Long-Term Memory of Past Events in Great Apes

Episodic-Like Memory in Apes

Episodic memory is memory for personally experienced past events (Tulving, 1972). It involves the subjective experience of mentally traveling back in time (autonoetic consciousness) and a self-awareness of one’s past, future, and present (chronesthesia; Tulving, 2002). Demonstrating conscious experience in populations without language is extremely challenging, and as of yet, there is no standard agreement on nonlinguistic markers of conscious recollection. Thus, episodic memory in animals has conventionally been investigated using criteria based on content, that is, the recall of what, where, and when from a single episode. Clayton and Dickinson (1998) termed this episodic-like memory, in acknowledgment of the lack of measures of conscious awareness.

One of the first demonstrations of chimpanzee memory for what and where was provided by Tinklepaugh (1932). He presented two chimpanzees with two containers and hid food (e.g., a banana) under one of them. Before the chimpanzees could make a choice, a screen was positioned over the two containers, and the hidden banana was substituted with a different type of food (e.g., an orange). The chimpanzees successfully selected the container under which they had seen the banana being hidden but hesitated when they saw that an orange was there. This behavior suggests that they remembered not only where the food was hidden but also what had been hidden because the hesitant behavior was observed only when the food was substituted. Similarly, a chimpanzee spontaneously communicated to a person (using lexigrams, gestures, gazing, and vocalizations) about what food had been hidden and where up to 16 hr after she witnessed the foods being hidden (Menzel, 1999). Indeed, there are numerous other examples of apes’ memory for what and where (see Scheumann & Call, 2006; Schwartz & Evans, 2001).

Regarding memory for when, Schwartz, Hoffman, and Evans (2005) showed that a gorilla, King, could remember what he had eaten and in what order. King was given three different foods, one at a time. After a delay, he correctly selected cards that depicted the foods he had eaten and in the reverse order in which he had eaten them, suggesting he remembered the order of events. Martin-Ordas, Haun, Colmenares, and Call (2010) tested apes using a paradigm developed from Clayton and Dickinson’s (1998) test of episodic-like memory in scrub jays. Apes needed to remember under which container (where) frozen juice (what) had been hidden. However, to obtain the frozen juice before it had melted, they needed to recall whether it was hidden 5 min or 1 hr ago (when). Demonstrating memory of all three what, where, and when components, the apes successfully chose the container where the frozen juice was hidden after the short delay and switched to the container in which less preferred, but edible, raisins had been hidden after the long delay (but see Dekleva, Dufour, de Vries, Spruijt, & Sterck, 2011).

Other research has investigated what-who combinations. Because human episodic memory often contains more details than what was where and when, investigating the memory of other parts is important. For instance, King (the gorilla) could remember what he had eaten and who had given him the food after a delay of 24 hr (Schwartz, Colon, Sanchez, Rodriguez, & Evans, 2002). However, a recent study by Lewis, Berntsen, and Call (2018) suggests that the who information must be salient to be successfully remembered, with apes failing to encode and remember who rewarded what, probably because the person was not a salient feature in the paradigm. This finding highlights the importance of carefully interpreting results in animal research in which subjects receive no instructions regarding what to attend to, and future research should clarify that a failure to remember is not just a failure to attend.

Despite numerous examples of apes’ memory for elements of past events, the criteria deemed as necessary to show episodic-like memory has changed substantially from Clayton and Dickinson’s (1998) original definition to include incidental encoding (Zentall, Clement, Bhatt, & Allen, 2001), binding, and flexibility (Clayton, Yu, & Dickinson, 2003), among others. Consequently, the criteria used in different studies have become more variable and the connection to underlying memory concepts less consistent across studies and with the literature on human memory. We therefore review an alternative approach that draws on recent developments in the conceptualization of human memory for past events (Rubin & Umanath, 2015) and research on involuntary episodic memories in humans (Berntsen, 2009, 2010).

Involuntary Episodic Memories

Involuntary episodic memories are memories of personal past events that are brought to mind spontaneously, without immediately preceding attempts at retrieval (Berntsen, 1996). They are well established in human autobiographical memory and provide an alternative approach to the study of past events in nonhuman animals. Involuntary memories often occur when features in the current environment overlap with features of the remembered event (Berntsen, Staugaard, & Sørensen, 2013). When this overlap is diagnostic of a specific event, the event is likely to be spontaneously remembered, consistent with the notion of encoding specificity (Tulving & Thomson, 1973), cue overload (Watkins & Watkins, 1975), and cue distinctiveness (Hunt & Smith, 1996). There are numerous ways in which a cue can successfully isolate a specific memory (Berntsen, 2009). Here, we present three examples (they should be seen as examples and not an exhaustive account). For instance, a single cue can trigger a memory if it is unique to one memory (i.e., it is not overloaded; Fig. 1, Cue Environment 1). Even if the cues taken individually are relevant to multiple memories, a combination of cues can create a unique constellation that is specific to one memory (Fig. 1, Cue Environment 2). Cues that are not unique to one memory (even when combined) can also result in successful retrieval of a specific event if one of the memories that the cue matches is distinctive from all other memories (Fig. 1, Cue Environment 3; Hunt & Worthen, 2006).

OPEN IN VIEWER

Fig. 1.

An illustration of spontaneous memory retrieval as a result of diagnostic cuing and event distinctiveness. In Cue Environment 1, the cue is unique to Memory 2, and thus Memory 2 is retrieved. In Cue Environment 2, none of the cues are unique to a single memory, but the combination of cues is unique to Memory 3, resulting in retrieval of Memory 3. In Cue Environment 3, the cues in isolation are not unique to a single memory and when combined, match both Memory 1 and 3. However, because the event in Memory 1 is distinctive because of the novel pineapple feature, this memory is likely to be retrieved. (Photo used with permission from the Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany.)

Involuntary memories are less cognitively demanding than voluntary memories. This finding is supported by studies showing shorter retrieval times for involuntary memories than voluntary memories in behavioral experiments (e.g., Berntsen et al., 2013; Schlagman & Kvavilashvili, 2008). Furthermore, a brain-imaging study showed that compared with involuntary memories, voluntary memories of scenes elicited greater activity in dorsal frontal regions, typically associated with retrieval effort, whereas areas typically associated with successful episodic recollection were similarly engaged by both types of memories (Hall et al., 2014). Thus, the two types of retrieval appear to sample from the same underlying memory structures but through different retrieval routes, with the involuntary route being less effortful. This is consistent with research showing that young children can spontaneously recall a past event when presented with distinct overlapping cues but have difficulty intentionally retrieving a past event (Krojgaard, Kingo, Jensen, & Berntsen, 2017).

Consequently, it has been proposed that animals may also have the involuntary mode of remembering (Berntsen, 2009). In the following, we present recent research with apes that is consistent with this proposal. Specifically, the research shows successful retrieval of specific past events as a result of diagnostic cuing and distinctiveness, irrespective of reinforcement. These findings complement research with chimpanzees showing the use of gesturing, vocalization, and symbols as a means to communicate about past events (Menzel, 1999) and are consistent with recent findings that rats can mentally replay a stream of unique past events (Panoz-Brown et al., 2018).

Spontaneous Memories in Apes

Apes can remember details from past events when presented with cues that provide a diagnostic overlap with the remembered event (Fig. 1, Cue Environments 1 and 2). For instance, Kano and Hirata (2015) presented chimpanzees with a film in which a King Kong–like character entered through one of two doors and attacked a human. While watching the film, an eye-tracker monitored the ape’s gaze. Twenty-four hours later, the apes watched the film again and made anticipatory looks toward the door in which “King Kong” had previously entered; that is, they looked at the door before the character entered, suggesting that they recalled what happened next. The apes received food regardless of their looking behavior and saw the film only once; thus, associative learning through reinforcement is unlikely. In a second film, an experimenter selected one of two items and attacked “King Kong.” The apes made anticipatory looks toward the item that was selected even when the items’ locations were switched from the original viewing. Thus, the apes demonstrated memory of where and what when they were cued with preceding parts of the film (see also Mendes & Call, 2014).

Examining distinctiveness (Hunt & Smith, 1996; Hunt & Worthen, 2006), we previously demonstrated (Lewis, Call, & Berntsen, 2017a) that apes can remember a distinctive event when probed by features that overlap with the distinctive event (Fig. 1, Cue Environment 3). Apes remembered which container food had previously been hidden under when later presented with cues that occurred during the hiding event, but only when the event was distinctive. When the hiding event occurred in a location experienced daily, with familiar apparatus, experimenters, and baiting procedure, the apes failed to remember the location of the hidden food. However, when the hiding event was made distinctive, by increasing the amount of food and using a novel baiting procedure, the apes successfully remembered the location. Moreover, a third experiment showed that performance was not explained by reinforcement. Apes in a reinforced condition received a banana during the hiding event, whereas apes in a nonreinforced condition received no food and instead saw a distinctive act of a banana being thrown away. The apes in the reinforced condition failed to remember the location of the banana, whereas the apes in the nonreinforced-but-distinctive-event condition successfully remembered this (Fig. 2; see also Beran, 2011, for a distinctiveness effect in recognition memory in a study with chimpanzees).

OPEN IN VIEWER

Fig. 2.

Percentage of apes that chose the correct location by condition. A correct choice was the container that had previously been baited with food. As there were three containers, chance was 33%. The asterisk indicates that the result was significantly different from chance (p < .05). (Adapted from Lewis, Call, & Berntsen, 2017a.)

Apes can also remember past events after very long delays. For instance, apes remembered a tool-hiding event that occurred 3 years ago when presented with the same tool task, experimenter, and location as during the initial hiding event (Martin-Ordas, Berntsen, & Call, 2013). Likewise, in a recent study (Lewis, Call, & Berntsen, 2017b), we found that apes remembered a specific hiding event after a delay of up to 50 weeks when presented with cues diagnostic of the hiding event. Apes saw distinctive food being hidden in a novel location. After a retention period of 2, 10, or 50 weeks, they returned to the room in which the event took place. When they discovered a piece of the previously hidden food on the ground, they immediately searched the location where the food was previously hidden (Fig. 3), despite no food being present there. Conversely, apes that had not witnessed a hiding event did not search. The immediate behavioral response of the apes in response to diagnostic cues is consistent with the fast retrieval times observed in involuntary episodic retrieval in humans (e.g., Berntsen et al., 2013). Importantly, the apes witnessed the event only once and retrieved the memory in the absence of any task requiring goal-directed retrieval, suggesting similarities with human involuntary memories (Berntsen, 2010). Additional similarities were also found in the rate of forgetting over time. In humans, forgetting often follows a logarithmic function (Rubin & Wenzel, 1996) in which information is lost rapidly at first but then levels over time (Ebbinghaus, 1885/1964). We observed this same result in apes (Fig. 4).

OPEN IN VIEWER

Fig. 3.

Still frames of an orangutan (Tanna) during a retrieval session. Tanna finds a cardamom-flavored pellet on the ground (a), smells the pellet and eats it (b), and then climbs and searches the location (indicated by the red circle) in which cardamom pellets were hidden 2 weeks previously (c).

OPEN IN VIEWER

Fig. 4.

Percentage of searching over time. Dots indicate the retention periods (i.e., the first dot shows memory after 5 min, the second after 2 weeks, the third after 10 weeks, the fourth after 50 weeks). The dashed line shows retention as a logarithmic function. (Adapted from Lewis, Call, & Berntsen, 2017b.)

Conclusions

Episodic memory in animals has conventionally been investigated using Clayton and Dickinson’s (1998) episodic-like paradigm. However, the extensive and varying criteria required to meet the definition of episodic-like memory has made progression difficult. We, among other researchers (e.g., Basile, 2015; Jelbert & Clayton, 2017), propose that further advances will be made by investigating the similarities and differences between human and animal memory using alternative approaches. Here, we have described an approach based on insights from research on involuntary memories of past events in humans.

We have shown that apes remember past events when probed by features that are diagnostic of the past event, consistent with encoding specificity and cue-overload explanations of involuntary memory retrieval in humans. Although this research documents behavior consistent with the occurrence of spontaneous event memories in apes, we acknowledge that we cannot comment on whether apes consciously recollect a past event (because they provide no verbal descriptions). However, the findings reviewed point to a behavioral reliving of a memory (a reenactment of a past event) that is triggered in ways consistent with involuntary memories in humans. These findings seem to indicate important similarities between memories for past events in apes and humans, suggesting that the ability to spontaneously recollect past events may be shared across species.

Future research may use the present approach for research on event memory in other species. Moreover, other aspects of episodic memory, such as retrieval effort or the influence of emotion or age, deserve some research attention from a comparative perspective. These aspects have received considerable research attention in the human literature but remain largely unexplored in animals. Such comparative insights may enrich the field by mapping the mechanisms and evolution of episodic memory in humans and other animals.

Recommended Reading

Clayton, N., & Dickinson, A. (1998). (See References). This article introduced the what-where-when paradigm as a way of assessing episodic-like memory in nonhuman animals.

Crystal, J. D. (2018). Animal models of episodic memory. Comparative Cognition & Behavior Reviews, 13, 105–122. doi:10.3819/ccbr.2018.130012. A comprehensive review of recent developments in comparative episodic-memory research.

Dere, E., Kart-Teke, E., Huston, J. P., & De Souza Silva, M. A. (2006). The case for episodic memory in animals. Neuroscience & Biobehavioral Reviews, 30, 1206–1224. doi:10.1016/j.neubiorev.2006.09.005. A comprehensive review of episodic-memory research in numerous species, including primates.

Lewis, A., Call, J., & Berntsen, D. (2017b). (See References). A representative study that illustrates original research about spontaneous memory retrieval in great apes.

Noser, R., & Byrne, R. W. (2015). Wild chacma baboons (Papio ursinus) remember single foraging episodes. Animal Cognition, 18, 921–929. doi:10.1007/s10071-015-0862-4. Application of episodic-memory research to a wild population of baboons.