reconsolidation. When memories are recalled, they undergo a brief period in which they are once again susceptible to disruption by many of the same biochemical inhibitors that affect the initial conversion of short- into long-term memory. This means that long-term memories are not immutable, but can be converted into short-term memory, and then reconverted back into long-term memory. If this reconversion doesn't happen, the specific long-term memory is effectively disrupted.

The second is the discovery of a few agents that do indeed erase long-term memories. These include inhibitors of the persistently active enzyme PKMzeta and of a protein translation factor with prion-like properties of perpetuation. Conversely, increasing the activity of the molecules enhances old memories. The persistent changes in synapse number that so strongly correlate with long-term memory may therefore be downstream of persistent biochemical changes. That memory erasing agents are so few suggests that there may be a relatively simple mechanism for long-term memory storage involving not hundreds of molecules as in short-term memory, but only a handful, perhaps working together.

Memory reconsolidation allows specific long-term memories to be manipulated. Memory erasure is extraordinarily potent and likely disrupts many, if not all long-term memories at the same time. When these two fields are combined, specific long-term memories will be erased or strengthened in ways never conceivable in prior theories.