Hoerl & McCormack (H&M) offer a two-system account of temporal cognition, arguing that only humans evolved a second, more advanced system, which allows reasoning about temporal relations. This view falls into a tradition of views on human uniqueness traceable back to at least C. Lloyd Morgan (Reference Morgan1903) and perhaps even to Aristotle (Sorabji Reference Sorabji1993), who held that the representation of relations is a uniquely human cognitive achievement (Clatterbuck Reference Clatterbuck2016). Morganian skepticism can be applied domain-generally (Penn et al. Reference Penn, Holyoak and Povinelli2008) or only to relations in a particular domain, such as meta-cognition (Carruthers Reference Carruthers2009). H&M seem to offer a version specific to temporal cognition, arguing that whereas animals can represent objects and properties that they encounter at different times in a maplike representation of their environment, they cannot represent those events as occurring at different times in a systematic way.
A generic strength of Morganian skepticism is that nonrelational cognition has intuitively clear signature limits. This is where H&M's view shines; in particular, they suggest that animals cannot learn information about events in an order other than that in which those events actually occur. Morganian skepticism, however, faces a generic challenge. After conceding to animals the representational flexibility required to explain their flexible behavior, it becomes difficult to explain why they cannot learn to represent relations. The boring answer to this question – boring because it renders all other cognitive differences between humans and animals derivative from the most obvious one – is that representing relations requires language, and animals lack language. The challenge facing H&M is particularly tricky, for they seem to concede that nonlinguistic animals possess mechanisms to represent spatial relations. Here, we ask why these mechanisms could not be bootstrapped to represent an additional temporal dimension as well, albeit less precisely than humans do so with temporal language.
To put some pieces on the workbench, let us briefly explore how animals might represent spatial dimensions. In mammals, at least, the most popular story has to do with the way that place cells and grid cells in the medial temporal lobes cooperate to build a maplike representation of their environment (Moser et al. Reference Moser, Kropff and Moser2008). Skipping over many details, place cells represent locations by binding together snapshots taken from different viewpoints in the same spot, and grid cells represent the locations of these bundles with respect to one another in spatial dimensions by linking them to a spatially organized array. Visible landmarks probably play an important role in establishing these links. The grids are anchored by landmarks, and the same landmarks are visible from different egocentric viewpoints.
To introduce a new component to the conversation, some psychologists have argued that temporal landmarks play a similar role in anchoring a system of temporal representation in children (Shum Reference Shum1998; Tartas Reference Tartas2001). Children may start by placing events with respect to an especially memorable life transition (“before I started preschool,” “after we moved to the new house”), and then begin placing events in an order with respect to a few indices sequenced by major events like birthdays (“when I was 3,” “when I was 4”). Only later, after learning more ordered sequences of temporal landmarks, can they progress to a series of hierarchically nested indices that enable the level of fine-grained, monotonically ordered temporal representation that characterizes adult human cognition (times of the day, days of the week, months of the year, and so on [Jack et al. Reference Jack, Friedman, Reese and Zajac2016]). Neuroscientific and psychological evidence further supports the idea that some of the mechanisms that represent spatial relations can be redeployed to represent temporal relations as well (Casasanto & Boroditsky Reference Casasanto and Boroditsky2008; Eichenbaum Reference Eichenbaum2017).
This view on development in humans suggests a different perspective on the human-animal divide in temporal cognition. These structural and functional parallels make it fairly plausible that some animals that use spatial landmarks might thereby possess mechanisms to use temporal landmarks, too (Buzsáki & Moser Reference Buzsáki and Moser2013; Ekstrom & Ranganath Reference Ekstrom and Ranganath2018; Naya & Suzuki Reference Naya and Suzuki2011). However, we might attribute the absence of human-like levels of temporal systematicity in animals not to the monolithic inability to represent temporal relations, but rather to the comparative scarcity of temporal landmark cues in animals’ lives. Animals live shorter lives with brief neotony and fewer stark autobiographical transitions, and without language, it is harder to provide animals with temporal landmark cues at learning or recall. In short, rather than animals failing to evolve a temporal reasoning system because they would not need it, perhaps it is instead because they typically fail to develop one, despite possessing the requisite neural machinery, because they are not exposed to enough of the necessary scaffolding.
This alternative perspective also suggests a different tack for future research. Before putting animals in experiments that test their ability to deploy temporal relations, we should expose them to temporal landmarking cues that could let them place events as occurring before or after the landmark, and provide them with tasks where they would be rewarded for ordering those events with respect to those landmarks and, consequently, each other in systematic and novel ways. The generic challenge facing this response to Morganian skepticism is that the landmark cues must not simultaneously provide animals with information about repeatable abstract properties of their world that have been previously rewarded with outcomes that could allow them to independently solve the task. This is a very difficult experimental challenge, but one not so different from similar difficulties in other domains that have already been overcome, for example, providing animals with evidence of mental relations without providing confounding behavior-reading cues (Bugnyar et al. Reference Bugnyar, Reber and Buckner2016; Karg et al. Reference Karg, Schmelz, Call and Tomasello2016; Penn & Povinelli Reference Penn and Povinelli2007). This might mean that we should focus on temporal cues that are associated not with things like times of day or seasons, which recur and carry many associations to rewarded outcomes themselves, but rather to personally significant events that may occur only once in an animal's life, such as a dominance reversal in a social hierarchy or movement to a new enclosure.
You have
Access
Hoerl & McCormack (H&M) offer a two-system account of temporal cognition, arguing that only humans evolved a second, more advanced system, which allows reasoning about temporal relations. This view falls into a tradition of views on human uniqueness traceable back to at least C. Lloyd Morgan (Reference Morgan1903) and perhaps even to Aristotle (Sorabji Reference Sorabji1993), who held that the representation of relations is a uniquely human cognitive achievement (Clatterbuck Reference Clatterbuck2016). Morganian skepticism can be applied domain-generally (Penn et al. Reference Penn, Holyoak and Povinelli2008) or only to relations in a particular domain, such as meta-cognition (Carruthers Reference Carruthers2009). H&M seem to offer a version specific to temporal cognition, arguing that whereas animals can represent objects and properties that they encounter at different times in a maplike representation of their environment, they cannot represent those events as occurring at different times in a systematic way.
A generic strength of Morganian skepticism is that nonrelational cognition has intuitively clear signature limits. This is where H&M's view shines; in particular, they suggest that animals cannot learn information about events in an order other than that in which those events actually occur. Morganian skepticism, however, faces a generic challenge. After conceding to animals the representational flexibility required to explain their flexible behavior, it becomes difficult to explain why they cannot learn to represent relations. The boring answer to this question – boring because it renders all other cognitive differences between humans and animals derivative from the most obvious one – is that representing relations requires language, and animals lack language. The challenge facing H&M is particularly tricky, for they seem to concede that nonlinguistic animals possess mechanisms to represent spatial relations. Here, we ask why these mechanisms could not be bootstrapped to represent an additional temporal dimension as well, albeit less precisely than humans do so with temporal language.
To put some pieces on the workbench, let us briefly explore how animals might represent spatial dimensions. In mammals, at least, the most popular story has to do with the way that place cells and grid cells in the medial temporal lobes cooperate to build a maplike representation of their environment (Moser et al. Reference Moser, Kropff and Moser2008). Skipping over many details, place cells represent locations by binding together snapshots taken from different viewpoints in the same spot, and grid cells represent the locations of these bundles with respect to one another in spatial dimensions by linking them to a spatially organized array. Visible landmarks probably play an important role in establishing these links. The grids are anchored by landmarks, and the same landmarks are visible from different egocentric viewpoints.
To introduce a new component to the conversation, some psychologists have argued that temporal landmarks play a similar role in anchoring a system of temporal representation in children (Shum Reference Shum1998; Tartas Reference Tartas2001). Children may start by placing events with respect to an especially memorable life transition (“before I started preschool,” “after we moved to the new house”), and then begin placing events in an order with respect to a few indices sequenced by major events like birthdays (“when I was 3,” “when I was 4”). Only later, after learning more ordered sequences of temporal landmarks, can they progress to a series of hierarchically nested indices that enable the level of fine-grained, monotonically ordered temporal representation that characterizes adult human cognition (times of the day, days of the week, months of the year, and so on [Jack et al. Reference Jack, Friedman, Reese and Zajac2016]). Neuroscientific and psychological evidence further supports the idea that some of the mechanisms that represent spatial relations can be redeployed to represent temporal relations as well (Casasanto & Boroditsky Reference Casasanto and Boroditsky2008; Eichenbaum Reference Eichenbaum2017).
This view on development in humans suggests a different perspective on the human-animal divide in temporal cognition. These structural and functional parallels make it fairly plausible that some animals that use spatial landmarks might thereby possess mechanisms to use temporal landmarks, too (Buzsáki & Moser Reference Buzsáki and Moser2013; Ekstrom & Ranganath Reference Ekstrom and Ranganath2018; Naya & Suzuki Reference Naya and Suzuki2011). However, we might attribute the absence of human-like levels of temporal systematicity in animals not to the monolithic inability to represent temporal relations, but rather to the comparative scarcity of temporal landmark cues in animals’ lives. Animals live shorter lives with brief neotony and fewer stark autobiographical transitions, and without language, it is harder to provide animals with temporal landmark cues at learning or recall. In short, rather than animals failing to evolve a temporal reasoning system because they would not need it, perhaps it is instead because they typically fail to develop one, despite possessing the requisite neural machinery, because they are not exposed to enough of the necessary scaffolding.
This alternative perspective also suggests a different tack for future research. Before putting animals in experiments that test their ability to deploy temporal relations, we should expose them to temporal landmarking cues that could let them place events as occurring before or after the landmark, and provide them with tasks where they would be rewarded for ordering those events with respect to those landmarks and, consequently, each other in systematic and novel ways. The generic challenge facing this response to Morganian skepticism is that the landmark cues must not simultaneously provide animals with information about repeatable abstract properties of their world that have been previously rewarded with outcomes that could allow them to independently solve the task. This is a very difficult experimental challenge, but one not so different from similar difficulties in other domains that have already been overcome, for example, providing animals with evidence of mental relations without providing confounding behavior-reading cues (Bugnyar et al. Reference Bugnyar, Reber and Buckner2016; Karg et al. Reference Karg, Schmelz, Call and Tomasello2016; Penn & Povinelli Reference Penn and Povinelli2007). This might mean that we should focus on temporal cues that are associated not with things like times of day or seasons, which recur and carry many associations to rewarded outcomes themselves, but rather to personally significant events that may occur only once in an animal's life, such as a dominance reversal in a social hierarchy or movement to a new enclosure.