The terminology used here is potentially confusing. In engineering terms, "positive feedback" should lead to something becoming more pronounced - the microphone too near the speaker is a classic example. "Negative feedback" usually results in a dampening or reduction if something is too large, and an increase if it is too small - resulting in a self-regulating behavior. The authors use these terms in a manner that suggests just the opposite.

Similarly, in the psychological or learning theory literature, if a response is "reinforced" its probability increases in the future. The authors suggest that the reinforced behavior will become less likely. In fact, an instrumental learning analysis might suggest that the behavior that yields more mating should increase (i.e., be "reinforced") yet the authors find the opposite.

All in all an interesting article and finding. I just point out that the terms used might be confusing to some readers coming from a learning theory background.