Winkler's analysis of Apuleius' Metamorphoses attempts to display the author's "[extraordinary sensitivity] to distinctions of faith from fact and truth from conjecture."  I have chosen to examine several themes which will, I hope, reveal a similar consciousness shaping his earlier work, the Apologia pro se de magia. I begin with his acts of construction: of a persona, a dramatic situation, and of an ideal audience. Throughout I shall use the later "apology" of Apollonius of Tyana (produced by Flavius Philostratus in the third-century Vita Apollonii) as a rhetorical control, a choice which I justify through the speeches' choice of similar topoi.
Apuleius' speaker frequently reports people's reactions to his rhetoric. A few examples include Tannonius Pudens' aporia (inquam, Tannoni, quid taces, quid cunctaris, quid respectas? Apol. 46.3), Aemilianus' recognition of his letter (Estne haec tua epistola? quid palluisti? nam erubescere tu quidem non potes, Apol. 69.6), and his stepson Pudens' embarrassment (quid abnuis, quid recusas, postquam sollicitudinem de hereditate materna reppulisti? Apol. 101.1). The speaker also calls attention to the passage of time during his speech, an impression which the reader usually ignores but which the listener is sensitive to. The literary anecdote which he recounts concerning Sophocles (Apol. 37.1-2), though actually pertinent to his argument, is supposedly only offered to fill up the time while court attendants fetch his book (interea, dum hic quaerit [sc. librum], ego exemplum rei competens dixero... Invenisti tu librum? beasti, Apol. 36.5-37.3). 
The attention which the speaker draws (with a semblance of spontaneity) to the dramatic situation occuring around him encourages belief that the Apologia is not a text identical to the one which Apuleius prepared before he spoke, nor the text he revised after delivery, but the verbatim recording of his immediate performance. Winter 1969's examination of the role of the exceptor or court stenographer plausibly suggests that the Apologia could indeed be nothing but a transcript of court proceedings (though it certainly does not exclude any possibility of subsequent revision by the author). The text which the speaker prepared may not be complete, since (following his account) it was thrown together in the "five or six days" since Aemilianus' patroni began to disseminate their false accusations:
...quamquam istae calumniae ut prima specie graves, ita ad difficultatem defensionis repentinae fuere. nam, ut meministi, dies abhinc quintus an sextus est, cum me causam pro uxore mea Pudentilla adversus Granios agere aggressum de composito necopinantem patroni eius incessere maledictis et insimulare magicorum maleficiorum ac denique necis Pontiani privigni mei coepere.
By referring to the rapidity of Aemilianus' attack (istae calumniae repentinae) and his own (apparent) lack of preparedness (me necopinantem), the speaker reinforces the idea that he improvises on a partially prepared text, rather than a masterpiece over which he labored for months, or a showpiece which he trots out at any opportunity.
The example of Apollonius of Tyana's "apology," however, counsels the interpreter to regard these dramatic cues as literary devices, not improvised reactions to actual circumstances. Similar "reactions" to a (non-existent) audience also occur in this undelivered  speech. A few examples would include the speaker's account of a supposed objection raised by the accuser:
epikoptei me ho kategoros: akoueis gar pou kai su, o basileu, kai phesin, ouk epeide soterias aitios Ephesiois egenomen...
Phil. VA 8.7
and his description of the accuser's impatience contrasted with Domitian's attentiveness:
xuniemi, o basileu, paroxunon ton kategoron, epeide sophoteron se akroaten eirgasmai, kai moi dokeis kai prosechein to logo... Phil. VA 8.7
Here the dramatic cues are clearly not reactions to real circumstances, but rhetorical mechanisms which allow the speaker, in the first case, to change subjects rapidly, and to conclude his argument with particular emphasis in the second. Even if Apuleius' addressees (Tannonius Pudens, Aemilianus, his stepson Pudens) never actually exhibited the reactions he assigns to them, his cues function in similar ways. His description of their figurative aporia, recognition, and embarrassment urges audience members who cannot see their reactions (such as, for example, the later reader) to trust in the irrefutability of his arguments.
The very title of the speech and its use of Plato (most notably Apol. 65.5) shows that the speaker intends this speech to be received in the tradition of Plato's Apology. Two other features emphasize the Greek qualities of Apuleius' self-defense: his lack of advocate recalls Greek practice, where the defendant was expected to act as his own advocate, as opposed to the "almost universal" advocacy in Roman defense (Kennedy 1968: 427); the misunderstanding of his research as magic, so he claims, means to him that he endures what many famous Greek philosophers were forced to endure (Apol. 27).
But Apuleius certainly does not pose as the gadfly of Oea, and his invocation of classical Greek tradition is contradicted by his second-century Roman circumstances and his Ciceronian style. His target is a single obscure farmer, not an entire body of citizens. Nor, though he claims to be destitute (Apol. 19), can the possessor of nearly two million sesterces (Apol. 23) be as poor as Socrates. His sophistic peers (as Bowie 1982 shows) all came from the highest social class. Apuleius also recalls the practice of Cicero with his use of statements that create and reinforce the dramatic fiction.  As the dramatic cues create the illusion of spontaneity and improvisation, they also serve the speech's thematic purposes. In non-dramatic passages Apuleius' speaker creates for the audience the image of an improvising, though cautious, self-defender (himself) battling a prepared, if reckless, aggressor (Aemilianus). As discussed above, he contrasts Aemilianus' unexpected, long-prepared attack with his own hasty defense. To this pattern he adds further contrasting images: light and dark, visibility and concealment:
id adeo factum, quod et tu rusticando obscurus es et ego discendo occupatus. ita et tibi umbra ignobilitatis a probatore obstitit, et ego numquam studui male facta cuiusquam cognoscere, sed semper potius duxi mea peccata tegere quam aliena indagare. igitur hoc mihi aduersum te usu uenit, quod qui forte constitit in loco lumine conlustrato atque eum alter e tenebris prospectat. nam ad eundem modum tu quidem, quid ego in propatulo et celebri agam, facile e tenebris tuis arbitraris, cum ipse humilitate abdita et lucifuga non sis mihi mutuo conspicuus.
Apuleius stands in a well-lighted place (in loco lumine conlustrato), engaged in appropriate and honorable pursuits (discendo occupatus, ego numquam studui male facta cuiusquam cognosecere) which can be proudly displayed in public places (quid ego in propatulo et celebri agam). Aemilianus, by contrast, is concealed by his rural setting (rusticando obscurus) and thus concealed by an umbra ignobilitatis. Thus when he attacks the philosopher, he comes from the darkness which has been equated with ignominy and embarrassment (e tenebris, e tenebris tuis, humilitate abdita et lucifuga). Apuleius continues the image of darkness in the next section by describing Aemilianus' "blindness of lying" (mentiendi caecitatem, Apol. 17.2).
Aemilianus had all the time he required to plot his attack, while Apuleius only had "five or six days" to muster his defense. Yet a reasonable observer (in the speaker's conceit) sees Apuleius so prepared that he or she would think the philosopher invented the whole charade of justice in order to discredit his enemies (Apol. 67.6). In discussing the formal characteristics of the "apology" written (though never delivered) by his hero Apollonius of Tyana, Philostratus also recommends that the fully- prepared rhetor appear spontaneous in his delivery (logou te kateskeuasmenou men, me dokountos de, Phil. VA 8.6). Several other correspondances with Apuleius' performance appear in his review of Apollonius' "apology":
sopho de andri apologoumeno, ou gar kategoresei ge ho sophos, ha epitman errotai, ethous te dei heterou para tous dikanikous andras, logou te kateskeuasmenou men, me dokountos de, kai huposemnos esto kai me polu apodeon tou huperoptes einai, eleos te apesto legontos.
Phil. VA 8.6
The persona which Apuleius' speaker adopts closely conforms to the sophos described in Philostratus' prescriptions. His frequent arrogant jabs at his accusers declare him a huposemnos. His description of Maximus Claudius as his equal (not his superior) in learning shows he has no need of the judge's eleos; if he asks for sympathy, it is for philosophy, not her individual exponent. He exposes his enemy Aemilianus to public humiliation, but does so only in the guise of a self-defender, lest he seem to commit the aggression which is presumably beneath the sophos. Thus he reminds the audience who brought the charge:
quae omnia tam falsa [sc. the charges], tam nihili, tam inania ostendam adeoque facile et sine ulla controuersia refutabo, ut medius fidius uerear, Maxime quique in consilio estis, ne demissum et subornatum a me accusatorem putetis, ut inuidiam meam reperta occasione palam restinguerem. mihi credite, quod reabse intellegetur: oppido quam mihi laborandum est, ne tam friuolam accusationem me potius callide excogitasse quam illos stulte suscepisse existimetis.
Though he has disgraced his enemies through his exposure of their malice and stupidity, they deserve his assaults since they brought the accusation against him. Apuleius' qualification is a contradictory attempt to revise the history of the case, however, since Aemilianus might never have actually gone to court if Apuleius had not taken the role of provocateur (ultro eos ad accusandum crebris flagitationibus provocavi, Apol. 1.4).
Apuleius frequently separates the audiences of his argument by his use of invective and praise. The members of the general audience, Apuleius claims, are ignorant of much more than just philosophy (imperitos, Apol. 1.2; tam rudibus, tam barbaris, Apol. 91.1). Thus he intends Maximus Claudius, whose wisdom he frequently praises, to serve as his judge in his simultaneous defense of philosophy (sustineo enim non modo meam, verum etiam philosophiae defensionem, Apol. 3.4). The prologue to Apuleius' de deo Soc. shows a similar awareness of audiences separated by their relative ignorance and knowledge. Thus the speaker claims he will attempt to satisfy both Greek- and Latin-speaking audiences:
nam et in principio vobis diversa tendentibus ita memini polliceri, ut neutra pars vestrum, nec qui Graece nec qui Latine petabatis, dictionis huius expertes abiretis.
de deo Soc., Prol.
Here, however, the speaker will satisfy the audience which understands Greek (Maximus Claudius) and frustrate that which is limited to Latin only (Aemilianus). He taunts Aemilianus for his ignorance of Greek (Apol. 38.4-5) and farcical logic (Apol. 67.5), but praises Maximus Claudius for the depth of his learning in Greek literature (e.g., Apol. 36.3, 64.2) and patience with absurd arguments (Apol. 35.4).
Whether Apuleius' appraisal of the ignorance and specialized knowledge of the different audiences is accurate must remain speculative. His separation of the audience, however, affects the reader's perception of his argument. Tempted to choose between audiences by the speaker's exhortations, the reader naturally attempts to align himself or herself with Maximus Claudius, whom Apuleius singles out as the most intelligent, informed, and learned audience, and therefore the one most qualified to criticize his argument. Maximus clearly knows more than the reader, however, especially with regard to the prosecution's case. Though its substance may be guessed at, the reader cannot know what words the accusers actually used against the speaker, while Maximus must. The speaker exploits the reader's ignorance by putting implausible words in Aemilianus' mouth. He claims, for example, that the following statement, which the reader sees as transparent nonsense:
haec quoniam ignoro quae fuerint, idcirco magica fuisse contendo, crede igitur mihi quod dico, quia id dico quod nescio...
was actually almost a verbatim account of Aemilianus' speech (his enim paene verbis, Apol. 53.3).
An inopportune side-effect of Apuleius' privileging of Maximus Claudius as his logical arbiter, however, is that it calls the reader's attention to the contradictions occuring later in the speech. This attention might otherwise have been diverted, if not for the speaker's frequent appeals to Maximus for careful analysis of his argumentation as compared to that of his opponents. By contrast, Apollonius of Tyana claims to defend himself through the authority of Pythagoras (apologou de huper touton, theie Puthagora, Phil. VA 8.7) and thus the logical inconsistencies in his "apology" may not be as glaring.
Early in his Apologia Apuleius suggests that the use of logical analysis will help his audience distinguish between the absurd slander of his accusers and the true nature of his actions. As he stands accused of magia... quae facilius infamatur quam probatur (Apol. 2.2), he has the more difficult task of attempting a probatio of his innocence. In a demonstration of the difference between Aemilianus' infamia and the "reality" he presents, he will accomplish one of his central tasks: to separate the serious and respectable business of the philosopher from the feared and vilified practices of the magician. To achieve this separation Apuleius uses several logical strategies, which differ depending on the particular object of his scrutiny.
Some considerations recommend against submitting the Apologia to the dispassionate logical analysis which follows. Kennedy's general dictum that Roman audiences, as opposed to Greek, did not appreciate displays of "naked logic,"  is reinforced by the speaker's military metaphors and his view of the courtroom as a battleground, where emotion should take precedence over ratiocination.  Sometimes it appears, however, that Apuleius intends to teach his listeners something about logic (e.g., the lack of connection between similar-sounding names and similar properties, Apol. 34; the technique of reductio ad absurdum, Apol. 54), and thus invites the following analysis.
Apuleius proves the mundanity of some of his actions by singling out aspects of his performance which he claims would be impossible for a magician to accept. As he argues, for example, the prosecution believes that magic is to be performed in secrecy and its implements carefully guarded from the uninitiated. Thus the fact that fifteen slaves know about a practice makes it inconceivable that it could be magical:
sciunt ergo aliquid XV serui et occultum est. an occultum non est et magicum est? alterum horum fatearis necesse est, aut inlicitum non fuisse in quo tot conscios non timuerim, aut si inlicitum fuit, scire tot conscios non debuisse.
Therefore the accessibility of his basileus  and the publicity of his sojourn at Crassus' house (Apol. 58) would also make it impossible for either that image or his supposed "rites" to be magical.
Apuleius makes other items and practices innocuous by redefining their functions. He defends his ownership of a mirror by claiming he uses it in optical research and to check his appearance (not to appease his vanity, rather to follow the moralizing exhortations of Socrates and the examples of earlier rhetors, cf. Apol. 15.2-6). His acquisition and dissection of strange fish is necessary in his anatomical and medical research (Apol. 40). Though Apuleius does not record his accusers' explanation of which magical rituals require the use of these objects, he forestalls suspicion by explaining their mundane purposes. Following Apuleius' argument, his researches into fish become respectable and public (which, as argued above, is important in expelling suspicion of magic) because of the work of earlier anatomists such as Aristotle, Nicander, and Theophrastus (Apol. 41). Other practices also become defensible by virtue of their familiarity and traditional nature. Earlier philosophers, such as Solon and Plato, wrote erotic poetry, thus Apuleius claims suspicion should not fall on his own nugae (Apol. 9.5, 10.5-6).
The concept that the speaker, rather than independent observers, is the witness with final authority for his own behavior is introduced as early as his equation of peccatum and nefas (nam omne peccatum semper nefas habui, Apol. 5.3). The speaker claims that he can talk about everything he has done; therefore everything he has done must be fas, and therefore he has never committed a peccatum. The claim that crime is unspeakable, however, means that if he had committed a crime, he would not be able to tell anyone about it, since it would be impossible for him to speak about it; therefore no-one would know if he was really innocent or not. Having elegantly reserved final authority for himself, he submits his description of his behavior for his audience's appraisal.
Weaknesses appear in Apuleius' arguments, however, when he attempts to unincriminate his possessions and actions by means of a blanket defense.  His discussion of ownership of potentially dangerous objects provides a good example. Since every object can be made potentially dangerous, no object necessarily makes its owner dangerous:
nihil in rebus omnibus tam innoxium dices, quin id possit aliquid aliqua obesse, nec tam laetum, quin possit ad tristitudinem intellegi.
Also, even objects used in mundane ways can have pernicious side-effects, such as the potter's wheel which can make healthy people dizzy (Apol. 45). Every object, therefore, can be potentially incriminated and unincriminated, depending on the viewer's perspective toward it (Apol. 54). Thus every object can be advanced as simultaneous proof and disproof of its owner's magical practices, and the speaker's argument that he used the dentifrice for personal hygiene, the mirror to examine his appearance, and the fish for research becomes unsupportable. (This reductio ad absurdum does support its immediate context, however, in that it allows him to avoid describing the precise nature of the sacred objects hidden under the handkerchief.)
The speaker's discussion of the openness of his behavior (opposed to the nefarious skulking of his accusers, as discussed above) involves a similar contradiction. He claims that the fact that his business is conducted in public, with many actual or potential witnesses, means that it is mundane, not magical. Magical practices, by his limited definition, go on in secret, without witnesses; if people know about them, they must be mundane. A practice does not lose its magical nature, however, as soon as someone besides the magician and his sworn confidantes learns about it. Though as a display of logic Apuleius' argument concerning secrecy is endangered by the inconsistencies observed above, the argument again serves its immediate context as well as the earlier one, as it allows Apuleius to preserve a veneer of piety by avoiding description of his basileus. 
A similarly artificial limiting of the accuser's options also occurs in the speaker's analysis of Pudentilla's letter (Apol. 80). The argument proceeds as follows: "if she was sane when she wrote the letter, she cannot have been bewitched; if she was insane, her letter cannot be used as evidence against me." Though Apuleius' view of magic might have been more moderate in an abstract discussion, for his immediate rhetorical purposes it has become extreme. Either the magician wields complete control over the victim's sanity, leaving him or her no mental capacities, or the magic has no effect whatsoever. He allows the audience to receive no concept of partial bewitchment, although it would seem to be the goal of most erotic magic -- since what good to the magician would a thoroughly incapacitated lover be? 
Apollonius of Tyana's "apology"
The formal correspondances of these apologiae (dramatic cues, pose of spontaneity, pose of non-aggression) have been discussed above. I will conclude by examining three similar topoi raised by both speeches.
The unprosecutable magician
Before the trial begins, Apollonius wonders how his accusers can imprison him if they really think that he is a magician; surely magic must be a more powerful practice than sycophantic slander: 
ei men goes, ephe, ego, pos krinomai; ei de krinomai, pos goes eimi; ei me ara to sukophantein ischuron houtos einai phesin, hos mede ton goeteuonton hettasthai auto.
Phil. VA 7.17
Apuleius extends this argument with examples at Apol. 26.3- 5. Murderers, poisoners, and thieves can be protected against by hiring bodyguards, eating carefully, and guarding one's possessions; but what possible defense could there be against supernatural powers? Therefore no-one who really believes his opponent to be a magician would have the confidence to prosecute them.
There is a skeptic response, however, to the fears on which Apuleius and Apollonius attempt to play. One of Apollonius' persecutors offers to test the philosopher's magical powers by cutting off his head. If Apollonius is just an ordinary man, he may go free (though minus his head) after the test; but if he is a magician, then he will terrify the tribune so much that he will drop his sword -- but then the charge of magic will, unfortunately, be proven (Phil. VA 7.21). Apuleius similarly spoofs the concept of the "unprosecutable magician" in the first inset tale of his Metamorphoses. The townsfolk attempt to prosecute Socrates' lover Meroe as a witch; before they can do so, however, she magically confines them in their houses until they swear to let her go in peace (Apul. Met. 1.10).
The frugal philosopher
As the career of Lucian's Alexander of Abonoteichos demonstrates, magic can be a lucrative endeavor. If it were not, it would not attract so many charlatan practitioners. Ever since Socrates made a virtue of poverty, however, the attractions of wealth are supposed to have no effect on the philosopher. Thus Apollonius attempts to defend himself from the charge of magic by arguing that magicians are philochrematoi, but since he is a true philosopher he has no interest in monetary gain:
philochrematoi pantes, ha gar kompseuontai, tauth' huper misthou sphisin heuretai masteuousi d' huperbolas chrematon, hupagomenoi tous hotoude erontas hos hikanoi panta. tina oun, o basileu, plouton peri hemas idon pseudosophian epitedeuein me oiei, kai tauta tou sou patros kreitto me hegoumenou chrematon;
Phil. VA 8.7 
Immediately afterwards, Apollonius quotes a letter from Vespasian praising the wisdom of his voluntary (authairetos) choice of poverty. Since Apuleius is not an entirely fictional character who can survive on a vegan diet, he has to go to somewhat greater lengths to display his disinterest in wealth. His attempt to pose as a pauper and praise the virtues of poverty (Apol. 18-22) is somewhat weakened by his revelation that he actually has two million sesterces (Apol. 23). While he cannot claim that he suffers the exigencies of Socrates, he can defend himself from the charge of magic by proving that his courtship and marriage of Pudentilla did not enrich him. Pudentilla's money was protected from him, and left entirely to her sons, except for the pittance for her husband (tenue nescio quid honoris gratia, Apol. 100.1). Skeptical readers will note, of course, that a speaker who associates the poverty of Diogenes with possession of two million sesterces could probably also call a generous bequest tenue without blushing. My purpose, however, is not to speculate on the actual size of the bequest, merely to note that Apuleius' description of it supports the theme of philosophical frugality.
The tradition of persecution
For these later writers of Apologiae, the trial of Socrates occupies an honored place in a tradition of philosophical resistance to persecuting politicians. To glorify his hero's battle with Domitian, Philostratus lists other philosophers who opposed tyrants.  At Apol. 27 Apuleius includes a similar list of natural philosophers (rerum naturae patronos: Anaxagoras, Leucippus, Democritus, Epicurus) who were accused of being irreligiosos and deniers of existence of the gods. To these he couples a list of philosophers (Epimenides, Orpheus, Pythagoras, Ostanes, Empedocles, Socrates) who were accused of being magos. Having included himself in such an august tradition, he ridicules Aemilianus' charges by claiming that they are the standard assaults that always get hurled by the ignorant at philosophers and researchers (verum haec ferme communi quodam errore imperitorum philosophis obiectantur, Apol. 27.1)
Philostratus, like Apuleius, redefines Apollonius' magical practices as demonstrations of his sophia, as opposed to goeteia which he regards as pejorative (cf. VA 5.12). But Apollonius' miracles are surely too frequent and too mysterious to admit a defense identical to Apuleius'. They include the simultaneous occupation of two different places (VA 4.10); teleportation (VA 8.10); communication with the dead (VA 4.12); raising the dead (VA 4.45) battle with various magical spirits (a daimon, VA 4.20; a lamia or empousa, VA 4.25; a satyr, VA 6.27); and acts of healing and prognostication too numerous to detail. Philostratus may have been directly influenced by Apuleius:  if so, his attempt to expand the domain of philosophy to include such wonders is a radical reinterpretation of his source.
1. Winkler 1985: 21.
2. Compare, for example, the overlap of the reader's experience of time, the author's, and that of the fictional characters in Laurence Sterne's The Life and Opinions of Tristram Shandy (ed. Ian Campbell Ross, Oxford 1983):
"It is about an hour and a half's tolerable good reading since my uncle Toby rung the bell, when Obadiah was order'd to saddle a horse, and go for Dr. Slop, the man-midwife; -- so that no-one can say, with reason, that I have not allowed Obadiah time enough, poetically speaking, and considering the emergency too, both to go and come; ---- tho', morally and truly speaking, the man, perhaps, has scarce had time to get on his boots.
If the hypercritick will go upon this; and is resolved after all to take a pendulum, and measure the true distance betwixt the ringing of the bell and the rap at the door; -- and, after finding it to be no more than two minutes, thirteen seconds, and three fifths, ---- should take upon him to insult over me for such a breach in the unity, or rather probability, of time; -- I would remind him, that the idea of duration and of its simple modes, is got merely from the train and succession of our ideas..." (83-84).
3. epei de kai logos men auto xunegraphe tis hos pros hudor es ten apologian aphesomnoi, xuneile de auton ho turannos es has eireka eroteseis, anagegraphtho kai ho logos (Phil. VA 8.6).
4. Cf. Kennedy 1968 for the importance of this feature to Ciceronian rhetoric.
5. Cf. Kennedy 1968: 426.
6. Analyzed by McCreight 1990: 44-48.
7. It is left where a freedman has access to it, Apol. 53; it was made in the open by a named craftsman, Apol. 62.
8. An example of this figure would be: nihil enim, quod salutis ferendae gratia fit, criminosum est (Apol. 40.4).
9. The emphasis on secrecy in this passage, however, misled Hermann to believe that Apuleius was hiding his Christianity. Cf. Mortley 1972.
10. A similar contradiction occurs at Phil. VA 8.7 [341-3], where Apollonius gives an alibi for the time when he was supposedly killing the Arcadian boy. A man who can be in two places at once, however, can never offer an alibi!
11. Apollonius asks his judge Domitian a similar question (Phil. VA 7.34).
12. Zeno of Elea against Nearchus; Plato against the enslaved Sicilians; Phyton against Dionysius of Sicily; Heraclides and Python against Cotys the Thracian; Callisthenes of Olynthus against the Macedonians; Diogenes of Sinope against Philip, Crates of Thebes against Alexander (Phil. VA 7.1-3).
13. Cf. Dzielska 1986: 92 (who does not, however, comment on the extent of this influence).