You are on page 1of 10

Brain and Cognition 55 (2004) 148157 www.elsevier.

com/locate/b&c

Development of hot executive function: The childrens gambling task


Aurora Kerr and Philip David Zelazo*
Department of Psychology, University of Toronto, Toronto, Ont., Canada M5S 3G3 Accepted 22 September 2003 Available online 5 March 2004

Abstract Development of aective decision-making was studied in 48 children at two ages (3 and 4 years) using a simplied version of the Iowa Gambling Task (Bechara, Damasio, Damasio, & Anderson, 1994). On each of 50 trials, children chose from 1 of 2 decks of cards that, when turned, displayed happy and sad faces, corresponding to rewards (candies) won and lost, respectively. Cards in 1 deck oered more rewards per trial, but were disadvantageous across trials due to occasional large losses; cards in the other deck oered fewer rewards per trial, but were advantageous overall. On later trials, 4-year-olds made more advantageous choices than 3-year-olds, and 4year-olds made more advantageous choices than would be expected by chance, whereas 3-year-olds made more disadvantageous choices than would be expected by chance. These ndings, which were especially pronounced for girls, indicate that aective decisionmaking develops rapidly during the preschool period, possibly reecting the growth of neural systems involving orbitofrontal cortex. 2003 Elsevier Inc. All rights reserved.
Keywords: Executive function; Hot executive function; Decision-making; Gambling; Young children; Orbitofrontal cortex; Sex dierences

1. Introduction Executive function (EF) is an ill-dened but important construct that refers generally to the psychological processes involved in the conscious control of thought and action. Traditionally, research on EF in human beings has focused almost exclusively on the more purely cognitive, cool aspects usually associated with dorsolateral prefrontal cortex (DL-PFC; Zelazo & Mller, 2002; cf. Metcalfe & Mischel, 1999; u Miller & Cohen, 2001). Recently, however, there has been growing interest in the relatively hot aective aspects of EF associated with orbitofrontal cortex (OFC), as wellin particular in what might be called aective decision-making, or decision-making about events that have emotionally signicant consequences (i.e., meaningful rewards and/or losses). To study affective decision-making, researchers have developed a

* Corresponding author. E-mail address: zelazo@psych.utoronto.ca (P.D. Zelazo).

number of useful measures, including measures of gambling (e.g, Bechara et al., 1994; this issue), risky decision-making (e.g., Rogers et al., 1999), and guessing with feedback (e.g., Elliott, Frith, & Dolan, 1997; for comparisons among measures, see Manes et al., 2002; Monterosso, Ehrman, Napier, OBrien, & Childress, 2001). One of the most widely used measures of aective decision-making is the Iowa Gambling Task. In an initial study (Bechara et al., 1994), OFC patients and healthy controls were presented with four decks of cards that, when turned, revealed a combination of gains and losses (measured in play money). Participants were given a stake of $2000 and asked to win as much money as possible by choosing cards from any of the four decks (one card per trial). They were not told how many trials there would be (100), but they were told that some of the decks were better than the others. In fact, the task was designed so that choosing consistently from two of the decks (the advantageous decks) would result in a net gain, whereas choosing consistently from the other two (the disadvantageous decks) would result in a net loss.

0278-2626/$ - see front matter 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0278-2626(03)00275-6

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

149

Each card from the disadvantageous decks provided a higher reward than each card from the advantageous decks ($100 vs. $50), but the variable (and unpredictable) losses associated with cards from disadvantageous decks were much larger on average than the losses associated with the advantageous decks. Bechara et al. (1994) found that both patients and controls typically preferred the disadvantageous cards at the outset. Over trials, however, controls were increasingly likely to select from the advantageous decks, whereas patients were not. Subsequent studies (e.g., Bechara, Damasio, Tranel, & Damasio, 1997) conrmed and extended these ndings (but see Manes et al., 2002, for some qualications regarding lesion focus and extent), and similar impairments on the Iowa Gambling Task have been documented in pathological gamblers (Cavedini, Riboldi, Keller, DAnnucci, & Bellodi, 2002) and individuals abusing cocaine (Monterosso et al., 2001), heroin (Petry, Bickel, & Arnett, 1998), alcohol (Mazas, Finn, & Steinmetz, 2000), and a combination of drugs (Bechara et al., 2001; Grant, Contoreggi, & London, 2000). Moreover, Reavis and Overman (2001) have shown that men outperform women on the taska nding consistent with evidence of sex dierences in OFC function early in ontogeny (see next section). 1.1. Development of OFC and hot EF Prefrontal cortex exhibits considerable growth during the course of childhood and into adolescence, as indicated by age-related changes in myelination (Pfeerbaum et al., 1994; Yakolev & Lecours, 1967), interhemispheric connectivity (Thompson et al., 2000), synaptic density (Huttenlocher, 1979, 1990), and metabolic and electrical activity (e.g., Luciana & Nelson, 1998; Travis, 1998; Rubia et al., 2000), among other measures. Although data on the growth of particular regions within prefrontal cortex are scarce, there is some suggestion that OFC develops earlier than DL-PFC (e.g., Orzhekhovskaya, 1981). Most research on the functional implications of these structural changes has focused on cool EF. Current evidence indicates that cool EF rst emerges early in ontogeny, probably around the end of the rst year of life, and undergoes important changes between 3 and 4 years of age (see Zelazo & Mller, 2002, for a review). For example, in one widely u used measure of cool EF, the Dimensional Change Card Sort (Frye, Zelazo, & Palfai, 1995), children are asked to sort a series of colored shapes rst by one dimension (e.g., color), and then by the other (e.g., shape). Whereas 4-year-olds switch exibly, 3-year-olds systematically perseverate on the pre-switch rules during the postswitch phase, despite being able to describe the rules they fail to use (e.g., Bialystok, 1999; Carlson & Moses, 2001; Jacques, Zelazo, Kirkham, & Semcesen, 1999;

Kirkham, Cruess, & Diamond, 2003; Munakata & Yerys, 2001; Perner, Stummer, & Lang, 1999; Zelazo, Frye, & Rapus, 1996). Much less is known about the development of hot EF. However, work by Overman and colleagues (Overman, Bachevalier, Schuhmann, & Ryan, 1996; see also Overman, this issue) has examined the development of performance on object reversal, which has long been known to depend on intact OFC (e.g., Iverson & Mishkin, 1970). In this task, participants are rst required to learn the reinforcement values of two objects using a nonverbal operant procedure (e.g., choosing one object is rewarded with food and praise, choosing another is not). Then the reward contingencies are reversed and participants must learn the reversed values. Overman et al. (1996) demonstrated age-related improvements in performance on this task in infants and young children. In addition, these authors found that prior to 30 months of age, boys performed better than girls. A ceiling eect may have obscured sex dierences beyond 30 months. The presence of a sex dierence in object reversal is particularly intriguing in light of earlier work using this task that suggests that OFC develops more slowly in female monkeys than in male monkeys, and that this sex dierence is under the control of gonadal hormones (Clark & Goldman-Rakic, 1989; Goldman, Crawford, Stokes, Galkin, & Rosvold, 1974). A similar sex dierence in rate of OFC growth may underlie the behavioral dierences documented by Overman et al. (1996). The primary aim of the present study was to examine the development of hot EF using a task that would be sensitive to age-related changes between 3 and 4 years, a period marked by major changes in cool EF. For this purpose, we created an age-appropriate version of the Iowa Gambling Task, which we refer to as the Childrens Gambling Task. Compared to the Iowa Gambling Task, the Childrens Gambling Task was modied in several ways: it involved two decks of cards instead of four; rewards were candy instead of play money; information about gains and losses was conveyed using happy and sad faces instead of written text, and involved smaller quantities; children were provided with a sample of cards from each deck; there were only 50 trials instead of 100. As in the Iowa Gambling Task, however, children were presented with a choice between advantageous and disadvantageous decks. Cards in one deck oered more rewards per trial, but were disadvantageous across trials due to occasional large losses; cards in the other deck oered fewer rewards per trial, but were advantageous overall. In light of the research reviewed earlier, it was expected that 3-year-olds would tend to make disadvantageous choices, whereas 4-yearolds would tend to make advantageous choices. Moreover, it was expected that any sex dierence would favor boys.

150

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

2. Method 2.1. Participants Twenty-four children at each of two ages, 3 years (M 42:5 months, range 3747) and 4 years (M 53 months, range 4859), were recruited from daycare centers and a database containing names of parents who had expressed an interest in research. Half of the children at each age were boys. An additional 5 children (three 3-year-old boys, one 4-year-old boy, and one 4-year-old girl) were tested but not included in the nal sample because they stopped playing before completing a minimum of 45 trials (see below). Data from one child (a 4-year-old girl) were excluded due to parental interference. Written informed consent was obtained from the parents or authorized representatives of all children. 2.2. Materials The Childrens Gambling Task, modeled after the Iowa Gambling Task (Bechara et al., 1994), involved one of two sets of laminated cards (18 cm 29 cm), with each set containing two decks of 50 cards. The back side of one deck was covered with vertical stripes (black and white), whereas the back side of the other deck was covered with black dots on a white background. The front sides of all cards were divided into white top half and a black bottom half. Black happy faces appeared on the top half of the card, and white sad faces appeared on the bottom half (see Fig. 1). The number of happy faces on each card indicated the number of rewards gained, and the number of sad faces indicated the number of rewards lost. During testing, the bottom half of the front side of each card was covered with a blue Post-It paper that the experimenter needed to remove in order to reveal any sad faces. This feature of the task was intended to clarify the presentation of information associated with each card, and to force children to attend rst to the reward information. Rewards consisted of mini M&M chocolate candies. When children gained rewards, candies were drawn from a plastic container kept by the experimenter and added to a 10-ml glass graduated cylinder positioned between the two decks of cards. When children lost rewards, candies were removed from the graduated cylinder and returned to the plastic container. In each set of cards, one deck of cards was advantageous over trials, and the other deck was disadvantageous. Consistently choosing from the advantageous deck resulted in a net gain of rewards, whereas consistently choosing from the disadvantageous deck resulted in a net loss. In both decks, the number of gains was constant across cards, but the number of losses was variable. Cards in the advantageous deck always provided a gain of one reward (i.e., they showed one happy

Fig. 1. Sample card face from the disadvantageous deck (Gain 2, Loss )4).

face) together with zero or one losses (with a net average of 5 candies gained per block of 10 cards). Cards in the disadvantageous deck always provided a gain of two rewards together with losses of 0, 4, 5, or 6 candies (with a net average of 5 candies lost per 10 cards). The order of cards in each deck was xed, and the gain/loss contingencies associated with each card in each deck (see Fig. 2) were proportional to those used by Bechara et al. (1994). The two sets of cards were identical except for the fact that in one set the advantageous deck was striped, whereas in the other set it was dotted. 2.3. Procedure The procedure involved six demonstration trials (involving three selections from each deck) and 50 test trials during which children were allowed to select from either deck. Each child was tested individually by a female experimenter in a single session lasting approximately

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

151

Fig. 2. Outcomes associated with each card in the advantageous and disadvantageous decks. Cards in the disadvantageous deck always provide a gain of 2 rewards, whereas cards in the advantageous deck always provide a gain of 1 reward.

25 min. Once the child was comfortable with the situation and the experimenter, he or she was seated at a small table. The two decks of cards were then placed face down in front of the child (separated by about 20 cm), and the graduated cylinder was placed between the decks. The experimenter sat beside the child and kept the container of rewards in front of her. Card set was counterbalanced, and the left right positioning of the decks was determined randomly for each child. The child was given one candy to eat to motivate him or her to play, and then introduced to the task. The child was told, Okay, in this game we get to put the M&Ms that you win in this tube here. Im going to give you 10 to play the game, and Ill show you how the game works and how you can win some more. Ten candies were counted out and deposited into the graduated cylinder. The experimenter then demonstrated how the task worked by selecting three cards consecutively from each deck, starting with three striped cards, which were disadvantageous for one card set and advantageous for the other. For example, when the striped deck was disadvantageous, on the third demonstration trial, the experimenter said, Look: there are two happy facesthat means that you win 2 M&Ms. Two candies were then taken from the container, placed demonstratively on top of the two happy faces to illustrate the one-to-one correspondence between number of faces and number of rewards gained, and then deposited into the graduated cylinder. The experimenter then proceeded to check for losses, pointing to the Post-It covering the bottom half of the card and saying, Okay, now we have to open this up and see if there are any sad faces. Oh look, there are 4 sad facesthat means that you lose 4 M&Ms so we have to give 4 back. The experimenter then removed 4 candies from the graduated cylinder, placed them on top of the four sad faces, and then deposited them in the plastic container. At this point, the experimenter said, We dont like the sad faces do we? Because we lose M&Ms, but we like the happy faces right? Because we get to win M&Ms! This procedure was repeated for all of the demonstration trials. Thus, although the child was not told that one of the decks was better than the other, he or she was forced to sample from each deck, and the xed sequence of gain/loss contingencies (see Fig. 2) ensured that a representative sample was obtained. After the six demonstration trials, the experimenter said,

Okay, now were ready to start playing the game. You get to choose whichever card you want to play with every time. You can play from the dots or the stripes or from both. You get to choose one card every time and you can pick as many cards as you want until I say STOP and then the game will be done. So remember, you want to make sure that you win as many M&Ms as possible! Lets see if we can ll this tube right up to the top with candy! Whatever you have in the tube at the end of the game you can eat or take home with you. Okay? Which card do you want to pick rst?

Test trials were administered exactly like the demonstration trials. Every time a card was turned, the experimenter announced the number of candies won, placed them on the happy faces, and then deposited them in the graduated cylinder. Then the Post-It was removed, the number of losses (if any) was announced, and the corresponding number of candies was removed from the graduated cylinder, placed on top of the sad faces, and then deposited into the plastic container. The child was not allowed to eat any of the candies he or she won until after the last test trial (trial 50), and the child was not told how many test trials there would be. Any child who stopped playing before completing at least 45 trials was replaced. Because there were only 50 cards in each deck and the rst 3 cards per deck were used during the demonstration trials, it was possible for the cards in one deck to run out. This happened in two cases, and because there were only two decks, children were not encouraged to select from the other deck. If children made a preponderance of disadvantageous choices, it was also possible for them to incur losses that they were unable to pay. When this happened, children were told, We dont even have that many to give back, so well just have to give back everything we have and play again.

3. Results The primary dependent measure was whether children made an advantageous or a disadvantageous choice on each trial. However, 14 children who completed at least 45 trials failed to complete all 50 trials, either because they selected all of the cards from a given deck, they were restless, or the experimenter miscounted. These children included: nine children who completed 49 trials (two 3-year-old boys, two 3-year-old girls, three 4-year-old boys, and two 4-year-old girls); two children who completed 48 trials (one 3-year-old girl and one

152

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

4-year-old girl); and one child each who completed 47, 46, and 45 trials (a 3-year-old girl, a 4-year-old boy, and a 3-year-old boy, respectively). To permit analysis of performance across blocks of 10 trials, including the fth block, proportion scores were used. In particular, we computed the proportion of advantageous choices per block minus the proportion of disadvantageous choices per block, which yielded dierence scores ranging from )1 to 1. Positive dierence scores indicate relatively advantageous performance, whereas negative scores indicate relatively disadvantageous performance. For analysis, data were transformed using an arcsine transformation, which is recommended when variances are proportional to mean scores (Winer, Brown, & Michels, 1991), but raw means are presented in the text and gures. Preliminary analyses conrmed that card set (i.e., whether the advantageous deck was striped or dotted, and hence whether or not the advantageous deck was demonstrated rst) had no eect on performance, so this variable was dropped from subsequent analyses. Difference scores were then analyzed using a 2 (age: 3 vs. 4 years) 2 (sex) 5 (blocks 15) mixed model analysis of variance (ANOVA). The only signicant eects to emerge from this analysis were a signicant main eect of age, F 1; 44 5:5, p < :05, and a signicant Age Block interaction, F 4; 176 2:74, p < :05 (GreenhouseGeisser adjusted). Means and standard errors for the interaction are presented in Fig. 3. Tests of simple eects indicated that scores increased across blocks for 4-year-olds, F 4; 92 3:09, p < :05 (GreenhouseGeisser adjusted), but not 3-year-olds, and that 4-year-olds made more advantageous choices than 3-year-olds on blocks 3, F 1; 44 5:49, p < :05, and 5, F 1; 44 7:73, p < :01. The simple eect of age on block 4 barely missed reaching the conventional level of signicance, F 1; 44 3:82, p < :06.

The t distribution was used to compare means for each age and block to the mean expected based on random responding (i.e., 0). Three-year-olds mean scores were signicantly lower than chance (i.e., more disadvantageous) for blocks 3 and 4 (ps < .05, onetailed). Four-year-olds means were signicantly higher than chance (i.e., more advantageous) for blocks 3 and 5 (p < :05 and p < :01, respectively, both one-tailed). There were no signicant eects involving sex, and in particular, the interaction of age, sex, and block failed to reach signicance, F 4; 176 < 1. However, given the a priori rationale for examining performance as a function of sex, we proceeded to make a series of individual comparisons of the means for boys vs. girls on blocks 3, 4, and 5, separately for each age group (see Fig. 4). Although this data analytic strategy is generally considered appropriate (e.g., see Winer et al., 1991), results should nonetheless be interpreted with caution due to the possibility of Type 1 error. One-way ANOVAs revealed statistical trends only for 3-year-olds for block 4, F 1; 23 2:38, p :14 (Cohens d :63) and block 5 F 1; 23 2:12, p :16 (Cohens d :59), indicating that on these blocks, 3-year-old boys tended to make more advantageous choices than 3-year-old girls. The eect sizes associated with both comparisons are considered medium eect sizes (Cohen, 1988). 3.1. Individual performance In addition to analyzing mean scores, we also examined the performance of individual children. The binomial theorem was used to determine whether each child made more advantageous or disadvantageous choices across all 5 blocks than would be expected based on random responding (assuming an a of .05). For children who completed 50 trials, this yielded a criterion of 32 or more choices of one type (criteria for other children were

Fig. 3. Mean (and standard error) proportion advantageous choices minus proportion disadvantageous choices as a function of age and block of 10 trials.

Fig. 4. Mean (and standard error) proportion advantageous choices minus proportion disadvantageous choices as a function of age, sex, and block of 10 trials.

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

153

adjusted accordingly). Among 3-year-olds, 10 children (5 boys and 5 girls) made more disadvantageous choices than would be expected by chance, whereas only 3 children (all boys) made more advantageous choices. Among 4-year-olds, 4 children (2 boys and 2 girls) made more disadvantageous choices, whereas 7 children (4 boys and 3 girls) made more advantageous choices. A second-order application of the binomial theorem was used to determine whether more children at each age responded nonrandomly than would be expected by chance based on an a of .05. It was determined that the number of 3-year-olds who made signicantly more disadvantageous choices than would be expected by chance was higher than would be expected by chance (p < :0001), as was the number of 4-year-olds (p < :05). In addition, signicantly more 4-year-olds made more advantageous choices than would be expected by chance (p < :0005).

4. Discussion The current study examined the development of affective decision-making during the preschool years using a simplied version of the Iowa Gambling Task (Bechara et al., 1994). On each of 50 trials, children were required to select a card from one of two decks, one of which was advantageous and one of which was disadvantageous. An age dierence emerged over trials. Analysis of mean scores indicated that 4-year-olds performance improved over trials, and that they made more advantageous choices than 3-year-olds during the last 3 blocks of trials (blocks 35). On two of these blocks, they made more advantageous choices than would be expected by chance. In contrast, 3-year-olds performance did not improve across trials, and on blocks 3 and 4, they made more disadvantageous choices than would be expected by chance. Analysis of individual childrens performance considered across all 50 trials generally corroborated these ndings, but also suggested that there were marked individual dierences in performance within each age group. That is, despite the age-related improvement in performance, there were some children at each age who made a majority of advantageous choices and some who made a majority of disadvantageous choices. In adults, performance on a modied version of the Iowa Gambling Task was found to be correlated with self-report measures of personality (Peters & Slovic, 2000). Individuals rated high in negative aectivity tended to avoid high losses, and individuals high in positive aectivity tended to seek high gains. Future research using the Childrens Gambling Task might usefully explore the correlates of individual dierences in childrens performance. As can be seen in Fig. 4, there was some evidence that the age dierence in performance on later trial blocks

was especially pronounced for girls, although this sex dierence was not statistically signicant and should be interpreted with caution. There was a lot of variability in childrens performance on this task, as the individual analyses make clear. While this variability seems to have obscured medium-sized eects of sex in the sample of 3-year-olds (on blocks 4 and 5), it may also have increased the likelihood of Type 1 error. Future research should address the issue of sex dierences using a larger sample size. Overall, then, the results indicate that aective decision-making develops rapidly during the preschool period. The age dierence, obtained using a new measure of hot EF, parallels well-established age dierences in cool EF (Zelazo & Mller, 2002), and extends earlier u work by Overman et al. (1996) using a simpler measure of hot EFobject reversal. Like object reversal, aective decision-making appears to depend importantly on the integrity of OFC. In the current study, many 3-year-olds performed like OFC patients insofar as they reliably made disadvantageous choices on a version of the Iowa Gambling Task (Bechara, this issue). The age-related improvement in performance between 3 and 4 years may reect continuing growth during the preschool years of neural systems involving OFC. Indeed, using a slightly more complex version of the Iowa Gambling Task than the one used here, Garon and Moore (this issue) found evidence of improvements in performance at even older agesat around 6 years of age. The nonsignicant sex dierence in this measure of hot EF is intriguing, but clearly needs to be studied further. Earlier work with monkeys (e.g., Clark & Goldman-Rakic, 1989) and younger children (Overman et al., 1996) indicates that OFC develops more rapidly in males than in females. Moreover, Reavis and Overman (2001) found that men outperform women on the Iowa Gambling Task. At the same time, however, there is evidence that women have larger OFC volume relative to amygdala volume, which has been taken to suggest that women may be better able to modulate basic limbic functions, such as those involved in aggression (Gur, Gunning-Dixon, Bilker, & Gur, 2002). Further work may reveal that the development of OFC follows a complicated developmental course, but it will be necessary in any case to characterize more fully the functional consequences of any anatomical dierences. More work is also needed to arrive at a more complete functional characterization of 3-year-olds diculty on the Childrens Gambling Task, and to better understand what changes between 3 and 4 years of age that allows children to choose advantageously despite the meretricious allure of the disadvantageous deck. The nding that 3-year-olds tended to respond systematically (i.e., disadvantageously) at both the group level and the level of individual children, renders unlikely the

154

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

possibility that they simply failed to understand the basic objectives of the task. This nding also indicates that 3-year-olds were motivated to obtain rewards. However, the task is relatively complex, and there are numerous possible reasons why 3-year-olds may have chosen disadvantageouslynone of which are mutually exclusive. Recall that in this task, the magnitudes of the gains are consistent across trials, whereas losses are variable and must be averaged over a number of trials. Moreover, information about gains is always presented rst. Consequently, among the many possibilities, 3-year-olds may choose disadvantageously because they are biased to attend to consistent information and/or are better able to process it; avoid attending to variable information and/or have diculty averaging information across trials; are biased to attend to immediate information and/or respond impulsively; are hypersensitive to rewards and/or hyposensitive to punishers; and/ or are more likely to take risks than older children and/ or have diculty shifting between motivational sets. Despite the plethora of possibilities, there are at least three broad perspectives on OFC function and/or the development of hot EF that might usefully be considered. 4.1. Flexible representation of the reinforcement value of stimuli One proposal, based on work by Rolls (e.g., 1999), is that age-related changes in childrens gambling reect increases in the functionality of OFC neurons underlying the exible reinterpretation of the reinforcement values of the two decks. The task is structured so that information about the gains associated with each deck is presented before information about losses, both across trials and within trials. That is, the rst two demonstration trials from each deck oer only gains, no losses (see Fig. 2), and on each trial, children rst receive the gains and then, after removing the Post-It, discover whether there are any losses. Under these circumstances, it seems reasonable to suppose that all children might initially represent the disadvantageous deck as more reinforcing than the advantageous deck. However, younger children may then have more diculty than older children reversing these representations. If this were the case, then the Childrens Gambling Task might in eect be seen as relatively complex version of an object reversal task perhaps one that is sensitive to age-related changes occurring relatively late during the preschool period (cf. Overman, this issue). 4.2. The somatic marker hypothesis Another possibility is that age-related changes in childrens gambling reect development of the ability to

use somatic markers (e.g., Damasio, 1994). On this account, OFC is important for learning to associate somatic states with particular situations. Re-enactment of the somatic states (either directly or indirectly) can serve to bias future-oriented decision-making in an adaptive fashion. Consistent with this proposal, Bechara et al. (1997) found that after about 1020 trials on the Iowa Gambling Task, healthy controls began to exhibit anticipatory skin conductance responses (SCRs) prior to choosing from the disadvantageous decks, even though their verbal reports indicated that they did not yet know which decks were disadvantageous. In contrast, OFC patients never exhibited anticipatory SCRs. The anticipatory SCRs could serve as somatic markers eventually steering healthy individuals away from the disadvantageous decks. Due to functional immaturity of OFC, 3-year-olds may fail to develop somatic markers associated with the disadvantageous decks. This possibility is currently being investigated in our lab. 4.3. Development of reection and higher-order rules A nal proposal focuses on the complex structure of the Childrens Gambling Task. Recall that the gains and losses are put into conict in this task: consideration of gains alone favors the disadvantageous deck, whereas consideration of gains in combination with losses favors the advantageous deck. Integrating gain and losses arguably requires the consideration of two dimensions in contradistinction. According to one theory (e.g., Zelazo & Frye, 1998), there are systematic age-related changes in childrens ability to formulate complex, higher-order rules for operating on (and selecting among) lower-order rules, and the formulation of higher-order rules requires a degree of conscious reection on lower-order rules. For example, research using measures of cool EF indicates that 3-year-olds have diculty formulating a higher-order rule that integrates two incompatible pairs of rules (e.g., If playing the color game, then red ones here and blue ones there, but if playing the shape game, then cars here and owers there; Zelazo et al., 1996). As a result, in the absence of integration, the particular rule pair that controls behavior is determined by relatively local, immediate associations, resulting in dissociations between reported knowledge and behavior. Although such dissociations were not observed in the current study, Bechara et al. (1997) found that toward the end of the Iowa Gambling Task, OFC patients continued to choose from the disadvantageous decks even after they were able to indicate verbally that these decks were disadvantageous. On this account, the Childrens Gambling Task may correspond in complexity to the cool EF tasks that are sensitive to age-related changes between 3 and 4 years (e.g., the DCCS; Zelazo et al.,

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157

155

1996). Three-year-olds learn an initial discrimination between the decks, such as striped has high gains, dotted has low gains, and then have diculty coordinating this discrimination with emerging evidence regarding losses (striped has high losses, dotted has low losses). Four-year-olds may be better able to reect on these conicting discriminations and formulate a higher-order rule that allows them to appreciate net gains and/or select the appropriate discrimination on which to base their behavior. In general, however, it might be noted that task complexity, or rather, the complexity of the cognitive processes that a task requires, is likely an important aspect of EF that may or may not be orthogonal to the hotcool dimension (e.g., Dias, Robbins, & Roberts, 1996; Stuss et al., 1999; Wise, Murray, & Gerfen, 1996). From this perspective, a task such as object reversal may be a relatively simple measure of hot EF, whereas the Childrens Gambling Task is relatively complex. Similarly, delayed response may be a relatively simple measure of cool EF, whereas the DCCS is relatively complex. The importance of complexity has long been recognized in the developmental literature, and taking a developmental perspective may help reveal its implications for understanding brain function (cf. Luria, 1966). 4.4. Conclusion Although EF can be understood in domain-general terms, a distinction may be made between the relatively hot aective aspects of EF associated with OFC and the more purely cognitive, cool aspects associated with DL-PFC (Zelazo & Mller, 2002; cf. Metcalfe & u Mischel, 1999; Miller & Cohen, 2001). On this account, whereas cool EF is more likely to be elicited by relatively abstract, decontextualized problems, hot EF is required for problems that involve the regulation of aect and motivation (i.e., regulation of basic limbic system functions). The term EF has traditionally been used only to refer to cool EF, and there is some lingering controversy about whether hot EF is really EF at all. However, the characterization of hot EF in contradistinction to cool EF is consistent with several recent proposals regarding the function of OFC. For example, based on single-cell recordings of OFC neurons together with neuroimaging data and evidence that OFC damage impairs performance on simple tests of object reversal and extinction, Rolls (e.g., Rolls, 1999, 2000; this issue) suggests that OFC is required for the exible representation of the reinforcement value of stimuli. A rather dierent theory has been proposed by Damasio (e.g., 1994; see also Bechara, this issue). According to this theory, the somatic marker theory, OFC is required for processing learned associations

between aective reactions and specic scenarios, and this processing plays a crucial but often overlooked role in decision-making. Despite their dierences, however, both approaches capture the fact that the control of thought and action depends on dierent cortical systems, depending on whether or not it occurs in motivationally signicant contexts. The present study serves as a rst step toward understanding of the development of aective decisionmaking during the preschool years, and, as such, it has the potential to add to the literature on the development of hot EF and the growth of neural systems involving OFC. The results indicate that hot EF develops rapidly between the ages of 3 and 4 years, in parallel with corresponding changes in cool EF. This correspondence is perhaps to be expected given that, in the normal case, OFC and dorsolateral prefrontal cortex are parts of a single coordinated system and probably work togethereven in a single situation. Thus, it seems likely that decision-making is routinely inuenced by aective reactions (e.g., Damasio, 1994) and the representation of reward value (e.g., Rolls, 1999). Conversely, it seems likely that a successful approach to solving some aective problems is to reconceptualize the problem in relatively neutral, decontextualized terms, and try to solve it using cool EF. Although it is useful for the purpose of analysis to pull hot EF and cool EF apart, it will ultimately be important to put them back together again.

Acknowledgments This research was supported by a grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada to P.D. Zelazo. The authors thank Dana Liebermann for her help in preparing this article, and Keith Happaney for helpful comments on a previous draft.

References
Bechara, A., Damasio, A., Damasio, H., & Anderson, S. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition, 50, 715. Bechara, A., Damasio, H., Tranel, D., & Damasio, A. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275, 12931295. Bechara, A., Dolan, S., Denburg, N., Hindes, A., Anderson, S. W., & Nathan, P. E. (2001). Decision-making decits, linked to a dysfunctional ventromedial prefrontal cortex, revealed in alcohol and stimulant abusers. Neuropsychologia, 39, 376389. Bialystok, E. (1999). Cognitive complexity and attentional control in the bilingual mind. Child Development, 70, 636644. Carlson, S. M., & Moses, L. J. (2001). Individual dierences in inhibitory control and theory of mind. Child Development, 72, 10321053.

156

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157 Overman, W. H., Bachevalier, J., Schuhmann, E., & Ryan, P. (1996). Cognitive gender dierences in very young children parallel biologically based cognitive gender dierences in monkeys. Behavioral Neuroscience, 110, 673684. Perner, J., Stummer, S., & Lang, B. (1999). Executive functions and theory of mind: Cognitive complexity or functional dependence. In P. D. Zelazo, J. W. Astington, & D. R. Olson (Eds.), Developing theories of intention: Social understanding and self-control (pp. 133 152). Mahwah, NJ: Erlbaum. Peters, E., & Slovic, P. (2000). The springs of aection: Aective and analytical information processing in choice. Personality and Social Psychology Bulletin, 26, 14651475. Petry, N. M., Bickel, W. K., & Arnett, M. (1998). Shortened time horizons and insensitivity to future consequences in heroin addicts. Addiction, 93, 729738. Pfeerbaum, A., Mathalon, D. H., Sullivan, E. V., Rawles, J. M., Zipursky, R. B., & Lim, K. O. (1994). A quantitative magnetic resonance imaging study of changes in brain morphology from infancy to late adulthood. Archives of Neurology, 51, 874887. Reavis, R., & Overman, W. H. (2001). Adult sex dierences on a decision-making task previously shown to depend on the orbital prefrontal cortex. Behavioral Neuroscience, 115, 196 206. Rogers, R. D., Everitt, B. J., Baldacchino, A., Blackshaw, A. J., Swainson, R., Wynne, K., Baker, N. B., Hunter, J., Carthy, T., Booker, E., London, M., Deakin, J. F. W., Sahakian, B. J., & Tobbins, T. W. (1999). Dissociable decits in the decision-making cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and Tryptophan-depleted normal volunteers: Evidence for monoaminergic mechanisms. Neuropsychopharmacology, 20, 322339. Rolls, E. T. (1999). The brain and emotion. Oxford: Oxford University Press. Rolls, E. T. (2000). The orbitofrontal cortex and reward. Cerebral Cortex (Special Issue: The mysterious orbitofrontal cortex), 10, 284294. Rubia, K., Overmeyer, S., Taylor, E., Brammer, M., Williams, S. C., Simmons, A., Andrew, C., & Bullmore, E. T. (2000). Functional frontalisation with age: Mapping neurodevelopmental trajectories with fMRI. Neuroscience and Biobehavioral Reviews, 24, 1319. Stuss, D. T., Toth, J. P., Franchi, D., Alexander, M. P., Tipper, S., & Craik, F. I. M. (1999). Dissociation of attentional processes in patients with focal frontal and posterior lesions. Neuropsychologia, 37, 10051027. Thompson, P. M., Giedd, J. N., Woods, R. P., MacDonald, D., Evans, A. C., & Toga, A. W. (2000). Growth patterns in the developing brain detected by using continuum mechanical tensor maps. Nature, 404, 190192. Travis, F. (1998). Cortical and cognitive development in 4th, 8th and 12th grade students: The contribution of speed of processing and executive functioning to cognitive development. Biological Psychology, 48, 3756. Winer, B. J., Brown, D. R., & Michels, K. M. (1991). Statistical principles in experimental design (3rd ed.). New York: McGrawHill. Wise, S. P., Murray, E. A., & Gerfen, C. R. (1996). The frontal cortexbasal ganglia system in primates. Critical Reviews in Neurobiology, 10, 317356. Yakolev, P. I., & Lecours, A. R. (1967). The myelogenetic cycles of regional maturation of the brain. In A. Minkowski (Ed.), Regional development of the brain in early life (pp. 370). Oxford, England: Blackwell. Zelazo, P. D., & Frye, D. (1998). II. Cognitive complexity and control: The development of executive function. Current Directions in Psychological Science, 7, 121126.

Cavedini, P., Riboldi, G., Keller, R., DAnnucci, A., & Bellodi, L. (2002). Frontal lobe dysfunction in pathological gambling patients. Biological Psychiatry, 51, 334341. Clark, A. S., & Goldman-Rakic, P. S. (1989). Gonadal hormones inuence the emergence of cortical function in nonhuman primates. Behavioral Neuroscience, 103, 12871295. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Mahwah, NJ: Erlbaum. Damasio, A. (1994). Descartes error: Emotion, reason, and the human brain. New York: Grosset/Putnam. Dias, R., Robbins, T. W., & Roberts, A. C. (1996). Dissociation in prefrontal cortex of aective and attentional shifts. Nature, 380, 6972. Elliott, R., Frith, C. D., & Dolan, R. J. (1997). Dierential neural response to positive and negative feedback in planning and guessing tasks. Neuropsychologia, 35, 13951404. Frye, D., Zelazo, P. D., & Palfai, T. (1995). Theory of mind and rulebased reasoning. Cognitive Development, 10, 483527. Goldman, P. S., Crawford, H. T., Stokes, L. P., Galkin, T. W., & Rosvold, H. E. (1974). Sex-dependent behavioral eects of cerebral cortical lesions in the developing rhesus monkey. Science, 186, 540 542. Grant, S., Contoreggi, C., & London, E. D. (2000). Drug abusers show impaired performance in a laboratory test of decision making. Neuropsychologia, 38, 11801187. Huttenlocher, P. R. (1979). Synaptic density in human frontal cortex developmental changes and eects of aging. Brain Research, 163, 195205. Huttenlocher, P. R. (1990). Morphometric study of human cerebral cortex development. Neuropsychologia, 28, 517527. Iverson, S. D., & Mishkin, M. (1970). Perseverative interference in monkeys following selective lesions of the inferior prefrontal convexity. Experimental Brain Research, 11, 376386. Jacques, S., Zelazo, P. D., Kirkham, N., & Semcesen, T. (1999). Rule selection versus rule execution in preschoolers: An error-detection approach. Developmental Psychology, 35, 770780. Kirkham, N., Cruess, L., & Diamond, A. (2003). Helping children apply their knowledge to their behavior in a dimension-switching task. Developmental Science, 6, 449476. Luciana, M., & Nelson, C. A. (1998). The functional emergence of prefrontally-guided working memory systems in four- to eightyear-old children. Neuropsychologia, 36, 273293. Luria, A. R. (1966). Higher cortical functions in man (2nd ed.). New York: Basic Books (Original work published in 1962). Manes, F., Sahakian, B., Clark, L., Rogers, R., Antoun, N., Aitken, M., & Robbins, T. (2002). Decision-making processes following damage to prefrontal cortex. Brain, 125, 624639. Mazas, C. A., Finn, P. R., & Steinmetz, J. E. (2000). Decision-making biases, antisocial personality, and early-onset alcoholism. Alcoholism: Clinical and Experimental Research, 24, 10361040. Metcalfe, J., & Mischel, W. (1999). A Hot/Cool-System analysis of delay of gratication: Dynamics of willpower. Psychological Review, 106, 319. Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167 202. Monterosso, J., Ehrman, R., Napier, K. L., OBrien, C. P., & Childress, A. R. (2001). Three decision-making tasks in cocainedependent patients: Do they measure the same construct? Addiction, 96, 18251837. Munakata, Y., & Yerys, B. E. (2001). All together now: When dissociations between knowledge and action disappear. Psychological Science, 12, 335337. Orzhekhovskaya, N. S. (1981). Fronto-striatal relationships in primate ontogeny. Neuroscience and Behavioral Physiology, 11, 379385.

A. Kerr, P.D. Zelazo / Brain and Cognition 55 (2004) 148157 Zelazo, P. D., Frye, D., & Rapus, T. (1996). An age-related dissociation between knowing rules and using them. Cognitive Development, 11, 3763.

157

Zelazo, P. D., & Mller, U. (2002). Executive function in typical and u atypical development. In U. Goswami (Ed.), Handbook of childhood cognitive development (pp. 445469). Oxford: Blackwell.

You might also like