About the Meaning of Φάντασμα in Aristotle

1 Introduction: Image, Content, or Motion?

Aristotle’s notion of φαντασία is central to his theory of soul and of animal life, even if φαντασία does not qualify as a part of the soul on his view.^[1] As Aristotle is deeply committed to the claim that animals have no rational capacity and rely exclusively on perception for cognition, φαντασία is the key notion that allows his theory to account for animal behaviour whenever it presupposes something that goes beyond immediate perception, such as memories or expectations. Aristotle repeatedly acknowledges its importance, stating that “animals live according to their φαντασίαις and memories” (Metaphysica Α.1, 980^b25, about animals that are incapable of learning; see also De anima II.3, 415^a10–11) or that the only form of universal cognition they enjoy is restricted to their “φαντασία and memory of particulars” (Ethica Nicomachea VII.5, 1147^b4–5).^[2]

This makes the absence of a definitive treatment of φαντασία (comparable, say, to the extensive account of perception in De anima) in the texts of Aristotle which we now have at our disposal all the more disappointing. The text that comes closest to a proper treatment is the third chapter of the third book of De anima. The chapter distinguishes φαντασία from other related concepts (perception, belief) and even provides a sort of definition, but explaining what Aristotle means by φαντασία is by no means its main goal. The causal account of φαντασία with which it ends (429^b10–430^a9) only comes after a number of twists and turns during which Aristotle seems to be grappling with various ways of using the same term that differ widely from his own understanding of it, and which modern readers often have difficulties delineating and articulating.

The issue has been aggravated by subsequent historical developments. Whereas our concept of perception, for instance, is still close enough to Aristotle’s αἴσθησις for us to feel comfortable with his claim that there are five sense modalities, Aristotle’s φαντασία has no immediate counterpart in everyday modern English, certainly not one with similar morphological connections (e.g. with φαίνεσθαι, “to appear”). Finding an adequate translation has proven challenging:^[3] scholarship tends to waver between the traditional rendering, “imagination,”^[4] and a more modern one, “representation” or “presentation”.^[5] Neither is fully satisfactory: the latter does not normally refer to a capacity, the former conveys, among others, connotations of innovation and fantasy that are foreign to Aristotle’s notion.^[6] As a result, both the content of Aristotle’s concept and its unity have been questioned for decades.^[7]

What has especially stirred up debates has been the claim, first put forward in modern days by Nussbaum,^[8] that φαντασία necessarily plays an “interpretative” role in the process of desire-formation in animals. The main tenet of the various “interpretative” theories of φαντασία that have subsequently been developed is the existence of a division of labour between perception and φαντασία in desire-formation, according to which perception is responsible for the animal’s mere grasp of the object of desire, while φαντασία would then be required for the animal to be able to envisage this object as worth pursuing or fleeing, thus leading to the formation of a desire (ὄρεξις) about it. Any such theory clashes head-on with the account of φαντασία offered in De anima III.3, according to which φαντασία is only an after-effect of perceptual activity and cannot have any special power that perception does not already have.^[9] As the textual evidence in favour of the “interpretative” view is flimsy even in De motu animalium,^[10] I shall in what follows rely exclusively for my understanding of φαντασία as an Aristotelian technical notion on the evidence from De anima III.3.

The present paper is not intended to pursue this controversy about the alleged “interpretative” role attributed by some to φαντασία or to offer a comprehensive account of φαντασία in Aristotle. Its purpose is only to shed new light on some of the difficulties that surround Aristotle’s notion by examining its relationship with its quasi-twin, the concept of φάντασμα, which is equally difficult to translate and, I argue, has often been misunderstood. The scope of the paper is limited to Aristotle’s use of φαντασία and φάντασμα in the explanation of non-intellectual phenomena.^[11]

Their couple is especially central to Aristotle’s theories of remembering and dreaming in the two relevant treatises in his so-called Parva naturalia, namely De memoria et reminiscentia and De insomniis. As with De motu animalium (6, 700^b21–22), both treatises contain unmistakable references to De anima III whenever the technical notion of φαντασία is employed, suggesting that the three texts are meant to offer a coherent doctrine.^[12] Still, the ways φαντασία features in De memoria et reminiscentia and in De insomniis contrast with its uses in De anima III.3 and De motu animalium in significant respects. To begin with, the word φάντασμα barely appears in the latter two, whereas it features prominently in Aristotle’s definition of both memories (“the possession of a φάντασμα as a likeness of that of which it is a φάντασμα”, De memoria et reminiscentia 1, 451^a15–16) and dreams (“a φάντασμα that results from the motion of αἰσθήματα when one is asleep, insofar as one is asleep,” De insomniis 3, 462^a29–31). As neither memories nor dreams are central to Aristotle’s theory of perception (and perhaps of φαντασία), the difference between φαντασία and φάντασμα sometimes goes by unregistered.^[13]

One immediate difference is that φαντασία can refer either to a capacity (“φαντασία is not found in some animals, for instance bees,” De anima III.3, 428^a10) or to its actual instances (“some φαντασίαι remain in the sense-organs even after the objects of perception are gone,” De anima III.2, 425^b15), whereas φάντασμα always refers to a particular instance. The rest is much more controversial.

The interpretation to which I will subsequently refer as the “traditional view” regarding Aristotle’s notion of φάντασμα is composed of the following claims. (1) Φάντασμα is in Aristotle a technical term tied to his own theory of φαντασία: being a product of φαντασία is part of its meaning.^[14] (2) A φάντασμα is the mental content of which the subject is aware in a determinate φαντασία, what one is mentally presented with during an instance of φαντασία.^[15] In modern-day parlance, φάντασμα roughly translates “representational content”. (3) The most adequate renderings for φαντασία and φάντασμα are respectively “imagination” and “image”.^[16] It is then often claimed that each and every φάντασμα, being an image, possesses a representational content.^[17]

The traditional view has been met with resistance. Caston (2021), for one, has recently challenged simultaneously (A) any attempt at translating φάντασμα by means of “image”-terms and (B) the claim that a φάντασμα is the content of awareness occurring to the subject in a φαντασία, arguing that both lead to the same disastrous result of committing Aristotle to a form of indirect realism.^[18] Note, however, that if A plausibly goes along with B, the reverse is not true: a content-interpretation of the concept of φάντασμα does not entail an “imaginistic” understanding of it. Building on several passages where Aristotle relies on physiological aspects of the experiencing of φαντάσματα involved in memories and dreams, Caston argues that any interpretation that makes Aristotelian φαντάσματα contents of awareness necessarily entails the frightful thesis that experiencing φαντάσματα involves being aware of what goes on, physiologically speaking, in one’s sense-organs. There would be only one remedy: the notion that φαντάσματα are contents of any awareness should be abandoned in favour of a strictly physiological interpretation of the concept.^[19] On Caston’s “physiological view,” when a φάντασμα occurs in the perceptual system, the subject is presented with some mental content without being aware of the underlying physiological process, i.e. the φάντασμα as such.

The claim that φαντάσματα are contents of awareness, which is part of the traditional view, fits nicely with the observation that Aristotle frequently speaks of φαντάσματα as being objects of cognition. It also allows for a clear-cut contrast between φαντασία and φάντασμα, the latter being either the susceptibility to having mental representations or a particular representation generated on its basis and the former the content of such a representation. Insistence on the rendering “image” (or “representation,” for that matter), however, may be in tension with this construal of the distinction, given that an “image” usually refers to something that has representational content itself. The view also leads to a precarious situation when passages where Aristotle provide physiological descriptions of the coming-to-be of φαντάσματα are taken into account:^[20] as stressed by Caston, the assumption that φαντάσματα are potential or actual contents of awareness may then result in the prima facie implausible claim that cognizing a φάντασμα means for the subject to be aware of the physiological process her perceptual system is undergoing. By contrast, adopting a strictly physiological interpretation of φάντασμα in Aristotle certainly avoids this result. It deprives, however, Aristotle’s lexicon of a word to pick out the content of which one is aware in a given instance of φαντασία,^[21] whereas his theories of memory and dream-interpretaton crucially depend on a description of the way in which the subject envisages this content. The view also struggles to make the sense of the primary textual basis for the traditional view, as passages where Aristotle speaks of a φάντασμα “appearing” or “being perceived” would have to be given a strictly physiological paraphrase. Both views agree on the initial claim that φάντασμα is a piece of Aristotle’s philosophical jargon that is part and parcel of his theory of φαντασία.

Two questions must here be kept apart. I/ Is a φάντασμα the content of awareness in a given φαντασία or its underlying physiological process (if not yet something else)? II/ Is it appropriate to translate φάντασμα by means of “imaginistic” vocabulary? I claim (a) that Aristotle’s technical term φάντασμα is to be construed against a pre-existing linguistic background (and not, for example, as a neologism derived from φαντασία), and that both Aristotle’s and this pre-existing usage have the word refer to what manifests itself in a given apparition (φάντασμα means roughly τὸ φαινόμενον, “what appears”), so that a φάντασμα is necessarily something of which one is aware. I also claim (b) that, despite several passages in Aristotle where the term is associated with words such as εἴδωλον, ζῷον or εἰκών, it is quite misleading to translate φάντασμα “(mental) image”. As a translation for φάντασμα “image” is somewhat ambiguous, especially when associated with cognitive terms: “seeing” or “having a (mental) image” may either imply that the subject having the image is aware that what she is presented with has some representational content (when deliberately picturing a tree, one is aware that one’s mental picture of a tree is not a real tree but is standing for one), or have no such implication, in which case “(mental) image” operates in much the same way as “mental representation”, providing a third-person description regarding the state the subject finds herself in. In the first case, I argue that the translation is wrong: φάντασμα picks out the content of a quasi-perceptual representation experienced by a subject, who may or may not envisage this content as bearing some relationship with something else (for instance with past objects of perception, in the case of remembering). The subject need not envisage her representations as representations in order to have them, even though having a representation necessarily entails being aware of its content. In the second case, the translation is infelicitous to say the least, as it does not capture the distinction made in Aristotle’s Greek between what presents quasi-perceptual contents of awareness to the subject (φαντασία) and these contents of awareness themselves (φάντασμα). In both cases, such translations become disastrous for the various passages where Aristotle draws analogies between φαντάσματα and actual images (such as reflections and paintings), which will be analyzed below. The interpretation I defend agrees with Caston against the traditional view^[22] on the denial of any intrinsic representational aspect to Aristotle’s φάντασμα and with the traditional view against Caston on the claim that one is always aware of the φάντασμα one experiences.

2 Tragic φαντάσματα

It has seldom been noted that φάντασμα is not a word that Aristotle (or any of his fellow members of the Academy or of the Lyceum) coined. The word φαντασία is first attested in Plato’s Republic (II, 382e10)^[23] and must have been, in Aristotle’s days, a very theoretically loaded term, especially in the wake of Plato’s Sophist. This is arguably not the case with φάντασμα. Several occurrences of the word φάντασμα before Plato have been preserved. All three belong to tragic plays, they all have to do with preternatural manifestations, especially of deceased individuals.

1. In Aeschylus’ Septem, Eteocles, before meeting his brother on the battlefield, observes that his dreams had already warned him about the fatal outcome of the conflict. We do not know what exactly these dreams are supposed to be, their content had probably been mentioned in the previous play in the cycle which is now lost. Eteocles uses the word φάντασμα to refer to what he has foreseen in his past dreams: ἄγαν δ’ ἀληθεῖς ἐνυπνίων φαντασμάτων / ὄψεις, πατρώιων χρημάτων δατήριοι (710–11), “there was too much truth in these specters I saw in my dreams that were dividing up my father’s possessions”.^[24]

2. In Euripides’ Hecuba, the plot’s main driving forces are at the outset (a) the request made by Achilles’ ghost, demanding the sacrifice of Hecuba’ daughter Polyxena, and (b) the murder of another of her children, her son Polydorus, by his host king Polymestor. Hecuba, being already aware of the former, learns about the latter during the first scenes, whereby the audience is informed of both. The play opens with Polydorus’ ghost narrating his fate before leaving the stage to make way for his mother. At this exact point, Polydorus’ ghost presents Hecuba as being “scared by [his] φάντασμα” (φάντασμα δειμαίνουσ’ ἐμόν, 54). One could take this comment to mean that Hecuba, who is getting out of Agamemnon’s tent, is supposed to catch a glimpse of the ghost, but this would not sit well with the fact that Hecuba, in her speech, does not make any mention of a ghost, suggesting instead that only the audience is supposed to have seen it.^[25] What is troubling Hecuba are rather her dreams, as she will now explain: she has had visions of her son’s death, but still wants to believe him to be alive. The φάντασμα in the ghost’s comment is therefore much more likely to be Polydorus’ apparition in his mother’s dreams.

3. In the following scene in the same play, Hecuba laments the request made by Achilles’ ghost, which she also describes as a φάντασμα: ἦλθ’ ὑπὲρ ἄκρας τύμβου κορυφᾶς / φάντασμ’ ̓Αχιλέως, (93–4), “the ghost of Achilles has come at the top of his grave”. Although it takes place after the person’s death, this epiphany has occurred publicly. The word φάντασμα does not refer here to something that is privately revealed to someone in her dreams, but to a preternatural apparition that has been witnessed by sailors (109–119).

This is all elevated speech, but the way the word φάντασμα is used is not tied up with any specific piece of philosophical reasoning. Rather, these three occurrences point to the existence of a pre-Aristotelian ordinary meaning of the word: one speaks of a φάντασμα whenever there is an appearance or manifestation of some person or object that does not exist or no longer exists (especially when the person is now dead). As the thing or person that “appears” in this process is known or suspected by its audience to be not completely real (Eteocles and Hecuba are retrospectively aware that they have been dreaming, Achilles is known by the sailors to have already died), its existence is, so to speak, presumed to be exhausted by its apparition.^[26] As a result, such “appearances” may be true or false, and one can believe them or refuse to do so. Believing them might amount to mistaking them for the real thing, it can also involve an awareness of this distinction and be about their relationship to the real thing. Hecuba’s sailors believe what they see to be really expressing Achilles’ posthumous will and it is decided that his wish should be granted, whereas Hecuba, when the play begins, still refuses to believe that her son has been murdered, despite her dreams.

I take it that the philosophical potential of the concept did not escape Aristotle. Once the connection with the peculiarities of such ghost stories is severed, what we have here is a word that captures special quasi-perceptual situations in which something is experienced by a subject without being real or even assumed to be real.

3 De anima III.3 and Its Contexts

The puzzling way in which the topic of φαντασία is introduced in De anima III.3 has long been noticed.^[27] Aristotle’s aim is, at first, to dispel a mistake that has confounded his predecessors for too long, that of failing to distinguish thought from perception. As a result, they have been incapable of explaining how errors come to be (427^b2–4): whereas perception, or at least some part of it, is always true due to its being a causal process, some types of thought, such as discursive thinking (διάνοια) and belief (δόξα), enjoy no such immunity to errors. Aristotle introduces φαντασία at this very point, almost as an after-thought, pointing out that it differs from both perception and thought (427^b14).

It is difficult indeed to understand why φαντασία is introduced at this point if one assumes that Aristotle is already operating with his own technical notion. Aristotle, however, must have known very well that he was not the first one to invest φαντασία with a special philosophical meaning. The Stranger in Plato’s Sophist defines δόξα and φαντασία jointly (264a1–7):^[28] according to his stipulations, whenever the soul reaches, after internal discussion (διάνοια), a resting point and pronounces something to be true or false, this state is called δόξα; whenever the conclusion is reached by means of perception (διʼ αἰσθήσεως), it is called φαντασία. There are strong echoes of the same concept of δόξα in other dialogues of the same period, for instance in the Theaetetus, where δόξα is already described as a silent statement made to oneself (189e4–190a7), or in the Philebus (38e1–5). The Stranger’s concept of φαντασία, on the other hand, remains an outlier in Plato.^[29] Even in the Sophist, the distinction between δόξα and φαντασία is not completely clear, what is meant by the clause διʼ αἰσθήσεως is not explained any further (does any perceptual input whatsoever make the resulting state a φαντασία instead of a δόξα?).^[30] The Stranger does not use his concept of φαντασία again after 264b, suggesting that the finer points of the distinction do not matter a great deal to him after all. Instead, he decides to call any kind of belief a δόξα.

The introduction of φαντασία in De anima III.3 offers clear echoes of the main concepts in the Stranger’s definitions. Aristotle sets out to distinguish perceiving and “reasoning” (φρονεῖν, 427^b7), pointing out the differences between various kinds of thinking. Whereas the first item is always called αἴσθησις throughout, the umbrella-term for thinking switches to νοεῖν (^b9) and finally to διανοεῖσθαι (^b13; διάνοια, ^b15), thus reiterating the opposition found in the Sophist. It is only at this point that φαντασία enters the stage.

Although the clause is introduced by means of the particle γάρ, the argumentative purpose of the claim that φαντασία is different from both perception and διάνοια (^b14–15), in which φαντασία makes its first appearance in the chapter, is not immediately clear. One would expect it to offer support for a thesis that would have been just expressed in what precedes, but it is hard to identify a suitable claim there: this status of φαντασία does not seem to support the previous claims that διανοεῖσθαι may only occur in animals that have λόγος (^b13–14) or that διανοεῖσθαι may be false (in contradistinction to perception, ^b12–13). It is worth pointing out, however, that the previous sentence is also introduced by means of the same particle (ἡ μὲν γὰρ αἴσθησις …, ^b11): this opens up the possibility that both occurrences of the particle γάρ could perform in parallel the function of supporting the same claim.^[31] This previous claim would, then, be the thesis (T) that thinking should not be identified with perception (οὐδὲ τοῦτό ἐστι ταὐτὸ τῷ αἰσθάνεσθαι, ^b11; see also ^b6–7). It would receive two different justifications: (1) διάνοια is sometimes false and requires reason, perception does not (^b12–14); (2) φαντασία is different from both perception and thought (^b14–15). What kind of support is being offered for (T) in (2) depends, obviously, on what φαντασία is taken to mean in that context.

Placing the Sophist in the background is helpful here. In Plato’s dialogue, φαντασία, far from being clearly demarcated from perception and thought, is presented as a sort of intermixing of both (“φαίνεται” δὲ ὃ λέγομεν σύμμειξις αἰσθήσεως καὶ δόξης, 264b1–2, reactivating the connection between φαντασία as just defined by the Stranger and the φαίνεται-locution). Aristotle is going to attack this very phrase in what follows (428^a24–26), and the next section of the chapter (until 428^b10) reads like an aggressive attack against any such theory of φαντασία that fails to separate it from both perception and thought.^[32]

Still, it is difficult to imagine that Aristotle would devote so much energy to fighting against a concept that, at the end of the day, barely features in the Sophist. Aristotle’s various arguments in De anima III.3 suggest that he has a much larger target in mind, one that radiates in various dialogues of Plato. Among the three descriptions of φαντασία quoted side-by-side in the polemical section of the chapter in 428^a24–26, two are directly inspired by the Sophist (δόξα δι’ αἰσθήσεως: cf. 264a4; συμπλοκὴ δόξης καὶ αἰσθήσεως: cf. 264b2), but the remaining one (and the first one to be mentioned), δόξα μετ’ αἰσθήσεως, originates in the Timaeus (28a2).^[33] Strikingly enough, although the phrase is used in the Timaeus to describe one of the main modes of cognition, it is never associated with the word φαντασία in the dialogue. Similarly, one of Aristotle’s examples of a situation captured by the φαίνεται-locution is that of having trouble seeing something and having the impression, without being quite sure, that it is a human being (428^a12–14). The example is strongly reminiscent of a situation described with much greater detail in the Philebus, whereby one sees a human figure from afar and wonders whether it is a human being or a statue (38c2–d11). The description of one’s thinking process in such a situation provided in the Philebus makes explicit use of every main feature of the Sophist’s account: what happens is an inner dialogue within one’s soul prompted by perception, which culminates in one’s affirming or denying a proposition, i.e. in a δόξα.^[34] Let us call this theory of φαντασία that Aristotle sees at work behind all these passages in Plato the “Academic” theory.

Aristotle does not take his own understanding of φαντασία for granted at the outset of De anima III.3 but goes at considerable length to dispel a different notion bearing the same name that threatens to disturb his own. My suggestion is that we still find ourselves in a gray area when φαντασία enters the stage in 427^b14. The “Academic” theory uses φαντασία to refer to beliefs as co-products of perception and thought. Such φαντασίαι can be true and false, they obviously have a role to play in the explanation of error. Even if the “Academic” theory does not amount to conflating perception and thought as crudely as Aristotle takes, for instance, Empedocles to have done, it still presupposes that the two must typically come together for errors to occur, as if perception were polluting thought and leading it astray. I suggest that this background of an “Academic” theory accounts for the way φαντασία is introduced in De anima III.3 in the context of a discussion of how errors come about. Identifying thinking, or even the types of thinking that cannot be mistaken, with perception, as Empedocles did, is a non-starter when it comes to explaining errors. Claiming that errors occur when perception mingles with thinking, i.e. having an “Academic” theory of φαντασία that describes ordinary beliefs as involving both perception and thinking, will not do either. With the “Academic” context, claim (2), i.e. “φαντασία is different from both perception and thought,” becomes relevant for thesis (T), i.e. “perception is not identical with thought,” given the existence of theories that have perception and thought “mingle” in φαντασία.

To move from the “Academic” theory to his own technical notion of φαντασία, Aristotle uses several arguments that are based on ordinary speech. As the “Academic” theory has a tendency to generalize the usage of φαντασία to cover any kind of belief or impression whatsoever as long as perception is involved, Aristotle points out that the φαίνεται-locution with which the noun is morphologically associated is only linguistically appropriate in situations in which one’s perception is known to be impeded (428^a12–14: λέγομεν … ὅτι φαίνεται), and that it would not be appropriate in situations where one clearly sees something and acquires the corresponding belief.

The first and only occurrence of the word φάντασμα in the chapter belongs to a similar context.

Εἰ δή ἐστιν ἡ φαντασία καθ’ ἣν λέγομεν φάντασμά τι ἡμῖν γίγνεσθαι καὶ μὴ εἴ τι κατὰ μεταφορὰν λέγομεν, μία τίς ἐστι τούτων δύναμις ἢ ἕξις καθ’ ἃς κρίνομεν καὶ ἀληθεύομεν ἢ ψευδόμεθα. Τοιαῦται δ’ εἰσὶν αἴσθησις, δόξα, ἐπιστήμη, νοῦς.

If, then, φαντασία is that on the basis of which we say that a φάντασμα occurs to us and unless we are speaking metaphorically, φαντασία is one of the capacities or states on the basis of which we discriminate and issue truths and falsehoods. Perception, belief, knowledge and thinking are such. (De anima III.3, 428^a1–5)

Although the transmission of the text does not present significant variant readings, the main clause has regularly been challenged on the grounds that Aristotle cannot mean to claim that φαντασία is a genuine cognitive capacity, starting with Trendelenburg (1833), 475.^[35] The passage is often read as an indication that Aristotle intends at this point to switch to his own understanding of φαντασία, for the reason that it rests on the explicit connection of φαντασία with the coming-to-be of a φάντασμα: Modrak (1987), 82, sees in this connection a “nominal definition” of φαντασία, Corcilius (2014), 72, a “stipulation” of how the word is to be used. It would be strange, however, for Aristotle to presuppose that his audience is already familiar with what he means by φάντασμα, especially if he has a special technical meaning of his own for it. The word φάντασμα does not occur again in the chapter, Aristotle’s final account of φαντασία remains remarkably silent about the topic.

There is a special use of φάντασμα in Plato’s Sophist. It has, however, little relevance for De anima III.3: the Stranger uses φάντασμα to single out a species of products of imitative arts. Images (εἴδωλα – the Stranger’s lexicon is borrowed from visual arts but is meant to have a much wider extension) are divided into likenesses (εἰκόνες) and φαντάσματα, according to whether their intrinsic properties (such as their inner proportions) reproduce those of their model or merely seem to do so (236a4–c8). By contrast with what happens in Aristotle, the notion of φάντασμα is never connected with that of φαντασία,^[36] even if it seems obvious enough that, based on the Stranger’s definitions, a φάντασμα aims at deceiving by creating an erroneous belief based on perception, that is, that the end of every artifact of the φάντασμα-type is to generate a false φαντασία.

The definition of φάντασμα in the Sophist remains close to the way the word is used in Attic tragedies. There is a common core: a φάντασμα is in both cases a thing that purports to be something that it is not. This cannot be exactly what Aristotle has in mind in De anima III.3, given that this φαντασία that generates a φάντασμα is said to result in either truth or falsity (although Aristotle is going to claim that φαντασία is usually false, 428^a12). What, then, does Aristotle mean by φάντασμα in De anima III.3, and how does it help him get closer to the notion of φαντασία he intends to deploy?

One common way of tackling these questions is to assume that φάντασμα already has its full-blown technical sense (whatever that is) here and that restricting φαντασία to cases where a φάντασμα is involved enables Aristotle to depart from a much broader notion of φαντασία he wants to reject. I offer a slightly different answer. I agree that the passage is meant to clarify what Aristotle means by φαντασία, its purpose is to distinguish it from what I have been calling the “Academic” notion of φαντασία which is meant to apply to virtually any perception-based belief. I very much doubt, however, that Aristotle’s way of achieving this goal is to parachute a concept of his own making.

One of Aristotle’s main strategies against the “Academic” theory of φαντασία is to drive a wedge between its jargon and ordinary speech. The φαίνεται-locution, Aristotle points out, is normally used to refer to what one has the impression of perceiving if and only if the speaker wants to indicate that she is unsure about the objects of her perception, especially due to suboptimal or suspicious perceptual conditions. “X φαίνεταί μοι, appears to me” usually means something along the lines of “I think I’m seeing X but I might be wrong”, being aware of some special optical conditions. Under normal circumstances, it is not an appropriate way to report one’s perceptual experience. By contrast, the “Academic” theory, Aristotle claims (relying on a Theaetetus-style equivalence whereby “I have a φαντασία that …” is tantamount to “it appears to me that …”), would have us generalize the use of the φαίνεται-locution to express practically any belief whatsoever as long as it bears some connection to perception. To be sure, proponents of such a theory may have valid reasons to discard perception as a way of acquiring knowledge. Aristotle, however, is content with pointing out that the way they speak is not the way people normally speak.

The one and only appearance of φάντασμα in De anima III.3 can be read along the same lines as another way of opposing the “Academic” theory to ordinary speech.^[37] Modern readers of Aristotle may not be terribly familiar with any “ordinary” meaning of φάντασμα and believe the word to be a purely technical term. Yet the evidence offered above suggests that Aristotle’s contemporaries are already familiar with the word and that its usage does not necessarily imply an intricate theory of perception and belief. One speaks of a φάντασμα, i.e. of a presumably misleading “apparition,” when one encounters something that goes well beyond everyday perception, such as a ghost. By contrast, the “Academic” theory, Aristotle claims (using once again an equivalence between “having a φαντασία” and “seeing or perceiving a φάντασμα”), would have us speak of a φάντασμα about any object of perception.

This is where Aristotle draws the line: his theory of φαντασία will not depart from ordinary speech in its use of φάντασμα or the φαίνεται-locution. Something needs to really “appear” for the experience to be a genuine case of φαντασία. Aristotle claims, on this basis, that φαντασία is a cognitive capacity or state,^[38] because φαντασία is now made responsible for our access to these special objects that anyone would call φαντάσματα. It will, of course, turn out later in the chapter that φαντασία shares its objects with perception and is causally dependent upon perception for their cognition (428^b11–13), so that there is no reason to posit it as a distinct cognitive capacity. That, however, cannot be presupposed in 428^a1–5, so that the fact that φαντασία gives access to φαντάσματα remains, at this point, a valid reason to envisage it as a cognitive capacity.

4 Dreams

The isolated use of the word φάντασμα in De anima III.3 is thus, as I read it, non-technical in that (a) it relies on ordinary intuitions about its meanings, and in that (b) it occurs in a context where the concept of φαντασία is still open for discussion, so that the word cannot yet mean “object” or “product” of a genuine full-blown Aristotelian φαντασία. The passage is still too terse to yield any definitive result. The decisive piece of evidence in favour of a non-technical meaning for φάντασμα in Aristotle is provided by his scientific study of dreams.

The investigation begins in De insomniis 1 with a question about the psychic capacity responsible for dreams, which could be either the perceptual or the intellectual part of the soul (458^a33–^b2). Aristotle toys with his audience for a while,^[39] showing that dreams should not be ascribed to either of them, before eventually pulling φαντασία out of his hat (starting from 458^b29 and once and for all in 459^a9–22). Before that, however, Aristotle feels already perfectly comfortable describing dreams as φαντάσματα throughout the entire discussion. It might well be that this usage is partly intended as a stepping-stone towards the eventual attribution of dreams to φαντασία and to the perceptual part of the soul. Still, if one assumes that the word φάντασμα carries with itself the full-blown Aristotelian theory of φαντασία, it just would not make sense for Aristotle, having just asked whether dreams are to be attributed to the perceptual part or not, to casually assume that dreams are products of φαντασία (if this is what being a φάντασμα amounts to) while still exploring alternative options. Describing dreams as a φαντάσματα would immediately entail that they must be attributed to the perceptual part in virtue of its connection with φαντασία. Aristotle would be giving away his answer much too soon.

Dreams are first called φαντάσματα in the course of an argument that purports to show that dreams should not be ascribed to the capacity for thought. The argument makes no use of the connection between φάντασμα and φαντασία, it rests on the observation that one can, when dreaming, think about different things than what one is presented with, just as one can think about something different when perceiving (hearing a talk sometimes does not prevent me from wondering about lunch). The phenomenon is expressed by the phrase “having other things in mind along with one’s φαντάσματα” (παρὰ τα φαντάσματα ἄλλα ἐννοῦμεν, 458^b18). The identification of these φαντάσματα with what Aristotle calls dreams, ἐνύπνια, is secured by the fact that a similar phrase, παρὰ τὸ ἐνύπνιον ἐννοῦμεν ἄλλο τι, has just been used (^b15) to describe the same situation.

Some clarification about Aristotle’s concept of ἐνύπνιον is here required. In modern-day English, the concept of “dream” usually includes not only our quasi-perceptual nocturnal experiences, but also our concomitant mental reactions and thoughts: if I dream of a scary monster chasing me, my thinking about ways to escape is part of that dream. This is not the case with the Greek ἐνύπνιον: an ἐνύπνιον is limited to what one “sees” during sleep, one’s thoughts about it are not part of it.^[40] The same goes for the other main Greek word for “dream,” ὄναρ, which originally refers to the figure that visits the sleeper at night,^[41] and not to her global nocturnal experience. This is why ἐνύπνιον, a word that is obviously derived from the corresponding adjective, is regularly associated with vision words in Greek literature of the classical period.^[42] Even Aristotle continues to associate ἐνύπνιον with the verb ὁράω,^[43] although he has no intention of claiming that dreams involve one sense-modality only.^[44] While we say that we see something in a dream, Aristotle and his contemporaries would say that they see an ἐνύπνιον.

It is legitimate for Aristotle to call an ἐνύπνιον a φάντασμα at this stage of the argument only if it does not already presuppose any theory about what dreams, ἐνύπνια,^[45] are and how they come to be. This is not the case if φάντασμα is assumed to mean “product of φαντασία”. The condition is only met if φάντασμα is assumed to remain close enough in its meaning to the usage observed in tragic plays: understood in this way, the term only emphasizes that what one sees when dreaming is presumably not a real object, without adding any further commitment regarding its nature or causal origin.

Let us go back to the argument in De insomniis 1. Aristotle has started by claiming that what he calls an ἐνύπνιον cannot be grasped by means of δόξα, because δόξα is not self-sufficient as a capacity and requires a prior cognitive capacity, such as perception, that presents it with its object (458^b10–15). Δόξα, however, is not the only kind of activity of which the intellectual part is capable. Two types of such activities are said to be liable to occur in the course of a dream (^b15–16): one can form a belief about the ἐνύπνιον one sees (“this is a horse”, according to Aristotle’s example), one can also entertain thoughts (ἐννοεῖν) about something else entirely. Thus, the intellectual part is able to be active about a different object at the same time as the ἐνύπνιον is being cognized or quasi-cognized.

This intellectual activity, once again, is not part of the ἐνύπνιον, it is only prompted by its “seeing”. One could still try to maintain that none of this prevents our thinking capacity from being responsible for the ἐνύπνιον one sees, provided that the same capacity can be simultaneously active with regard to two different objects. This option is ruled out, however, by means of an analogy with ordinary perception, which often leads in the same way to thoughts about different things (^b16–17): what triggers this kind of thinking when one is awake is perception, the ἐνύπνιον plays the same role when one is asleep. As a result, if there is to be any identity between the activities for which the intellectual part is responsible when perceiving and when dreaming, it cannot be responsible for the coming-to-be of the ἐνύπνιον. The argument does not rely at any point on any connection between φάντασμα and φαντασία, the reader remains in complete darkness regarding the capacity responsible for generating one’s dream.

Aristotle’s final definition of dreams continues to characterize them as φαντάσματα. At no point in the text does Aristotle signal that he is switching from his initial notion of φάντασμα to something else, suggesting that the meaning of the term φάντασμα in this context in which it is connected with technical notions such as αἴσθημα and φαντασία is not envisaged by Aristotle as departing from ordinary usage. This observation runs counter to the thesis that an Aristotelian φάντασμα is a strictly physiological process, as ordinary φάντασματα are things that are “seen” and of which one is aware.^[46]

Definition of dreams: a dream is “the φάντασμα originating in the motion of αἰσθήματα, when it occurs in sleep, insofar as one is asleep” (τὸ φάντασμα τὸ ἀπὸ τῆς κινήσεως τῶν αἰσθήματων, ὅταν ἐν τῷ καθεύδειν ᾖ, ᾗ καθεύδει, De insomniis 3, 462^a29–31)

Moreover, the definition includes a causal component (ἀπὸ + genitive expressing causal origin) indicating that dream-φαντάσματα result from a motion of αἰσθήματα. This causal component seems redundant if one assumes that being a product of φαντασία is part of the meaning of φάντασμα in the definition and that every instance of φαντασία has an αἴσθημα as its cause.

Again, there is some controversy concerning the meaning of the term αἴσθημα: even if it is generally agreed that it refers to something that results from perceptual activity, it can be understood as referring to something strictly physiological (a physical motion or state in the sense-organs or in the blood they contain brought about by perception) or to something of which the subject is necessarily aware. One further passage in De insomniis 2, 406^b2–3, has often been adduced in the debate, as Aristotle seems there prima facie to say that the αἰσθήματα that persist after the external sense-objects have ceased to cause perception are themselves perceptible (ἀπελθόντος τοῦ θύραθεν αἰσθητοῦ ἐμμένει τὰ αἰσθήματα αἰσθητὰ ὄντα).^[47] On a physiological reading, the claim is alarming: if one believes that every perception involves the formation of an αἴσθημα (the sense-object must have an effect on its perceiver if it is to be perceived), claiming that every αἴσθημα is perceptible seems to lead to the introduction of a second and superfluous object of perception (every time I see something red, I would also be aware of something happening in my eye), or, perhaps even more worryingly, to the claim that the awareness of the αἴσθημα is the basis for the perception of the proper αἰσθητόν.^[48]

Fortunately, the passage does not commit Aristotle to any such view, the αἰσθήματα in question are only said to be perceptible (and sometimes actually perceived) in the absence of the external objects the perception of which has caused their formation. The passage does not entail the thesis that an αἴσθημα must be perceived for perception to occur, but only that some αἰσθήματα may be perceived in possibly exceptional circumstances. The sentence is backward-looking, and one of Aristotle’s standard examples for perceptible αἰσθήματα in what precedes is the residual visual impression left behind by one’s looking at the sun (459^b7–18). The point of the example is not that, when I look at the sun, I also perceive that my retina is being affected (although it will be damaged if I look at the sun for too long and I am likely to notice it then): one’s seeing the sun, as with any normal object of vision, does not necessarily involve perceiving the αἴσθημα it results in, the latter is not constitutive of the former. That being said, the bright spot one sometimes keeps on seeing afterwards still deserves to be called an αἴσθημα, inasmuch as it is a direct outcome of the perceptual process,^[49] and it is clearly something that one perceives.

One could worry about having both φάντασμα and αἴσθημα refer to contents of awareness that are grasped by means of the perceptual capacity and result from one’s having engaged in perceiving previously, as the two notions now seem to compete for the same conceptual spot. Aristotle himself does not always draw an air-tight distinction between the two, as he is willing to describe dreams as being αἰσθήματα “in a certain way” in De somno et vigilia 2, 456^a26. His preferred usage, when dealing with a content of awareness that results from earlier perceptual activity, seems to be to reserve αἴσθημα for immediate effects of perception, whereas φάντασμα applies to cases in which what “appears” is much more distant from what one has been perceiving, both in terms of content and with respect to the causal sequence. What one sees when dreaming is sometimes quite far-fetched and does not always bear an obvious relationship with what one has perceived during the day, which corresponds to the fact that a lot can happen between the first imparting of the motion to the sense-organs and its resurfacing at night.

Therefore, even if they sometimes converge, the terms αἴσθημα and φάντασμα have fundamentally different meanings. An αἴσθημα is defined by its causal ancestry: it is an immediate effect brought upon the perceptual system by the sense-object when it is perceived. This usually goes unnoticed, apart from marginal cases (e.g. remanent bright spots), what the subject becomes aware of in perception is the sense-object. By contrast, a φάντασμα is necessarily something of which one is aware when it comes to be in one’s perceptual system, although its reality and nature may be questioned by the subject. It is not defined by its causal ancestry: Aristotle claims that a dream-φάντασμα normally results from daytime perception but does not completely rule out (at least on conceptual grounds) that some people may exceptionally receive dreams from gods.^[50] Such dreams would obviously have a special causal origin, they would nonetheless remain φαντάσματα.

This explains why Aristotle’s final definition of dreams as φαντάσματα includes αἰσθήματα as their causal antecedents. Dreams occur because one’s perceptual system has been previously affected, i.e. because some sense-objects have previously caused αἰσθήματα when they were being perceived. These αἰσθήματα need not be noticed for perception to occur, and they will typically not survive in one’s perceptual system for very long. Some, however, will persist in some form and be a part of physiological chain reactions in the perceptual system that will result in dreams. The final definition of dreams highlights this causal relation because it is not included in the notion of φάντασμα as such.

5 Reflections and Paintings

I have argued that the connection with ordinary speech which Aristotle intends to maintain entails that a φάντασμα should necessarily be grasped by the subject to whom it occurs. This awareness is expressed by means of different verbal forms in Aristotle. The word φάντασμα can operate either as the subject of a manifestation process, expressed by verbs such as φαίνεσθαι (De insomniis 3, 462^b8), ἐπελθεῖν (De memoria et reminiscentia 2, 453^b9) or γίγνεσθαι (De anima III.3, 428^a2; De insomniis 2, 460^b17), or as the object of some cognitive activity, especially in the case of dreams and memories, expressed by verbs that are usually attached to either sense-perception or thought (αἰσθάνεσθαι: De insomniis 1, 458^b9; De memoria et reminiscentia 1, 450^b14, ^b18; θεωρεῖν: De memoria et reminiscentia 1, 450^b18; ἰδεῖν: De insomniis 1, 458^b20, De divinatione per somnum 2, 464^a3–4, Historia animalium IV.10, 537^b18).^[51] Some of these verbs present a strong visual component, which might seem to support the traditional rendering “image”. The fact that a φάντασμα is not necessarily exclusively or primarily visual does not present a significant obstacle, as Aristotle himself highlights the connection between φαντασία and φῶς (De anima III.3, 429^a2–4) even though φαντασία ranges across all sense-modalities.^[52]

An “image”, however, is not just any kind of visual content. To qualify as an image, an object usually needs (1) to be perceptible by means of sight and (2) to depict or represent something else (no matter whether this relationship is thought of in terms of resemblance, denotation, etc.). “Seeing an image” as an image entails envisaging it as possessing a representational content on the basis of some of its visual properties:^[53] I see the image I, which is a perceptible object, and I grasp that I represents some content C. C is is not something I perceive, it need not even be something that ever existed. Any talk of my seeing C in this context is necessarily derivative on my seeing I. Without its content, an image is just an ordinary object of vision. As a result, one may fail to grasp the representational content of an image (one might, say, fail to spot the man or the guitar in Braque’s Man with a guitar and find it to be an abstract painting) or misidentify it.^[54]

I very much doubt that the notion of φάντασμα involves any such representational content. There is no threat of failure to access any representational content in the case of a basic φάντασμα: one either sees one’s child in one’s dream, hallucination, or daydreaming – or one does not see it. Only the analyst might wonder, retrospectively, if what the patient reports having seen stands for her child.

Still, proponents of the “image” rendering will insist, Aristotle himself is prone to using what are, surely, some of the most “imaginistic” Greek substantives, εἴδωλον and εἰκών, as analogous, perhaps even interchangeable with φάντασμα.^[55] Let us have a closer look at the relevant passages in order to assess whether Aristotle’s own understanding of φαντάσματα requires them to have some intrinsic representational content.

6 Conclusions

Aristotle’s use of the term φάντασμα is rooted in ancient Greek linguistic practice: it refers to what appears in a quasi-perceptual way in situations that go beyond normal perception, without presupposing any particular explanation of the phenomenon. According to Aristotle’s theory of φαντασία, the subject of perception is presented with a φάντασμα whenever φαντασία occurs as some motions are triggered in the subject’s perceptual system. These motions originate in prior perceptions of sense-objects whose physiological effects on the perceptual system, the relevant αἰσθήματα, have led to the persistence of some similar physiological processes. This physiological feature of remanence in the perceptual system is supposed to explain how the subject of perception may be presented with some perceptual content without being affected by the corresponding sense-objects at the time. Readers of Aristotle should not assume, however, that Aristotle’s theory of φαντασία as a by-product of the activity of sense-perception is implied as soon as they encounter the term φάντασμα, which just picks out the content of awareness in such quasi-sensory situations.

The meaning of φάντασμα in Aristotle comes close to what one could nowadays describe as the content of a mental representation. Translators may, then, feel tempted to take the shortcut of using “(mental) image” or “representation” as its rendering. This little trick is harmless in some contexts, but it becomes devastating as soon as Aristotle focuses on the relationship between a φάντασμα and its causal antecedents or on the complex ways in which subjects may envisage their own φαντάσματα. The notion of “mental representation,” like that of “mental image,” is intentional. In “imaginistic” and “homuncular” language, it conveys the idea that there is an important distinction between what one “sees” with the mind’s eye, namely the “mental image”, and what the image represents: the representation is directly present in the mind, the content is not. Aristotle would certainly agree that picturing a horse does not involve having a literal horse in one’s soul.^[70] However, there is no such duality between a φάντασμα and its content. When one remembers a tune, the φάντασμα is straightforwardly the tune one “hears” or “sings” in one’s head insofar as it “appears”. It is a mental quasi-sense-object that plays a role analogous, not to the modern notion of “mental representation,” but to its content.

The intentionality Aristotle is concerned with whenever he explicitly describes a φάντασμα as a picture happens at the level of the attitude one adopts towards the φάντασμα, and not within the φάντασμα itself. Every φάντασμα bears traces of its causal ancestry in virtue of the way it comes to be (it is in some cases a near-perfect replica of the original sense-object it has as its cause) but experiencing it does not necessarily require envisaging these traces as traces. Such complex attitudes only take place in specific cases, such as dream-interpretation and remembering. Aristotle illustrates them by means of pictorial analogies, not because every φάντασμα is picture-like in that it possesses a representational content, but because, as a copy, a φάντασμα can be envisaged in its relationship with its own model. These intricacies, as well as the force of Aristotle’s pictorial analogies, are obliterated when “image” or “representation” are substituted for φάντασμα: there is nothing tremendously stimulating in comparing an image to a painting or even to a reflection. Conflating the content of a representation with the representation itself makes it near-impossible to talk with some precision about attitudes towards representational content. If philosophical accuracy is to be given priority, I suspect that we should perhaps not settle for these seemingly convenient translations of φάντασμα in Aristotle.^[71]