Temporal information plays a central role in discussions of the nature of episodic memory (EM) (e.g., Tulving Reference Tulving, Tulving and Donaldson1972; Reference Tulving1984; Reference Tulving2002b) and mental time travel (MTT) (target article). For this reason it is important to analyze the meaning of time in EM and MTT and to consider psychological research about memory for time and ways of thinking about the future. Among the different types of temporal information that humans and animals could process are: when an event occurred (or is expected to occur) within some time pattern (“temporal locations”), how long ago an event occurred (“temporal distances”), and before–after relations and other relations of the order of events (see Friedman Reference Friedman1993). Multiple representations and processes are involved in humans' memory for the times of past events and in their thinking about the times of future events (Friedman Reference Friedman, Hoerl and McCormack2001; Reference Friedman and Kail2003). The impression we sometimes have that time is a seamless, linear continuum is at odds with the findings of research on temporal abilities and the processes that underlie them (Friedman Reference Friedman1993). For example, memory for time is often inaccurate, systematically distorted, and even inconsistent with remembering time in an integral way – we sometimes remember the time of day of an event but not the month or year.

Adults' sense of the times of past events depends mainly on inferring when the event must have happened by relating the content of the memory to one's general knowledge of personal, conventional, and natural time patterns. But adults, as well as children as young as 4 years, also have available impressions of the ages of events that provide limited information about their distances in the past (Friedman Reference Friedman and Medin1996; Reference Friedman, Hoerl and McCormack2001). A differentiated sense of the future depends on mental representations of time patterns, probably supplemented from early childhood onward with propositions that are active in memory (e.g., that particular events are coming soon or won't happen for a long time) (Friedman Reference Friedman and Kail2003). This patchwork of processes, and the fact that humans remember and anticipate times separately on multiple time scales, reveals the complexity of memory for and anticipation of the times of events.

What temporal abilities are necessary to possess EM? At various stages in the development of his theory, Tulving referred to the following as critical features: temporally dated events (though not in conventional time units; Tulving Reference Tulving, Tulving and Donaldson1972), coding the temporal relations among experienced events (Tulving Reference Tulving, Tulving and Donaldson1972; Reference Tulving1984), and having a subjective sense of time (Tulving Reference Tulving2002b). Others have pointed to different time-related criteria (e.g., the ability to discriminate recent from remote events [de Kort et al. Reference de Kort, Dickinson and Clayton2005] or the capacity to replay the flow of experience [Eichenbaumet al. Reference Eichenbaum, Fortin, Ergorul, Wright and Agster2005]), or they have maintained that temporal information is not necessary (Suddendorf & Corballis [S&C] in the target article; Zentall Reference Zentall2005). In light of research on memory for the times of events, S&C's and Zentall's positions may be the wisest, at least for describing human memory. There is no evidence that events are automatically coded by the times of their occurrence or that memory is temporally organized (Friedman Reference Friedman1993; Reference Friedman2004); many older events are difficult to discriminate by their ages (e.g., Friedman & Huttenlocher Reference Friedman and Huttenlocher1997) but are still presumably EMs; and it seems likely that we are poor at remembering the internal order of some EMs. It might be best to think of the relevant quality of EM as experiences that are remembered as occurring on a particular occasion.

In MTT, what does it mean to say that one is traveling through time? The metaphor can unintentionally imply a unity and continuity of time that is quite at odds with the fragmentary, manifold way humans experience it. The finding, mentioned earlier and cited by S&C, that adults sometimes remember the time of day of an event but not its time on longer scales, is difficult to reconcile with the metaphor of “traveling through time.” The limitations of this metaphor may be even clearer when one considers related developmental research. From about 4 years of age onward, many children are able to recall unique happenings when asked about events such as “your last birthday” or “last Christmas,” but this ability appears years before children are aware of when these events had occurred relative to one another (Friedman Reference Friedman1992). What appear to be genuine EMs are more like “islands in time” than memories one reaches by mentally traveling through some temporally organized representation. Similarly, children anticipate particular events (and plan for them) before they have a clear understanding of when in the future the events will occur (Friedman Reference Friedman and Kail2003). Finally, 5-year-olds, who can remember specific past events and anticipate specific events to come, sometimes confuse the past–future status of these events (Friedman Reference Friedman and Kail2003).

Research on humans' memory for times and on thinking about the times of future events shows that there are some common processes (e.g., the use of representations of time patterns) and some differences (using impressions of the ages of memories). In my view, it remains an open question to what extent common processes underlie EM and future-directed thinking in general. The developmental and neurological evidence that S&C cite is suggestive, yet developmental changes can co-occur but be rooted in different processes, and the deficits that hippocampal patients show in EM and anticipation could be due in part to problems other than the capacity to engage in MTT (such as ones related to constructing spatially coherent representations; Hassabis et al. Reference Hassabis, Kumaran, Vann and Maguire2007). Even if temporal information is not a defining feature of EM, as I and others have suggested, it is not clear that the remaining criterion – autonoetic consciousness of particular autobiographical episodes – is necessary to flexibly plan specific future events. Throughout development, planning may rely to a greater extent on information abstracted from repeated episodes (the commonalities of which are more relevant to the future than the particularities) and from semantic memory. S&C have raised interesting questions which merit further research.