DEJA VU
by Robert M. Best
Deja vu (French for "already seen") is the illusion of remembering scenes or
events that are actually being experienced for the first time. This is
common enough that normal people have experienced it. Here is my
explanation for this illusion.
To benefit from experience, we must recognize prior situations in our lives
whenever we encounter them again. Episodic memory would be useless if each
repetition of a recurring situation were experienced as entirely new. But
each recurrence is almost never exactly the same. If our brains required
that a present situation match every trivial detail in a prior situation to
qualify as similar, we would never learn from experience. The brain
mechanism that recognizes a present situation as similar to a prior
experience must match only the essential features and disregard trivial
details.
There are several areas of cerebral cortex that recognize familiar patterns
such as faces, objects, sounds, odors, movements, etc. and these areas of
cortex are able to recognize patterns that are incomplete, distorted, noisy,
or altered. There is a corresponding brain mechanism that recognizes
similar situations in episodic memory, and equally important, it is able to
recognize situations that are entirely new, even when trivial details are
disregarded.
The brain formation that recognizes new episodes and commits them to memory
is the right hippocampus (HC). In the book "Neural Networks and Brain
Function" by Edmund Rolls and Alessandro Treves, the authors describe in
great detail how the parts of the HC operate (pages 95-135). Comparing each
episode to prior experience is performed by a HC layer of neurons called
CA3. Inputs to CA3 are first preprocessed by another layer of HC neurons
called the dentate gyrus (DG). The DG neurons act as a competitive learning
neural network which removes redundancy and trivia from detailed inputs and
outputs a small set of signals representing categories of essential
features. CA3 matches and recalls prior episodes based on this set of
categories and signals that a match has occurred or not occurred.
By removing redundancy and compressing the input pattern to essential
features, the DG discards information, just as compression of a digital
picture in a JPG file discards information. Without details, sooner or
later a new pattern will match a prior pattern based on only the essential
features, but not match the details in the uncompressed memory. Since the
details do not match, the prior memory does not become conscious, but the HC
has already signaled that the pattern is a familiar one because the
essential features do match. This is the deja vu feeling. We feel that a
new situation has occurred before but cannot recall when or where.