Tinbergen defined structure as the momentary condition of an organism. Inference, interpretation, and theory were reserved for answering the “why” questions. Mason, however, observed that the study of behavior is incorrigibly mentalistic. Description seems barren compared to the richness supplied by creative intelligence. Exciting intervening variables such as learning, intention, and selective pressures can only be examined using inference, logic, and powerful deductive theories. For more than 100 years people have argued the two extremes: should science be restricted to the directly observable, or expanded to include what professional judgment could provide? I have devoted my life to the study of aggression and dominance, yet I have directly observed neither. I point to some behavior as an example of aggression, but I cannot provide an operational definition of what I mean. I will never know if another intends harm. I will never know if another is aware of the functions that behavior serves, and the outcomes that I confuse with past selective pressures and motivation. In a world where our subjects face extinction, perhaps it is wise to use computer simulations, to ask colleagues to rate their impressions, and to use logic to deduce behavior. The new “data” may be very useful. Still, I will miss actually watching what animals really do, even if I always describe it in terms of what I infer. Am. J. Primatol. 60:77–84, 2003. © 2003 Wiley-Liss, Inc.