|Your take on value related terms, such as "should," can be summarized as that they merely refer to the relationships between ends and means. You "should" eat, IF you want to live. Specify a goal condition, and the values fall into place in the analysis of a path to that goal.|
This is true so far as it goes. However, it is also true that many goal conditions are simply impossible, due to the nature of reality. I might wish that I had the power to change lead into gold by merely chanting over it, but there is no path to that goal.
In other cases, goal conditions are mutually exclusive, as in eating ones' cake, etc.
Thus, right from the outset, one is challenged by the nature of reality itself to analyze and evaluate ones goals themselves. In the process of eliminating meaningless, impossible and mutually exclusive goals, it should become clear to a rational being that goals that are inconsistent with life itself cannot be successfully pursued.
However, this is taking the issue out of a real context and treating it like a floating abstraction. Living, conscious, rational entities do not suddenly "poof!" into existence seeking answers to existential questions. Such entities as human beings have a history, of organizing primitive sensations as a fetus, reorganizing the sensations into percepts on the basis of testing and then further organizing the percepts into concepts, etc. That history is important in understanding the is/ought issue.
The driving force in all this process of building a consciousness, involving millions, probably trillions of choices and paths, was the successful integration of data into a model of reality that could be continuously tested and verified via the feedback of action and observation. It was an evolutionary process that selected for the integrations that provided the most complete and coherent and efficient organization of data from the real world.
All the intellectual virtues were created, called into play and reinforced during this process, which was not arbitrary, but rather guided implicitly by the natural needs of the living entity.
Thus, at the point that a person starts asking the question "why," as in the example you gave, they have implicitly already given the answer - "to live." Otherwise, why ask "why" to begin with. That entity, that child, is attempting to extent the process of integration that started in the womb to the next natural level.
In a parallel to the character in Roger Rabbit, who explained her character in the famous phrase, "I'm just drawn that way," we just grew up that way. It wasn't very much a matter of choice. The mistakes - the deviations from the central inherent goal of integrating the data of the universe in the service of our well being - were erased from our sensory system, forgotten by our perceptual system, and identified as error, and thereby removed from normal consideration as potential premises, by our conceptual understanding.
Thus, it is legitimate to tell someone they are wrong, not just 'IF' they want to live, but in the sense that they have an error condition in their processing if they do not value life itself, as such a conclusion is inconsistent with what they as an evolved consciousness have been attempting to do with every breath and every synapse firing since they were conceived.