Quilty-Dunn et al. claim that “complex infant and animal reasoning […] all implicate LOT-like structures” (target article, long abstract). The authors explain that recent evidence in comparative psychology shows that “The use of abstract content in physical reasoning is arguably present throughout the animal kingdom” (target article, sect. 5.1, para. 4), and, they say, these findings are compatible with their proposal that language-of-thought hypothesis (LOTH)-based accounts have the potential to explain all animal cognition. We will comment on the authors' claim, given the authors' interpretation of the findings they choose to exemplify in comparative psychology and given their account of animal reasoning and the nature of mental representations they take it to involve.
We agree with the authors that the mental life of animals can be explained by referring to various formats and architectures; we even agree that representational approaches are well suited to standout as flexible and useful explanatory tools. However, we suggest that from accepting that representationalist accounts of the mind are suitable to explain animal minds, to claiming that mental representations are sentence-like in nonhuman animals, is yet a big step. We want to comment on the authors' proposal that LOTH accounts must hold to six core properties, and in particular, we question the sixth property: Abstract conceptual content. We do not deny that complex cognitive processes take place in much animal mental life, on the contrary. But we disagree that such complexity is best explained by appealing to abstract concepts.
According to LOTH, the language-of-thought hypothesis, mental representations are formatted like sentences (Fodor, Reference Fodor1975, Reference Fodor1987). A LOTHA, language-of-thought hypothesis for animals, has since been discussed (see Beck, Reference Beck, Andrews and Beck2017), also on the ground that LOTH can be split up into strong-LOT, that understand the compositionality of representations as involving the mechanisms proper of natural languages, and weak-LOT, that maintains the compositionality of representations (Camp, Reference Camp2007, Reference Camp and Lurz2009). To date, there is no direct evidence for LOTHA, but there is evidence that some animal mental representations are not sentence-like (see work on analogue magnitudes in Beck, Reference Beck2015, Reference Beck, Andrews and Beck2017).
We agree with Fitch's (Reference Fitch2020) mentalistic, yet physicalist, perspective that a concept is simply “a nonlinguistic psychological representation of a class of entities in the world.” Abstract concepts, in particular, refer to abstract entities, that is, those entities that are at least not directly perceptually available. We argue that animals may not need abstract concepts to engage in complex reasoning. Their mental life is based on representational mechanisms that need not involve abstract concepts. We argue that animals can realize complex activities relying on “minimal beliefs” without involving abstract concepts. Those consist of informational states that are sufficiently decoupled from motivational states, which allows for informational and motivational states to be combinable and organizable. In addition, these representations need to interact with epistemic dispositions to allow for the acquisition of novel information, for their (perception-based) categorization, and for constant updates (Newen & Starzak, Reference Newen and Starzak2020). We will put these minimal beliefs at work through an example from primate ethology: Orangutans' long calls (Askew & Morrogh-Bernard, Reference Askew and Morrogh-Bernard2016; Lameira & Call, Reference Lameira and Call2018; Spillman et al., Reference Spillman, van Noordwijk, Willems, Mitra Setia, Wipfli and van Schaik2015; van Schaik, Damerius, & Isler, Reference van Schaik, Damerius and Isler2013). These long calls not only lack any involvement of abstract concepts but, as we will elaborate, there are also neither concrete constituents nor a predicate–argument structure involved in them.
An observational study by van Schaik et al. (Reference van Schaik, Damerius and Isler2013) examined the extent to which the direction of long calls emitted by flanged male Sumatran orangutans (Pongo abelii) and Bornean orangutans (Pongo pygmaeus wurmbii) indicated the direction of their future travel. These animals live in a very dense tropical forest and are semi-solitary, thus often out of sight from other members of their population. For this reason, their communicative repertoire is distinctively (though not exclusively) more vocal than that of other apes. Flanged male orangutans use long calls to indicate to female members their future travel direction. These male individuals perform these calls when stationary. And these vocalizations can anticipate the direction of their travel 1 day ahead. In response, females show receptive behaviour by travelling in the direction indicated by the long calls. The study of this communicative strategy focused on three questions: First, testing to what extent the direction in which flanged male Sumatran orangutans give spontaneous long calls predicts their travel direction accurately. Second, if the initial calls are followed by additional spontaneous long calls that indicate the subsequent travel direction with more precision than the old one would if no new call had been given. Third, the extent to which long calls that are given in the evening from the night nest or in its proximity still indicate travel direction during the next day, thus indicating future planning independent of the current motivational state.
The capacity displayed by flanged male orangutans to communicate their future travel directions, and the corresponding ability displayed by the females to be receptive to such communicative intentions, is readily explained through the framework of minimal beliefs.
The long calls communicate spatial and temporal information about future travels. This information is first processed, then stored, and eventually reactivated and integrated at the time of the day they will need to be used to guide the travelling. We suggest that picking up different categories of information (direction = inferred by the orientation of the male performing the call; distance = inferred by the loudness of the call; time = inferred by the intervals between one call and the following ones) about the same event ( = travel), then processing, storing, and retrieving them can be managed by combinable and organizable informational states like minimal beliefs and that is important evidence of the representational capacities of orangutans. But this does not imply any LOTH-like structure of the mental representation involved in the long calls. More precisely, we do not need to presuppose any language-like syntactic structure, and, specifically, no subject–predicate structure for the long calls.
Even if the authors suggest understanding all communication and reasoning through language-like structures in a wide sense, to justify compositionality, this use of the LOTH would result inflationary and the hypothesis itself would lose its explanatory power. Thus, the LOTH still is not the key explanatory framework to understanding complex cognition in nonhuman animals.
You have
Access
Quilty-Dunn et al. claim that “complex infant and animal reasoning […] all implicate LOT-like structures” (target article, long abstract). The authors explain that recent evidence in comparative psychology shows that “The use of abstract content in physical reasoning is arguably present throughout the animal kingdom” (target article, sect. 5.1, para. 4), and, they say, these findings are compatible with their proposal that language-of-thought hypothesis (LOTH)-based accounts have the potential to explain all animal cognition. We will comment on the authors' claim, given the authors' interpretation of the findings they choose to exemplify in comparative psychology and given their account of animal reasoning and the nature of mental representations they take it to involve.
We agree with the authors that the mental life of animals can be explained by referring to various formats and architectures; we even agree that representational approaches are well suited to standout as flexible and useful explanatory tools. However, we suggest that from accepting that representationalist accounts of the mind are suitable to explain animal minds, to claiming that mental representations are sentence-like in nonhuman animals, is yet a big step. We want to comment on the authors' proposal that LOTH accounts must hold to six core properties, and in particular, we question the sixth property: Abstract conceptual content. We do not deny that complex cognitive processes take place in much animal mental life, on the contrary. But we disagree that such complexity is best explained by appealing to abstract concepts.
According to LOTH, the language-of-thought hypothesis, mental representations are formatted like sentences (Fodor, Reference Fodor1975, Reference Fodor1987). A LOTHA, language-of-thought hypothesis for animals, has since been discussed (see Beck, Reference Beck, Andrews and Beck2017), also on the ground that LOTH can be split up into strong-LOT, that understand the compositionality of representations as involving the mechanisms proper of natural languages, and weak-LOT, that maintains the compositionality of representations (Camp, Reference Camp2007, Reference Camp and Lurz2009). To date, there is no direct evidence for LOTHA, but there is evidence that some animal mental representations are not sentence-like (see work on analogue magnitudes in Beck, Reference Beck2015, Reference Beck, Andrews and Beck2017).
We agree with Fitch's (Reference Fitch2020) mentalistic, yet physicalist, perspective that a concept is simply “a nonlinguistic psychological representation of a class of entities in the world.” Abstract concepts, in particular, refer to abstract entities, that is, those entities that are at least not directly perceptually available. We argue that animals may not need abstract concepts to engage in complex reasoning. Their mental life is based on representational mechanisms that need not involve abstract concepts. We argue that animals can realize complex activities relying on “minimal beliefs” without involving abstract concepts. Those consist of informational states that are sufficiently decoupled from motivational states, which allows for informational and motivational states to be combinable and organizable. In addition, these representations need to interact with epistemic dispositions to allow for the acquisition of novel information, for their (perception-based) categorization, and for constant updates (Newen & Starzak, Reference Newen and Starzak2020). We will put these minimal beliefs at work through an example from primate ethology: Orangutans' long calls (Askew & Morrogh-Bernard, Reference Askew and Morrogh-Bernard2016; Lameira & Call, Reference Lameira and Call2018; Spillman et al., Reference Spillman, van Noordwijk, Willems, Mitra Setia, Wipfli and van Schaik2015; van Schaik, Damerius, & Isler, Reference van Schaik, Damerius and Isler2013). These long calls not only lack any involvement of abstract concepts but, as we will elaborate, there are also neither concrete constituents nor a predicate–argument structure involved in them.
An observational study by van Schaik et al. (Reference van Schaik, Damerius and Isler2013) examined the extent to which the direction of long calls emitted by flanged male Sumatran orangutans (Pongo abelii) and Bornean orangutans (Pongo pygmaeus wurmbii) indicated the direction of their future travel. These animals live in a very dense tropical forest and are semi-solitary, thus often out of sight from other members of their population. For this reason, their communicative repertoire is distinctively (though not exclusively) more vocal than that of other apes. Flanged male orangutans use long calls to indicate to female members their future travel direction. These male individuals perform these calls when stationary. And these vocalizations can anticipate the direction of their travel 1 day ahead. In response, females show receptive behaviour by travelling in the direction indicated by the long calls. The study of this communicative strategy focused on three questions: First, testing to what extent the direction in which flanged male Sumatran orangutans give spontaneous long calls predicts their travel direction accurately. Second, if the initial calls are followed by additional spontaneous long calls that indicate the subsequent travel direction with more precision than the old one would if no new call had been given. Third, the extent to which long calls that are given in the evening from the night nest or in its proximity still indicate travel direction during the next day, thus indicating future planning independent of the current motivational state.
The capacity displayed by flanged male orangutans to communicate their future travel directions, and the corresponding ability displayed by the females to be receptive to such communicative intentions, is readily explained through the framework of minimal beliefs.
The long calls communicate spatial and temporal information about future travels. This information is first processed, then stored, and eventually reactivated and integrated at the time of the day they will need to be used to guide the travelling. We suggest that picking up different categories of information (direction = inferred by the orientation of the male performing the call; distance = inferred by the loudness of the call; time = inferred by the intervals between one call and the following ones) about the same event ( = travel), then processing, storing, and retrieving them can be managed by combinable and organizable informational states like minimal beliefs and that is important evidence of the representational capacities of orangutans. But this does not imply any LOTH-like structure of the mental representation involved in the long calls. More precisely, we do not need to presuppose any language-like syntactic structure, and, specifically, no subject–predicate structure for the long calls.
Even if the authors suggest understanding all communication and reasoning through language-like structures in a wide sense, to justify compositionality, this use of the LOTH would result inflationary and the hypothesis itself would lose its explanatory power. Thus, the LOTH still is not the key explanatory framework to understanding complex cognition in nonhuman animals.
Acknowledgments
We thank the editors of Behavioural and Brain Sciences for their support and editorial supervision.
Financial support
This research was supported by the DFG-project (NE 576/14-1) “The structure and development of understanding actions and reasons” (Gefördert durch die Deutsche Forschungsge-meinschaft [DFG] – Projekt NE 576/14-1).
Competing interest
None.