Episodic memory in animals, Are they capable of mental time travel?

Sarah-Jane Fahed

American University Of Beirut

Mental Time Travel in animals Episodic memory is a type of declarative memory, it’s the memory for personal events and is distinguished from semantic memory: memory for facts.
What characterizes episodic memory is that it involves mental time travel also termed “Chronesthesia”: it is the capacity to mentally project oneself in the past to remember events that took place and projecting oneself to try and predict the future.
It is commonly thought to be specific to humans but some studies have been done to research this specific type of memory and see if it can be attributed to animals.
The study of episodic memory in non-humans led to many contradictory results and depends on how it is defined. The main focus of this paper is to study the different researches done on the Chronesthesia component of episodic memory in animals: mental time travel to the past and to the future and show their limitations.

Tulving originally defined episodic memory in terms of the kind of information it appears to store: what where and when something happened (the www criterion) and later added the concept of autoneotic awareness to the definition (as cited in Suddendorf & Corballisb, 2007): the sensation that a memory was personally experienced
In their book Missing Link in Cognition: Origins of Self-Reflective Consciousness Terrace and Metcalfe (2005) distinguish between two definitions for episodic memory: a broad and a narrow definition, the narrow definition comprises 4 main features that are unique to episodic memory and distinguish it from semantic memory: mental time travel (the ability to mentally project oneself in the past to relive a personal experience or in the future to prelive an event), the self, autoneotic consciousness and that it doesn’t only cover the past but also the future. Now that we clearly defined episodic memory and identified its different features, we can see how it applies to non-humans. Since most episodic-memory and mental time travel are studied and analyzed through language: by interviewing the subject; it’s hard to measure for animals since they lack this capacity and the results have therefore to be inferred from behavioral and cognitive experiments.
In their paper “The comparative study of mental time travel”, Roberts and Feeney (2009) studied mental time travel in animals based on two perspectives: the www criterion and the capacity to plan for the future a behavioral indication for mental time travel in animals.
The www criterion is the memory about what, where and when an event happened. According to Suddendorf and Busby the www criterion features must form an integrated structure allowing to differentiate analogous events that happened at different times as well as being flexible as for the semantic information given to the subjects (as cited in Cheke & Clayton, 2010). One of the most famous studies to show the www in animals is Clayton and Dickinson’s: scrub jays were given the option to hide two types of foods in different places: wax worms and peanuts and could retrieve it after either 4 hours or 5 days. An important thing to note is that scrub jays prefer wax worms to peanuts. After the long delay (5 days) worms weren’t eatable anymore while the peanut were still good; the scrub jays acted accordingly: After 5 days they would go where peanuts are located even though it is less preferred than worms. So they met the www memory criteria: they remembered where they stored the food, what food was stored and when since they chose the location based on how long the food was hidden (if it decayed or not yet). What was argued is that by satisfying the www criterion, scrub jays formed “episodic like” memories that are necessary but not sufficient for mental time traveling; thus the term “episodic like” (Roberts, W., & Feeney, M. 2009) . In addition, scrub jay also knew information about who saw them hide the food, in an experiment by Dally (2006), if a dominant male saw them hide the food, the birds were more likely to change location. (as cited in Suddendorf & Corballisb, 2007) and that also shows us a capability of future planning: they are decreasing the chance of their food being eaten in the future.
The experiment shows us structure in the www memory of scrub jay (they recovered food according to they delay between caching and recovery) but doesn’t satisfy the other criterions such as flexibility. So Babb and Crystal experiment on rats studied the flexibility criterion of www memory: two out of eight arms of a maze contained two different preferred flavors whereas the rest contained tasteless food. The tasty food was replenished after a long period and the locations were varied throughout trial so that rats could remember www of a specific occasion. The rats went to the preferred location after a long delay but not after the short delay and that shows us that they have a www memory: They knew what was placed, where it was placed (what arm) and how much time had passed (enough to be replenished or not). As for flexibility, the appeal of one of the preferred flavor was manipulated through satiety and the visits to the location of that food decreased: with an altered information given between encoding and retrieval that is a change in the value of the food cached (decrease in its value due to satiety) the subjects modified their behavior accordingly (decreased visits to the location) and that flexibility shows us that the www memory isn’t a learned response and has an episodic-like component. (as cited in Cheke & Clayton, 2010)
But many critics were raised: First, According to Suddendorf & Busby, studies have shown that one is capable of mentally revisiting an event without adhering to the www criterions and equally one can know what, where and when an event happened without mentally traveling back to that event (like birth). (as cited in Suddendorf & Corballisb, 2007) Furthermore the experiment doesn’t prove autoneotic consciousness, a necessary element for mental time travel.
Another critique pointed out is that this episodic-like memory may result from “ feed-forward mechanism that need not to be about the past at all” (Suddendorf & Corballisb 2007) the subject might know the www without having to remember it (semantic rather than episodic-like memories). But scrub jay had no visual access to the place where the food was hidden so how they remembered www criterions is unclear and one of the explanation is their memory of the episodic event of hiding the food (Cheke & Clayton, 2010), an alternative explanation may be the strength of memory-traces that caused the animal to remember where, when and what he hid rather than mental time travel, however an argument against this explanation was raised by Eichebaum et al. in an experiment on rats, based on the belief that the hippocampus is the main site for episodic memory : they showed that after a lesion to the hippocampus, rats were still capable of recognition ( semantic memory) but their faculty to discriminate temporal order ( www memory) was impaired and this means that this capacity to discriminate isn’t based on memory strength because if it was then the observation would be the same as for recognition : It wouldn’t be impaired. (as cited in Suddendorf & Corballisb, 2007). But we shouldn’t dismiss the trace-strength hypothesis because the results may be due to the spatial nature of the www discrimination task (Cheke & Clayton, 2010)
Another concern was brought to light by Williams, that animals in both experiments (scrub jay and rats) might remember how much time passed rather than when in absolute time. This concern was later tested on rats by Roberts et al. , by altering the time of the day and how long ago a favored treat was replenished in Babb& Crystal’s experiment discussed earlier. In the group 1 of rats the food was replenished according to the time of the day (when), in a second group it was based on the time passed (how long ago). The findings showed that rats used the how long ago cue since in the when condition rats went back to the arm that contained the preferred treat when it wasn’t replenished yet. So a modification to WWW memory must be made and when should be replaced with how long ago. (as cited in Roberts , 2012)
Thus as Suddendorf and Busby supposed, the WWW memory cannot be sufficient to prove mental time travel in animals, the memory may be episodic but may as well involve cognitive processes (as cited in Cheke & Clayton, 2010)
That brings us to the second perspective used to study mental time travel in animals: future planning.
According to Suddendorf and Corballisb (2007) there’s continuity between mental time travel to the past and planning for the future. Indeed, a study by Klein and Tulving showed that people who suffer from amnesia and can’t remember events from the past are also unable to anticipate the future. So experiences showing that animals are capable of planning for the future can be a way to invoke mental time travel in animals but it’s hard to distinguish between semantic information about the future and episodic future planning.
Suddendorf & Corballisb (2007) added conditions for the former statement to be correct: First, the subject should not feel in the present, the event to be anticipated (hunger for example); also, innate predispositions, procedural and semantic memory should be excluded as an explanation to future planning; finally the subject must show flexible and adaptive behaviors in new situations.
I’m going to report a couple of experiments showing the capacity of future planning in animals and present their limits.
Roberts and Feeney (2009) examined two experiments: First, scrub jay breakfast experiment by Raby et al.: Scrub jays were given breakfast in a certain compartment “A” some days and not given anything in another compartment “C” other days, the next day when they were given the opportunity to cache food in either A or C before breakfast they chose compartment C and that shows their ability to anticipate a future need. Cheke and Clayton (2010), discuss in their article a deepening of Raby et al.’s breakfast experiment done by Correia et al. where satiety was controlled: scrub jay were nourished with a specific type of food (let’s call it food 1) and then hid food 1 as well as food 2. Later, before the subjects were asked to retrieve the food previously hidden they were fed food 2, so during the hiding stage, the scrub jay were not feeling the event to be anticipated yet they retrieved food 2 a significant number of time.
The second experiment is that of Osvath where two chimpanzees and an orangutan were given the choice between four objects, one of which (the hose) could be used to drink a favorite beverage. Even though the objects were offered in a setting unrelated to their possible use and temporally distant (the opportunity to use the hose was presented 70 min after their choice) the animals chose the hose. And even when the animals were presented with a preferred but not favorite food, they still chose the hose and that demonstrate their ability to anticipate a future need.
But these experiments are not enough to prove episodic memory and planning for the future, they do show flexibility and adaptability but they might be based on semantic memory and may be a “ rough combination of predispositions and specific learning algorithms that have evolved in the caching context” (Suddendorf & Corballisb, 2007). Regarding the hose they could have chose it because they know (semantic) its usual use without necessarily predicting its future use (Roberts, 2012) however in another experiment by Osvath and Osvath, the apes were presented with 3 novel object, one of which could be used to extract fruit juice and a bamboo stick that is associated with food (used for honey extraction) and they chose the novel object that will lead to future reward rather than the bamboo stick that is semantically paired with food. (as cited in Roberts , 2012).
The most famous opposition to future planning in animals is the Bischof-Kohler hypothesis that states that animals act on immediate needs and can’t anticipate future needs; they therefore are incapable of mental time travel. The hypothesis can be disproved by observing the behavior of many animals who cache food for the future, hibernate, migrate… but that may be explained by instinctive behavior and doesn’t necessarily show mental time travel. So in response to this concerns, many experiments were done controlling this instinctive behavior by choosing tests that don’t involve typical behaviors: like the following experiment by Flaherty and Checke, they trained rats to drink from a saccharin beverage, one of the group was given a preferred beverage after 1, 5 or 30 minutes (sucrose) and the control group was only given saccharin. What was observed was that rats in the experimental group decreased their intake of saccharin in anticipation of the more preferred beverage whereas rats in the control group kept on consuming the same amount. A critique raised was that saccharine is non-nutritive whereas sucrose is nutritive and that might explain the rats’ behavior so Feeney, Roberts, and Sherry with chickadees responded to that concern with an experiment offering to the control group a nutritive: (sunflower seeds) meal uniquely and to an experimental group sunflower seeds followed at a 5,10 or 30 minutes interval by a more preferred meal: mealworms. The results were the same as Flaherty and Checke’s experiment; the birds in the experiment group decreased their intake of seeds in anticipation of a more preferred meal (as cited in Roberts, 2012). So they ignored a present state in anticipation of a future one and that directly contradicts the Bischof-Kohler hypothesis.
Other experiments offered a choice to animals: either a small immediate reward vs. larger delayed reward and “to the extent an animal can anticipate a larger delayed reward, such experiments may be seen to involve future thinking.” (Roberts, 2012). Let’s consider one experiment in this category: McKenzie, Cherman, Bird, Naqshbandi, and Roberts gave squirrel monkeys the choice between either 10 or 20 peanuts, 90% chose 20. A pilfering manipulation entered the equation during 12 successive trials: the experimenter removed the tray of peanuts after 15 min for the monkeys who chose the 20 unit tray so they only had the time to consume 6 to 8 peanuts which is less than if they had chosen the 10 unit tray. After this manipulation the monkey anticipated the future for a better reward and the choice for 20 peanuts dropped to 50%. Although their initial instinct would be to choose the bigger quantity, the monkeys were able to plan for their long-term benefit and chose the smaller quantity (as cited in Roberts, 2012). A limitation to this experiment is that it could be explained by associative learning and therefore may represent a manifestation of semantic memory rather than episodic so other experiments also controlled the associative learning element by asking an unexpected question. Like Zentall et al. experiment where pigeons had to respond to whether they pecked or not at a certain response key; first the pigeons had to acquire some semantic information: they learned to choose the red stimulus if they had recently pecked at an initial stimulus and the green otherwise (association between red and pecking; green and not pecking). In the second part of the experiments the pigeons were exposed to different stimulus, the idea behind this second exposure is to make them peck at one stimulus that will be followed by food (yellow stimulus) and to abstain them from pecking at another stimulus (blue) that’s not followed by any reinforcement. They were then re-exposed to a green and a red stimulus, when the pigeons had received food in the second part of the experiment (pecked when exposed to yellow stimulus) they had a significant tendency to chose the red and when they didn’t they chose the green. In other words they learned to associate pecking with red and not pecking with blue and instinctively pecked when the second stimulus yellow was followed by food and refrained from doing so when the blue stimuli wasn’t reinforced. So their final choice (either red or green) is a way of responding to the question “ Did you or did you not peck in the past? ” and the fact that the subjects didn’t encounter this association before (there’s no association between whether they pecked or not and the yellow or blue stimuli) and were still capable of doing so when asked shows episodic-like memory. (as cited in Zentall, 2006) Even though the experiments were able to manipulate flexibility, current state of the subject, associative learning it’s hard to distinguish between semantic knowledge and episodic knowledge in animals specially when planning for the future; an explanation of the previous experiment may be an ongoing trace memory from the first phases of the experiment. So in a nutshell, experiments studied mental time travel in animals based on two main perspectives: www memory and planning for the future. Both showed episode-like components in animal’s memory but this can’t be taken as evidence for mental time travel. The main challenge for studying mental time travel in animals is their incapacity to generate language that is the main technique used to study mental time travel in humans. However according to Morris and Frey the automatic part of episodic memory is present in any creature that has hippocampal long-term potential, responsible for encoding the whole setting (time and space) of an event so animals posses this capacity (as cited in Cheke and Clayton 2010). Also, many insights to further research this issue were proposed, they need to be further studied but they show potential to prove or disprove the capacity of episodic memory in animals. Feeney and Roberts pointed out that anticipation for the future was showed in animals whose survival depend on it so mental time travel may be a specific trait for these animals acquired through evolutionary adaptation vs. the trait being common to all species and contrasting these options may lead to interesting findings (as cited in Roberts, 2012). Another option would be to rely on the distinction between making a personal decision or a decision for someone else, which may be an indicator of the episodic vs. semantic memory (Cheke & Clayton, 2010). But until today no research was able to prove mental time travel in animals but if they do find conclusive evidence it would raise many ethical issues as for animals anticipating future or re-living past suffering.

References:

Cheke, L. G. , & Clayton, N. S. (2010). Mental time travel in animals. WIREs Cognitive Science

Roberts, W. , & Feeney, M. (2009). The comparative study of mental time travel. Trends in Cognitive Science.

Roberts, W. (2012). Evidence for future cognition in animals. Elsevier

Suddendorf, T. , & Corballisb, M. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and Brain Sciences (in Press) Cambridge University Press.

Terrace, H. S. , & Metcalfe, J. (2005). Missing Link in Cognition: Origins of Self-Reflective Consciousness. Cary, NC, USA: Oxford University Press, Incorporated.

Zentall, T. R. (2006). Mental time travel in animals: A challenging question. Elsevier