Содержание материала

Human Behavior Model (HBM) overview


The model is implemented as a computer program (Figure 1).

english1
Figure 1. HBM ver. 2.00 screenshot (Java implementation).

The program has a modular architecture (Figure 2) which allows connection and disconnection of entire logical blocks and ensures dynamism, flexibility and interactivity to the program.

english2
Figure 2. Modules of the HBM.

The current stage of model’s development can be described as laying the foundation. At this stage the core model is being designed, i.e. those elements whose absence make model’s work absolutely impossible. Furthermore, during the foundation stage all major features and capabilities of an artificial human being that are expected in the model, will be developed and described. However, description of a human at this stage will be greatly simplified, since the main goal is to form the structure of the model.
By and large, the model provides an interactive environment for a variety of humanoid objects, a kind of digital anthill with computer people instead of ants.
As for the logical structure, the main element of the model is an artificial human with his various properties and characteristics that determine his behavior in the environment. All people in the model are variations (sometimes very different from one another) of this archetypal artificial human.
Conceptually the activity choosing mechanism is the heart of artificial human and of the entire model. The choosing mechanism ensures meaningful, non-chaotic dynamics of the whole model.
In addition, at this stage, artificial human is endowed with an incentive system, with some resources and with physiological systems. But first things first.
Certainly, such problem as modeling human behavior can’t be solved with a clean slate. Many researchers from different areas of knowledge whose ideas, among others, are used in the HBM were or still are involved with it in varying degrees (Figure 3).

english3
Figure 3. Theoretical filling of the HBM.

But back to the logical structure. If an artificial human is the central element of the model, then his activity would be the second most important element. Activity is implemented as an independent class (i.e. a blueprint for an object in Java language), so it has certain autonomy with respect to the human engaged (implemented as a separate class as well), while at the same time activities can only be performed by humans. Activities have a number of properties, such as duration, type of result or energy costs for execution. A specific goal is associated with each “activity-human” tuple. For simplicity and to reduce conflicts, some results of activities present in the current version of the model are fixed and goals often coincide with them. At this stage, 9 activities are available in the model (Figure 4). They were not chosen at random: time-budget surveys conducted in several countries suggest that an average person dedicates about 90% of his time to these 9 activities (along with the “travel/commuting”* activity). All other activities require considerably less time.
* The “travel” activity is not represented in the model for technical reasons, namely due to the fact that humans in the model are not related to spatial coordinates.

english4
Figure 4. Valid activities of the HBM.

Due to persistent availability of alternatives the problem of choosing activity comes to the forefront for a model human. Factors determining current importance (needness, i.e. the willingness to spend time and effort on activities) of activities are described for every activity. Activity is then selected on the basis of the current value (which is dynamically calculated) of its needness: the one with the greatest needness is to be chosen.
For example, activities such as “sleeping” or “eating” are determined by physiological processes, so their needness is linked to human’s fatigue, circadian rhythms (for “sleeping” activity) and energy balance (for “eating” activity). In turn, fatigue and energy consumption depend on daily activity.
Whereas needness of activities such as “self-care” is trigger-based: a man wakes up - a man washes. In other words, the needness is guaranteed to hit some level if some simple condition holds.
In short, activities are designed in different ways, they produce different results and their neednesses are formed in various patterns.
However, time factor is among determining factors for all activities. Firstly, because time is relatively easy to measure and track (in comparison, for example, to psychological factors). And secondly, because lack of time is crucial for choosing. More specifically, activity selection problem actually vanishes without time pressure, because it means that a human has sufficient time to carry out all his plans. Non-instantaneousness of activities leads to the problem of time shortage.
Every activity brings a certain result when it succeeds or fails. Technically, the model human is described by a set of parameters characterizing his physical condition, his needs, his resources, and so on. Therefore, activity results are interpreted as changes in the values of these parameters.
Obtained result’s compliance with the goal of the performed activity improves human’s mood (makes him happy) while discrepancy worsens it (makes him upset). Human’s emotional dynamics form the basis for assessment of his happiness level (whose measuring became so trendy in some social sciences in recent years).
Since the model human is able to gain experience (remember the results of activities taken), he is able to adjust his behavior in order to avoid disappointments. While model human cannot significantly vary his activity performance, he reduces the needness of those activities that often used to make him upset and increases the needness of those that used to please him. Thus, experience (and emotions indirectly with it) states itself as one of the choosing factors.
Model human couldn’t be even tagged human if he didn’t interact with others of his kind. In the model there are two types of interactions: in-group and out-group.
People in the model form groups based on their common interests (interest refers to the average long-term needness of a specific activity). An even greater convergence of interests (including those which initially didn’t coincide) is the result of membership in such groups. This makes mutual influence of group members another factor of activity choosing.
Out-group interactions don’t result in change of interests. These include contacts during jointly performed “trade” activity, for example.
Besides artificial people there are firms (production companies) in the model. A firm (implemented by a separate Java class) follows independent logic but like with activities it pointless without humans and requires people – employees – for production to start. In turn, the employees is a kind of group. Firms produce goods. For now there is one provisory good in the model (it’s one of compulsory resources of an artificial human and because of that it can be related to food) which constitutes firms’ output. For simplicity firms exist in a low competition environment, therefore they form their sale price by setting a desired profit relative to the costs.
Everything concerning firm behavior is not as interesting at this stage of HBM building as individual behavior, that’s why companies are presented in a simplistic and very traditional way for economic theory (focused on profit, Cobb-Douglas production function, etc.). Most likely one will only need to open a textbook in order to advance firm behavior - new ideas won’t be required.
There is still environment in the model. It is represented very rudimentary as an inexhaustible supply at a fixed price (as in the case of small economies) of the only good existing in the model. In other words, the environment is reduced to the external market which is significantly larger than the model internal market (hence the inability to influence the price of goods).
Thus, HBM is an environment in which every arbitrary artificial human sequentially chooses and performs one activity after another for a certain period of time specified by researcher (Figure 5). All choices of every artificial human, as well as their effects and consequences (and more generally, any value of any model parameter required for analysis) are documented in separate files. Thus, a kind of chronicles of an artificial community comes into our possession.

english5
Figure 5. General logic of the HBM dynamics.

  • Комментарии не найдены
Чтобы оставить комментарий, необходимо зарегистрироваться

© 2015 humanmodel.ru