Professional Documents
Culture Documents
Frank Saunders
31 Jan 2011
In this paper, I will argue in favor of an analog model of memory. I will first explore the
commonsense notion of memory and expose its problems and the reasons behind them. I will
then present the philosophical distinction between digital and analog memory. From there,
I will present what I will call the analog theory of memory and its various implications, and
argue for its validity using evidence ranging from philosophy to neuroscience.
1
The Analog Theory of Memory 1.1 Physical Limitations
these traditions are perfectly acceptable. However, accepting the ideas that memories exist
within the brain regardless of our awareness of them raises some serious issues.
2
The Analog Theory of Memory 1.1 Physical Limitations
human thought ability, which is even less likely based on what we know about the slow and
gradual nature of evolution. Furthermore, the only physical link between two organisms
brain function is brain size, which would imply that if our “hard-drives”’ capacities had
grown one hundredfold, our brains would be much larger now than they were 80,000 years
ago, which is not the case. Unfortunately, this is what we must accept if we believe that
memories themselves exist within the brain like files on a computer.
An obvious objection to my argument can be posed at this point. Why should we accept
that humans remember now more than they did hundreds of thousands of years ago? After
all, in order to stay alive, humans had to be extremely aware of their surroundings. Our
knowledge of how to navigate a subway terminal or program a computer may very well be
qualitatively equivalent with primitive people’s knowledge of a particular area of land or how
to hunt effectively.
This objection, however, does not pose a threat; it merely calls for my pushing the
timescale back a bit and expanding the argument. Conscious memory occurs in the brain.
The brain is an evolved organ. Therefore, memory should be subjected to the same laws of
natural selection that all other life functions are subjected to. If memory is physical, and
memories are “written” in the brain, then there should be a direct correlation between the
ability to remember and brain size.
So far as my first example is concerned, if we accept that humans hundreds of thousands
of years ago remembered less than they do today, then memories cannot be physical, since no
physical difference in the two brains exists. If one does not accept the first premise, however,
one can easily turn to the animal kingdom for more compelling evidence.
The recent research in the cognitive abilities of animals, most notably birds, shows that
memory and brain size do not always correlate, which is more than enough evidence to prove
my point. For years, birds were believed to be incapable of intelligent behavior or complex
learning due to “the poor development in birds of any brain structure clearly corresponding
to the cerebral cortex of mammals,” which, “led to the assumption among neurologists not
only that birds are primarily creatures of instinct, but also that they are very little endowed
with the ability to learn,” (Thorpe 1964, p. 336).
It is becoming very clear to scientists, however, that this simply is not the case. As it
turns out, birds do have the ability to learn and remember and their abilities far exceed
what their brain sizes would imply. “Pigeons could maximally memorize between 800 and
1,200 picture-response association before reaching the limit of their performance. In contrast,
baboons minimally memorized 3,500-5,000 items and had not reached their limit until after
3
The Analog Theory of Memory 1.1 Physical Limitations
more than 3 years of testing.”1 What is interesting to note, however, is that a pigeon’s brain
weighs on average 0.4 grams whereas that of a baboon is on average 137 grams (over 300
times larger than the pigeon brain).2 However, the difference in memory ability between the
two animals does not reflect this large of a difference in brain size. Studies like these show
that brain size and memory do not necessarily correlate. This is as compelling evidence as
any that memories cannot be physical properties written into an organism’s brain.
Another obvious objection could be made based off of my response to the first one; since
there is a limit to the amount that baboons and pigeons (and humans) can remember, does
that imply that memories are physical? If I am denying that memories exist in the brain
physically, then must I take the position that every organism’s memory would be infinite?
The short answer is no. Just because memories are not stored in code or written in mentalese
in the brain does not mean that human memory is infinite. Writing something down is not
the only way to store a piece of information.
When a friend of mine buys a new softball bat, she first needs to “break it in,” before
actually using it in a game. Every time she hits a ball with it, the physical structure of the
bat itself becomes altered, allowing it to hit balls further and with greater speed. Clearly
when the first game comes around, the bat will “remember” all of that training and will
hopefully hit a home run. However, no encoded record or inscription of every hit taken
before that day exists anywhere in the bat itself. The memory of all of the break-in period
is a physical property of the bat itself. But still, does that imply that the bat’s “memory” is
infinite? Again, no it does not. After so much use, the bat’s performance will reach a plateau
and eventually taper off. The physical limitations of the bat put a cap on the amount of
benefit it can get from being broken in (i.e. the amount of hits it can “remember”). This
type of example will become more clear in my section about the distinction between digital
analog memory. But for now, I can avoid the infinity objection with the simple truth that
finite memory need not be written or inscribed anywhere.
Putting evolution aside for a moment, the varying ability of memory among people today
is hardly a product of variation in brain size. Your brain and mine are more or less the same
size as that of Albert Einstein, Neils Bohr, Nikola Tesla, Confucius, Buddha, and Aristotle.
However, I am certain that all of those men were much better at storing, retaining, and
recalling information than I am as well as most other humans. This kind of evidence as well
as the examples above should clearly show that memory and brain size does not necessarily
correlate.
4
The Analog Theory of Memory 1.2 Application Issues
5
The Analog Theory of Memory The Analog Theory of Memory
remember certain things and not being able to as well as remembering particularly useless
bits of information that we would be much better off having the ability to delete. Like
the other problems with the computer model of memory, the solution to this one makes an
appeal to infinity or unknowns: there must me some unseen force picking and choosing what
memories are more important than others and which get stored or not.
Here, the fairly obvious objection could be made: why should the computer model force
us to think we have control over our memories? My response to the question is quite simply
that if we are not in control of what gets remembered or not, then who is? Is there, perhaps, a
ghost in the ghost in the machine controlling which memories get kept or not? I doubt it. But
if all memories are uniform, written in the same mental language, and kept somewhere in the
brain, why should we not have access to them? There would need to be some undiscovered
force that resides in each of us to account for this phenomenon. Perhaps the computer model
does not necessarily imply that we have limited access to our memories, but it certainly does
not account for the fact that we do have limited access. And the only way to salvage the
model and account for this phenomenon is to make an appeal to some hitherto unknown
force, which diminishes the model’s credibility. Should such a force be discovered, it will
surely be one of the greatest scientific discoveries of our time. But for now, I think it might
be best to look for a theory that can account for this on its own.
It should be fairly obvious that this model has some serious problems that either imply a
structural flaw, or an appeal to infinity (or unknowns). If the computer model of memory is
correct, then we must have an infinite hard drive, an infinitely fast processor, and some piece
of hardware in our brains that selects which memories to keep and which to throw out that is
often at odds with our desires. Since this seems implausible at best, we must construct a new
model of memory based off of a much less technological philosophy. The digital revolution
has trained people into believing that brains and computers are synonymous in spite of the
obvious differences between the two. This is why we must look to another philosophy as
our basis for a definition of memory, one that eliminates the problems discussed above and
advocates the least unknowable and infinite elements.
6
The Analog Theory of Memory The Analog Theory of Memory
or imprinted into a selected medium, and that memory exists until it is replaced by another or
the medium is destroyed. The inscriptions can vary, however, in both content and mediums.
They can range from footprints in mud to pictures on a wall, writing in books, and even
binary code on a hard-drive.
The main property of digital memory to note, however, is that its contents are isomorphic
and they must be interpreted. A digital memory of a hurricaine would be written documen-
tation of the event written in some language; a record of the event. Note that this is the
working assumption of the computer model of memory, that memories are written, etched,
or in some way “stored” in the brain.
Analog memory is something completely different altogether. Unlike digital memories,
analog memories are not written or recorded anywhere at all. They are physical, structural
changes that an object undergoes due to some repeated action. A simple example of analog
memory is the bending of a wire. Initially, a wire may be difficult to bend. However, after
repeated efforts, the wire is broken in, so to speak, to the point that bending it becomes
effortless. This type of memory occurs with any object that requires some sort of breaking
in (recall the softball bat above), whether it be a pair of speakers, or a car engine. Analog
memory occurs when an object becomes naturally accustomed to a certain behavior through
repeated use so much that a disposition to have such behavior becomes a property of the
object itself.
Another object that uses analog memory is muscle tissue. When someone lifts weights at
the gym, he tears his muscles to the point that they are reasonably weaker than they were
at the beginning of the workout. When he returns a few days later and repeats the same
workout, his muscles have regenerated and come back stronger than they were before. They
“remembered” being torn and returned with a bit more strength this time. Note that there
is no written record anywhere in the muscles themselves that tells them exactly how much
weight was lifted for how many repetitions. The muscles’ physical structure has changed,
however, due to the repeated action of lifting weights.
For analog memory, the memories themselves do not exist within the object. Just because
the wire is now predisposed to being bent (or the muscles are more predisposed to lifting
heavy weights), does not mean that it is bent in half at this very moment. To be more
explicit, just because a forest is dry and is predisposed to catching fire does not imply that
fire exists anywhere in the forest at this very moment in time. Analog memory occurs when
objects are inherently disposed to performing certain acts and it becomes an organic property
of them. It is a proclivity to respond a certain way to certain stimuli, a habit.
7
The Analog Theory of Memory 2.1 The Analog Theory of Memory
To summarize, the two types of memory can be distinguished in the following way: digital
memories exist as pieces of information written on a medium. Analog memories are habits
formed by repeated actions.3
8
The Analog Theory of Memory 2.3 Associationism
of the bell?
According to my theory, not at all. There was nothing but habit formation occuring in
the dogs’ brains. There was no searching, computing, examining past stimuli, calculating the
probable outcome of events to come after the bell was rung or “bringing forward,” of past
similar events. Rather, upon the ringing of the bell, brainstate A was induced. Immediately
afterwards, the dogs would eat and brainstate B would be induced. After enough times, a
causal relationship was created due to the habit formation of brainstate A naturally leading
into brainstate B. Therefore, when brainstate A was induced, B naturally followed.
The two examples differ in that the first has instantaneous association of the two brain
states while the second has a more dynamic and fluid scenario. However, the implications of
the theory are exactly the same. Whether brainstate A and B happen simultaneously or with
a slight delay is not important; the same phenomenon is occurring. The brain habitually
associates the two brainstates and when one is induced in reality, the other is induced by
brain itself.
2.3 Associationism
The analog theory is starting to look quite a bit like associationism, which says that intelli-
gent behavior is the product of associative learning. Furthermore, it says that animals learn
cause and effect relationships in the world by analyzing and associating events that regularly
coincide. My theory, however, takes these claims a step further. So far as psychology is con-
cerned, associationism may very well be true in some sense, but the underlying phenomenon
that allows us to even make associations is what my theory addresses. My theory states that
operant conditioning and association takes place on the neural level.
I am not simply claiming that people associate school with homework or shoes with
socks, but that the brain associates brainstate A with brainstate B, causing us to remember
something. Associationism (and behaviorism) in its classical sense could still work with
the computer model of memory. Perhaps after a threshhold was met, Pavlov’s dogs’ brains
copied “salivation” into the “bell-ringing” experience folder and from that point onward, the
two events would coincide. In this way, associationism could still be true. However, I claim
that the events happening within the dogs’ brains are not at all what the computer model
advocates. The brainstates themselves associate with one another after being conditioned
enough so much that it is a physical property of the brain that one state follows the other.
9
The Analog Theory of Memory 2.4 Mental Momentum
10
The Analog Theory of Memory 2.5 Neurological and Psychological Evidence
The work of both Gerald Edelman and Israel Rosenfield shows that neurons very often
work in groups after being exposed to certain repeated stimuli. In Edelman’s book Bright
Air, Brilliant Fire, he explains this process as it takes place in a developing embryo. As
an embryo develops, nerve cells become associated with one another and specialize based
on how they are stimulated. This specialization process merely involves moving around and
linking up with the proper neurons so that they are accustomed to certain types of stimuli.
For this reason, no two brains are alike, not even those of twins. The epigenetic development
of the brain causes specialization that is solely based on the individual’s environment.
This is why it is so difficult to “diagram” the brain and divide it up into specialized
sections. Where the frontal lobe ends and the temporal lobe begins varies greatly from
person to person, showing that our genetic makeup has a limited amount to say about how
our brains actually turn out. This type of variety in neuronal group specialization even takes
place in the case of identical twins.
The most important idea here for our purposes is the adaptive ability of neurons. Cer-
tain repeated stimuli and conditioning that occur on the neural level determine what each
neuron’s function is. In Rosenfield’s The Strange, Familiar, and Forgotten, he quite clearly
states that, “after repeated excitation by similar stimuli, its (the neuronal group) response
will be reinforced: some of the groups will come to respond better to certain stimuli on
later occasions and some that had initially responded weakly will not respond at all. Thus
environmental stimuli select neuronal groups.” (p. 82)
That neurons specialize and form working neuronal groups through repeated stimuli
during development lends an inordinate amount of credibility to my theory because unlike
stem-cells, whose ability to specialize is lost once it “chooses” a particular cell, a neuron’s
ability to specialize and form groups with other neurons is not lost as the brain develops.
Recently, a baby, Dylan Catania, underwent a surgery to treat a rare condition that
gave him as many as 100 epileptic siezures per day, which involved a procedure that severed
half of his brain.4 The reason this procedure worked was because of the plasticity of the
human brain, the ability for neurons to specialize long after the embryonic stage. Although
he will only be able to use half of his brain, nearly all of the vital functions, as well as the
ability to think and reason, will be preserved. People who suffer from strokes also very often
find themselves being saved by the plasticity of the brain, although the property diminishes
significantly as the person grows older.
Rosenfield also writes that, “Recent neurophysiological evidence appears to confirm that
the brain creates coherent patterns of responses to stimuli.”(p. 84) This claim was made
11
The Analog Theory of Memory Conclusions
in light of a study that found that neuronal groups (in cats) would oscillate at the same
frequency in response to particular visual stimuli. The most baffling consequences was that
these neurons would oscillate at the same frequency even in different parts of the brain.5
It should be very clear to the reader how all of this fits into the analog theory of memory.
Neurons, through repeated stimuli, specialize, group up, and even network themselves in
ways that certain neurons are more likely (or less likely) to respond to particular stimuli than
others. The result is that neuronal groups have physical structures based on environmental
stimuli. This is analog memory. It is the operant conditioning, associationism, and habit
formation of neuronal groups.
Notice that at this point, there is no room for the computer model of memory whatso-
ever. There is no room for memories to be inscribed or stored anywhere. There is no room
for conclusions such as that of John Searle who claims that, “When, for example, we store
memories, it seems we must store them somehow in the synaptic connections between neu-
rons.” Memory simply comes from the way that the neurons group together and react with
one another after certain repeated stimuli, just like they do in developing embryos, and just
like they do in developing people. Just because certain neuronal groups become predisposed
after repeated stimuli to respond in certain ways does not mean that the stimuli or response
is “written” or exists anywhere in the brain independent of the event itself; it is just more
likely to respond a certain way due to habit. Likewise, just because somebody is a smoker
does not necessarily imply that he is smoking at this very moment in time.
3 Conclusions
Though on the surface this model of memory as simply habit of the brain seems very different
from the traditional digital model, it really is not too much at odds with what we know about
the brain. Scientists searching for the area of the brain where memories are stored are making
a category mistake. There is no one area where memories are kept in the brain and sifted
through each time someone wants to remember something; memory simply is a physical
property of the brain itself.6
In conclusion, memory is the result of neuronal specialization and the development of
cause and effect relationships in the brain on the neural level itself. It is the ability to induce
brainstate B from brainstate A out of habit.
12
The Analog Theory of Memory NOTES
Notes
1
(Jol Fagot and Robert G. Cook Proc Natl Acad Sci U S A. 2006 November 14; 103(46): 1756417567.
Published online 2006 November 6. doi: 10.1073/pnas.0605184103. PMCID: PMC1634836)
2
http://faculty.washington.edu/chudler/facts.html
3
Digital and analog memories are not to be confused with the terms digital and analog in their other
applications. Brainstates are digital. Neurons either fire or do not and operate in a sort of neural binary.
However, the sequences themselves are not written or stored anywhere in the brain, the memories themselves
are analog whereas the brainstates are digital.
4
http://www.latimes.com/health/la-me-dylan-html,0,6113116.htmlstory
5
C.M. Gray and W. Singer, “Stimulus-Specific Neuronal Oscillations in Orientation Columns of Cat
Visual Cortex,” PNAS 86 (1989): 1698-1702.
6
Although there is general agreement that long-term and short-term memory is where memories occur,
we have no reason to believe that these areas are where the memories are “written,” for they are not written
anywhere in the brain at all. The functional roles of long-term and short-term memory are not at odds with
my theory of memory. Memories do occur in the brain, and there may very well be localized areas that are
responsible for inducing them. I am simply saying that they are not written in these areas but, rather, are
physical properties of the areas themselves.
13