Wednesday, January 16, 2008

Intelligent Memory: Understanding the conceptualization process


Myself, with Jo Wood and Peter Fisher, discussed the idea of conceptualization uncertainty in the SDH2006. We argued that conceptualization uncertainty is introduced during the conceptualization of a phenomenon rather than due to measurement error.

Barry Gordon, professor of neurology and cognitive science, presented the concept of Intelligent Memory which is the mostly unconscious, lighting-fast thought process that connects pieces of memory and knowledge in order to generate new ideas. It's the memory that aids us in making everyday decisions, gives us the chuckle of a good joke, sparks a "Eureka!" solution to a problem, and enables us to enjoy a work of art. Intelligent Memory is what powers most of our mental life.

I personally think that understanding intelligent memory can give us some ‘clues’ about how people conceptualise and argue about indeterminate phenomena such as town centres.

The Science Behind Intelligent Memory

What follows is an in-depth explanation of the neuroscience behind the Intelligent Memory concept. It's for readers who want to understand the scientific underpinnings of memory and learning.
All memories, along with every perception, action and thought, arise from the activity of nerve cells. However, the memories that we are conscious of and which are important to us, obey somewhat different rules than nerve cells. This is makes sense, given that our important memories generally require the coordinated action of thousands, if not millions, of nerve cells.
Paradoxically, some of these nerve cells help generate memories by not being active or not communicating with other neurons. They're somewhat like the essential patches of blank canvas that an artist uses to suggest clouds or a piece of reflected light. Another analogy can be found in the printing that you are reading at this moment. The letters and the words the ink forms are meaningful because of where the ink is, and is not.
The key to memory is time. In essence, memory is a displacement of knowledge a little bit into the future. Or, from a future perspective, it's the retrieval of knowledge from the past. This knowledge can be latent, or unused, or active and available. When nerve cells are firing, they are actively carrying information, and so the memory is active, and usable.
But this form of memory is also transient and by itself, it can exist only a few fractions of a second. What makes a memory permanent is not a nerve cell constantly firing but over time acquiring more potential for being able to fire. In other words, a nerve cell becomes more sensitive to firing or to staying quiet. This sensitivity to being triggered into action can be varied up or down. The processes that change susceptibility are built into nerve cells. There are many of these processes, including temporary changes in the permeability of the nerve cell membrane and permanent changes within its DNA. Correspondingly, they can take place over different time scales. Changes in the permeability of the nerve cell membrane can occur in fractions of the second, while changes in the proteins within a nerve cell may take hours to days to generate. And DNA may take weeks to months or years to change.
One of the crucial contributors to nerve cell sensitivity is individual experience - whether and how often they've fired before. If a nerve cell has been triggered to fire in the past, in general it will be more sensitive to those triggers in the future. Yet, if a nerve cell has been active over long periods of time, it gradually becomes less sensitive and needs increasingly more stimulation to set it off or produce changes.
Oddly enough, this intrinsic regulation is basic to creating intelligent memories. This regulation, when repeated over and over, produces particular kinds of memories - memories that arise through practice. Repeating a thought or action strengthens and weakens individual connections between nerve cells, and the upshot of many connections is learning. By and large, this learning happens relatively slowly. It takes a fair amount of repetition to convince nerve cells to be more sensitive the next time. Doing something once doesn't do it. Doing something twice or three times doesn't do it. But doing something hundreds or thousands of times definitely does.
You know these kinds of memories well. They are the memories you acquire when you learn how to ride a bicycle, to drive a car, to play golf or to add 2 + 2. As you acquire them, you can strengthen them quickly if each time you think about the precise right way and immediately correct your mistakes. However, if a task is complicated, you need a great deal of practice.
Although so far the focus has been on individual nerve cells, keep in mind that most of the memories and activities that mean anything to us take long chains of nerve cells. Catching a ball requires chains for seeing as well as chains for hand control. Nevertheless, individual nerve cells and connections between them are the basis for these activities.
Getting back to how nerve cells form memories and learn: on their own, individual nerve cells don't decide whether to learn. Brains as complex as ours have additional circuits of nerve cells that monitor what's important and what needs to be repeated and remembered. Such circuits control how other neural circuits learn. They can even force neural circuits to learn quickly. ("Enough daydreaming - remember this!") Or, they can stop them from learning at all. These control circuits also dictate how the more basic neural circuits are wired together, which get inputs and which do not, and which chains of circuits are beefed up and which are broken up and rewired.
And, as you may have guessed, our brains also have circuits that monitor and control the controlling circuits. And there are undoubtedly monitors and controls for the monitoring and controlling circuits, and so forth. Neuroscience doesn't completely know how many levels of controls our brains possess. They're hard to identify or track down because there is not a strict hierarchy. Instead, some controlling circuits seem to influence other controlling circuits at the same level and sometimes lower-level processes can boss around their controllers.
Our brain's basic wiring plan governs how we perceive, act, think, and remember. But to understand intelligent memories, we need to elaborate beyond this basic scheme and look at the links between nerve cells and nerve circuits. It's these connections which are the true building blocks of thoughts, and Intelligent Memory. ("Intelligent Memory" is our shorthand term for all the different intelligent memories. They all work much the same way; it's just their specific contents - such as words or images - that differ.)
What we think of as a single thought in our mind - "ball" for instance - is composed of many fragments of thoughts. If you think about a ball, you do not normally separate its color from its roundness or its bounciness. However, your brain does. Its color and shape and function are stored in different regions of the brain, although not every distinct element has its own region.
In the brain, these elements of thought are represented by patterns of activity in many nerve cells. These patterns can be active and the nerve cells firing, or they can be latent, existing in the pattern and strengths of connections between sets of nerve cells. An idea in our mind -- whether it's the color or the shape or movement of a ball -- is represented in the activity or latent activity of these sets of nerve cells as a whole. And thoughts that we are very interested in are likely to involve thousands if not tens of thousands or more nerve cells.
Most complex thoughts have to be learned; they are not innate. When elemental thoughts arise from the senses, its usually constant exposure, like playing with balls as a child, that gradually produces the whole idea inside our minds. The same process seems to be at work for thoughts or concepts that have no obvious sensory or other correlates.
Elements of thoughts are linked in many ways. Sometimes they are linked just by being part of the same entity in the outside world, as in the case of the ball. In this case, there are linked by experience. But the most interesting links for our purposes -- the links that make up intelligent memories -- are ones we discover and put into place. They are the links, for example, that allow a child to see the similarity between the ball he is throwing and the planet he is standing on.
The links between elements of thoughts, or between thoughts themselves, are patterns of neural activity, either active or latent. Therefore, they can be learned.
Links between thoughts produce thinking. Some kinds of thinking generated by these links may seem so ordinary that we don't call it thinking at all. Being hungry, passing a candy machine, and stopping to put in a coin is hardly a Nobel prize-winning connection. But even these thoughts required having the elements inside of our head (some coming internally, from our hunger; others coming externally, from the image of the candy machine) and then making the connection between them. (It also involved acting upon that connection.)
Solving harder, more complex problems requires more and better connections. But this should not obscure the fact that elements of thought and the links between them are nevertheless necessary. Moreover, it is easy to understand that creative thinking occurs when the links go in unpredictable directions or towards goals we did not set in advance. But they are still links, and they still arise from the same nerve cell activity and the same learning process.
Links are the streets that take us from thought to thought. But finding connections between thoughts, or finding the best ones, can be like trying to find the best route to a destination. The first route we explore may have many false starts or roads that look good on paper but don't work in practice. With time, though, we find a shorter work or faster route. So it can be with thinking. Over time, we can prune away the false starts and wrong directions, and eliminate the links that look good originally but prove to be rocky or laborious or time-consuming.

This process of finding the best mental route is the essence of training our thinking. But from the perspective of what nerve cells must do to be trained to think, it is also learning. Memory mediates mental training. This memory, this learning, is what helps make us intelligent. It's also a basis for intelligent memories.
Nerve cells also comprise the circuits that monitor the links and open and close the routes, and these, too, can learn and can improve. The controlling systems, these guidance providers inside of our heads, can be trained and so form another site for intelligent memories.

At least two more physical facts about memory and our brains figure into an understanding of our thinking, learning and creativity, and how they can be improved. One of them relates to how learning can be enhanced. The other relates to how we create miniature intelligences in our minds to help eliminate the bottlenecks of certain kinds of thinking.
Nerve cells learn when they are exercised. Practice, which stimulates connections, makes nerve cells learn. However, nerve cells also learn when we tell them to. When we deliberately activate the circuits that signal something is important, the circuits pass on the message and tell the appropriate other nerve cells that what is happening is important and should be learned well. This happens, for instance, with the learning involved in memorizing facts, names or faces.
While it is less clear that the circuits involved in learning connections between thoughts can be revved up this way, it seems almost certain that interest and motivation synergistically tickle nerve cells and make them learn much faster. So this is another mechanism we can use to enhance our Intelligent Memory.
The bottleneck mentioned earlier arises with our conscious thinking and attention. When we are consciously and fully alert, we can keep no more than a few thoughts in our mind at once. (Perhaps just only one thought at a time can be maintained consciously.) Our unconscious, automatic minds, on the other hand, do not have such a bottleneck or limitation. And fortunately, much of our mental activity takes place unconsciously and automatically. When you walk, you don't think about every irregularity in the pavement, or every curb you step on. Those perceptions, decisions, and actions are handled automatically and unconsciously.
Your mind did not always perform such mental tasks automatically. There was a time when you had to learn them. As an infant, you had to learn to walk, which required paying attention to the terrain in front of you and coordinating what you saw and felt to how your body reacted. A better example of the process may be when you learned how to drive a car.
When you learned to drive, you had to learn to pay attention to everything going on and everything you had to do. You watched your hands on the steering wheel, the hood of the car, each sign and traffic light, the other cars on the road, and every pedestrian. You also had to think about what to do in situations: the stop sign or the yield sign, a car getting too close, a pothole. But as you practiced driving and became better, your ability to detect what was happening on the road as well as your reactions became more automatic. You didn't have to consciously look for a stop sign or a red light in order to notice it and automatically respond the right way. And if a pothole suddenly appeared, you knew you would immediately see it and not only swerve but check your mirrors for other cars nearby and slow down.
What you did through all this practice and attention was create automatic mental abilities. You used your conscious mind and deliberate intention to instruct your brain on what to attend to, what decisions to make, and what to be done. Your conscious mind programmed the necessary circuits in your brain. It instructed your vision to pay attention to the color red on a light or a sign. In addition, your mind established a network of override circuits so that the need to stop would take precedence over almost everything else. It also set up a watchdog circuit, so you would not stop too quickly if a car was on your tail. Finally, it programmed what you have to do to stop: take your foot off the gas and push the brake pedal. All these mental processes had to be laid down and practiced to the point that they became instinctive, like a separate intelligence or "minimind" operating on its own.
Now that you are an experienced driver, this minimind is vigilant whenever you're behind the wheel, ready to respond to any stop sign or stop light. You don't have to think about it and it no longer requires your conscious attention. Because it's automated, it works in parallel with your conscious mind. It augments your abilities. It augments your intelligence.
Elementary mental processes are relatively rapid. They operate in hundreds of a second, or at their slowest, tenths of a second. However, these elementary mental processes are often strung together in chains and loops and these strings of processes often take a fair amount of time to unfold. Conscious minds may need more than a second to appreciate a situation, and several seconds of backwards and forwards thinking to come up with a response. Our unconscious, automatic minds, on the other hand, are much simpler and more direct, and can work much faster. A baseball thrown by a professional pitcher moves too quickly from the pitcher's mound to the plate for a batter's conscious thought to react (which takes a minimum of 1/4 of the second). But the batter can preprogram his miniminds to watch the pitcher's throw and to watch the ball, so that his swing has a decent chance of connecting.
All of your thinking, all of your decisions, all of your creativity comes from the same kind of miniminds you apply to skillful driving. But these miniminds cannot always substitute for careful, deliberate thinking. Sometimes, the information they use is too limited, and the judgments they make are too quick. Still, they augment the powers of your conscious mind, which usually does not have the luxury of unlimited evidence and slow, deliberate thinking.
These miniminds, which represent intelligent memories, take time to be constructed, but they are extremely persistent once they have been built. This is often an advantage, since a useful mental tool should be kept around. However, this persistence can also cause major problems. Problems can arise when a minimind has not been constructed properly or when its operation has taken wrong turn that becomes permanent. For example, making a snap judgment using these miniminds is a big reason people make errors on everyday problems, particularly those involving statistics and logical thinking.
A first step in enhancing your miniminds is to understand what types you have available. The ones that work well can be left alone, while the ones that repeatedly make mistakes need to be retrained. When you survey your mental abilities and needs, you may well discover that you need certain abilities -- miniminds -- that you do not currently have. These gaps need to be identified and filled, and to take their place alongside your high-functioning miniminds. And, of course, you need to train the intelligent memories that orchestrate these particular miniminds, so the right ones can be used in the right situations.
Now you know more of the details about why we can have Intelligent Memory, and why we can consciously exercise this memory and make it stronger.