The previous section explained how concepts are created by the intelligent mind. This section will explore the second aspect of the mind’s interaction with concepts: concept retrieval.
In this document, concept creation implies storage whether that be in the human in brain or in the volatile or permanent storage of a computer and so forth. Retrieval of concepts immediately demands discussion on the mechanism of memory.
What is Memory?
For any physical makeup of an intelligent mind the storage and retrieval of concepts is dependent on memory. I am not qualified to write extensively on the details of human memory and that topic is not the focus and intent of this document. However, there are fundamental aspects of memory that are relevant to the ideas discussed in this document.
Memory is where concepts are stored and from which they are retrieved as needed. Every thought in our mind, from the simplest to most complex, is stored as a concept. As I will discuss more in the section regarding concept application, retrieval of these concepts is necessary to solve problems, understand our environment and make sense of new perceptions (information vectors). Therefore:
It seems to me that some of my concepts have high importance or magnitude. Concepts involving my family, my well being, and my career work are constantly forefront in my mind. Some might say that my emotions make these concepts important. But then how would an emotionless mind, such as a computer program give importance to these kinds of things? I believe considering a goal-driven perspective could be used in place of emotion (or is it closely related to emotion?). If I have ongoing goals that involve the well being of my family, this will force those concepts to be high priority as the following sections will detail.
Frequency of occurrence is one factor that supports concept retrieval. If I see the same movie 100 times I am more familiar with it than if I saw it once. Making coffee every morning is a stronger, more accessible concept for me than making waffles with a waffle iron because I do the former activity much more frequently. Perhaps our mind recognizes the recurrence of concepts and configures our thought network to make these particular concepts easily accessible. It is fairly easy to quantify frequency in the context of concepts because it is driven by the number of times a concept has been experienced or triggered. In other words, it is straightforward to determine which concepts have been considered more frequently than others.
Recency is another factor that drives concept accessibility and retrieval. Computers are designed to use cache mechanisms to store recently used data in quickly accessible locations.
Recently utilized concepts are often repeatedly relevant to our current situation so it makes sense for the mind to optimize their retrieval. Again, maybe it is that pattern of life: recent concepts are typically strongly relevant, that caused our thought network to give recent concepts the optimized accessibility. In other words, over time, our minds realized that recent concepts are often reused, so in general it should design retrieval to make recent concepts easily accessible. That is how the concept of computer caching was designed: engineers realized that “recent” data is typically required repeatedly so they came up with the idea of an accessible cache to store recently used information. It is easy to quantify recency in the context of concepts because it involves a measure of time. In other words, it is simple to determine which concepts have been considered more recently than others.
Relevance also supports concept retrieval, but the mysteries behind how this works are the most interesting and difficult to understand. Unlike frequency and relevance, there is no apparent numeric way we measure relevance, although it must exist. In some sense, relevance may be driven by frequency. If you see smoke when you see fire frequently, there will be a strong relevance between smoke and fire. If you always have a hot dog when you go to a baseball game, you may think of (retrieve the concept of) baseball if you are eating a hot dog. It will be clear that relevance is an important retrieval influencer as we discuss concept application and goal driven behavior.
Compression and Concept Storage and Retrieval
I asked my wife to draw a picture of a smiley face. Once she was convinced I wasn’t setting her up for some practical joke she drew this picture in a couple of seconds:
If I had been asked to perform the same task, I would have drawn something very similar. I probably would have omitted the nose, just based on my concept of a smiley face. And chances are, your drawing would be somewhat similar to the sketch above — your drawing would probably more closely resemble the sketch above than that of a tree or motorcycle.
Throughout my life, I have probably perceived thousands of information vectors (images) that I would conceive as smiley faces and the details of these images vary. I mentioned that my wife drew this picture right away. I am confident her mind did not retrieve the thousands of memories of smiley faces that she has experienced in her life. It’s safe to say that most people people are not capable of doing that at all.
So how are we able to represent our concept of something like a smiley face so quickly? It seems that our mind models a “template” version of the concept that is an amalgam of our collective experiences. This concept template is another example of our minds using compression for the purpose of efficiency.
When I see the image above my thought process is: this image is very close to my template concept of a smiley face, but it has a nose. At this point I believe my mind expands its search of my memory for different stored concepts of smiley faces (based on different information vector inputs it has perceived over time) and I decide: yes, that is definitely a smiley face. It matches up very well to my concept of a smiley face after all.
I asked my four and a half year old son what this same picture was. He said it was a happy face. I asked him how he knew it was a happy face. He told me: because it is a circle with two eyes a nose and a mouth. He interpreted the information vectors and retrieved the concept from his memory.
If I ask you think about a tree, or a cat, or a hot dog, you quickly recall some template of that concept in your mind. This recall occurs when we experience information vectors, provided we do have some concept of the image, sound or sensory perception we are experiencing. In the next section, we will discuss application of concepts and how creation, recall and application work together to build our intelligence.