Fig. 7. (see bigger image)
6. Scheme of Information Interactions: The Main Blocks
Institutional structures set the limits of people’s interaction in the process of their socioeconomic activity (see above sections). Separate institutional structure could be imagined as a meta-model or a framework of a kind of interactions between individual agents that conduct activities “inside” it. The typical functions of such framework are the following three:
1. The switch of the joint activity participants (agents) from the “free” state that precedes their amalgamation within a certain organization, to a constrained state when they are already organization members and have formed the necessary relationships with its members. Here, institutional mechanism is working of coordination of opportunities and intentions of the agents relative to their potential joint activity.
2. The use of organizational relationships between agents for attaining the goals which were the reason for the establishment of the relationships. Coordination of the organization members’ activity in the process of joint activity which at this stage has been exercised via organizational mechanism.
3. Termination of membership in organization (breach of relationship) and agent return to the free state.
Further we shall examine the work principles of the first item in detail, and second – in general.
There arises a question: what is the content of specific human activity that creates an efficient institutional mechanism?
According a general picture of institutional structure, of direct information interactions, of "cognition cycle", and etc that was discussed in sections above, we can summarize the following:
By means of interaction exchange, agents participate in the collective formation of a certain information image (model) of their environment (Michael Lachmann, Guy Sella, end Eva Jablonka; Scott Moss and Bruce Edmonds) and possible joint activity. Agents use this information model for simulating different variants of their activity. Doing this they verify and coordinate their viewpoint of the content of their joint activity as well as define the place of each agent within the system of labor division between them. At a certain stage of coordination and specification the agents begin to consider the current state of information image of the joint activity to be acceptable for practical implementation. After that they are able to make the appropriate re-configuration of the links, and to readjust the type of their activity. In the process of re-configuration production relations are changed through which intermediate results of agent production activity are disseminated (in accordance with technological series). Also, distributive relationships can be changed through which the final results of the agent collective activity are fed back to them in the form of resources for maintaining their activity.
The outline of agents information interactions in a more detailed and formalized form is represented in following sub-sections:
1. Let us specify a set A consisting of n agents À(i), where i = 1…n.
2. Specify space Ð in which agents conduct their interactions and which contains the whole agent set A as well as set Z of the rest of economic system players different from the set A: Let P{A, Z}. More detailed description of the features of the space for economic interactions see in above sections.
3. Set macro-technology Ì. Its description sees above in the previous section. Let the number of jobs in the macro-technology be n and, consequently, it will coincide with the number of agents in set A. For each macro-technology job Ì(j) (where j = 1,…,n) its output is V(j,i), defined for each agent À(i) from the set À. Note that "output" here characterizes a production process result, “output” in the sense of final consumption is defined below. Thus, a certain agent k has a coefficient vector V(j,i=k) of his personal outputs that the macro-technology will produce if he takes each of the jobs Ì(j). Assume that agents “know” their output values. E.g., they experimentally found those values in the past.
Surmise the existence of differences between agents in their skills relative to different jobs. Let there exist a clearly defined agent specialization within the macro-technology set of jobs for the existence of solutions for optimal agent distribution over the jobs. Thus, we do not tolerate that a large number of agents from the set A hold maximum output values for the same macro-technology jobs. A small number of such coincidences will not, apparently, influence the existence of at least one optimal solution.
For the sake of simplicity, for the time being we ignore such variable as agent costs per unit of output. Let all the agents from set A have equal cost values in the process of their activity.
4. At the start point of the interaction scheme, the set of agents will be randomly distributed over the jobs M(j). As a result, for a certain current moment we recorded what output is “produced’ by each job. Total output V of the macro-technology M is defined by the sum of values V(j,i) for all the jobs, where i is the number of the agent who took job j (items in this sum are equal to the corresponding output of agents occupying certain jobs). Assume that there exists at least one distribution of agents by the jobs – any deviation from this distribution means diminution V (Pareto optimum).
5. After the total output V on the use of macro-technology M is formed, this final product of the production process becomes a resource for maintaining living conditions and development. Now it is designed for distribution among agents. Let V be distributed in a certain manner over the whole agent's set A. Denote the distribution function as F. Then the share of total return received by each agent in the process of distribution, is F(V, A(i)). Assume that function F is linear by the agent return factors (i.e., the higher the agent’s output, the higher share he receives in the process of distribution V).
6. Each agent À(i) has a mental model (D. North, 1994, p. 360) of space Ð{A,Z}, which we denote as Î(À(i),Ð). This model is initially created and exists in the human mind. The mental model as a result of psychological reflection of the environment by humans, is a copy (to a certain degree of accuracy) of the structure of real space Ð (including the relationships between agents). The difference of mental model from its real prototype can be equal to zero (in this case, mental model identically reflects all the economic system components as well as their changes in time), or can have significant deviations. To describe the general case, let us introduce the distortion function Ñ(A(i)) that can be individual for every agent. Then identity O(A(i),P)=C(A(i))*P{A,Z} holds. Let us assume (for simplicity) that there are no distortions, i.e., C(A(i))=1.
In our description of the mental model we based on some related research:
According to D. North, “humans perceive the outside world by means of processing the information with the help of pre-existing mental patterns that enable for understanding of the environment and solving the problems arising (D. North, 1997, p.7). "Ideas and ideologies do form subjective mental patterns; with their assistance individuals interpret the surrounding world and make their choice” (D. North, 1997, p.143).
B. Edmonds in his article (Bruce Edmonds) using the techniques close to genetic programming, designs the simulation description of the agent behavior that includes the population of the agent internal models describing the environment that surrounds him. There is also similar research on sharing information to support an evolution of collective actions (Michael Lachmann, Guy Sella, Eva Jablonka).
Mental model will be wholly functional only when there is a mechanism that continuously keeps it in the actual state. For this purpose, information images of partner agents in the mental model should be sufficiently well fit with the state of real partners. Under small group conditions, a member of the group makes an actualization of his mental model via continuous direct information exchange with the other partners. Interactivity of actualization procedure enables to view the mental model as some collectively maintained substance that is not already the product of an individual agent; however, it can exists only in his consciousness. Starting from a certain level of information technology development, mental model can be estranged from the agent's mind to a certain information carrier; after that it becomes one of the objects in the agent information habitat.
Conceptions related to the agent mental model can be implemented as the following interaction scheme items.
7. Mental model Î(À(i),Ð{A,Z}) is used by agent to design and simulate new configurations of relationships (to take a new "job" in macro-technology) with the purpose to enhance its output. The deviation of the current agent distribution over the jobs from the theoretically optimal is a reserve of future output enhancement. In the process of using the model agents can find other variants of their distribution by jobs that are more preferable than the current variant by output of each of them. Thus, the agent generates a new configuration of the system's space in his mental model À(i): Î(À(i),Ð{A,Z}) -> Î'(À(i),Ð{A,Z}).
In his mental model, an agent can construct the variants of changes in the current relationship pattern desirable for himself. This means that, as a result of accomplishing his local task of increasing the return from his activity, the agent is able to define a desirable version of reconfiguring the links between the agents. In the general case, those proposals can pertain not only to relations of a particular agent but any other combinations of relationships. Local optimization task of an agent can be written as: max F(V, A(i)), by variable V obtained by sorting out the set of possible variants Ð{A,Z}.
Rather similar approach to the study of social systems based on the agent computer modeling is proposed by S. Moss and B. Edmonds (Scott Moss and Bruce Edmonds). In their work "Modeling Learning as Modeling" they describe how agents can lower the errors in their forecasts via the change in the structural forms of the environment models used by them. This is attained by providing the agents with the opportunity to build the model of their habitat. "Specification of our agents’ models is based on their observations of the data and the success with which their models forecast the needed variables".
Collective information model of the environment
8. By definition, one of the aspects of agent interaction under direct information exchange consists in the mutual information on their capabilities and intentions. By means of such information exchange of the “everybody with everybody” type in the real time mode, mental images of capabilities and intentions of each of them become known to all the rest of the agents.
As a result, mental model of each agent is, actually, a certain combination of mental models of all the agents. Information activity of the partners creates their virtual presence in the private mental model of each agent from the set A.
| Figuratively speaking, each of the partners is “responsible” for the content and actuality of a certain portion of the agent’s mental model. Each model fragment is filled and renovated (with certain accuracy and periodicity) as a result of information activity of one or several agents that have been reflected in the given fragment. Thus, the formal notation for a mental model should be rewritten as: Î(À(i),Ð{Î(À(l),Ð{A,Z}),Z}), for all values of l = 1,…,n, but not equal to i. |
Fig. 7. (see bigger image) |
9. It has been already discussed above that there are several parameters that determine if a direct information exchange within the given agents' group is possible at all. In this connection, assume that information exchange intensity combined with the intensity of changes in the states of agents and other objects of space P, as well as the ability of each agent from set A to process the information, are sufficient for the direct information exchange being necessary for maintaining the mental models of each of the players in the actual state. If this is feasible, than within the given agents' group there should be a tendency to the increased similarity of their private mental models.
Assume that the whole set of agents constructs and plays through the possible variants of reconfiguring the relationships between them in the form of a unified, collectively maintained information image of their habitat. Let us denote such image as “collective information model" of an economic system (their living environment). In the general case, this information model is stored in the mind of each agent; however, other variants are also possible.
Assume that the information technology development level makes it possible for the agents to estrange their mental models into the outside environment in the form of certain information objects. If this is done by a unified “standard”, then it is possible to make a composition of the set of such information objects. As a result, the agents receive information model of the system that exists as a substance being outside their consciousness. In a sense, internal models of the agents are materialized as an external model common for all the agents from set A. In this case, we can speak of the existence of a virtual variant of the model in addition to its mental variant.
It is characteristic of the interaction process under the direct relations between agents that they “discuss” their actions between themselves, “agree” upon mutually acceptable actions and then “coordinate” their joint activity that they agreed upon. Subject to our hypothesis on the existence of such a substance as “collective information model of a system”, the general outline of interactions described above could be specified as follows:
10. If the agents have built a new information model of their living system that would provide for the higher output from their activity compared to the current system, the real system is changed in accordance with this model. Here, the pattern of information links established by the agents in the process of formation and coordination of the new information model of the system, becomes the prototype of organizational structure and the basis for organizational mechanism. This mechanism would ensure the maintenance and support of the new link system (organization) in the process of its practical use.
Thus, institutional mechanism having accomplished its function on moving the agents from the "free" to the "bound" state, as a final result of its work, provides possible all the necessary components that agents need for starting their joint activity within a particular established organization.
Material and information sub-spaces of interactions
Going back to the content above on institutional structures as sub-spaces of agents' interactions, we can note that there are also two another sub-spaces that mediate different kinds of agent activity:
First - material – includes real processes of creation, distribution and consumption of resources.
Second - information – is the result of mental (psychological) reflection of the first sub-space. It includes the processes of forming the information image of the environment as well as collective construction by the agents of a new information image of the first sub-space on this basis.
Agents form in the information sub-space the image of a new desirable material space and then rebuild the latter in accordance with this image. There is a certain cyclic recurrence in this process: information images of new links and types of agents' activity being born in the second sub-space, are partially materialized within the structure of the first sub-space changing its current state. On the other hand, the new state of the first sub-space being reflected in the agents’ mind, becomes the basis for generation of new states and information images filling the second sub-space.
In compliance with those sub-spaces, agents network of links formed at the “exit” of institutional mechanism, can also be divided into two types: first – network of links for resource exchanges, second – for information exchanges.
Next section - 7. General picture of economic system functioning through information interactions
Previous section - 5. Scheme of Information Interactions: Causes and Motives of Interactions
