Kohonen neural networks are able to reproduce important features of human
cognition, such as the ability to form and modify mental categories
developing classification criteria on their own. Thus, they are
applicable in domains where decision-makers face novel situations,
e.g. in investment decision-making when emerging technologies are
involved.
I will describe a productive system by means of a Kohnen neural network where each neuron represents a decision-maker. This model extends and unifies several previous models of investment decision-making. I present results of computer simulations and compare them to analytical approximations. Possible applications on real data are indicated. Further, I will point out that the topic offers numerous possibilities for collaborations between social and natural scientists. |