Did reasoning evolve because it is best suited for communicating arguments? In the target article, Mercier & Sperber (M&S) claim this might be the case even though individuals poorly generate and recognize valid arguments. If, however, individuals are poor at generating and recognizing valid arguments, there are several – in our view, insurmountable – obstacles for this type of thinking evolving for any social function.
First, reasoned argumentation would fail to achieve the goal of persuasion – changes in beliefs and attitudes. One of the most common pitfalls of everyday reasoning is that nonrational factors (e.g., believability of an argument, confidence and status of the arguer, and self-interest of the receiver) trump logical validity in argument evaluation (for review of believability effects, see Evans et al. Reference Evans, Newstead and Byrne1993). The emergence of a trait for recognizing valid arguments, therefore, would offer no particular benefit in a context of like-headed reasoners: Valid arguments would only convert the converted. Examples of this come from the phenomenon of belief polarization (Lord et al. Reference Lord, Ross and Lepper1979), where open exchanges of arguments serve only to pull disputants away from recognizing valid alternatives.
Another example of valid reasoning failing to persuade comes from studies of dyadic learning in children. A common finding in this literature (e.g., Ames & Murray Reference Ames and Murray1982) is that when conservers and nonconservers share their reasons for conservation problems, the former persuade the latter. But is this a case of reason triumphant – or of less-confident students yielding to their more-confident peers? Suggesting this might be a victory for confidence over reason, Levin and Druyan (Reference Levin and Druyan1993) found that confidence was much higher among conservers than nonconservers. Further, when problems were switched to science problems with high rates of misconceptions, the competence/confidence correlation was flipped – and children found the confident incompetents more persuasive than their better-reasoning peers. (From this perspective, it's easy to see why dogmatic creationists want scientists to “teach the controversy.”)
Could it be (as argued by M&S) that reasoning evolved to help people detect untrustworthy sources by flagging inconsistency in their arguments? Developmental evidence suggests that this is unlikely because children detect trustworthiness long before they detect argument inconsistency. For example, when Morris and Hasson (Reference Morris and Hasson2010) presented children with simulated arguments of the simplest logical form (e.g., Puppet 1 claims “There is a sticker in the box!” and Puppet 2 claims “There is no sticker in the box!”), nearly 100% of 4- and 5-year-olds failed to detect an inconsistency between the two claims (Morris & Hasson Reference Morris and Hasson2010). At the same time, ability to detect a source as trustworthy emerges by 3 or 4 years of age (e.g., Jaswal & Neely Reference Jaswal and Neely2006; Koenig et al. Reference Koenig, Clement and Harris2004). Given this, it cannot be that detecting trustworthiness requires the ability to detect argument inconsistency.
Second, reasoned argumentation is expensive and may compete for limited cognitive resources with less expensive “hot” cognition. Recognizing and generating valid arguments (that are not already believable) requires substantial cognitive resources, including a heavy cost to working memory (Barrouillet et al. Reference Barrouillet, Grosset and Lecas2000). This cost increases dramatically with an increase in the number of premises and introduction of quantifiers. At the same time, if the goal is persuasion, “hot” cognition (e.g., appeals to emotionally laden examples) can offer a less expensive alternative (Petty & Cacioppo Reference Petty and Cacioppo1996). Given that the cost of generating and comprehending logical arguments outweighs those of “hot” cognition, without offering substantial benefits, a mutant with an elevated ability for logical argument would have no competitive advantage against her demagogic disputants. Thus, it is difficult to see how the argumentative context would provide the ideal environment for the evolution of logical argument.
Third, operator/receiver parity precludes benefits of social learning or knowledge transfer. Although it is often tempting to draw analogies between language and reasoning (e.g., Braine & O'Brien Reference Braine and O'Brien1998), the difference between the two is profound. Unlike reasoning, language proficiency is (more or less) universal in human adults. Consequently, in linguistic communication, more proficient language users (operators) can pull less proficient language learners (receivers) along the path to greater proficiency. This is not the case with reasoned argumentation, however. Here, operator and receiver characteristics are more symmetrical: When a person who is receptive to invalid arguments is put in the role of the operator, invalid arguments follow, and when a person who produces invalid arguments is put in the role of the receiver, valid and invalid arguments are not discriminated. Consequently, communicating reasons across individuals cannot, by itself, add anything to argumentation. Indeed, one of the most striking findings in cognitive development concerns how greatly change in language proficiency outpaces that of logical thinking, as well as how little reasoned argumentation is affected by observing a more proficient reasoner (as observed by Levin & Druyan Reference Levin and Druyan1993).
The failures of everyday reasoning that we think would make the argumentative context an inhospitable environment for the evolution of reasoning are seen in a strange light by M&S. According to them, these failures support their account because they arise mostly outside an argumentative context. Yet, even if we were to stipulate this as true, superior reasoning in an argumentative context does not support their claim about the evolution of reasoning: It would imply that arguments facilitate reasoning, not that reasoning facilitates arguments. Yet, if reasoning is designed for arguments, as M&S contend, quality of reasoning must facilitate or hinder quality of arguments, whereas the reverse is unnecessary. To take M&S's analogy, to show that the structure of the foot is designed for walking, one must show that a different structure would facilitate or impede walking – not that process of walking strengthens the foot of the walker.
In our view, the kind of argument that is optimally designed for social communication – that is, would have the largest effect on manipulating another's behavior – is not necessarily a reasoned argument but an argument that is believable, emotive, easy to generate on the fly, and clear to others. Put simply, reasoned argumentation is no more likely to have evolved for social communication than is the posture of the foot to have evolved for disco.
Did reasoning evolve because it is best suited for communicating arguments? In the target article, Mercier & Sperber (M&S) claim this might be the case even though individuals poorly generate and recognize valid arguments. If, however, individuals are poor at generating and recognizing valid arguments, there are several – in our view, insurmountable – obstacles for this type of thinking evolving for any social function.
First, reasoned argumentation would fail to achieve the goal of persuasion – changes in beliefs and attitudes. One of the most common pitfalls of everyday reasoning is that nonrational factors (e.g., believability of an argument, confidence and status of the arguer, and self-interest of the receiver) trump logical validity in argument evaluation (for review of believability effects, see Evans et al. Reference Evans, Newstead and Byrne1993). The emergence of a trait for recognizing valid arguments, therefore, would offer no particular benefit in a context of like-headed reasoners: Valid arguments would only convert the converted. Examples of this come from the phenomenon of belief polarization (Lord et al. Reference Lord, Ross and Lepper1979), where open exchanges of arguments serve only to pull disputants away from recognizing valid alternatives.
Another example of valid reasoning failing to persuade comes from studies of dyadic learning in children. A common finding in this literature (e.g., Ames & Murray Reference Ames and Murray1982) is that when conservers and nonconservers share their reasons for conservation problems, the former persuade the latter. But is this a case of reason triumphant – or of less-confident students yielding to their more-confident peers? Suggesting this might be a victory for confidence over reason, Levin and Druyan (Reference Levin and Druyan1993) found that confidence was much higher among conservers than nonconservers. Further, when problems were switched to science problems with high rates of misconceptions, the competence/confidence correlation was flipped – and children found the confident incompetents more persuasive than their better-reasoning peers. (From this perspective, it's easy to see why dogmatic creationists want scientists to “teach the controversy.”)
Could it be (as argued by M&S) that reasoning evolved to help people detect untrustworthy sources by flagging inconsistency in their arguments? Developmental evidence suggests that this is unlikely because children detect trustworthiness long before they detect argument inconsistency. For example, when Morris and Hasson (Reference Morris and Hasson2010) presented children with simulated arguments of the simplest logical form (e.g., Puppet 1 claims “There is a sticker in the box!” and Puppet 2 claims “There is no sticker in the box!”), nearly 100% of 4- and 5-year-olds failed to detect an inconsistency between the two claims (Morris & Hasson Reference Morris and Hasson2010). At the same time, ability to detect a source as trustworthy emerges by 3 or 4 years of age (e.g., Jaswal & Neely Reference Jaswal and Neely2006; Koenig et al. Reference Koenig, Clement and Harris2004). Given this, it cannot be that detecting trustworthiness requires the ability to detect argument inconsistency.
Second, reasoned argumentation is expensive and may compete for limited cognitive resources with less expensive “hot” cognition. Recognizing and generating valid arguments (that are not already believable) requires substantial cognitive resources, including a heavy cost to working memory (Barrouillet et al. Reference Barrouillet, Grosset and Lecas2000). This cost increases dramatically with an increase in the number of premises and introduction of quantifiers. At the same time, if the goal is persuasion, “hot” cognition (e.g., appeals to emotionally laden examples) can offer a less expensive alternative (Petty & Cacioppo Reference Petty and Cacioppo1996). Given that the cost of generating and comprehending logical arguments outweighs those of “hot” cognition, without offering substantial benefits, a mutant with an elevated ability for logical argument would have no competitive advantage against her demagogic disputants. Thus, it is difficult to see how the argumentative context would provide the ideal environment for the evolution of logical argument.
Third, operator/receiver parity precludes benefits of social learning or knowledge transfer. Although it is often tempting to draw analogies between language and reasoning (e.g., Braine & O'Brien Reference Braine and O'Brien1998), the difference between the two is profound. Unlike reasoning, language proficiency is (more or less) universal in human adults. Consequently, in linguistic communication, more proficient language users (operators) can pull less proficient language learners (receivers) along the path to greater proficiency. This is not the case with reasoned argumentation, however. Here, operator and receiver characteristics are more symmetrical: When a person who is receptive to invalid arguments is put in the role of the operator, invalid arguments follow, and when a person who produces invalid arguments is put in the role of the receiver, valid and invalid arguments are not discriminated. Consequently, communicating reasons across individuals cannot, by itself, add anything to argumentation. Indeed, one of the most striking findings in cognitive development concerns how greatly change in language proficiency outpaces that of logical thinking, as well as how little reasoned argumentation is affected by observing a more proficient reasoner (as observed by Levin & Druyan Reference Levin and Druyan1993).
The failures of everyday reasoning that we think would make the argumentative context an inhospitable environment for the evolution of reasoning are seen in a strange light by M&S. According to them, these failures support their account because they arise mostly outside an argumentative context. Yet, even if we were to stipulate this as true, superior reasoning in an argumentative context does not support their claim about the evolution of reasoning: It would imply that arguments facilitate reasoning, not that reasoning facilitates arguments. Yet, if reasoning is designed for arguments, as M&S contend, quality of reasoning must facilitate or hinder quality of arguments, whereas the reverse is unnecessary. To take M&S's analogy, to show that the structure of the foot is designed for walking, one must show that a different structure would facilitate or impede walking – not that process of walking strengthens the foot of the walker.
In our view, the kind of argument that is optimally designed for social communication – that is, would have the largest effect on manipulating another's behavior – is not necessarily a reasoned argument but an argument that is believable, emotive, easy to generate on the fly, and clear to others. Put simply, reasoned argumentation is no more likely to have evolved for social communication than is the posture of the foot to have evolved for disco.