ILO as a target of learning Needless to say, choosing ILO as the learning target is certainly a simplification of the potentially more complex and multifaceted learning. Nonetheless, statistics from our experimental results (Table 2) show that the intensity of using the limit order is positively correlated with the realized profits. Let (pj,t) be the profit earned by subject (j) at the end of the (tth) round of one market experiment. By summing up the submission and the associated profit over all rounds, one can have where T is the total number of rounds in that market experiment. Table 2 gives the correlation between (ILOj) and (pj). To take into account the possible divergence, both the Kendall and Spearman correlations are provided, while, qualitatively, the result is not much different. They both clearly show the significant positive correlation that exists, which indicates that individuals who use the LOS more intensively also tend to earn a higher profit.3 Furthermore, Figure 1 gives the dynamics of the ILO of the best-performing five subjects and the worst-performing five subjects in Markets A, C, and E.4 From these figures, we can see that the ILO dynamics vary among subjects. Both fluctuating patterns and converging patterns exist. The general feature is that those subjects whose ILO converges to or fluctuates around a higher value of ILO tend to be among the few best performers, and those subjects whose ILO converges to or fluctuates around a lower value tend to be among the few worst performers. This feature, which is shown in all markets, justifies ILO as a learning target.