InterJournal Complex Systems, 1811
Status: Accepted
Manuscript Number: [1811]
Submission Date: 2006
Emergence of altruism as a result of cognitive capabilities involving trust and intertemporal decision-making
Author(s): Tibor Bosse ,Martijn Schut ,Jan Treur ,David Wendt

Subject(s): CX.35



A basic assumption in the evolutionary explanatory framework is that an organism’s behaviour serves its own interests, to improve its own well-being and production of offspring. For the explanation of the development of altruistic behaviour from an evolutionary perspective, one easily encounters a paradox; see, e.g., [Sober and Wilson, 1998, pp. 17-23]. As by definition altruistic behaviour is behaviour that is against one’s own interests, this paradox can be formulated as: ‘an organism serves its interests by behaviour which is against its interests’. One way to solve this paradox is by widening the scope in the temporal dimension. Then the occurrences of the concept ‘interest’ in the paradox get different time points attached, as follows: altruistic behaviour serves the organism’s interests at a future time point by the organism’s behaviour which is against its interests at the present time point. So, the organism’s present behaviour is seen as an investment to obtain future revenues for the organism itself. As long as the future revenues are at least as important for the organism as the present investment, this may work out fine. It is this approach that is analysed further in this paper; see also [Dennett, 2003, Chapter 7]. In this case a basic assumption is that the environment of an organism has the potentiality or regularity to provide future revenues in return for present investments. This is a nontrivial characteristic of an environment, that often depends on the presence of other organisms in the environment. For example, for species developing agriculture, the activity of sowing in the present, depending on the potential of the seed, leads to growth of food or other products that are in the organism’s interest. Another example, which is taken as a case study in this paper, is that other agents are present in the environment that offer the future returns when they are favoured by an agent, depending on their own intertemporal decision making. Godfrey-Smith [1996, p. 3] relates environmental complexity to the development of cognition, as briefly formulated in his Environmental Complexity Thesis as: ‘The function of cognition is to enable the agent to deal with environmental complexity’. For the case considered here, the agent needs a cognitive system that is able to make a decision where a current investment has to be compared to a future revenue. So, it needs cognitive facilities to predict future revenues based on the present world state and the world’s regularities, and to compare such predicted future revenues to investments to be made in the present. These processes require nontrivial cognitive capabilities, the more so as the world’s regularities usually have a probability aspect in them, that also has to be accounted for in the decision. These cognitive processes are usually called ‘intertemporal decision making’; cf. [Loewenstein and Elster, 1992]. To cope with the world’s risks that in some cases predicted revenues will not come true, in such decision making the future revenues have to be estimated higher than the present investment, for example, by taking into account a certain interest rate. In the literature on intertemporal decision making, the environmental regularity or probability to indeed provide revenues in return usually is not modelled in a detailed manner, and not adapted on the basis of the agent’s experiences. Experiments and models often focus on one subject and its expectations, and do not address how these relate to the real environment. In fact, to estimate the risk of not getting the future revenues in return, the model of intertemporal decision making for the subject should be combined with an environment-dependent model describing how based on its experiences the subject estimates when the environment indeed returns revenues for investments of the subject. In this way the agent can learn and adapt itself to the world’s regularities or potentialities. In this paper, these issues are analyzed and tested by creating an artificial society. As part of the model, for any of the agents also the environment is modelled in a detailed manner as the rest of the society. By formal analysis and simulation it is investigated how agents endowed with a cognitive model for intertemporal decision making can choose for altruistic behaviour by providing services (for free) to other agents in the present and provide revenues in their own interest in the future. The guarantee or probability that revenues indeed are returned by the environment, depends in this case on other agents in the environment receiving the services, that in the future may or may not provide services in return. To estimate the risk of not getting the future revenues in return, the model of intertemporal decision making is combined with an environment model, which in this case is a model for evolution of trust in other agents based on experiences with them (adopted from [Jonker and Treur, 1999]). If the agent experiences over time that another agent does not provide services, the trust in this agent becomes lower; if it does provide services, trust becomes higher. Having such a dynamic environment model enables the agent to become better adapted to the environment. One of the main properties to verify is how agents with different variants of a cognitive system for trust-based intertemporal decision making perform over time. 1. Dennett, D.C. (2003). Freedom Evolves, New York: Viking Penguin. 2. Godfrey-Smith, P. (1996). Complexity and the Function of Mind in Nature. Cambridge University Press. 3. Jonker, C.M., and Treur, J. (1999). Formal Analysis of Models for the Dynamics of Trust based on Experiences. In: F.J. Garijo, M. Boman (eds.), Multi-Agent System Engineering, Proceedings of the 9th European Workshop on Modelling Autonomous Agents in a Multi-Agent World, MAAMAW'99. Lecture Notes in AI, vol. 1647, Springer Verlag, Berlin, pp. 221-232. 4. Loewenstein, G.F., and Elster, J. (1992). Choice over time. Russel Sage Foundation, New York. 5. Sober, E., and Wilson, D.S. (1998). Unto Others: The Evolution and Psychology of Unselfish Behaviour. Harvard University Press, Cambridge, MA.

Retrieve Manuscript
Submit referee report/comment

Public Comments: