views updated


Contrary to a still widely held belief, the adult brain routinely generates new neurons. Strikingly, one of the two brain regions to which this adult neurogenesis is apparently restricted is the hippocampus. The hippocampus plays a key role in learning and memory, so the role that new neurons can contribute to its function is an inviting area for research. Most current theories on how the hippocampus processes information for storage consider the brain to be a network that is static at the level of neuronal numbers and plastic only at the level of neurites and synapses. However, mounting evidence indicates not only that the hippocampus produces new neurons but also that adult neurogenesis is tightly linked to hippocampal function. New neurons might contribute not only to hippocampal function, but they might also be indispensable to it.

What Is Adult Neurogenesis?

Neurogenesis is the development of neurons from neural stem or progenitor cells (see Figure 1). This developmental process begins with the division of the stem or progenitor cell and ends with a mature and functioning new neuron. Neurogenesis of the vast majority of neurons occurs during intrauterine development and ceases postnatally. In the hippocampus and the olfactory system, however, neurogenesis continues throughout life. Compared to the billions of neurons in the brain, the number of new neurons produced in the postnatal brain is very low and even decreases with age. In the hippocampus new neurons are generated only in the granule cell layer of the dentate gyrus, which increases in size with age. Under special circumstances single new neurons might be generated in brain regions outside the hippocampus or the olfactory system, most notably in the neocortex. It would be interesting to learn whether cortical neurogenesis under normal conditions would could contribute to learning and memory, because memory is stored in the neocortex. Elizabeth Gould from Princeton University has reported adult neurogenesis in the neocortex of primates, but Pasko Rakic and David Kornack from Yale University published a study in which they found no evidence of adult cortical neurogenesis. Although the function of neurogenesis in the olfactory system is not known, it appears that adult neurogenesis pertaining to learning processes occurs only in the hippocampal dentate gyrus.

Adult hippocampal neurogenesis in rats was first described by Josef Altman in 1965, but at that time the new neurons were considered to be a curiosity or an atavism. The methods then available did not allow a quantitative approach and a detailed phenotypic analysis. Consequently, adult hippocampal neurogenesis was "rediscovered" several times, most notably by Michael Kaplan in the late 1970s and in the early 1990s by Elizabeth Gould, then with Bruce McEwen at Rockefeller University in New York City. Georg Kuhn at the Salk Institute in La Jolla, California, was the first to use confocal microscopy in connection with immunofluorescent labeling techniques to investigate adult hippocampal neurogenesis. As a result, new cells could be unambiguously identified as neurons, and quantitative analyses became more feasible. In 1998 Peter Eriksson from Gothenborg, Sweden, used this technique to demonstrate that adult hippocampal neurogenesis occurs in humans.

In a parallel development, Fernando Nottebohm and his coworkers from Rockefeller University studied adult neurogenesis in birds. They were the first to add a strong functional context to this research. They found that song learning in adult canaries correlates strongly with adult neurogenesis in the brain center responsible for this behavior. They later showed that, in some food-storing bird species, neurogenesis is enhanced in those seasons in which the birds have to remember the location of their food caches.

How Is Adult Hippocampal Neurogenesis Regulated?

The development of a functional interpretation of adult hippocampal neurogenesis in mammals seemed more complex. Gould's early work emphasized a suppression of adult neurogenesis arising from chronic, severe stress and mediators of a stress reaction such as NMDA receptor activation and glucocorticoids; on this account adult hippocampal neurogenesis was due to stress-related damage to the hippocampus. Did this explanation imply that under normal conditions adult hippocampal neurogenesis contributes to hippocampal function? The late 1990s brought a wealth of studies showing the effects of numerous manipulations on adult hippocampal neurogenesis. While no conclusive theory of the regulation of adult hippocampal neurogenesis emerged from this research, it became apparent that adult neurogenesis reacts to a wide range of stimuli with great sensitivity but low specificity.

One solution to this puzzle lies in the definition of neurogenesis. The generation of a new neuron from a neuronal stem or progenitor cell entails a series of developmental steps. What is sometimes seems to be an effect on neurogenesis is often an impact on cell proliferation in which there is no certainty of a detectable change in the number of new mature neurons at a later time. Regulation of adult neurogenesis, however, occurs on several levels of neuronal development, including cell proliferation. Progenitor-cell proliferation might represent a rather nonspecific part in this regulation and can be influenced by many different factors.

Learning stimuli and experience, however, apparently do not influence progenitor-cell proliferation in the dentate gyrus but exert a survival-promoting effect on the progeny of the dividing cells. Animals that live in a so-called "enriched environment" (e.g., a cage providing much more sensory, physical, and social stimulation than the regular housing in laboratories) have a net increase in neurogenesis. This increase did not show a mandatory correlation with changes in cell proliferation, but was due to a survival-promoting effect on the progeny of the dividing cells. In old mice, no effect on the total number of surviving cells was detectable, but a phenotypic shift among these cells occurred toward more new neurons and at the expense of new glial cells.

Physical activity enhances adult neurogenesis. In addition to providing a hint of potential links between physical and mental activity and health, this finding suggests that active participation, in contrast to passive exposure to sensory stimuli, might affect adult neurogenesis. This finding dovetails with psychological learning theories that also emphasize active exploration and pursuit. Together these findings support the hypothesis that, in response to functional demand, new neurons can be recruited into hippocampal neuronal circuitry. The cellular machinery needed to achieve this goal goes far beyond a nonspecific mitogenic effect on the progenitor cells in the dentate gyrus.

Hippocampal Function and Adult Neurogenesis

The question of how the new neurons contribute to hippocampal function and thereby to learning and memory is linked to the more fundamental issue of the overall function of the hippocampus. Although the hippocampus is possibly the best-investigated structure in the brain, with a rich store of data available from researchers in anatomy, molecular biology, biochemistry, electrophysiology, and behavioral psychology, no "grand unifying theory" of hippocampal function has yet emerged. The hippocampus is classically characterized as "the gateway to memory" through which all information must pass to be remembered. Careful analysis of famous patient H.M. and animal research have indicated that bilateral hippocampal damage results in the inability to acquire new memory. Recall of older information, however, is spared. There is nonetheless a brief period of retrograde amnesia prior to the damage, indicating that a certain amount of time is needed for storage of information in cortical regions. During this time the hippocampus processes information for storage.

Not all kinds of information require hippocampal treatment. While declarative memory is considered hippocampus-dependent, procedural information, such as brushing your teeth (which is much more difficult to "declare" in words than to do), can be learned without hippocampal involvement. In the context of experimental research with animals, declarative memory is difficult to assess. Spatial navigation, however, is hippocampus-dependent and can be easily investigated in animals. Certain types of conditioning tasks are hippocampus-dependent, too, and therefore tests of spatial memory and conditioning tasks dominate animal experiments on hippocampal function.

In 2000 Tracey Shors, in collaboration with Elizabeth Gould, showed that the elimination of cell proliferation in the dentate gyrus by means of a cytostatic drug is followed by reduced performance on a hippocampus-dependent conditioning task, whereas performance on the hippocampus-independent form of the test was spared. While this result conforms well to theories linking adult hippocampal neurogenesis to hippocampal function, the study does not explain the function of the new neurons. This problem is similar to one faced in many knockout studies. Elimination of a particular gene often provides strong evidence that this gene is involved in a biological process, even that it is necessary for this process, but it rarely explains the specific nature of this involvement. One reason why the Shors study does not yet provide the full answer is that hippocampal function certainly goes beyond mediating eyeblink conditioning. In this sense the study even raises a new, puzzling question: Why would the hippocampus require new neurons for a task as simple and redundant as eyeblink conditioning?

Regarding spatial memory as an indicator of hippocampal function, our studies have shown that an increase in hippocampal neurogenesis correlates with an improvement on the Morris water maze task, a widely used test of hippocampal function in relation to spatial memory. However, this correlation never seemed to be linear. Levels of adult neurogenesis do not seem to be indicative of memory as storage capacity. Rather, adult hippocampal neurogenesis seems linked to the learning process itself; differences between animals living in an enriched environment emerged during the acquisition phase of the test. This makes sense if one considers the hippocampus as a gateway to memory, a processing unit that consolidates information for storage but not as the brain structure that provides long-term memory. But how can new neurons contribute function to this processing unit?

Joe Z. Tsien's group from Princeton University found that presenilin-1 knockout mice had reduced experience-induced neurogenesis without any signs of impaired memory formation. The group reasoned that new neurons might be necessary to "forget" old information, in the sense that the hippocampus would periodically need to clear the structure from "outdated hippocampal memory traces after cortical memory consolidation." This is a new twist on the idea that new neurons increase (or modify) the processor's working memory (RAM in a computer analogy). New neurons would allow the hippocampus to clear the structure for the processing of new information. Our own theory, explained below, is similar in emphasizing the modification of the processor network. In our view, however, cells need not be replaced to enable forgetting because information is not stored in single cells but is rather distributed over the synaptic weights in a network of neurons. Synaptic plasticity, which is essentially changing synaptic weights, is at the heart of the cellular mechanisms underlying memory. Moreover, synapses come and go. Altered connections cannot explain the recruitment of new neurons. But neurons are more than the sum of their connections. Neurons process information.

A Theory of New Neurons in Learning and Memory

If the hippocampus is the "gateway to memory," then adding new neurons in the hippocampus might amount to adding new gatekeepers to this portal (see Figure 1). The essentially three-synapse circuit within the hippocampal formation includes a bottleneck within the connection between the dentate gyrus and area CA3. The axons of granule cell neurons in the dentate gyrus form the mossy fiber tract, and adult neurogenesis adds new granule cells that rapidly seek connection to CA3.

Henriette van Praag and Alejandro Schinder at the Salk Institute in La Jolla studied the electrophysiological properties of newly generated granule cells and found that the new neurons are indeed functional. Information has to pass the narrows of the mossy fiber tract, and there appears to be a good reason for this forced concentration. A small network is fast and flexible. But the small network here has to cope with a huge range of input information. It has to adjust to novelty and complexity. If optimization here means having the smallest possible network that can handle the situation, and if single new neurons can be strategically introduced to the network, it becomes clearer how even a low rate of adult neurogenesis can be beneficial for the functional system. This view is also compatible with the fact that neurogenesis decreases in older age. If adult neurogenesis were to add units of memory to the hippocampus, old animals, who learn quite well, would have to rely on new neurons as much as younger animals. If the modification of the network were cumulative, however, older animals would require fewer new neurons because, given their greater experience, optimization of their network is already further advanced and requires fewer strategic additions.

The "new gatekeeper" theory might explain how the hippocampus could benefit from adult neurogenesis. It does not explain what the new neurons actually do. So far no conceptual links have been established between adult neurogenesis and existing theories about the many specialized hippocampal neurons, most notably place cells, with their response to a specific, recognized spatial pattern.

There is strong evidence that new neurons play a role in hippocampal function and thus in learning and memory. Their contribution most likely lies in a refinement of the processing unit, not in an extension of the storage capacity. But the precise manner in which this refinement takes place (and how a growing hippocampal network actually works) remains a question for current and future research.



Altman, J., and Das, G. D. (1965). Autoradiographic and histologic evidence of postnatal neurogenesis in rats. Journal of Comparative Neurology 124, 319-335.

Feng, R., Rampon, C., Tang, Y. P., Shrom, D., Jin, J., Kyin, M., Sopher, B., Martin, G. M., Kim, S. H., Langdon, R. B., Sisodia, S. S., and Tsien, J. Z. (2001). Deficient neurogenesis in forebrain-specific presenilin-1 knockout mice is associated with reduced clearance of hippocampal memory traces. Neuron 32, 911-926.

Gould, E., Tanapat, P., Hastings, N. B. and Shors, T. J. (1999). Neurogenesis in adulthood: A possible role in learning. Trends in Cognitive Science 3, 186-192.

Greenough, W. T., Cohen, N. J., and Juraska, J. M. (1999). New neurons in old brains: Learning to survive? Nature Neuroscience 2, 203-205.

Kempermann, G., and Gage, F. H. (1999). New nerve cells for the adult brain. Scientific American 280, 48-53.

Kempermann, G., Kuhn, H. G., and Gage, F. H. (1997). More hippocampal neurons in adult mice living in an enriched environment. Nature 386, 493-495.

Rakic, P. (2002). Neurogenesis in adult primate neocortex: An evaluation of the evidence. Nature Review Neuroscience 3, 65-71.

Shors, T. J., Miesegaes, G., Beylin, A., Zhao, M., Rydel, T., and Gould, E. (2001). Neurogenesis in the adult is involved in the formation of trace memories. Nature 410, 372-376.

van Praag, H., Kempermann, G., and Gage, F. H. (2000). Neural consequences of environmental enrichment. Nature Review Neuroscience 1, 191-198.

van Praag, H., Schinder, A., Christie, B., Toni, N., Palmer, T. D., and Gage, F. H. (2002). Functional neurogenesis in the adult hippocampus. Nature 415, 1,030-1,034.


Fred H.Gage