**6. Novel object recognition task**

The paradigm used most often to assess what/where/which memory is the novel object recognition task. This clever but simple task takes advantage of a predisposition in many species to explore novel objects over familiar ones. Ennaceur and Delacour (1988) first reported the object recognition task, in which rats were exposed to objects during an acquisition trial and then tested on their ability to discriminate between familiar and novel objects, as a test of working memory. The object recognition test has been used to show that rats are sensitive to the location of objects (Dix & Aggleton, 1999; Ennaceur et al., 1997; Poucet, 1989), to the topological relationship between objects (Dix & Aggleton, 1999; Goodrich-Hunsaker et al., 2008; Harley et al., 2001; Lemon et al., 2009), to changes in the distance between objects (Goodrich-Hunsaker et al., 2008), to the context in which objects have been experienced (Dix & Aggleton, 1999; Eacott & Norman, 2004), and to changes in object compounds (Norman & Eacott, 2004).

In addition to the innovative what/where/which definition, Eacott and Norman's (2004) unique method of testing episodic-like memory meets the requirements of spontaneous recall (Menzel, 1999) and recall during an unexpected test (Zentall et al., 2001). Eacott & Norman (2004) found that rats can integrate memories of a specific object (what), its spatial location (where) and the context in which it occurs (which) to discriminate the more novel of two object/location/context combinations. Rats explored the locations (left or right) of each

Spontaneous Object Recognition in Animals: A Test of Episodic Memory 31

rats, pigs naturally tend to explore novel aspects of their environment (Wood-Gush & Vestergaard, 1991). Pigs also have good spatial memory abilities and they are able to learn tasks quickly (e.g., Croney, 1999; Held et al., 2002; Held et al., 2005; Puppe et al., 2007; reviewed in Gieling et al., 2011). As well, wild and feral pigs have a life history in which memory is valuable; particularly because they live socially, have large foraging ranges, and have foraging habits/movement patterns/nesting sites that change with season and food availability (Graves, 1984). Since domestic pigs have retained many natural behaviours despite the domestication process, particularly in foraging (Gustafsson et al., 1999), it is reasonable to speculate that domestic pigs retain the memory abilities possessed by their wild ancestors. These factors indicate that episodic-like memory in pigs may be more

The physiological similarity between humans and pigs is likely responsible for the recent increased popularity of pigs as biomedical models of human disease and cognition (for reviews, see Gieling et al., 2011; Kornum & Knudsen, 2011; Lind et al., 2007). Accordingly, pigs may also provide a more effective comparison than other species between human episodic memory and episodic-like memory in animals. Specifically, the pig brain is more similar to the human brain in structure (gyration), myelination and electrical activity than are the brains of rodents and other small laboratory animals (Dickerson & Dobbing, 1966; Pond et al., 2000). Also similar to humans, the pig brain develops perinatally, with a growth spurt extending from mid-gestation to about 40 days after birth (Dickerson & Dobbing, 1966; Dobbing & Sands, 1973; Pond et al., 2000). Such similar physiological brain development may be particularly valuable in studies of changes in memory with age.

Prior to our work (Kouwenberg et al., 2009), the existence of episodic-like memory in pigs remained virtually unexplored. However, there were several studies that demonstrated pigs can perform spontaneous object recognition, using modifications of the Ennaceur and Delacour (1988) protocol (Gifford et al., 2007; Kornum et al., 2007; Moustgaard et al., 2002). We explored episodic-like memory in pigs by examining their ability to discriminate between objects according to the location and context in which they occurred (Figures 1 and 2). On each trial, a pig was confined to a holding pen for 2 min prior to a 10 min exposure to one context containing two objects (e.g., Context 1 with Object A on the right and Object B on the left). After an additional 5 min in the holding pen, the pig was given a 10 min exposure to another context containing the same objects but in opposite locations (i.e., Context 2 with Object A now on the left and Object B on the right). The test trial was administered after another 5 min in the holding pen, and consisted of a 10 min exposure to one of the contexts with two identical objects (e.g., Context 2 with two copies of Object A). If the pigs remembered the location and context in which each object occurred during the two exposure phases, they should allocate their exploration time differentially, based on the

familiarity of the object/location/context configurations during the test phase.

Pigs spent more time with the less familiar object/location/context during the test phases of the episodic-like memory trials, indicating that they were able to simultaneously recall memories of what (object), where (location) and which (context). Since the separate aspects (object, location, and context) of each object/location/context configuration are equally familiar, it is only the combination of all three aspects that makes one configuration less familiar than another. Therefore, the pigs' significant preference for the *less* familiar configuration cannot be attributed to object preference alone, location preference alone, or

developed than in some other species.

of the objects (A or B) in each of the two contexts (1 or 2). During the test, the rat was placed in one of the contexts with two copies of the same object (e.g., A and A), and the amount of time the rat spent exploring each object was recorded. Since identifying the more novel of two configurations requires the simultaneous recall of what, where and which (object/location/context) information, Eacott and her colleagues argued that novel object recognition tasks test episodic-like memory (Eacott et al., 2005; Kart-Teke et al., 2006). In fact, they argued that object recognition is superior to other methods because it requires very little training before subjects are tested, which reduces potential confounds caused by reinforced learning (Eacott & Norman, 2004). Furthermore, since exploring novelty is a natural response for many species, recall of the more novel object/location/context appears to be spontaneous, which meets Menzel's (1999) criterion for episodic-like memory. As well, explicit cues or rewards are not needed to prompt memories, which meets Zentall et al.'s (2001) criterion that episodic-like memory tests must be unexpected.

Similarly, others have shown that rats (Kart-Teke et al., 2006) and mice (Dere, et al., 2005) integrate what and where information with the order in which stimuli are presented. An object recognition task was used that required the animals to discriminate more novel objects based on a combination of the objects' locations and the order in which they were presented. The animals spent more time exploring a less recently presented object compared to a more recently presented object, which indicates that they had integrated "what and when" memory. The authors found that "what and when" memory was integrated with "where" because the animals responded differently to displacement of more recent and less recent objects. When presented with two more recently experienced objects, the animals spent more time with the object that had been displaced to an unfamiliar location as compared to the object in a familiar location. In contrast, when presented with two less recent objects, the animals spent more time with the object in the familiar location than with the object in an unfamiliar location. They concluded that these findings provided evidence for integration of what/where/when memories into a single tripartite code because they show that these three dimensions are not encoded, stored and retrieved independently (Dere, et al., 2005; Kart-Teke et al., 2006). As well, they argue that animals could not use relative memory strengths to discriminate whether an object was displaced because spatial information was obtained on a single trial.

The use of object recognition/preference to study episodic memory has also been extended to the study of recall of information without the stimuli being re-presented in the test phase (Eacott et al., 2005). Rats were trained to explore an E-shaped maze with two objects, followed by exposure to a different E-shaped maze with the same objects in opposite locations. After the two exposures, the rats were placed in a different context with one of the objects for a habituation session. When the rats were placed back into one of the E-shaped mazes, they tended to go to the non-habituated object, which was not visible from the middle stem of the E. In order to explore the more novel (non-habituated) object, the rats must remember which of the two objects (what) is in each arm (where) in which context. This recollection task, like the caching task used with scrub jays, 'asks' rats about objects that they cannot see and is more akin to the types of recall used in human measures of episodic memory.

#### **7. Episodic-like memory in pigs**

Eacott and Norman's (2004) successful demonstration of what/where/which memory in rats has led to an interest in applying this definition and method to other species. Similar to 30 Advances in Object Recognition Systems

of the objects (A or B) in each of the two contexts (1 or 2). During the test, the rat was placed in one of the contexts with two copies of the same object (e.g., A and A), and the amount of time the rat spent exploring each object was recorded. Since identifying the more novel of two configurations requires the simultaneous recall of what, where and which (object/location/context) information, Eacott and her colleagues argued that novel object recognition tasks test episodic-like memory (Eacott et al., 2005; Kart-Teke et al., 2006). In fact, they argued that object recognition is superior to other methods because it requires very little training before subjects are tested, which reduces potential confounds caused by reinforced learning (Eacott & Norman, 2004). Furthermore, since exploring novelty is a natural response for many species, recall of the more novel object/location/context appears to be spontaneous, which meets Menzel's (1999) criterion for episodic-like memory. As well, explicit cues or rewards are not needed to prompt memories, which meets Zentall et al.'s

Similarly, others have shown that rats (Kart-Teke et al., 2006) and mice (Dere, et al., 2005) integrate what and where information with the order in which stimuli are presented. An object recognition task was used that required the animals to discriminate more novel objects based on a combination of the objects' locations and the order in which they were presented. The animals spent more time exploring a less recently presented object compared to a more recently presented object, which indicates that they had integrated "what and when" memory. The authors found that "what and when" memory was integrated with "where" because the animals responded differently to displacement of more recent and less recent objects. When presented with two more recently experienced objects, the animals spent more time with the object that had been displaced to an unfamiliar location as compared to the object in a familiar location. In contrast, when presented with two less recent objects, the animals spent more time with the object in the familiar location than with the object in an unfamiliar location. They concluded that these findings provided evidence for integration of what/where/when memories into a single tripartite code because they show that these three dimensions are not encoded, stored and retrieved independently (Dere, et al., 2005; Kart-Teke et al., 2006). As well, they argue that animals could not use relative memory strengths to discriminate whether an object was displaced because spatial

The use of object recognition/preference to study episodic memory has also been extended to the study of recall of information without the stimuli being re-presented in the test phase (Eacott et al., 2005). Rats were trained to explore an E-shaped maze with two objects, followed by exposure to a different E-shaped maze with the same objects in opposite locations. After the two exposures, the rats were placed in a different context with one of the objects for a habituation session. When the rats were placed back into one of the E-shaped mazes, they tended to go to the non-habituated object, which was not visible from the middle stem of the E. In order to explore the more novel (non-habituated) object, the rats must remember which of the two objects (what) is in each arm (where) in which context. This recollection task, like the caching task used with scrub jays, 'asks' rats about objects that they cannot see and is more

Eacott and Norman's (2004) successful demonstration of what/where/which memory in rats has led to an interest in applying this definition and method to other species. Similar to

akin to the types of recall used in human measures of episodic memory.

(2001) criterion that episodic-like memory tests must be unexpected.

information was obtained on a single trial.

**7. Episodic-like memory in pigs** 

rats, pigs naturally tend to explore novel aspects of their environment (Wood-Gush & Vestergaard, 1991). Pigs also have good spatial memory abilities and they are able to learn tasks quickly (e.g., Croney, 1999; Held et al., 2002; Held et al., 2005; Puppe et al., 2007; reviewed in Gieling et al., 2011). As well, wild and feral pigs have a life history in which memory is valuable; particularly because they live socially, have large foraging ranges, and have foraging habits/movement patterns/nesting sites that change with season and food availability (Graves, 1984). Since domestic pigs have retained many natural behaviours despite the domestication process, particularly in foraging (Gustafsson et al., 1999), it is reasonable to speculate that domestic pigs retain the memory abilities possessed by their wild ancestors. These factors indicate that episodic-like memory in pigs may be more developed than in some other species.

The physiological similarity between humans and pigs is likely responsible for the recent increased popularity of pigs as biomedical models of human disease and cognition (for reviews, see Gieling et al., 2011; Kornum & Knudsen, 2011; Lind et al., 2007). Accordingly, pigs may also provide a more effective comparison than other species between human episodic memory and episodic-like memory in animals. Specifically, the pig brain is more similar to the human brain in structure (gyration), myelination and electrical activity than are the brains of rodents and other small laboratory animals (Dickerson & Dobbing, 1966; Pond et al., 2000). Also similar to humans, the pig brain develops perinatally, with a growth spurt extending from mid-gestation to about 40 days after birth (Dickerson & Dobbing, 1966; Dobbing & Sands, 1973; Pond et al., 2000). Such similar physiological brain development may be particularly valuable in studies of changes in memory with age.

Prior to our work (Kouwenberg et al., 2009), the existence of episodic-like memory in pigs remained virtually unexplored. However, there were several studies that demonstrated pigs can perform spontaneous object recognition, using modifications of the Ennaceur and Delacour (1988) protocol (Gifford et al., 2007; Kornum et al., 2007; Moustgaard et al., 2002). We explored episodic-like memory in pigs by examining their ability to discriminate between objects according to the location and context in which they occurred (Figures 1 and 2). On each trial, a pig was confined to a holding pen for 2 min prior to a 10 min exposure to one context containing two objects (e.g., Context 1 with Object A on the right and Object B on the left). After an additional 5 min in the holding pen, the pig was given a 10 min exposure to another context containing the same objects but in opposite locations (i.e., Context 2 with Object A now on the left and Object B on the right). The test trial was administered after another 5 min in the holding pen, and consisted of a 10 min exposure to one of the contexts with two identical objects (e.g., Context 2 with two copies of Object A). If the pigs remembered the location and context in which each object occurred during the two exposure phases, they should allocate their exploration time differentially, based on the familiarity of the object/location/context configurations during the test phase.

Pigs spent more time with the less familiar object/location/context during the test phases of the episodic-like memory trials, indicating that they were able to simultaneously recall memories of what (object), where (location) and which (context). Since the separate aspects (object, location, and context) of each object/location/context configuration are equally familiar, it is only the combination of all three aspects that makes one configuration less familiar than another. Therefore, the pigs' significant preference for the *less* familiar configuration cannot be attributed to object preference alone, location preference alone, or

Spontaneous Object Recognition in Animals: A Test of Episodic Memory 33

context preference alone. Objects in the test phase were identical, pigs had been equally exposed to both locations before the test phase, and pigs had been equally exposed to both contexts before the test phase. Furthermore, the preference for the less familiar object/location/context could not be attributed to object and location alone because objects and locations were counterbalanced for each pig. Thus, each pig received an episodic-like memory trial with two objects "A" in the test phase and a trial with two objects "B" in the test phase. If pigs were ignoring context and making decisions based solely on object and location, half of the time pigs would spend more time with the left-hand object and the other half of the time the pig would spend more time with the right-hand object. This would have resulted in no significant overall preference for either object/location/context. Our data indicate that this is not the case, leading to the conclusion that pigs formed integrated memories of what/where/which information. Whether pigs can also do the recollection

The above findings indicate that the formation of a tripartite code of either "what/where/when" or "what/where/which" seems to be within the compass of animals when species-typical preferences are taken into account. While the what/where/when model of Clayton and Dickinson (1998) is an elegant demonstration of episodic-like memory, its usefulness may be restricted due to the limited number of animals that cache food. While some researchers, notably Babb and Crystal (2006), have attempted to modify this task with rodents it requires a significant amount of pre-training because it does not use behaviours that naturally exist within the repertoire of some species. The recent findings from object exploration indicate that this may be a powerful way to study the formation of a tripartite code in animals. The paradigm allows for the testing of a tripartite code of what/where/which but not what/where/when memory (Eacott & Norman, 2004). It takes advantage of the tendency to explore novel objects, seen in many species, to demonstrate spontaneous recall (Menzel, 1999). It tests an animal's recall of an event when the test is not expected (Zentall et al., 2001) and may even allow for a test of past events over a long period of time (Schwartz & Evans, 2001) although no such long term tests have yet been carried

While a model of episodic memory based on object recognition is applicable to a greater variety of animals than a model based on food caching behaviour, we acknowledge that many of the criticisms that have been lodged against the what/where/when model (e.g., Suddendorf & Busby, 2003) also apply to the what/where/which model. For example, evidence for future planning and mental time travel would greatly improve both models of episodic-like memory. Clayton et al., (2003a) have recognized that their basic what/where/when criteria no longer adequately define the evolving concept of episodiclike memory. In response, they have refined their definition of episodic-like memory to include three particular behavioural criteria. Specifically, they state that a solid demonstration of episodic-like memory requires *content* (what/where/when details of a specific past event), *structure* (integration of the what/where/when details into a consolidated memory), and *flexibility* (ability to change how information gained from an episodic-like memory is used). Eacott et al. (2005) have argued that these three criteria are

task used by Eacott et al. (2005) remains to be determined.

also met in their modified task examining recall of objects.

**8. Conclusion** 

out.

Fig. 1. A diagram of the episodic-like memory trial used in our study. An example of a possible configuration for the rst exposure phase (a), second exposure phase (b), and test phase (c). Shading indicates a different floor colour (i.e., different context). The black circle indicates the novel object/location/context in the test phase of this trial.

Fig. 2. A pig interacting with an object during episodic-like memory test.

context preference alone. Objects in the test phase were identical, pigs had been equally exposed to both locations before the test phase, and pigs had been equally exposed to both contexts before the test phase. Furthermore, the preference for the less familiar object/location/context could not be attributed to object and location alone because objects and locations were counterbalanced for each pig. Thus, each pig received an episodic-like memory trial with two objects "A" in the test phase and a trial with two objects "B" in the test phase. If pigs were ignoring context and making decisions based solely on object and location, half of the time pigs would spend more time with the left-hand object and the other half of the time the pig would spend more time with the right-hand object. This would have resulted in no significant overall preference for either object/location/context. Our data indicate that this is not the case, leading to the conclusion that pigs formed integrated memories of what/where/which information. Whether pigs can also do the recollection task used by Eacott et al. (2005) remains to be determined.
