You describe a tension that a lot of scientists (myself included) wonder about, but few have addressed head on. The implications here are huge: in encouraging moral behavior, should we encourage people to think about the moral acts in which they've already engaged, or nudge them away from this type of humble-bragging? I think you (and Gneezy and Conway and Cornelissen) are absolutely on the right track in answering that (as with so many other psychological phenomena) it depends. Critically, we need to identify factors that tip people between consistency—doubling down on an initial moral act and licensing—feeling that you’ve done your good deed for the day and giving up. I also think you’re right that one critical factor here is the amount that a prosocial act "gets into" your identity, becoming something part of who you are. This reminds me of some great work done by Christopher Bryan and some of my colleagues, in which asking people to consider past voting as part of their identity ("I am a voter") instead of simply an action they performed ("I tend to vote") led to more voting behavior down the road. In other words, it's easy to fulfill one's quota on actions, but once those actions become a long-standing, abstract, and important feature of your self, those quotas disappear.
Another tension that's been on my mind lately is that between negative and positive motivators of prosocial behavior. Whenever I see ads for Save the Children, St. Jude's Hospital, and so forth, I'm always struck by the appeal to one or both of a pair of approaches: a harrowing depiction of lives poorer or sicker than our own as pure, uncut suffering, or an uplifting sense that you—yes, you—can make that suffering go away. Both messages are problematic: they present simplified accounts of others' lives, which can cause people to feel LESS similar to the targets of their helping, and even lead them to subtly dehumanize others. We both have gripes with this, for instance with the KONY campaign, but I'll leave you to open that can if you like. Another issue with using negative emotion to promote good deeds is that it can encourage people to perform a moral act as a form of avoidance: an attempt to tamp down negative feelings. Daryl Cameron and Keith Payne recently demonstrated that negative emotion can exacerbate the identifiable victim effect: under which people help one individual in need but not many. This is because people anticipate feeling somewhat bad when faced with one person’s suffering, but feeling extremely bad when faced with the suffering of many. In order to escape this pain, they, in essence, “regulate away” their empathy: a topic to which I hope we can return later.
For now, my question is: do positive or negative messages serve as stronger catalysts to donation? And does one type of message lead to more immediate donation—followed by subsequent licensing—whereas the other leads to more sustained adoption of a "donating identity?" I have some thoughts, but let’s hear yours first.