A Brain-Based Account of the Development of Rule Use in Childhood

The ability to follow explicit rules improves dramatically during the course of childhood, but relatively little is known about the changes in brain structure and function that underlie this behavioral improvement. Drawing from neuroscientific studies in human adults and other animals, as well as from an emerging literature in developmental cognitive neuroscience, we propose a brainbased account of the development of rule use in childhood. This account focuses on four types of rules represented in different parts of the prefrontal cortex: simple rules for reversing stimulus–reward associations, pairs of conditional stimulus–response rules (both univalent and bivalent), and higher-order stimulus–response rules for selecting among task sets. It is hypothesized that the pattern of developmental changes in rule use reflects the different rates of development of specific regions within the prefrontal cortex. KEYWORDS—cognitive control; executive function; task set; complexity; neuroimaging The use of explicit rules to control behavior is one of the hallmarks of executive function—the conscious control of thought, action, and emotion—and it develops gradually over the course of childhood. As children get older, they typically become increasingly adept at using explicit rules to solve problems, play games, and interact with others. Behavioral research has now established that the development of rule use follows a reliable pattern: Children first acquire the ability to use a single rule, then the ability to switch flexibly between two rules, and then the ability to switch flexibly between two incompatible pairs of rules (Zelazo, Muller, Frye, & Marcovitch, 2003). The neural correlates of these developmental changes are not yet well understood, mainly because until recently there were no measures of brain function suitable for use in young children. But given what is known about the neural basis of rule use in human adults and nonhuman primates, together with what is known about the growth of the prefrontal cortex (PFC) in childhood and adolescence, we propose a brain-based account of the development of rule use in childhood. Although this brief overview focuses on the role of the PFC, other brain regions are also needed to represent rules and implement them flexibly (Bunge, 2004). THE PREFRONTAL CORTEX The PFC is a large expanse of cortex at the front of the brain that has been closely associated with rule use in human adults and nonhuman animals. Through its interactions with numerous other brain regions, the PFC processes information about an individual’s current context and about his or her goals and motivations. The PFC also plays an important role in retrieving rules for governing behavior in the current context—for example, the rule that if one is sitting in a classroom, one should raise one’s hand to be excused. The involvement of the PFC is particularly important when rules are not yet overlearned or automatic and when ad hoc rules must be formulated to govern behavior in an unfamiliar setting. Indeed, patients with damage to the PFC have particular difficulty planning and controlling their behavior when faced with novel challenges. The PFC consists of a number of subregions, including the orbitofrontal, ventrolateral, dorsolateral, and rostrolateral prefrontal cortices (Fig. 1). Evidence that these regions have different functions comes from several sources. First, structural investigations reveal that these regions differ in their cellular composition and in their connections to other brain regions. Second, neuropsychological studies in humans and focal-lesion Address correspondence to Silvia Bunge, Department of Psychology and Center for Mind and Brain, University of California at Davis, 267 Cousteau Place, Davis, CA, 95616, USA, e-mail: sbunge@ucdavis. edu; or Philip David Zelazo, Department of Psychology, University of Toronto, Toronto, Ontario, M5S 3G3, Canada; e-mail: zelazo@psych. utoronto.ca. CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 118 Volume 15—Number 3 Copyright r 2006 Association for Psychological Science (selective removal) experiments in nonhuman primates reveal different cognitive deficits following damage to each of these regions. Finally, brain-imaging experiments in school-aged children and adults, using techniques like functional magnetic resonance imaging (fMRI), show that these regions are engaged differently for different kinds of cognitive demands. PREFRONTAL REGIONS AND RULE COMPLEXITY A growing body of evidence indicates that the different regions of the PFC are involved in representing rules at different levels of complexity—from single rules for new stimulus–reward associations (orbitofrontal cortex), to sets of conditional rules (ventrolateral and dorsolateral PFC), to task sets (rostrolateral PFC; see Fig. 1). Orbitofrontal Cortex: Stimulus–Reward Associations From infancy onward, individuals learn that some stimuli (such as a high chair) are associated with rewarding experiences, and others (such as a pediatrician’s office) with experiences that are perceived to be detrimental or less rewarding. These associations need not be represented explicitly (i.e., an individual need not be conscious of them). However, if there is a sudden change in the outcome associated with a stimulus, it may be helpful to represent the new stimulus–reward association explicitly. In humans, the ability to reverse stimulus–reward associations improves dramatically during the first 3 years of life (Overman & Bachevalier, 1999). Although the hypothesis has yet to be explicitly tested, this improvement is likely to be related to structural changes in the orbitofrontal cortex. The ability to reverse stimulus–reward associations is impaired by damage to the orbitofrontal cortex in both human adults and nonhuman primates (e.g., Rolls, Hornak, Wade, & McGrath 1994; Dias, Robbins, & Roberts, 1996). Electrophysiological findings in nonhuman primates indicate that orbitofrontal neurons encode the values of rewards associated with specific stimuli and strongly suggest that this reward information is passed on to dorsolateral prefrontal neurons, which then select a response on the basis of this information (Wallis & Miller, 2003). In contrast to its crucial role in representing single stimulus–reward associations, the orbitofrontal cortex does not appear to have to update rules that do not explicitly assign a value (amount of reward or punishment) to a stimulus (e.g., one might learn that an apple is a rewarding stimulus whereas broccoli is an unrewarding stimulus), as in the conditional stimulus–response rules discussed below (Dias et al., 1996). The Ventrolateral and Dorsolateral PFC: Sets of Conditional Rules The simplest set of conditional rules consists of a pair of univalent stimulus–response associations—rules in which each stimulus is associated with a different response. For example, drivers learn to associate a green light with driving and a red light with stopping. More complex rule sets involve bivalent conditional rules, or rules in which two different responses may be associated with a single stimulus, depending on the context in which the stimulus occurs. Neuroscientific studies implicate both the ventrolateral and dorsolateral PFC in the representation of sets of conditional rules, although the precise roles of these regions may differ (for a review, see Bunge, 2004). Lesion studies in nonhuman primates show that the ventrolateral PFC is critical for learning pairs of univalent and bivalent conditional rules. fMRI studies in humans show that both the ventrolateral and dorsolateral PFC are active during the maintenance (i.e, keeping in mind) of sets of conditional rules, and that they are sensitive to rule complexity. Specifically, these regions are more active when participants consider stimuli that have been associated with different responses depending on the rule relevant to the current context (bivalent stimuli) than when they consider stimuli with fixed responses (univalent stimuli; Crone, Wendelken, Donohue, & Bunge, 2006). These regions are also more active for more abFig. 1. A hierarchical model of rule representation in the lateral prefrontal cortex (PFC). A lateral view of the human brain is depicted at the top of the figure, with regions of the PFC identified by different Brodmann areas (BA): the orbitofrontal cortex (BA 11), the ventrolateral PFC (BA 44, 45, 47), the dorsolateral PFC (BA 9, 46), and the rostrolateral PFC (BA 10). Rule structures are shown below along a continuum from simple to complex, in colors corresponding to the brain regions that represent them. S 5 stimulus; check 5 reward; cross 5 nonreward; R 5 response; C 5 context, or task set. Brackets indicate a bivalent rule that is currently being ignored. The lateral PFC regions are shown in various shades of blue, with darker shades indicating regions that represent more complex rules. The orbitofrontal cortex is shown in red, to indicate that the rules represented in this region are qualitatively different from the types of rules represented in the lateral PFC, in that the rules provide information about the value of a stimulus. (Differences between the medial and the lateral orbitofrontal cortex with respect to the representation of rewards and punishments are not depicted here.) Volume 15—Number 3 119 Silvia A. Bunge and Philip David Zelazo