Once upon a time, cognitive scientists employed the notion of “cognitive system” to put forth substantive hypotheses about cognitive architecture. There was much debate then as there is now as to what it takes for a certain system that causally brings about a behavior to count as cognitive, let alone what it counts for two such systems to be distinct (De Brigard Reference De Brigard2017). Nevertheless, many would agree that at the very least, two postulated systems count as distinct if (1) there is a difference in the nature of the representations upon which they operate, and (2) they carry out different computations. These minimal requirements, for instance, undergird Atkinson and Shiffrin's (Reference Atkinson, Shiffrin, Spence and Spence1968) model of a short-term memory system – involving temporally limited computations and modality-dependent representations – distinct from a long-term memory system, which involves modality-independent representations and no temporal constraints on computations.
Our concern here is that Hoerl & McCormack's (H&M's) proposal may not meet these two minimal requirements. To illustrate this point, we offer two challenges. The first challenge pertains to the temporal updating system (TUS). According to H&M, the TUS operates upon representations of how things are in the world in the present. This does not mean, however, that the representations of the TUS are not time-sensitive: They are, as they can be updated to convey information about change. Simply put, the representations of the TUS can represent change because change is part of the representing, not the represented. As such, H&M argue, a creature with just a TUS can still represent an object as desirable (e.g., a worm, a pacifier) even when it is no longer visible in its surroundings. Moreover, by being able to capture information about non-present but desirable objects, the TUS enables creatures to represent primitive goals.
But how does a creature know that a representation has been updated? Consider the case of, say, a scrub jay returning to a previously cached worm, as in the Clayton and Dickinson (Reference Clayton and Dickinson1998) study. According to H&M's proposal, the scrub jay may decide to go for a worm rather than a peanut because the updated representation supplied by its TUS – unlike the representation of its current surroundings supplied by the perceptual system – represents the worm as an open possibility, that is, as a goal that is still available in the organism's future. By contrast, the scrub jay that goes for the peanut instead of the worm may do so because the TUS delivers an updated representation in which the worm is absent. Not only is this possibility closed, but this scrub jay is, presumably, no longer aware of its past existence. Their suggestion is that the modal profile of these scenarios is time-locked to the timing of the cache; indeed, H&M explicitly acknowledge that the TUS cannot handle updating that does not conform to the order in which the information was received. But there is plenty of animal research showing that whether an option is seen as available to an organism in the future is not clearly tied to the time in which it is learned. For example, rodents can return to prior less desirable choices (e.g., taking a long but likely open path in a maze) upon learning that a more desirable one (e.g., taking a short but likely closed path) is unavailable (Tolman & Honzik Reference Tolman and Honzik1930). In other words, rodents seem to be able to revisit a previous option upon learning that what appeared to be an open possibility is, sadly, closed. The flexibility in the rodent's updated representations suggests that the modal profile—whether an option counts as an open versus a closed possibility – is not fully accounted for by the timing of learning, and further suggests that the animal is either capable of representing alternatives as being in a possible future or a possible past (a representational constraint not afforded by the TUS), or else is capable of drawing contrastive inferences between outdated and updated representations (a move that makes the computations of the TUS suspiciously similar to the temporal reasoning system [TRS]).
The second challenge pertains to the TRS. In their discussion of infant and preschool children, H&M argue that the TRS represents “times arranged in a linear array such that each time occupies a unique unrepeated location in the array” (p. 39). It is implied that such is the proper concept of time. However, there is currently cross-cultural evidence of non-linear ways of thinking about time, including – but not limited to – cyclical, branching, and volumetric (Casasanto & Boroditsky Reference Casasanto and Boroditsky2008). Are these non-linear ways of thinking about time still handled by the TRS? What about reasoning about time within the framework of the theory of relativity, whereby under certain conditions – which physicists have reasoned about – linearity breaks down? Are these improper concepts of time or are they concepts of time that are not handled by the TRS? Given that we can reason about time in all sorts of non-linear ways, we think that it may be unnecessary to postulate the existence of a dedicated TRS that only operates with a particular kind of temporal representation. A more parsimonious account may ask us instead to focus on how acquired information about linear time can be manipulated in the same way as acquired information about non-linear time, and in turn how such information can be used for reasoning. Under this lens, children learn to reason about time thanks to the development of the same representational, attentional, and memory systems that enable them to reason about all sorts of things, rather than through the maturation of an independently dedicated TRS. Furthermore, under this account, the intra- and cross-cultural heterogeneity in conceptions of time is explained by variations in acquired representations and inferential strategies, rather than competing TRSs.
In sum, we think that there is need for further argument as to how H&M's proposal meets the minimal requirements (1) and (2).
You have
Access
Once upon a time, cognitive scientists employed the notion of “cognitive system” to put forth substantive hypotheses about cognitive architecture. There was much debate then as there is now as to what it takes for a certain system that causally brings about a behavior to count as cognitive, let alone what it counts for two such systems to be distinct (De Brigard Reference De Brigard2017). Nevertheless, many would agree that at the very least, two postulated systems count as distinct if (1) there is a difference in the nature of the representations upon which they operate, and (2) they carry out different computations. These minimal requirements, for instance, undergird Atkinson and Shiffrin's (Reference Atkinson, Shiffrin, Spence and Spence1968) model of a short-term memory system – involving temporally limited computations and modality-dependent representations – distinct from a long-term memory system, which involves modality-independent representations and no temporal constraints on computations.
Our concern here is that Hoerl & McCormack's (H&M's) proposal may not meet these two minimal requirements. To illustrate this point, we offer two challenges. The first challenge pertains to the temporal updating system (TUS). According to H&M, the TUS operates upon representations of how things are in the world in the present. This does not mean, however, that the representations of the TUS are not time-sensitive: They are, as they can be updated to convey information about change. Simply put, the representations of the TUS can represent change because change is part of the representing, not the represented. As such, H&M argue, a creature with just a TUS can still represent an object as desirable (e.g., a worm, a pacifier) even when it is no longer visible in its surroundings. Moreover, by being able to capture information about non-present but desirable objects, the TUS enables creatures to represent primitive goals.
But how does a creature know that a representation has been updated? Consider the case of, say, a scrub jay returning to a previously cached worm, as in the Clayton and Dickinson (Reference Clayton and Dickinson1998) study. According to H&M's proposal, the scrub jay may decide to go for a worm rather than a peanut because the updated representation supplied by its TUS – unlike the representation of its current surroundings supplied by the perceptual system – represents the worm as an open possibility, that is, as a goal that is still available in the organism's future. By contrast, the scrub jay that goes for the peanut instead of the worm may do so because the TUS delivers an updated representation in which the worm is absent. Not only is this possibility closed, but this scrub jay is, presumably, no longer aware of its past existence. Their suggestion is that the modal profile of these scenarios is time-locked to the timing of the cache; indeed, H&M explicitly acknowledge that the TUS cannot handle updating that does not conform to the order in which the information was received. But there is plenty of animal research showing that whether an option is seen as available to an organism in the future is not clearly tied to the time in which it is learned. For example, rodents can return to prior less desirable choices (e.g., taking a long but likely open path in a maze) upon learning that a more desirable one (e.g., taking a short but likely closed path) is unavailable (Tolman & Honzik Reference Tolman and Honzik1930). In other words, rodents seem to be able to revisit a previous option upon learning that what appeared to be an open possibility is, sadly, closed. The flexibility in the rodent's updated representations suggests that the modal profile—whether an option counts as an open versus a closed possibility – is not fully accounted for by the timing of learning, and further suggests that the animal is either capable of representing alternatives as being in a possible future or a possible past (a representational constraint not afforded by the TUS), or else is capable of drawing contrastive inferences between outdated and updated representations (a move that makes the computations of the TUS suspiciously similar to the temporal reasoning system [TRS]).
The second challenge pertains to the TRS. In their discussion of infant and preschool children, H&M argue that the TRS represents “times arranged in a linear array such that each time occupies a unique unrepeated location in the array” (p. 39). It is implied that such is the proper concept of time. However, there is currently cross-cultural evidence of non-linear ways of thinking about time, including – but not limited to – cyclical, branching, and volumetric (Casasanto & Boroditsky Reference Casasanto and Boroditsky2008). Are these non-linear ways of thinking about time still handled by the TRS? What about reasoning about time within the framework of the theory of relativity, whereby under certain conditions – which physicists have reasoned about – linearity breaks down? Are these improper concepts of time or are they concepts of time that are not handled by the TRS? Given that we can reason about time in all sorts of non-linear ways, we think that it may be unnecessary to postulate the existence of a dedicated TRS that only operates with a particular kind of temporal representation. A more parsimonious account may ask us instead to focus on how acquired information about linear time can be manipulated in the same way as acquired information about non-linear time, and in turn how such information can be used for reasoning. Under this lens, children learn to reason about time thanks to the development of the same representational, attentional, and memory systems that enable them to reason about all sorts of things, rather than through the maturation of an independently dedicated TRS. Furthermore, under this account, the intra- and cross-cultural heterogeneity in conceptions of time is explained by variations in acquired representations and inferential strategies, rather than competing TRSs.
In sum, we think that there is need for further argument as to how H&M's proposal meets the minimal requirements (1) and (2).