Suddendorf & Corballis (S&C) argue that humans are unique in possessing the ability to mentally travel in time. Episodic memory allows us to mentally travel into the past, and so, it is claimed, nonhuman animals also lack episodic memory. In making this claim, it is important to be clear about one’s definition of episodic memory. S&C write that episodic memory is “about reconstructing particularities of specific events that have happened to the individual” (sect 2.1, emphasis in original). But what particularities might be relevant to this claim? S&C discuss the widespread claim in the literature that at least a precursor of episodic memory may be found nonhuman animal’s memory for the what, where, and when of an event.

Yet S&C reject evidence of such episodic-like memory from nonhuman animals on two bases. First, they criticise examples of episodic-like memory in nonhuman animals that depend on coding when an event happened, which may, they claim, depend on other cues such as trace strength. Second, they claim that this type of memory could be known rather than remembered. We address these claims using examples of work with rats from our laboratory.

First, we consider what “particularities of specific events that have happened to the individual” episodic-like memory might represent. The problematic claim is that nonhuman animals might code when (or, strictly speaking, how long ago) a remembered event occurred. Yet human episodic memory rarely codes when, and even with all the advantages of verbal labels to help code such information (e.g., “last Wednesday,” “19th March”), we are extremely poor at such memory (Friedman Reference Friedman1993). Nevertheless there is evidence that even without such labels, nonhuman animals may code this type of information, although as S&C point out, the difficulty remains of ensuring that trace strength itself cannot provide a sufficient cue without recourse to mental time travel. Yet, could this difficulty be overcome if we asked nonhuman animals about when a remembered event happened in a different way? When we code when something happened, we rarely code the exact time point; instead, we often use convenient contextual cues. For example, I may remember that I mislaid my car keys on Wednesday, but only because of the contextual cue that it happened in the lecture theatre in which I teach only on Wednesday mornings; and I may also differentiate this from a similar event that happened in the afternoon in a different room.

Recently in our lab we have used such contextual information to cue animals to one of two recently experienced events (Eacott et al. Reference Eacott, Easton and Zinkivskay2005; Easton et al. Reference Easton, Eacott, Zinkivskay and Jimenez-Rodriguez2006). Rats experience two highly similar events, finding two different objects each day hidden within an E-shaped maze. In the first event each day, object A is found in the left arm of the maze, while object B is in the right arm with a particular visuospatial context present in the maze. In the second event each day, the position of these objects is reversed in the presence of a second visuospatial context. The visuospatial context serves as a contextual cue, like the lecture theatre in the previous example. After the rats have experienced the two events, one of the objects (e.g., object A) is overexposed, giving the rats a natural preference for seeking out the other, nonexposed object (in this case, object B). We then ask whether rats can use their memory to seek out a preferred object when returned to the maze with one of the previous contexts present. The context serves in place of a verbal label for the particular occasion being asked about, so this provides a way of asking the rat “where was object B when you saw it in the black context?” Intact rats efficiently “answer” such a question by seeking out object B (Eacott et al. Reference Eacott, Easton and Zinkivskay2005), and it cannot be claimed that such ability can be reduced to a time-dependent process such as trace decay.

However, S&C cite this work and claim that, like other examples of memory for the peculiarities of events, this type of memory “may be known rather than remembered” (sect. 3.1, para. 5). Yet how can one address such a criticism? The essential difference between remembering and knowing about a past, personally experienced event lies in the subjective feeling associated with a memory, and clearly this cannot be investigated in nonverbal animals. However, in humans the subjective feelings of knowing and remembering are associated with different neural systems and different patterns of performance when analysing receiver operating curve (ROC) curves (Yonelinas Reference Yonelinas1994; Yonelinas et al. Reference Yonelinas, Kroll, Dobbins, Lazzara and Knight1998). Recently we have produced direct behavioural evidence (Easton et al. Reference Easton, Eacott, Zinkivskay and Jimenez-Rodriguez2006) of this dissociation in rats. In the task described earlier, intact rats were capable of answering a question equivalent to “where was object B when you saw it in the black context?” by seeking out a hidden object in the appropriate context. Because the object was hidden, this ability must rely on recall, and not recognition memory. Moreover, this recall ability is dependent on the fornix, the main efferent of the hippocampus (Easton et al. Reference Easton, Eacott, Zinkivskay and Jimenez-Rodriguez2006). Yet the same fornix-transected rats that could not seek out the preferred object on the basis of recall, showed entirely normal recognition memory of the same objects in the same trials of the task. As these animals had impaired recall, we can assume that their intact recognition ability was mediated by intact familiarity processes. Thus, by using behavioural tasks in conjunction with lesion-based evidence, we can dissociate recall (“remembering”) from familiarity-based recognition (“knowing”). This behavioural evidence supports other recent work in rats performing a food-rewarded, old-new, odour recognition paradigm in which statistical analysis of ROC curves show recollection, and familiarity can be dissociated from one another with lesions of the hippocampus specifically impairing recollection (Fortin et al. Reference Fortin, Wright and Eichenbaum2004).

Of course, we cannot say with any confidence that this “recall” in a rat was associated with the phenomenological experience that accompanies recall in humans. Yet, as S&C point out, there is as yet no agreement as to how to study such subjective phenomena in humans, and so we should not set the evidential bar for demonstrating recall in rats so high that it cannot be satisfied even for other humans. Nevertheless, by showing increasing evidence for similarity between phenomena in rats and in humans, we can at least claim that we have demonstrated a dissociation between “familiarity-like” memory and “recall-like” memory in the rat.