Where is knowledge stored within the brain and living systems? The concept of monitor-and-act units provides a framework for understanding this.
In my last post I have listed three properties that intelligent systems must satisfy in the context of hierarchical adaptations. I will now focus on the first one: the storing of knowledge within the brain, or within living systems in general. Let’s start with a cell. How can we generalize all possible ways a biological cell can store knowledge?
Considering cellular processes in terms of monitor-and-act units
Knowledge is stored within cells in a number of ways from that not only include synaptic connections, but also intrinsic plasticity, genetic materials and so on. The way the theory of hierarchical adaptations (aka, practopoiesis) generalizes across these various elements is through the concept of monitor-and-act units. By introducing monitor-and-act units, the theory emphasizes the need to consider not only the information storage (like memory) but also the parts of the machinery that uses that information to execute action (something like a CPU). As it will turn out, in many cases the two cannot be physically distinguished. For example, the synapse contains information but also acts by opening and closing its channels and letting ions flow across the membrane. (The emphasis on action has been expressed in the “practo” part of the name in the theory as this is the Ancient Greek word for acting).
Properties of monitor-and-act units
Monitor-and-act units do something or ‘act’ and thus, impose changes on the system. The ‘actions’ can span high level things like moving muscles or low level things like synthesizing proteins, or anything in between. Importantly, monitor-and-act units possess knowledge. This knowledge is used to determine when one needs to act and in what way. Essentially monitor-and-act units exhibit a certain level of autonomy to monitor the situation and act when necessary to serve a certain purpose. In general, any purpose will broadly relate to increasing the wellbeing of the organism and its fitness. Survival of the organism and the species as a whole is dependent on monitor-and-act units acting at right times in right ways.
One more important property of monitor-and-act units is that they need to learn. The knowledge that they have does not come from a thin air, the knowledge has to instilled in some way. And, after it has been instilled it may need to change as the properties of the world change. Whatever rules of acting did well for the organism at one point in time, they may not do as well later. The knowledge of the monitor-and-act units needs be adjustable.
Finally, to be able to act, monitor-and-act units have to have their own storage of energy. Every action requires a physical work and physical work cannot be done without energy. Hence, energy storage is an integral part of a monitor-and-act unit. To be able to produce work, this has to be the so-called free energy.
An example of such a monitor-and-act unit is the patellar reflex, about which I have written before. The reflex monitors the tension of a muscle. For that it needs to receive some inputs. The reflex also acts when necessary by contracting the muscle. As a result, the reflex may change the position of the body. A simple patellar reflex is achieved through a whole system. The system comprises of multiple neurons, and each neuron has multiple components. The system together possesses knowledge and when and how to act. [Add one sentence on the learning aspect and free energy].
Components of monitor-and-act units
To understand the theory of hierarchical adaptations is to identify the minimal possible component of a monitor-and-act machinery — components that still acts as a monitor-and-act unit but as soon as further reduction takes place, the remaining parts no longer have monitor-and-act properties. That way we can identify elementary monitor-and-act units. These elementary monitor and act units will be our elementary physical components that make the building blocks of biological intelligence.
Let us continue with the example of the patellar reflex. If we take only one neuron out of the reflex arc and remove all the other neurons and the muscle, this neuron alone still has monitor-and-act properties: input, action, knowledge, learning capability. But if we continue disassembling the neuron, sooner or later we will end up with parts that do not have such properties. For example, individual ions do not have such properties. A messenger does not do that. An ATP molecule is a source of energy but does not have monitor-and-act functionality. DNA alone does not monitor and act. And so on.
So, where is the border? We need to find the border line so that we identify the minimum number of physical components that still exhibit monitor-and-act functionality. These elementary monitor-and-act units are interesting because we can then identify the physical properties that give them the functionality that they have. And identification of these physical properties is critical for understanding their creation — the way they come into existence. These physical properties constitute the knowledge of the unit and whichever chemical or physical mechanisms puts those physical properties into place, are the mechanisms of their poiesis. This then becomes the touching point of the adaptive (poietic) hierarchy, which we will discuss in detail in a later blog post.
Monitor-and-act units within a neuron
So, let us consider some examples of monitor-and-act units within a neuron. Is a synapse a monitor-and-act unit? Yes, it seems so. It checks all the boxes: monitoring, acting, having knowledge, can learn, and free energy gets to be used in the form of the difference in electric potential across the membrane. What is the elementary monitor-and-act unit of synapse? This seems to be the individual ion channel. For if we decompose an ion channel any further it will lose its functionality and its monitor-and-act properties will be lost.
Is axon a monitor-and-act unit? This is questionable. Certainly, the mainstream brain theory does not consider axon as monitor-and-act unit properties. An axon does monitor electric potential at the hillock and does act by letting ions in and out and moving the action potential along its length. Also, axon consumes energy. But axon may be missing one key property: It may not be learning. It may not posses its own unique knowledge that helps and contributes to the overall fitness of the organism. The axon may only be a “wire” to transfer information unchanged. At least this is how the mainstream brain theories treat the axon. However, it remains open whether the axon itself may also be adjusting its properties as the organisms need arise or as the properties of the environment change. A simple adaptive change in the speed with which action potentials are being conducted would suffice to qualify an axon as a monitor-and-act unit.
Neurons also possess plasticity mechanisms—plasticity is necessary to put synapses into place but also other intrinsic changes to the neuron outside of the synapses. Can plasticity mechanisms be considered as monitor-and-act units? It seems so. They monitor what is going on with the neuron, exert changes or adaptations and consume energy during that process. But to qualify as monitor-and-act units, plasticity mechanisms also need to learn how become better intrinsic plasticity mechanisms. To answer this question, we have to repeat our exercise of finding the elementary units of those mechanisms. What seems to pop-out is gene expression.
Plasticity involves expression of genes (at least some types of plasticity). Therefore, the elementary monitor-and-act units of those mechanisms have to do with DNA and all the associated machinery including transcription to RNA, and synthesis of proteins by ribosomes. This entire factory is the elementary monitor-and-act unit for the neuron’s plasticity. So, can it learn? Yes, this mechanism learns through the process of evolution by natural selection. Darwinian evolution is a slow learning process that does not take place during the life of a single specimen but rather across multiple specimens—including ancestors and offspring. Thus, this is a learning mechanism whose effects can be seen only across a species, not within an individual. Nevertheless, it is a learning mechanism that qualifies neural plasticity as a monitor-and-act unit.
Which other monitor-and-act units exist in the neuron or in living cells in general? To list them out we need to carefully analyze structure by structure. There are likely many others. We know that there is fast adaptation of neuron (fast = 100 milliseconds or so), which I also wrote about in previous post on how the brain deals with novelty. The mechanisms of those adaptations need be identified too.
Combining monitor-and-act units into macro monitor-and-act units
Can monitor-and-act units combine in other ways to create bigger monitor-and-act units? Yes, they can. There are many ways in which monitor-and-act units can combine. For example, ion channels can can combine into a synapse, or into an axon. Multiple neurons can link into a chain that forms a full reflex. Out of countless different ways to combine there one special way that is interesting for the theory of hierarchical adaptations. This is where some of the monitor-and-act instill knowledge into other monitor-and-act units. This is the process of poiesis and we will cover it in one of the next posts.
This is part of a Mind/Brain series by Danko Nikolić. You can read the full series here.