Suddendorf & Corballis (S&C) have done a great service to the scientific community by introducing the concept of mental time travel as an integrated process in their 1997 paper (see Suddendorf & Corballis Reference Suddendorf and Corballis1997) and in restating it now in the present article. This concept has provided a background theoretical framework for our own research on memory and future thinking in animals, which we would like to acknowledge.

Here S&C make a comparative hypothesis between the abilities of human and nonhuman animals with respect to mental time travel. However, they do not articulate this hypothesis in terms of specific behavioural criteria; and without such criteria, it is not possible to empirically evaluate their claim that mental time travel is unique to humans, because, in the absence of language, the cognitive capacities of animals can only be inferred from their behaviour. Some of the components of mental time travel (MTT) that S&C identify in their theatre metaphor, for example, autonoesis and a sense of subjective time, are intrinsically phenomenological and hence their presence or absence is impossible to demonstrate in a nonhuman animal. Other criteria such as recursion or secondary representation may be possible to test empirically in other contexts but not, we think, in considering behaviour demonstrating mental time travel. We also question whether all the criteria listed – recursion, causal understanding, and rehearsal, for example – are essential for mental time travel as opposed to being skills that humans have and use incidentally while engaged in MTT.

The type of criteria we have in mind can be illustrated by our own analysis of “episodic-like” memory. We used Tulving's original definition of episodic memory in terms of spatio-temporal relations (Tulving Reference Tulving, Tulving and Donaldson1972) to investigate whether western scrub jays have “what, where, when” (www) memory. We subsequently proposed additional behavioural criteria for “episodic-like” memory; namely, that the memory must not only be shown to have www content but also to have an integrated structure and that the information can be deployed flexibly (Clayton et al. Reference Clayton, Bussey and Dickinson2003). Although our term “episodic-like” acknowledges the difficulty of interpreting behaviour in another species as a manifestation of a specific phenomenological experience of remembering, the important point here is that the what, where and when must be bound together in order that the animal can discriminate one event from another. For this reason even a ‘yes’ in every box in Table 1 of the target article is not sufficient to establish the episodic-like character of memory in the absence of the criteria of integration and flexibility.

Importantly, S&C do offer an empirically tractable criterion for foresight, the Bischof-Köhler hypothesis. This is the claim that an animal cannot take action for a future motivational state, and we have recently used this criterion to evaluate the MTT hypothesis by investigating whether scrub-jays who are not hungry cache more food in a place in which they are likely to experience hunger the following morning than in a place that they are not. Not only did they meet this criterion of MTT, but they also cached more of a type of food that they will not receive in a particular place the following morning than of a type of food that they will (Raby et al. Reference Raby, Alexis, Dickinson and Clayton2007). In Shettleworth's view this is the first unambiguous example of future planning in an animal (Shettleworth Reference Shettleworth2007).

What we do not know, and think is not testable empirically, is whether this type of future planning depends upon episodic personal projection or semantic knowledge. S&C are inclined to dismiss the semantic cognitive system as self-evidently inferior and more limited than the episodic cognitive system with a number of statements that there is no room to debate here. A contrasting view is that the only distinguishing difference between episodic and semantic memory is the phenomenological sensation of remembering rather than knowing about the event. The semantic memory system allows individuals to acquire information about themselves and their world through different sensory modalities and flexibly retrieve and use this information. This includes personal information and autobiographical facts. We agree with S&C that there is evidence that the episodic and semantic memory systems are closely linked to their respective future thinking counterparts and suggest that the only significant difference between episodic and semantic future thinking is the sensation of pre-experiencing one's own personal future. Humans may always combine episodic and semantic future thinking in assessing alternatives, but this does not prove that the episodic element is essential for future thinking. Using one of their own examples, it is quite possible to semantically arrange a seating plan for a forthcoming event without needing mental time travel to envisage oneself at the table.

S&C argue that a demonstration of mental time travel in nonhuman animals must exclude behaviours that are species typical or domain specific, without adequately explaining why. They describe a possible test of episodic-like prospection in scrub-jays and then immediately dismiss it as explicable by “predispositions and specific learning algorithms,” although, in their view, the use of language by humans – surely a case of predisposition and learning if ever there was one – “provides the clearest evidence of [mental time travel].” Most behaviours in any animal, cognitive or otherwise, are likely to be species typical and a result of some form of predisposition and learning. They use a different yardstick in assessing children than in assessing animals. Caching and retrieval may develop in a predictable fashion in scrub-jays, but so does the emergence of autonoesis, theory of mind, and indeed the episodic cognitive system in children.

In relying so heavily on phenomenology, S&C effectively define mental time travel by the exclusion of animals other than humans. If they are going to propose a comparative hypothesis it is incumbent on them to offer relevant, applicable, behavioural criteria which are empirically tractable. In the absence of such criteria, the comparative aspect of MTT is not a matter of scientific enquiry but of intellectual preference.