Memory and Cellular Control Processes
235
to be a basis for long-term memory. A cogent objection to it is the small
effect on well-established memories of electroconvulsive shock treatment
or grand mal epileptic fits. One would expect either of these to disrupt
orderly cycles of circulating action potentials and hence, if the theory were
correct, to erase all memory.
Other mechanisms of a macroscopic kind involve permanent alterations
of cellular membrane properties, growth of new axonal or dendritic branches and growth of new synaptic knobs or alteration in the size of old ones.
However, in each of these examples, although the final result can be described only in macroscopic terms, the alteration might be triggered and
possibly maintained by a simple molecular mechanism [16, 51].
At the molecular limit it is possible to distinguish between two broadly
different kinds of theory. Both would suppose that memory is represented
by some lasting change in the metabolic balance of the cell but one would
implicate the control processes operating at the DNA-RNA-protein
synthesis level [16, 30, 47, 51], and the other would not. An example of
the second would occur if, after each time a nerve cell fired, the restoration
of the membrane potential was coupled biochemically to the synthesis of a
small quantity of a compound X. If X were metabolically stable and only
synthesized in these circumstances, then the total quantity of X in a nerve
cell would record the total number of times that cell had fired. The first
kind of molecular theory is probably the most interesting at the present
time, both because of current concern with genetic regulatory mechanisms
in cells and because it gives one a framework within which one can think
precisely and constructively. Accordingly we shall concentrate our attention
upon it as giving potentially the most fruitful approach although, in our
present state of ignorance, we should not entirely ignore the other possible
types of theory.
b) Quantity Requirements
Various authors have made attempts to give approximate estimates of the
memory capacity in "bits" of the human brain. There are two fundamentally
different assumptions upon which such estimates can be based. One is
that the brain stores a record of the sequence of action potentials arriving
from every single input cell. The other is that the storage system is arranged
so that only pieces of information which would be of interest at the "psychological" or "whole organism" level are extracted from the input and stored.
The difference is rather like that between the storage in picture form of a
letter, such as A, on video tape and its retention in coded form as one
character on a five hole paper tape record. Obviously, much more capacity
is needed in a store operating in the first way than in the second.
I do not think we can be sure at the moment which of these extremes is
nearer the truth for the human brain. However, the second seems more
235
to be a basis for long-term memory. A cogent objection to it is the small
effect on well-established memories of electroconvulsive shock treatment
or grand mal epileptic fits. One would expect either of these to disrupt
orderly cycles of circulating action potentials and hence, if the theory were
correct, to erase all memory.
Other mechanisms of a macroscopic kind involve permanent alterations
of cellular membrane properties, growth of new axonal or dendritic branches and growth of new synaptic knobs or alteration in the size of old ones.
However, in each of these examples, although the final result can be described only in macroscopic terms, the alteration might be triggered and
possibly maintained by a simple molecular mechanism [16, 51].
At the molecular limit it is possible to distinguish between two broadly
different kinds of theory. Both would suppose that memory is represented
by some lasting change in the metabolic balance of the cell but one would
implicate the control processes operating at the DNA-RNA-protein
synthesis level [16, 30, 47, 51], and the other would not. An example of
the second would occur if, after each time a nerve cell fired, the restoration
of the membrane potential was coupled biochemically to the synthesis of a
small quantity of a compound X. If X were metabolically stable and only
synthesized in these circumstances, then the total quantity of X in a nerve
cell would record the total number of times that cell had fired. The first
kind of molecular theory is probably the most interesting at the present
time, both because of current concern with genetic regulatory mechanisms
in cells and because it gives one a framework within which one can think
precisely and constructively. Accordingly we shall concentrate our attention
upon it as giving potentially the most fruitful approach although, in our
present state of ignorance, we should not entirely ignore the other possible
types of theory.
b) Quantity Requirements
Various authors have made attempts to give approximate estimates of the
memory capacity in "bits" of the human brain. There are two fundamentally
different assumptions upon which such estimates can be based. One is
that the brain stores a record of the sequence of action potentials arriving
from every single input cell. The other is that the storage system is arranged
so that only pieces of information which would be of interest at the "psychological" or "whole organism" level are extracted from the input and stored.
The difference is rather like that between the storage in picture form of a
letter, such as A, on video tape and its retention in coded form as one
character on a five hole paper tape record. Obviously, much more capacity
is needed in a store operating in the first way than in the second.
I do not think we can be sure at the moment which of these extremes is
nearer the truth for the human brain. However, the second seems more