7 Heuristics and biases: insights from developmental studies – The Development of Thinking and Reasoning

7
Heuristics and biases Insights from developmental studies

Kinga Morsanyi and Simon J. Handley

When people are looking for a solution to a problem or they have to decide between choice options, they often rely on simplifying heuristics rather than on formal logic or rule-based argumentation. Moreover, the use of heuristics is just as common in real-life settings as in psychology laboratories, and it pervades all aspects of life. For example, when reading a piece of text, people judge the writer to be more intelligent if the text is easier to process (Oppenheimer, 2006), rhyming versions of aphorisms seem to be truer than non-rhyming aphorisms with the same meaning (McGlone & Tofighbakhsh, 2000), and arguments are judged to be stronger if they correspond to the reader’s beliefs (Stanovich & West, 2008). Heuristics do not only influence our judgements, but also affect the ways in which we try to tackle problems. For example, children in a classroom might choose the operations to solve a mathematical word problem based on the title of the chapter that the problem appears in (Van Dooren, De Bock, Evers, & Verschaffel, 2009).

According to dual-process theorists (e.g., Evans, 2006), the fact that people often rely on heuristics instead of cognitively demanding reasoning demonstrates the operation of two kinds of cognitive processes: type 1 (fast, automatic and effortless) and type 2 (slow, conscious and effortful). Indeed, it seems that there are two fundamentally different ways of solving problems: through following our intuitions, or through engaging in a thorough and careful reasoning process. Whereas type 1 processes are assumed to utilize pragmatic and contextual cues, type 2 reasoning is supposed to rely on participants’ knowledge of relevant normative rules (cf., Stanovich & West, 2008). Not surprisingly, the two approaches can sometimes lead to very different outcomes. However, instead of envisioning “two minds in one brain” (cf., Evans, 2003), it is also possible to conceive the use of mental shortcuts and the reliance on conscious, rule-based reasoning as simply the two ends of a continuum, where cognitive processes are performed through the same mechanisms, although the amount of cognitive resources (e.g., attention, working memory) allocated to these processes might vary considerably (cf., Keren & Schul, 2009; Osman & Stavy, 2006). Finally, there is a third possibility; namely, that all novel reasoning and decision-making tasks require conscious, effortful processing. However, as the tasks and situations become more and more familiar, people develop mental habits to tackle them, and this process is qualitatively different from the processes used to approach novel problems. Indeed, once people have well-practised shortcuts to solve a problem, they might even need conscious effort to override well-learnt processes and responses, and to be able to approach familiar problems from a new perspective.

In this chapter, we argue that people acquire and develop a wider and wider range of heuristics as they get older (see also e.g., Klaczynski, 2001a, 2001b; Reyna & Farley, 2006). Although some of the most commonly used heuristics (such as the ones regarding social stereotypes about gender or age) might develop early, others (which are more specific and, thus, used less frequently) might develop later in life, or they might only be used by people who get specific training in a certain domain. The purpose of this chapter is to investigate examples for the development of different kinds of heuristics, and to identify the driving forces behind the development of heuristic reasoning. We will do this by providing a short review of the relevant literature. Then we will present some of our own results.

Developmental trends in heuristic reasoning

Developmental research on reasoning has been relatively sparse. One possible reason for this might be the implicit assumption that, given that adults perform poorly on reasoning and decision-making tasks, children’s performance must be even worse (cf., Klaczynski, 2009). However, although there is a general consensus that the capacity for rule-based, normative reasoning develops with age, the predictions regarding age-related changes in heuristic reasoning are not very clear. The general assumption is that heuristic reasoning is simple, and it appears early in the course of development (see, e.g., Barrouillet, 2011). Thus, many theorists do not expect the heuristic system to change considerably with age (but see, e.g., Fischbein & Schnarch, 1997; Reyna & Farley, 2006; Klaczynski, 2009, for exceptions). Nevertheless, some developmental theorists (Stanovich, West, & Toplak, 2011) acknowledge that rules and decision-making principles might be practised to automaticity. Thus, whereas heuristic reasoning is often contrasted with analytic or normative reasoning, heuristic and conscious, rule-based reasoning might produce similar output, and this tendency increases with development.

Most studies that have investigated the development of heuristic-based reasoning have used tasks where there is a conflict between a response which is based on a heuristic (i.e., which “feels right” – see, e.g., Thompson, 2009) and another one which is based on the conscious application of normative rules (for typical examples see, e.g., Gilovich, Griffin, & Kahneman, 2002; Kahneman, Slovic, & Tversky, 1982). For example, consider the following problem (based on Frederick, 2005):

In a lake, there is a patch of lily pads. Every day, the patch doubles in size. If it takes 48 days for the patch to cover the entire lake, how long would it take for the patch to cover half of the lake? _____ days

Although the correct response to this question is 47 days, the response “24 days” springs to mind very easily, and people have a strong tendency to give this response. This is also true in the case of highly educated adults, although

Table 7.1 Examples of the tasks used by Morsanyi and Handley (2008)

Belief bias (syllogistic reasoning) Trees have red trunks.
Oaks do not have red trunks.
Does it follow that oaks are not trees?
yes   no   not certain
Sunk cost fallacy Consider the following two situations and answer the questions that follow each.
A. You want to try out the new skates that you’ve just bought, so you go to the ice rink. You stand in the queue for 40 minutes to get inside. When you finally start to skate you find the rink so crowded with people that it’s impossible to move around and you don’t enjoy the whole thing at all.
What do you do?
(a) go home immediately
(b) skate for a little while, then go home
(c) stay for two hours as you’ve planned
B. You want to try out the new skates that you’ve just bought, so you go to the ice rink. When you start to skate you find the rink so crowded with people that it’s impossible to move around and you don’t enjoy the whole thing at all. What do you do?
(a) go home immediately
(b) skate for a little while, then go home
(c) stay for two hours as you’ve planned
Conjunction fallacy Tim is ten. He lives in a house with a garden. He has many friends, he likes to play sports in the park, and he collects football cards. Your task is to mark the statement which is the most likely to be true with number 1, the next one with number 2, and so on. Mark the statement which is the least likely to be true with number 4.
(a) Tim has a rabbit.
(b) Tim has a sister.
(c) Tim has a rabbit and he often plays football.
(d) Tim goes to chess competitions.
If-only fallacy Jessica went to a concert with her parents. She asked her father not to take the usual way to the concert hall, but to drive through the city centre instead to pick up her friend who lived in the city centre and who they promised to take with them to the concert. Unfortunately, they got stuck in a traffic jam and they missed the first 30 minutes of the concert. Now consider the next story. Linda went to a concert with her parents. She asked her father not to take the usual way to the concert hall, but to drive through the city centre instead because she wanted to see the Christmas decorations and lights. Unfortunately, they got stuck in a traffic jam and they missed the first 30 minutes of the concert.
What do you think about the two stories?
(a) Jessica made a worse decision than Linda.
(b) Linda made a worse decision than Jessica.
(c) It wasn’t their fault, it was just bad luck.

For the belief bias task the heuristic response (based on beliefs) is “no”, whereas the normatively correct response is “yes”. In the sunk cost task giving the same response in both scenarios is considered to be normatively correct, whereas staying for longer after standing in the queue for 40 minutes is considered a heuristic response. For the conjunction fallacy task participants commit the fallacy if they consider response (c) to be more probable than response (a). In the case of the if-only task response (c) is considered normatively correct, whereas response (b) is considered to be a heuristic response.

individuals with higher IQ are more likely to respond correctly (Frederick, 2005). Given that the knowledge of normative rules, as well as the ability to retrieve and apply these rules is assumed to increase with development, in the case of tasks where there is a conflict between intuitions and normative reasoning, the obvious expectation would be that normative responding should increase with age and intuitive responding decrease. Thus, on the above task, one would expect adults to perform better than children.

Nevertheless, the evidence regarding age-related changes in heuristic and normative responding is mixed. Some studies have found that normative responding increases with age on certain tasks, but this increase is not apparent on other tasks (e.g., Fischbein & Schnarch, 1997; Klaczynski, 2001a; Kokis, Macpherson, Toplak, West, & Stanovich, 2002). Moreover, development is sometimes associated with increases in non-normative responses and violations of formal rules of inference. In the following sections, we will focus mainly on findings which show an age-related increase in heuristic reasoning, with the purpose of gaining a better insight into what might be the basis of these counterintuitive findings.

Davidson (1995) reported that susceptibility to the conjunction fallacy (see Table 7.1 for an example) increased during the primary school years. In the conjunction fallacy task, a response which is cued by a social stereotype is in contrast with a very simple rule of probability (i.e., that the conjunction of two events – A and B occurring together – cannot be more probable than either A or B occurring alone). Thus, in our example in Table 7.1, although the description of Tim cues the answer that “he often plays football,” based on logical consideration it cannot be more likely that he plays football and he has a rabbit than that he has a rabbit. Another study with primary school children (Jacobs & Potenza, 1991) found that children increasingly relied on their own experience as opposed to probabilistic information in making decisions about social situations, although in non-social situations the trend was the opposite (they increasingly favoured probabilistic information with age). Take this example:

In Juanita’s class 10 girls are trying out to be cheerleaders, and 20 are trying out for the band. Juanita is very popular and very pretty. She is always telling jokes and loves to be around people. Do you think Juanita is trying out to be a cheerleader or for the band?

Whereas younger children tended to give the response which corresponded to the base rates (i.e., Juanita is more likely to be trying out for the band), older children tended to assume that she was trying out to be a cheerleader, which was more in line with the description. One way to explain the increasing effect of stereotypical information is to assume that younger children are simply unfamiliar with the stereotypes which are evoked by the descriptions, and, thus, obviously, they will be less likely to base their decisions on stereotypical information (cf., Kokis et al., 2002; Stanovich et al., 2011). However, in a later study with very similar problems, De Neys and Vanderputte (2011) manipulated stereotype familiarity (relying on the results of a pre-test, where preschoolers were asked questions such as “What are daddies/mommies doing at home?”; “What do thin/fat children like to eat most?”). De Neys and Vanderputte (2011) found that the tendency to rely on social stereotypes increased not only in the case of stereotypes that younger children were less familiar with (e.g., “Dutch people like to eat cheese” or “workmen drink beer”), but also in the case of tasks which utilized stereotypes that were equally familiar to both preschoolers and older children (e.g., “grannies like to knit” or “fat kids eat candy”). Furthermore, these authors also reported an increase in the tendency for children to justify their responses with reference to social stereotypes. That is, these results suggest that older children show an increasing tendency to activate and apply relevant knowledge to solve problems.

In another study, Chiesi, Gronchi, and Primi (2008) investigated the conjunction fallacy using non-social materials (i.e., the question was about whether a coin, which had been lost, was more likely to be found under a blue flower, or a blue flower with a bee on it), and frequency information was provided in a pictorial format. The overall number of blue flowers was always higher than the number of blue flowers with a bee on it, but the number of blue flowers with (A&B) and without a bee (A&notB) was manipulated, resulting in three conditions: A&B < A&notB; A&B = A&notB; and A&B > A&notB. The participants were from three age groups – seven-year-olds, ten-year-olds, and young adults – and for each scenario they had to choose from three response options: the coin is most likely to be under a blue flower, the coin is most likely to be under a blue flower with a bee, or both of the above are equally likely. One important finding is that whereas the youngest children were not influenced by the relative frequencies of A&B and A&notB (i.e., the number of normative responses was the same in each condition), older participants were strongly affected by frequency information, and this effect was stronger in the case of adults. Specifically, whereas for A&B < A&notB, the older participants predominantly gave normative responses, for the other two cases this trend reversed, and the predominant response was a non-normative one, which corresponded to a pragmatic interpretation (i.e., participants clearly interpreted the question as referring to a comparison between A&B and A&notB, as opposed to A&B and A). That is, the sensitivity to pragmatic cues (see also Feeney, Scrafton, Duckworth, & Handley, 2004; Noveck, 2001), as well as to frequency information (see also De Neys & Vanderputte, 2011; Jacobs & Potenza, 1991) increased with development.

The application of stereotype-knowledge, and pragmatic cues are not the only examples of the increasing tendency for children to base their decisions on contextual cues. Webley and Plaisier (1998) using a problem similar to Tversky and Kahneman’s (1981) “lost ticket scenario” (see Table 7.2) found that older children (between the ages of 8 and 12) were increasingly influenced by past investments whereas young children (age five to six) were not. According to traditional economic analysis, past investments should not influence present economic decisions. In Tversky and Kahneman’s (1981) lost money/ticket task, there are two scenarios which are equivalent in terms of the expected utility and required financial investment. However, in one scenario (the lost ticket scenario), people associate a past investment with the ticket. By contrast, in the lost money scenario, the past investment is not associated with the ticket. This gives people

Table 7.2 Lost ticket and lost money scenarios (Tversky & Kahneman, 1981)

Lost money scenario Lost ticket scenario

Imagine that you have decided to see a play where admission is $10 per ticket. As you enter the theatre you discover that you have lost a $10 note.
Would you still pay $10 for a ticket for the play?
Yes              No
Imagine that you have decided to see a play and paid the admission price of $10 per ticket. As you enter the theatre you discover that you have lost the ticket. The seat was not marked and the ticket cannot be recovered.
Would you pay $10 for another ticket?
Yes              No

the impression that the actual price of the ticket is $10 in the lost money scenario, whereas it is $20 in the lost ticket scenario. As a result, 88% of participants in the original study were willing to pay for the ticket in the former, as opposed to only 46% in the latter case. The effect is based on making links between events (i.e., considering the value of the ticket in the context of another event), which seems to be absent in the case of young children (in Webley & Plaisier’s 1998 study 80% of the children indicated that they would buy the ticket in the lost ticket scenario, and 70% indicated that they would buy the ticket in the lost money scenario).

In addition to the previous examples, Reyna and Ellis (1994) reported that the framing effect (i.e., people are more risky in the domain of losses than in the domain of gains) does not emerge until roughly ten years of age. Age increases in the tendency to ignore denominators on ratio problems (see below – e.g., Brainerd, 1981), and to draw non-logical “transitive” inferences (e.g., “A is a friend of B. B is a friend of C. Therefore, A and C are friends”; Markovits & Dumas, 1999) have also been reported.

In adolescence and adulthood there is less evidence for age-related increases in heuristic reasoning. A notable exception is the domain of education in mathematics, probability, and science. Here we give a couple of examples of heuristics that increase with education, such as the “same A same B” rule (e.g., Stavy & Tirosh, 2000), “the overuse of proportionality” (Van Dooren et al., 2009), and the “equiprobability bias” (Lecoutre, 1985).

In the cylinders task (Tirosh & Stavy, 1999; see Figure 7.1), students are presented with two identical sheets of paper, one of which is rotated by 90 degrees. First they are asked whether the area of the two sheets is equal. Young children (below the age of eight) focus on the difference in the length/width of the two sheets, and they tend to claim that the area of the two sheets is different (see also Piaget, Inhelder, & Szeminska, 1960). Thus, on this part of the task, younger children perform worse than older ones. In the second part of the task, the sheets are folded to form two cylinders, and students are asked if the volume of the two cylinders is equal, or if it is different. Tirosh and Stavy (1999) found that whereas young children who focus on the perceptual properties of the cylinders tend to (correctly) assume that the volume of the smaller but wider cylinder is larger, older children typically argue that as the areas of the sheets are the same, the volume of the cylinders should

Figure 7.1 The cylinders task.

also be equal. There is a sharp increase in the equivalence response with age during the primary school years, and equivalence responses on the area and volume questions are strongly associated. Thus, although the students get better on one problem, at the same time they get worse on another one.

Now consider the following problem (based on Cramer, Post, & Currier, 1993): “Sue and Julie are running equally fast around a track. Sue started first. When she had run 9 laps, Julie had run 3 laps. When Julie completed 15 laps, how many had Sue run?” Van Dooren, De Bock, Hessels, Janssens, and Verschaffel (2005) reported that whereas between the third and sixth grades of primary school the ability for correct proportional reasoning on mathematical word problems steadily developed, at the same period the erroneous use of proportionality also increased. Indeed, whereas 30% of third graders gave the erroneous proportional answer (i.e., “5 laps”) to the above problem, in sixth grade 51% of the students committed the proportionality error.

Finally, consider the following problem (Konold, 1989):

Five faces of a fair die are painted black, and one face is painted white. The die is rolled six times. Which of the following results is most likely?

  • (a)To get five blacks and one white.
  • (b)To get six blacks.
  • (c)Same chance of getting six blacks or five blacks and one white.

Although the correct response is (a), many students assume that both outcomes should be equally likely, because random outcomes are “equiprobable by nature” (Lecoutre, 1985). What is more striking is that this assumption and the prevalence of erroneous responses based on this concept of equiprobability tend to increase with age and education (e.g., Batanero, Serrano, & Garfield, 1996; Chiesi & Primi, 2009; De Neys, 2007; Lecoutre, 1985; Morsanyi, Primi, Chiesi, & Handley, 2009).

Overall these findings seem puzzling from a dual-process perspective where development and the acquisition of knowledge (and especially formal education in a given area) is expected to reduce the reliance on heuristics (see e.g., Stanovich, Toplak, & West, 2008). Indeed, given that there is a clear association between using formal rules correctly and incorrectly (e.g., to apply correctly the proportionality rule and to overgeneralize it to situations where it does not apply – e.g., Tirosh & Stavy, 1999; Van Dooren et al., 2005), it would be hard to argue that normative and heuristic reasoning are necessarily underpinned by different representations and processes. Nevertheless, it seems that individuals with higher cognitive capacity and with a stronger spontaneous tendency for effortful thinking are less susceptible to the erroneous application of formal rules (Morsanyi et al., 2009). By contrast, the tendency for incorrect overgeneralization increases in individuals who are under cognitive load or time pressure (Gillard, Van Dooren, Schaeken, & Verschaffel, 2009). Thus, whereas relevant knowledge can both improve and impair reasoning performance, higher ability participants are better able to use their knowledge selectively. That is, to apply it only when it is useful, and to suppress seemingly relevant knowledge when a task requires a different approach.

Given these considerations, the findings of studies that have reported simultaneous improvements and decreases in performance on different tasks within a particular domain (e.g., Fischbein & Schnarch, 1997) are less puzzling. One important factor is that, in general, the impact of intuitions increases with age and with education. When a problem is easy to conceptualize, and the relevant rule is readily available, misconceptions will diminish with age. However, if the task is hard to conceptualize, the effects of misconceptions or irrelevant strategies activated by superficial cues will increase (cf., Fischbein & Schnarch, 1997).

In summary, there is much evidence that as children get older, they increasingly rely on contextual and pragmatic cues, as well as their relevant knowledge, when they make decisions or solve problems. This can lead to both increases and decreases in responses which are considered to be normatively correct. In fact, it is often the case that when children get better at solving a certain type of problem, at the same time they get worse on other tasks in the same domain, due to a tendency to overgeneralize strategies to situations where they cannot be applied. Besides an age-related increase in knowledge about the world and about rules which can be applied in different domains, and an increase in the tendency and ability to activate this knowledge, there are a number of other factors which affect an individual’s reasoning performance. In the next section we give a brief overview of these factors.

The factors that affect the prevalence of heuristic reasoning: knowledge, cognitive capacity, instructions, and thinking dispositions

In a recent paper, Stanovich et al. (2011) emphasized the complexity of making predictions regarding age-related changes in heuristic reasoning. Indeed, there are many factors that can affect the quality and outcome of people’s reasoning. For example, cognitive capacity is often positively related with normative responding (e.g., Chiesi, Primi, & Morsanyi, 2011; Handley, Capon, Beveridge, Dennis, & Evans, 2004; Kokis et al., 2002; and see Stanovich & West, 2000 for an extensive review), whereas cognitive load can lead to performance decrements (e.g., De Neys, 2006; Gillard et al., 2009). Additionally, dual-process theorists often emphasize the need for type 2 reasoning to inhibit and override erroneous automatic (type 1) response tendencies (e.g., DeNeys & Van Gelder, 2008; Stanovich et al., 2008). Interestingly, there is hardly any direct evidence for a relationship between measures of inhibition skills and normative reasoning on tasks with a conflict between intuitive and normative response options (for an exception, see Handley et al., 2004).

Another manipulation which might increase heuristic use is instructing participants to respond to a reasoning problem within a short period of time, which is determined by the experimenter (e.g., Evans & Curtis-Holmes, 2005; Roberts & Newton, 2001). By contrast, education and training usually improves reasoning performance (e.g., Fong, Krantz, & Nisbett, 1986; Lehman, Lempert, & Nisbett, 1988), although, as we have discussed above, education can also increase the misapplication of rules, which might lead to a simultaneous increase and decrease in correct responding across conceptually related tasks (e.g., Morsanyi, Handley, & Serpell, 2013; Tirosh & Stavy, 1999; Van Dooren et al., 2009).

Now consider the following problem, which was used by Pacini and Epstein (1999):

You are playing a lottery in which you can win $1000. There are two jars from which you can select a winning ticket. In the first jar, there are only ten tickets, and one of these is the winning ticket. In the second jar, there are 100 tickets and ten winning tickets.

Which jar, if either, would you select from to have a better chance of winning the lottery?

  • (a) The jar with one winning ticket.
  • (b) The jar with ten winning tickets.
  • (c) It would not matter to me.

Although the proportions of targets in options (a) and (b) are identical, both children and adults often fail to consider the differences in denominators and favour option (b) with the greatest absolute number of winning tickets. Interestingly, most participants choosing the heuristic response are aware that this response does not make sense statistically. So why do people make suboptimal choices against their better judgement? Pacini and Epstein (1999) found that the tendency to neglect denominators was negatively related to a self-report measure of rational thinking styles. Indeed, there is much evidence that self-report measures of participants’ spontaneous tendency to rely on effortful and flexible reasoning are powerful predictors of actual reasoning performance. Besides the rational-experiential inventory, need for cognition (Cacioppo & Petty, 1982) and actively open-minded thinking (Kokis et al., 2002) were also found to be positively related to reasoning performance (e.g., Kokis et al., 2002; Morsanyi et al., 2009). Furthermore, the variance explained by thinking dispositions is (at least partly) independent of the effect of cognitive capacity. Another dispositional measure, superstitious thinking (e.g., Kokis et al., 2002; Stanovich & West, 1997), which is the tendency to reason on the basis of superstitious beliefs about chance and luck, was found to be negatively correlated with reasoning performance (e.g., Chiesi et al., 2011; Toplak, Lui, Macpherson, Toneatto, & Stanovich, 2007).

Whereas thinking dispositions indicate participants’ spontaneous tendency for effortful, rule-based, or rational reasoning, it is also possible to increase participants’ tendency for normative responding by instructing them to reason logically (e.g., Denes-Raj & Epstein, 1994; Ferreira, Garcia-Marques, Sherman, & Sherman, 2006; Klaczynski, 2001a, 2001b). Given the effect of thinking styles on reasoning, and the fact that people might be aware that the option that “feels right” might not be the best solution to a particular problem (cf., Sloman 1996), it is not surprising that rational instructions can improve reasoning. What is more interesting is that instructions have a stronger effect in the case of participants with higher cognitive ability (Evans, Handley, Neilens, & Over, 2009; Morsanyi et al., 2009), and the ability to follow instructions also increases with development (Chiesi et al., 2011). This might reflect an increased aptitude for higher ability participants to consciously adopt a particular mindset. Another, related possibility is that cognitive development and higher cognitive capacity is associated with an increased tendency to recognize that problems can be approached from different points of view.

In line with this claim, Klaczynski (2009) suggested that the most important aspect of cognitive development during adolescence is that thinking becomes more flexible. Thus, instead of a development towards increasingly normative reasoning, adolescents do not adhere to rules as rigidly as children, and, as a result, their reasoning becomes more sensitive to subtle changes in task context. At the same time, there is evidence that adolescents, especially those with higher cognitive capacity, increasingly develop the ability to reason consistently across tasks with a similar logical structure but with different context (Handley, Newstead, & Neilens, 2011; Moshman, 1999). However, initially this process might be rigid, and adolescents might find it hard to discriminate between situations where contextual cues provide useful information, and where context is irrelevant (Simoneau & Markovits, 2003).

In summary, in the previous sections we presented evidence that during childhood and adolescence there is an increasing influence of contextual and pragmatic cues on reasoning performance. At the same time, knowledge about normative rules, and the ability to apply these rules, also increases. When there is a conflict between normative rules and contextual information, a number of factors can affect the eventual outcome of the reasoning process. These include participants’ cognitive capacity, their spontaneous tendency to rely on effortful reasoning, and the way the tasks are presented (i.e., whether there is time pressure, whether participants are instructed to reason logically). In the following section, we present a study which was aimed at assessing age-related changes in heuristic and normative responding on a number of tasks where the two types of response are in conflict. The purpose of this study was to identify some typical trends in reasoning development, together with the cognitive correlates of these changes.

Another aim of the following section is to compare our findings with recent dual-process conceptualizations of cognitive development. Barrouillet (2011) highlighted that dual-process theorists tend to focus almost exclusively on adult cognition, and he emphasized the need for dual-process theorists to spell out the implications of their ideas for cognitive development. In response to this, Stanovich et al. (2011) proposed that development proceeds from random responding through “smart errors,” followed by normative responding, whereas Evans (2011) presupposed the following developmental sequence: (1) non-contextualized reasoning; (2) contextualized reasoning without inhibition; (3) contextualized reasoning with inhibition. In the following section we will contrast our findings with these predictions.

The study

Although there are some theorists who have investigated dual-process predictions in developmental contexts (e.g., Brainerd & Reyna, 2001; Klaczynski, 2000), these studies usually suffer from several weaknesses. One issue is that the bulk of research into the development of reasoning processes has been conducted with children above the age of ten. If we assume that heuristic reasoning is a basic process which develops at a young age, we can expect that by early adolescence heuristics have already appeared, making it hard to identify the cognitive processes that played a role in their development. Furthermore, although a number of studies have investigated the development of reasoning heuristics, these studies typically focused on a small number of problems or just one particular type of task (e.g., De Neys & Vanderputte, 2011; Klaczynski & Cottrell, 2004), a restricted age range (e.g., Handley et al., 2004; Kokis et al., 2002), or, on the contrary, they compared participants from markedly different age groups, which makes it hard to draw inferences about the intervening developmental patterns (e.g., De Neys & Van Gelder, 2008). Finally, unlike the majority of the studies with adults, developmental dual-process studies generally failed to explore the individual differences correlates of reasoning performance (apart from some rare exceptions – e.g., Chiesi et al., 2011; Handley et al., 2004; Kokis et al., 2002).

In order to overcome these typical weaknesses of developmental research on reasoning, the present study was conducted with participants between the ages of 5 and 25 with no age gaps, and with a substantial number of participants. Besides measuring reasoning performance, we also assessed our participants’ general intelligence, working memory capacity, and their inhibition skills. The problems that we used were designed to elicit some classic biases of judgement and reasoning: the belief bias in syllogistic reasoning (e.g., Evans, Barston, & Pollard, 1983), the sunk cost fallacy (e.g., Arkes & Ayton, 1999), the if-only fallacy (Epstein, Lipson, Holstein, & Huh, 1992), and the conjunction fallacy (Tversky & Kahneman, 1983) – see Table 7.1. for illustrations. Given that our if-only fallacy tasks have been criticized on the basis that it is not clear whether the decontextulized response is more normative than the context-based one (see Stanovich et al., 2011), we will just focus here on the other three tasks. Although these problems are very different in terms of their structure and the domains that they tap into, they share one important characteristic: each of these tasks evokes an intuitively compelling response which is based on people’s general knowledge and beliefs, or the contextual cues presented in the tasks. At the same time, all of these problems can also be answered by a non-intuitive response which is based on normative considerations.1

In the syllogisms that we used, the problems’ logical structure is in conflict with beliefs and general knowledge. In the case of the sunk cost fallacy, the “waste not” heuristic (e.g., Arkes & Ayton, 1999; Klaczynski & Cottrell, 2004) is in contrast with the more rational consideration that investing further effort and resources into something that turned out to be worthless and undesirable is more wasteful than abandoning the activity altogether. Finally, as we described above, in the conjunction fallacy task, a response which is cued by a social stereotype is in contrast with a very simple rule of probability.

Dual-process researchers (e.g., Evans & Over, 1996; Kahneman & Frederick, 2002; Stanovich, 1999) have used these and other conflict tasks to demonstrate the “struggle” between type 1 (i.e., intuitive) and type 2 (i.e., conscious, effortful and, sometimes, rule-based) processes. Given that type 2 processes are postulated to be heavily dependent on working memory resources (e.g., Evans, 2011) and that they can be applied more efficiently when people possess knowledge of the relevant normative rules or principles (e.g., Stanovich & West, 2008), one could expect that children should be less likely to give normatively correct responses to such tasks as compared to adolescents and adults. Indeed, it has been suggested that the potential for type 2 processing to override type 1 processing should increase with development. Moreover, the tendency for override should be positively associated with differences in cognitive capacity among individuals of the same age (cf., Kokis et al., 2002; Stanovich et al., 2011). By contrast, as the implementation of heuristics requires minimal effort, even children should be able to utilize them. In fact, dual-process researchers argue that (in some form) both type 1 and 2 processing are available at all points of development – at least after an early age (e.g., Jacobs & Klaczynski, 2002; Kokis et al., 2002). As we have shown above in our review of the literature, the actual developmental findings are more complex than one could expect on a purely theoretical basis. Below we will try to explore the reasons for this, and to identify some general trends in the development of reasoning performance.

Method

Participants

We collected data from primary school, secondary school, and university students from Plymouth, UK. Results from the two youngest age groups were also reported in Morsanyi and Handley (2008). We have divided our sample into four age groups: younger children, older children, adolescents, and young adults. The younger children (n=43) were between the ages of 5 years 2 months and 8 years 6 months (M=7.08, SD=.85). The older children (n=41) were between the ages of 8 years 7 months and 11 years 7 months (M=9.99, SD=.87). The adolescents (n=49) were between the ages of 11 years 11 months and 15 years 7 months (M=13.82, SD=1.15). Finally, the young adults (n=72) were between the ages of 17 years 10 months and 24 years 11 months (M=20.99, SD=2.15).

Materials

Reasoning and judgement tasks. The three types of task that we used are known to elicit a high proportion of heuristic responses in the case of adults. We used four syllogistic reasoning, two sunk cost fallacy, and two conjunction fallacy tasks. Examples of each type of task can be found in Table 7.1 (all tasks are described in Morsanyi & Handley, 2008). For children, we presented the stories as a PowerPoint slide show that included photographic illustrations, and the text was read out by the experimenter. In all age groups, participants were given a booklet with the response options for each task, and they responded by circling what they considered to be the most appropriate response. In the case of two types of problem (the sunk cost fallacy, and the syllogistic reasoning tasks), there were three response options: heuristic, normative, and “other.” For the conjunction fallacy task, there were only two types of response: heuristic and normatively correct. All the tasks that we used were so-called “conflict” problems, that is, heuristic and normative responses were mutually exclusive.

Measures of general intelligence. Children and adolescents were administered the short form of the Wechsler Intelligence Scale for Children-III (WISC-III; Wechsler, 1991) consisting of the Vocabulary and Block Design subtests. This short form is reported to have the highest validity and reliability compared to any other two subtest short forms of the WISC (used jointly, these two subscales have a reliability of rtt=.91 and validity of r =.86; Sattler, 2001, Table A-16). As we were interested in our participants’ absolute computational capacity rather than their IQs, we used the raw scores on the tasks for our analyses.

Measures of verbal working memory. The verbal working memory tasks that we employed in the present study measured participants’ ability to temporarily store information while they process information. Children and adolescents were administered the counting span task (for a detailed description, see Handley et al., 2004). In this task children were presented with red and blue dots on the computer screen, and they had to count out loud the number of red dots. After a series of screens (the number of consecutive screens gradually increased from two to five), children were asked to recall the number of red dots on each screen in the order that they were presented in. A working memory global score was calculated by adding up the number of instances when the child recalled the number of dots on the preceding screens in the correct order.

In the case of adults, we used the operation span task (based on Turner & Engle, 1989; see Morsanyi & Handley, 2012 for a detailed description). The processing component consisted of verifying the solution to a two-step math problem (e.g., [2 x 5] – 4 = 6). A word was presented together with the math problem, and participants were instructed to memorize this word to recall at a later stage. After completing a set of problems (set length varied between two and five), participants had to recall the words in a written form and in the serial order in which they had been presented previously. We calculated the reliability of the working memory tasks by performing a split half correlation between the first and third trials at each set level, and then we used the Spearman–Brown formula to give an estimate of reliability. The reliability of the counting span task was.77, and for the operation span task it was.73.

Measures of inhibition. For adolescents and adults we also administered measures of inhibition skills. These tasks assess participants’ ability to make an appropriate response when they are presented with two conflicting cues. These tasks have been linked to the functioning of the anterior cingulate cortex (e.g., Swick & Jovanovich, 2002), which is presumably involved in detecting conflicts between beliefs and normative considerations during reasoning (De Neys, Vartanian, & Goel, 2008). In the adolescent sample, the stop-signal task (see Handley et al., 2004 for a detailed description) was used. This task consists of two types of trial: primary task trials and stop signal trials, which are randomly mixed together. On primary task trials, participants are presented either with an X or with an O on the computer screen, and they have to respond by pressing an X or an O button on one of two button boxes. On stop signal trials, the X or the O is presented along with a tone (the stop signal). On these trials, participants are instructed to withhold their response. The main measure of performance on the task was accuracy on the stop signal trials.

Adults completed a computerized version of the Stroop task. They were administered three blocks of 60 trials: the colour task, the word task, and the interference task. The order of administration for the three blocks was randomized across participants, and each block was preceded by ten practice trials. In each task, participants had to click with the mouse on one of four buttons displayed on the computer screen, labelled “red,” “blue,” “green,” and “yellow.” In the colour task, a patch of colour was displayed on the screen and participants had to click on the corresponding button. In the word task, a word (red, blue, green, or yellow) was displayed on the screen, and participants had to click on the button with the same label. Finally, in interference trials, the colour names were presented in an ink which conflicted with the colour name, and participants had to click on the button corresponding to the colour of the ink. Performance was indexed by the RT difference between the non-interference tasks and the interference task. The combined z scores of each participant’s average performance across these measures (i.e., the RT difference between the word task and the interference task, and the RT difference between the colour task and the interference task) was used.

We computed Cronbach’s alpha for both the stop-signal and the Stroop task (including correct trials only). In the case of the stop-signal task, performance was indexed by the mean number of correct responses to the stop signal trials. Consequently, when computing Cronbach’s alpha for this task, we included the stop-signal items only which resulted in r =.90. The measure of reliability for the Stroop task was based on the RTs for correct trials, and Cronbach’s alpha for this task was.82.

Procedure

Children and adolescents completed the tasks in two testing sessions, each lasting approximately 25 minutes. The first session consisted of the reasoning and judgement tasks. Participants were tested in groups of five to ten. The second session consisted of the counting span task, and the Vocabulary and Block Design subtests of the WISC-III. Adolescents additionally completed the stop-signal task. The participants performed these tasks individually, supported by the experimenter. University students completed all the tasks in a single session which was conducted in groups of five to six. Participants were instructed to consider the problems carefully and to always select one response only, which they found to be the most appropriate. However, they were not specifically instructed to think logically, as such instructions would have been inappropriate in the case of young children.

In order to facilitate reasoning in the youngest age groups, we presented the syllogisms in a fantasy context. Specifically, participants were instructed to imagine a distant planet, “which is similar to Earth in many respects, but there are some differences as well,” and to consider the problems with reference to this imaginary context (see also, e.g., Dias & Harris, 1988, 1990; Markovits & Lortie-Forgues, 2011). Such instructions have been found to facilitate reasoning on the basis of empirically false premises, and to make it possible for children to disregard their beliefs. For the sake of consistency, we presented the problems in this context for all age groups.

Results

Table 7.3 displays the proportion of heuristic, normative, and “other” responses on the different tasks in each age group. We have analyzed the age trends in heuristic, normative, and “other” responding separately for each task with one-way ANOVAs, and used Tukey’s B test to perform post-hoc contrasts.

In the case of the syllogistic reasoning task, heuristic (i.e., belief-based) responding remained stable across age groups. There was a marginal increase in normative responding (F(3,201)=2.19, p=.09, h 2p=.03), and a significant decrease in “other” (i.e., neither belief-based nor logical) responses – F(3,201)=7.96, p<.001, h 2p=.11. Tukey’s B test indicated that young children gave significantly more “other” responses than participants in the other age groups.

Table 7.3 Proportion of different types of responses across age groups

In the case of the sunk cost fallacy task, heuristic responding showed a U-shaped pattern with adolescents giving fewer heuristic responses than children and adults (F(3,201)=5.83, p=.001, h 2p=.08). Normative responding significantly increased with age (F(3,201)=12.78, p<.001, h 2p=.16). That is, children gave significantly fewer normative responses than adolescents and adults. The number of “other” responses significantly decreased with age (F(3,201)=17.81, p<.001, h 2p=.21). Young children gave significantly more “other” responses than participants from the other age groups, and older children gave significantly more “other” responses than adults. The number of “other” responses given by adolescents was not significantly different from that given by adults and older children.

Finally, we analysed the age-related changes on the conjunction fallacy task. In the case of the heuristic responses, we observed a reversed U-shaped pattern with lower levels of heuristic responding in the case of young children and adults than in older children and adolescents (F(3,201)=7.78, p<.001, h2p=.10). As on this task there were only two possibilities for participants in how they could respond to the task (i.e., they either committed the fallacy or not), the pattern for normative responses was the exact opposite of the pattern of heuristic responses.

We also analysed the age-related changes on the overall pattern of heuristic, normative and “other” responses collapsed across the three tasks (see Figure 7.2). The ANOVA on the overall heuristic responses indicated a reversed U-shaped pattern, with older children giving more heuristic responses than participants in the other age groups (F(3,201)=3.50, p=.016, h2p=.05). In the case of the normatively correct responses (F(3,201)=9.14, p<.001, h2p=.12) we observed a U-shaped pattern with older children giving fewer normative responses than adolescents and adults, and younger children giving fewer normative responses than adults. The performance of younger children was between that of older children and adolescents (and it was not significantly different from the other groups). There was also no significant difference between the performance of adolescents and adults. Finally, the proportion of “other” responses steadily decreased with age (F(3,201)=21.92, p<.001, h2p=.25), with younger children giving significantly more

Figure 7.2 Age trends in the overall proportion of heuristic, normative, and “other” responding across the different age groups.

“other” responses than participants in the other age groups, and older children giving significantly more “other” responses than adults.

We were also interested in the individual differences correlates of these developmental trends. In order to explore this issue, we computed correlations between heuristic responses and age, general intelligence, working memory capacity, and inhibition separately in each age group (see Table 7.4). Probably the most interesting finding is that whereas working memory capacity and general intelligence were positively correlated with heuristic responding in children,2 the reverse of this trend emerged in the case of adolescents and adults, although the correlations were not significant in the older age groups. Nevertheless, when we compared the correlation coefficients across children and adolescents (who were administered the same working memory and intelligence measures) using the Fisher r-to-z transformation, we found that the correlation coefficients across these samples were significantly different in the case of both general intelligence (z=3.03, p=.002) and working memory (z=3.02, p=.003). Another notable finding is the negative correlation between heuristic reasoning and inhibition in adolescence. This is in line with an earlier study by Handley et al. (2004).

Table 7.4 Correlations between heuristic responding and the individual differences variables in each age group

Discussion

The present study looked at age-related changes in heuristic reasoning across three types of task: syllogisms with a belief–logic conflict, and problems designed to elicit the sunk cost fallacy, and the conjunction fallacy. First we should note that the age trends on these tasks were somewhat similar, but not exactly the same. Nevertheless, it is still possible to identify some characteristic changes with development.

The results regarding “other” responses are quite clear. The tendency to give responses which neither feel right intuitively nor are correct in a normative sense decreases with development (see also Klaczynski, 2001b). In line with traditional theories of the development of reasoning skills (e.g., Piaget, 1976) normative responding tended to increase with age. This is especially true if we compare the performance of older children with that of adolescents or adults (this is, by the way, the comparison that we most often find in developmental studies). However, if we also consider the performance of young children, the picture is less clear. In fact, if anything, normative responding appeared to decrease during childhood. Stanovich et al. (2011) proposed that an apparent decrease in normative performance could be simply the result of a shift from random responding to a more systematic, heuristic-based responding during childhood. This shift is supposed to be the consequence of children acquiring more relevant knowledge, which they can use to inform (i.e., contextualize) their reasoning. Indeed, as we have described above, a number of studies found that the tendency to base responses on social stereotypes increases during childhood (see, e.g., Jacobs & Potenza, 1991), and there is also evidence that stereotype knowledge increases with development (De Neys & Vanderputte, 2011). Nevertheless, De Neys and Vanderputte (2011) found that older children were more likely to rely on social stereotypes (as opposed to statistical information) when solving problems, even in the case of tasks where both younger and older children were familiar with the relevant stereotype.

Another aspect of our findings regarding normative responding deserves attention. This is the fact that although adults gave the highest proportion of normative responses, their normative performance was not markedly different from that of adolescents’. Indeed, we only found a difference in normative performance between adolescents and adults in the case of the conjunction fallacy task.3 It should be noted that although adults have higher cognitive capacity than adolescents, and they presumably possess more relevant knowledge of a wide range of different normative rules, they might not be inclined to spontaneously utilize their knowledge and capacity for reasoning. In this respect, it is important to stress that in the present study participants were not specifically instructed to reason logically or to try to follow normative considerations. Thus, we have investigated their performance under unconstrained (i.e., typical) performance situations (cf. Stanovich et al., 2011). By contrast, when performance is investigated under optimal performance situations (i.e., when overt instructions to maximize performance are given – see, e.g., Ackerman, 1996; Sternberg, Grigorenko, & Zhang, 2008), we could expect that higher ability participants perform better, at least when they possess relevant knowledge (cf., Stanovich & West, 2008). Indeed, in a recent study (Chiesi et al., 2011) we found that university students performed better than secondary school students on a set of probabilistic reasoning tasks where normative rules conflicted with heuristically cued responses. However, this was only the case when participants were explicitly instructed to reason like “a perfectly logical and rational person.” Without such instructions, the performance of the two groups was at the same level. Thus, although the capacity for normative reasoning and individuals’ relevant knowledge increases with development, this does not necessarily lead to increases in normative reasoning.

Finally, it is also possible that university students were less inclined to reason carefully, as they underestimated the difficulty of the tasks. This could be the consequence of the childish content of the tasks, and the fact that university students were informed that their performance on the tasks would be compared to that of children. In this respect, it is interesting to note that in adulthood there seems to be a tendency for individuals to adopt heuristic approaches to deciding whether they should reason about problems effortfully or in a more casual manner. Such powerful heuristic cues include the fluency of processing, the ease of retrieval of relevant information, and emotional responses accompanying the processing of information (see e.g., Hertwig, Herzog, Schooler, & Reimer, 2008; Thompson & Morsanyi, 2012; Thompson, Prowse-Turner, & Pennycook, 2011; Topolinski & Reber, 2011).

Now we will discuss our findings regarding heuristic responding. First of all, in childhood there seems to be a general trend for increasing contextualization, and an increasing tendency for children to use relevant knowledge when they reason. In the case of the syllogistic reasoning and the conjunction fallacy tasks, this corresponded to an increase in children’s tendency to base their responses on their beliefs. For the sunk cost problems the heuristic response is based on the application of the “waste not” rule.

In striking contrast with the assumption of dual-process theories that heuristic reasoning is automatic and effortless, the responses that are presumably based on heuristic reasoning in adulthood were found to be effortful in the case of children. This was evidenced by the positive relationships between our indices of cognitive capacity and heuristic responding in childhood. This is even the more interesting, as we found a negative relationship between heuristic responding and cognitive capacity in adulthood (although this trend did not reach significance). Thus, the tendency to contextualize is stronger in the case of higher capacity children, whereas higher capacity adults tend to decontextualize. De Neys and Vanderputte (2011) proposed that this pattern in childhood might simply reflect the fact that higher ability children possess more relevant knowledge, and, thus, they are more likely to use this knowledge. Nevertheless, in their study, whilst they included both familiar and non-familiar stereotypes, they did not measure children’s cognitive capacity. Thus, it is impossible to tell if higher ability children actually possessed more relevant knowledge. By contrast, there is much evidence that higher ability children (and adults) are more able to employ relevant knowledge when they are engaged in a reasoning task (e.g., Barrouillet, Markovits, & Quinn, 2002; De Neys, Schaeken, & d’Ydewalle, 2005; Janveau-Brennan & Markovits, 1999). Thus, it is likely that the increase in knowledge-based and contextualized reasoning in childhood does not simply reflect an increase in children’s knowledge of the world, but is also the consequence of an increased ability to apply this knowledge. Furthermore, the ability to apply relevant knowledge is positively associated with cognitive ability in childhood.

In adolescence and adulthood, this tendency to use relevant knowledge is also paired with an increasing ability to discriminate between situations when knowledge-based reasoning is appropriate, and when it is not. Indeed, real-life knowledge can support as well as hinder reasoning (e.g., De Neys & Everaerts, 2008; Janveau-Brennan & Markovits, 1999; Markovits, 2000). In line with Handley et al. (2004), we found that the ability to suppress the non-normative use of heuristics was negatively correlated with inhibition skills in adolescence. Interestingly, although it is generally assumed that adults have to inhibit the tendency for automatic contextualization during reasoning in order to be able to give normative responses to conflict problems, we did not find a correlation between heuristic responding and inhibition in the case of adults. One possibility is that although inhibition skills are needed for suppressing context-based reasoning, the inhibition demands are relatively greater for adolescents than for adults. Another factor that might moderate the role of inhibition skills is individual differences in the spontaneous tendency for effortful thinking in the university student sample. Indeed, it is possible that some students with a high capacity for inhibition chose not to employ this capacity when reasoning in the absence of explicit instructions to think effortfully. It is also possible that the contextualization of materials is optional in adulthood, and that in cases when the logical structure of the problem is simple, the contextualization of the problem does not necessarily take place (cf., Handley, Newstead, & Trippas, 2011).

Finally, as we used two different inhibition tasks, it is also a possibility that the stop-signal task is a better predictor of performance on the particular types of problem that we used in the present study than the Stroop task. Although both tasks can be conceptualized as decision-making paradigms involving a trade-off between fast responding and accurate inhibition, the inhibitory demands of the two tasks differ. Whereas the Stroop task requires sustained inhibition in the interference trials (i.e., participants have to disregard the colour names and just focus on the colour of the ink), the stop-signal task demands that participants stay tuned to the possibility that they have to inhibit a response, although this only happens at random intervals (thus, whereas the Stroop task assesses baseline performance and inhibition separately, in the stop-signal task the two types of trials are intermixed). Regardless of the interpretation of the present findings, it is an interesting question if problems with a conflict between a heuristic and normative response generally share their inhibitory demands, as it is often assumed (e.g., De Neys & Van Gelder, 2008; Stanovich & West, 2008), or the nature of the inhibitory demand of different tasks varies. Alternatively, it remains a possibility that conflict resolution does not necessarily demand inhibitory control.

One more aspect of our findings regarding heuristic reasoning deserves attention. Whereas there was a clear trend for increases in normative responding from (late) childhood to adulthood, and the reverse of this trend was observed in the case of “other” responses, the patterns for heuristic reasoning were more complex. Although during childhood the tendency for heuristic reasoning increased and this increase was positively correlated with children’s cognitive capacity, between late childhood and adolescence there was no change in the proportion of stereotype-based responses on the conjunction fallacy task, and belief-based responses in syllogistic reasoning. At the same time, there was a dip in committing the sunk cost fallacy (i.e., adolescents were less influenced by past investments than children and adults). Interestingly, older adults are also less susceptible to the sunk cost fallacy than younger adults (Strough, Mehta, McFall, & Schuller, 2008). Thus, it seems that besides linear and (reversed) U-shaped patterns, some heuristics might show an undulating pattern across the lifespan. Finally, although adults possess the highest capacity for normative responding, this is not necessarily paired with an increased spontaneous tendency to avoid heuristic reasoning. This is in line with the fuzzy-trace theory (e.g., Reyna & Brainerd, 2011) which claims that gist-based, intuitive processing increases with age and experience. That is, more experienced reasoners prefer a qualitative assessment of information (e.g., about probabilities and risk) to precise, step-by-step computations.

Concluding comments

In this chapter we proposed that whereas reasoning is often conceptualized as a “struggle” between effortful and automatic processes, it is also possible to conceive reasoning problems as a collection of cues that can evoke more or less effortful problem-solving strategies. A choice between these strategies might depend on the way in which problems are presented (i.e., whether there is time pressure, whether participants are internally or externally motivated to reason carefully, etc.). Additionally, the selection of available strategies to tackle problems also increases with development as a result of increases in individuals’ knowledge, their familiarity with different domains, and their cognitive capacity.

Stanovich et al. (2011) claimed that development proceeds from random responding through “smart errors,” followed by normative responding, whereas, according to Evans (2011), non-contextualized reasoning is followed by a stage of contextualized reasoning without inhibition, and later on by contextualized reasoning with inhibition. With regard to Stanovich et al.’s (2011) position, although random responding might happen in the case of complex or unfamiliar tasks, as our study and several examples in the literature (e.g., regarding proportional or probabilistic reasoning) suggest, an increase in heuristic responses is most often associated with the acquisition and conscious application of novel knowledge, regardless of whether people had a dominant strategy for solving the problem before or not. Moreover, the same knowledge can form the basis of both normative and non-normative responding (e.g., Tirosh & Stavy, 1999; Van Dooren et al., 2005) although higher ability participants might be better able to discriminate between situations when a rule applies and when it does not (e.g., Gillard et al., 2009; Morsanyi et al., 2009). It is also important that although the capacity for normative reasoning increases with development, the tendency for heuristic responding remains strong during adulthood. In this respect it is notable that heuristics are aimed at reducing the computational demands of solving complex problems (e.g., Shah & Oppenheimer, 2008; Simon, 1990), and that people might be motivated to rely on heuristics whenever possible. Given that heuristics are informed by experiences and learning, it is also not surprising that they often result in correct responses. Indeed, people might even have correct intuitions about logical solutions for very complex tasks, such as syllogisms (Morsanyi & Handley, 2012). Finally, although normative reasoning increases with development, there is an equally strong tendency for developing reasoning heuristics.

The developmental stages identified by Evans (2011) correspond nicely to the patterns that we found in younger children, older children, and adolescents. Nevertheless, adult reasoning might not be best characterized by an interaction between contextualization and inhibition. Although, in fact, adults tend to predominantly rely on contextualized reasoning, and they also have the ability to suppress this tendency, this might not necessarily mean that non-contextualized reasoning is always based on the suppression of a contextualized response (see Handley, Newstead, & Trippas, 2011). It might be the case that adults are able to shift between context-based and rule-based strategies, depending on the available time, cognitive resources, and motivational factors. In summary, the story of cognitive development is only partly about an increase in the capacity for normative reasoning. There is an equally compelling story to tell about the development of an increasing range of heuristics which pervade all domains of reasoning. In fact, given that heuristics often rely on normative knowledge (such as in the case of some heuristics in the domain of scientific reasoning – e.g., the overuse of proportionality, see Van Dooren et al., 2008), and that they seem to originate in effortful processing (as evidenced by a positive relationship between heuristic responding and cognitive ability in childhood – see above), these two stories might not be so dissimilar after all.

Notes

1 Here we use the label “normative” to describe responses that are generally regarded as normatively correct, although it should be noted that some of these normative standards have been debated (e.g., Hertwig & Gigerenzer, 1999)

2 In fact, when we partialled out the effect of age from the correlations between general intelligence and heuristic reasoning, and working memory capacity and heuristic reasoning in the child samples (collapsed across the two age groups), the correlations were still significant: r(81) =.34 p<.01 and r(81) =.26 p<.05.

3 Although there was no significant difference in normative responding between adolescents and adults, Figure 7.2 shows a trend for age-related increases in normative responding. This is accompanied by a decrease in atypical (“other”) responses, whereas there is no change in heuristic responding. Klaczynski (2001b) reported very similar age trends, using ratio bias, sunk cost, and if-only problems.

References

Ackerman, P. L. (1996). A theory of adult development: Process, personality, interests, and knowledge. Intelligence, 22, 227–257.

Arkes, H.R. & Ayton, P. (1999). The sunk cost and concorde effects: Are humans less rational than lower animals? Psychological Bulletin, 125, 591–600.

Barrouillet, P. (2011). Dual-process theories of reasoning: The test of development. Developmental Review, 31, 151–179.

Barrouillet, P., Markovits, H. & Quinn, S. (2002). Developmental and content effects in reasoning with causal conditionals. Journal of Experimental Child Psychology, 81, 235–248.

Batanero, C., Serrano. L. & Garfield, J. B. (1996). Heuristics and biases in secondary school students’ reasoning about probability. In L. Puig & A. Gutiérrez (Eds.), Proceedings of the 20th Conference on the International Group for the Psychology of Mathematics Education (vol. 2, pp. 51–59). University of Valencia.

Brainerd, C. (1981). Working memory and the developmental analysis of probability judgment. Psychological Review, 88, 463–502.

Brainerd, C. J. & Reyna, V. F. (2001). Fuzzy-trace theory: Dual processes in memory, reasoning, and cognitive neuroscience. Advances in Child Development and Behavior, 28, 49–100.

Cacioppo, J. T. & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116–131.

Chiesi, F., Gronchi, G. & Primi, C. (2008). Age-trend related differences in task involving conjunctive probabilistic reasoning. Canadian Journal of Experimental Psychology, 62, 188–191.

Chiesi, F. & Primi, C. (2009). Recency effects in primary-age children and college students using a gaming situation. International Electronic Journal of Mathematics Education, Special issue on “Research and Development in Probability Education,” 4, www.iejme.com.

Chiesi, F., Primi, C. & Morsanyi, K. (2011). Developmental changes in probabilistic reasoning: The role of cognitive capacity, instructions, thinking styles and relevant knowledge. Thinking & Reasoning, 17, 315–350.

Cramer, K., Post, T. & Currier, S. (1993). Learning and teaching ratio and proportion: Research implications. In D. Owens (Ed.), Research ideas for the classroom (pp. 159–178). New York: Macmillan Publishing Company.

Davidson, D. (1995). The representativeness heuristics and the conjunction fallacy in children’s decision making. Merrill Palmer Quarterly, 41, 328–346.

Denes-Raj, V. & Epstein, S. (1994). Conflict between intuitive and rational processing: When people behave against their better judgment. Journal of Personality and Social Psychology, 66, 819–829.

De Neys, W. (2006). Dual processing in reasoning: Two systems but one reasoner. Psychological Science, 17, 428–433.

De Neys, W. (2007). Development of decision making: The case of the Monty Hall Dilemma. In J. A. Elsworth (Ed.), Psychology of decision making in education, behavior, and high risk situations (pp. 271–281). Hauppage, NY: Nova Science Publishers.

De Neys, W. & Everaerts, D. (2008). Developmental trends in everyday conditional reasoning: The retrieval and inhibition interplay. Journal of Experimental Child Psychology, 100, 252–263.

De Neys, W., Schaeken, W. & d’Ydewalle, G. (2005). Working memory and everyday conditional reasoning: Retrieval and inhibition of stored counterexamples. Thinking & Reasoning, 11, 349–81.

De Neys, W. & Vanderputte, K. (2011). When less is not always more: Stereotype knowledge and reasoning development. Developmental Psychology, 47, 432–441.

De Neys, W. & Van Gelder, E. (2008). Logic and belief across the life span: The rise and fall of belief inhibition during syllogistic reasoning. Developmental Science, 12, 123–130.

De Neys, W., Vartanian, O. & Goel, V. (2008). Smarter than we think: When our brains detect that we are biased. Psychological Science, 19, 483–489.

Dias, M. G. & Harris, P. L. (1988). The effect of make-believe play on deductive reasoning. British Journal of Developmental Psychology, 6, 207–221.

Dias, M.G. & Harris, P. L. (1990). The influence of the imagination on reasoning by young children. British Journal of Developmental Psychology, 8, 305–317.

Epstein, S., Lipson, A., Holstein, C. & Huh, E. (1992). Irrational reactions to negative outcomes: Evidence for two conceptual systems. Journal of Personality and Social Psychology, 62, 328–339.

Evans, J. St. B. T. (2003). In two minds: Dual process accounts of reasoning. Trends in Cognitive Sciences, 7, 454–459.

Evans, J. St. B.T (2006). The heuristic-analytic theory of reasoning: Extension and evaluation. Psychnomic Bulletin and Review, 13, 378–395.

Evans, J. St. B. T. (2011). Dual-process theories of reasoning: Contemporary issues and developmental applications. Developmental Review, 31, 86–102.

Evans, J. St. B. T., Barston, J. L. & Pollard, P. (1983). On the conflict between logic and belief in syllogistic reasoning. Memory and Cognition, 11, 295–306.

Evans, J. St. B. T. & Curtis-Holmes, J. (2005). Rapid responding increases belief bias:

Evidence for the dual-process theory of reasoning. Thinking & Reasoning, 11, 382–389.

Evans, J. St. B. T., Handley, S. J., Neilens, H. & Over, D. E. (2009). The influence of cognitive ability and instructional set on causal conditional inference. Quarterly Journal of Experimental Psychology, 63, 892–909.

Evans, J. St. B. T. & Over, D. E. (1996). Rationality and reasoning. Hove, UK: Psychology Press. Feeney, A., Scrafton, S., Duckworth, A. & Handley, S. J. (2004). The story of some:

Everyday pragmatic inference by children and adults. Canadian Journal of Experimental Psychology, 58, 90–101.

Ferreira, M. B., Garcia-Marques, L., Sherman, S. J. & Sherman, J. W. (2006). Automatic and controlled components of judgment and decision making. Journal of Personality and Social Psychology, 91, 797–813.

Fischbein, E. & Schnarch, D. (1997). The evolution with age of probabilistic, intuitively based misconceptions. Journal for Research in Mathematics Education, 28, 96–105.

Fong, G. T., Krantz, D. H. & Nisbett, R. E. (1986). The effects of statistical training on thinking about everyday problems. Cognitive Psychology, 18, 253–292.

Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19, 25–42.

Gillard, E., Van Dooren, W., Schaeken, W. & Verschaffel, L. (2009). Proportional reasoning as a heuristic-based process: Time constraint and dual-task considerations. Experimental Psychology, 56, 92–99.

Gilovich, T., Griffin, D. & Kahneman, D. (Eds.). (2002). Heuristics and biases: The psychology of intuitive judgment. Cambridge: Cambridge University Press.

Handley, S., Capon, A., Beveridge, M., Dennis, I. & Evans, J. St. B. T. (2004). Working memory, inhibitory control and the development of children’s reasoning. Thinking & Reasoning, 10, 175–195.

Handley, S. J., Newstead, S.E. & Neilens, H. (2011). Matching bias requires analytic reasoning. In K.I. Manktelow, D. Over & S. Elqayam (Eds.), The science of reason: A festschrift for Jonathan St. B. T. Evans (pp. 167–189). Hove, UK: Psychology Press.

Handley, S. J., Newstead, S. E. & Trippas, D. (2011). Logic, beliefs, and instruction: A test of the default interventionist account of belief bias. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37, 28–43.

Hertwig, R. & Gigerenzer, G. (1999). The “conjunction fallacy” revisited: How intelligent inferences look like reasoning errors. Journal of Behavioral Decision Making, 12, 275–305.

Hertwig, R., Herzog, S. M., Schooler, L. J. & Reimer, T. (2008). Fluency heuristic: A model of how the mind exploits a by-product of information retrieval. Journal of Experimental Psychology: Learning, Memory, & Cognition, 34, 1191–1206.

Jacobs, J. E. & Klaczynski, P. A. (2002). The development of decision making during childhood and adolescence. Current Directions in Psychological Science, 4, 145–149.

Jacobs, J. E. & Potenza, M. (1991). The use of judgment heuristics to make social and object decision: A developmental perspective. Child Development, 62, 166–178.

Janveau-Brennan, G. & Markovits, H. (1999). The development of reasoning with causal conditionals. Developmental Psychology, 35, 904–911.

Kahneman, D. & Frederick, S. (2002). Representativeness revisited: Attribute substitution in intuitive judgement. In T. Gilovich, D. Griffin & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp. 49–81). New York: Cambridge University Press.

Kahneman, D., Slovic, P. & Tversky, A. (1982). Judgment under uncertainty: Heuristics and biases. New York: Cambridge University Press.

Keren, G. & Schul, Y. (2009) Two is not always better than one: A critical evaluation of two-system theories. Perspectives on Psychological Science, 4, 533–550.

Klaczynski, P. A. (2000). Motivated scientific reasoning biases, epistemological beliefs, and theory polarization: A two-process approach to adolescent cognition. Child Development, 71, 1347–1366.

Klaczynski, P. A. (2001a). The influence of analytic and heuristic processing on adolescent reasoning and decision making. Child Development, 72, 844–861.

Klaczynski, P. A. (2001b). Framing effects on adolescent task representations, analytic and heuristic processing, and decision making: Implications for the normative-descriptive gap. Journal of Applied Developmental Psychology, 22, 289–309.

Klaczynski, P. A. (2009). Cognitive and social cognitive development: Dual-process research and theory. In J. St. B. T. Evans & K. Frankish (Eds.), In two minds: Psychological and philosophical theories of dual processing (pp. 265–292). Oxford: Oxford University Press.

Klaczynski, P. A. & Cottrell, J. E. (2004). A dual-process approach to cognitive development: The case of children’s understanding of sunk cost decisions. Thinking & Reasoning, 10, 147–174.

Kokis, J., Macpherson, R., Toplak, M. E., West, R. F. & Stanovich, K. E. (2002). Heuristic and analytic processing: Age trends and associations with cognitive ability and cognitive styles. Journal of Experimental Child Psychology, 83, 26–52.

Konold, C. (1989). Issues in assessing conceptual understanding in probability and statistics. Journal of Statistics Education, 3, www.amstat.org/publications/jse/v3n1/konold.html

Lecoutre, M. P. (1985). Effect d’informations de nature combinatoire et de nature fréquentielle sur les judgements probabilistes. Recherches en Didactique des Mathématiques, 6, 193–213.

Lehman, D. R., Lempert, R. O. & Nisbett, R. E. (1988). The effects of graduate training on reasoning: Formal discipline and thinking about everyday-life events. American Psychologist, 43, 431–442.

Markovits, H. (2000). A mental model analysis of young children’s conditional reasoning with meaningful premises. Thinking & Reasoning, 6, 335–347.

Markovits, H. & Dumas, C. (1999). Developmental patterns in the understanding of social and physical transitivity, Journal of Experimental Child Psychology, 73, 95–114.

Markovits, H. & Lortie Forgues, H. (2011). Conditional reasoning with false premises facilitates the transition between familiar and abstract reasoning. Child Development, 82, 646–660.

McGlone, M. S. & Tofighbakhsh, J. (2000). Birds of a feather flock conjointly(?): Rhyme as reason in aphorisms. Psychological Science, 11, 424–428.

Morsanyi, K. & Handley, S. J. (2008). How smart do you need to be to get it wrong? The role of cognitive capacity in the development of heuristic-based judgment. Journal of Experimental Child Psychology, 99, 18–36.

Morsanyi, K. & Handley, S. J. (2012). Logic feels so good – I like it! Evidence for intuitive detection of logicality in syllogistic reasoning. Journal of Experimental Psychology: Learning, Memory and Cognition, 38, 596–616.

Morsanyi, K., Handley, S. J. & Serpell, S. (2013). Making heads or tails of probability: An experiment with random generators. British Journal of Educational Psychology (in press).

Morsanyi, K., Primi, C., Chiesi, F. & Handley, S. (2009). The effects and side-effects of statistic education. Psychology students’ (mis-)conceptions of probability. Contemporary Educational Psychology, 34, 210–220.

Moshman, D. (1999). Adolescent psychological development. Mahwah, NJ: Lawrence Erlbaum Associates.

Noveck, I. A. (2001). When children are more logical than adults: Investigations of scalar implicature. Cognition, 78, 165–188.

Oppenheimer, D. M. (2006) Consequences of erudite vernacular utilized irrespective of necessity: Problems with using long words needlessly. Applied Cognitive Psychology, 20, 139–156.

Osman, M. & Stavy, R. (2006). Development of intuitive rules: Evaluating the application of the dual-system framework to understanding children’s intuitive reasoning. Psychonomic Bulletin & Review, 13, 935–953.

Pacini, R. & Epstein, S. (1999). The relation of rational and experiential information processing styles to personality, basic beliefs, and the ratio-bias phenomenon. Journal of Personality and Social Psychology, 76, 972–987.

Piaget, J. (1976). The grasp of consciousness. Cambridge, MA: Harvard University Press. Piaget, J., Inhelder, B. & Szeminska, A. (1960). The child’s conception of geometry. London: Routledge & Kegan Paul.

Reyna, V. F. & Brainerd, C. J. (2011). Dual processes in decision making and developmental neuroscience: A fuzzy-trace model. Developmental Review, 31, 180–206.

Reyna, V. F. & Ellis, S. C. (1994). Fuzzy-trace theory and framing effects in children’s risky decision making. Psychological Science, 5, 275–279.

Reyna, V. F. & Farley, F. (2006). Risk and rationality in adolescent decision making: Implications for theory, practice, and public policy. Psychological Science in the Public Interest, 7, 1–44.

Roberts, M. J. & Newton, E. J. (2001). Inspection times, the change task, and the rapid response selection task. Quarterly Journal of Experimental Psychology, 54, 1031–1048.

Sattler, J. M. (2001). Assessment of children: Cognitive applications (4th ed.). San Diego, CA: Jerome M. Sattler.

Shah, A. K. & Oppenheimer, D. M. (2008). Heuristics made easy: An effort-reduction framework. Psychological Bulletin, 134, 207–222.

Simon, H. A. (1990). Invariants of human behavior. Annual Review of Psychology, 41, 1–19.

Simoneau, M. & Markovits, H. (2003). Reasoning with premises that are not empirically true: Evidence for the role of inhibition and retrieval. Developmental Psychology, 39, 964–975.

Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119, 3–22.

Stanovich, K. E. (1999). Who is rational? Studies of individual differences in reasoning. Mahwah, NJ: Lawrence Erlbaum Associates.

Stanovich, K. E., Toplak, M. E. & West, R. F. (2008). The development of rational thought: A taxonomy of heuristics and biases. Advances in Child Development and Behaviour, 36, 251–285.

Stanovich, K. E. & West, R. F. (1997). Reasoning independently of prior belief and individual differences in actively open-minded thinking. Journal of Educational Psychology, 89, 342–357.

Stanovich, K. E. & West, R. F. (2000). Individual differences in reasoning: Implications for the rationality debate? Behavioral and Brain Sciences, 23, 645–665.

Stanovich, K. E. & West, R. F. (2008). On the relative independence of thinking biases and cognitive ability. Journal of Personality and Social Psychology, 94, 672–695.

Stanovich, K. E., West, R. F. & Toplak, M. E. (2011). The complexity of developmental predictions from dual process models. Developmental Review, 31, 103–118.

Stavy, R. & Tirosh, D. (2000). How students (mis-)understand science and mathematics: Intuitive rules. New York: Teachers College Press.

Sternberg, R. J., Grigorenko, E. L. & Zhang, L. (2008). Styles of learning and thinking matter in instruction and assessment. Perspectives on Psychological Science, 3, 486–506.

Strough, J., Mehta, C. M., McFall, J. P. & Schuller, K. L. (2008). Do older and younger adults make different decisions about sunk costs? Psychological Science, 19, 650–652.

Swick, D. & Jovanovic, J. (2002). Anterior cingulate cortex and the Stroop task: Neuropsychological evidence for topographic specificity. Neuropsychologia, 40, 1240–1253.

Thompson, V. A. (2009). Dual process theories: A metacognitive perspective. In J. Evans and K. Frankish (Eds.), In two minds: Dual processes and beyond (pp. 171–196). Oxford: Oxford University Press.

Thompson, V. A. & Morsanyi, K. (2012). Analytic thinking: Do you feel like it? Mind & Society, 11, 93–105.

Thompson, V.A., Prowse-Turner, J. & Pennycook, G. (2011). Intuition, reason and metacognition. Cognitive Psychology, 63, 107–140.

Tirosh, D. & Stavy, R. (1999). Intuitive rules: A way to explain and predict students’ reasoning. Education Studies in Mathematics, 38, 51–66.

Toplak, M. E., Liu, E., Macpherson, R., Toneatto, T. & Stanovich, K. E. (2007). The reasoning skills and thinking dispositions of problem gamblers: A dual-process taxonomy. Journal of Behavioral Decision Making, 20, 103–124.

Topolinski, S. & Reber, R. (2010). Gaining insight into the “aha” experience. Current Directions in Psychological Science, 19, 402–405.

Turner, M. L. & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28, 127–154.

Tversky, A. & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211, 453–458.

Tversky, A. & Kahneman, D. (1983). Extentional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Bulletin, 90, 293–315.

Van Dooren, W., De Bock, D., Evers, M. & Verschaffel, L. (2009). Students’ overuse of proportionality on missing-value problems: How numbers may change solutions. Journal for Research in Mathematics Education, 40, 187–211.

Van Dooren, W., De Bock, D., Hessels, A., Janssens, D. & Verschaffel, L. (2005). Not everything is proportional: Effects of age and problem type on propensities for overgeneralization. Cognition and Instruction, 23, 57–86.

Webley, P. & Plaisier, Z. (1998). Mental accounting in childhood. Children’s Social and Economics Education, 3, 55–64.

Wechsler, D. (1991). Wechsler intelligence scale for children–third edition (WISC-III). San Antonio, TX: Psychological Corporation.