April 2008 PAEI
In Part 1 of this blog report I summarized the background of my research into PAEI foundations. I described how a very broad literature scan led to my uncovering 130 or so PAEI-type models in 15 different disciplines. I call these kinds of models “Structures of Concern”, and I still find new ones from time to time today, without even looking for them.
In many of the 130-or-so cases I reviewed, PAEI-like concern structures are *independently* discovered by investigators in their respective studies. This suggests something those of us in the Adizes network already know and see every day – PAEI seems to describe an objective reality. PAEI must be objectively real in some way, since people who study different aspects of reality continue to stumble upon PAEI, expressing it in different terms whenever they re-discover it.
It may be possible to save everybody the trouble of re-discovering PAEI, by developing a theory of PAEI that will allow us to predict the kinds of situations that will give rise to it. Creating such a description would increase the parsimony of human knowledge, and would further validate the power and accuracy of the Adizes Methodology, not just in the field of organizational studies, but in the field of general knowledge as well. Adizes concepts would echo throughout many fields of human investigation.
Why Adizes concepts? If I discovered 130 PAEI-like models, why should the Adizes version of the structure of concern have any special status among them?
There is an objective reason for according the Adizes PAEI concept special status among the others. Other concern structure models I uncovered almost always apply at only one level of analysis. In the Adizes Methodology, however, PAEI describes individual psychology, role/task types, organizational structures and organizational lifecycle dynamics. So in the Adizes Methodology, the same concern structure is described at the subjective, objective, functional, macro and micro levels. Furthermore, there is an account of how each level of analysis intersects with all the other levels.
It is this integrated, multi-level quality that makes the Adizes PAEI model the best reference model against which to compare all other concern structure models. It creates a common frame across many levels of analysis, allowing us to draw comparisons between things as disparate as learning styles and forest ecosystems. If PAEI did not play such a multi-level role in the Methodology itself, it could not have become the reference model for my study. The multi-level applicability of the Adizes concern-structure model (or at least the one on the “decision” side) is what made my study possible. My choice was not just driven by my own personal familiarity with the Adizes Methodology.
So, to return to the question, how do we create a causal, PAEI-based explanation that allows us to group together learning styles and ecosystems into one category? How do we link those things together the same way that PAEI links together management styles and organizational lifecycle dynamics?
I have no definitive answer as of yet. All I can deliver is a progress report.
I think a theory of the structure of concern will have to be some kind of general systems theory. These concepts apply to systems of different types and of different orders. A famous general systems concept is the concept of “feedback”, which applies in many different contexts. For example, I can give an employee “feedback”, but household appliances also get “feedback”. An oven gets feedback on the adequacy of its output (heat) from the thermostat, and turns the heating element on or off to maintain the right temperature. In a food web with predators and prey, the predator population gets feedback regarding its reproductive behaviour from the size of the prey population. If there are fewer prey, the hungry predators do not reproduce as much. If prey is abundant, predators can store energy and they reproduce more. Then if they reproduce too much, the number of prey diminish, so predators get the feedback not to reproduce as much again.
All of these are examples of “feedback” in very different domains. PAEI also applies across domains, like the concept of feedback. It probably needs to be expressed at the same level of abstraction and generality. So the ultimate explanation can’t rely on human personality or human needs. Even before there were human beings, systems had to fulfill PAEI functions.
Besides domain-neutrality, another clue towards building a theory of PAEI is the clue of scale. I did not find a whole lot of PAEI-like dynamics in fundamental physics – either the tiny level of quantum theory, or the huge level of relativity. Concern structures may operate at those scales, but not in any screamingly obvious way that I have found. PAEI seems to be a middle-scale phenomenon – the scale at which systems concepts like thermodynamics and evolution are quite visible. It seems specifically to relate to the coordination of systems with multiple interacting elements.
In terms of the characteristic of complexity, PAEI models often emerge in descriptions of systems that are much simpler than human cognition or personality. The concept spans many orders of complexity at the medium-scale. So the theory of PAEI is not a theory about the human brain. It operates at a more fundamental causal level than that. In terms of the characteristic of agency, we do find PAEI-like dynamics described in totally inanimate system, but they are less conspicuous. In animate systems and agent-based system, PAEI-like dynamics become more strongly visible. So the explanation can be born from an explanation of multi-agent systems, but it can’t be described in a way that would exclude application to inanimate systems.
Theses are the qualities that a theory of PAEI would have to exhibit: domain-neutrality, medium-scale and applicable to both simple/inanimate and intelligent/animate systems. I still don’t have a theory to offer you, and I hope some of you may be able to help me find it. However, I have singled out a few components for a potential final theory. These are some of the explanatory ideas that might fit the requirements listed above.
Hierarchical causation deals with part-whole relationships; also called nested or inclusive hierarchies, (e.g. molecules within cells, within organisms, within ecological niches, etc.). Smaller or lower-level events enable and constrain what happens at higher levels in the hierarchy, for example the shape of a stone tower is dependent upon the shapes, sizes and strengths of all the little stones it is composed of. Higher-level or larger events also impose constraints upon a system, so the shape of the “tower-system” is also determined in part by very large-scale conditions: the gravitational field of the earth, the solidity of the ground under the tower, and the prevalence of high winds and storms in that region all play a role in determining what kind of tower it is possible to build at that site. This combination of bottom-up and top-down causation upon the “focal system” is hierarchical causation.
Hierarchical causation plays a big role in theoretical ecology. Take the example of global warming. Carbon is being released into the atmosphere from billions of natural and industrial sites to increase global temperatures (upwards causation). Those higher temperatures then act on the planet, causing the earth to release carbon at an even faster rate (from the soil, from heat-accelerated decomposition, etc.). This is downward causation – a change in boundary conditions for all systems. In this case the two levels are connected in a feedback loop – a positive feedback loop, otherwise known as a vicious circle: more carbon release increases temperatures which accelerates carbon release.
All medium-scale events have a hierarchical structure in time and in space. This is an important point. A song, for example, is composed of notes which make figures within bars to add up to a song. Our lives are like a song, in this sense. We engage in a lot of little activities which add up to bigger activities and which ultimately add up to our lives. We have evolved to live on many time scales at once, from the very short timescale of our single cells which are born and die every day, to the cyclical periods of days and seasons, to the timescale of our careers or marriages, etc. You have active concerns at all of these levels at once, in each and every moment of your life. Organizations also have a hierarchy of efforts, ranging from the task to the project all the way up to the corporate mission. In any time-slice, activity is happening that is relevant to all of these different scales, simultaneously. PAEI and other similar models emerge in part because of this general feature of event structure, I suspect.
This should come as no surprise. In the Adizes Methodology, one of the first things we do when we are explaining PAEI is to talk about the short-term versus the long-term concerns of the different styles. Short and long time-scales are thus already central to the PAEI concept. This core Adizes insight can serve as the basis for further theoretical development by equating it with the slightly more elaborate models of hierarchical causation that have been developed by prominent theoretical ecologists (e.g. Stanley Salthe). In other words, we expand the discussion of what “short term/long term” time structure really is.
ADAPTIVE RENEWAL CYCLES
Adaptive renewal cycles are another construct from theoretical ecology, more specifically from within population ecology – which has already made inroads as a school of thought within organizational studies. Adaptive renewal cycles deal with the availability of energy and other resources in an ecological system, complementing the concept of hierarchical causation, which deals with the structure of events in time.
Adaptive renewal cycles describe a progression of resource environments, and the kinds of organization that produce the most effective and efficient use of resources under those conditions. For example, say a farmer ploughs his fields in the spring. This creates a muddy field lying open on the plains, which is a rich source of food for plants. This kind of resource environment favours a certain kind of organizational action – namely, a gold-rush. Plants will soon take root there – mainly fast-growing, fast-spreading weedy plants, who access the food quickly and reproduce quickly, scattering millions of seeds everywhere hoping that at least some of them will sprout and grow. That is the best strategy for getting at those free resources before any other species does. The advantage will go to those species of plants that execute their lifecycle as quickly as possible. Accuracy is not important. Seeds can be tossed to the winds by the handful, so long as just a few of them hit the ground before competing seeds do. This is a P-style selection regime (called r-selection in population ecology, which we can think of as “rapid” selection).
However, a P-style explosion of activity cannot last forever. Eventually the success of the P-strategy becomes its own downfall, because the field is full of weeds and there is nowhere else to spread. Furthermore, easy-to-reach resources (soil for plants, sales for corporations) have been used up, and with growth, there won’t be enough for everybody. At that point, more specialized plants appear, with more internal differentiation to support their more specialized metabolisms. They put down deeper roots. They tend to be larger because they contain food and water stores as a buffer against periods of time with no resource availability. They create larger, more complex seeds that also contain food stores, and may have protective coatings like shells, so they can rest dormant until times are good and resources are available to start growth.
This is a more A-style phase, all about getting a niche in order to avoid conflict (differentiation), building up buffers and back-up plans, and reducing waste by investing more on each seed (that also carry buffers and thicker boundaries). In population ecology these are called K-selection regimes, which we can think of as the selection regime that is in place when the “Carrying capacity” of the region has been reached.
As the community of differentiated specialists evolves, integration dynamics emerge. Integration becomes the next available dimension for growth and access to energy, through the development of more complex food webs. A community of differentiated species starts to get “sewn together” as it matures and complementary species move in to ensure that every available resource is used. Interdependence increases, amplifying both differentiation and integration simultaneously. The Amazon rain forest is a very good example of a very mature ecological community which has become a fantastically complex web of highly differentiated, highly interdependent species. This interdependent web is an I-style selection regime.
As the integrated web gets more saturated, more and more innovative specializations become needed to find available niches, but those opportunities become very rare in what ecologists call a “climax” community. At that point of saturation, there is only one way to release the energy and matter bound up in the forest – through an act of creative destruction. Either a new trait evolves for a local species, or a new species evolves, or a species from another region arrives, and these organisms are not bound into the web. They sweep through the region, consuming everything in their path, or perhaps a meteor or something crashes through the forest (or there is a fire, etc). Some act of change, novelty or creative destruction releases energy and resources bound up in the old food web. Old dependencies are disrupted, and as a result of that event a new P-like phase starts all over again. This disruption of the AI web to unleash a new P-like growth period is an E-like event.
If hierarchical causation covers the temporal aspect of PAEI (short and long-term), the adaptive renewal cycle speaks to the effectiveness/efficiency concerns – of different selection pressures that make different strategies optimal. We remain on very familiar ground: PAEI is about effectiveness and efficiency in the short and long term.
GENETIC ALGORITHMS/EVOLUTIONARY PROGRAMMING
There is a new approach to computer programming that is slowing making an impact on human and economic sciences. It is called evolutionary or genetic programming. In this style of programming, you do not write a program that tells the computer what to do. Instead, you define the parameters of a problem, create a population of possible solutions, have rules which allow the solutions to “breed” or cross-fertilize each other, and set up selection forces that weed out unfit solutions. These solutions can be modeled as agents, who have to carry out certain tasks under various conditions using a diverse set of strategies. A program like this is run over thousands of iterations, and solutions take their form over time. This kind of programming was not possible until cheap high-speed computing became widespread, and it is already starting to change the way we understand economic life. Eric Beinhocker reviews some of this economic work in his book The Origin of Wealth.
I believe that hypotheses about PAEI will be ones we can model as evolutionary problems or multi-agent optimization problems. Thus it should be possible to set up conditions in a program that will draw attention to how PAEI dynamics arise as a response to certain aspects of the hierarchical spacetime and energy structure of events. From there it should be possible to illustrate exactly how, for example, the brain evolved to produce PAEI-type motivations in humans; or how/why a particular organization needs to overcome a particular lifecycle challenge, etc.
AND THERE IT SITS
Three years of searching brought me to this inconclusive point, and I could not afford to steal any more time away to study the problem. Plus, I am not a programmer of any stripe, so I reached the limit of how far I could progress alone. Those of you who are still reading will hopefully understand that if we manage to explain PAEI in objective terms in other disciplines, this will be pretty good for the Adizes “brand” and public recognition of the brand. However, that is a happy byproduct of the research.
This research was undertaken to learn something true about the universe. I have already said that the Adizes PAEI concept was chosen not because of my association with people in this network, but simply because it was the most multi-level, multi-order model I found. I think there is something true and useful here, that can help us deepen our awareness of what exactly it is that Adizes Associates *do* with client organizations. And that is just the tip of the iceberg. So I remain excited about the prospects for this research, even though for the foreseeable future it must sit on the shelf.
And there it sits.