The adaptive resonance theory examines how the brain retains information and then uses it. The human brain can categorize data, use information to recognize familiar items, and even predict future outcomes based on what has been learned. Humans learn massive levels of information throughout life on their own. This data is then integrated into individualized self-concepts.
Unlike other forms of data retention, there is no need to delete data to make room for new data.
The adaptive resonance theory proposes through natural mechanisms that there are mechanic links between various brain processes that make it possible for humans to be constantly learning, intentionally or unintentionally. That makes every conscious state a resonant state.
Stability-Plasticity Dilemma and the Adaptive Resonance Theory
The stability-plasticity dilemma is a constraint for every neural system, including artificial systems. For new knowledge to be obtained and integrated, there must be a level of plasticity to the neural system. If there is too much plasticity involved, then the data that was previously encoded to the system will begin to erode.
If there isn’t enough plasticity to the system, then it will become unstable and that makes it difficult for new knowledge to be obtained.
As a knowledge system ages, it naturally loses some of its plasticity. That’s why as people age, their ability to recall new information or reference previous information can be delayed or obstructed for some individuals. It is also why a developing network, such as that of a toddler, absorbs more information than mature networks.
Younger networks have higher plasticity because there is more natural stability within the system. Older networks have less natural stability, which means there must be less plasticity to balance the learning process.
When those two processes become out of balance, whether through disease, defect, deformity, or some other reason, then the mechanisms become ineffective and the information is difficult to retain or recall.
What Are the CLEARS Mechanisms?
A human is an intentional being. Individuals learn their expectations for their environment and that learning process allows them to make predictions about what may happen.
At the same time, a human is also an attentional being. Individuals focus on finding and processing resources based on the data that is being processed at any given moment.
The ability to perform these actions in the adaptive resonance theory is done through the CLEARS mechanisms. The acronym stands for the following.
Through the connections that are made, the human brain is able to map out how numerous items can belong to the same category or how an individual object compares to others that are similar to it.
This is accomplished through a salient combination of cues, focusing attention on a specific subject through top-down expectations that are formed from previous learning processes. Supervision isn’t required for this learning process as it relies on comparison fields, recognition fields, a threshold of recognition, and a reset module.
As data is gathered, it goes through the mechanisms and learning processes to retain the information as it was perceived.
Training Through Adaptive Resonance Theory
There are two methods of adaptive resonance theory training when focused on the neural networks. They are categorized as being slow or being fast. When the slow learning method is being used with this theory, then the amount of time the information is being presented to the individual will be a reflection of how much information is retained. This time is combined with the training of the individual’s brain to retain information and that creates accuracy in its recall.
In the fast learning method, equations are used to determine what needs to be learned. Those equations are then used to reflect larger information values that can be disseminated from the information that was retained.
Let’s say an individual is learning how to use a dishwasher for the first time.
In the slow learning method, the individual would need to watch someone load the dishwasher with dirty dishes because they had no experience with it before. They would learn where to place dishes based on how someone teaches them to place it. Then they would need to be shown where the detergent would need to go, how to close the dishwasher door, and the final wash setting that is appropriate for that load.
In the fast learning method, an individual would use their past experiences to create equations that would help them figure out how to use the dishwasher. They might recognize that the two dishwasher trays look similar to dish drying racks, so it would be logical to place dirty dishes into it. They would recognize the need for soap from hand-washing dishes in the past, so would look for a detergent receptacle. Then, because they’ve used appliances in the past, they would infer that there was an on switch or knob to find, locate it, and use to engage the unit.
Slow learning is a better option for learning in continuous time environments for skill retention. Information that is learned over time with direct data makes it easier to recall that data when it needs to be used in the future.
Fast learning is efficient for when a variety of tasks needs to be completed quickly. It helps individuals get the job done while gaining experience, but makes it difficult to recall specific task aspects if asked.
For example: in the slow method, a person learning how to use the dishwasher could potentially recall the type of detergent used, the wash setting that was appropriate, and how long a complete washing cycle would take. In the fast method, only the basics of loading the dishwasher have the potential to be retained.
How Data Is Gathered in ART
Data is gathered through an environmental awareness of all 5 senses. Many ART descriptions use the process of how the human eye gathers information from a visual input and transmits it from the retina to the visual cortex, but sounds, smells, touch, and taste all transmit data to the brain as well.
Once that data is collected and transmitted, patterns and engrams are produced that recall the input. The brain can then compare the expectation of future events to the input pattern being received at that moment.
Using the dishwasher example, the individual might grab regular dish soap to put into the dishwasher after learning how to put detergent into the appliance. The brain compares the previous input to the current one, recognizes the discrepancy, and then understands that the dish soap is the wrong item to grab.
By understanding how we learn new information and retain it, we can gain new insights into how the brain works and maximize our learning potential. That is what is significant about the adaptive resonance theory.