Mental time travel and the philosophy of memory

The idea that episodic memory is a form of mental time travel has played an important role in the development of memory research in the last couple decades. Despite its growing importance in psychology, philosophers have only begun to develop an interest in philosophical questions pertaining to the relationship between memory and mental time travel. Thus, this paper proposes a more systematic discussion of the relationship between memory and mental time travel from the point of view of philosophy. I start by discussing some of the motivations to take memory to be a form of mental time travel. I call the resulting view of memory the mental time travel view. I then proceed to consider important philosophical questions pertaining to memory and develop them in the context of the mental time travel view. I conclude by suggesting that the intersection of the philosophy of memory and research on mental time travel not only provides new perspectives to think about traditional philosophical questions, but also new questions that have not been explored before. acknowledgements I’m grateful to Kirk Michaelian for comments on a previous draft and for various discussions on the topics related to this paper. I’m also grateful to the participants in the philosophy of memory seminar in 2017 at the University of Otago, and to participants in the Mental Time Travel e Agência Moral workshop at Unisinos, where I had the opportunity to discuss some of these topics in more detail.


Introduction
The idea that episodic memory is a form of mental time travel has played an important role in the development of memory research in the last couple of decades.Mental time travel, according to Su dendorf and Corballis (1997, p. 133), "comprises the mental reconstruction of personal events from the past (episodic memory) and the mental construction of possi le events in the future"."The real importance of mental time travel", they a d, "ap lies to travel into the future rather than into the past; that is, we predominantly stand in the present facing the future rather than looking back at the past" (Su dendorf and Corballis, 1997, p. 147).
Traditiona ly, memory has been taken to be primarily about the past, in the sense that it a lows us to reca l things that hap ened.However, the su gestion that episodic memory is just a form of mental time travel cha lenges this idea, for "the primary role of mental time travel into the past is to provide raw material from which to construct and imagine possi le futures" (Su dendorf and Corba lis, 2007, p. 302).These considerations raise a number of important philosophical questions.A first relevant question refers to whether memory requires an ap ropriate causal connection to past experiences or events.Since Martin and Deutscher (1966), it has been standard to assume that remembering requires such a connection (see, e.g., Bernecker, 2008;Debus, 2008;Michaelian, 2011;Robins, 2016b).A second relevant question is whether episodic memory can be a source of knowledge of the past (see Debus, 2014;Michaelian, 2016b).Since mental time travel into the past, or episodic memory, is in the service of providing raw material to simulate future scenarios, it is not clear whether or under what conditions it can provide us with relia le information about past hap enings.A third and more general question refers to the relationship between memory and other forms of mental time travel, such as imagining future events.Because both are a result of similar cognitive capacities, the question of whether they belong to the same metaphysical kind becomes central (see Per in and Michaelian, 2017).
These and other questions have attra ed attention from philosophers concerned with memory (see, e.g., De Brigard, 2014;Debus, 2014;Michaelian, 2016b;Per in, 2016).In this paper, I wi l explore some of the implications that the mental time tra el iew of me ory, as I wi l refer to it, has for the philosophy of memory.I wi l start by discussing some motivations to consider episodic memory as a form of mental time travel.Subsequently, I wi l explore the implications of this idea for the philosophy of memory.

Episodic memory and mental time travel
Before we discuss the relationship between episodic memory and mental time travel, it wi l be helpful to first clarify what e iso ic me ory is.The term was initia ly introduced by Endel Tulving (1972, p. 385) and, roughly eaking, cor esponds to the memory system responsi le for receiving and storing "information about tempora ly dated episodes or events, and temporal-spatial relations among these events" .2So, when you episodica ly remember an event, your memory contains information about the hat, the here, and the hen associated with that event.That is the so-ca led whatwhen-where view of episodic memory, or simply the www iew.Episodic memories, in Tulving's initial formulation, contrast with semantic memories, which refer to memories about general facts that were not necessarily experienced.For example, when I remember that the Second World War ended in 1945, I am semantica ly remembering a fact by using lan uage.The semantic memory system, Tulving says, refers to the "organized knowledge a person possesses about words and other verbal symbols, their meanings and referents, about relations among them, about rules, formulas, and algorithms for the manipulation of these symbols, concepts, and relations" (1972, p. 386).Thus, in contrast to episodic memories, semantic memories do not require the previous experience of the relevant events.
The important thing to note about this definition of episodic memory is that it is primarily based on the kind of information that is processed and stored.And, because of this, it faces some important pro lems.One such pro lem refers to the fact that some semantic memories possess the relevant "www" information; for example, my memory that the Waterloo battle was fought in 1815.Thus, it is not entirely clear whether episodic memories and semantic memories can be distin uished solely on the basis of the information possessed by them.Another pro lem refers to the phenomenological dimension of episodic memories.Remembering a particular event that was previously experienced seems to involve more than the retrieval of information.Episodica ly remembering seems to have a distinctive phenomenology, involving a "feeling of pastness" (Russe l, 1921, p. 161-162) and a "feeling of warmth and intimacy" (James, 1890).In other words, besides the information car ied, episodic memories seem to make reference to the past ("feeling of pastness") and to belong to subjects in a unique way ("feeling of warmth and intimacy").For example, when I remember my tenth birthday party, the memory not only presents the event as having occur ed in the past, but also as being "mine" , in the sense that I seem to own the memory.These and other difficulties led Tulving to reformulate his first chara erization of episodic memory.Later on, he proposed a definition that took into account the phenomenological a ects described above.According to him, besides car ying "www" information, episodic memories involve a unique kind of consciousness, which he ca led autonoetic consciousnes or simply autonoesis (see Tulving, 1985Tulving, , 2005)).Autonoesis, Tulving says, "refers to the kind of conscious awareness that chara erizes conscious reco lection of personal hap enings"; that is, it is what makes subjects "aware that the present experience is related to the past experience in a way that no other kind of experience is" (2005, p. 15). 3he definition of episodic memory as involving autonoesis is very important.Because "[t]he act of remembering [...] is chara erized by a distinctive, unique awareness of reexperiencing here and now something that hap ened before, at another time and in another place" (Tulving, 1993, p. 68), remembering makes subjects "capa le of mental time tra el: [...] [a] person can transport at wi l into the personal past, as we l as into the future" (p.67, my emphasis).So, besides being responsi le for the unique feeling associated with episodic memories, autonoesis gives subjects a more general capacity to "travel" in subjective time.This is not difficult to motivate on phenomenological grounds.As Klein (2015, p. 21) notes, there is a "perceived temporal symmetry between movements toward (future) and away (past) from the present" .To i lustrate, imagine that you are thinking about your holidays at the beach next year.Similarly to episodic memories, you have the feeling that the thought is owned by you, in the sense that the holidays are yours and not someone else's.However, because the event is something that can hap en, it is presented to you as being "future" to your cur ent thought.Thus, it looks like we can "relocate" ourselves to the future in the same way that we can do it in relation to the past.
The capacity endowed to us by autonoesis to travel both to past subjective time and to future subjective time consists in an important motivation to take episodic memory to be just one form -among others -of mental time tra el.Despite giving emphasis to phenomenological considerations above, there are also good empirical reasons to endorse this view.In a recent survey, Per in and Michaelian (2017) discuss similarities between episodic memory and future mental time travel found in different domains.In developmental studies, for example, it has been shown that the children's capacity to remember the past and imagine the future arise at ap roximately the same time (Su dendorf and Busby, 2005;Atance, 2008;Fivush, 2011).In studies with patients with memory impairments, it has been found that deficits in memory incur in similar deficits in the ability to think about future scenarios (Klein et al., 2002;Rosenbaum et al., 2005;Hassabis et al., 2007).Moreover, imaging studies also show that there is a strong overlap in the brain regions associated with episodic memory and future mental time travel (A dis et al., 2007;Scha er et al., 2007Scha er et al., , 2012)).
I wi l not attempt to review the relevant literature here.4I sha l, instead, point out an important development of the mental time tra el iew of me ory.More recently, some researchers have su ge ed that the primary function of mental time travel is not to a low us to remember the past.Su dendorf and Corba lis, for example, ar ue that "[t]he real importance of mental time travel ap lies to travel into the future rather than into the past; that is, we predominantly stand in the present facing the future rather than looking back at the past" (1997, p. 147).In a similar spirit, De Brigard says that "remembering is a particular operation of a cognitive system that permits the flexi le recombination of different components of encoded traces into representations of possi le past events [...] in the service of constructing mental simulations of possi le future events" (2014, p. 158, my emphasis).And, more recently, Michaelian says that "remembering is not different in kind from other episodic constructive processes" (2016b, p. 103); thus, "[w]hat it is for a subject to remember [...] is for him to imagine an episode belonging to his personal past" (2016b, p. 111).
The idea that the primary function of mental time travel is not to remember the past but to imagine the future has important consequences.One such consequence is that our common sense conception of memory, according to which memory's function is to store information of what hap ened, seems to be threatened.It is compati le with the mental time travel view that our representations of the past be inaccurate as long as they are beneficial for future a ions.So, as De Brigard notes, "many ordinary cases of misremembering should not be seen as instances of memory's malfunction" (2014, p. 158, his emphasis).This raises a further question, which is of particular interest to philosophers, about whether, and if so, how, memory provides knowledge of the past.Because the primary function of remembering is not to recover information about the past, we need a proper account of how knowledge can be formed on the basis of memory.Similarly, the mental time travel view poses important questions pertaining to the relationship between memories and the past events.The causal theory of memory, which has been predominant in philosophy for the past four decades, stipulates Filosofi a Unisinos -Unisinos Journal of Philosophy -19(1):52-62, jan/apr 2018 that remembering requires the preservation of an ap ropriate causal connection to past events.However, if memory is a form of mental time travel in the same way that imagination is, and "if imagining need not draw on stored information ultimately originating in experience of the relevant episode" (Michaelian, 2016b, p. 111), there is no principled reason to say that such requirement holds for memory.
In summary, the mental time travel view of memory raises a lot of important questions for philosophers concerned with memory.In an attempt to motivate those pro lems, I wi l consider, in the next section, some implications that the mental time travel view of memory has for the philosophy of memory.

Mental time travel and the philosophy of memory
The mental time travel view of memory not only challenges important traditional conceptions about memory, but also offers pro ects for future research on the subject.In this section, I wi l consider some topics that are of potential interest to philosophers of memory concerning the mental time travel view of memory.However, because the interest of philosophers on those topics is sti l very recent, there are not many works dealing systematica ly with the questions that I discuss below.For this reason, rather than attempting to survey the debate, I wi l try to motivate some pro lems of potential interest.

The causal theory of memory
After the pu lication of Martin and Deutscher's seminal paper "Remembering" (1966), philosophers in the analytic tradition started to develop an increasing interest in philosophical questions pertaining to memory.Martin and Deutscher proposed what is now known as the causal theory of me ory (CTM).The CTM has been very influential and it sti l shapes, to a large extent, the way how philosophers think about memory today.5However, if cor ect, the mental time travel view raises important concerns about the CTM.
The CTM provides us with a set of criteria to determine whether a given mental state counts as remembering or not.For the CTM, a subject S counts as remembering an event e iff: To clarify these points, consider my putative memory of my tenth birthday party.In order for me to count as remembering this event, I need to have experienced it previously.That is the pa re resentation condition.A ditiona ly, I need to be a le to represent the same event in the present.That is the current re resentation condition.But my past and cur ent representations can only be representations of the same event if their contents are sufficiently similar (the content condition); for example, if the contents of both representations contain members of my family and friends, a chocolate cake, etc.7 Fina ly, remembering requires that my cur ent representation of my tenth birthday be caused, in an ap ropriate way, by my past representation of the same event (the causal connection condition).The requirement for such causal connection consists in the main novelty of the CTM.Moreover, since it is also the source of the pro lems that arise in the context of the mental time travel view of memory, I wi l focus on it more closely.
The causal condition is sup osed to rule out cases that, intuitively, we do not count as remembering, but that are allowed by (1)-(3).To see this, consider the case of Kent described by Martin and Deutscher (1966, p. 174): A man whom we shall call Kent is in a car accident and sees particular details of it, because of his special position.Later on, Kent is involved in another accident in which he gets a severe blow on the head as a result of which he forgets a certain section of his own history, including the first accident.He can no longer fulfil the first criterion for memory of the first accident.Some time after this second accident, a popular and rather irresponsible hypnotist gives a show.He hyp-Filosofi a Unisinos -Unisinos Journal of Philosophy -19(1):52-62, jan/apr 2018 notizes a large number of people and suggests to them that they will believe that they had been in a car accident at a certain time and place.The hypnotist has never heard a thing about Kent nor the details of Kent's accident, and it is by sheer coincidence that the time, place, and details which he provides are just as they were in Kent's first accident.Kent is one of the group which is hypnotized.The suggestion works and [...] [Kent] believes firmly that he has been in an accident.The accident as he believes it to be is just like the first one in which he was really involved.
Kent's case satisfies (1) and ( 2), as he had a past representation of the car accident and has a cur ent representation of the same event.Moreover, it satisfies (3) too, for Kent's cur ent representation is sufficiently similar to his past representation.Nevertheless, it seems wrong to say that Kent is genuinely remembering.The reason is that his cur ent representation does not preserve the right kind of causal connection to his past representation.To use Martin and Deutscher's (1966) term, the past representation is not "operative" in producing the cur ent representation.In Kent's case, the operative cause, so to eak, is the hypnotist.For the CTM, then, remembering is not only a matter of getting the details of a past experience of an event right, but also of standing in an ap ropriate causal relation to that experience.
Besides offering a way to rule out cases not contemplated by (1)-(3), the causal connection condition has also been used to provide a taxonomy of memory.As it stands, the CTM is an answer to the general question of what it takes for a subject to remember.However, there is more than one way in which one can successfu ly or unsuccessfu ly remember something, which requires an account of those differences.For example, it is consistent with remembering my tenth birthday party that I get some of its details wrong.8I can correctly remember that my whole family was there and that the party took place at a certain location, but I can simultaneously remember, incor ectly, that I had strawber y cake.In this case, we can say that I am misre e be ing my tenth birthday party.Thus, Robins (2016b) has recently ar ued that, given the constructive chara er of memory (see Bartlett, 1995;Scha er et al., 2007Scha er et al., , 2012;;Michaelian, 2011;De Brigard, 2014), we need to ap eal to a causal connection between past and cur ent representations to distin uish remembering from misremembering.9In a similar spirit, Bernecker (2017) has su ge ed that one can only distin uish successful remembering from confabulations (see Hirstein, 2005) if one requires that the former, but not the latter, preserves a causal connection to past experiences (see also Robins 2016bRobins , 2017)).The causal connection, therefore, is not only important to provide an adequate analysis of remembering, but also of the different kinds of successful and unsuccessful remembering.
The mental time travel view of memory cha lenges the central status given to the causal connection condition in a theory of memory.As I discussed above, in the mental time travel view, the primary function of memory is not to remember the past (see Su dendorf and Corba lis, 1997;De Brigard, 2014;Michaelian, 2016b).But, if that is the case, then it is hard to see why we should endorse the CTM.There are multiple reasons to think so.One reason is that, as Michaelian (2016b, p. 111) notes, because other forms of mental time travel need not have such a causal connection to past experiences, there is no principled way to require it in the case of memory.This does not mean, of course, that there cannot be such a connection, but only that it is not necessary.
Another reason is that, from the per ective of the mental time travel view, straightforward occur ences of remembering would be ruled out by the CTM.The causal connection a lows us to preserve the intuition that, in cases such as Kent's, subjects do not count as remembering.However, intuitively we do not seem to require that all occurrences of remembering preserve an ap ropriate causal connection to past events.Consider the fo lowing case.Imagine that I experienced my tenth birthday party in the past and that I now have a putative memory of it.I remember my friends and family being there and I remember having chocolate cake.However, sup ose that my cur ent representation is not being caused by my previous representation of my tenth birthday party, but rather by two different experiences that involved the relevant elements of my cur ent representation.In this case, the content of my cur ent representation is partly derived from, say, my experience of my ninth birthday party, which was attended by the same individuals, and partly derived from my experience of another party that I attended, where there was a chocolate cake.In this case, there is no causal connection of the sort required by the CTM, but it seems too stringent to say that the subject is not remembering the relevant event only because the content of his cur ent representation is not derived from the content of the original experience. 10 third reason why the mental time travel view challenges the CTM is that the latter is incompati le with the constructive chara er of mental time travel.Because mental time travel is in the service of simulating events to assist subjects in future intera ions with the environment, it seems too restrictive to require that our representations of the past have to draw content from only one sin ular source.For example, in thinking about how I should act in my job interview next week, my cur ent representation of the past wi l benefit more from drawing on different past experiences of job interviews than drawing on only one sinular experience. 11n sum, the CTM has occupied a central position in philosophical theorizing about memory for the past fifty years.Besides providing an analysis of remembering that accounts for a wide range of cases, it provides a useful principle to conceive of a taxonomy of remembering.However, if the mental time travel view of memory is right, the centrality of the CTM might not be war anted.

Mental time travel and our knowledge of the past
One direct consequence of abandoning the causal condition can be seen in the epistemology of memory.Because the causal condition is no longer necessary to remember, there is no uarantee that the content of our cur ent representations derives from the content of our past representations.That being the case, the question that poses itself is whether, and if so, how, we can form knowledge of what hap ened in the past on the basis of our cur ent representations.Is mental time travel capa le of providing such knowledge?Before I turn to this question, it is important to distin uish between two senses in which it can be asked.On the one hand, we can ask the pragmatic question of whether memory provides us with information that, in pra ical contexts, a lows for useful inferences about how things were in the past.Ca l this the pragmatic e iste ic question.On the other hand, we can ask whether memory actua ly provides knowledge of the past, in the sense that it serves as grounds for our justified beliefs about it.Ca l this the st ict e iste ic question.
This distinction is important because a positive answer to the pragmatic epistemic question does not necessarily give us a positive answer to the strict epistemic question.It might be the case that the content of my memory of my tenth birthday party is the same or very similar to the content of the memories that other people have of this event, such that I can make useful inferences about the event in relevant contexts, but it does not fo low from this that my memory a lows me to know anything about this event.An answer to the strict epistemic question, in contrast, requires identifying what makes it possi le that our cur ent memories serve as grounds for our justified beliefs about the past.
The causal condition provides an answer to the strict epistemic question.Because the content of my cur ent representation of an event is caused by my past representation of it, the causal connection makes it possi le for memory to ground my knowledge of the past.Otherwise put, the beliefs that we form on the basis of memory are justified because there is an ap ropriate causal connection between memories and past events.However, if, as the mental time travel view su gests, this condition is not necessary for remembering, how can we explain the relationship between the content of our past and cur ent representations?
It is not entirely clear what the alternatives for defenders of the mental time travel view are here.In fact, because he is the most systematic critic of the causal condition, Michaelian (2016b) has been the only one so far to provide an explicit treatment of the question.His ap roach consists in adopting a broad reliabilist framework in epistemology, according to which "the epistemic status of a belief is determined by the reliability of the process that produced it" (Michaelian, 2016b, p. 39; see also Goldman, 2012).Roughly, the idea is that one is justified in holding a certain belief if that belief was produced by a relia le process.On Michaelian's proposal, then, we can explain why memory serves as grounds for forming knowledge of the past in terms of the reliability of its underlying processes.This solution, however, wi l not be ap ealing if one is not already inclined to a form of reliabilism.The reason is that, as Michaelian (2016b, p. 40) recognizes, it takes reliabilism as a starting point and then proceeds to explain how memory is relia le.However, if one is skeptical of the idea that reliability itself can provide an account of epistemic justification, an account of how memory is relia le wi l not suffice to a dress the strict epistemic question.
The question of whether reliabilism is a good account of epistemic justification is beyond my scope here.However, given the question at hand about how memory can form knowledge about the past, it might be useful to explore other alternatives.One possi le ap roach might be to adopt an eternalist view of events (e.g., Bernecker, 2008).According to eternalism, past events do not cease to exist when they become past.Eternalism is promising because it a lows one to say that past events are constitutive parts of memories.To see this, consider an analogy with perception.Relationalists about perception claim that mid-sized objects are constitutive parts of perception, in the sense that I could not have a visual experience of the chair in my office if this object were not there (see, e.g., Campbe l, 2002;Martin, 2004;Brewer, 2007;Fish, 2009).An important motivation for acknowledging the constitutive role played by objects in perception is that it a lows one to explain how they ground our knowledge of the world (see Sche lenberg, 2016, for a recent discussion).Similarly, it might be ar ued, acknowledging the constitutive role played by past events in memories a lows one to explain how they ground our knowledge of the past. 12ternalism faces important pro lems.It is not obvious, for example, how our memories can be constituted by events located in a different spatiotemporal location.While it makes room, at least in principle, for that relation to take place by recognizing the existence of past events, an account of how they relate to our cur ent mental representations is sti l required.The pro lem is that it is hard to see how such an account would look like.Another pro lem for eternalism is that it requires us to pay a high metaphysical price to account for how remembering grounds our knowledge of the past.Because we are required to postulate the existence of past events, some might view this solution with skepticism (e.g., Michaelian, 2016b, p. 63).
Another alternative, which I sha l ca l the pragmatist solution, is to deny that the pragmatic epistemic question is different from the strict epistemic question.On such view, having knowledge about the past is simply a matter of making useful inferences about how things were back then.Whether or not we have knowledge of the past, the pragmatist wi l say, depends on how our memories can inform our future behavior.If memories a low for behaviors that lead to coordinated a ion with other individuals in relevant settings, such as discussing who attended your birthday party, or more primitively, discussing where food can be found, then that is a l that is required to say that we have knowledge of the past.The pragmatist wi l deny, therefore, that there needs to be, necessarily, a causal connection to past representations, as long as the cur ent representations a low for useful inferences about the past.
The pragmatist solution also faces important pro lems.The first pro lem is similar to the one raised above to reliabilism.In other words, it wi l only look ap ealing for those who are already inclined to a pragmatist view in epistemology.The second pro lem is that the pragmatist solution seems arbitrary, in the sense that it seems to imply that our knowledge of the past depends on what certain individuals "agree" to be the case.However, it is not clear who the relevant individuals are in each situation, or even if there is a principled way to identify them.Moreover, the focus on usefulness might lead to counterintuitive results, for a memory might be useful to uide the cur ent behavior of different individuals without being true of the past.In other words, it is completely plausile that subjects might misremember some or a l details of an event in a similar way, such that their memory reports agree with each other, but nonetheless fail to effectively describe what hap ened.
To conclude this part, it seems that an account of how we form knowledge of the past according to the mental time travel view might require some controversial commitments.While these commitments might take place in different domains -e.g., in metaphysics, as in the eternalist solution, or in epistemology, as in the reliabilist and the pragmatist solutions -a convincing answer to this question wi l inevita ly require a proper motivation of those commitments.

The objects of mental time travel
The mental time travel view of memory also raises important questions about the objects of mental time travel.If memory is only one form of mental time travel, then an account of the objects of memory wi l inevita ly depend on a more general account of the objects of mental time travel.Traditiona ly, philosophers have a dressed the question of the objects of memory in quite some detail.Inspired by Hume (2011) and Locke (1975), re resentational or indirect realist views hold that the objects of memory are internal representations of events (see, e.g., Russe l, 1921;Byrne, 2010).Relational or direct realist views, in contrast, say that the objects of memory are the past events themselves (see, e.g., Reid, 2000;Laird, 2014;Russe l, 2001;Debus, 2008).Given this framework, one natural su gestion here to address the question of the objects of mental time travel would be to take one's prefer ed account of the objects of memory and ap ly it to mental time travel.However, this seems to get things backwards.On the mental time travel view of memory, the mental time travel category is more basic than the category of memory, so we first need an account of the objects of mental time travel, which wi l only then inform our account of the objects of memory.
The question of the objects of mental time travel has not been a dressed in the literature so far.So, there are no e a lished views about it.However, this should not prevent us from thinking about how an answer to the question might look like.One way to start a dressing it is to distin uish between different forms of mental time travel.Although this is not always made explicit in discussions on the subject, there is more than one way in which mental time travel into the past and into the future can hap en.Besides episodic memory, which refers to mental time travel to past events that occur ed, and episodic future thinking, which refers to mental time travel to events that might occur, we may also think about counterfactual events located in subjective time (see De Brigard, 2014).For example, I can think about how my life would be right now if I had not gone to co lege.In this case, I am thinking about an event that could have hap ened in the past and that would influence the present, but that is no longer possi le.Similarly, I can think about how my life would be in ten years if I had not gone to co lege.In this case, I am thinking about an event that would be the case in the future if some other event in my past had been different.In both cases, then, I am entertaining thoughts about counterfactual situations oriented to the past and to the future.
The above su gests that an account of the objects of mental time travel needs to take into account not only episodic memory and episodic future thinking, but also forms of episodic counterfactual thought (see De Brigard, 2014) directed to the past and to the future.This makes the initial question significantly harder, for now we have to explain how things that can no longer be the case can somehow be the objects of our thoughts.One promising line of investigation might be to ap eal to the notion of intentional objects.Intentional objects, as origina ly introduced by Brentano (2014), are non-existent objects which are the direct objects of awareness of the mind.Although this is a promising line, no one has pursued it systematica ly as of yet. 13 Another alternative might be to look at the traditional accounts of the objects of memory as starting points.While relational views have been defended more consistently in the context of memory, they do not seem to offer promising pro ects for a more general account of the objects of mental time travel.The reason is that the objects of mental time travel, except ar ua ly for the objects of memory, do not exist, which makes it impossi le for us to be related to them.So, unless one is wi ling to commit to more controversial metaphysical views, such as the view that there are intentional objects (e.g., Crane, 2001Crane, , 2013) ) or some form of modal realism (Lewis, 1986), it is not clear whether relational views can be coherently sustained.In contrast, representational views might be more promising.Because the objects that are represented by mental time travel need not exist to be represented, there is no need to wor y about the metaphysical status of those events.What is relevant to explain how we are aware of the relevant events is the existence of representations, which would serve as proxies for the events.It is not clear, however, what the pro lems for a representational account of the objects of mental time travel would be.Since this question has not been explored in enough detail, it remains to be seen whether representationalism can stand up to a more detailed analysis.

The metaphysics of mental time travel
The consideration of the questions above fina ly puts us in a position to consider a more general question about the metaphysics of mental time travel.As we saw, the mental time travel view of memory raises a lot of different issues regarding the epistemology and the metaphysics of memory.But how pressing those questions are wi l depend on how one sees the category of memory in relation to the broader category of mental time travel.Until now, I have taken for granted that there are good reasons to accept that memory is just another occur ence of mental time travel.However, some philosophers have resisted this view.Debus (2014), for example, ar ues that memory and future future-oriented mental time travel -or what she ca ls sensory imagination -are occur ences of different kinds because there are important metaphysical dissimilarities between them.
The debate about the metaphysics of mental time travel is sti l very recent and, as with some of the other questions above, there are not we l-e a lished views in the literature.Despite this fact, I wi l fo low Per in and Michaelian (2017) here and distin uish between continuist and discontinuist metaphysical views of mental time travel.Continuists accept that the similarities between memory and other forms of mental time travel sup ort the more general view that they are occur ences of the same kind.Discontinuists, in contrast, say that those similarities are not enough to say that memory and other forms of mental time travel are occur ences of the same kind.
Reasons for endorsing continuism vary.The general motivation, though, seems to stem from different strands of research in the empirical sciences.As I discussed in the section "Episodic memory and mental time travel" , there is a great variety of empirical work that highlights important similarities between episodic memory and mental time travel.Perhaps the most distinctive motivation comes from the fact that mental time travel into the past and mental time travel into the future draw on very similar cognitive resources, which su gests that a common or "core" cognitive mechanism responsi le for mental time travel wi l be eventua ly identified (A dis et al., 2007;Scha er et al., 2007Scha er et al., , 2012)).In more philosophical terms, then, we can see continuism as relying on a more naturalistic stance towards the question of the relationship between episodic memory and mental time travel.In other words, for continuists views, because there is a lot of different empirical evidence su gesting that episodic memory is just another occur ence of mental time travel, we should take this evidence seriously when thinking about the metaphysics of mental time travel.
Discontinuist views, in contrast, seem to be motivated by more general a priori considerations about the metaphysics of mental time travel.This is not to say, of course, that discontinuists simply ignore the empirical evidence on which continuism relies. 14Instead, they believe that other considerations, such as whether mental time travel e a lishes an ap ropriate causal connection to the events in question, are also important to provide an ap ropriate picture of the metaphysics of mental time travel.Debus (2014), for example, arues that episodic memory and other forms of mental time 13 See, however, Crane (2001Crane ( , 2013) ) for potentially helpful discussions about intentional objects in philosophy of mind. 14See, for example, Perrin (2016) for a more modest discontinuist view that takes into account the similarities highlighted by empirical research.
Filosofi a Unisinos -Unisinos Journal of Philosophy -19(1):52-62, jan/apr 2018 travel are occur ences of two fundamenta ly distinct kinds. 15o sup ort this claim, she says that, unlike episodic memory, other forms of mental time travel fail to put subjects in an experiential relationship with the relevant events.The notion of an experiential relationship is a technical one, which refers to the causal and spatiotemporal relationship that subjects have to the events that their thoughts are about.In episodic memory, this relationship obtains because the relevant events did occur and we can, at least potentia ly, draw the causal connection between the cur ent memory and the past event.In other forms of mental time travel, in contrast, the relationship does not obtain because the relevant events do not exist.
Besides reflecting different metaphilosophical attitudes towards the same question, the dispute between continuism and discontinuism reflects different commitments taken in relation to the questions discussed in previous sections.Consider the question of whether episodic memory requires an ap ropriate causal connection to past events.While continuism is compati le with the CTM, it does not give the causal connection condition a central place in its metaphysical theorizing of mental time travel.For continuism, the presence (or the absence) of a causal connection reflects, at best, only a difference of degree between episodic memory and other occur ences of mental time travel.For discontinuists, however, this question is central to the metaphysics of mental time travel.The presence (or the absence) of a causal connection is sufficient to separate two mental occur ences as being of two different kinds.
The same ap lies to the question of our knowledge of the past and the objects of mental time travel.For continuists, like Michaelian (2016b), a proper account of how episodic memory provides us with knowledge of the past can be given by looking at the reliability of the mechanisms that produce memories, which, in turn, do not require causal connections to the past.Thus, the things that make us aware of past events are ar ua ly the internal representations, which are detacha le from those events.In this sense, continuists might be more inclined to adopt a representational view of the objects of mental time travel.For discontinuists, in contrast, episodic memory is capa le of providing subjects with knowledge in a way that other forms of mental time travel cannot.This is because it puts us in a relationship to past events, which necessarily involves a causal connection to them that is not possi le by means of other forms of mental time travel.Thus, discontinuists might not be satisfied with a representational view of the objects of mental time travel, as representations of events can occur in the absence of causal connections to the relevant events.A direct realist or relational view of memory (see Debus, 2008) wi l, therefore, seem more ap ealing for discontinuists, which Debus (2014) recognizes to be central to her discontinuist account.

Conclusion
The view that memory is a form of mental time travel offers exciting pro ects for new research in the emerging subfield of the philosophy of memory.Traditional views of memory, such as the causal theory of memory, and traditional questions about memory, such as how it provides knowledge of the past and what is the nature of its objects, need to be reconsidered in the broader framework of mental time travel.These questions, however, are inter-related with more general and new questions that arise only in the context of the research on mental time travel, i.e., what the objects of mental time travel are and what is the metaphysical status of those mental states.Thus, the intersection of the philosophy of memory and research on mental time travel not only provides new per ectives to think about traditional questions, but also new questions that have not been explored before.

( 1 )
S represented e in the past (Pa re resentation condition); (2) S has a cur ent mental representation of e (Current re resentation condition); (3) The content of the cur ent mental representation of e is sufficiently similar to the content of the past representation of e (Content condition); (4) There is an ap ropriate causal connection between the cur ent representation of e and the past representation of e (Causal connection condition). 6