"After ten rounds of playing the prisoners' dilemma game, subjects were given the possibility of eliminating one of the two actions by plurality. Each subject voted on whether to keep all actions, eliminate D, or eliminate C. After voting, the subject participated in ten more rounds depending on the decision made by plurality. To study the effect of subjects' understanding of the game on voting decisions I modify how game is presented to the subjects. In half the sessions, the computer screen shows the payoff matrix with the subject action as rows and their partners as columns. Feedback about the outcome is also provided by highlighting the chosen row and column. The other half of the sessions did not see the payoffs displayed as a matrix and feedback did not stress the behavior of the partner by highlighting his/her behavior in the matrix (but this behavior was reported). Figure 1 shows a screen shot of each treatment (payoffs are set in cents). I hypothesize that not showing the game as a matrix may diminish subjects' understanding of the structure of the game and the likely effect of modifying the game by eliminating a strategy. I called these two treatments as "See Matrix" and "Do Not See Matrix" treatments, respectively. The participants were 80 Brown University or RISD undergraduates. Half the subjects participated in each of the treatments. As Figure 2 shows, in the first ten rounds the evolution of cooperation is consistent to what has been found in the literature: a significant cooperation rate that decreases with experience (see Andreoni and Miller 1993, and Dal B