In World Hypotheses: A Study in Evidence (1942), Chapter V: Root Metaphors, the second section describes Root Metaphor Theory as an elaboration of analogy, not conceiveable as permulations of logical postulates.
digraph PepperCh05LogicalPostualtes { // Global setup rankdir = BT // Node list gwt [shape=oval color="blue" label="Generating World Theories"] {rank=same cs ana plp dt} ana [shape=oval color="blue" label="Analogizing"] plp [shape=oval color="blue" label="Permuting\nLogical\nPostulates"] cs [shape=box color="green" label="Common\nSense"] dt [shape=box color="green" label="Data"] {rank=same rmt dtt} rmt [shape=box color="green" label="Root\nMetaphor\nTheory"] dtt [shape=box color="green" style="dotted" label="Deductive\nSystem\nwith\nTheorems"] // Adding edges ana -> gwt [arrowhead="obox" label="is an\ninstance\nof"] plp -> gwt [arrowhead="obox" label="is an\ninstance\nof"] cs -> ana [arrowhead="dot" label="handles"] dt -> plp [arrowhead="dot" label="handles"] rmt -> cs [arrowhead="obox" label="is an\ninstance\nof"] dtt -> dt [arrowhead="obox" style="invis" label="is an\ninstance\nof"] }
Root metaphor theory can be associated with analogy, and common sense
> ยง2. _Can logical postulates make world theories?_ -- How could world theories be generated? Barring the refined account from world theories themselves, and sticking to the levels of common sense and data, two suggestions emerge. > * One of these is typical of common sense, > * the other of data. > * The first suggestion is analogy; > * the second, permutations of logical postulates. > The root-metaphor theory is an elaboration of the first suggestion. It has the advantage of being practically a common-sense theory and therefore inviting refinement and self-development along the lines of structural corroboration, so that each refined interpretation of the root-metaphor theory by a relatively adequate world theory appears as simply the natural and fully detailed exposition of precisely what a root metaphor is. [p. 87, editorial paragraphing added]
Building a Root Metaphor deductively doesn't work, says Pepper.
> The idea is to conceive a world theory in the form of a deductive system with theorems derived from postulates. Once obtain such a system, and new world theories might then be generated like new geometries by simply adding or dropping or changing a postulate and noting the result in the self-consistency of the system and in the application of the theorems to _all_ the observed facts of the world. [p. 88] > ... it does not seem likely that adequate world theories will be generated in the postulational way. Subsidiary theories of limited scope can be generated in this way; but probably not world theories, for the cogent reason that an uncritical acceptance of data at their face value already commits a man to one structural world theory, and all the permutations of postulates he can make will never get him out of that theory. If he accepts the interpretation of data in terms of some other structural world theory the same condition will hold there. [p. 90]