Significantly, neurogenesis takes place in only one region of the hippocampus - the dentate gyrus.
Our computational model imparts a unique role to this region in encoding the specific details of episodic memories. Moreover, the constant neural turnover in the dentate region ensures that each new event is encoded uniquely, without interfering with previously or subsequently stored memories.

The associational pathways in the CA3 and CA1 regions of the hippocampus can integrate this novel experience into prior learning episodes and perform associative retrieval. The unique feature of the new neurons that enables them to generate distinctive episodic memories without interference is their turnover.
This turnover relies on two processes: selective cell death, which eliminates redundant units, and maturation, which transforms young, plastic units into less plastic ones. Both groups are continuously replaced by neurogenesis; hence the turnover

Experimental manipulations that reduce the number of new neurons, such as irradiation , have contributed further to our understanding of possible functions of neurogenesis in the normal brain. Although many hippocampus- dependent tasks involve different aspects of associative memory, not every task that requires the hippocampus also requires the new neurons
For example, spatial learning by rats in the Morris water maze is disrupted by hippocampal lesions but not by irradiation . However, although irradiated animals learn the water maze at a normal rate, their long-term memory retention of the hidden-platform location is greatly impaired relative to that in controls when they are re-tested four or more weeks later. This finding is consistent with predictions of our computational model highly distinctive memories for individual episodes, thereby protecting them against retroactive interference.

In addition to this role in encoding specific details of events, the new neurons seem to be crucial for linking events across time
when these events are part of the same context.
Thus, animals that lack new hippocampal neurons show deficits on tasks that seem to require contextual memory abilities, including trace conditioning , contextual fear conditioning and delayed non-match to sample (DNMS) with long delays .
However, they perform normally on corresponding non-hippocampal control tasks: delay conditioning , cued fear conditioning and DNMS with short delays .These neurons also have a role in linking events across time when the events are part of a common context.

A novel proposal for the role of neurogenesis in temporal context: the functional cluster hypothesis Understanding the role of the new neurons in temporal coding requires a more elaborate model.
Traditionally, the hippocampus is thought to be responsible for associating multiple stimuli into a single episodic memory.
Synaptic integration of multiple inputs carrying sensory information can occur via spatial summation of individual synaptic potentials in dendrites of granule neurons.

Such synaptic responses are usually mediated by two principal types of glutamate receptors, AMPA and NMDA. AMPA is responsible for short-term interactions and NMDA for long-lasting changes in excitability, such as during learning.

However, temporal summation beyond the range of milliseconds cannot be explained using traditional biophysical mechanisms.
Temporal summation of events on the order of minutes, hours or days might be required to solve the learning tasks described here. Neurogenesis is ideally suited to encode such events; it is an ongoing process that begins with proliferation of neural precursors and ends with fully functional mature neurons . One striking feature of proliferation is that it occurs in clusters. The dividing precursors are often seen in groups of 2-10 cells, tightly packed in the subgranular zone (SGZ) of the dentate gyrus . These clusters disperse along the SGZ within several days, presumably by migration and/or attrition due to cell death. Differentiation of cells within the clusters into neurons is characterized by the expression of specific proteins, extension of axons and dendrites, and synaptogenesis .

Importantly, the excitatory influences, in the form of depolarizing GABA-mediated responses, are formed long before the new neurons integrate with the dense inhibitory circuitry in the dentate gyrus, which enables new neurons to sustain much higher activity levels than mature granule cells .
One can envisage 'waves' of neurons that respond to afferent stimulation and send impulses from neurons belonging to a cluster, via mossy fibres, to CA3 for association of their common inputs by CA3 axon collaterals.

New neurons within a cluster, innervated by different perforant path inputs, will respond to different features of an event.

Some will fire in response to persistent aspects of the environment, such as odours, stationary objects and boundaries, which we shall refer to as the context.
Other neurons might respond to more transient aspects, such as a tone or a shock. The highly plastic new neurons will become tuned to this constellation of features and should respond consistently when they experience the same context again. ,BR>Using plastic recurrent connections, targets in CA3 can link the transient features with the context, thus temporally linking items into a single episode. This enables cued recall of the entire event from a single item, which provides the basis of episodic-memory encoding and retrieval .

The new neurons will then either die or mature and become less plastic, which will protect the memory from interference by later learning. Subsequent events could be encoded by other 'waves' of generations of new neurons.
This assumption of 'superior plasticity' of the new neurons is consistent with a recently proposed model of a mechanism that separates ongoing experience into temporally tagged, unique event memories. More specifically, the cluster model proposed here (not to be confused with the 'clustered plasticity model' of Govindarajan et al (which is a single-neuron model) assigns a unique role to the clusters of cells born at approximately the same time and their impact on the encoding of event memories in CA3.

back to Home Page