Introduction
Relative to research on working memory, attention, or inhibition, a concerted effort devoted toward understanding “conflict monitoring” as a neurocognitive concept and phenomenon in its own right is a recent endeavor that only emerged in the past decade. Early behavioral findings from the 1970s foreshadowed the notion that, the detection of “conflict” in the processes of cognitive operations (e.g., such as error detection) plays an important role in information processing (e.g., Higgins & Angel,
1970; Rabbitt,
1966). Contemporary cognitive monitoring research over the past 10 years has established that monitoring mechanisms—in terms of outcome, processing or response conflict—are important adaptive signaling functions that could serve the purpose of online dynamical reconfigurations of the neurocognitive system in order for it to be sufficiently flexible and goal-oriented (cf. Ridderinkhof & van den Wildenberg,
2005).
Mechanisms subserving conflict monitoring operate in a variety of related functions, such as outcome monitoring, error detection, or conflict resolution. Emerging evidence suggests that various regions of the prefrontal cortex (PFC) as well as the anterior cingulate cortex (ACC) are implicated in processes of outcome monitoring, error processing, reward-based learning, and conflict resolution (e.g., Cohen, Aston-Jones, & Gilzenrat,
2004; Egner & Hirsh,
2005; Ridderinkhof, Ullsperger, Crone, & Nieuwenhuis,
2004; Ullsperger & von Cramon,
2004; Yeung, Botvinick, & Cohen,
2004). According to some accounts, activities in ACC send signals about detected conflicts (Botvinick & Cohen, 2004) or error likelihoods (Brown & Braver,
2005) to the PFC in order to generate stronger, compensatory PFC representations to support or direct task-relevant processing. In a related vein, PFC cognitive control mechanisms may bias processing to enhance target-relevant information, thus resolving conflicts. For instance, activations in dorsal lateral PFC have also been shown to display increased functional connectivity to perceptual regions (e.g., fusiform face area when face stimuli were used) under high conflict conditions (Egner & Hirsh,
2005). Furthermore, empirical evidence from animal models (e.g., Schultz,
2002) and human studies (e.g., Berns, McClure, Pagnoni, Montague,
2001; Dolan et al.,
1995; Schott et al.,
2004; see also Marschner et al.,
2005) as well as various neurocomputational models (e.g., Frank, Seeberger, & O’Reilly,
2004; Holroyd & Coles,
2002; Yeung et al.,
2004) all point to the involvements of frontal and midbrain dopaminergic systems in modulating cognitive monitoring mechanisms (see also Montague, Hyman, & Cohen,
2004, for review).
Whereas evidence of neural correlates of different aspects of conflict monitoring as well as theories of neuromodulation of conflict monitoring are gradually consolidating (e.g., Botvinick, Braver, Barch, Carter, & Cohen,
2001; Brown & Braver,
2005; Holroyd & Coles,
2002; Yeung et al.,
2004; see also Botvinick, Cohen, & Carter,
2004; Montague et al.,
2004, for reviews), research on the maturation (e.g., Davies, Segalowitz, & Gavin,
2004; Ladouceur, Dahl, & Carter,
2007; Rueda, Posner, Rothbart et al., 2004; Santesso, Segalowitz, & Schmidt,
2006) and senescence (Fernandez-Duque & Black,
2006; Mathewson, Dywan, & Segalowitz,
2005) of conflict monitoring mechanisms has only started. The commonly used conflict monitoring tasks (e.g., the Stroop, Flanker, and Go/NoGo tasks) have been applied to address questions of child cognitive development or cognitive aging independently. Thus, so far a synopsis of age-related differences in conflict monitoring and their neural correlates across the lifespan needs to pieced together from different sets of evidence that have been separately collected at the two ends of the lifespan. Children’s performance levels on the so-called “conflict tasks” (e.g., flanker and Go/NoGo tasks) are below the level of younger adults (e.g., Davies et al.,
2004; Rueda et al.,
2004; Williams, Ponesse, Schachar, Logan, & Tannock,
1999). Similarly, older adults’ performances also compare unfavorably to those of younger adults (e.g., Mathalon et al.,
2003; Mathewson et al.,
2005; West,
2004; West & Moore,
2005). Initial evidence also suggests that children (Davies et al.,
2004) and older adults (Nieuwenhuis et al.,
2002) are less able in detecting errors than younger adults. However, there is a lack of direct comparisons of children’s performance with that of older adults. Hence, thus far, little is known about the similarities and differences between the maturation and senescence of conflict monitoring.
As for neural correlates of conflict monitoring, recent findings from cognitive neuroscience of child development (e.g., Sowell, Thompson, Holmes, Jernigan, & Toga,
2003) and aging (e.g., Raz et al.,
2005) reveal that PFC and ACC mature late during child development and decline relatively early during aging, relative to other brain areas, such as the medial temporal structures. Specifically, the maturational gradients of PFC and ACC are very protracted, with continuing development until late adolescence (Sowell et al.,
2003). At the neurofunctional level, the magnitude of event-related potentials associated with incorrect responses (i.e., error-related negativity) has been found to increase from childhood to adulthood (Davies et al.,
2004) and to decrease during aging (Nieuwenhuis et al.,
2002). At the neurochemical level, age-related changes in dopaminergic mechanisms during child development and aging that underlie various aspects of cognitive processing are also well established (for reviews, see Bäckman, Nyberg, Lindenberger, Li, & Farde,
2006b; Diamond,
1996). At the same time, a related but distinct line of research also indicates that individual differences in within-person trial-by-trial processing fluctuations may be attributable to development-, aging-, or pathology-related deficiencies in dopaminergic modulations (e.g., see Castellanos & Tannock,
2002; Diamond, Briand, Fossella, & Gehlbach,
2004; Li, Lindenberger, & Sikström,
2001; see MacDonald, Nyberg, & Bäckman,
2006; Winterer & Weinberger,
2004, for reviews). With respect to processing fluctuations, children and older adults tend to show higher levels of within-person trial-by-trial RT variability (e.g., Li et al.,
2004; Williams, Hultsch, Strauss, Hunter, & Tannock,
2005). Furthermore, older individuals who show greater short-term trial-by-trial variability tend also to show greater extent of longitudinal declines in tasks assessing executive control (e.g., Lövdén, Li, Shing, & Lindenberger,
2007; MacDonald, Hultsch, & Dixon,
2003) and less efficient functional brain activation in regions (e.g., inferior parietal cortex) supporting episodic memory (MacDonald, Nyberg, Sandblom, Fischer, & Bäckman,
2008).
Given that studies on the development and aging of conflict monitoring mechanisms have thus far been conducted separately, direct comparisons between the child developmental and aging gradients are lacking. To bridge the gap, we investigated lifespan development of conflict monitoring in a sample that covered the age range from 6 to 89 years. Specifically, we assessed conflict cost associated with the incongruence between stimulus and response to indicate individual differences in the efficacy of conflict monitoring mechanisms broadly defined. Furthermore, in order to understand the ontogenetic antecedents and consequences of age-related differences in conflict monitoring, at one level we examined the relations between the efficacy of conflicting monitoring and indicators of processing fluctuation, which could reflect the efficacy of dopaminergic modulation; and at another level, we examined the relation between conflict monitoring and fluid intelligence, which is supported by basic cognitive processes.
Discussion
Findings from this study reveal that, similar to other basic cognitive processes and intellectual abilities, the age gradient of conflict monitoring follows a U-function across the lifespan: with a slightly steeper growth gradient during childhood and adolescence and a more gradual declining gradient during late adulthood and old age. Furthermore, lifespan age gradient of conflict monitoring parallels very closely to the gradient of processing fluctuation at one level and the gradient of fluid intelligence at another level. Individuals at both ends of the lifespan yielded larger conflict cost and at the same time displayed a greater extent of processing fluctuations in elementary cognitive processes and a lower level of performance in tasks assessing fluid intelligence. Individuals in adulthood showed least conflict cost, least processing fluctuation, and best performance in measures of fluid intelligence.
The details of how these three aspects of functions related to each other, however, differed across the lifespan. During childhood and adolescence, the associations between conflict monitoring and process fluctuation and between conflict monitoring and fluid intelligence are primarily shared with individual differences in baseline processing speed and chronological age. After controlling for individual differences in processing speed and age, conflict monitoring no longer correlates with either processing fluctuation or cognitive mechanics. In adulthood, after controlling for individual differences in processing speed and age, conflict monitoring still accounts for a small but reliable amount of variance in fluid intelligence. In late adulthood and old age, conflict monitoring remained to be significantly correlated with fluid intelligence and, importantly, with processing fluctuation, even after controlling for baseline processing speed and age. The fact that, during childhood and adolescence, the contribution of conflict cost to fluid intelligence is not independent of individual differences in baseline processing speed and in other factors correlated with chronological age indicates less differentiated processes in premature cognitive systems. Thus, measures of conflict monitoring mechanisms in children may be less task specific, and reflect additional processes that affecting processing speed and other factors. More generally speaking, this finding of lifespan differences in the correlations between conflict cost and other aspects of cognition highlights the developmental dynamics in conflict monitoring mechanisms. As brain and behavioral processes undergo functional reorganizations throughout development across the lifespan (Johnson,
2001; Li et al.,
2004; Lindenberger, Li, & Bäckman,
2006; Park et al.,
2004), conflict monitoring mechanisms may involve different brain functional circuitries and may affect different constellations of high-level cognitive abilities.
The significant relation between conflict cost and fluid intelligence in adulthood and old age suggests that conflict monitoring is a putative basic cognitive mechanism that subserves intellectual abilities. In childhood, conflict monitoring also contributes to the functioning of primary intellectual abilities; however, processing speed and other factors covariate with age play more important roles. The unique association between conflict cost and processing fluctuation in late adulthood and old age lends further support to the neuromodulation of neuronal noise theory of cognitive aging (e.g., Li et al.,
2001; Li, von Oertzen, & Lindenberger,
2006) and theories on dopamine’s involvement in conflict monitoring (e.g., Frank et al.,
2004; Holroyd & Coles,
2002; Yeung et al.,
2004). On the one hand, empirical data and simulation results suggest that aging-related deficit in dopaminergic modulation contributes to greater processing fluctuation, and on the other hand, dopamine is postulated to be involved in different aspects of conflict monitoring, including the more basic process, response conflict monitoring as examined here (Frank et al.,
2004; see Montague et al.,
2004, for review). Specifically, dopamine D1 receptor has been found to implicate the binding of perception with action (Colzato & Hommel,
2008, in press), which could be directly relevant to individual differences in conflict cost that arises from the incongruence between the stimulus feature and response action as examined here. Effects of dopaminergic modulation on cognitive processes have also been investigated with respect to child development (e.g., Castellanos & Tannock,
2002; Diamond,
1996; Rueda, Rothbart, McCandliss, Saccomanno, & Posner,
2005b). We found a relation between processing fluctuation and conflict cost in children and adolescence as well. However, this effect is mostly shared with processing speed and age, indicating that mechanisms in addition to neuromodulation of processing noise may add further influences here.
Furthermore, there is also evidence showing that individual differences in brain activity related to conflict processing are genetically predisposed (Anokhin, Heath, & Mayers,
2004). Currently there are rapid developments in combining genomic approaches with functional brain imaging and cognitive experiments in understanding the relations between neuromodulation of functional brain processes and cognition in children and adolescence (e.g., Cornish and Hollis
2002; Rueda et al.,
2005a; Wahlstrom et al.,
2007), in adults (e.g., Frank, Moustafa, Haughey, Curran, & Hutchison,
2007), during aging (e.g., Bäckman, Nyberg, & Farde,
2006a; Nagel et al.,
2008) or in the case of pathology (e.g., Frank et al., 2002; Winterer & Weinberger,
2004). Future research on lifespan development of conflict monitoring could benefit from taking a combined genomic neurocognitive research to disentangle individual and developmental differences in gene–brain–behavior relations.