Cognitive Modeling as a Forecasting Tool

The purpose of this study is to provide research and history of cognitive research. During the study, it is planned to consider the works of leading Russian (Soviet) and foreign scientists in the field of theory, methodology, and practice of cognitive modeling. There will also be research into the capabilities of software products developed for cognitive modeling and the possibility of their application. Analysis of software products has not previously been presented in scientific publications. This article presents the results of a study that allows us to evaluate the possibilities of using cognitive modeling and it is potential.

Cognitive modeling is a scientific discipline that studies how the human brain processes information and creates internal representations of the world around us. It is based on the assumption that the work of the brain can be represented in the form of a computer model that can simulate its functioning.

Cognitive modeling itself is a branch of artificial intelligence, the purpose of which is to simulate human thinking and decision-making processes. In economics, cognitive modeling is used to understand and predict the behavior of complex economic systems. At the domestic level, the cognitive approach has found application in the management of various socioeconomic systems, such as healthcare, education, and business. For example, in healthcare, cognitive interventions such as cognitive behavioral therapy have been used to treat mental disorders. In education, the cognitive approach has been used to develop learning strategies that are better aligned with how people process and store information. In business, the cognitive approach is used to improve decision-making processes and optimize the effectiveness of the organization.

At the international level, the cognitive approach is applied in a similar way, with some countries paying more attention to its use in the management of socioeconomic systems. For example, in the Netherlands, the cognitive approach was used in the management of transport systems, where it was used to improve traffic flow and reduce accidents. In Japan, the cognitive approach has been applied in the management of disaster response systems, where it has been used to improve decision-making and communication between emergency services.

A cognitive map is a knowledge structure that is mathematically described by a sign-oriented graph G. Cognitive structuring, or cognitive mapping, refers to the process of identifying the target and undesirable states of the control object, as well as the most important management factors and environmental factors that can affect the transition of the object to these states. In addition, cognitive structuring involves establishing qualitative and quantitative relationships between factors, taking into account their interaction with each other.

Cognitive modeling of a large system can provide insight into the factors influencing the behavior of complex systems and can help identify opportunities for intervention or improvement. For example, understanding cognitive biases in decision-making processes can help decision-makers develop more effective development strategies.

One of the approaches to cognitive modeling of the history of a large system is to create computational models that simulate the behavior of the system and its various components. These models may include cognitive principles such as decision-making, learning, and perception to better understand the factors that influenced the development of the system and how it may evolve in the future.

Another approach is to use historical and statistical data and their analysis to identify patterns and trends in the system. This may include studying the behavior of individual elements in the system, as well as interactions and feedback loops between various components.

In decision support systems (DSS), cognitive modeling is used to analyze and optimize decision-making processes, as well as to develop more effective decision-making strategies.

Cognitive modeling allows to:

Assess the situation;

The position of describing the system in terms of entropy turns out to be the most suitable for the cognitive approach to modeling complex socioeconomic systems. Thus, it is possible to understand the essence of the system S only by studying its interaction with the environment. The view of the system as a unified whole is developed by introducing the concept of “connection”. Further, the whole complex of system connections and their characteristics leads to the concepts of system structure St and complexity. In structural studies, a set-theoretic description of the system in terms of the theory of relations, including the language of cognitive maps, is particularly effective. Graphs are a universal means of describing the structures of systems. Depending on the elements, structural diagrams and signal graphs are used.

The structural scheme is an oriented graph consisting of a set of vertices W = {w_i} and a set of arcs E = {x_i}.

The arcs of the structured graph correspond to the variables X_i, i = 1…, k, and at the vertices, the variables are summarized and the sums are transformed according to the transfer functions: Xis=wisΣj xjs {X_i}\left( {\rm{s}} \right) = {w_i}\left( {\rm{s}} \right)\Sigma j\;{x_j}\left( s \right)

In the language of binary relations, a stultified graph is defined as a pair of vertex sets w_i ∈ W and arc sets x_i ∈ X Gc=W,X {G_c} = \left\langle {W,X} \right\rangle

The arc (w_i, w_j) of a graph specifies causal relations; this arc corresponds to the variable x_i.

Signal graphs are a convenient form of representation of control system models.

A model in the form of a signal graph is defined as a binary relation W on the set of variables X.

Mathematically, a cognitive map is a sign-oriented graph: (1) G=V,E G = \left\langle {V,E} \right\rangle

где V — many factors of the situation (vertices, objects, and concepts), vertices V_i ∈ V,

The tools of cognitive modeling can var y depending on the goals and objectives of the DSS. Some of them include:

Cognitive maps (Mind maps) are graphical models that display connections between ideas and concepts that can be used to study decision-making. Maps can be used to visualize conceptual relationships and facilitate the decision-making process.

Causal relationship diagrams are a methodology that is used to determine causal relationships between events. Cause-and-effect diagrams help identify factors that influence decision-making and allow you to develop strategies more effectively.

Event and process modeling is a method that is used to create models that display the processes that occur in the system. These models can be used to analyze system performance and to determine the best decision-making strategies.

Cognitive agents are tools that are used to model human thought processes. These tools make it possible to analyze decision-making and predict the results of decisions made.

For the organization of research work and the development of a consistent action program, the “meta-laboratory of the description of the object (system)” (Gorelova & Pankratova, 2015) is used, which is also a “meta-model of research” (2). This model includes an “observer”—a meta-laboratory that takes into account the influence of the researcher on the system and its research. (2) M=MOY,U,P,MEX,MO,MDQ,MMO,MME,MU,MH,A M = \left\{ {{M_O}\left( {Y,U,P} \right),{M_E}\left( X \right),{M_O},{M_D}\left( Q \right),{M_{MO}},{M_{ME}},{M_U},{M_H},A} \right\}

где M_O(Y, U, P) – identifying the model of the system (object model) in which Y are endogenous variables,

The MO, ME, and MORE models are cognitive models. The MD system behavior model is a model of an impulse process that describes the evolution of the situation on the model when disturbing influences are introduced. The MH observer model is part of the cognitive modeling process that reflects the process of cognition by the subject of the studied object. Let’s consider the signal graph in the example of the graph by Magoro Maruyama. One of the historical examples of a cognitive map of a complex system is the model developed by Magoro Maruyama and designated by him as a “Landmark Digraph for Analyzing Solid Waste Disposal Problems” (Gorelova & Kalinichenko, 2018). To create this model, a cognitive modeling software system was used (a program for cognitive modeling and analysis of socioeconomic systems at the regional level).

When:

V₁ – Number of inhabitants in the city

The Cognitive map G1, “The Sign Orgraph for Analyzing Solid Waste Management Problems,” was developed by Magoro Maruyama using Formula 1.

For further modeling of the development of the situation, pulse modeling can be used (Gorelova et al., 2006). (3) xin+1=xvin+∑j=1k−1fijPjn+Qin {x_i}\left( {n + 1} \right) = {x_{vi}}\left( n \right) + \sum\limits_{j = 1}^{k - 1} {{f_{ij}}{P_j}\left( n \right) + {Q_i}\left( n \right)} where

x_i(n) – the magnitude of the pulse at vertex V_i at the previous moment—the simulation cycle – n,

In pulse modeling, the situation is determined by a set of values Q and X, which changes in each simulation cycle. A set of implementations of impulse processes is a “development scenario” that indicates possible trends in the development of situations. Formula 3 was used to show with arrows the effect of one vertex on another vertex in Figure 1, impulse impact = +1, −1 since it is impossible to quantify the impact of one node on another. To build this model, statistical data from sources was analyzed (Official website of the Territorial body of the Federal State Statistics Service for the Rostov region//Rostov region in numbers (2021): Brief statistical collection/Rostovstat. – Rostov-n/D, 2022. [Electronic resource]. – URL: https://61.rosstat.gov.ru/folder/30195.; Official website of the Federal State Statistics Service. [Electronic resource]. – URL: https://rosstat.gov.ru/folder/11109.; Official website of the Industrial Development Fund. [Electronic resource]. – URL: https://frprf.ru/ (access date 03/10/2023). These data allow us to determine the quantitative values of the vertices V₁, V₂, V₃, V₄, V₅, V₆, and V₇.

Figure 1:

Figure 2:

The analysis of statistical data has shown that such indicators V₂, V₃, and V₅ are absent in the analyzed collections. Therefore, the use of cognitive modeling allows for building a forecast without precise quantitative data (Official website of the Territorial body of the Federal State Statistics Service for the Rostov region//Rostov region in numbers (2021): Brief statistical collection/Rostovstat. – Rostov-n/D, 2022. [Electronic resource]. – URL: https://61.rosstat.gov.ru/folder/30195.; Official website of the Federal State Statistics Service. [Electronic resource]. – URL: https://rosstat.gov.ru/folder/11109.; Official website of the Industrial Development Fund. [Electronic resource]. – URL: https://frprf.ru/ (access date 03/10/2023). At the moment, there are several software tools for cognitive modeling of complex economic systems. They are based on various modeling methods, such as agent modeling, system dynamics modeling, and network modeling, to describe economic systems.

The analysis of programs for cognitive modeling of complex economic systems allowed us to identify the following (Makarenya et al., 2023):

NetLogo is an agent-based modeling software that is widely used in economics. It allows users to create and model complex economic systems such as market dynamics, social networks, and game theory models. NetLogo has open-source code and a large community of users who contribute to its development.

Russia also has a number of software tools for the cognitive modeling of complex economic systems. Among such tools, software complexes can be distinguished, which, unlike editors like Mind maps, have advanced functionality for analyzing the constructed model and pulse modeling.

Among them, one can single out the Canvas DSS (Kulinich, 2002), which allows for conceptual analysis and modeling of complex and insufficiently defined political, economic, or social situations. This system can be used to develop management strategies and mechanisms for their implementation, as well as to create policy documents for the strategic development of a country, region, enterprise, firm, etc. In addition, it can be used for continuous monitoring of the state of the situation, generating and testing hypotheses about the mechanisms of development and management of the situation (Makarenya et al., 2023).

This also includes the program being developed for cognitive modeling of complex CMCS systems (Gorelova & Kalinichenko, 2018) as well as full-fledged complexes to support strategic decision-making in the fields of economics, politics, sociology, military–political conflicts, medicine, etc., such as CoSMoS (Silov, 1995), IGLA (Korostelev et al., 2008), Compass (Kulinich & Maksimov, 1998), and others.

These tools allow you to quickly and comprehensively characterize complex and uncertain situations, as well as offer qualitative solutions to problems in these situations, taking into account environmental factors and a systematic approach.

But, none of the above software products has an information and analytical system for accounting for industrial production, which could be used to conduct PEST analysis and build input cognitive maps.

It can be seen that it is quite difficult to understand the factors, the impact on the factors, and the main directions of industrial development.

The authors propose the development of such an information and analytical product that would collect data from industrial enterprises, and form databases of industrial products in natural units of measurement that can be used for cognitive modeling and machine learning in order to build predictive development scenarios. The interface of this software product may look like this (Figure 3).

Figure 3:

Each of the modeling systems has its own strengths and weaknesses, and the choice of modeling tool depends on the subject area, modeling requirements, and user preferences.

Currently, the cognitive approach is widely used in intelligent management DSS. It allows you to combine formalized scientific knowledge with the experience of experts and the creative potential of decision-makers. The results of cognitive modeling of complex systems belong to the field of artificial intelligence, and are also used to create intelligent DSS (ISPR).

The disadvantages of quantitative forecasting methods are that there must be a database of data for at least 5 years on the basis of which forecasts are based (for example, the exponential smoothing method). Such databases are not always available. This makes it difficult to use quantitative forecasting methods. For cognitive modeling, it is necessary to carry out pest analysis, which does not require a data set for 5 years. Therefore, the cognitive modeling method is easier to use.

The disadvantage of the cognitive modeling method is that the resulting development forecasts provide a qualitative assessment, and not quantitative indicators of the development of the system under study. For example, it is impossible to predict how many pieces of a product should be produced using the cognitive modeling method. This is a disadvantage of this method. However, the cognitive modeling method makes it possible to determine the direction of development of a weakly structured system under the influence of regulatory factors—the tops of causes, that is, it is possible to predict what needs to be done in the future.

Therefore, it is possible to use cognitive modeling when modeling socioeconomic processes. This method can be used by both scientists and government authorities. In a situation where the political and economic situation changes dramatically, cognitive modeling makes it possible to build forecasts for the development of socioeconomic, regional, and sectoral systems

In general, the use of a cognitive approach in the management of socioeconomic systems has shown promising results in improving productivity and results. By better understanding the cognitive processes underlying behavior in the system, managers can identify areas for improvement and develop more effective strategies to achieve their goals. However, further research is needed to fully understand the potential benefits and limitations of this approach, as well as identify best practices for its implementation. The study showed that cognitive modeling is a tool for predicting loosely structured systems when there is no quantitative data from statistical reports.

Further directions of research are to compare the obtained cognitive models with quantitative predictions.