R2.1. Theatrical concerns

Although generally commending the Theater hypothesis, Hurford et al. query the role of the broadcaster (see also Sterck & Dufour). By adding the broadcaster to the metaphor, they suggest, we preclude possible research on episodic memory in nonhumans. The broadcaster can be most clearly identified with language, but we did mention the possibility that nonhuman animals might in principle be able to communicate past episodes (e.g., by using mime). In any case, we did not mean to imply that the broadcaster was a necessary component of the theater metaphor; we wrote, “Although language is not necessary for mental time travel, it provides the clearest evidence of it” (sect. 4.7).

The causal relation, though, may run the other way. That is, mental time travel may have been critical to the evolution of the broadcaster. Indeed, as we hinted in the target article, language seems exquisitely adapted to the description of events that are remote in place and time. So it is that language comprises symbols to represent nonpresent objects, actions, emotions, and the like, and has systems of tense for signaling past and future (see Corballis & Suddendorf, in press, for further discussion). It has become commonplace to refer to one component of memory as declarative memory; conversely, we suggest, much of language consists of memorial declaration. It also consists of declarations about the future, as evident in recent discussions about the consequences of global warming.

Dessalles, though, seems to think that the broadcaster may have come first, arguing that episodic memory evolved as an outgrowth of language (see also Nelson), providing us with worthwhile things to talk about. This seems to us the wrong way around, rather like suggesting that feet evolved to allow us to play football. It is surely more likely that language evolved to allow experiences and plans, and also knowledge, to be shared. Dessalles' position, moreover, appears to be somewhat circular: We can tell of events because they are tellable – hence episodic memory evolved in support of language performance. Although there may be individual differences in the extent that people disclose, we suspect that every clinically normal person has memories and anticipations that they do not share with others. In fact, we think it is at times very difficult to put into words what has happened or what we expect to happen. And although telling exciting stories of the unusual and surprising may have social (and perhaps sexual) pay-offs, it surely does not stop people from remembering the mundane and even talking about it. I well remember getting on the bus last week, but it is not a story I would tell. We agree with Dessalles that this is so because it conveys little new or unexpected information. But why is unexpectedness important? Nelson claims that it is only humans who share events they have experienced. We concur, and this broadcasting of one's mental time travel can indeed be extremely adaptive, in that it allows people to learn from others' mistakes and successes without even having to experience the events themselves. Those who tell such useful tales may be encouraged and rewarded by those who learn. Thus, these are stories worth telling and worth listening to.

Whereas Hurford et al. think the Theater hypothesis goes too far, Ainslie suggests that it does not go far enough, since it neglects the audience. In so doing, it fails to consider the question of what motivates us to indulge in mental time travel rather than live in the present. Questions about motivation for mental time travel are perhaps separate from questions about mental time travel itself, just as the description of a theater production is not a description of who will go to see it. Nevertheless, Ainslie raises interesting points. He notes that people have little tolerance for a bad lecture – or reading a boring article, we might add – and it may be precisely in such circumstances that they will indulge in some future fantasy, or revel in some past glory. Buckner and Carroll (Reference Buckner and Carroll2007) suggest that this may in fact be the human default mode. Even so, different scenarios must compete with each other and with present needs. Ainslie suggests that they may compete in terms of entertainment value and suggests that a primary motivator may be the very uniqueness of scenarios that we can envisage; that uniqueness may in turn result from our ability to use combinatorial principles to generate novelty (see sect. R2.5), and also perhaps from what Locke and Bogin (Reference Locke and Bogin2006) recently termed the “instinct for inventiveness” (see also Whiten & Suddendorf [in press] on the evolution of inventiveness).

R2.2. The role of emotion

D'Argembeau & Van der Linden offer some detail on the role of emotion in mental time travel. Their interesting work has documented links between past and future time travel (Janus hypothesis), adding depth to the study of its content. They suggest that in addition to enhancing future fitness, mental time travel serves to regulate emotion and create the sense of self. We suspect it has many other functions as well. In fact, many human emotions, such as remorse and hope, may fundamentally depend on mental time travel. Individual differences in the emotional content of imagined episodes may underlie various mental disorders (see also Brüne & Brüne-Cohrs); thus depressed people generate fewer positive episodes; and anxious patients, more negative episodes (MacLeod & Byrne Reference MacLeod and Byrne1996). On the other hand, they note that people with stronger visual imagery generate more detailed event representations and rate these as more meaningful (D'Argembeau & Van der Linden Reference D'Argembeau and Van der Linden2006), and they ask whether this may also be associated with a stronger ability to make adaptive choices. There is reason to suspect that this might be so, as other work has found that individuals with stronger concern for the future show better health-preserving behavior (e.g., Mahon et al. Reference Mahon, Yarcheski and Yarcheski1997). However, this comes at a price, as they are also more prone to stress (Zimbardo & Boyd Reference Zimbardo and Boyd1999). As Ainslie points out, there are serious problems associated with being, like Walter Mitty, too “fantasy-prone.” It is easy to see how the adaptive ability to concern oneself with the future confronts humans with all kinds of cognitive, motivational, and emotional complexities and dilemmas (e.g., Suddendorf & Busby Reference Suddendorf and Busby2005). For example, one may struggle to stop thinking about the same events (as in posttraumatic stress syndrome), or worry about events that may never occur. Brüne & Brüne-Cohrs make the interesting suggestion that obsessive-compulsive disorder might be the result of too much mental time travel activity of this latter kind (Brüne Reference Brüne2006).

Sometimes, though, the sense of time can be seen as a solution rather than a dilemma. Viola in Shakespeare's Twelfth Night, masquerading as a man, finds herself in an impossible predicament, and is moved to say

We often live in the hope that problems will simply go away.

According to Atance & Meltzoff, present affective state is not so much a consequence of mental time travel as a cause of it. They suggest that mental time travel arises because we experience some desire for a future goal in the present, and work toward it by constructing future scenarios. This is similar to the point raised by Ainslie that we need to consider what motivates a person to indulge in future thinking in the present, or to think about one scenario rather than another. Emotional considerations are no doubt important (e.g., D'Argembeau & Van der Linden), but it is not clear to us that mental time travel is always suffused with emotion, and the role of emotion in memory has a controversial history. Some of the things we remember seem banal and without obvious affect (see also our response to Dessalles), and they may serve more as part of the vocabulary from which to construct future scenarios.

R2.3. About time

Hurford et al. suggest what they call a “radically different” taxonomy from the Prospection-systems hypothesis we summarized in our Figure 1, although their alternative ideas do not really impinge on the notion of mental time travel itself. They see no distinctions between semantic memory and (semantic) prospection, or between procedural memory and procedural prospection, since both semantic and procedural memory are essentially timeless. Semantic memory is indeed timeless in the sense that it does not record and allow us to re-experience a specific event that occurred at a specific point in time, but we can of course have semantic memories that refer to time. We can know, for example, when we were born, or the date at which Columbus discovered America. Further, some semantic knowledge is clearly directed toward the future, such as knowledge of an upcoming birthday party, and influences our planning as much as the episodic constructions that may guide the specific form of the celebrations. As we noted in the discussion of the reminiscence bump in the target article, with increasing age we may rely less on mental time travel and increasingly on the semantic system. The notions of past and future can be embedded in semantic knowledge as well. We suggest, though, that the semantic representation of time itself depends on the initial emergence of mental time travel, both in evolution and development.

This possibility receives some support from Friedman, who rightly points out that our sense of the time at which events occurred in the past is fragmentary and generally based on inference rather than on direct experience (see also Eacott & Easton). He thus suggests that episodic memory should be understood as having to do with events that have occurred on a particular occasion, rather than at a specific time. In a sense, then, our concept of time is an extrapolation from incomplete data. Nevertheless, we certainly understand time to be a continuum, running smoothly from some infinite past (or perhaps from the big bang that created the universe) to a future that is hopefully distant, global warming or not. Time is therefore woven into our semantic knowledge about the world. But such notions, we argue, can only emerge from an initial understanding of a contextualized, event-filled time (cf. Gerrans Reference Gerrans2007).

Merker suggests that our “entire argument is built on a perfect symmetry between past and future,” despite our having written “There is a fundamental causal asymmetry [between past and future], and one simply cannot know the future as one knows the past”. More seriously, perhaps, he suggests that semantic prospection is all that is needed for optimal future planning (cf. Hegdé) and asserts that we have provided no concrete evidence that mental time travel ever yields greater efficacy. As noted earlier, mental time travel may be necessary to construct such semantic prospection. Furthermore, his view of prospection seems too passive, as though mental time travel into the future is little more than a guessing game. Thus he suggests that we cannot know which one of several possible future events will actually eventuate, and that particularities gleaned from past episodes might make prediction worse by adding to the number of possible alternatives. On a more proactive view, though, those particularities can surely help us actually shape future events, as distinct from merely trying to anticipate what will happen.

R2.4. Beyond past and future

Mental time travel may be based on components that are also used to achieve other goals (the Components hypothesis). The Theory-of-mind hypothesis, for example, suggests that humans use the same mechanisms to simulate the mental states of others. This idea is not new (as Bischof-Köhler & Bischof and Sterck & Dufour seem to imply), but was a central argument in our original article (Suddendorf & Corballis Reference Suddendorf and Corballis1997). Other authors have also pointed to links between theory of mind and mental time travels into past (Perner & Ruffman Reference Perner and Ruffman1995) and future (Hazlitt Reference Hazlitt1805). Bischof-Köhler & Bischof cite additional developmental support for this hypothesis, and make the interesting case for taking the Gestalt psychological concept of “frame-of-reference awareness” as a perspective for understanding mental time travel. Buckner and Okuda provide additional reasons why theory of mind and mental time travel (and possibly some other forms of “self-projection” such as in navigation) might draw on the same components, and none of the commentators made a case for conceiving of mental time travel as an informationally encapsulated module. In the target article, we proposed that mental time travel draws on a range of sophisticated components (the Theater hypothesis) that may serve a range of other cognitive functions, including theory of mind.

The relation between mental time travel and theory of mind may be quite complex. On the one hand, both may draw on basic component abilities such as metarepresentation and executive functions (e.g., Suddendorf Reference Suddendorf, Corballis and ea1999), or “frame-of-reference awareness”; on the other hand, each appears to be intimately intertwined with the other. For example, to understand what oneself and others might do in the future typically requires some appreciation of different mental states. Conversely, mental time travels are mental states, and understanding others' minds often depends on understanding what happened to them, and what their plans and goals might be. The recursion evident in theory of mind (e.g., I believe that you know what she wants) is also evident in mental time travel (e.g., I remember that I thought yesterday about what you are going to do tomorrow).

R2.5. Combinations and recombinations

Although we proposed that past episodes provide a vocabulary from which to construct possible future episodes, using recursive, combinatorial principles similar to those involved in theory of mind and language, Hegdé misinterprets us as implying that “one can mentally create only those future events that one has specifically experienced in the past.” Like Merker, Hegdé also argues that the information provided by episodic memory is not needed and can even place “untenable limitations on one's ability to imagine.” He suggests that prospection is most efficiently accomplished entirely through the use of Bayesian principles (which he does not specify) based on generic prior knowledge, although he does concede that the Bayesian approach can use episodic memory without being dependent on it. We agree that imagining future episodes is based on both semantic knowledge and episodic memory, but suggest that Hegdé may have underestimated the constructive, combinatorial aspect of our account. We recognize that humans can imagine whatever, wherever, and whenever (Suddendorf & Busby Reference Suddendorf and Busby2003b), and we emphasized the generative nature of mental time travel both in the target article and in the original formulation:

This generativity may be an asset for flexible foresight, but it may serve less well for accurately recording events of the past, lending support to our argument (the Future-first hypothesis) that episodic memory may in fact be a design feature of the foresight system (cf., D'Argembeau & Van der Linden). Schacter & Addis reinforce our claim that the error-prone nature of episodic memory is the result of a system geared towards flexible recombination of future events, rather than towards faithful reconstruction of the past – and suggest we should call this idea the Constructive episodic simulation hypothesis (Schacter & Addis Reference Schacter and Addis2007a). The malleability of memory that we have known of since Bartlett's (Reference Bartlett1932) pioneering studies is often forgotten by researchers who use correct recall as the dependent variable. Several commentators (e.g., Ainslie; Schacter & Addis) agree with us that the veracity of episodic memory cannot be used to define it. Just as visual illusions hold vital clues to studying normal vision, so errors of memory also hold clues to understanding the system (see also Brüne & Brüne-Cohrs), as is well illustrated by Schacter's (e.g., Reference Schacter1999) seminal work on these “sins of memory.”

In the playwright section of the theater metaphor, we focused especially on the peculiar human capacity for recursion, which drives the recombination of elements not only in mental time travel but also in other spheres such as language, theory of mind, and music (Corballis Reference Corballis2007). This may have distracted from other aspects of the story line. Nelson therefore argues for the importance of narrative in mental time travel. We are sympathetic to her view (see also the discussion in sect. R4) and have acknowledged previously (Suddendorf & Corballis Reference Suddendorf and Corballis1997) Freud's observation that even memories that reveal themselves as images require a story grammar if remembrance is to be distinguished from random hallucinations. We thought the use of the theater metaphor itself would have gone a long way to emphasizing the importance of the story or narrative being played out.

R5.1. Are behavioral tests possible?

Raby et al. lament that we did not articulate our hypotheses in specific behavioral terms, and complain that we rely too heavily on phenomenology. They also argue that phenomenology may be the only difference between episodic and semantic memory. We do not think that this is the only difference. Clearly, loss of episodic memory in amnesic patients is not merely the loss of a phenomenological sensation, or quale, but has fundamental consequences for the sufferer. In any case, the charge that we are relying too heavily on phenomenology is misleading. We took pains here and elsewhere (Suddendorf & Busby Reference Suddendorf and Busby2003b; Reference Suddendorf and Busby2005) to move beyond first-person experience and towards objective indicators. Indeed, we pointed out that natural selection needs something to get its teeth into, and an internal feel alone is inadequate for that. We suggest that future-directed action is the key towards mapping animals' competence and understanding the evolution of mental time travel.

Of course, what Raby et al. want are predictions that are instantly translatable into tests with nonhuman species, such as scrub jays. Clayton and colleagues (e.g., Clayton et al. Reference Clayton, Bussey, Emery and Dickinson2003; Dally et al. Reference Dally, Emery and Clayton2006b; Raby et al. Reference Raby, Alexis, Dickinson and Clayton2007) have been prolific in developing clever ways of turning mental time travel questions into experiments involving jays caching and retrieving food. The components we discuss in the Theater hypothesis offer several new avenues, but Raby et al. dismiss components such as recursion and secondary representation that may not so easily be translated into this caching context. Yet, these are behavioral criteria that lend themselves to comparative inquiry. Suddendorf and Whiten (Reference Suddendorf and Whiten2001), for example, have shown that there is convincing evidence for secondary representation (the stage component) in great apes, and we see no reason why scrub jays could not be tested similarly. We thus reject the charge that we do not provide enough relevant behavioral criteria and are happy to suggest more.

Raby et al. further charge that we did not adequately explain why we advocate a focus on non-species-typical behavior in attempts to demonstrate mental time travel in animals. We made the case that the key adaptive advantage of mental time travel is the flexibility it offers for acting now to increase future survival and reproduction chances (the Flexibility hypothesis), and note that the human capacity is not restricted in content in any apparent way, but instead is open-ended and domain general. We thus argued that comparative research should focus on non-species-typical behavior because it (a) provides evidence for such flexibility and (b) rules out any role of non-cognitive behavioral predispositions governing future-directed behavior (as appears to be the case, for example, in hibernators). Scrub jays, like various other bird species, develop caching in predictable stages that suggest a prominent role of innate factors. For example, they have an early tendency to place food, as well as many other objects, without recovering these caches. Raby et al. counter that human children develop mental time travel also in a predictable fashion. However, it is not yet clear whether or not this is so. It has certainly been argued that this development is not uniform, but that there are individual and cultural differences in the way children first acquire mental time travel (see Nelson). More developmental research is required to settle this, and we maintain that studies with nonhuman animals (as well as with children) are best advised to employ novel contexts to ascertain that the behavior of interest is the result of an individual's faculty for mental time travel.

R5.2. New data on animal foresight

As we acknowledged in the target article, the Bischof-Köhler hypothesis cannot be true in the superficial sense that food caching is directed toward future gratification. Many species secure future needs, whether in food caching, nest building, or migration. These behaviors, though, are largely instinctive, and the hypothesis may still hold for “individual, flexible situations involving non-instinctive behaviors” (sect. 3.3 of the target article).

Several commentators (Bar; Carr & Viskontas; Hurford et al.; Osvath & Gärdenfors; Raby et al.; Schacter & Addis; Sterck & Dufour) refer to the study by Raby et al. (Reference Raby, Alexis, Dickinson and Clayton2007) as evidence that caching by scrub jays does display sufficient flexibility to refute the Bischof-Köhler hypothesis. In this study, eight scrub jays received breakfast in one cage and no breakfast in another cage, on alternating days. When later given the opportunity to store food while currently satiated, they cached more in the place where no breakfast was to be expected. In a second experiment, they were given different kinds of food in different locations, and later they cached more food of each particular kind in the location not associated with that food. These were taken to be clear cases in which a nonhuman species, currently satiated, cached food in anticipation of a future state of hunger or food preference. An alternative, perhaps, is that the birds based the caching, not on anticipated hunger, but on a general heuristic to balance food sources, based on previous experience of imbalance.

A closer look at the procedure raises some doubt about whether the birds were in fact acting with specific future hunger in mind. Because of counterbalancing, half the jays had received breakfast the morning of the test and the other half had not. Thus, four of the birds had reason to expect no breakfast the next day, whereas the other four should have expected to receive breakfast. Yet, the results for these two groups were combined, revealing only that, overall, the jays were caching more in the cage that was associated with no breakfast. To conclude that the jays were planning for the next morning rather than having acquired a more general rule, these two conditions would need to be compared rather than amalgamated.

Nevertheless this experiment does show flexibility beyond that expected of a purely instinctive behavior. If jays can be shown to act for specific future events, it will be interesting to see how far into the future they can “plan” and if they can also discriminate between multiple future events. In all this, however, there remains also the nagging suspicion that the flexible behavior associated with caching in scrub jays is restricted to the particular speciality of food storage. We agree with Shettleworth (Reference Shettleworth2007) that it remains to be seen whether this ability is available to jays in other contexts. As we noted previously, if the same animals pass different versions of such tasks (involving different contexts and needs), then this evidence for flexibility would make the Bischof-Köhler hypothesis increasingly untenable (Suddendorf & Busby Reference Suddendorf and Busby2005).

Some of the same commentators (Bar; Hurford et al.; Osvath & Gärdenfors; Sterck & Dufour) also referred to a study by Mulcahy and Call (Reference Mulcahy and Call2006), who claimed to show that apes save tools for future use, thereby refuting the Bischof-Köhler hypothesis. The limitations of this study were addressed in the target article, as well as by Suddendorf (Reference Suddendorf2006), and Hurford et al. praise our assessment. Sterck & Dufour, however, misunderstood our main critique to be that tool selection was based on current needs because the apparatus was visible, and report that in their own unpublished work, chimpanzees managed to select the right tool even when the apparatus was out of sight. But we already noted in the target article that Mulcahy and Call's “third experiment showed that the apes could pass the task even when they could not see the apparatus during tool selection”. The issue of need in this context is not about seeing the apparatus, but about the internal motivational state (see also Bischof-Köhler & Bischof), and we merely observed that these were neither controlled nor manipulated (and that it is likely that the apes desired grapes throughout).

Osvath & Gärdenfors point to another recent test of the Bischof-Köhler hypothesis in which Naqshbandi and Roberts (Reference Naqshbandi and Roberts2006) manipulated the supply of drinking water available to their subjects. They found that over repeated trials two squirrel monkeys (but not rats) gradually reversed their natural preference for four pieces of dates over one piece, when choosing four pieces resulted in provision of water after a long delay (3 hours), and choosing one piece was followed by water supply after a short delay (1/2 hour). Eating dates induces thirst, and subsequently not having access to something to drink for a while causes discomfort. Thus, one explanation for the gradual reversal of choice from four pieces of dates to one piece would be that the monkeys took their future thirst into account (i.e., evidence against the Bischof-Köhler hypothesis). As Shettleworth (Reference Shettleworth2007) noted, however, the repeated exposure to the conditions means that associative learning could possibly explain the results. We would add that gradual change in preference from four pieces of dates to one piece would not necessarily be the behavior we would predict from mental time travel. If the monkeys did think ahead about the best outcome, why did they not continue to select four date pieces and simply choose to eat one or two and keep the others until enough water was available?

Eacott & Easton report on clever experiments with rats in which they compare behavioral measures thought to tap recall with those based on familiarity. Fornix-;[?tjl]>dissected rats fail on the “recall” task, yet still succeed on a “recognition” task. This innovative approach is a nice illustration of a comparative neuroscientific approach. However, the distinction between recall and recognition is not equivalent to the distinction between episodic and semantic memory (e.g., Wheeler et al. Reference Wheeler, Stuss and Tulving1997). We can recall that Wellington is the capital of New Zealand, but that is because we know it, not because we can remember being told it. Thus, although the task may tap an important difference between procedural and declarative memory, no mental time travel may be involved. Nonetheless, comparing the neuronal implementation of analog behavior such as that between rats and humans, as Buckner submits, may hold vital information to clarifying the component processes of the Theater hypothesis (cf. our response to Brüne & Brüne-Cohrs in sect. R3).

R5.3. Continuity in animal competence

Hurford et al., in spite of agreeing that there is a large gulf between animal and human mental time travel, are not convinced that animals lack it entirely. They point to an anecdote from Kanzi and to the studies with the gorilla King, which they interpret appropriately cautiously. We are of course also not convinced and acknowledge that we cannot prove the Stuck-in-time hypothesis. We believe that there is something unique about human mental time travel (the Uniqueness hypothesis) and suggest that only extensive further study can determine what, if anything, that might be. The Traveler hypothesis and the Bischof-Köhler hypothesis are two proposals. In the target article we put forward the theater metaphor and suggest that various deficits in the components we discuss could limit mental time travel (Multiple limits hypothesis). For example, foresight may be severely constraint without a capacity for recursive thought to generate content (Corballis Reference Corballis2007; Suddendorf & Corballis Reference Suddendorf and Corballis1997).

Bar suggests that we should consider foresight as a continuum, with nonhuman animals having as much foresight as their environment and life require, and since animal lives are less complex, they need less foresight than humans. In response, we note first that the causality may well be the other way around: Mental time travel makes human lives more complex than those of animals. Second, the idea that foresight is based on a continuum runs counter to the case we made for the Prospection systems hypothesis. Bar's notion of foresight is equivalent to what we called “future-directed capacities.” However, there are reasons to distinguish between mental time travel and the various other kinds of future-directed capacities, such as instinctual, procedural, and semantic prospection. It may be possible to make a case for a unitary model of foresight, just as some scholars have done for a unitary model of memory. But such models have to explain why there are dissociations between different kinds of prospection and how they come about. For example, why does an amnesic patient, such as D.B. (Klein et al. Reference Klein, Loftus and Kihlstrom2002b), have selective impairment of mental time travel, but little problem providing semantic information about past and future?

Nonetheless, a case for quantitative differences in mental time travel could still be made (e.g., Carr & Viskontas; Hurford et al.). In fact, some of the proposals we discuss could be regarded as setting quantitative limits. The Bischof-Köhler hypothesis, for example, suggests a limit to the extent of foresight (i.e., anticipation of future needs). The difference between being and not being limited in such a way could of course amount to a qualitative shift – without future need anticipation, there is little point in considering any genuinely remote future events. The Multiple limits hypothesis suggests that species might be limited by a variety of other deficits that may have both quantitative and qualitative effects.

The clever experiments we discussed in the previous section suggest that nonhuman animals may have some competence, and future research may well whittle away at some suggested limits. There are perhaps two questions here (see also sect. R5.1). One is whether such phenomena remain tied to narrow specialties, such as food-caching in birds, or whether they can cover the broad range of contexts that characterizes our own ability to travel forward and backwards in time. The second is whether examples from nonhuman animals can display the generativity of our own ability to conjure imagined events. Bar suggests that even human mental time travel may reduce to associations, but this again neglects the inventive component of human imagination, just as animal communication fails to display the generativity of human language.

Sterck & Dufour provide further evidence about future-oriented capacities in chimpanzees, who can learn to hold a desired cookie for up to 8 minutes in order to exchange it for a larger one. Research with children (e.g., Mischel & Mischel Reference Mischel and Mischel1983) suggests that they can delay gratification for longer when the salience of the reward is decreased (e.g., when looking at a photo of the reward rather than at the actual reward), suggesting that the extent of the delay reflects inhibition or executive control, rather than the capacity to travel mentally to the future event per se. Still, this is a fairly impressive performance for chimpanzees compared to other animals. But these data do not challenge the Bischof-Köhler hypothesis, and it remains unclear how such relatively brief delay of gratification relates to the flexible capacity of humans to travel mentally to any future scenarios, be they a year or only a few minutes away. We agree of course with Sterck & Dufour that we should not prejudge animal competence, and we acknowledged that “it remains to be established what exactly the nature and extent of modern apes' foresight (and that of other animals) is” (target article, sect. 3.3, para. 7). We also stated that “a substantial body of comparative data is required before it can be concluded that any trait is uniquely human” (target article, sect. 3.3, para. 6).

We do, however, need to start with the null hypothesis that nonhuman animals do not have this faculty or the necessary components, and we encourage researchers to attempt to falsify these statements one by one. We think it remains important, though, to stay cautious and not be too quickly impressed by apparent positive results. We recommend keeping Occam's razor sharp. Note that negative findings are seldom published, and hence Osvath & Gärdenfors' observation that most studies have reported positive results for animal foresight may not be surprising. There are, of course, good epistemological reasons not to emphasize failures to reject the null hypothesis, but we might be missing out on instructive information about the systematic failures of nonhuman animals. We would therefore like to encourage comparative researchers to also investigate and report the limits of their subjects' capacities.

R5.4. The future of animal studies of mental time travel

In stark contrast to our call for more studies, Shettleworth thinks this is all heading in the wrong direction and suggests a moratorium on animal studies of mental time travel (at least until developmental psychology has made some progress). Shettleworth thinks our questions are flawed because we can never know what is on an animal's mind, and because it moves the field in non-productive directions. The first charge is easily dispelled. We can also never know for certain what even another human's mental experience is like (language or not), but that does not stop us studying the mind (though it did during the behaviorist era). With respect to the second charge, quite contrary to Shettleworth's assertion, recent work on mental time travel in animals has been very productive indeed (in addition to our review, see also Dere et al. Reference Dere, Kart-Teke, Huston and Silva2006; Roberts Reference Roberts2006; Zentall Reference Zentall2006). In fact, Shettleworth (Reference Shettleworth2007) herself proceeds to demonstrate this when discussing Raby et al.'s (2007) recent study. If these researchers had followed her suggestion, that paper would not exist. As Raby et al. acknowledge at the beginning of their commentary, our analyses have provided a theoretical framework for their work. Whether one agrees or disagrees with our proposals and interpretations, there can be little argument that the questions we have raised have sparked very productive research programs.

We agree with Shettleworth that questions like “what components of process X are and are not shared among what species and why, in evolutionary, functional, and perhaps neurological terms?” are productive, and indeed our target article raised many more questions than the Stuck-in-time hypothesis (e.g., Table 1). Since humans can travel mentally in time, the natural question of whether nonhuman animals can do so, too, can inspire the pursuit of more detailed analyses of animal capacities. Raising such big questions about consciousness and its evolution can be very useful. Consider the case of theory of mind, a field instigated by Premack and Woodruff's (Reference Premack and Woodruff1978) seminal Behavioral and Brain Sciences article in which they searched for an existence proof. The article sparked a flurry of comparative research that has led to real progress. In the process, it kick-started perhaps one of the most productive research areas in developmental psychology and galvanized research in neuroscience, philosophy, and other disciplines. Would it be more productive to retreat to questions that dare not mention mind or consciousness for fear of a return to anthropomorphism?

Table R1.

Key hypotheses discussed in the target article

Of course, we advocate that comparative psychologists be cautious with their interpretation – after all, we have been promoting such caution in regard to claims about mental time travel. But that does not mean we would want to abandon these efforts. There are various reasons why such questions are popular, as Shettleworth (Reference Shettleworth2007) herself noted: “These include Donald Griffin's exhortations to ethologists to tackle animal consciousness, legitimisation of the study of consciousness in the cognitive sciences generally, and the prospect of understanding the neural and genetic bases of conscious processes using ‘animal models’” (p. 826) Whereas the first point merely offers a glimpse at the author's position, the latter two are very compelling reasons to pursue these research questions, we think. We may learn a lot about cognition and the brain in the process (see Buckner; Carr & Viskontas; Eacott & Easton; Hurford et al.; Tulving & Kim).

Another good reason to study what is unique about the human mind, and what is not, is that this can in various ways inform us about the evolution of the human mind (Suddendorf Reference Suddendorf2004). In this context, convergent evolution of some capacities (as perhaps in jays) may be less relevant than the potentially homologous abilities of our closest living primate relatives.