Although we believe that a list of cognitive capacities involved in human tool use might be fruitful in other respects, we fail to see the gain from using it in the perspective of comparative cognition. Vaesen's list has been reverse engineered from modern human cognition. How does it help us to describe the evolution of human tool use to know whether or not a modern chimpanzee has a similar list? Vaesen claims that the list can at least explain the discrepancy between chimpanzee and human tool use. This may be, but no cognitive capacities need invoking to argue that humans and, for example, chimpanzees use tools differently. The paper could have been an analysis of the cognitive capacities themselves, or a paper on the evolution of human technology – but using a comparative approach does not tie the paper together.
We also find problems with Vaesen's treatment of the primatological research on cognitive capacities per se. Specifically, we want to comment on the analysis of foresight, function representation, and social intelligence.
Vaesen regards a study by Osvath and Osvath (Reference Osvath and Osvath2008) on chimpanzees and orangutans as coming closest to showing inhibition in relation to foresight in nonhuman primates. One of the study's four experimental conditions included an inhibition task where the subject could select either an immediate, favoured reward (a grape) or a tool that might be used 70 minutes in the future for retrieving a large, also highly favoured liquid reward (rosehip berry soup). All subjects performed significantly above what would be expected of an animal with no foresight. However, Vaesen dismisses the results as a consequence of mere associative learning, based on the misapprehensions of Suddendorf and Corballis (Reference Suddendorf and Corballis2009) in their commentary on the study. Not only does the original study contain an experiment specifically designed to control for associative learning; there also exists an extensive response to Suddendorf and Corballis (Osvath Reference Osvath2010) that Vaesen does not account for. There is little doubt that the inhibitory behaviour was not caused by purely associative learning.
Vaesen does, for sake of argument, suppose that associative learning can be excluded; but he then finds that the different rewards are not qualitatively different, which would imply that the apes are not taking a future state into account when selecting their tool. Vaesen argues that wanting to eat a grape and wanting to drink rosehip berry soup are the result of the same desire: namely, hunger.
First, hunger is not a desire; it is a response to a physiological state. In any case, the subjects in this study were not particularly hungry or thirsty: Their daily feeding routines had not been interrupted. Second, eating and drinking are dissimilar activities, with different physiological outcomes. For the time being, allow the most parsimonious interpretation to be that the apes select between two different desires.
Vaesen concludes from very few studies – two, it appears – that great ape foresight is highly limited compared with that of humans. However, the few great ape studies that exist simply have not tested for a range of foresight skills allegedly present in humans. Much work needs to be done before great ape foresight abilities can be delineated. We are plainly ignorant of significant facts about great ape foresight and also know little about the mechanisms of foresight in themselves. If a certain, uniquely human type of foresight is indeed essential for superior tool use, the evidence for such uniqueness, in the experimental literature that Vaesen cites, is insufficient.
Third, we have concerns with how Vaesen handles the abilities of great apes to represent tool function. We will here limit ourselves to the previously mentioned foresight study by Osvath and Osvath (Reference Osvath and Osvath2008). The fourth experimental condition in that study included novel items from which the apes were to select. Of the four items, only one was functional for retrieving the future reward. The subject had to decide, by visual inspection of each novel item, whether that item would be functional in the future. Stimulus generalisation – of length, colour, size, and shape – was precluded. The apes performed significantly above chance.
It is difficult to explain these results without ascribing to the apes some representation of function. This novel use would imply a type of causal understanding, according to Vaesen's line of argumentation. When it came to ascribing to a tool a particular function, it is obvious that, in the other three conditions in the study, the apes consistently chose the functional tool. The experimental condition for associative learning controlled more specifically for whether the tool was selected because of its function – rather than representing an arbitrary, reinforced stimulus. These results are not compatible with Vaesen's statement that primates do not attach particular functions to particular objects.
Fourth, Vaesen adopts Premack and Woodruff's (Reference Premack and Woodruff1978) characterization of theory of mind in the section on social intelligence. Since this definition was first proposed, the capacity for theory of mind has been shown to consist of several components (e.g., Call & Tomasello Reference Call and Tomasello2008; Gärdenfors Reference Gärdenfors2001; Reference Gärdenfors2003). One should at least distinguish understanding the emotions of others from understanding their attention, understanding their intentions, and understanding their beliefs. Only the last capacity is normally called a theory of mind. With the other capacities, it is fully possible for the agent to react to visible behaviours directly. Evidence for both the capacity to understand emotions and the capacity to understand attention can be found in non-human primates (Preston & de Waal Reference Preston and de Waal2002; see Call & Tomasello Reference Call and Tomasello2008). The situation is less clear for understanding intentions, although Tomasello et al. (Reference Tomasello, Carpenter, Call, Behne and Moll2005) claim the capacity to form joint intentions as the hallmark of humans. Only with respect to understanding the beliefs of others is there, so far, no evidence from non-human primates. A less anthropocentric research methodology might change the situation, as it did for research on understanding the attention of others.
Although we believe that a list of cognitive capacities involved in human tool use might be fruitful in other respects, we fail to see the gain from using it in the perspective of comparative cognition. Vaesen's list has been reverse engineered from modern human cognition. How does it help us to describe the evolution of human tool use to know whether or not a modern chimpanzee has a similar list? Vaesen claims that the list can at least explain the discrepancy between chimpanzee and human tool use. This may be, but no cognitive capacities need invoking to argue that humans and, for example, chimpanzees use tools differently. The paper could have been an analysis of the cognitive capacities themselves, or a paper on the evolution of human technology – but using a comparative approach does not tie the paper together.
We also find problems with Vaesen's treatment of the primatological research on cognitive capacities per se. Specifically, we want to comment on the analysis of foresight, function representation, and social intelligence.
Vaesen regards a study by Osvath and Osvath (Reference Osvath and Osvath2008) on chimpanzees and orangutans as coming closest to showing inhibition in relation to foresight in nonhuman primates. One of the study's four experimental conditions included an inhibition task where the subject could select either an immediate, favoured reward (a grape) or a tool that might be used 70 minutes in the future for retrieving a large, also highly favoured liquid reward (rosehip berry soup). All subjects performed significantly above what would be expected of an animal with no foresight. However, Vaesen dismisses the results as a consequence of mere associative learning, based on the misapprehensions of Suddendorf and Corballis (Reference Suddendorf and Corballis2009) in their commentary on the study. Not only does the original study contain an experiment specifically designed to control for associative learning; there also exists an extensive response to Suddendorf and Corballis (Osvath Reference Osvath2010) that Vaesen does not account for. There is little doubt that the inhibitory behaviour was not caused by purely associative learning.
Vaesen does, for sake of argument, suppose that associative learning can be excluded; but he then finds that the different rewards are not qualitatively different, which would imply that the apes are not taking a future state into account when selecting their tool. Vaesen argues that wanting to eat a grape and wanting to drink rosehip berry soup are the result of the same desire: namely, hunger.
First, hunger is not a desire; it is a response to a physiological state. In any case, the subjects in this study were not particularly hungry or thirsty: Their daily feeding routines had not been interrupted. Second, eating and drinking are dissimilar activities, with different physiological outcomes. For the time being, allow the most parsimonious interpretation to be that the apes select between two different desires.
Vaesen concludes from very few studies – two, it appears – that great ape foresight is highly limited compared with that of humans. However, the few great ape studies that exist simply have not tested for a range of foresight skills allegedly present in humans. Much work needs to be done before great ape foresight abilities can be delineated. We are plainly ignorant of significant facts about great ape foresight and also know little about the mechanisms of foresight in themselves. If a certain, uniquely human type of foresight is indeed essential for superior tool use, the evidence for such uniqueness, in the experimental literature that Vaesen cites, is insufficient.
Third, we have concerns with how Vaesen handles the abilities of great apes to represent tool function. We will here limit ourselves to the previously mentioned foresight study by Osvath and Osvath (Reference Osvath and Osvath2008). The fourth experimental condition in that study included novel items from which the apes were to select. Of the four items, only one was functional for retrieving the future reward. The subject had to decide, by visual inspection of each novel item, whether that item would be functional in the future. Stimulus generalisation – of length, colour, size, and shape – was precluded. The apes performed significantly above chance.
It is difficult to explain these results without ascribing to the apes some representation of function. This novel use would imply a type of causal understanding, according to Vaesen's line of argumentation. When it came to ascribing to a tool a particular function, it is obvious that, in the other three conditions in the study, the apes consistently chose the functional tool. The experimental condition for associative learning controlled more specifically for whether the tool was selected because of its function – rather than representing an arbitrary, reinforced stimulus. These results are not compatible with Vaesen's statement that primates do not attach particular functions to particular objects.
Fourth, Vaesen adopts Premack and Woodruff's (Reference Premack and Woodruff1978) characterization of theory of mind in the section on social intelligence. Since this definition was first proposed, the capacity for theory of mind has been shown to consist of several components (e.g., Call & Tomasello Reference Call and Tomasello2008; Gärdenfors Reference Gärdenfors2001; Reference Gärdenfors2003). One should at least distinguish understanding the emotions of others from understanding their attention, understanding their intentions, and understanding their beliefs. Only the last capacity is normally called a theory of mind. With the other capacities, it is fully possible for the agent to react to visible behaviours directly. Evidence for both the capacity to understand emotions and the capacity to understand attention can be found in non-human primates (Preston & de Waal Reference Preston and de Waal2002; see Call & Tomasello Reference Call and Tomasello2008). The situation is less clear for understanding intentions, although Tomasello et al. (Reference Tomasello, Carpenter, Call, Behne and Moll2005) claim the capacity to form joint intentions as the hallmark of humans. Only with respect to understanding the beliefs of others is there, so far, no evidence from non-human primates. A less anthropocentric research methodology might change the situation, as it did for research on understanding the attention of others.