108 Review TRENDS in Cognitive Sciences Vol.9 No.3 March 2005 childhood into early adolescence during performance of a Box 3.Questions for future research Go/No-go task.There was a developmental shift in patterns of cortical activation from diffuse to focal activity Do the biological substrates of learning and development differ? similar to those reported in large cross-sectional studies On what timescale can we see neuroanatomical and physiological changes with learning and development? [36].Regions uncorrelated with task performance were What neural changes coincide with enhanced learning during recruited less with age,whereas a region in the ventral development that could hinder learning with maturity? prefrontal cortex that correlated with performance(speed How can we move beyond claims of causality between coinciden- and accuracy)in this study and prior studies [28,42,45,46] tal changes in brain and behavioral development with converging showed enhanced recruitment.This change in pattern of methods? Can typical human development provide important clues on the activity was associated with enhanced performance of the type and timing of interventions with atypical development? cognitive control task.Activation in earlier developing regions such as the primary motor cortex remained unchanged.A parallel cross-sectional analysis based on behavioral and cognitive interventions on developmental a second group of children,the same ages as the disorders such as dyslexia 51,52]and attention deficit- longitudinal group,showed less specific results,support- hyperactivity disorder [53]. ing the importance of longitudinal studies in evaluating Another future direction of developmental imaging cortical changes with age. studies will be in the combined use of complementary These studies,together with the longitudinal MRI- imaging methods.For example,a method not mentioned based morphometry studies [6],suggest differential in this review,but one that has been used to address developmental trajectories for sensorimotor relative to questions about cognitive and brain development,is association areas such as the prefrontal cortex.Both the electrophysiology.This technique records brain electrical imaging-based neuroanatomical studies and the func- activity that can be time-locked to presentation of a tional studies highlight the importance of examining stimulus or to the production of a response (i.e.event- changes in ability within individuals over time.As such, related potentials or ERPs).These electrophysiological this work can begin to delineate processes specific to measures have millisecond temporal resolution that can learning or development. provide important information on the developmental changes in the temporal characteristics of neural and Cortical organization with learning cognitive processes beyond that provided by fMRI [541. The importance of tracking cortical changes in individuals Clearly,the combination of these two techniques,together over time is perhaps most evident in the area of learning. with other methods such as DTI,will strongly enhance our Using repeated scanning of the same individuals,Karni ability to understand neural and cognitive development and others(e.g.[47,48])have shown rapid learning effects and constrain current developmental theories.Investi- in primary motor cortex of adults during motor sequence gators have begun to move in that direction with parallel learning that were apparent within a single imaging fMRI and DTI studies in adults and children [25.261. session,but that increased over weeks of training.This use of fMRI to trace learning-related changes in cortical Conclusions areas is currently being used by others in adults Findings from both cross-sectional and longitudinal (e.g.[49,50]).Across these studies,the pattern of results imaging studies of late childhood and adolescence show shows that activity in task-relevant regions becomes that brain regions associated with more basic functions increasingly enhanced with training,whereas task-irre- such as motor and sensory processes mature first,followed levant regions become less activated over time [50].This by association areas involved in top-down control of pattern of change during adult learning studies mimics thoughts and action.This pattern of development is the observed changes in cross-sectional [36]and longi- paralleled by a shift from diffuse to more focal recruitment tudinal [37]developmental studies.The findings highlight of cortical regions with learning and cognitive develop- the importance of examining and distinguishing between ment.Fine-tuning of cortical systems occurs with pro- contributions from experience and learning compared to tracted refinement of association cortex (e.g.prefrontal those associated more with maturational development,in cortex)relative to sensorimotor cortex.These findings driving the cortical patterns of activation observed. seem to parallel changes observed over shorter time periods in studies of adult learning.The reported shift in Future directions cortical architecture and function is presumably an Future research will no doubt take advantage of the experience-driven maturational process that reflects ability to image children multiple times with fMRI over fine-tuning of neural systems with experience and devel- the course of learning to delineate developmental and opment,but future work delineating how learning during experience-based processes in cortical activity(see Box 3). development affects this pattern is needed.These imaging For example,to determine whether the immature brain studies map onto findings from animal and human after extended practice engages in the same neural postmortem studies indicating that pruning and elimin- processes as the mature brain,we could compare brain ation of connections,in combination with strengthening of activity in the mature system with brain activity in the relevant ones,occur during this time period,and illustrate immature system both before and after extended experi- the subtle interplay between neuroanatomical and phys- ence.Investigators are already beginning to take advan- iological changes in neural circuitry and cognitive tage of this approach in investigating the impact of maturation. www.sciencedirect.comchildhood into early adolescence during performance of a Go/No-go task. There was a developmental shift in patterns of cortical activation from diffuse to focal activity similar to those reported in large cross-sectional studies [36]. Regions uncorrelated with task performance were recruited less with age, whereas a region in the ventral prefrontal cortex that correlated with performance (speed and accuracy) in this study and prior studies [28,42,45,46] showed enhanced recruitment. This change in pattern of activity was associated with enhanced performance of the cognitive control task. Activation in earlier developing regions such as the primary motor cortex remained unchanged. A parallel cross-sectional analysis based on a second group of children, the same ages as the longitudinal group, showed less specific results, support￾ing the importance of longitudinal studies in evaluating cortical changes with age. These studies, together with the longitudinal MRI￾based morphometry studies [6], suggest differential developmental trajectories for sensorimotor relative to association areas such as the prefrontal cortex. Both the imaging-based neuroanatomical studies and the func￾tional studies highlight the importance of examining changes in ability within individuals over time. As such, this work can begin to delineate processes specific to learning or development. Cortical organization with learning The importance of tracking cortical changes in individuals over time is perhaps most evident in the area of learning. Using repeated scanning of the same individuals, Karni and others (e.g. [47,48]) have shown rapid learning effects in primary motor cortex of adults during motor sequence learning that were apparent within a single imaging session, but that increased over weeks of training. This use of fMRI to trace learning-related changes in cortical areas is currently being used by others in adults (e.g. [49,50]). Across these studies, the pattern of results shows that activity in task-relevant regions becomes increasingly enhanced with training, whereas task-irre￾levant regions become less activated over time [50]. This pattern of change during adult learning studies mimics the observed changes in cross-sectional [36] and longi￾tudinal [37] developmental studies. The findings highlight the importance of examining and distinguishing between contributions from experience and learning compared to those associated more with maturational development, in driving the cortical patterns of activation observed. Future directions Future research will no doubt take advantage of the ability to image children multiple times with fMRI over the course of learning to delineate developmental and experience-based processes in cortical activity (see Box 3). For example, to determine whether the immature brain after extended practice engages in the same neural processes as the mature brain, we could compare brain activity in the mature system with brain activity in the immature system both before and after extended experi￾ence. Investigators are already beginning to take advan￾tage of this approach in investigating the impact of behavioral and cognitive interventions on developmental disorders such as dyslexia [51,52] and attention deficit– hyperactivity disorder [53]. Another future direction of developmental imaging studies will be in the combined use of complementary imaging methods. For example, a method not mentioned in this review, but one that has been used to address questions about cognitive and brain development, is electrophysiology. This technique records brain electrical activity that can be time-locked to presentation of a stimulus or to the production of a response (i.e. event￾related potentials or ERPs). These electrophysiological measures have millisecond temporal resolution that can provide important information on the developmental changes in the temporal characteristics of neural and cognitive processes beyond that provided by fMRI [54]. Clearly, the combination of these two techniques, together with other methods such as DTI, will strongly enhance our ability to understand neural and cognitive development and constrain current developmental theories. Investi￾gators have begun to move in that direction with parallel fMRI and DTI studies in adults and children [25,26]. Conclusions Findings from both cross-sectional and longitudinal imaging studies of late childhood and adolescence show that brain regions associated with more basic functions such as motor and sensory processes mature first, followed by association areas involved in top-down control of thoughts and action. This pattern of development is paralleled by a shift from diffuse to more focal recruitment of cortical regions with learning and cognitive develop￾ment. Fine-tuning of cortical systems occurs with pro￾tracted refinement of association cortex (e.g. prefrontal cortex) relative to sensorimotor cortex. These findings seem to parallel changes observed over shorter time periods in studies of adult learning. The reported shift in cortical architecture and function is presumably an experience-driven maturational process that reflects fine-tuning of neural systems with experience and devel￾opment, but future work delineating how learning during development affects this pattern is needed. These imaging studies map onto findings from animal and human postmortem studies indicating that pruning and elimin￾ation of connections, in combination with strengthening of relevant ones, occur during this time period, and illustrate the subtle interplay between neuroanatomical and phys￾iological changes in neural circuitry and cognitive maturation. Box 3. Questions for future research † Do the biological substrates of learning and development differ? † On what timescale can we see neuroanatomical and physiological changes with learning and development? † What neural changes coincide with enhanced learning during development that could hinder learning with maturity? † How can we move beyond claims of causality between coinciden￾tal changes in brain and behavioral development with converging methods? † Can typical human development provide important clues on the type and timing of interventions with atypical development? 108 Review TRENDS in Cognitive Sciences Vol.9 No.3 March 2005 www.sciencedirect.com