Inverse relationship between working memory and neurogenesis

This is interesting. Researchers at Columbia have established that restricting neurogenesis in the hippocampus improves working memory:

New research from Columbia University Medical Center may explain why people who are able to easily and accurately recall historical dates or long-ago events, may have a harder time with word recall or remembering the day's current events. They may have too much memory -- making it harder to filter out information and increasing the time it takes for new short-term memories to be processed and stored.

Published in the Proceedings of the National Academy of Sciences (March 13, 2007 issue), the research reinforces the old adage that too much of anything -- even something good for you -- can actually be detrimental. In this case, the good thing is the growth of new neurons, a process called neurogenesis, in the hippocampus, the region of the brain responsible for learning and memory.

Results of the study, conducted with mice, found that the absence of neurogenesis in the hippocampus improves working memory, a specific form of short-term memory that relates to the ability to store task-specific information for a limited timeframe, e.g., where your car is parked in a huge mall lot or remembering a phone number for few seconds before writing it down. Because working memory is highly sensitive to interference from information previously stored in memory, forgetting such information may therefore be necessary for performing everyday working memory tasks, such as balancing your check book or decision making.

"We were surprised to find that halting neurogenesis caused an improvement of working memory, which suggests that too much memory is not always a good thing, and that forgetting is important for normal cognition and behavior," said Gael Malleret, Ph.D., a research scientist at the Center for Neurobiology and Behavior at Columbia University Medical Center and the paper's co-first author. "Altogether, our findings suggest that new neurons in the hippocampus have different, and in some cases, opposite roles in distinct types of memory storage, and that excess neurogenesis can be detrimental to some memory processes."

...

Many scientists had believed that neurogenesis in the hippocampus, and specifically, the dentate gyrus region, was wholly beneficial to memory. Previous research by Dr. Malleret with co-first author Michael D. Saxe, Ph.D., who was at Columbia when the research took place and is now at the Salk Institute in San Diego, Calif., found that reducing neurogenesis causes long-term memory deficits.

Based on this research, Drs. Malleret and Saxe hypothesized that the growth of too many new neurons could actually be more harmful than helpful to working memory. To examine this hypothesis, they designed working memory tests for two independent groups of mice in which neurogenesis in the hippocampus regions was suppressed. Results of the tests, in which mice had to locate food within specific areas of a maze, showed that mice in which neurogenesis had been halted made more correct choices and found the food faster.

The paper looked cool, so I went and read it. (You can to. It is available for free here.)

The researchers use two methods to ablate new neurons in the dentate gyrus of the hippocampus: x-ray irradiation and a transgenic mouse that can inducibly kill them. In both, they use tasks like the 8 arm radial maze with delay to look and see how neurogenesis impairs or improves working memory.

I have three comments about this paper:

1) The first thing you should be thinking when you read this paper is "does the task that they are using require the hippocampus"? The answer is yes.

If the task that you use to test an experimental manipulation does not require the structure you are manipulating, it makes the data rather difficult to interpret. You would be shocked how many papers ignore that fact.

2) When someone says that they are testing "working memory" you need to be very skeptical. Working memory is different than testing performance on a particular spatial task.

In a experimental sense, we define working memory in two ways: 1) resistance to delay and 2) resistance to interference. Resistance to delay means that working memory is taking place when the individual or animal can hold a memory for an extended period. Experimental manipulations that decrease performance after a delay is instituted are said to affect working memory. Resistance to interference means that working memory is taking place when an individual or animal performs well in spite of the presence of distracting information. Experimental manipulations that decrease performance in the presence of distraction are also said to affect working memory.

However, in this case using those two standards, we can actually argue that these experiments affected working memory.

What they find is that, in general, the improvement in performance for the animals where neurogenesis was prevented is not observed on increases in delay -- unless the delay is quite long and then not always -- but it is observed in the presence of distractors during the delay. (It would satisfy the second but not the first criterion I listed above.)

This suggests that the new neurons are playing a different role in working memory than older neurons, and that there introduction into the system provides some basis for interference. This would not be ridiculous. New neurons in the hippocampus do behave differently. The researchers mentions this idea:

The fact that opposite effects on working memory were found after eliminating neurogenesis than after more complete lesions of the dentate gyrus suggests that young dentate granule cells have a different function than mature granule cells. Recent computational models have suggested that the addition of new neurons to the stable network of the adult hippocampus may impact memory processes such as information encoding, storage, or retrieval adversely. This could result from the replacement of existing synapses by new ones. Moreover, because young neurons are more excitable than mature granule cells, their response to related but distinct stimuli, such as repetitive spatial representations, may contribute to memory interference or reduced discrimination but, at the same time, enhance encoding of highly distinct spatial information. (Citations removed. Emphasis mine.)

3) I would reiterate just because I think it is an idea present in popular culture that new neurons do not equal new memories. Neurons are not atomic units to which is assigned a new memory as they are created. The hippocampus is a neural network, and the properties of that network are changed by the addition of new members. However, there is not a one-to-one relationship between neurons and memories.

I hear people saying things like that sometimes, and it just makes me cringe. That is not what you should take from this work.