THEORIES AND MODELS 247 



trying to convey to the novice some idea of potential energy, and 

 activation energy, the physicist may sketch analogy with familiar 

 experience of balls rolling downhill. But, exactly as in the use of the 

 heat engine in the exposition of thermodynamics, this is no trivial 

 pedagogic device. In his research the physicist may investigate po- 

 tential energy "wells" and "barriers," and he speaks of a "tunnel 

 effect." Thus, as Hebb quite properly observes: 



The worker in the laboratory does not merely report and expound 

 by the aid of analogy; that is how he thinks, also. 



Is it not preposterous to suggest that the physicist has no physical 

 images in mind when he speaks, as invariably he does, of spin, inter- 

 ference, resonance, scattering, shell-structure, liquid-drop model, 

 strong and weak interactions, cross-sections, waves, particles, and so 

 forth? To be sure, the quantum physicist can no longer suppose that 

 in the microcosm it is any simple matter of wave or particle. But^ 

 taking care only not to mix the metaphors, he productively conceives 

 microcosmic entities now in one way, now in the other, as the occa- 

 sion demands. There is then overwhelming evidence for the conclu- 

 sion Schrodinger puts shortly: 



Most physicists, whether or no they confess to it, are using some kind 

 of model-picture ... 



"Confession" may be withheld for three reasons. (1) The user of 

 an implicit hierarchic model may easily fail to recognize it as such. 



(2) He may not ivant to recognize his use of a model: positivist 

 criticism holds models unsophisticated— and who would not be so- 

 phisticated? (3) Even if one would confess, often he cannot. Modern 

 editorial policy of scientific journals suppresses accounts of the 

 course of scientific discovery— in favor of highly compressed, strictly 

 analytic accounts of the nature of the discovery. The published arti- 

 cle is then often a complete inversion of the actual process of dis- 

 covery, likely to omit all mention of heuristically powerful models 

 and analogies that are retrospectively superfluous. Points (2) and 



(3) together are self -reinforcing: the fewer the references to models 

 in published papers, the greater will be the feeling that their use is 

 somehow disreputable. But even though explicit confession may thus 

 be inhibited, the quantum physicist's vocabulary makes confession 

 for him— and rather strongly suggests that the heuristic power of 



