The Guess Worker

It's complicated.

First there's long-term memory which can hold information indefinitely. Long-term memory is either implicit, which aids in performing motor tasks, or explicit, which is the conscious recollection of previous experiences. Explicit memory is itself subdivided into episodic, semantic and autobiographical memory. Then there's short-term memory which allows us to remember a few items for up to thirty seconds. Even briefer than short-term memory is sensory memory, which lasts less than a second after information has been received. At least three kinds of sensory memory exist – iconic, echoic and haptic memory, storing visual, auditory and tactile information respectively.

That's a lot of types of memory. But is this how it really is?

Memories are made of this

Imagine you are one of Pavlov's dogs. You have been conditioned to associate the sound of a bell with food. Whenever you hear the bell you think you are going to get food and start to salivate. The memory of your past experiences is making you react in this way. How does this memory come to your brain? The neurons of the two stimuli would link up to form a concept like this*1:

When the bell is rung, some impulses from the bell neuron will be diverted across the linkage towards the up-dopaminergic neuron for food. Because of this diversion, the sound of the bell causes the same kind of pleasure as the appearance of food by itself would. In other words, the sound of the bell has evoked the memory of the food.*2

The long, the short and the faint

Is this memory long-term or short-term? The answer is, of course, "long-term": you should be able to remember the association between the two stimuli for days or perhaps even weeks.

But what if you were at the start of the experiment and hadn't yet been fully conditioned? Suppose you had only experienced the two stimuli together only once or twice before. Would you have a long-term or a short-term memory? Seeing as you've experienced the stimuli previously, your memory probably still is long-term. But it won't last as long as it would have been had you been fully conditioned. Perhaps in a day or sooner it will have gone. Why is this?

At this stage in the experiment your neurons would look something like this:

"

Here the linkage from the bell neuron to the up-dopaminergic neuron for food is still weak. After hearing the bell far fewer impulses reach the up-dopaminergic neuron than they would had you been fully conditioned. Your feeling of pleasure will be less intense and you will only have faint memory of the food.

Strong linkages last longer

A faint memory, however, doesn't necessarily mean a short-lived memory. My earliest memories are faint but have lasted decades. What is it, then, that makes the memory of the food in the second diagram less durable than in the first?

Previously I have suggested that if two linked neurons are rarely activated together, the linkage between them weakens.*3 In other words, linkages degenerate if they're not used. Now, let's assume that both weak and strong linkages degenerate at the same rate. Let's also assume that linkages can degenerate to the point that the two linked neurons become disconnected. It would therefore take less time for two linked neurons to disconnect when the linkage is weak and more time when the linkage is strong. Once the two neurons disconnect, no impulses can pass between them and there is no memory. To put this in a nutshell: weak linkages take less time to degenerate and so make short memories; strong linkages, on the other hand, take longer to degenerate and so give longer lasting memories.

Two ways to breakdown

Linkages could degenerate in one of two ways. Either they break down in jumps or they break down steadily over time. If they break down in jumps, their strengths would remain stable for a period of time and then at a certain moment, they would quickly degrade into a weaker state. A graph plotting linkage strength against time would show steps on the curve as the linkage suddenly becomes weaker.

Alternatively, if linkages break down steadily, their strengths would range from very strong to very weak and any strength in between. We could think of any particular linkage strength as being part of a continuum of strengths. Here the graph of linkage strength against time would be a smooth curve.

Which of the two ways above is more likely? If our memories were vivid for a long time and then suddenly disappear, it would be more likely that linkages break down in jumps. That's because memories depend on linkage strengths and if linkages stay stable for long periods and then suddenly degrade, then memories will also stay stable for long periods and suddenly degrade.

Of course, this is not what we experience. Memories don't disappear suddenly; on the contrary, they fade gradually. We can also say something else about our memories: their durability is very variable. Some memories only last for a few minutes, others last for life and all the rest last any length of time in between. Both this range of durability and the fading of memories point to one thing: linkages break down steadily over time. It seems, at least when it comes to long-term memory, linkages can have any strength ranging from very strong to very weak and any strength in between.

A separate memory?

So what about short-term memory? Short-term memory is thought of as being a separate system to long-term memory. It's generally believed that the two types of memory store information independently and that after a short period of time memories are transferred from the short-term store to the long-term store. In line with the idea that the two memories are separate, it's also believed that they each use a distinct method to encode and retrieve information.

The reason for all these beliefs is that the two types of memory seem to have different functions. Short-term memory can hold a small amount of information for a few seconds whereas long-term memory can hold a large amount of information for a long period of time. If we observe two different biological functions, then it's odds on that they come from two different systems.

But let's think about this: do the two types of memory really have different functions? Actually they don't. Both hold information, both encode information and both retrieve information. That's the same function. Where they do differ, however, is in two attributes: how long they last and how much information they can hold. Can we explain these attributes other than by assuming the two memories are separate? Yes we can. Both can be explained by linkage strength.

Strengths and attributes

While long-term memory is caused by stronger linkages, short-term memory could be caused by very weak linkages which degenerate and disconnect within a few seconds. These weak linkages would therefore give very short-lived memories. Why, though, does short-term memory hold so little information? The answer is that since very weak linkages degenerate very quickly, at any moment there will be very few of them in the brain. A few linkages can hold only a little information, whereas a lot of linkages can hold a lot of information. So seeing as short-term memory has only a few linkages at any time, it won't be able to hold much information.

Because both types of memory have the same function and their different attributes can be explained by linkage strength, it doesn't make any sense to think of them as being separate systems. They are the same memory with the same store and the same linkages.*4 And how do short-term memories become long-term ones? Simply by strengthening linkages.

Practical considerations

Let's consider how this might work in practice. Suppose you are shown pictures of twenty objects during a laboratory test for short-term memory. Neurons fire and very weak linkages are made in the brain. Because these linkages are very weak, they degenerate in a short time. That means your memory of these objects disappears very quickly. If you are asked a minute later to remember what you saw, you will recall perhaps six objects. That's because in that minute most of the linkages have become disconnected.*5

Suppose now that you are given half an hour to study the same pictures. The chances are that after this time you will be able to remember all the objects. You should also be able to remember them for much longer than before. That's because during those thirty minutes you are firing your neurons again and again. With each refiring, the initially weak linkages become stronger and longer lasting. Your short-term memory has changed into long-term memory.

Sensory memory isn't memory

If long-term memory is memory with stronger linkages and short-term memory is memory with weaker linkages, then it makes sense to think of sensory memory, which lasts less than a second, as memory with extremely weak linkages. It makes sense, that is, until we look closer at what sensory memory is.

Sensory memory is the impression that a sensory stimulus makes on the brain – an impression which can last a fraction of a second after the stimulus is no longer present. So a light flashed in the dark will leave a temporary image in your brain even after the light is switched off.

What is happening in your brain when you see the light? Why do you still see an image even after the light is switched off? The answer has to do with time. It takes time for the sensory receptors to start and stop sending signals. It takes time for impulses to be sent to the dopaminergic neurons. It takes time for dopamine levels to rise and then fall again. And it takes time for these rises and falls to be registered in the consciousness. Not a lot of time – but enough to make a small time lag between the light flash and the conscious appreciation of it.

Sensory memory doesn't encode, store or retrieve information. By any usual definition, sensory memory shouldn't be called a "memory" at all. It's just sensory awareness of stimuli for short periods of time.*6

Senseless divisions

In all likelihood, then, there's only one type of memory. It works according to simple principles. It encodes information by forming linkages, stores information by strengthening linkages and retrieves information by diverting impulses across linkages.

If it's all so simple, how can we explain those subdivisions of long-term memory scientists have come up with? After all, there is evidence that implicit and explicit memory as well as episodic, semantic and autobiographical memory occur in separate regions of the brain. Surely that's proof enough that these types of memory exist, isn't it?

There are definitely differences between the types of memory. My implicit memories for motor tasks, let's say, don't feel the same as my autobiographical memories. But both implicit memories and autobiographical memories, as well as the other types of memory, are made out of concepts. And all concepts probably work in a similar way. The reason the subdivisions of memories don't feel the same, therefore, has nothing to do with their concepts. Rather, the reason has to do with consciousness. Consciousness converts messages from concepts into feelings, and what kind of feelings they are probably entirely depends on what part of the brain the messages are being sent from. In other words these subdivisions of memory feel different only because consciousness makes them feel different; in all other respects they are identical.

Contrary to what scientists suggest, then, memory probably doesn't consist of multiple, complex systems, each operating separately. The theoretical framework scientists have worked out is complicated and doesn't help us understand in the least what memory is. Why, then, did they think it up?

The answer, I guess, lies in their psychology. Little is known about the brain but there is a strong desire to make sense of the mystery. Add to the mix the very human need to categorise, leave out any good ideas and what we get is an ordered list of observations without any connecting logic. The divisions, subdivisions and sub-subdivisions of memory are a result of scientists classifying for the sake of classifying. Not surprisingly such classification doesn't mean very much. It's nothing more than classification run riot.

Working memory: a project for the future

The one type of memory I haven't dealt with is working memory. Working memory allows us to deal with new information using information already stored in our brains. It's a kind of interface where the two kinds of information come together and are manipulated to give a result. We reason, solve problems and read using working memory. My guess is that the memory used in these thought processes is exactly the same type of memory as described in this post. But it's not a straightforward task to describe how even the most simple of procedures, such as adding up two numbers, is accomplished using memory. That, though, is my goal, even if it takes me a lot of posts before I get there.

*1. See Connections

*2. You might doubt that feeling the pleasure of food is the same thing as remembering food. You would expect that if you remember food you would have a visual image of the food in your brain or, especially if you are a dog, a memory of its smell. Surely an image or the smell of food isn't the same thing as the feeling of pleasure.

They aren't of course, but I would argue that consciousness makes its mental constructs out of two sets of information, both of which come from the same dopaminergic neuron. The first comes from the signals of pain or pleasure the neuron is sending. The second comes from the location of the neuron in the brain.

For example, if a neuron in the visual cortex sends a signal to the consciousness about the pleasure of food, consciousness can tell it's a visual signal from the location in the brain the signal is coming from. It can also tell precisely what kind of visual signal it is from the relative position of the neuron to other visual neurons. So the same neuron which gives information about pleasure of the food can also give information to the consciousness on how to construct a mental image of the food.

*3. See Hope

*4. Eric Kandel of the Kalvi Institute won the Nobel Prize in Physiology or Medicine in 2000 for, amongst other things, elucidating the difference between short term and long term memory. It's not comfortable to be in disagreement with a Nobel laureate, especially given his extensive knowledge and his in-depth research going back decades. But I have to say I don't disagree with Eric Kandel's experimental results. What I do disagree with is his interpretation of them. He starts with the assumption that long term and short term memories are distinct and draws his conclusions based on this "fact". I believe, though, his findings prove neither they are distinct nor, for that matter, that they are part of a continuum. I hope to explain why in a future post.

*5. You might wonder why only some of the objects are remembered. Shouldn't it be either all or none? If the time interval after seeing the pictures were very short, the linkages would not have broken down yet, in which case we should remember all the objects. On the hand if the time interval were too long, all the linkages should have broken down, in which case we would remember none of the objects.

We have to consider, though, that these linkages are so weak that they break down very quickly after they are formed. That means that by the time you have looked at the last pictures you have already forgotten the first.

Shouldn't that mean, though, that you would remember only the last six objects you were shown? Clearly that doesn't happen – you might remember some of the objects you were shown at the beginning. What's the reason? It has to do with salience. Some of the objects might be slightly more interesting than others and that means the neurons they activate make stronger linkages. These stronger linkages will, of course, last slightly longer, and so you'll keep your memory of these objects for slightly longer too.

For more on how salience strengthens linkages see Learning.

*6. Sensory memory might have been misdefined because of the reluctance of scientists at the time these divisions were devised to embrace the idea of consciousness. Consciousness has only recently become a fashionable topic of scientific research; previously the subject was viewed as a suitable only for pseudoscientists and transcendental meditators. The articles I have read on sensory memory carefully avoid the word consciousness.