Henrich et al.'s thought-provoking article discusses two major issues – how psychologists should pursue research, and how culture leads to cognitive variability. In what follows, we address these issues in turn, and argue that any cognitive theory ought to account for not only culture, but also other sources of cognitive variability.
First, Henrich et al. compare empirical data obtained from individual samples that fall under the umbrella description WEIRD (Western, Educated, Industrialised, Rich, and Democratic) with those obtained elsewhere, and draw the conclusion that “WEIRD subjects may often be the worst population from which to make generalizations” (sect. 7.1.2). For example, they report that people in many non-Western samples do not experience the Müller-Lyer illusion as strongly as do Americans, and query that “if visual perception can vary, what kind of psychological processes can we be sure will not vary?” (sect. 3.1, para. 5). However, while Henrich et al. caution against making sweeping generalizations from limited sets of data, they do not explain why they are permitted to make sweeping generalizations of relativism from their own data – the Müller-Lyer illusion is but one single phenomenon in visual perception, hardly representative of all visual perceptual processes. What would count as a representative sample of human psychology? The assertion that WEIRD participants are least representative of human psychology implies that there is a more representative sample, but Henrich et al. have failed to specify it. We believe that such a specification can only be arrived at empirically, and that it is impossible to specify a priori what a representative sample of human cognition should be.
Psychological research is theory-driven. Hence, in the absence of any evidence or theoretical rationale suggesting otherwise, WEIRD samples are a convenient proxy for conducting research, and they allow researchers to draw tentative conclusions about the matter of investigation. We acknowledge that certain specific psychological phenomena observed in WEIRD samples may not occur in other populations, and such discrepancies may help researchers make more accurate predictions in future experiments. Henrich et al.'s results underline the point that tentative conclusions are needed in order to support or contravene a theory.
Second, Henrich et al. have identified culture as a major source of cognitive variability, but we believe it is important to examine other sources of variability, as well. Cognitive processes are by nature non-deterministic: Children do not employ a deterministic strategy to perform cognitive tasks (e.g., Siegler Reference Siegler1996), and patients with dementia, head injury, ADHD, and schizophrenia are even less consistent in their thinking (for a review, see MacDonald et al. Reference MacDonald, Nyberg and Backman2006). Likewise, the same individual may perform a task differently at different times. In one of our studies (Bucciarelli & Johnson-Laird Reference Bucciarelli and Johnson-Laird1999), a group of participants had to draw deductions from syllogistic premises by using cut-out paper shapes representing syllogistic terms. Results showed that individuals' strategies differed from trial to trial in terms of which premises to interpret first, how to interpret the premises, and how to diligently search for counterexamples. As a result, it was impossible to predict individuals' cognitive operations based on their previous performance.
What other sorts of factors affect the way we think? Maturational and psychopathological factors are clear determinants, but the content on which a cognitive process operates may affect the process itself. Individuals think about different contents because they differ in their experiences, education, and beliefs. Culture may explain variability in these factors only to a certain extent, and hence psychologists ought to develop theories that explain how a cognitive process (a) can be modulated by content, and (b) develops and decays under normal and pathological conditions, respectively.
Consider the case of bicultural individuals. The behaviour of these individuals is guided by one internalized culture or the other at different moments (e.g., Ng & Lai Reference Ng and Lai2009; Pouliasi & Verkuyten Reference Pouliasi and Verkuyten2007), and they organize their cultural identities differently (Haritatos & Benet-Martínez Reference Haritatos and Benet-Martínez2002). When bicultural individuals' cultures contain inconsistent moral values, they will experience moral dilemmas such as the following described by a 19-year-old second-generation Indian American: “I enjoy my Indian culture, I feel that it is rich in tradition, morality, and beauty; confused because I have been in many situations where I feel being both cultures is not an option … I feel like you have to choose one or the other” (Haritatos & Benet-Martínez Reference Haritatos and Benet-Martínez2002). How might a theory explain this phenomenon? Mental models theory (e.g., Johnson-Laird Reference Johnson-Laird2006) allows for individuals to entertain inconsistent beliefs, because we tend to rely on separate sets of beliefs in separate contexts (Johnson-Laird et al. Reference Johnson-Laird, Legrenzi, Girotto and Legrenzi2000). This is evident in moral reasoning, in which moral intuitions and conscious moral reasoning are based on beliefs that are neither complete nor consistent (Bucciarelli et al. Reference Bucciarelli, Khemlani and Johnson-Laird2008). Our conception of culture therefore differs from that in cross-cultural psychology, which considers culture as a network of discrete, specific constructs that guide cognition only when they come to the foreground in an individual's mind (Hong et al. Reference Hong, Morris, Chiu and Benet-Martínez2000).
How do cross-cultural differences in thinking emerge in a society? Henrich et al. explain the development of these differences by appealing to content (data perceived, norms, and connotations) and context (individuals' contemporary environment, the environment during development, and the immediate experimental environment). We emphasize that an analysis of reasoning strategies can explain variability within the same individual. Therefore, if content, context, and strategy drive cultural differences, then those factors are of primary interest, whereas cultural differences are merely incidental. Cognitive theories should distinguish between the universal processes they propose and the specific contents on which they operate. For instance, our own theory of moral reasoning (Bucciarelli et al. Reference Bucciarelli, Khemlani and Johnson-Laird2008) posits that moral reasoning is simply normal deontic reasoning (Bucciarelli & Johnson-Laird Reference Bucciarelli and Johnson-Laird2005) applied to moral contents and contexts. Moral contents and contexts may differ across cultures, but the theory of deontic reasoning we propose is, and ought to be, domain-general. Such a dissociation between general computational operations and the contents they operate on allows researchers to construct theories that are sensitive not just to cultural differences, but to age-related, social, personality, and strategic differences, as well.
You have
Access
Henrich et al.'s thought-provoking article discusses two major issues – how psychologists should pursue research, and how culture leads to cognitive variability. In what follows, we address these issues in turn, and argue that any cognitive theory ought to account for not only culture, but also other sources of cognitive variability.
First, Henrich et al. compare empirical data obtained from individual samples that fall under the umbrella description WEIRD (Western, Educated, Industrialised, Rich, and Democratic) with those obtained elsewhere, and draw the conclusion that “WEIRD subjects may often be the worst population from which to make generalizations” (sect. 7.1.2). For example, they report that people in many non-Western samples do not experience the Müller-Lyer illusion as strongly as do Americans, and query that “if visual perception can vary, what kind of psychological processes can we be sure will not vary?” (sect. 3.1, para. 5). However, while Henrich et al. caution against making sweeping generalizations from limited sets of data, they do not explain why they are permitted to make sweeping generalizations of relativism from their own data – the Müller-Lyer illusion is but one single phenomenon in visual perception, hardly representative of all visual perceptual processes. What would count as a representative sample of human psychology? The assertion that WEIRD participants are least representative of human psychology implies that there is a more representative sample, but Henrich et al. have failed to specify it. We believe that such a specification can only be arrived at empirically, and that it is impossible to specify a priori what a representative sample of human cognition should be.
Psychological research is theory-driven. Hence, in the absence of any evidence or theoretical rationale suggesting otherwise, WEIRD samples are a convenient proxy for conducting research, and they allow researchers to draw tentative conclusions about the matter of investigation. We acknowledge that certain specific psychological phenomena observed in WEIRD samples may not occur in other populations, and such discrepancies may help researchers make more accurate predictions in future experiments. Henrich et al.'s results underline the point that tentative conclusions are needed in order to support or contravene a theory.
Second, Henrich et al. have identified culture as a major source of cognitive variability, but we believe it is important to examine other sources of variability, as well. Cognitive processes are by nature non-deterministic: Children do not employ a deterministic strategy to perform cognitive tasks (e.g., Siegler Reference Siegler1996), and patients with dementia, head injury, ADHD, and schizophrenia are even less consistent in their thinking (for a review, see MacDonald et al. Reference MacDonald, Nyberg and Backman2006). Likewise, the same individual may perform a task differently at different times. In one of our studies (Bucciarelli & Johnson-Laird Reference Bucciarelli and Johnson-Laird1999), a group of participants had to draw deductions from syllogistic premises by using cut-out paper shapes representing syllogistic terms. Results showed that individuals' strategies differed from trial to trial in terms of which premises to interpret first, how to interpret the premises, and how to diligently search for counterexamples. As a result, it was impossible to predict individuals' cognitive operations based on their previous performance.
What other sorts of factors affect the way we think? Maturational and psychopathological factors are clear determinants, but the content on which a cognitive process operates may affect the process itself. Individuals think about different contents because they differ in their experiences, education, and beliefs. Culture may explain variability in these factors only to a certain extent, and hence psychologists ought to develop theories that explain how a cognitive process (a) can be modulated by content, and (b) develops and decays under normal and pathological conditions, respectively.
Consider the case of bicultural individuals. The behaviour of these individuals is guided by one internalized culture or the other at different moments (e.g., Ng & Lai Reference Ng and Lai2009; Pouliasi & Verkuyten Reference Pouliasi and Verkuyten2007), and they organize their cultural identities differently (Haritatos & Benet-Martínez Reference Haritatos and Benet-Martínez2002). When bicultural individuals' cultures contain inconsistent moral values, they will experience moral dilemmas such as the following described by a 19-year-old second-generation Indian American: “I enjoy my Indian culture, I feel that it is rich in tradition, morality, and beauty; confused because I have been in many situations where I feel being both cultures is not an option … I feel like you have to choose one or the other” (Haritatos & Benet-Martínez Reference Haritatos and Benet-Martínez2002). How might a theory explain this phenomenon? Mental models theory (e.g., Johnson-Laird Reference Johnson-Laird2006) allows for individuals to entertain inconsistent beliefs, because we tend to rely on separate sets of beliefs in separate contexts (Johnson-Laird et al. Reference Johnson-Laird, Legrenzi, Girotto and Legrenzi2000). This is evident in moral reasoning, in which moral intuitions and conscious moral reasoning are based on beliefs that are neither complete nor consistent (Bucciarelli et al. Reference Bucciarelli, Khemlani and Johnson-Laird2008). Our conception of culture therefore differs from that in cross-cultural psychology, which considers culture as a network of discrete, specific constructs that guide cognition only when they come to the foreground in an individual's mind (Hong et al. Reference Hong, Morris, Chiu and Benet-Martínez2000).
How do cross-cultural differences in thinking emerge in a society? Henrich et al. explain the development of these differences by appealing to content (data perceived, norms, and connotations) and context (individuals' contemporary environment, the environment during development, and the immediate experimental environment). We emphasize that an analysis of reasoning strategies can explain variability within the same individual. Therefore, if content, context, and strategy drive cultural differences, then those factors are of primary interest, whereas cultural differences are merely incidental. Cognitive theories should distinguish between the universal processes they propose and the specific contents on which they operate. For instance, our own theory of moral reasoning (Bucciarelli et al. Reference Bucciarelli, Khemlani and Johnson-Laird2008) posits that moral reasoning is simply normal deontic reasoning (Bucciarelli & Johnson-Laird Reference Bucciarelli and Johnson-Laird2005) applied to moral contents and contexts. Moral contents and contexts may differ across cultures, but the theory of deontic reasoning we propose is, and ought to be, domain-general. Such a dissociation between general computational operations and the contents they operate on allows researchers to construct theories that are sensitive not just to cultural differences, but to age-related, social, personality, and strategic differences, as well.