Hooray for Hoerl and McCormack's (H&M's) project on temporal cognition (TC). Their distinction between “temporal updating” (TU) and “temporal reasoning” (TR) adeptly demonstrates the sufficiency of the lower-order system, TU, to explain the TC of animals. To modify a trope from Dan Dennett: In order to keep perfect track of the changing states of affairs in the world, it is not requisite to know a thing about time (see Dennett Reference Dennett2009, for a discussion of Charles Darwin and Alan Turing's similar and respective “strange inversions of reasoning”).
Povinelli and colleagues have previously shown that H&M's analysis, mutatis mutandis, holds true across most (if not all) other domains of cognition (Penn et al. Reference Penn, Holyoak and Povinelli2008). Their “reinterpretation hypothesis” (RH) was initially advanced to explain the evolution of social cognition, and its central claims tightly parallel H&M's account of TC (Povinelli & Giambrone Reference Povinelli and Giambrone1999; Povinelli & Vonk Reference Povinelli and Vonk2004):
1. Human social cognition is composed of phylogenetically ancient mechanisms for reasoning about behavior (analogous to those that H&M describe for the TU system), and a uniquely human system that reinterprets those behavioral relations in terms of abstract mental states.
2. The two systems continue to operate in concert in modern humans.
3. The newer system is dependent on the older one, but the causal power of the older system can completely explain the results of tests with animals.
4. Because of (3), RH is not an ad hoc alternative to higher-order accounts of animal social cognition.
The RH was later extended to other domains of cognition, including concept formation, physical causality (tool use), reasoning about weight, and even TC (Povinelli Reference Povinelli2000; Reference Povinelli2012; Vonk & Povinelli Reference Vonk, Povinelli, Wasserman and Zentall2006). Finally, Penn et al. (Reference Penn, Holyoak and Povinelli2008) specified the domain general format of the RH, arguing that the ability to cognize over higher-order, role-based analogical relations is a uniquely human capacity cutting across every domain of cognition.
In this view, “time” is one of myriad, higher-order relations the human mind constructs. The bedrock distinction of TR is the ability of humans to group innumerable (indeed, any) individual perceptual relations (leaves falling, sands running through an hourglass, gray hairs erupting on one's head, etc.) as temporal relations. H&M note that the human naïve (or folk) theory of time is yet to be fully explicated, and offer the interesting claim that one feature might be the idea that time “flows.” This may be true, but all human babies share the capacity to be enculturated into any theory of time (scientific or otherwise). Why? Because the human mind allows for disparate perceptual relations to be grouped under common thematic or argumentative roles—a hallmark signature of all higher-order, role-based relations (Penn et al. Reference Penn, Holyoak and Povinelli2008).
Since the most general statement of the RH was published in the pages of this journal a decade ago, dozens of empirical studies with animals have challenged the view that only humans reinterpret first-order perceptual relations in terms of higher-order relations. But all the demonstrations we have examined suffer from the same logical limitation that Povinelli and colleagues (and herein, H&M) have identified—namely, that first-order relational reasoning is necessary, but not sufficient for higher-order relational reasoning:
Same/different judgments?
Animals are presented with a sample of two (or more) objects that are either all the same (AA) or different (BC), and then can learn to select alternatives that match the relation (i.e., DD or EF). Are such performances evidence that animals possess the higher-order relations of same/different as some have claimed (e.g., Flemming et al. Reference Flemming, Thompson and Fagot2013)? No, because to form such higher-order relations, a cognizer must first detect the amount of perceptual variability in the displays (zero variability for same, higher variability for different). Once such perceptual variability is detected, however, this information can be used to sort novel exemplars.
Spatial analogies?
Haun and Call (Reference Haun and Call2009) claim that chimpanzees can recognize relational similarity between perceptually distinct predictors of food location. Subjects were confronted with a tilted table that contained three equally spaced out beyond their reach (“far”) cups and three within reach (“near”) cups. The second and third near cups were increasingly spatially misaligned with the far cups. In one condition, opaque tubes connected the experimenter's cups to the subjects’ cups. In another, painted lines “connected” them. The chimpanzees saw food dropped in a far cup and successfully searched in the near cups that were connected by the tubes or the lines. These apes were clearly tracking spatial relations (e.g., “if food is placed to extreme right, orient to that side” or “if cups are touched by a tube, pick it”), but there is no reason to think they constructed an analogy between the spatial relations of lines and tubes as suggested by the authors. Christie et al. (Reference Christie, Gentner, Call and Haun2016) recently claimed that chimpanzees are sensitive to the spatial analogy between a three-tiered shelf and another, identical one located nearby. While this is evidence that space guides searching (“food located low, continue to search low”), it is a far cry from higher-order relational reasoning.
Analogy in tool use?
Taylor et al. (Reference Taylor, Hunt, Holzhaider and Gray2007) demonstrated that crows use a short stick to retrieve a longer, functional stick. They suggest the crows may have done so by cognizing over the causal analogy between short and long sticks (“tools access out of reach objects”). This task may or may not be “cognitively demanding,” but it can certainly be solved by detecting the spatial distance between the subject and the goal object and the length of the stick.
Theory of mind?
Bugnyar et al. (Reference Bugnyar, Reber and Buckner2016) showed that ravens that hear (but do not see) a conspecific in an adjoining room are sensitive to the presence/absence of a small hole in the wall between the rooms. They interpret this as evidence that the subjects can imagine the mental state of the other raven seeing them. This experiment was designed to rule out the deflationary account of previous studies, wherein animals need only track and use the relations between conspecific location and unobstructed geometric paths. But there is nothing higher-order about an organism constructing a geometric relation based on “hearing” (as opposed to “seeing”) a nearby conspecific.
A flood of additional claims for higher-order thinking in animals have surfaced on topics such as the appearance-reality distinction, metacognition, intentional communication, water displacement, logical inference, false beliefs, love, morality, maps, gravity, altruism, mourning the dead, self-recognition, teaching, cooperation, and physical cognition (Povinelli & Barker Reference Povinelli and Barker2019). We contend each of these claims can be dismantled in the manner that H&M have for TC, and we have done for other cognitive domains.
Given that such a straightforward issue lies at the heart of innumerable confusions in animal cognition (Penn & Povinelli Reference Penn, Povinelli, Watanabe, Huber, Blaisdel and Young2009), why is it consistently ignored by comparative psychologists? While we encourage others to remain hopeful that H&M's master class on TC will lead to a sudden sea change, we remain cautious. Is there something so folk-psychologically compelling about tales of higher-order thinking in animals that even scientists cannot escape them? If so, comparative psychologists may well go on telling such animal tales as long as humans go on telling stories (Barker & Povinelli Reference Barker and Povinelli2019). H&M's heroic efforts would then be destined to sink into the mythic sea of “lost knowledge”—that ever-receding ocean of hard-won truths humans are fated to continually rediscover.
Hooray for Hoerl and McCormack's (H&M's) project on temporal cognition (TC). Their distinction between “temporal updating” (TU) and “temporal reasoning” (TR) adeptly demonstrates the sufficiency of the lower-order system, TU, to explain the TC of animals. To modify a trope from Dan Dennett: In order to keep perfect track of the changing states of affairs in the world, it is not requisite to know a thing about time (see Dennett Reference Dennett2009, for a discussion of Charles Darwin and Alan Turing's similar and respective “strange inversions of reasoning”).
Povinelli and colleagues have previously shown that H&M's analysis, mutatis mutandis, holds true across most (if not all) other domains of cognition (Penn et al. Reference Penn, Holyoak and Povinelli2008). Their “reinterpretation hypothesis” (RH) was initially advanced to explain the evolution of social cognition, and its central claims tightly parallel H&M's account of TC (Povinelli & Giambrone Reference Povinelli and Giambrone1999; Povinelli & Vonk Reference Povinelli and Vonk2004):
1. Human social cognition is composed of phylogenetically ancient mechanisms for reasoning about behavior (analogous to those that H&M describe for the TU system), and a uniquely human system that reinterprets those behavioral relations in terms of abstract mental states.
2. The two systems continue to operate in concert in modern humans.
3. The newer system is dependent on the older one, but the causal power of the older system can completely explain the results of tests with animals.
4. Because of (3), RH is not an ad hoc alternative to higher-order accounts of animal social cognition.
The RH was later extended to other domains of cognition, including concept formation, physical causality (tool use), reasoning about weight, and even TC (Povinelli Reference Povinelli2000; Reference Povinelli2012; Vonk & Povinelli Reference Vonk, Povinelli, Wasserman and Zentall2006). Finally, Penn et al. (Reference Penn, Holyoak and Povinelli2008) specified the domain general format of the RH, arguing that the ability to cognize over higher-order, role-based analogical relations is a uniquely human capacity cutting across every domain of cognition.
In this view, “time” is one of myriad, higher-order relations the human mind constructs. The bedrock distinction of TR is the ability of humans to group innumerable (indeed, any) individual perceptual relations (leaves falling, sands running through an hourglass, gray hairs erupting on one's head, etc.) as temporal relations. H&M note that the human naïve (or folk) theory of time is yet to be fully explicated, and offer the interesting claim that one feature might be the idea that time “flows.” This may be true, but all human babies share the capacity to be enculturated into any theory of time (scientific or otherwise). Why? Because the human mind allows for disparate perceptual relations to be grouped under common thematic or argumentative roles—a hallmark signature of all higher-order, role-based relations (Penn et al. Reference Penn, Holyoak and Povinelli2008).
Since the most general statement of the RH was published in the pages of this journal a decade ago, dozens of empirical studies with animals have challenged the view that only humans reinterpret first-order perceptual relations in terms of higher-order relations. But all the demonstrations we have examined suffer from the same logical limitation that Povinelli and colleagues (and herein, H&M) have identified—namely, that first-order relational reasoning is necessary, but not sufficient for higher-order relational reasoning:
Same/different judgments?
Animals are presented with a sample of two (or more) objects that are either all the same (AA) or different (BC), and then can learn to select alternatives that match the relation (i.e., DD or EF). Are such performances evidence that animals possess the higher-order relations of same/different as some have claimed (e.g., Flemming et al. Reference Flemming, Thompson and Fagot2013)? No, because to form such higher-order relations, a cognizer must first detect the amount of perceptual variability in the displays (zero variability for same, higher variability for different). Once such perceptual variability is detected, however, this information can be used to sort novel exemplars.
Spatial analogies?
Haun and Call (Reference Haun and Call2009) claim that chimpanzees can recognize relational similarity between perceptually distinct predictors of food location. Subjects were confronted with a tilted table that contained three equally spaced out beyond their reach (“far”) cups and three within reach (“near”) cups. The second and third near cups were increasingly spatially misaligned with the far cups. In one condition, opaque tubes connected the experimenter's cups to the subjects’ cups. In another, painted lines “connected” them. The chimpanzees saw food dropped in a far cup and successfully searched in the near cups that were connected by the tubes or the lines. These apes were clearly tracking spatial relations (e.g., “if food is placed to extreme right, orient to that side” or “if cups are touched by a tube, pick it”), but there is no reason to think they constructed an analogy between the spatial relations of lines and tubes as suggested by the authors. Christie et al. (Reference Christie, Gentner, Call and Haun2016) recently claimed that chimpanzees are sensitive to the spatial analogy between a three-tiered shelf and another, identical one located nearby. While this is evidence that space guides searching (“food located low, continue to search low”), it is a far cry from higher-order relational reasoning.
Analogy in tool use?
Taylor et al. (Reference Taylor, Hunt, Holzhaider and Gray2007) demonstrated that crows use a short stick to retrieve a longer, functional stick. They suggest the crows may have done so by cognizing over the causal analogy between short and long sticks (“tools access out of reach objects”). This task may or may not be “cognitively demanding,” but it can certainly be solved by detecting the spatial distance between the subject and the goal object and the length of the stick.
Theory of mind?
Bugnyar et al. (Reference Bugnyar, Reber and Buckner2016) showed that ravens that hear (but do not see) a conspecific in an adjoining room are sensitive to the presence/absence of a small hole in the wall between the rooms. They interpret this as evidence that the subjects can imagine the mental state of the other raven seeing them. This experiment was designed to rule out the deflationary account of previous studies, wherein animals need only track and use the relations between conspecific location and unobstructed geometric paths. But there is nothing higher-order about an organism constructing a geometric relation based on “hearing” (as opposed to “seeing”) a nearby conspecific.
A flood of additional claims for higher-order thinking in animals have surfaced on topics such as the appearance-reality distinction, metacognition, intentional communication, water displacement, logical inference, false beliefs, love, morality, maps, gravity, altruism, mourning the dead, self-recognition, teaching, cooperation, and physical cognition (Povinelli & Barker Reference Povinelli and Barker2019). We contend each of these claims can be dismantled in the manner that H&M have for TC, and we have done for other cognitive domains.
Given that such a straightforward issue lies at the heart of innumerable confusions in animal cognition (Penn & Povinelli Reference Penn, Povinelli, Watanabe, Huber, Blaisdel and Young2009), why is it consistently ignored by comparative psychologists? While we encourage others to remain hopeful that H&M's master class on TC will lead to a sudden sea change, we remain cautious. Is there something so folk-psychologically compelling about tales of higher-order thinking in animals that even scientists cannot escape them? If so, comparative psychologists may well go on telling such animal tales as long as humans go on telling stories (Barker & Povinelli Reference Barker and Povinelli2019). H&M's heroic efforts would then be destined to sink into the mythic sea of “lost knowledge”—that ever-receding ocean of hard-won truths humans are fated to continually rediscover.