Revisiting the problem of satisfaction conditions and the indispensability of i-desire

Gregory Currie has argued for the indispensability of i-desires – a kind of imaginative counterpart of desires – by drawing a distinction between the satisfaction conditions of the desire-like states involved in our emotional responses to tragedies and those of genuine desires. Nevertheless, Fiora Salis has recently shown that the same sort of distinction can also be found in nonfictional cases and has proposed a solution to the issue of satisfaction conditions that dispenses with i-desires. In this paper, I refute Salis’s stance and argue for the indispensability of i-desires. For this aim to be achieved, I first argue that the distinction between the satisfaction conditions of i-desires and those of desires can be given a different explanation, and that in this case, the same sort of distinction cannot arise in nonfictional cases; Secondly, I argue that we cannot make sense of the conflict between our desire-like states triggered by fictions and our background desires, and therefore i-desires should be introduced to avoid this conflict.


Introduction
Recently, several philosophers have argued that imagining can also be understood as an ability to copy or simulate any mental state. For example, Gregory Currie and Ian Ravenscroft (2002) distinguish creative imagination from recreative imagination. The latter is a capacity for persp ective-shifting. Through recreative imagination, people can put themselves in nonactual situations and produce some imaginative states that are not perceptions, beliefs, or desires but are like these states. Alvin Goldman (2006aGoldman ( , 2006b) introduces the concept of enactment-imagination that is a matter of creating or trying to create the simulation or the facsimile of a selected mental state in one's own mind. Therefore, it seems that there are imaginative perception, imaginative belief (make-believe), imaginative emotion, and of course imaginative desire (hereafter, i-desire).
The simulation theory of mindreading provides a powerful argument for the existence of i-desires. Since the 1980s, several philosophers Ravenscroft, 2002;Goldman, 2006a;Gordon, 1986;Heal, 1986) claim that we explain the mental states of others and predict their behaviors through an imaginative simulation, a capacity to mentally project ourselves into others' situations. It seems that to do so, we must pretend or simulate the target's states. This means that we must bring myself to imaginatively desire what the target desires. Consider the following example: for a costume party, I made myself up as a vampire. While leaving the house, I met my neighbor. He threw his head back, bellowed a wordless roar, and dashed into his house. Why does my neighbor perform such behaviors? I need to put myself in his situation: I imaginatively believe that I am with a vampire and imaginatively desire to escape from the vampire. The two imaginative states can lead me to run or scream. So I understand why my neighbor acts this way. It seems that an imaginative desire is essential for understanding others. The "imaginative desires" are what we refer to as i-desires (Goldman, 2006a, p. 48). It should also be noted that this argument does not depend upon whether the simulation theory is the best theory of mindreading. In fact, as Kind (2016) said, those competing theories, such as the theory-theory, "(…) tend to admit that we engage in imaginative simulations of the sort postulated by simulation theorists -what's primarily at issue between the two sides in the mindreading debate is not whether we simulate but whether such simulations are theory-laden" (Kind, 2016, p. 166).
The current debate is focusing on whether or not i-desires can account for our emotional responses to fictional charact ers and events. 2 Some philosophers argue that genu-ine desires cannot account for these, citing three reasons to support their claim: (1) Desires are governed by normative constraints: one cannot desire something that is unattainable. As we never believe that fictional situations or charact ers are real, we do not have relevant desires about fictional objects (Velleman, 2000, p. 260). Similarly, Currie claims that ''Desires can be shown to be unreasonable, or at least unjustified, if they fail to connect in various ways with the facts'' (Currie, 2002, p. 211). 3 However, my desire that Desdemona not die does not become unreasonable even though I know that Desdomona does not exist.
(2) Desires are intrinsically motivational: to desire p is to be disposed to act in ways that will bring about p. However, in engaging with fictions, people are not always disposed to act (Currie, 2002, p. 211). (3) Egan (2007, 2012) argue that there is no adequate way to understand the content of these desire-like states if they are genuine desires. Consider my desire that Desdomona not die. It cannot be understood as a desire that, according to the story, Desdomona is safe, because I have a contrary desire that, according to the story, Desdomona dies (maybe because I think that this is better for the story). It also cannot be understood as a desire that the fictional charact er Desdomona be safe because the content of this desire rationally requires that one has the corresponding desire about the fiction. This entails that I have a desire that, according to the story, Desdomona is safe. Opponents of i-desire claim that these reasons are implausible. Reason (1) implies that we cannot desire things that are unattainable. However, in daily life, we often have a variety of desires toward things that are not actual. One often has desires about the past, the future, counterfactual events, and so forth (Kind, 2011, p. 425). Reason (2) claims that desires are intrinsically motivational. However, we can find many cases in which desires are inert. Alfred Mele (1995, p. 394) suggests that a person driving to the airport to pick up friends may desire that the plane took off on time. In this case, the person's desire is inert because it is about past events. In addition, Carruthers (2003) also suggests that ''real desires will normally lead to real act ions only when interact ing with real beliefs'' . I am not motivated to save Desdemona because I does not believe that Desdemona is a real person (see also Kind, 2011, p. 426-427). Reason 3) claims that contradictory desires make us irrational. But Kind (2011, p. 429) noticed that contradictory desires are commonplace. For instance, a mother may want her only child to go away to university because she believes that it is necessary for his own good. Simultaneously, she may want her son to stay home because she fears having an empty nest.
Recently, Gregory Currie (2010) proposed a new argument in favor of the indisp ensability of i-desires. He argues that desire-like states triggered by fictions must be introduced 253 to account for our emotional responses to tragic fictions, but if these desire-like states are understood as desires, then there is no way to sp ecify their satisfact ion conditions; so these desire-like states should be regarded as i-desires. Fiora Salis (2016) has proposed a solution to the problem of satisfact ion conditions that disp enses with i-desires; she has shown that the same sort of discrepancy between the satisfact ion conditions of i-desires and those of desires can also be found in nonfictional cases, so there is no need to introduce i-desires to account for our emotional engagement with fictions.
In this paper, I rebut Salis's argument and argue for the indisp ensability of i-desires. The second and third Sections state, as exactly as possible, Currie's argument for i-desires and Salis's objection. The fourth Section draws a distinction ignored by Currie and Salis between the satisfact ion conditions of i-desires and of desires and argues that this sort of distinction cannot arise in nonfictional cases. The fifth Section proposes another argument for the indisp ensability of i-desires: if the desire-like states triggered by fictions are understood as desires, then our emotional responses to tragic fictions will involve a conflict that we cannot make sense of from our daily experience, so i-desires should be introduced to avoid this conflict.

Currie's argument for i-desires
Currie's argument is based on his analysis of our emotional responses to tragedies, which he noted involve two contradictory mental states. We often want the fiction to go a certain way; in other words, we want a tragedy to end tragically in order to read an amazing story. For example, we wish for Anna Karenina to unfold in such a way that Anna commits suicide. Additionally, in being engaged in fiction, we feel sympathy, pity, anxiety, and so forth, which implies that we wish that tragic events do not occur in the fiction. In the case of Anna Karenina, we also do not want Anna to lose her life. Thus, our emotional responses can be charact erized by the following two contradictory states: (1) We want the fiction to be such that something, E, occurs in it.
(2) We react in ways that make it tempting to say that we want E not to occur (Currie, 2010, p. 632).
Currie claims that (2) raises a question of whether the tempting thing to say is the right thing, and he suggests that there are three solutions to interpret (2). According to the simple solution, 4 (2) is equivalent to: (3) We desire that E not occur. According to the simple solution, the tempting thing to say is the right thing. (3) is a desire and its content is that the fictional event does not occur. In the case of Anna Karenina, by combining (1) and (3), we have a desire for the novel to unfold in such a way that Anna commits suicide and also a desire that Anna not die.
According to the change-of-content solution, 5 (3) is false. The right thing should be to say: (4) We desire the fiction to be such that E not occur. The change-of-content solution requires that the content of the desire includes reference to the fiction; our desire is not directed to the event included in the fiction, but rather the fiction itself. By combining (1) and (4), we have a desire for Anna Karenina to unfold in a way that Anna commits suicide and further desire it to be a novel wherein she does not die.
According to the change-of-attitude solution, (3) gets the content right but the attitude wrong. Thus, (3) should be replaced with: (5) We i-desire that E not occur. Our mental state is directed to the fictional event itself. The state is not a desire, but rather an i-desire, an imaginative counterpart of desire. By combining (1) and (5), we desire Anna Karenina to be a story wherein Anna die, and we have an i-desire that Anna not die.
Currie argues resp ectively that both the change-of-content solution and the simple solution are problematic. The change-of-content solution does not properly account for our emotional responses to tragedies. It indicates that we want a tragedy to end tragically but also want it not to include tragic endings. That requires that we are ambivalent about the tragic ending we want, but most people feel no ambivalence about what they want for the ending. Moreover, the change-of-content solution also implies that we must always be disappointed because we cannot always get what we want in having two conflicting states. However, disappointment is not our emotional response to tragedies. Doggett and Egan explain: ''If you were so conflicted, the fiction would disappoint you by not giving you everything you want. But Romeo and Juliet was not disappointing for us in this way'' Egan, 2012, p. 281-282).
To rebut the simple solution, Currie imagines a counterexample in which the charact er in fiction is real:  (Currie, 2010, p. 633).
It seems that, according to the simple solution, i) and ii), resp ectively, are equivalent to 1) and 3): I have a desire for the play to be such that Mrs. Thatcher is killed and also have a desire that Mrs. Thatcher is not killed. However, Currie claims that i) and ii) cannot elicit a tragic response to the drama, because (i) and (ii) ''leave it entirely open how I respond to the fiction'' (Currie, 2010, p. 634). Although I really admire Mrs. Thatcher, I may still be on the side of the assassin if the story is so excellent and changes my persp ective. In this case, I might have a desire (iii) that Mrs. Thatcher die. The combination of (i) and (iii) cannot elicit a tragic response, because (i) corresponds to (1) but (iii) contradicts (3); it is not consistent with the simple solution.
According to Currie, a simple theorist may tweak their theory by claiming that I have a background desire that Mrs. Thatcher has a good life, but also have a condition-dependent desire that Mrs. Thatcher is killed, and neither desire is dominated by more salient contrary desires. 6 The desire that Mrs. Thatcher have a good life is a stable, long-term and background desire. But when watching the drama, I come to have an occasional desire that she die. This desire is triggered by watching the drama and is regarded as a condition-dependent desire.
Currie's objection is that the condition-dependent desire that Mrs. Thatcher is killed has a different satisfact ion condition. My background desire that Mrs. Thatcher have a good life is satisfied if she has a good life, but the so-called "desire triggered by tragedy" is not satisfied even if Mrs. Thatcher is killed. In other words, my desire-like state triggered by fiction is not satisfied by what happens but by what happens in the fiction; my "desire" that Mrs. Thatcher is killed is satisfied if and only if she is killed according to the fiction. The satisfaction conditions of a desire-like state triggered by fiction must include reference to fiction. However, the propositional content of my desire-like state does not include reference to fiction. In other words, what I "desire" is that Mrs. Thatcher is killed, but is not that she is killed in the fiction. Currie (2010, Note 7) notices that we can have genuine desires about the story. For instance, one person can desire that Mrs. Thatcher is killed in the fiction. Although my desire-like state that Mrs. Thatcher is killed and the desire that she is killed in the fiction have the same satisfact ion condition, they are not the same attitude, because the propositional content of the latter includes reference to the fiction, but the former does not. At this point, Currie proposes a distinction between desires and i-desires: (SC) A putative desire, A, is an i-desire and not really a desire if A has satisfact ion conditions, a canonical statement of which makes reference to a fiction which is not also the object of A (Currie, 2010, p. 635).
My desire that Mrs. Thatcher thrive is a desire because its satisfact ion conditions does not include reference to a fiction; my desire-like state that Mrs. Thatcher is killed is an i-desire because its satisfact ion conditions include refer-ence to a fiction, and the fiction is not the object of my state. Therefore, the change-of-attitude theory is right, but the simple theory is not plausible.
Salis's objection to Currie's argument Salis (2016) counters that Currie's argument about the satisfact ion conditions is based on a restrictive notion of desires. To show that the two conflicting states involved in our emotional response to fiction can also be found in some nonfictional cases, she imagines the following example: Suppose that I attend for the first time a training session on the impact of stress on decision-making in a laboratory for behavioural neuroscience. The experiment includes brain surgery on subjects, usually rats, which eventually causes their death. When I attend the experiment, I have the following desire: (iv) I want the experiment to be such that the rat dies (suppose that is the only way to gather the necessary data). Yet, I am also perturbed by the death of the rat. In other words, I react in ways that make it tempting to say that I have another desire: (v) I want that the rat not die (Salis, 2016, section 5, p. 112).
Salis claims that the combination of (iv) and (v), like (i) and (ii) in the case of Mrs. Thatcher, leaves it completely open how I react to the experiment. Although I support the wellbeing of animals, I may still hope that the rat is killed because I focus on the lab experiment and have a desire that the rat die. So, my response to the lab experiment would not be tragic. 7 Like Currie, Salis claims that we can modify the proposal by claiming that the lab experiment is tragic for me if I have a background desire that the rat not die and a condition-dependent desire that the rat die, and neither desire is dominated by more salient contrary desires. In the case, I have two desires: one is a background desire that the rat not die; another is a condition-dependent desire that the rat die. The condition-dependent desire is an occasional state triggered by the experiment. The two desires can be represented as: (6) I desire that the rat die.
(7) I desire that the rat not die.
(7) is a background desire, and 6) is a condition-dependent desire triggered by the experiment. Thus, the combination of (6) and (7) elicits a tragic response to the lab experiment. 6 Currie claims that proponents of the simple solution introduce the notion of a condition-dependent desires. It seems that those desire-like states about fictional characters or events are triggered by the fiction; while we do not engage with fictions, we do not have such desires. But it should also be noted that some proponents of the simple solution such as Carruthers (2003), Kind (2011) or Spaulding (2015), do not mention the concept of condition-dependent desires. 7 It should be noted that Salis does not really think that the combination of desires regarding the lab rat elicits a tragic response. Salis attaches an asterisk to "tragic response" in her paper.
According to Salis, the combination of the two desires is similar to that of a background desire and a condition-dependent desire triggered by the fiction in Currie's example. My background and long-term desire (7) that the rat not die is satisfied if the rat does not die, but my condition-dependent desire (6) that the rat die is not satisfied while the rat escapes from the cage and is killed in another way. In other words, my desire that the rat not die is satisfied by what happens to the rat, but my desire that the rat die is satisfied not by what happens to the rat, but by what happen in the experiment. So, the satisfact ion conditions of the desire (6) that the rat die would not be simply that the rat dies, but that it dies in the experiment. Moreover, the propositional content of the desire (6) does not include reference to the lab experiment. It seems that (6) corresponds to the state triggered by the fiction in Currie's example. As a consequence, we should also introduce a novel desire-like state: x-desire.
(SC*): A putative desire, B, is an x-desire and not really a desire if B has satisfact ion conditions, a canonical statement of which makes reference to a lab experiment that is not also the object of B.
The same discrepancy between the satisfact ion conditions of desires and those of desire-like states triggered by some particular reference can be found not only in fictional cases but also in nonfictional cases. By (SC), we conclude that the desire-like state triggered by fiction is an i-desire; similarly, it seems that by (SC*), we should conclude that the desire-like state triggered by the experiment is not really a desire but an x-desire, an e-desire. Moreover, according to Salis, postulating a novel kind of mental state that is relativized to the context of a lab experiment would be highly implausible. She said "(…) no one would introduce a new mental state to explain our react ion to the lab experiment or to any other cases that do not engage our imagination" (Salis, 2016, Section 5, p. 114). Similarly, we have no reason to introduce an i-desire to explain our react ion to the fictional case. So, we cannot affirm the indisp ensability of i-desires by drawing a distinction between the satisfact ion conditions of desires and of i-desires.

The propositional contents of desire-like states and their satisfaction conditions
If Currie' s purpose is to distinguish desires from i-desires by showing that the satisfact ion conditions of i-desires include reference to fictions and those of desires do not include it, the same sort of distinction is also found in nonfictional cases. However, the distinction between the satisfact ion conditions of i-desires and those of desires can also be seen as a relationship between the satisfact ion conditions and the truth values of propositional contents: if a desire is satisfied, then the content of the desire is true, but when an i-desire is satisfied, then the propositional content of i-desire is not always true. I propose that the distinction circumvents Salis' s objection.
Consider the following four sentences: (a) The rats die in the lab experiment. (b) The rats die in the real world. (c) Mrs. Thatcher dies in the fiction. (d) Mrs. Thatcher dies in the real world. It seems that "the real world" is a larger set of events and facts that include "the lab experiment" . In other words, what has happened in the lab experiment has also happened in the real world; "the lab experiment" is included in "in the real world" . Therefore, although the sentence (a) does not literally include "in the real world" , it is also bound by the operator "in the real world" . It follows that proposition (b) can be inferred from proposition (a): if the rats die in the lab experiment, then the rats also die in the real world. But proposition (a) cannot be inferred from proposition (b) (it is possible that the rats die of hunger in the real world instead of being killed in the lab experiment).
Yet what has happened in the fiction is isolated from the real world. Romeo dies in fiction instead of in the real world; Sherlock Holmes does not live in the real world but rather in the London of the novel; Captain Haddock is Tintin's friend in comics, not in the real world. Works of fiction can sometimes reveal some of the features of the real world but it cannot be shown that what has happened in fiction can truly happen in the real world. Another difference between the operators "in the experiment" and "in the fiction" is that the lab experiment has a causal link with real-world features, but the fiction does not have such a link. The lab experiment can change real-world features. In Salis's example, the rats that died in the experiment reduce the number of rats living in the real world, but the rats that die in fiction cannot do that. So it seems that sentence (d) cannot be inferred from sentence (c): "Mrs. Thatcher dies in the fiction" does not entail that "Mrs. Thatcher dies in the real world" .
The distinction between the two concepts "in the fiction" and "in the experiment" can also apply to the satisfact ion conditions of desire-like states. In Salis's example, the subject has a desire that the rats die; as a condition-dependent state, it is triggered by the experiment. According to Salis, the satisfact ion conditions of the desire include reference to the experiment, which is not the propositional content of the desire: if the desire that the rats die is satisfied, then the rats die in the lab experiment. Thereafter, since what has happened in the lab experiment has also happened in the real world, while the desire that the rats die is satisfied, the rats die not only in the lab experiment but also in the real world. According to the rule of conjunction elimination, one can conclude that the rats die in the real world. On the other hand, it seems that while the rats die in the real world, the proposition that the rats die is true (Concretely, we ought to claim that the proposition that the rats die in the real world is true. But while we don't emphasize that a proposition is bound by the fiction, we can remove the operator "in the real world" without changing its truth value and meaning). Therefore, we have reached a temporary conclusion: if the condition-dependent desire that 256 the rats die is satisfied, then its propositional content that the rats die is true. 8 Usually, while one desires that A is Ф (A is Ф is the propositional content of the desire), if the desire is satisfied, then the propositional content A is Ф is true; otherwise, the desire is not satisfied or is not a desire. In Currie's example, the subject has a condition-dependent desire triggered by the fiction: a desire that Mrs. Thatcher die. The satisfact ion condition of the desire includes reference to fiction. So we say that if the desire that Mrs. Thatcher die is satisfied, then Mrs. Thatcher dies in the fiction. Since what has happened in the fiction is isolated from the real world, one cannot infer that Mrs. Thatcher dies in the real world from the proposition that she dies in the fiction. While Mrs. Thatcher is killed in the fiction, she might die in the real world but might also have a good life in the real world. The proposition that Mrs. Thatcher is killed in the fiction is independent of the real-world features. Hence, if the desire that Mrs. Thatcher die is satisfied, its propositional content that Mrs. Thatcher dies is not always true. 9 It seems that the satisfact ion conditions of a desire triggered by the fiction are different from a normal desire.
If a condition-dependent desire triggered by the fiction has satisfact ion conditions different from normal desires, why do we still insist that it is a kind of desire? It seems that we should introduce i-desires to account for the desire-like states. So I propose a distinction between desires and i-desires: (SC1) If a desire-like state "A is Ф" is a desire, then if the desire-like state is satisfied, "A is Ф" is true ("A is Ф" is the propositional content of a mental state); (SC2) If a desire-like state "A is Ф" is an i-desire, then while the desire-like state is satisfied, the truth value of the propositional content "A is Ф" is not fixed ("A is Ф" is either true or false). 10 In Currie's example, the desire that Mrs. Thatcher thrive is a desire because if it is satisfied, then its propositional content is true, but the state that Mrs. Thatcher is killed is not a desire but an i-desire because it satisfies (SC2): while my desire-like state that Mrs. Thatcher is killed is satisfied, the truth value of the proposition that Mrs. Thatcher dies is not fixed; we don't know if Mrs. Thatcher is truly killed or not. In Salis's example, the desire that the rats die is a desire, not an i-desire or x-desire, because it satisfies (SC1): while my desire that the rats die is satisfied, the proposition that the rats die is true.
To summarize, I have proposed a distinction between desires and i-desires: if a desire is satisfied, then its proposi-tional content is true, but if an i-desire is satisfied, then its content is either true or false. The distinction indicates that a desire-like state triggered by fictions, which is introduced by Currie, should be regarded as an i-desire, but a desire-like state triggered by the lab experiment, which is introduced by Salis, should not be regarded as an i-desire but a genuine desire. In such a way, I have circumvented Salis's objections. I admit that my argument cannot decisively est ablish that the thesis of i-desires is the best account of our engagement with fiction. But, at least, the argument suffices to undermine Salis' objections. In the next section, I would argue why i-desires are indisp ensable in the account of our engagement with fiction.
Why we cannot make sense of the conflicting desires about fictions In this section, independently of the problem of the satisfact ion conditions of desires, I argue that in being engaged in tragedy fictions, the subject has two conflicting desire-like states, and if these desire-like states are understood as desires, we cannot make sense of them because it cannot be understood in a way that we account for the conflicting desires in daily life. To avoid this conflict, i-desires must be introduced. Reconsider Currie's example:  (Currie, 2010, p. 633).
A proponent of the simple solution may think that i) and ii) are, resp ectively, equivalent to e) and f): (e) I desire that Mrs. Thatcher die.
(f) I desire that Mrs. Thatcher not die.
(e) is a desire triggered by the fiction, (f) is a background and long-term desire. It seems that being engaged in fiction, one often has two conflicting attitudes toward charact ers: one wants the fiction to be such that some tragic events occur, but as a matter of fact, one also wants that the events do not 8 Yet, when the propositional content of the desire that rats die is true, the desire is not always satisfied, because we can infer the proposition that the rats die in the real world from the proposition that the rats die in the experiment, but cannot reversely make an inference. 9 You might think that the propositional content is that Mrs. Thatcher die in the fiction. In other words, the subject has a desire that Mrs. Thatcher die in the fiction. But if the propositional content of a condition-dependent desire includes the reference "in the fiction", the simple solution is equivalent to the change-of-content solution, which is implausible in explaining our emotional response to tragedies. It should be noted that I do not modify the content of desire-like state, but only consider their truth-conditions. 10 (SC1) and (SC2) are not the definition of i-desires or desires. They are only a necessary distinction between desires and i-desires. I admit that mere truth-conditions cannot individualize i-desires or desires. In daily life, while we judge that a desire-like state is an i-desire, we need not only its truth-condition, but also to understand the objects of the attitude, the current situations and the relevant cognitive states.
occur. In Currie's example, the conflicting attitudes are represented as (e) and (f), two contradictory desire-like states. In this section, we do not consider if they are enough to elicit my tragic response but do consider whether or not the case of the conflicting desires can be explained in a way that we explain conflicting desires in daily life.
Conflicting desires are entirely possible in daily life. One can have a desire for a short-term goal that is contrary to a long-term goal. For example, one person can desire to smoke while desiring to remain healthy. In addition, our desires can also be directed to one object by virtue of its different asp ects; for example, somebody can desire to drink beer by virtue of its taste but not desire to drink it by virtue of its consequences on his or her health. The proponents of the simple solution also state that one person can have a condition-dependent desire that contradicts his or her background desires. A person with a background desire to stay sober may have a desire to drink when being confronted with alcohol. Il seems that (e) can be regarded as a short-term or condition-dependent desire, and (f) can be seen as a long-term or background desire. Thus, the tension which I feel in watching the drama should comes from a conflict between two kinds of desires.
Nevertheless, conflicting desires in daily life cannot be applied to cases of fiction. One reason in favor of my thesis is that conflicting desires cannot be satisfied simultaneously. If someone drinks beer, his or her desire to drink was satisfied, while if he or she did not drink beer, his or her desire to stay sober may be satisfied, but both of these desires cannot be satisfied simultaneously. In Salis' s example, one person has a desire, triggered by the lab experiment, that the rats die and also has a background desire that the rats do not die. The two desires are in conflict with each other, and they also cannot be simultaneously satisfied. While the rats are killed in the experiment, one' s background desire is not satisfied; while the rats thrive, one' s desire for the lab experiment cannot be satisfied. There are no states of affairs that satisfy the two desires. However, it is surprising that both (e) and (f), two conflicting desires about charact ers, can be satisfied simultaneously. Suppose that at the end of my story, Mrs. Thatcher is murdered; Death of a Prime Minister ends tragically, so my desire (e) is satisfied. But suppose that, simultaneously, I know that Mrs. Thatcher is living a good life in real life. My desire (f) is then also satisfied, which entails that both of my conflicting desire-like states are satisfied simultaneously, a feature that the conflicting desires in daily life cannot have. Therefore, the combination of (e) and (f) cannot be understood as a combination of background desires and condition-dependent desires. I conclude that conflicting desires in daily life cannot be applied to cases of fiction, and the conflicting desires about fictional charact ers are utterly different from anything in daily life.
There are seemingly two ways to account for the problem. One is to say that the two desires are not intrinsically conflicting; another is to admit the two states are conflicting, but to say that one of them is not a desire.
According to the first view, opponents can give two proposals against my thesis: A) Mrs. Thatcher in Death of a Prime 258 ent kinds of attitudes. By not introducing conflicting desires that cannot be understood in daily life, one can better explain the difference between our attitudes toward fiction and those in daily life. I indirectly argue for the indisp ensability of i-desires: if i-desires are not introduced, one cannot make sense of the cases of conflicting desires about charact ers. Dogget and Egan (2012) argue that there is no adequate way to understand the content of my desire that Desdemona not die. It cannot be understood as a desire that, according to the story, Desdemona is safe, because I have a contrary desire that, according to the story, Desdemona dies, and because it's irrational for one person to have two conflicting desires. But Kind (2011, p. 429) rebutted this argument because she noticed that contradictory desires are very commonplace. My point is not that it is irrational or impossible for having two conflicting desires, but rather that the conflicting states about fictional charact ers cannot be understood as the case of conflicting desires, because the two conflicting states can be satisfied at the same time. 11 Unlike Currie's argument, my argument does not rely on our tragic emotions; we don't need to analyse the constitution of our tragic emotions in order to affirm the existence of i-desires. Hence, I conclude that my argument is better than other those involving conflicting states. 12

Conclusion
Our emotional responses to tragedies are charact erized by two conflicting desire-like states. Suppose that E refers to tragic events. The tragic responses imply that we have a desire-like state, triggered by the tragedy, that E occur. I have argued that the desire-like state triggered by the tragedy has satisfact ion conditions that are different from genuine desires: while a desire is satisfied, its content is true, but while a desire-like state triggered by the tragedy is satisfied, its content is not always true. So I proposed that the desire-like state is an i-desire for the reason that its satisfact ion conditions cannot be understood in a way that we understand a genuine desire. Salis has recently argued for the disp ensability of i-desires by showing that a genuine desire sometimes has satisfact ion conditions similar to a desire-like state triggered by the tragedy. My argument can avoid her objections and justify the indispensability of i-desires.
On the other hand, I have also proposed another argument for i-desires, independent of the problem of the satisfact ion conditions of desires: as demonstrated by Currie, our emotional responses to tragedies are charact erized by two conflicting desire-like states, but the conflicting states cannot be understood as desires because they have a feature -namely, that they can be satisfied at the same time -that conflicting desires in daily life cannot have. So i-desires must be introduced to make sense of them.