Introduction:
The hippocampus (meaning seahorse, named for its curved
shape) is an important component of the brain. Much is still unknown
about the hippocampus, but it is now widely accepted that the hippocampus plays
important roles in
- Episodic memory,
- Navigation
Anatomy:
The hippocampus is an elaboration of the edge of the medial
temporal cortex. It is a paired structure and there is one found in each
hemisphere of the brain.
A number of adjacent areas of the brain interact in important
ways with the hippocampus and these areas all together are known as the
hippocampal formation.
A cross section of the hippocampus reveals these important
areas:
The major groups of cell bodies which make up the hippocampal
formation are:
- CA areas, these
together are the hippocampus proper, they are named Cornu ammonis, (meaning
rams horns due to their shape). There are four of these areas, and they
progress in sequence, CA4, CA3, CA3, CA1. The main cells in this part of
the hippocampus are pyramidal cells.
- The next area is the
dentate gyrus, so named because of its resemblance to a tooth. The cells
in this area are called granule cells, so named because they have tiny
cell bodies.
- The next area is the subiculum;
this is continuous with the Cornu Ammonis areas and is also composed of pyramidal
neurones.
- The final area is the entorhinal
cortex; this is an area of the cerebral cortex adjacent to the
hippocampus.
Circuitry:
All input into the hippocampal formation enters through the
entorhinal cortex. The axons of the entorhinal cortex project mostly to the
cells in the dentate gyrus (but also into CA3 and CA1). This fibre tract is
known as the perforant path (as it perforates the subiculum)
The cells in the Dentate Gyrus then project their axons
(known as mossy fibres) to the spiny dendrites of the cells in CA3 .
The cells in CA3 then send their axons (known as Shaffer
collaterals) to the cells in CA1. There are also many recurrent connections
with CA3.
The cells in CA1 then send their axons to the cells in the
subiculum. And the cells in the subiculum complete the loop by sending their
projections back to the entorhinal cortex.
This can be schematically represented as below:
Hippocampal
indexing theory:
One of the most popular theories concerning the role of the
hippocampus in episodic memory is hippocampal
indexing theory.
This theory states that;
When we have a conscious experience, many different areas of
the neocortex are activated, corresponding to the different aspects of that
experience. (for example the visual cortex for the visual aspect the auditory
cortex for the auditory aspects, ect..)
When we remember then remember that episode later on, similar
areas of the neocortex are reactivated, and this is what results in our
re-experiencing of the event.
The hippocampus acts as an index, storing the different
patterns of neocortex activity, associated with all our different memories.
The process is thought to be as follows;
The Entorhinal
cortex receives inputs from all parts of the neocortex, in a compressed manner.
It then projects through the perforant pathway projects
to the dentate gyrus, and then from the dentate gyrus to CA3.
The CA3 autoassociator:
The CA3 area is thought to act as a autoassociator, due to
its dense reciprocal connections.
Each pattern of neocortex activity is results in a unique
input pattern activating a unique subpopulation of neurones. Due to the dense reciprocal
connections, when these neurones are activated at the same time, the connections
between them are strengthened.
This allows for a process called pattern completion. in the
future whenever a feature of the original experience is presence, it actives a
portion of the previous neurones, these then activate any other neurones they
are strongly connected to, allowing recall of a whole memory from just a part
of an experience.
The dentate
gyrus pattern separation:
However in order for the CA3 autoassociator to work
correctly, inputs need to be relatively unique. If inputs to an autoassociator
are very similar, interference can occur.
For example:
Imagine we
have autoassociator which is presented with the number sequences 3,4,5 and
5,6,7. Each activate a unique set of neurones and result in separately indexed
memories.
Now when pattern completing, there is an overlap between the
two inputs. This means the outputs may be either of the memories and the
network will get confused. To avoid these we need a process known as pattern separation.
This is essentially where a network will take in two patterns, and produce an
output which is less similar.
The mechanism of pattern separation in the dentate gyrus is
not completely understood, but multiple possibilities have been proposed.
One possible mechanism may be that there are many more cells
in the dentate gyrus which project onto relatively fewer cells in CA3, meaning
the chance that any two populations of cells in the dentate gyrus project to
the same cells in CA3 is low.
Therefore the dentate gyrus ensures that unique populations
of cells get activated each time.
However evidence suggests this may not be the only mechanism;
differences in spiking frequency are another possibility.
The dentate gyrus cells change their rate of firing with new
information, with some increasing their frequency and some decreasing their
frequency. The higher frequency a DG neurone fires the more neurotransmitter it
releases and the more likely it is to activate the downstream CA3 neurone, so
changes in frequency with some DG neurones increasing their frequency and some decreasing
their frequency could also activate unique populations of cells
Hippocampal
output and memory recall:
So now we’ve seen how a memory consist of patterns of
neocortical activation, which can be condensed and sent to the entorhinal
cortex, undergo pattern separation and then bound together and indexed in the
CA3 autoassociator. Then, when a feature of the original stimulus is presented
it activates a subpopulation of the original neurones activated in CA3 and
their recurrent connections allow reactivation of the remaining neurones making
up the pattern.
The final step is to see how the hippocampus reactivates the
appropriate areas of the cortex.
This is thought to be mediated by CA1.
The Entorhinal cortex not only presents to the DG but also to
CA1,
This means that when the neurones
are activated in CA3, also activated in CA1 is another representation of the
cortical pattern. As these two populations of neurones are activated at the
same time they undergo synaptic plasticity and the connections between them are
strengthened. This means that the neurones from CA3 activate the neurones in
CA1 corresponding to the correct cortical areas. These then project back to the
entorhinal cortex, which has reciprocal connections to many areas of the
cortex, reactivating the same combination of cortical areas as the input and
causing us to re-experience the event as a memory.
