Experiences of Absence that Get the World Right

Abstract

The topic of absences and their ontological status has long been the focus of intense philosophical debate. Recent years have witnessed the burgeoning of a related discussion concerning the phenomenon of experiencing absences. A lot of this discussion revolves around the question of whether such experiences are best construed as literal perceptions or as some other kind of mental state. Rather than try to settle that ongoing debate, I take as my starting point a claim that seems to be granted by virtually all of the participants in that debate, namely, that experiences of absence are capable of representing reality accurately and of misrepresenting reality. But if they can represent reality accurately, they can do so in a manner that is merely a lucky coincidence and they can do so in a way that is noncoincidental, and I offer reasons for thinking that the latter is more valuable than the former. The burden of this paper, then, is to try to offer an account of the conditions under which absence experiences can be noncoincidentally accurate representations of reality – something that only one other author in the current literature has thus far attempted. To begin with, Section 1 outlines various kinds of experiences of absence and singles out the kind that will be my focus throughout the rest of the paper. In Section 2, I survey the current debate over whether experiences of absence are best construed as perceptual or as something else, and I outline several assumptions I shall be making in the remainder of the paper. Section 3 motivates the project of trying to understand the conditions in which experiences of absence are noncoincidentally accurate, and Sections 4–5 develop an account of those conditions. Finally, in Section 6, I discuss the connections between experiences of absence and justifiedly believing and knowing that a given object is absent from a given location.

1. Varieties of absence experience

Roy Sorensen’s writings (2008) are a treasure trove of examples of experiencing absences. Among the various cases he discusses are ones that, so he argues, involve visual perception of shadows, black holes, silhouettes, and gaps. More controversially he contends that it is possible to see total darkness and hear perfect silence. Sorensen (2013, 2015) also discusses cases which appear to involve the expectations or memories of the percipient, such as the case of the Parisians who queued up seemingly to see the absence of the Mona Lisa shortly after its theft from the Louvre in 1911 or that of the visitors to Ground Zero who apparently came to view the absence of the Twin Towers.

These various cases can be grouped fairly naturally into three categories. Firstly, there are those cases involving perceptual contact with shadows, holes, gaps, silhouettes, black holes, and the like, amidst a scene populated with positive entities. Although they emit no light or sound or any other form of energy, if such “negative entities” as these really are perceived, it is plausibly because of the contrast between them and their energy-emitting surroundings.¹ Secondly, there are those cases in which the percipient registers no stimulus at all via a certain perceptual modality (sight, hearing, etc.), despite being suitably sensitive to stimuli (light, sound) via the modality in question.² Thirdly, there are those cases in which the experience of absence seems to depend upon the subject’s having had some prior idea of a given object O’s association with a given location L, be that a memory of O previously being in L, an expectation that O will be in L, or a desire for O to be in L. Perhaps the most celebrated case of this third variety is one described by Jean-Paul Sartre in a passage in Being and Nothingness:

I have an appointment with Pierre at four o’clock. I arrive at the café a quarter of an hour late. Pierre is always punctual. Will he have waited for me? I look at the room, the patrons, and I say, “He is not here.” Is there an intuition of Pierre’s absence, or does negation indeed enter in only with judgment? At first sight, it seems absurd to speak here of intuition since to be exact there could not be an intuition of nothing and since the absence of Pierre is this nothing. Popular consciousness, however, bears witness to this intuition. Do we not say, for example, “I suddenly saw that he was not there”? (Sartre 2003 [1943]: 33)

Sorensen’s cases of “absence tourists” flocking to see the place where the Twin Towers once stood or the empty wall on which the Mona Lisa had previously hung would seem to belong in the same category as Sartre’s café experience. What unites these cases is that the experience of absence depends on the subject’s already having entertained an association between that location and the missing object. For convenience, we can refer to such cases as Sartrean experiences. It is this third category that will be my focus for the remainder of the paper.

2. Do Sartrean experiences really involve perception of absences?

The past few years have witnessed a lively debate around this intriguing question. Broadly speaking the views that have emerged in the literature can be grouped into three: (1) perceptual accounts, according to which, yes, absences are literally perceived during Sartrean experiences; (2) cognitive accounts, according to which Sartrean experiences involve the perception of present objects plus an inference to the belief that something is missing; (3) accounts on which absences are not literally perceived, but according to which the mental state that represents an object as missing is something other than a belief. For want of a better term, we can call these third-way accounts.

To get a better handle on what is being claimed by perceptual accounts, it is useful to recall Fred Dretske’s (1969) distinction between epistemic and nonepistemic perception. Epistemic perception involves perceiving that p, for some proposition p – for example, seeing that the sun is shining or hearing that a train is approaching. Epistemic perception involves coming to know the truth of a proposition via perception. Nonepistemic perception, by contrast, involves perceiving an object O without necessarily having any beliefs about O, though having such beliefs is compatible with nonepistemic perception. For Dretske, nonepistemically seeing O means O appears in some way to you as distinct from its surroundings. You might nonepistemically see an ammonite despite not believing it is an ammonite or nonepistemically hear bison while mistaking them for horses. It is fairly uncontroversial to claim that we can epistemically perceive that p for many propositions p which are about absences – for example, that someone can see that the Twin Towers in New York are gone. What is far more contentious is the claim made by perceptual accounts of Sartrean experiences: namely, that nonepistemic perception of absences can occur – for example, that someone can literally see the absence of the Twin Towers. Anna Farennikova (2013, 2015, 2018a, 2018b), Roy Sorensen (2008, 2013, 2018, 2015), and Dan Cavedon-Taylor (2017) all argue for perceptual accounts of Sartrean experiences. Their arguments include the following: (i) that Sartrean experiences have a felt immediacy that is best explained by their involving nonepistemic perception of absences, as opposed to involving inferences from perception of positive objects; (ii) that conceptually unsophisticated creatures (human infants, nonhuman animals) are evidently able to detect absences in an instinctive fashion, which is again held to be best explained by there being nonepistemic perception of absences; (iii) that one can experience the absence of something despite believing that it is present, such as when the North Tower of the World Trade Center was attacked by terrorists in 1993 and lost its electricity supply so that at night it appeared absent to New Yorkers despite their knowing (and hence believing) that it was still standing; (iv) that tourists sometimes flock to locations from which a famous object (e.g. the Mona Lisa, the Twin Towers) has disappeared, and that this so-called “absence tourism” is hard to make sense of unless the tourists are looking to have nonepistemic perception of absences, given that the tourists already knew that the object in question is missing. Farennikova offers a further argument that relies on her own particular account of Sartrean experiences. On her account, Sartrean experiences involve a “mismatch” between incoming perceptual information and working memory, in virtue of the way that the latter mechanism “projects” a perceptual template of the object that the person was expecting to be present in the scene at hand. She contends that,

The two elements related via a mismatch structure are a visual representation of the missing object and positive representation of the world. Given that both representations are perceptual in format, and the comparing process occurs in perception, it seems plausible to regard the entire mechanism as perceptual. (2018b: 147)

Against (i), (ii), and (iii), it can be pointed out that whilst the various data alluded to are hard to square with a cognitive account of Sartrean experiences, they can be quite happily explained on an account on which the mental state that represents the scene as missing something is not a belief but also is not perceptual – for example, an intellectual seeming (more on this suggestion shortly).³ Against (iv), positing literal absence perception is hardly the only way to make sense of the behaviour of absence tourists. Arguably, a hypothesis that is at least as plausible is that the tourists simply desire to have first-hand experience of the place where a famous object used to be located. More broadly, a standard objection against perceptual accounts is that they require a dubious ontological commitment to the reality of absences and, moreover, to the causal efficacy of absences. At any rate, the latter commitment is required unless one is willing to give up on the widely held view that genuine perception requires having causal contact with that which is perceived.⁴ Another worry about perceptual accounts is that they posit a difference in perceptual phenomenology where there isn’t obviously one. For example, consider Farennikova’s case in which someone leaves her laptop on a table in a coffee shop in order to step outside for some fresh air and when she returns to her table she is confronted with the shocking scene of an empty tabletop (2013: 430). Compare this with a variant case in which the same person momentarily leaves the same coffee shop table to step outside for air but instead takes her laptop in her backpack and so isn’t at all surprised when she returns to find an empty tabletop. Perceptual accounts of Sartrean experiences are committed to the idea that there is a difference in visual phenomenology between these two cases; that in the first case, but not in the second case, there is literally a way that the laptop’s absence looks to the person returning to her table. It is not easy to discern what this alleged difference in visual phenomenology might amount to. At any rate, positing such a difference entails the controversial view that visual perception is capable of being richly representational (see, e.g., Siegel 2011).

The antithesis of perceptual accounts are cognitive accounts, on which absences cannot be nonepistemically perceived; rather, in Sartrean experiences perception of positive objects prompts one to draw an inference to the belief that something is missing. There are in fact very few defenders of thoroughgoing cognitive accounts of absence experience. Brian O’Shaughnessy (2000: 333) may be one of the few. The chief advantage of cognitive accounts is that they avoid the contentious ontological and phenomenological commitments that perceptual accounts involve. A key objection to cognitive accounts is that they fail to do justice to the felt immediacy of Sartrean experiences. Another objection, one that Laura Gow views as “decisive,” is that

[A]bsence experience exhibits resilience to a change of belief. If you are told that your laptop is still on the table and a magician is playing an elaborate trick with mirrors, you will still experience the absence of your laptop even though you now believe that your laptop is not absent. A proponent of the belief/judgement version of the cognitive view must therefore allow that a subject can simultaneously (and consciously) believe or judge p and not p. (Gow 2021: 176)

Jean-Rémy Martin and Jérôme Dokic’s (2013) view can be seen as a third-way account, in that it denies that absences are perceived but also denies that the mental state at the heart of Sartrean experiences is a belief. Rather, they hold that Sartrean experiences involve the perception of positive objects plus a feeling of surprise, which is occasioned by virtue of one’s expectation that a given object would be present. The difficulty is that not all Sartrean experiences involve surprise. Cavedon-Taylor’s (2017) Dentist case involves the tactual feeling of one’s missing tooth, which one fully expected to be missing after one’s dental procedure.⁵ Gow (2021: 175) also worries that this account fails to capture the difference in phenomenology (not necessarily perceptual phenomenology) between a case in which one is searching for and failing to find one’s wallet and a case in which one is searching for and failing to find one’s wallet and one’s keys. In short, the phenomenology of all Sartrean experiences is flattened by Martin and Dokic’s account into a single undifferentiated experience of surprise.

To my mind, the two accounts that strike the best balance between ontological caution and doing justice to the phenomenology of Sartrean experiences are the third-way accounts of Gow (2021) and Clare Mac Cumhaill (2018). Gow’s account posits that Sartrean experiences involve the perception of positive objects and an accompanying intellectual seeming whose content is that the thing in question is missing. An intellectual seeming is crucially different from a belief, in that intellectual seemings have a felt quality and in that it can intellectually seem to you that p even when you know (and hence believe) that not-p. For example, it can intellectually seem to you that for any possible combination of objects, there must be a corresponding set, even though you are fully aware of Russell’s paradox and accept the standard lessons drawn from it. Gow contends that her account can make sense of the differences in phenomenology between different Sartrean experiences because the content of your intellectual seemings differs between cases: in one case, for example, you have the intellectual seeming that your wallet is absent, and in another, the intellectual seeming that your wallet and keys are absent. Crucially, though, the difference in phenomenology is not held to be perceptual. Perhaps the overriding advantage of Gow’s view is that intellectual seemings are representational and can thus be accurate (or inaccurate), and yet no dubious ontological or phenomenological commitments are required. As Gow puts it, on her view “one’s absence experiences can get the world right even though absences do not exist” (2021: 172).

In a not-dissimilar vein, Mac Cumhaill (2018) suggests that Sartrean experiences involve an experience of a “figureless ground.” In ordinary perceptual experience an object which is the focus of your attention is the figure and the object’s surroundings are the ground. In Sartrean experiences, because you expected (or in some other way had the idea of) O’s presence in a given location, when O turns out not to be there you experience that location as a mere ground upon which O should have appeared as figure but has failed to do so. Crucially, on this account one does not perceive absences themselves; “one only perceives absential locations the bounds of which are circumscribed by reference to attitudes and intentions” (2018: 44). The view, in effect, is that your prior mental states (expectations, desires, memories, etc.) cause you to see a present scene as an “absential location,” that is, a location from which O is absent. Similar to Gow’s view, experiencing a scene as a location from which O is absent is compatible with your believing that O is in fact present there. This is because seeing x as F does not entail believing that x is F. For example, one can see one line as longer than the other despite believing that really the two lines are of equal length.

In what follows I shall remain agnostic on the question of whether Sartrean experiences involve genuine nonepistemic perception of absences, although I have indicated that I am doubtful about that – my sympathies lie with certain accounts of the third way variety. There are, however, a few important assumptions that I shall be making in what follows, namely:

• • Sartrean experiences possess a felt quality of immediacy. They are not plausibly construed as involving inferences, unless the threshold for what constitutes an inference is set very low.

• • The possibility of having a Sartrean experience depends upon your already having some representational mental state that associates the missing object O with a location L – be that mental state an expectation of O’s being in L, a memory of O’s being in L in the past, a desire for O to be in L, or whatever.⁶

• • Even if absences don’t literally exist and Sartrean experiences don’t literally involve nonepistemic perception of absences, Sartrean experiences can nevertheless “get the world right” and can fail to do so. That is, Sartrean experiences represent the world as being a certain way and can therefore be assessed for accuracy. Again, I am not assuming that the mental state doing the representing is a perception.

3. Conditions for noncoincidentally accurate Sartrean experiences

All three of the families of accounts surveyed in the previous section agree that Sartrean experiences represent the world as being a certain way – specifically, they represent a given location as missing a particular object – and can thus be accurate or inaccurate. The disagreement is over what kind of mental state does the representing. Given the apparent consensus that Sartrean experiences are representational and thus assessable for accuracy, it is perhaps surprising that only one author in this literature has sought to investigate the conditions under which Sartrean experiences are noncoincidentally accurate, namely, Roy Sorensen (see especially Sorensen 2013). I shall argue shortly that Sorensen’s account is not fully adequate, but nevertheless it provides a valuable jumping-off point for our discussion.

Before going further, it is worth seeing why advocates of all three families of accounts of Sartrean experiences surveyed in Section 2 have strong reasons to care about the question of what is needed for a Sartrean experience to represent the world in a noncoincidentally accurate manner. First, consider cognitive accounts, on which it is a belief that something is absent which is doing the representing. Beliefs, of course, can be true or false. Since at least as far back as Plato’s Theaetetus (1987: 201c–10b), however, it has been recognized that mere true belief is not sufficient for knowledge. Knowledge involves true belief that is related in a certain way to the world, such that the belief has a secure “tether” to the truth. There is a widespread intuition that knowledge is more valuable than mere true belief, albeit that there is disagreement over how best to account for this.⁷ Suffice it to say, cognitivists about Sartrean experiences have excellent reasons for being interested in what it takes for the beliefs that they see as being at the heart of Sartrean experiences to be noncoincidentally true.

As for perceptualists like Sorensen and Farennikova, they too have strong reasons to care about the noncoincidental accuracy of Sartrean experiences. A considerable amount of work has been done in the past few decades on the question of what exactly distinguishes genuine veridical perception from veridical hallucination.⁸ An example of veridical hallucination would be a case in which you take psychoactive mushrooms which temporarily blind you and cause you to randomly hallucinate an elephant in front of you, and it just so happens that there is an elephant in front of you. Your perceptual experience accurately resembles the scene in front of you but its doing so is a lucky fluke – you do not genuinely perceive the elephant. In this way, perception is analogous to knowledge, and it is similarly intuitive that genuine veridical perception is more valuable than veridical hallucination.

As for third-way-ers about Sartrean experiences, at least the two accounts for which I indicated sympathy make it very natural to care about the noncoincidental accuracy of Sartrean experiences. Again, Gow is emphatic that these experiences “can get the world right” (2021: 172); it is natural to wonder about the conditions under which they get the world right noncoincidentally. Similarly, Mac Cumhaill’s notion of a location being seen as an absential location (a location from which O is absent) seems to be assessable for accuracy and thus for noncoincidental accuracy. In general, one accurately sees x as F only if x is indeed F. Intellectual seemings and seeing-as bear important analogies to both knowledge and straightforward perception, in that it is intuitively plausible that it is valuable for a seeming state or an episode of seeing-as to accurately represent the way the world is, but it is better still for it to do so in a way that manifests a secure tether to reality.

The following are five cases of Sartrean experience. All of them get the world right and, moreover, do so in a manner that seems noncoincidental. I shall give the cases labels to make it easier to refer to them again later.

#1 (Twin Towers). Tom moved away from New York City in 2000 and returned again in 2002. His visual memory of the Manhattan skyline as it was in 2000 is still vivid, and as he walks across the Brooklyn Bridge towards Manhattan for the first time in two years he is mesmerized by the gap in the skyline where the Twin Towers once stood.

#2 (Classroom). Ryan is a notorious troublemaker who regularly aggravates other boys in his class and occasionally starts fistfights with them. Unfortunately for the class teacher, Ryan almost never misses school. Glancing over at Ryan’s desk one morning as the students are opening their textbooks, the teacher notices Ryan’s absence and is immensely relieved.

#3 (Santa). Like many children her age, Alice believes in Santa Claus. One Christmas Eve, in the middle of the night, Alice hears the sound of jingling bells coming from the direction of the front yard and, overwhelmed with excitement at the prospect of seeing Santa and his sleigh, rushes to her bedroom window to look out into the front yard. Santa and his sleigh are nowhere to be seen. Alice is deeply saddened by their absence.

#4 (Dream House). Charlie has inherited a plot of land from his father and has spent years saving up in the hope of building his dream house on it. He spends his idle moments visualizing his dream house in great detail. Whenever Charlie drives past the plot the absence of his dream house is achingly apparent to him.

#5 (Portraits). Jane is walking down a long hallway gazing at the portraits which are hung at equal intervals and is surprised when she encounters a gap where it seems there ought to be another portrait.

Each of these cases involves a Sartrean experience wherein the subject S had already entertained some association between a given location L and an object O. The nature of that association is varied across the cases. In each case, the experience represents O as absent (again, I am not committed to the view that it is perception that represents the absence), and O is indeed absent. What is more, in none of these cases is there any deviant causal mechanism or any other factor that would seem to render the accuracy of the Sartrean experience a mere lucky coincidence. At this point, we can introduce our first condition on noncoincidental accuracy, which is simply the accuracy condition:

The accuracy condition: S has a representational mental state that represents a location L as missing an object O, and O is indeed missing from L.

I leave it open what the representational mental state is – whether it is a perception, a belief, or something else, such an intellectual seeming.

4. A causal condition on perception of the present scene

Consider two further cases:

#6 (Moggy). Sally’s cat Moggy spends the vast majority of his time curled up asleep on the sofa in the lounge. Sally walks into the lounge one evening fully expecting to see Moggy and is taken aback when his usual spot appears empty. Moggy is in fact there; it’s just that in the dim light he looks exactly like one of the fluffy cushions that adorn the sofa.

#7 (Ornament). On the plain white wall of Bill’s bedroom hangs an expensive ornament. One night while Bill is sleeping a thief steals the ornament. When he wakes up and opens his eyes Bill is struck by the absence of the ornament. Unbeknownst to Bill, just before he woke Bill developed a blind spot in the middle of his field of vision (which gives the impression of looking at a plain white surface), the very portion of his field of vision that would, once Bill opened his eyes, have been occupied by the ornament if it had still been there.

In case #6, Sally’s Sartrean experience fails to get the world right; it doesn’t meet the accuracy condition. Moggy is in fact there in the scene in front of her. In case #7, Bill’s experience does get the world right – the ornament is indeed gone – but it is a lucky fluke that it does so, given the facts about his blind spot occurring in just the place where the ornament would have been. This case is an analogue of veridical hallucination (or, according to perceptualists about Sartrean experiences, it is a case of veridical hallucination). Let us turn to look, then, at the only extant account of the noncoincidental accuracy of Sartrean experiences, namely, that of Sorensen (2013).

It isn’t clear that Sorensen means to offer necessary and sufficient conditions for noncoincidental accuracy, but we can discern the shape of his account, nonetheless. Sorensen contends that “You see the absence of something only if you occupy a perspective from which you would have seen it” (2013: 109). This counterfactual requirement is intended to express the thought that having a noncoincidentally accurate Sartrean experience requires that one is appropriately perceptually sensitive to the missing object O. Sorensen’s counterfactual requirement is met in all of cases #1–5. In each of these cases, it is true that the subject occupies a perspective from which he or she would have seen the object in question if it had been present. Sorensen’s account can also explain why the Sartrean experiences in cases #6 and #7 fail to be noncoincidentally accurate. In case #6 (Moggy), the cat is present and yet Sally fails to see him. It follows that if Moggy were present, Sally would fail to see him.⁹ Sorensen’s counterfactual requirement is not met. In case #7 (Ornament), assuming that Bill’s visual blind spot is properly thought of as a feature of the perspective that he occupies, Bill wouldn’t have seen the ornament if it had still been hanging on the wall when he woke up, given that perspective. Once again, Sorensen’s counterfactual requirement isn’t met.

There is a kind of case with which Sorensen’s account struggles, however. It involves what is plausibly taken to be a noncoincidentally accurate Sartrean experience, despite the fact that S wouldn’t have seen O if it were present:

#8 (Grandfather clock). During the night a thief breaks into Lucy’s house and steals her beloved antique grandfather clock. When Lucy comes down for breakfast in the morning she is stunned to find it gone. But if the thief hadn’t stolen the clock, he would have placed a highly convincing three-dimensional screen in front of the grandfather clock, making it appear as though the grandfather clock was missing whilst all its surroundings remained intact.

Lucy’s Sartrean experience fails to meet Sorensen’s counterfactual requirement: Lucy occupies a perceptive such that if the grandfather clock had been present then Lucy would not have seen it because it would have been behind the trick 3D screen.¹⁰ My judgment is that Lucy’s Sartrean experience is noncoincidentally accurate, however. I anticipate that many will share this judgment. Case #8 is an analogue of the pre-emption cases in the literature on causation (see Lewis 1986a: 193–212). Suppose that if John Wilkes Booth had not fired the shot that killed Abraham Lincoln then a waiting accomplice of Booth’s would have fired a fatal shot. Given such a setup, it would be false to say that Lincoln would not have been shot dead had Booth not fired. And yet Booth did cause Lincoln’s death. This strongly suggests the falsity of an account on which A causes B only if B wouldn’t have occurred if A hadn’t occurred. That is, pre-emption cases like this have generally been taken to render a simple counterfactual account of causation unviable. A similar lesson applies to absence experience.

Sorensen is no doubt aware of the issue of pre-emption and if pressed would likely offer refinements of his counterfactual requirement to try to deal with cases such as #8 into account. I want to propose, however, that rather than try to patch up the counterfactual requirement we take a different, arguably more straightforward tack: namely, appealing to a causal condition on the subject’s perceptual experiences of the positive entities in the scene before her. Given that perception in any case very plausibly requires causal contact with its objects, it is worth exploring whether we can capture what goes wrong in case #7 (Ornament) and what goes right in case #8 (Grandfather clock) by appealing to a causal condition on one’s perception of the positive entities that comprise the absential location L. The condition I have in mind can be stated roughly thus:

The causal condition: the positive entities present in L nondeviantly cause S to have an accurate perceptual experience of L.

Clearly, a lot is riding on the notion of “non-deviant” causation. One of the principle motivations for the development of a causal condition on perception was the desire to rule out veridical hallucinations from being counted as instances of perception (see, e.g., Grice and White (1961), Goldman (1977), Lewis (1986b)). As a starting point, a very straightforward causal condition on perception would require that one perceives an object O only if O enters into the causal chain that results in a perceptual experience that accurately resembles O. The earlier case of veridical hallucination, in which psychoactive mushrooms temporarily blind you and cause you to hallucinate an elephant just when an elephant happens to be in front of you, would be deemed not to be an instance of genuine perception by the lights of this straightforward causal condition. Put simply, the elephant in no way enters into the causal chain that results in your perceptual experience of an elephant.

There are, however, more challenging cases of veridical hallucination. Valtteri Arstila and Kalle Pihlainen present the following case:

[Y]ou come across an Indian Cobra while travelling in Asia. The cobra spits its venom into your eyes, forcing you to close them. The poison makes you ill and leads you to have a visual experience of a cobra in front of you. (2009: 400)

The Indian cobra does enter into the causal chain that results in your perceptual experience of an Indian cobra, and yet we surely don’t want to say that you genuinely perceive the cobra. A straightforward causal condition is therefore not adequate to the task of distinguishing genuine perception from veridical hallucination. How, then, should we deal with cases in which the hallucination is caused by an object that closely resembles the contents of the hallucination? A natural thought is that the ease with which a given causal pathway could have resulted in an inaccurate perceptual experience has something crucial to do with it.¹¹ In the cobra case, the causal pathway involving the cobra’s venom is one that could very easily have resulted in your having a perceptual experience that did not resemble the actual scene in front of you. For example, the venom could presumably quite easily have caused you to hallucinate something else, such as a donkey or an eagle; or, it could easily have been the case that the cobra slithered away immediately after spitting in your eyes, such that your perceptual experience of a cobra in front of you would have failed to resemble the actual scene in front of you. A nondeviant causal pathway might thus be understood as a pathway that could not easily have led to an inaccurate perceptual experience. This appeal to what could or could not easily have happened is of course a modal consideration.¹² We can state the causal condition thus refined as follows:

The modal–causal condition: the positive entities present in L cause S to have perceptual experience of L that accurately resembles those entities, and the causal pathway running from L to S’s experience of L is such that not easily could that pathway have resulted in S having perceptual experience that failed to accurately resemble the objects comprising L.

There is one further kind of case, however, which is thought to make trouble even for this modal–causal condition. Alva Noë (2003) has discussed a case in which a manipulative neurosurgeon causes you to have a visual experience of a clock on the mantlepiece before you, when there is in fact a clock on the mantlepiece before you. The crucial twist is that the neurosurgeon causes you to have this clock-on-the-mantlepiece visual experience because there is a clock on the mantlepiece. We can add that if the clock were different than it actually is (e.g., hands pointing to a different time, being a different colour, etc.) then the neurosurgeon would have given you a correspondingly different visual experience (see Arstila and Pihlainen 2009: 402). In short, the worry is that this causal pathway involving the manipulative neurosurgeon could not easily have resulted in an inaccurate visual experience of the scene before you, and yet we are uneasy with classifying it as a case of genuine perception. Noë contrasts this case with a case in which an angel is causing visual experiences in a blind person and “is committed to maintaining these regularities [i.e., regularities pertaining to the resemblance between the visual experience and the actual scene in front of the person] come Hell or high water” (2003: 98). Noë judges that whilst unusual, this “divine prosthetic vision” would be a genuine instance of perception (2003: 98). Arstila and Pihlainen concur with this judgment, and this moves them to elaborate their causal account of perception with the requirement that the causal pathway is “not susceptible to random manipulation” (2009: 413). That the causal pathway be free from such manipulation seems like a plausible refinement of the causal condition. However, I’m inclined to think that this point is in fact already implicit in the aforementioned modal requirement: that the causal pathway could not easily have resulted in an inaccurate perceptual experience. Look at it like this: if the neurosurgeon has a rock-solid resolve to induce only accurate perceptual experiences in you, then this case begins to look a lot like Noë’s angel case and thus perhaps a case of genuine (albeit prosthetic) visual perception; if, on the other hand, the neurosurgeon is whimsical, such that he could somewhat easily have decided to give you an inaccurate perceptual experience, then the causal pathway will not satisfy the modal–causal condition.¹³

Given the modal–causal condition, we can explain why case #8 (Grandfather clock) is a noncoincidentally accurate Sartrean experience whereas case #7 (Ornament) is not. In case #8, Lucy’s perceptual experience of the absential location – that is, the hallway from which the grandfather clock is missing – satisfies the modal–causal condition: the positive entities comprising the hallway cause Lucy to have a visual experience accurately resembling those items, and the causal pathway running from those objects to Lucy’s visual experience could not easily have resulted in Lucy’s having a visual experience that failed to accurately resemble those items.¹⁴ Put another way, Lucy has successful perceptual contact with the location from which the grandfather clock is missing, and this is irrespective of the fact that if the grandfather clock had still been present it would have been invisible to her owing to the trick 3D screen. By contrast, Bill’s visual experience of a white patch of wall where his favourite ornament used to hang does not satisfy the modal–causal condition, because the causal pathway that generated that experience – the one that involves a blind spot in Bill’s vision that causes him to see a plain white patch that luckily happens to match the colour of the wall – is one that could very easily have resulted in inaccurate visual experience of the scene in front of him.

A final clarification regarding the modal–causal condition is in order before we move on. I think we ought to say that the accuracy/inaccuracy that is relevant to the modal–causal condition pertains not only to the way things visually look to S with respect to colour, shape, spatial properties, and so on but also to the way that S conceptualizes or recognizes the contents of the visual field – for example, to the way that a given white round shape looks to S like a cat or a cushion. (To be clear, I am not taking any position on whether such conceptual recognition is part of the content of perception itself or is rather a cognitive response to the raw data of perception.) A variant of case #6 (Moggy) illustrates why this qualification is needed. Suppose that Moggy really is absent from the sofa, but as in the original case, the dim lighting is such that Moggy would look just like a cushion to Sally if he were present on the sofa. I would judge this not to qualify as a noncoincidentally accurate Sartrean experience. But in order to vindicate this judgment, the modal–causal condition needs to be understood as having in view not just Sally’s visual experience of the shapes, colours, and spatial properties of the things in front of her, with respect to which she is accurate and would still be accurate if Moggy were present; it needs to be understood also as pertaining to her conceptualization of the objects in front of her, with respect to which Sally would not be accurate if Moggy were on the sofa – Moggy would look to Sally like a cushion rather than a cat. Given this understanding of the modal–causal condition, the causal pathway that leads from the objects in front of Sally to Sally’s visual experience could easily have resulted in the inaccuracy of a relevant sort.

5. A modal–causal condition on one’s mental image of the missing object

So far, then, I have suggested that in order to qualify as noncoincidentally accurate, a Sartrean experience must meet an accuracy condition: the object that is represented as missing from the relevant location L really is missing from L. (And again I am not assuming that it is perception that plays the representational role here.) It must also meet a modal–causal condition, as discussed in the previous section. But there are cases that meet these conditions and yet intuitively fail to constitute noncoincidentally accurate Sartrean experiences.

#9 (Mona Lisa). Amelie hears the news that the Mona Lisa has been stolen from the Louvre. Amelie has never seen the Mona Lisa herself, not even in a reproduced form. Her father Pierre, however, has described the Mona Lisa to her. What he has told her is that the Mona Lisa is a life-sized marble sculpture of an old woman who is hunched over. Amelie heads to the Louvre and patiently lines up to get into the Salon Carré. When she finally gets to the head of the line, she gazes intently at the cordoned-off space in the gallery that was once occupied by the Mona Lisa.

#10 (Twin-New York). Josh lives in Twin New York, a city extremely similar to our New York, located on a faraway earth-like planet. The only readily noticeable thing that is different about Twin-New York is that its counterparts of the Twin Towers are still standing, whereas our New York’s Twin Towers were destroyed and (let’s suppose) haven’t yet been replaced by any other buildings. One night Josh gets kidnapped in his sleep and transported across space to our New York. The next morning, he wakes up unaware that anything has changed. When he ventures out to a location from which the Twin Towers used to be visible, he is shocked by the gaping hole in the skyline.

In case #9, Amelie has a mental state that correctly represents an object as missing from the relevant spot in the Salon Carré, and there is nothing awry with Amelie’s perceptual contact with that location. There is no need to doubt that she is experiencing the absence of something and is doing so in a noncoincidentally accurate manner. But it seems odd to say that she is experiencing the absence of the Mona Lisa, considering that her mental image of what she calls “the Mona Lisa” is so unlike the real Mona Lisa. As for case #10, one might think of this as a kind of Gettier case for Sartrean experience: it is a sheer coincidence that Josh’s mental image is of a pair of buildings that look just like the ones that used to be present in the scene before him. Again, we need not doubt that Josh noncoincidentally accurately experiences the absence of something (a pair of buildings; the Twin-Twin Towers, perhaps), but it isn’t the Twin Towers’ absence which he experiences. Cases like these strongly suggest the need for a further condition on noncoincidental accuracy for Sartrean experiences, one that links S’s mental image of O (i.e., S’s idea of what O would look like if it were present) to O itself by way of a certain sort of causal pathway. I propose the following condition, which is an analogue of the modal–causal condition on S’s perceptual contact with the absential location:

The mental image condition: O caused S to have a mental image that accurately resembles O, and the causal pathway running from O to S’s mental image is such that not easily could that pathway have resulted in S’s having a mental image that failed to accurately resemble O.

An important issue raised by this condition is that in some cases of Sartrean experience, including some of the cases considered earlier in the paper (#3, #4, #5), O is an object that never existed in the first place. This throws up a question about what sort of connection between O and S’s mental image of O we should be looking for. Broadly, we can divide cases of Sartrean experience into two classes: those involving an experience of the absence of a particular thing (of the F, where “the F” is a definite description, or of a, where “a” is a proper name); and those involving an experience of the absence of an unspecified token (or tokens) of some type (of an O or some Os).

Consider cases of the former kind. Within it, there are cases in which the O in question used to exist but no longer does. Case #1 (Twin Towers) is a paradigm of this. In cases like this, the facts about what O would be like if it were present are presumably grounded in the facts about what O was like when it used to exist. There are also cases in which the O still exists but is missing from a location in which it would normally be found (at least, at certain times). Case #2 (Classroom) is a paradigm. The facts that ground truths about what Ryan would be like if he were present in class today are those facts about what he is like right now. Then, there are cases in which O never really existed in the first place. Some such cases involve distinctively fictional characters. Case #3 (Santa) is one such case. There is a sense in which some statements about Santa Claus are true, and these fictional truths are grounded in actually existing features of the world, namely, the content of the various fictional writings, films, illustrations, and so on, that have grown up around the character of Santa.¹⁵ On the other hand, there are cases in which the O is entirely a figment of S’s imagination, such as case 4 (Dream House). In these sorts of cases, what grounds truths about what O would be like if it were present are simply those facts about how S imagines O to be. This explains why there isn’t a meaningful sense in which Charlie’s visual image of his dream house could fail to be an accurate representation of it: his dream house would be like whatever his imagination depicts it to be like. As for cases like #5 (Portraits), the Sartrean experience is of the absence of an unspecified token of a type. In this case, the type in question is portraits of a certain size and style. The facts that ground the truth about what a portrait would be like were one present in the gap in the row of portraits are presumably those facts about what the portraits that actually are present are like and about any patterns they exemplify. In light of these points, the mental image condition can be adjusted as follows, where the expression “the O-facts” stands for the facts in virtue of which it is true that O would have been a certain way if it had existed and been present in L:

The mental image condition: the O-facts caused S to have a mental image that accurately resembles O, and the causal pathway running from the O-facts to S’s mental image is such that not easily could that pathway have resulted in S’s having a mental image that failed to accurately resemble O.

Returning to cases #9 (Mona Lisa) and #10 (Twin-New York), which motivated the inclusion of the mental image condition, it is worth seeing briefly how this condition deals with these cases. Again, to be clear, both cases #9 and #10 are noncoincidentally accurate experiences of the absence of something, but not of the absence of the Mona Lisa or of the Twin Towers, respectively. In case #9, Amelie’s mental image of what she calls “the Mona Lisa” does not resemble the Mona Lisa closely enough for the mental image condition to be met where O is the Mona Lisa. In case #10, when Josh looks at the gap in the skyline where the Twin Towers once stood, he has in mind a mental image of a pair of buildings that he had expected to be there, but that mental image was in fact formed as a result of past visual contact with an entirely different pair of buildings – the Twin-Twin Towers – in an entirely different place – Twin-New York. Unlike Tom’s mental image of the Twin Towers in case #1 (Twin Towers), Josh’s mental image was not the result of a causal pathway that goes back to the Twin Towers, but rather, to the Twin-Twin Towers. Now, clearly Josh’s past visual contact with the Twin-Twin Towers has – by a hugely lucky coincidence – caused him to have a mental image that does closely resemble the Twin Towers. But that is why the second clause of the mental image condition is needed: given the causal pathway via which Josh formed his mental image (namely, past visual contact with the Twin-Twin Towers), Josh’s mental image could easily have failed to resemble the Twin Towers. After all, the Twin Towers could have been somewhat different than they actually were – for example, they could have been half as tall or dark blue or clad in smooth glass rather than unsightly aluminium columns – and yet Josh’s mental image would not have been correspondingly different.¹⁶ Hence, Josh’s mental image fails to satisfy the mental image condition where O is the Twin Towers.

6. Justifiedly believing and knowing that something is absent from a given location

Consider now a case that meets all the conditions proposed thus far, and yet intuitively has something wrong with it:

#11 (Fake New York). For her birthday Lisa is taken by her friend to what she believes to be New York and she immediately recognizes the portion of the skyline that (so she believes) used to be occupied by the Twin Towers. Lisa is thoroughly familiar with what the Manhattan skyline looked like at the time when the Twin Towers were still standing, having watched dozens of episodes of Friends and other New York-based TV and films from the 1990s. Surveying the scene, Lisa is mesmerized by the gap in the skyline where (she believes) the Twin Towers once stood. But Lisa is not in fact in New York at all. Her friend couldn’t afford to take her to New York but instead took her to a highly convincing studio lot that gives visitors a realistic experience of what the Manhattan skyline would have looked like in the mid-2000s.

What should be said about such a case? The accuracy condition is met: the Twin Towers are indeed missing from this mock-up of Manhattan. The modal–causal condition is met: Lisa’s visual experience of the studio lot is caused by the objects comprising the lot in such a way that the causal pathway running from those objects to her visual experience could not easily have led her to have an inaccurate visual experience. And the mental image condition is met: her mental image of the Twin Towers was caused ultimately by the Twin Towers themselves, via a causal pathway that could not easily have led her to have a mental image that did not closely resemble the Twin Towers. So, what is wrong? The diagnosis is in fact fairly straightforward. Lisa is having a noncoincidentally accurate experience of the absence of the Twin Towers from the location in front of her. It’s just that she mistakenly believes that that location is New York; she takes herself to be experiencing the absence of the Twin Towers from New York, when she is in fact experiencing their absence from a studio lot. If she were to believe on the basis of this Sartrean experience alone that the Twin Towers are gone from New York, never having heard about the destruction of the Twin Towers in 2001, then Lisa would fail to know that the Twin Towers are gone from New York. It is natural that we are interested in experiencing the absence of something from a particular location L, apprehended under a specific guise – for instance, “Ground Zero” or “the place where the Twin Towers once stood.” More generally, we can say that in order for S to know that she is experiencing the absence of O from a particular location L under a specific guise, S must know that it is L, apprehended under the guise in question, with which she currently has perceptual contact.

It is worth observing that the three different families of accounts of Sartrean experience surveyed in Section 2 will have somewhat different implications for the relationship between Sartrean experience and one’s ability to justifiedly believe or know that something is missing from a given location. For cognitivists about Sartrean experience, the relationship is seemingly quite straightforward. On cognitive accounts, the mental state playing the role of representing O as missing from L on the cognitivist view is a belief. Hence, having a noncoincidentally accurate Sartrean experience entails having a noncoincidentally true belief that O is missing from L. If knowledge just is noncoincidentally true belief,¹⁷ then having a noncoincidentally accurate Sartrean experience of O’s absence from L is identical with knowing that O is absent from L. However, many epistemologists think that knowledge is more than just noncoincidentally true belief. Indeed, it is quite common these days to hold that knowledge also has a no-defeater condition, such that even if S’s belief that p is in fact noncoincidentally true, S fails to know that p if she has acquired some reason to doubt the truth conduciveness of the basis on which she formed her belief that p.¹⁸ This would apply, for example, in a case where S is looking at a red object under normal lighting with good eyesight and forms the belief that the object is red, but where S is told apparently on good authority, though in fact misleadingly, that there is trick red lighting in the room that makes white objects appear red. If this no-defeater view of knowledge is correct, then there will be the possibility of a gap between noncoincidentally accurate Sartrean experience and knowledge, even on the cognitive account of Sartrean experience. That is, given cognitive accounts of Sartrean experience, it will be possible in principle for S to have a noncoincidentally accurate Sartrean experience of O’s absence from L and yet to fail to justifiedly believe and to fail to know that O is absent from L, in virtue of possessing an undercutting defeater.

Both perceptual accounts and third-way accounts (especially Gow’s and Mac Cumhaill’s accounts) allow for the possibility of a gap between noncoincidentally accurate Sartrean experience and justified belief and knowledge. Recall that on neither of these two families of accounts of Sartrean experience is it a belief that plays the role of representing O as missing from L. On both these families of views, but not on cognitive accounts, it will be possible for S to have a noncoincidentally accurate Sartrean experience of O’s absence from L despite S’s believing (even justifiedly believing) that O is really present in L. Indeed, this was suggested earlier to be a strong advantage of both these kinds of accounts over cognitive accounts, considering that cases like this are very plausibly possible. For example, suppose that in a variant of case #8 (Grandfather clock), Lucy has been informed apparently on good authority, albeit misleadingly, that the intruder placed a trick 3D screen in front of the grandfather clock that makes it appear as though the grandfather clock is missing from the hallway even though it is present. In actual fact the home intruder did not set up a trick 3D screen; he simply stole the grandfather clock, and Lucy now has successful perceptual contact with the location from which it was stolen. In such a case Lucy will justifiedly believe that the grandfather clock is really still present in the hallway, and yet she will have a Sartrean experience of its absence which will qualify as noncoincidentally accurate. If instead, Lucy persisted in her belief that the grandfather clock is missing from her hallway, in spite of having been told that the thief set up a 3D screen that would misleadingly make it look as though the grandfather clock were missing, then she would have a defeater for her belief that the grandfather clock is absent; her belief would not be justified. And yet, her Sartrean experience of the grandfather clock’s absence would still be noncoincidentally accurate. Clearly what is not possible, however, is to have a noncoincidentally accurate Sartrean experience of O’s absence from L and to know that O is present in L, given that knowledge is factive.

It is worth noting that there are broadly two sorts of undercutting defeater that S might acquire for her Sartrean experience-based belief that O is absent from L. The first sort consists of some reason to think that it is not L with which S has perceptual contact, but some distinct location L*. Case #11 (Fake New York) would be an example of this sort of undercutting defeater. The second sort of undercutting defeater consists of some reason for doubting that the apparently missing object would be visually detectable by you if it were present. An example would be a variant of case #8 (Grandfather clock), in which Lucy is told apparently on good authority that the home intruder placed trick 3D screens in front of various objects in her house – though which objects is not specified – making them appear to be missing when they are in fact still there. In such a case, Lucy would have reason to doubt the truth-conduciveness of any Sartrean experience she might have whilst inspecting her house after the break-in.

In view of all these points, I suggest that adherents of all three families of accounts of Sartrean experiences surveyed in Section 2 should be content to allow that the following conditions are sufficient for S to justifiedly believe that O is absent from L, where L is apprehended under a specific guise (e.g., “the hallway of my house,” “Ground Zero,” “the Salon Carré,” etc.). Given the ongoing disagreements among epistemologists about whether a no-defeater condition is genuinely necessary for justification or for knowledge, it seems wisest for now to offer these conditions as jointly sufficient but not as individually necessary:

• • S has a Sartrean experience of O’s presence from L.

• • S is justified in believing that she has perceptual contact with L.

• • S believes that O is absent from L on the basis of her Sartrean experience.

• • S has no defeaters: no reasons for thinking that O really is present in L (rebutting defeaters) or reasons for thinking that her Sartrean experience is not a good guide to whether O is absent from L (undercutting defeaters).

Similarly, I suggest that adherents of all three families of accounts of Sartrean experience should be content to allow that the following conditions are sufficient for S to know that O is absent from L (again, where L is apprehended under a specific guise):

• • S has a Sartrean experience of O’s presence from L that is noncoincidentally accurate: i.e., her Sartrean experience satisfies the aforementioned accuracy condition, modal–causal condition, and mental image condition.

• • S knows that she has perceptual contact with L.

• • S believes that O is absent from L on the basis of her Sartrean experience.

• • S has no defeaters: no reasons for thinking that O really is present in L (rebutting defeaters) or reasons for thinking that her Sartrean experience is not a good guide to whether O is absent from L (undercutting defeaters).

7. Conclusion

It seems to me that investigating the conditions in which Sartrean experiences get the world right noncoincidentally is a natural avenue for the burgeoning discussion on absence experiences to explore, and I hope that the coming years will see more such exploration. The account offered in this paper proposes that the conditions in question include an accuracy condition, a modal–causal condition on one’s perceptual contact with the absential location, and a mental image condition. No doubt there are lurking counterexamples,¹⁹ and I hesitate to claim that these three conditions are individually necessary and jointly sufficient. Nevertheless, I take it that the broad shape of an adequate account of noncoincidental accuracy for Sartrean experiences will involve both an appropriate link between S and the absential location L and an appropriate link between S’s idea of what the missing object O would be like and the facts in virtue of which O would be like that. If nothing else, I hope to have shown that the way is open to investigate the question of noncoincidental accuracy without first having to settle the question of whether absences really are perceived.