Several avenues remain open for future research. Perhaps most important, experimental research manipulating peer helping activities is needed to confirm and extend the conclusions offered here. As suggested previously, the correlational evidence linking helping to better recovery outcomes is subject to valid critique based on the possibility of third-variable associations, so experimental studies manipulating helping are indispensable to demonstrating a causal role for peer helping (and to ruling out reverse causality). Future research will also be important in identifying what forms of helping work best, and when. The Zemore and Kaskutas (2004) study described earlier uncovered some rather complex associations between length of recovery and helping in specific life domains: longer recovery was positively associated with community-oriented helping, negatively associated with recovery-oriented helping (outside of AA participation), and not associated with level of life helping. These results hint that, over time, recovering alcoholics may naturally transition from helping activities associated primarily with alcohol problems to helping activities concerned with the broader community--and that there may be some value to this transition. What works for a recovering alcoholic early in recovery may not work later in recovery. Further research is needed to explore how length of sobriety, type of helping, and other variables interact to affect the impact of helping activities on recovery outcomes. Another exciting question for further research will be why helping helps the helper. It has been suggested that helping enhances helpers' self-perceptions of independence, competence, usefulness, and/or social status; fosters relationships with others; and diverts helpers from excessive self-absorption. Some have argued that helping others contributes to building and expressing a positive identity. Cooperation may even be inherently rewarding. A recent study found that, when women cooperated (versus competed) in a Prisoner's Dilemma game, blood flow increased in areas of their brains rich in dopamine, the neurotransmitter involved in reward systems and famed in part for its role in addictive behaviors. Moreover, the longer the women engaged in the cooperative strategy, the stronger was the activation. The fact that people often compete rather than cooperate implies the existence of countervailing tendencies. But, the fact that one can (at least sometimes) help oneself by helping others suggests that the traditional dualities pitting competition against cooperation, and self against other, disguise a compelling third choice. The ability to see beyond such dualities may be part of what AA, NA, and other mutual help groups offer.