Systems Science 11 — Bio-Systems Examples
What is Life?
Erwin Schrödinger asked that question in 1944, one year before my birth. Of course he was by no means the first person to ask it in scientific query. But he was the first to link living processes to the ideas of thermochemistry and our modern notions of organization and order. He called life negentropy or the negative of entropy, implying that living systems had the capacity to self organize and maintain states far from equilibrium in spite of the Second Law of Thermodynamics. What some scholars had, in previous times, called the 'vital force' Schrödinger correctly identified the basic principles that underlie the motivation of living systems, the flow of energy through the system.
Inspired by Schrödinger, years later (1968), Harold J. Morowitz (sadly no Wikipedia entry to my dismay) extended this thinking in what I personally consider a seminal contribution at least as important as that of Ilya Prigogene's with respect to chaos theory. Whereas Prigogine posited some emergence of order out of an otherwise chaotic process, Morowitz provided a precise mechanism for how this might happen. Far from deterministic, in the ordinary sense of that word, Morowitz provided the low level details of how a stochastic process could be driven far from equilibrium and provided a deeper understanding of the means by which Prigogene's dissipative systems could produce higher levels of organization and maintain a 'vital' process. It was not so mysterious after all. In the hands of Morowitz we see how the flow of energy through a well defined unorganized but bounded system will drive that system toward higher internal organization without it heating up to a point where whatever structure obtains is obliterated (via dissipation). This seminal work is Energy Flow in Biology, and in my personal opinion it ought to be required reading for every biologist who claims to understand evolution and organization (see this Amazon.com page for a list of Morowitz's other books).
Living organisms, from single celled to human beings, are nearly incomparable in terms of complex, organized systems that appear to be tractable in terms of what goes on within the observable boundary of that system. They are not closed to either energy or matter. Yet they are sufficiently like Mech-Systems that we are able to study the internal workings and hope to have some decent ideas about how they work. Nor are they closed to information flows and that too makes them more interesting than Mech-Systems, but not as ultra-complicated as an Eco-System. Indeed, individual organisms along with their populations comprise the living aspect of an Eco-System. But unlike the latter, a Bio-System has evolved with elaborate control structures, a hierarchical control system, that provides for stability internally that is not always present in an Eco-System.
In this installment, I want to explore, through some examples, the principles of organization and dynamism for Bio-Systems, showing where they inherit some characteristics of Mech-Systems (origin of life) and where they behave, at times, like Eco-Systems (neural Darwinism and learning). But mostly I hope to show how, via the mechanisms of hierarchical control, an organism achieves a relatively long life of stability after development and manages to survive and procreate as a result of the control of energy flow through the body (cell).
Where Did Life Come From?
One of the greatest questions of all time. At one point early in Earth's history, there were no living things. Then, about 3.8 billion years ago, primitive organisms emerged, swimming in ancient oceans and kicking off the Darwinian process of variation in form followed by selection for fitness. Today the origin of life (abiogenesis) is still shrouded in mystery, not because we haven't got an idea of how it might have come about so much as the fact that we have a number of competing ideas and are in the process of testing them.
There are several clear, key elements that must be present for a system to be alive. Typically life is defined by this list of characteristics (see Life). But Schrödinger, Morowitz, Prigogine and many more have wanted to define life as a unified process. One of the most ambitious attempts to describe this process in a concise, coherent way was developed by Chilean biologists Humberto Maturana and Francisco Varela. They introduced the concept they called autopoiesis in 1973. More on this later.
The key to understanding the origin of life is to describe how a prebiotic chemical mixture could give rise to something like an autopoietic system, in essence a cell. Morowitz has suggested that there are clues hidden in modern metabolism, sort of like fossil chemical reactions that have been conserved from the earliest times because they are so good at producing a stable but once primitive metabolism.
The first requirement for some aggregate of chemicals to be alive is that the system be bounded. Fatty membranes seem to form relatively easily when the right molecules are available in an aqueous solution. These membranes can act to contain larger, less portable molecules while being permeable to smaller, component molecules. The second requirement is for an appropriate energy gradient to exist wherein the system can capture high potential energy from the environment and use it internally to do the work of creating (and breaking) chemical bonds in the process of maintaining the dissipative structure, the structure that then emits low potential heat and waste products. One of the reasons I favor a metabolism-first model (see this section of the abiogenesis article) is that appropriately coupled energy flow mechanisms internal to the system have to have been in place in order for RNA synthesis, as well as other complex molecules, could be achieved. Someday I will take a closer look at this, but my hunch, for what it is worth, is that the origin of adenosine triphosphate came very early (or at least adenosine diphosphate). ATP is an energy currency (carrier) for cellular metabolism. The third phosphate bond is very high in energy and can easily be broken in reactions, releasing that energy to drive syntheses of various kinds. Moreover, adenosine is one of the four code molecules in RNA and DNA (the others being guanine, cytosine, thymadine, and uricil in RNA in place of thymadine). This looks suspicious to me! And, if it weren't enough that adenosine is a key player in energy flow and information coding in genes, it is the end molecule in the CCA terminal of transfer RNA the molecule responsible for bringing amino acid molecules to the ribosome where protein synthesis takes place. Adenosine seems to show up in some very interesting places in the metabolism of cells (see also cyclic adenosine monophosphate for yet another intriguing role for the adenosine molecule!)
A second requirement is that the aggregate of chemicals have achieved some form of collective autocatalysis, especially a web with internal autocatalytic cycles that forms the basis of autopoiesis. Again, this suggests to me that a primitive metabolism, bounded by some form of membrane, emerged early from the pre-biotic soup and gave us the origin of life. Once these pre-bacterial forms found ways to encode their own structure (perhaps in RNA first) and maintain that code in primitive chromosomes life was on its way in terms of Darwinian evolution. Yet another step on the way to eukaryotic cells is suggested by Lynn Margulis' theory of endosymbiosis, wherein early bacteria having various 'talents', like photosynthesis, formed symbiotic relations with other bacteria that had, essentially complimentary talents, such as digestion of large molecules, to form cells with organelles.
The main point, however, is that energy flows into the system are captured (by photosynthesis and respiration in modern plants and animals), disbursed to working areas (metabolism) and the system is able to move things around, maintain and repair itself, and, well, live. Some time after primitive RNAs that actually coded for functional components, proteins that worked as catalysts (enzymes), and in turn, for structural components such as organelles and membranes, Darwinian evolution got a purchase. The rest is biological history.
What stands out in the story of the origin of life is the way in which the right combinations of Mech-Systems, interacting among themselves, can produce a system capable of interacting with other similar systems. That combination, capable of a degree of adaptability (later) transitions from non-life to life, an emergent property of all such systems. We'll take a peek under the hood of a living system to see how these play out.
Homeostasis and Autopoiesis
Homeostatic processes are at the core of living systems. You may recall from Systems Science 7 — Cybernetics: The Science of Control that homeostasis is a first order cybernetic system, feedback control. Homeostasis is generally a little more complex than I described it there. In metabolism there are critical factors, like pH (acidity), calcium concentrations in the protoplasm and so on. In multicellular organisms like us, there are larger scale factors that require managing within critical ranges. For example body temperature for us hot blooded mammals can't be allowed to get too low, nor can it get too high. There is an allowable range of temperatures (or any of the factors) that the body (cell) can tolerate and adjust to. But then there is a range outside of that usually narrow range where action must be taken to counteract whatever is causing the deviation from normal, what I will later call a deformation. Then there is a slightly wider range of values in which damage can be caused, the deformation reaches a breaking point. The homeostatic mechanism is the cybernetic controller that works to keep the values within the safe range. Its job is to prevent damage and if the system gets too far outside the safe zone, the mechanism must be able to muster more resources to wrestle the system back into alignment. For example in the case of body heat, if you find yourself exposed to cold temperatures and your core temperature is starting to go down (even a little bit) your muscles are commanded to shiver to produce heat. Your behavioral system also takes note and you move around attempting to get out of the cold. The shivering is an automatic operational control. The getting out of the cold is a higher level tactical control.
The homeostat (short for homeostatic mechanism, the controller in Fig. 1 of SS-7) has to be constructed (from genetic information) and maintained. So other systems are responsible for keeping the homeostat charged and ready for action. In fact there is a whole complex web of mechanisms that are there to keep the system, all of its various homeostats, operating within their proper ranges. Success is health. Failure is disease and possibly death. This larger complex of mechanism includes subsystems whose job it is to repair damage done to the system as a result of a homeostat's inability to maintain proper values. These subsystems are held in reserve, so to speak, against the need to do repair work, but obviously the system would prefer (if it had a brain!) to not have to activate them.
This larger network is the self-maintaining autopoietic system described in the Wikipedia article. The autopoietic network surrounds the homeostats and acts generally as the logistical control for all the lower level operational subsystems (and their homeostats). Actually it is more complicated than that might sound since every subsystem has its own control parameters, even the autopoietic components, so in a sense every subsystem is supporting every other subsystem, even if indirectly. Nobody ever said understanding the details of a living system would be easy!
Fig. 1 shows a highly simplified diagram of a system with autopoietic maintenance surrounding a homeostatic core process.
Figure 1. A homeostatic core is surrounded by multiple mechanisms for maintaining and repairing structures. Each such component in a complex network has its own homeostatic control.
The "maintenance processor", actually many such processors might be involved, actually has an internal structure similar to the example homeostatic core process. However, in its case the environment IS the pictured homeostatic core process. A distortion or disturbance in the pictured homeostatic core process acts as the stimulus to the maintenance processor and its homeostat kicks in to cause the maintenance work to commence (shown as the black & white arrow pointing toward the homeostatic core) as its response. Inside a living cell and in your body there are literally thousands of low-level mechanisms interoperating to maintain the system in its function within its optimal ranges of factors.
Under nominal environmental circumstances (the right environment for the particular organism) every thing works in concert and the organism is "happy". But what happens if one or more environmental factor gets out of a comfortable range? This is going to trigger the appropriate control mechanism to counteract the disturbance. The autopoietic system will channel resources to meet the emergency and with luck and good behavior the system may be able to reestablish ideal conditions. However, should it fail to counter the effect before damage is done, then other mechanisms will be brought on-line to repair the damage done. If the damage overwhelms the system's capacity to repair, then, of course death follows. Interestingly, as we will see, in the non-death scenario this is the basis of adaptivity. Here is the argument.
The autopoietic system maintains an energy budget for directing internal energy flows to where they are needed to do work. This is part of its logistic control function. The budget is flexible in that flows can be adjusted on demand and within reasonable ranges, much like the ranges of deforming stimuli we discussed above. So, for example, if there is an increase in demand to move molecules within the cytoplasm, more ATP can be routed to the actin-based fibers responsible for the movement. Thus they can respond to the demand. This of course would be at the short-term expense of some other subsystems that might get less ATP as a result. More importantly, the autopoietic system can keep track of these shifting demands over time and make longer term adjustments based on longer term demand.
A good example of this adjustment based on longer term demand is your muscles. Suppose after a lifetime of being a couch potato you decide one day to get into shape by working out, maybe lifting weights. Initially your muscles are not competent to handle the weight lifting without some pain (and some damage), even if you start easy and work your way up. If you only tried the weight lifting once and then quit, your body would not do anything except complain with pain or stiffness the next day. It would have certainly routed energy into the used muscles to accomplish the goal at the time. But then it would go back to the 'regular' energy budget and that would be all there was to that.
But if you work out regularly, your muscles send signals back to your growth control centers expressing an on-going need for additional resources, both real time energy input and reinforcements in the form of more muscle cells. Your muscles actually retain a kind of memory of the effort you have been expecting over a longer period of time and training. So the body responds by coordinating several processes. Your diet may start to include more protein, for example. Your growth hormones for muscle bulk may signal muscle cells to replicate to bulk up your muscleclature. You basically build a physique that is capable of responding to the long term demands that have been put on it. You adapt structurally within the limits possible for doing so. The autopoietic system has responded to the longer term demand by marshalling resources to build more prepared structure. This so that the system is better able to respond in a faster time and avoid damage. Of course a new balance in the energy budget is needed; more energy will be routed to maintain the new, larger subsystem. But the tradeoff between energy needed to maintain the subsystem and the energy needed to repair damage, and even to respond to the demand is quite favorable.
Here is a paper I did some years ago, "Foraging Search: Prototypical Intelligence", which lays out the case for why learning evolved in the brain based on these energy budget arguments. If you look at the graphs you will see an interesting aspect of adaptivity (in the case of neurons, adaptivity is the change in synaptic efficacy that leads to changes in behavior). The first graph shows a non-adaptive stimulus and response along with accumulations of energy costs resulting from three different components. Costs, in this sense are based on the notion of opportunity costs. The energy is used to take care of the response, but it might have been used elsewhere had the stimulus not occurred. The first component is the energy cost to respond to the stimulus. Without going too deeply into the biochemical aspects of this, it is the case that the more prepared a response subsystem is, the less total energy it will require to respond to the stimulus! This is because of the lag time between the onset of the stimulus and the time at which the response becomes strong enough to begin counteracting the stimulus. A more prepared (stronger) response subsystem can react sooner and with greater force, diminishing the time lag and the total amount of work that needs to be don to respond. Thus, a minimally prepared subsystem will end up expending more energy for a given stimulus-response. The basic idea in structural adaptation is to minimize this cost.
The second component of energy costs is for repair of damage. In principle, if a subsystem is better prepared, not only will it minimize response costs, but also quell the stimulus sooner and avoid or minimize damage repair expenses. Finally there is the energy cost of maintaining the structure at an elevated level of preparedness. Bigger muscles require more food, even when you are not working them hard. However, this cost is only incrementally higher than at, say, a baseline level of preparedness. So while it is higher over time, it is marginally higher compared with the costs of response and damage repair.
A key idea here is that it is the long-term frequency of stimulus events that determines the degree to which the response subsystem needs to build up. If you were to lift weights only once per week, or stopped after working out every day for one week, your muscles would not bulk up to the Schwarzenegger level or even look that different from your couch potato days. But if you worked out, say, three times a week, for months on end, you would expect to see some more 'definition' in your muscleclature and muscle tone. If you worked out as much as Arnold did in his youth, maybe you too would look like the Governator (or more like the Terminator; age has its costs to pay!) The more frequently a stimulus occurs, and the longer the time frame over which it maintains that frequency determines how much adaptation is required to be 'ready' and 'able' to respond.
As I show in this paper, the benefits of a memory system based on building adaptive (bigger, more powerful) structures, is a system that can respond to the first hint of a stimulus and by earlier quelling of the stimulus, reduce costs considerably. But that isn't all that can be accomplished, at least in neural tissue.
Neurons, at their synapses, have evolved long-term associative memory networks (for more details see my paper, "Toward a Theory of Learning and Representing Causal Inferences in Neural Networks". Associative memory traces allow a neuron to create, essentially, a causal linkage between a non-deforming stimulus, such as the sight of a predator, and a deforming stimulus, such as pain from a previous, non-lethal encounter with that type of predator, so as to anticipate the outcome of letting the deforming stimulus occur. The third graph in the first paper shows that having such an anticipatory stimulus causes the system to preemptively respond such that repair costs are not incurred and response costs themselves, which are proportional to the stimulus strength, are very much more minimized. Thus, the system's total costs have been reduced at not much greater cost of maintaining the larger response subsystem and the associated anticipatory response subsystem.
While this is most visible in neurons, brains, and behavior control, the same general principle applies to all adaptivity in Bio-Systems. At the end of the Foraging paper I present the case for why Bio-Systems (animals in this case) need to evolve anticipatory adaptive response. The case hinges on the argument that the real world is non-stationary. That means that things are always changing. By this I don't just mean things are moving, or that the environment is subject to cyclical behavior (like the seasons). Rather this refers to the fact that totally new conditions can come into effect in an environment (when I get to Eco-Systems I will develop this concept much more thoroughly). Completely new stimuli may be encountered. But more often, new causal relations between old stimuli may arise. A Bio-System has to be able to adapt to these new arrangements and causal relations. It does this by always being able to flexibly adjust the energy budget and 'learn' the new associations. It must forever be able to readjust as needed over varying time frames.
The Limits of Adaptivity, Growth, and Development
One of the main things that differentiates Bio-Systems from Eco-Systems is that the former have internally imposed limits on their capacity to modify themselves in light of radically new stimuli. They are bounded by internal controls that limit their adaptivity as well as their capacity to grow or develop. While our genes provide us with considerable flexibility in terms of how variant we are in our development, and how much we can learn, they still put constraints on us in all of these ways. Of course, even Eco-Systems have limits when considered as a stable system. An old growth, climax forest can be utterly altered in hours from a bad fire. However, Eco-Systems are not closed to the invasion of other species and life on Earth has withstood numerous massive die-offs yet managed to adapt and life continued to evolve. You can't say the same for an individual, a Bio-System. [Note that individuals are very often colonized by all manner of parasites, forming a kind of Eco-System! But to the individual so colonized, the effects of the activities of these foreigners is part of the environmental forces acting on it.]
There are obviously good reasons why Bio-Systems are self-limiting. First there is no reason that a Bio-System should have to adapt radically and constantly, taking on virtually any form needed to survive. The environment doesn't change that fast. Individuals are part of a population and whatever changes do occur don't necessarily impact all individuals. Further, because of the way genetics works, there is generally a reasonable amount of variability of traits and behaviors within a population such that if some individuals are affected by an environmental change, not all will necessarily be negatively affected. That is natural selection at work. The one rule you can count on in Bio-Systems is that no such system will expend energy on some capability that isn't absolutely necessary for survival.
Energy arguments can be brought to bear on growth limits as well. Bio-Systems, with a few exceptions that are reasonably understood from an evolutionary point of view, are programmed to grow to a needed size relative to their eco-niche and then stop. The same does not apply to populations of Bio-Systems, however. Populations sizes are controlled, as a rule, by Eco-System internal dynamics that I will cover in that installment. But at the individual level, growth is carefully regulated by the logistical control system so as to keep any single individual within a population from dominating completely and possibly wiping out other members. Populations need to maintain variability and so domination by one fast growing individual (or even a few) would be counter to the long-term health of the species. Hence, individuals are restrained and kept within an admittedly variant range, but with a relatively small variance.
Finally, it would be completely self-defeating for an individual to undergo form and behavioral modifications that effectively made it unrecognizable to other members of its species. In the end, given that the only means of continuation of Bio-Systems' patterns is through reproduction, and that requires successful mating, con specifics have to be recognizable to one another. Neighboring Eco-Systems hardly care at all about what the others look like, or what their composition might be. But individuals need to mate with similar individuals in order for the species to continue. This might seem like a silly thing to be concerned about. But back in the beginning of life, it is very likely that the first proto-cells and very early cells that invented sex or sex-like reproductive strategies had more capacity to take on new forms since they had not yet evolved internal control systems to ensure species continuity. Of course, those that changed radically enough to not be successful in reproduction soon disappeared from the scene leaving those that could maintain species-specific forms and behaviors. Evolution is the great leveler.
Why Reductionist Methods Work on Bio-Systems (and are hard in Eco-Systems)
I'd like to end this installment with an observation about some methodological aspects of studying Bio-Systems, especially versus Eco-Systems. The relevance to the latter may need to wait until I've finished that installment to be fully appreciated. But I'll at least introduce the ideas here.
The fact is that reductionist science has been wildly successful at producing our knowledge about Bio-Systems and natural Mech-Systems (although engineering of man-made Mech-Systems may involve some reductionist thinking). The reason is that these systems tend to have more definable boundaries and boundary transfers with functional attributes that are relatively easy to measure. Put bluntly, you can dissect a representative system and find its parts and their relations to other parts. It may take advanced technology (e.g. electron microscopes) to get there, but, in principle, all functions are measurable.
This is a pretty strong statement, to be sure. But I think if you explore the history of biology you will see repeated examples of functional decomposition illuminating the inner workings of system after system. The use of model building as a way to study whole systems dynamics, has been less useful in understanding Bio-Systems as it has in Eco-Systems precisely because access to the system internals is so readily available given that you have the dissection tools to do it. And the history of biological science is that when you don't have the tools, you either put off the questions or look about to create the tools. The invention and development of imaging technologies for brains (and other working tissues inside live organisms) is a case in point.
It isn't uncommon to hear a critic of reductionist methods complain that reductionism fails to explain the whole organism. The mantra one most often hears is something like "The whole is greater than the sum of its parts." It all depends on what you mean by the word 'sum'. If you mean that if you just throw a bunch of parts together they will fail to produce a living system, then DUH! Bio-Systems and natural Mech-Systems are the sum of their parts AND the functional relations between the parts. And reductionist methods have done a splendid job of discovering and explaining all of that. [BTW: if you read the article, above linked, be sure to note the difference between reductionism in science and the philosophical notion that reductionism means 'reducing' explanations of complex systems to simpler principles (rather than parts and relations), e.g. reducing biology to "mere" chemistry. Of course, as we saw in the origin of life above, you have to be able to explain the biological phenomena BASED on chemistry, but this isn't the same as reducing a complex system to just its chemistry. I'll probably write more about that issue at some point in the future since it is such a common argument leveled at science, motivated, I think, by a desire to knock science off its high horse as the best method to obtain knowledge. As you might guess, I have very little patience for those arguments!]
With Eco-Systems the situation becomes incredibly much more difficult (and remember, Eco-Systems include societies and economies, etc.). The very fact that Eco-Systems are highly variant and non-stationary in nature makes attempts to understand relations between components by taking such a system apart problematic. Relations may be continually changing over time in non-stationary ways. Sampling the components itself is messy and hard work. The science of ecology is still grappling with reductionist methodology and probably always will. Where the Eco-System includes humans sampling becomes even more problematic for ethical reasons (of course this can apply to Bio-Systems studies that involve testing on animals). Sometimes it is the case that attempts to dissect parts of an Eco-System do destroy the sought relations and that fact may go undetected by the scientist. Fortunately, another methodology involving the construction of models (mostly computer-based) can be brought to bear on Eco-Systems. This is generally regarded as an integrative or whole-systems approach, essentially the reverse of reductionist approaches. But it is also true that models need to start with some basic understanding derived from field studies. The model must try to capture what is known about the system from reductionist studies, but at some point it can become a powerful tool, if used correctly, in positing hidden relations and suggesting places to look for previously hidden relations. I will likely do an installment devoted to computer modeling in the future to explain this tool.
While I said that modeling has been less used in Bio-Systems studies, that is starting to change. Today very powerful computer models are used to study genetics, metabolism, and neural networks just to name a few. These models are used in a manner similar to how they are used in studying Eco-Systems. That is they can be used to develop hypotheses that can then be tested in the reductionist sense. Thus there is another region of overlap between Eco-Systems and Bio-Systems. Next we'll take a look at this from the other side, that of Eco-Systems.