The dynamics of collective decision making is not yet well understood. Its practical relevance however can be of utmost
importance, as experienced by people who lost their fortunes in turbulent moments of financial markets. In this paper we
show how spontaneous collective ‘‘moods’’ or ‘‘biases’’ emerge dynamically among human participants playing a trading
game in a simple model of the stock market. Applying theory and computer simulations to the experimental data generated
by humans, we are able to predict the onset of such moments before they actually happen.