To motivate the Figurative Simile theory, I begin with three standard objections to the Naïve Simile view.  First, Beardsley (1967) points out that metaphor is characterized by conceptual tension.  For example, "Juliet is the sun" is anomalous on its face because human beings differ categorially from suns and other immense astronomical items; yet metaphorical sentences are not only intelligible but perhaps even exceptionally informative or illuminating, and may express important truths.  Against the Naïve view, Beardsley complains that although that theory does explain the distinctive tension (it can say the tension arises from the move from likeness to actual ascription--"Juliet is the sun"), that explanation is very shallow.  If a metaphor is equivalent to the corresponding simile, then it should not be heard as anomalous or puzzling in the first place; on that view, the tension is the merest surface appearance.  But that seems wrong.  There is no conceptual tension in "Juliet is like the sun," even if one wants to be told more about the respects in which Juliet resembles the sun.  One feels that a metaphor works by containing an inherent tension that is more substantive.  (Davidson and Searle argue that in particular, the metaphor works by having the anomalous literal meaning that it does.)
    Second, Searle complains that a simile taken by itself is almost entirely uninformative.  "Similarity is a vacuous predicate: any two things are similar in some respect or other."   In what way is Juliet supposedly like the sun?  Not by being a gigantic ball of gas, or by consisting in large part of nuclear fusion, or by being 93 million miles from the earth.  As Searle points out, those properties are salient and well-known features of the sun; yet the Naïve Simile theory gives no hint as to why Romeo's metaphor imputes different properties to Juliet, rather than those.  Thus, the theory fails to offer any mechanism by which metaphorical significance might be conveyed.
    Third, even when we have identified the relevant respects of similarity, they often prove to be themselves metaphorical.  Searle gives the example, "Sally is a block of ice."  How, according to the Naïve Simile theorist, is Sally like a block of ice?  Perhaps she is hard and very cold.  But not, of course, literally hard or cold; "hard" and "cold" are themselves used metaphorically here.  So Sally is only like something that is hard and cold.  In what ways?  Perhaps she is unyielding, unemotional and unresponsive.  But, Searle points out, there is no sense in which blocks of ice are unyielding, unemotional and unresponsive but many other inanimate things are not.  Bonfires too are unyielding, unemotional and unresponsive; but neither "Sally is like a bonfire" nor "Sally is a bonfire" is metaphorically compatible with the original sentence.  The Naïve Simile theorist would have to insist that there is a further underlying literal similarity between cold things and unemotional things.  But we are given no evidence for that claim.  Searle conjectures that due to heaven knows what psychological factors, "people [just do] find the notion of coldness associated in their minds with lack of emotion."
    This last objection suggests a simple but radical modification of the Naïve theory, that preserves the central claim that metaphors are compressed similes but avoids the foregoing objections.  It is articulated and defended at length by Robert Fogelin in Figuratively Speaking (1988): that metaphors are equivalent, not to similes taken literally, but to similes themselves taken figuratively.
    Similes are often, perhaps usually, figures of speech.  Sally is only figuratively like a block of ice, for she is only figuratively hard and cold.  Juliet is only figuratively like the sun.  One way to see this (not Fogelin’s own way) is to note that literal similarity is symmetric: if A is literally similar to B, then necessarily B is literally similar to A.  But a block of ice is not literally like Sally, nor the sun literally like Juliet.  And no one would propose such comparisons as similes, as in "The sun? --Oh, the sun is like Juliet."  It is when similes are themselves nonliteral that they best paraphrase metaphors.  This suggests the hypothesis that a metaphor is just an abbreviated figurative simile, deriving from the corresponding simile taken figuratively.
    This Figurative theory easily sidesteps our three objections to the Naïve theory.  First objection:  Since the Figurative theorist does not reduce metaphors to literal and near-trivial assertions of similarity, it cannot be said that the Figurative theory treats the metaphors' conceptual tension as superficial.  There is already conceptual tension in the underlying simile.  Second objection:  Taken figuratively, the simile already carries one or more particular respects of similarity.  So it does not fail to explain how the metaphor brings out those same respects.  Third objection:  Of course the Figurative theorist is not committed to literal similarities between Juliet and the sun, Sally and a block of ice, etc.
    These three advantages come at an obvious price.  In each case, the Figurative theory remedies a deficiency of the Naïve theory by lodging the needed material in the now figuratively interpreted corresponding similes and letting the respectively derived metaphors inherit it.  But the danger here is that of only putting off the problem.  For now the explanatory work is being done by the figurative nature of the underlying similes, and so their figurative interpretations need explaining in turn.  Indeed, two standard main questions about metaphor arise for figurative similes as well:  What is it for such sentences to have figurative meanings, and how are those meanings conveyed to hearers?
    Fogelin exploits the notion of a salient feature of a thing.<1>   In that way he is able to mobilize an actually nonsymmetrical similarity relation (p. 78):  "A is similar to B just in case A has a sufficient number of B's salient features."  A may share a sufficient number of B's salient features without B's sharing a sufficient number of A's salient features, since the particular features of B that A shares need not be salient in A.  E.g., a chipmunk is very like a rat, except for being cute or perceived as such by humans; it has most of the rat's salient features, being a small scavenging rodent of loose morals.  But one would not say that a rat is like a chipmunk, because the cuteness of chipmunks is highly salient to humans and rats are not cute.
    According to Fogelin, the difference between a figurative comparison and a literal one is in the standard of salience, which in a way reverses.  It is (Fogelin says, p. 90) literally true that Winston Churchill looked like a bulldog, but literally false that Churchill was like a bulldog (he having been human rather than canine, two-legged, lacking in fur, given to talking rather than barking, far too big to crawl into burrows, etc.).  Yet it is figuratively true that he was like a bulldog.  In calling him one, Fogelin says, "we compare him to a bulldog (as opposed, say, to a French poodle), while at the same time trimming the feature space in terms of the subject's [Churchill's] salient features" (p. 91).  Unfortunately Fogelin does not go into detail about "trimming the feature space."  I believe the idea is that, having rejected the simile as literal, the hearer nonetheless charitably assumes that the alleged similarity does obtain, and now ignores the salient features of bulldogs that most obviously make the literal comparison false and looks for features that match the salient features of Churchill.  (I am not sure what these would be; toughness, tenacity, earthiness, and looking like a bulldog?)
    On this view, sentences have metaphorical meanings in context that differ from their literal meanings; yet it does not follow that any expression in the sentence has changed its meaning from literal to figurative use, or that the metaphorical meanings are spooky or magical.  Rather, resemblance is always and everywhere relative to a standard of similarity, a "feature space" that determines which properties are to be matched with which.  The standard of similarity is like an indexical in being determined by contextual factors, but also can take more than one value within a single context.  That is why the sentence can be both true (metaphorically) and false (literally) on one and the same occasion of utterance: because two different standards of similarity are in play -- much as "Muffie is small" can be both true and false if Muffie is an undersized moose.  This is a nice advantage of Fogelin's theory.
    However, Fogelin faces at least two more difficulties. First, a statement may continue to be accepted as metaphorically true even when the corresponding simile has proved to be false.  Searle offers the example, "Richard is a gorilla," which the Naïve Simile theory would parse as "Richard is like a gorilla."  Let us suppose that what is meant is that Richard is like a gorilla in being fierce, nasty, prone to violence, and perhaps not very bright.  But primatologists tell us that in fact, gorillas are not nasty or prone to violence; they are shy, rather sensitive and very intelligent animals.  Likewise pigs, which figure in many metaphors imputing messiness, filth, greed, obesity, crassness, or some combination of those:  I myself know of no evidence that pigs are either dirty (in fact I gather that pigs are rather clean animals) or particularly greedy, or that they are fatter relative to their skeletal size than other animals are.<2>
    One might think that Fogelin has easily avoided this new objection, for when a simile is figurative it does not require the actual correctness of the relevant stereotype.  "Sam acts like a gorilla" and "Merle eats like a pig" are correctly expressed and understood despite the fact that the two stereotypes are respectively simian and porcine slanders, because in the similes, "gorilla" and "pig" are themselves being used figuratively rather than literally.  But Fogelin' s picture of "trimming the feature space" presupposes or at least strongly suggests that the features relevantly shared by, e.g., Churchill and a bulldog are possessed literally by each of the two.  And in that sense, on Fogelin's theory a metaphor must still bottom out in a literal sharing of genuine properties.  In examples such as Searle’s (in which the stereotype is just wrong) it is far from obvious what the properties would be.<3>
    The second new objection is that some metaphorical statements are too convoluted to be parsed as similes.  Consider "When the blood burns, how prodigal the soul lends the tongue vows" [Hamlet, I,iii,116].  That sentence is not literally about anyone's blood, and blood cannot literally burn (while still within the body under even faintly normal conditions); "the soul" is probably itself being used metaphorically, and even if not, souls cannot literally lend anything to tongues; but "tongues" is not being used to mean tongues, either, and vows are not the sorts of things that can be lent.  So any Simile theorist,  Naïve or Figurative, faces the daunting task of translating all of those things at once into resemblance talk.  One would have to make free use of some sort of contextual placeholder.  A first pass might be: "When X, which is like a person's blood, does something that resembles burning, how prodigal[ly] Y, which is like a person's soul, does something similar to lending some things that are vowlike to Z, which resembles a person's tongue."  We are not much the wiser.  And refinement is needed, because for "the blood" metaphorically to burn is probably something distinctive to a bloodlike substance, not for it to do something that resembles the literal burning of, say, a piece of wood.  It is no wonder that Simile theorists have in the main stuck to simple subject-predicate examples.
 

1.  Here and elsewhere he draws on Tversky (1977).

2.  If you want greed, try cats.  But no one ever calls someone a cat as a metaphorical way of saying that that person is greedy.
    A further example is "bastard."  I know of no evidence that a male person whose parents were not married when he was born is any more likely to be callous or unscrupulous than is anyone else.

3.  Fogelin addresses this objection (pp. 44-45), but I think weakly.  He complains that "gorilla" is not a metaphor but a dead metaphor; if so, that seems inessential to the example.  Then he suggests that either the ellipsis is larger than usual, including "what most people think -----s are like," or the speaker "speaks from the perspective of common belief which he and his listerner know contains false beliefs they do not share."  The first of these moves is semantically desperate; the second, absent some independent motivation, is ad hoc.