Visualizations used in the practice of neuroscience, as one example of
a scientific practice, can be sorted according to whether they repres
ent (A) actual things, (B) theoretical models, or (C) some integration
of these two. In this paper I hypothesize that an assessment of a cha
in of visual representations from (A) through (C) to !B (and back agai
n) is used, as part of the practice of scientific judgment, to assess
the adequacy of the ''working fit'' between the theoretical model and
the actual thing or process that the model is intended to explain.