
An Economic Perspective on the Ant Colony
Ever since I explored the surrealism around Salvador Dali’s creative universe, I have been fascinated in particular by one of his prevailing iconographic motifs symbolizing death, putrefaction, moral degradation as well as his dark spectrum of sexual desires: the swarming ants. Dali’s artistry largely consisted of exposing his subconscious imagery through these seemingly unobtrusive symbols. Despite this, however, what do we actually know about ants?
Ant societies
Living in extremely complex structured hierarchical societies, renowned for their high level of organization, ants are the conventional emblem of eusocial animals. Ants have maximized the functionality of their colonies, making them extremely efficient – whether we are talking about ensuring favorable conditions for larval development or about the construction of ventilation systems as a result of high CO2 concentrations inside the nest chambers [7]. Consequently, being integrated in this what appears to be a perfectly organized community gives ants plenty of advantages ranging from foraging and group hunting to nest construction and protection against predators.
But what is at the base of this organization? One of the most significant factors responsible for the tremendous success of insect societies is the principle of division of labor. There is, for example, a division of labor on reproductive criteria that separates fertile individuals that reproduce, such as queen ants, from sterile and semi-sterile ones, such as worker ants [6]. This concept of “division of labor” has been borrowed by biologists from social philosophy, its role being widely promoted by Ludwig von Mises since the discovery of its valuable contribution within the “general law of association”. And as long as “society is division of labor and combination of labor” [10, p.143], there are indisputable similarities between our approach as social animals and social insects’ approach to society.
However, in one of his most famous works, Human Action: A Treatise on Economics, Mises has described human society as “the outcome of conscious and purposeful behavior” [10, p.143], social cooperation being the key concept that illustrates this essential complex of mutual relations, imperative to achieve specific purposes, called society. Mises is also drawing a dividing line between human beings and animals by invoking the argument of species consciousness, according to which only human beings are “capable of recognizing the mutual benefits of cooperation, while the animals lack this faculty” [10, p.144].
Interestingly, a debate about the social behavior of ants has gained fresh prominence among economists and entomologists. In several experiments undertaken, among others, by researchers such as Deneubourg and Pasteels [4] a context was created that depicts a seemingly symmetric situation, which consists of two identical food sources situated at the same distance from an ants’ nest. In addition, both sources were continually being replenished with food such that there was no distinction between them. Nevertheless, when ants were invited to explore the experimental area, they expressed their preference for one of the two sources, although, occasionally, they redirected their attention to the one that they had previously ignored. Accordingly, the results of the studies are somewhat counter-intuitive as they have indicated that, collectively, ants act in an asymmetric manner, the aggregate behavior of the group as a whole not being equivalent to the isolated behavior of identical individuals [8]. What is intriguing, though, is that exactly the same unbalanced exploitation has been noticed in humans’ restaurant selection process [8].
In his case study of restaurant pricing and other social influences on price, Becker [2] identifies a vast majority of people deliberately choosing one of the two available restaurants that are located on either side of a street despite their similarity in terms of food quality, prices charged and types of services. That being the case, the selected restaurant always has long queues, while the other has a lot of empty seats almost all the time. We therefore recognize this sort of herding behavior from the previous experiments performed on ants, the behavior of the group as a whole being impossible to explain just by analyzing an isolated individual in a state of equilibrium and thus subordinated to some rational expectations, but in terms of interactions between individual agents.
Furthermore, Becker argues that the same herding phenomenon not only underlies the popularity of books and the pricing of famous sporting events, but is also considered to be the origin of some “endogenous fluctuations in the price level in asset markets” as long as “it does not rely on exogenous shocks to the system” [8, p.138].
The creation of an ingenious algorithm
The foraging behavior of ants has served as the original model for the construction of the ant colony optimization (ACO) metaheuristic algorithm which has quickly proven to be a grand success, receiving recognition from the scientific community. This prodigious bio-inspired algorithm has been applied to several types of fundamental combinatorial optimization problems including routing problems (that we may encounter, for instance, in the supply chain distribution), assignment problems (in which case a certain number of resources are supposed to be assigned to an equal number of objects or activities, thus ensuring profit maximization as a result of minimizing the total cost), scheduling problems (that require us to find the answer to the question of how can we effectively allocate the limited resources we have to meet certain needs over time) and subset problems (which involve the selection of a subset consisting of different elements belonging to a given set in order to find the solution to a problem) [5]. “Strong robustness, good global optimization ability and inherent parallelism” along with its capability to easily combine with other heuristic algorithms in order to enhance their functionality are among the immediate advantages of this technique [3, p.2]. On the other hand, “the randomness of probabilistic transfer” and “the inappropriateness of pheromone intensity update” are the principal causes leading not only to “poor convergence”, but also to the fall of the traditional ACO “into the local optimum”. [3, p.2].
However, let us further discuss the main contributions of this metaheuristic to various branches of economic investigation.
First of all, as far as we know, the efficiency of a logistics system is primarily influenced by the locations of the depot and also by the routes of the vehicles. Consequently, in order to minimize the overall cost of the system, decisions regarding the facility location and vehicle routing are vital to the company’s survival in the fiercely competitive business environment. In this regard, a multiple ant colony optimization algorithm has been proposed with the aim of solving the location routing problem [11]. The hierarchical structure of ant colonies is basically the key concept behind the construction of such a brilliant algorithm whose purpose is to optimize a wide range of “subproblems” such as “location selection, customer assignment, and vehicle routing problem” [11, p.34]. In this situation, the ant colonies communicate and collaborate with each other through the exchange of information in the form of “pheromone updating between the location selection and customer assignment” [11, p.34].
Second of all, recent trends in global capital markets have led to a proliferation of studies on stock market prediction. Thus, an increasing number of deep learning algorithms have been created so as to improve as much as possible the forecasting process. Considering the intricate nature of the stock market data, neural networks are associated with a greater risk of performing “inconsistently and unpredictably”, while also facing the difficulty of “learning the data patterns” [1, p.52]. However, what is interesting is that Ahmed et al. [1] carried out a number of investigations into the efficiency of different major technical analysis strategies to obtain the most optimal forecasts of the following day’s closing stock price. The most striking result to emerge from this study is that of all the approaches, the ant colony optimization metaheuristic generated the most accurate predictions, outperforming the other three indicators (the Stochastic Algorithm, the Moving Average Algorithm and the Price Momentum Oscillator) in terms of “accuracy, sensitivity and specificity” [1, p.57]. In addition, the ACO-based feature selection algorithm also plays the cardinal role of an investment consultant who is responsible for providing “market signal” regarding the rise or fall of the stock prices in the following trading day [1, p.57].
Moreover, I think we all remember the disastrous consequences of the 2008 global financial crisis. Such devastating worldwide economic crises have heightened the need for a multitude of data classification models to be developed to predict when an organization will face a financial crisis depending on the company’s historical data. In this respect, in 2020, Uthayakumar et al. [12] attempted to draw fine distinctions between the ACO-FCP (financial crisis prediction) algorithm and three other popular computational methods of feature selection such as the Particle Swarm Optimization algorithm, the Grey Wolf Optimization technique and the Genetic algorithm. In summary, the results of the study indicated that not only “the ACO-FCP ensemble model is superior and more robust than its counterparts”, but it is also “highly competitive than traditional and other artificial intelligence techniques” [12, p.538]. Furthermore, the ant colony algorithm is also used in conjunction with the multimedia assisted back propagation neural network model in macroeconomic forecasting [9].
But how exactly does this algorithm work? You might be wondering how some ordinary ants could have contributed to the development of such an advanced technique that is now responsible for revolutionizing almost all the scientific fields from artificial intelligence and economics to bioinformatics, computational chemistry and urban planning. Well, in my fourthcoming paper in the Œconomica journal, we will be able to explore the ant colony optimization paradigm from both an economic and a biological perspective. In the first place, we will look with a magnifying glass at the genetic background that makes the chemical communication possible, simultaneously analyzing the process of signal transduction and the main families of complex macromolecules occupying a leading role in olfaction. Pheromones are the chemical messengers on which we will focus in our attempt to understand the social structure of ant colonies while the extremely sophisticated mechanism of stigmergy is going to be observed through the emblematic example of termite nest construction and whose applicability will later be extended not merely to urban economics, but also to a myriad of cognitive sciences and ultimately to the human activity (stigmergy 3.0). Afterwards, the double bridge experiments will help us to contextualize the pheromone laying strategy and the colony dynamics. At a later date, I will provide an illustrative example of the use of an ant colony optimization algorithm on real data (involving the distances between different cities in Romania) with the aim of solving a classical routing problem.
Photo source: flickr.com (illustration by Otokar Stafl, from the Czech children's book, Naměsíc a ještě dál, 1931).
References
- M.K. Ahmed, G.M. Wajiga, N.V. Blamah and B. Modi, Stock market forecasting using ant colony optimization based algorithm. American Journal of Mathematical and Computer Modelling, 4(3), 2019, p.52-57.
- G.S. Becker, A note on restaurant pricing and other examples of social influences on price, Journal Of Political Economy, 99(5), 1991, pp.1109-1116.
- X. Dai, S. Long, Z. Zhang and D. Gong, Mobile robot path planning based on ant colony algorithm with A* heuristic method, Frontiers in Neurorobotics, 13, 2019, 15.
- J.L. Deneubourg, S. Goss, J.M. Pasteels, D. Fresneau and J.P. Lachaud, Self-organization mechanisms in ant societies. II: Learning in foraging and division of labor, Experientia. Supplementum, (54), 1987, p.177-196.
- M. Dorigo, M. Birattari and T. Stutzle, Ant colony optimization, IEEE Computational Intelligence Magazine, 1(4), 2006, p.28-39.
- Furnica, Proiectul trans-european Elena finantat de Comisia Europeana, http://elena-project.eu/phocadownload/Modules/romanian/Furnica%20-%20RO.pdf
- F. Halboth, F. Roces, The construction of ventilation turrets in Atta vollenweideri leaf-cutting ants: Carbon dioxide levels in the nest tunnels, but not airflow or air humidity, influence turret structure, PLoS One, 12(11), 2017.
- A. Kirman, Ants, rationality, and recruitment, The Quarterly Journal of Economics, 108(1), 1993, p.137-156.
- Y. Kuang, R. Singh, S. Singh and S. P. Singh, A novel macroeconomic forecasting model based on revised multimedia assisted BP neural network model and ant Colony algorithm, Multimedia Tools and Applications, 76(18), 2017, p.18749-18770.
- L.v. Mises, Human action: A treatise on economics, The Scholar’s Edition. Auburn, Ala.: Ludwig von Mises Institute, 1998.
- C. J. Ting and C.H. Chen, A multiple ant colony optimization algorithm for the capacitated location routing problem, International Journal of Production Economics, 141(1), 2013, p.34-44.
- J. Uthayakumar, N. Metawa, K. Shankar and S.K. Lakshmanaprabu, Financial crisis prediction model using ant colony optimization, International Journal of Information Management, 50, 2020, p.538-556.