Dual-task processing in younger and older adults: Similarities and differences revealed by fMRI
Research highlights
► Dual-task interference in young and elderly adults. ► fMRI study: Neural substrates of age differences. ► Neural network: Lateral and medial frontal structures in both age groups. ► Networks employed differently in the two age groups. ► Results do not support Cabeza’s HAROLD.
Introduction
Carrying out more than one task at the same time is commonplace in everyday life, for example driving an automobile while conversing, either on a mobile telephone or with a passenger. Despite anecdotal claims that this can be done successfully, empirical evidence shows that under most circumstances, there is noticeable interference with both tasks. In the instance of telephoning while driving, both driving performance and conversation flow are significantly impaired (Charlton, 2009, Strayer and Drews, 2007). A very substantial body of evidence from controlled laboratory experiments confirms the validity of the findings from real world tasks (for reviews, see Pashler, 1994, Pashler, 1998). Further, the difficulty in managing overlapping tasks appears to increase with advancing age (for reviews, see McDowd and Shaw, 2000, Verhaeghen et al., 2003). The existence and nature of age-related differences in dual-task management are of both theoretical interest and practical concern.
The method for studying dual-task performance that provides the most leverage in understanding interference gives two simple tasks and systematically manipulates the onset time for the stimuli for each task. For example, in a study of simulated driving, Levy, Pashler, and Boer (2006) had participants carry out two tasks: The first task was to determine whether a brief auditory or visual stimulus had been presented once or twice; the second task was to press the brake pedal whenever the brake lights of a lead car illuminated. The SOAs between the auditory or visual stimulus and the brake light change ranged from 0 ms to 1200 ms. With single-task reaction times under 1000 ms for the auditory–visual task, this meant that there was substantial overlap between the tasks at the shortest SOAs whereas with the longest SOA the response to the first task would likely have been given before the stimulus for the second task appeared. Nevertheless, all of the processes involved in managing both tasks were present on each trial, unlike other approaches in which dual-task performance is simply compared to single-task performance. A large number of experiments using the varied-SOA procedure have been reported with very consistent results (for reviews, see Pashler, 1994, Pashler, 1998). As SOA decreases (and, therefore, task overlap increases), RTs to Task 2 are slowed dramatically. RTs to Task 1 show little or no effect of SOA. This period over which the RT to Task 2 is slowed has been called the psychological refractory period (PRP, Vince, 1948, Welford, 1952) in analogy to the period after an initial firing when a neuron is unresponsive; the general paradigm is often called the PRP procedure. The overwhelming weight of the empirical evidence is consistent with response-selection bottleneck models.1 These models assume that there are broadly three stages of processing, an early stage involving perceptual processing, a central stage involving response selection, and a final stage involving execution of the response. The critical assumption is that there is a bottleneck at the central processing stage, such that processes such as response selection can only be carried out for one task at a time (e.g., Lien, Ruthruff, & Johnston, 2006). Central processing of the other task must be postponed until central processing of the first task is complete. Unlike central processing, perceptual processing of the two tasks can occur largely in parallel as can execution of the responses for the two tasks (but see de Jong, 1993).
Age-related differences in cognitive function have been attributed to a reduction of executive control of cognitive processes in old age (e.g., Hull, Martin, Beier, Lane, & Hamilton, 2008). From one point of view, the PRP procedure provides an ideal vehicle to examine this hypothesis because of the need for additional executive control when the second task follows closely on the first (e.g., Erickson et al., 2005): (a) Two task sets must be maintained in working memory; (b) the order in which the two tasks are to be carried out must be prepared and then managed; (c) once processing of one task has begun, processing of the other task must be interrupted and delayed, while maintaining the results of processing already completed; (d) when processing can be returned to the second task, there must be a fast switch of attention back to that task and reinstatement of the processing; and (e) responses must be programmed and executed for two incompatible tasks. Any of these executive operations would be a plausible locus for age-related differences. In contrast to claims that the PRP procedure requires active, executive control, there is also a second and very different view point. In this point of view the behavioral slowing is simply due to passive queuing as the second task waits for necessary resources to become available (Jiang et al., 2004, Marois et al., 2006). In this view, the slowing is due not to increased executive demands but simply to postponement. If this point of view is correct, we might not expect any greater effect of task overlap in older adults than in younger adults, other than what would be expected from normal age-related slowing of all processes (e.g., Hartley, 2006, Salthouse and Miles, 2002).
Earlier studies of age differences in dual-task performance used procedures with little control over the relative onset of processing in the two tasks (e.g., McDowd & Craik, 1988). More recent age group comparisons have adopted variants of the PRP procedure with controlled onset of two simple tasks. The results for older adults, as for younger adults, have been well fit by response-selection bottleneck models. Allen, Smith, Vires-Collins, and Sperry (1998) concluded that interference in central stage response selection between the two tasks was greater in older than in younger adults. Glass et al. (2000) and Hartley and Little (1999), however, concluded that after general slowing was taken into account, the age differences were small, and could be localized to greater difficulty at input and to a slowed central process of the “unlocking” of processing in the second task. Consistent with Glass et al., Hein and Schubert (2004) concluded that older adults were more sensitive to interference in input modalities. Maquestiaux, Hartley, and Bertsch (2004) also implicated greater difficulty in the switching of central processing when they found that highly trained older adults—but not younger adults—were aided by shifting to tasks that were comparable but with simpler response selection rules. Hartley and Maquestiaux (2007) concluded that central operations were equivalent in younger and older adults, but that older adults showed greater output interference. Hartley (2001) showed that much of the age difference in switching between two different tasks could be eliminated by removing output interference. Thus there are indications of age-related differences at all three phases: input processes, central processes, and output processes.
Neuroimaging is very promising as a way to put constraints on theories of dual-task interference (Jiang et al., 2004). Marois and Ivanoff (225) reviewed a number of approaches that have been used, among them comparing dual-task performance to that of the two tasks done singly and comparison of dual-task performance with high task overlap to that with low task overlap. As they note, each approach has strengths and limitations.
Despite the variety of approaches that have been taken to neuroimaging of dual-task performance and although there are differences from study to study, the areas of activation have been relatively consistent (see Marois & Ivanoff, 2005, for a meta-analysis). Activations have been reliably found in lateral prefrontal cortex (Broadmann’s Areas—BAs—9, 44, 45, 46), supplementary motor areas (BAs 6, 8), and parietal areas (BAs 7, 40). Activations have frequently been observed in the anterior cingulate cortex (BAs 24, 32), posterior areas such as cuneus (BAs 18, 19), orbital frontal cortex and anterior insula (BA 47), polar prefrontal cortex (BA 10), temporal areas (BA 37), and subcortical structures such as cerebellum, the basal ganglia, and the thalamus. Activations in these regions have been obtained in the left hemisphere, in the right hemisphere, and bilaterally.
Although different researchers describe it somewhat differently, a consensus model has emerged of how these areas might be involved in executive control of dual-task processing. In this view, the lateral prefrontal cortex is optimized for rapid, adaptive, amodal control (Dosenbach et al., 2008, Dosenbach et al., 2006, Dosenbach et al., 2007, Marois et al., 2006) and is involved in the fast adaptation of response sets and the coordination of selection-for-action in situations with interfering information (Collette et al., 2005, Koechlin et al., 1999; Schubert & Szameitat, 2003; Szameitat, Schubert, Müller, & Von Cramon, 2002). Medial areas, including anterior cingulate, are optimized for stable set maintenance, maintaining and monitoring associations between actions and their outcomes and the implementation of task sets particularly in situations of conflict (Dosenbach et al., 2006, Dosenbach et al., 2008, Fleck et al., 2006, MacDonald et al., 2000, Rowe et al., 2008). The lateral and medial prefrontal areas interact to exercise top-down control, biasing signals to parietal areas that load, transmit, or implement the required task-set parameters (Dosenbach et al., 2006, Dosenbach et al., 2008, MacDonald et al., 2000, Sigman and DeHaene, 2006). The parietal areas can also operate to feed information forward for stimulus-driven bottom-up shifts of attention.
Other studies have found little or no evidence for recruitment of executive areas in the dual-task situation beyond those activated in the single-task situation (Adcock et al., 2000, Bunge et al., 2000, Erickson et al., 2005, Jiang et al., 2004; also see Sigman & DeHaene, 2008). These studies are consistent with the passive-queuing model in which the processes carried out at short SOAs are no different from those carried out at long SOAs. Rather than active monitoring and management of processes in the two tasks, a delay is simply injected into the processing stream as Task 2 is passively queued until Task 1 central processing is completed.
Only one neuroimaging study has examined dual-task interference in older adults. Erickson et al. (2007b) presented older adults with two tasks simultaneously, to determine the color (yellow or green) of one stimulus and to determine the letter identity (B or C) of another. They did not report results of whole-brain analyses comparing dual and single tasks, but rather focused selectively on two regions of interest that showed significant change from before to after five dual-task training sessions, ventrolateral PFC and dorsolateral PFC. When we compared their results for older adults with those previously published for younger adults (Erickson et al., 2007a), it suggested to us that brain activations prior to training were greater in younger adults than in older adults in left ventrolateral PFC but were greater in older adults than in younger adults in dorsolateral PFC bilaterally. It is difficult to draw conclusions from this study about age differences in the neurobiological substrate of dual-task interference both because the information provided is relatively limited and because the procedure compared dual-task conditions with single-task conditions, rather than comparing short and long SOAs within the dual-task situation as is done in the PRP procedure. The present study examined age differences in dual-task interference using contrasts of activations at short (100 ms) and long (1000 ms) SOAs and using whole-brain analyses rather than predetermined regions of interest.
Absent previous findings, what age differences can we expect to find in brain activations in the PRP procedure? The literature provides three alternative schemas. Cabeza (2002) summarized a number of studies showing that areas of activation in older adults were similar to those in younger adults, but with a more bilateral pattern of activity. Because the evidence was limited, he restricted the generalization—a schema that he termed hemispheric asymmetry reduction in older adults or HAROLD—to the prefrontal cortex, but speculated that it might apply more generally. Reuter-Lorenz and Lustig (2005) concluded that these patterns were also observed in more recent studies. Although HAROLD is an empirical generalization, it has led to a number of theoretical conjectures. The asymmetry reduction can be seen either as underactivation or overactivation in older relative to younger adults. Reuter-Lorenz and Lustig describe hypotheses about underactivation as postulating impairments in cortical areas (e.g., differential age effects in the right hemisphere, Dolcos, Rice, & Cabeza, 2002) or in cortical–cortical connections (e.g., the age-related breakdown of white matter structural integrity, Ardekani et al., 2007, Bartzokis et al., 2004). Reuter-Lorenz and Lustig describe hypotheses that attribute overactivation either to compensatory or to incidental recruitment resulting from age-related impairment. Young adults show patterns of recruitment with increasing task demands (e.g., Braver et al., 1997), and older adults may attempt to compensate for greater difficulty by recruiting other areas of the cortex at lower levels of task demands. Alternatively, the overactivations may be nonselective (e.g., Logan, Sanders, Snyder, Morris, & Buckner, 2002) or may be the result of age-related failures of inhibitory connections (Cabeza, 2002). In these cases, the overactivation could be epiphenomenal or could be interfering. Reuter-Lorenz (Reuter-Lorenz and Cappell, 2008, Reuter-Lorenz and Lustig, 2005) argue that the evidence is best explained by a schema somewhat broader than HAROLD, compensation-related utilization of neural circuits (CRUN) in older adults, the CRUN hypothesis or CRUNCH.
A very different schema to explain age differences in brain activation and cognitive processing has been proposed by Braver, Gray, and Burgess (2007). They propose dual mechanisms of cognitive control. When the management of task demands can be planned in advance and when resources are adequate, they argue that proactive control is likely to occur. Proactive control involves (a) the active maintenance of context information in PFC, allowing it to bias processing in other systems, (b) the augmenting of PFC functions through rapid bindings of representations in the medial temporal lobe, and (c) performance monitoring in the ACC. When proactive control is not possible or too costly, reactive processing is invoked which is engaged only as needed on a just-in-time basis, rather than consistently and in advance of critical events. Braver et al. hypothesize that older adults, with processing limitations, are likely to resort to reactive processing in situations that elicit proactive processing from young adults. They predict that this shift will lead either to activations of brain regions not typically activated in younger adults or to different patterns of activation in the same areas activated in younger adults, decreasing in conditions most dependent on control and increasing when control is less possible. In support, they cite evidence of their own from a continuous performance task in which a long delay led to an increase in left PFC activation in younger adults, but a decrease in older adults, as well as a finding by Jonides et al. (2000) that resistance to interference in a Sternberg task in younger adults resulted in an activation in left PFC not seen in older adults. Velanova, Lustig, Jacoby, and Buckner (2006) offered a similar hypothesis that older adults shift to less taxing strategies than those used by younger adults.
The theoretical notions we have reviewed lead to three quite different sets of predictions. First, (a) if we found activations consistent with reduced asymmetry or compensatory recruitment in older adults (i.e., consistent with the HAROLD generalization and CRUNC hypothesis), (b) if dual-task processing is characterized by greater executive processing demands, and (c) if older adults are subject to impaired executive processing, we could expect to see more bilateral activations in lateral prefrontal cortex and in anterior cingulate and medial frontal cortex in older adults. We might also expect more bilateral activations in parietal cortex and extrastriate areas as well as subcortical areas. Second, if older adults adopt more reactive processing in the PRP procedure, following the notions of Braver et al. (2007) we would expect to see a reduction or elimination of lateral PFC and ACC and medial FC activation, in sharp contrast to the HAROLD or CRUNCH predictions. We might also expect to see increased activity in more posterior areas and in subcortical areas concerned with rapid switching of attention such as basal ganglia. Third and last, if dual-task processing simply reflects passive queuing of processes, we would expect to find few if any age differences in activation, once normalized for overall level of activity.
Section snippets
Participants
Twelve older adults (7 female) age range: 65–77 years; M = 70.67) were recruited through the University of Michigan Institute of Gerontology and through newspaper advertisements. The 12 younger adults (6 female, age range: 19–25 years, M = 21.00) were recruited through University of Michigan subject pools and through newspaper advertisements. All participants were right-handed, free of positive neurological histories, and had normal or corrected-to-normal vision. The two age groups had similar median
Behavioral results
ANOVA was carried out on the median reaction times (RTs) to Task 2 (letter) on trials with correct responses to both the color task and the letter task as a function of the age group (younger or older) and the SOA (100 ms and 1000 ms). Medians were used rather than trimming to reduce the effect of outliers. There was a significant effect of age, F(1, 22) = 19.58, p < .001, = .46,2 with
Discussion
We will first discuss the neuroimaging results common to younger and older adults before returning to a discussion of the age differences.
Acknowledgments
This research was supported by Grant AG15-19195 from the US National Institute on Aging. We are particularly grateful to Michael Spezio for his extensive assistance with data analysis and to Amanda Thomas and James Christianson for testing the participants. We are also grateful for the advice and assistance of Deborah Little, Patricia Reuter-Lorenz, Edward E. Smith, Nicole Speer, and Tor Wager.
References (96)
- et al.
Exploratory voxel-based analysis of diffusion indices and hemispheric asymmetry in normal aging
Magnetic Resonance Imaging
(2007) - et al.
Heterogeneous age-related breakdown of white matter structural integrity: Implications for cortical “disconnection” in aging and Alzheimer’s disease
Neurobiology of Aging
(2004) - et al.
A parametric study of prefrontal cortex involvement in human working memory
NeuroImage
(1997) - et al.
Neural mechanisms of transient and sustained cognitive control during task switching
Neuron
(2003) - et al.
Cognitive and emotional influences in anterior cingulate cortex
Trends in Cognitive Science
(2000) Driving while conversing: Cell phones that distract and passengers who react
Accident Analysis and Prevention
(2009)- et al.
Involvement of both prefrontal and inferior parietal cortex in dual-task performance
Cognitive Brain Research
(2005) - et al.
Hemispheric asymmetry and aging: Right hemisphere decline or asymmetry reduction
Neuroscience and Biobehavioral Reviews
(2002) - et al.
A core system for the implementation of task sets
Neuron
(2006) - et al.
A dual-networks architecture of top-down control
Trends in Cognitive Science
(2008)
Neural correlates of dual-task performance after minimizing task-preparation
NeuroImage
Training-induced plasticity in older adults: Effects of training on hemispheric asymmetry
Neurobiology of Aging
The effect of finger-movement speed of the dominant and the subdominant hand on cerebellar activation: A functional magnetic resonance imaging study
Neuroimage
Functional grouping and cortical–subcortical interactions in emotion: A meta-analysis of neuroimaging studies
Neuroimage
An evaluation of distinct volumetric and functional MRI contributions toward understanding age and task performance: A study in the basal ganglia
Brain Research
FMRI of healthy older adults during Stroop interference
NeuroImage
Age-related differences in response regulation as revealed by functional MRI
Brain Research
Under-recruitment and nonselective recruitment: Dissociable neural mechanisms associated with aging
Neuron
Capacity limits of information processing in the brain
Trends in Cognitive Science
Psychiatry Research
Attentional control in the aging brain: Insights from an fMRI study of the Stroop task
Brain and Cognition
On the optimality of serial and parallel processing in the Psychological Refractory Period paradigm: Effects of the distribution of stimulus onset asynchronies
Cognitive Psychology
Brain aging: Reorganizing discoveries about the aging mind
Current Opinion in Neurobiology
Functional significance of age-related differences in motor activation patterns
NeuroImage
Functional neuroanatomy of interference in overlapping dual-tasks: An fMRI study
Cognitive Brain Research
Brain localization of attentional control in different age groups by combining functional and structural MRI
NeuroImage
Changing channels: An fMRI study of aging and cross-modal attention shifts
NeuroImage
Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain
Neuroimage
Neuroimaging studies of shifting attention: A meta-analysis
NeuroImage
Functional neuroanatomy of executive processes involved in dual-task performance
Proceedings of the National Academy of Sciences of the United States of America
The psychological refractory period: Evidence for attentional differences in time-sharing
Psychology and Aging
Explaining the many varieties of working memory variation: Dual mechanisms of cognitive control
A resource model of the neural basis of executive working memory
Proceedings of the National Academy of Sciences of the United States of America
Hemispheric asymmetry reduction in older adults: The HAROLD model
Psychology and Aging
The implications of cortical recruitment and brain morphology for individual differences in inhibitory function in aging humans
Psychology and Aging
Attentional modulation of neural processing of shape, color, and velocity in humans
Science
Superior parietal cortex activation during spatial attention shifts and visual feature conjunction
Science
xjView 4 manual
Multiple bottlenecks in overlapping task performance
Journal of Experimental Psychology: Human Perception and Performance
General and task-specific frontal lobe recruitment in older adults during executive processes: A fMRI investigation of task-switching
Neuroreport
Distinct brain networks for adaptive and stable task control in humans
Proceedings of the National Academy of Science
Dissociating the roles of the rostral anterior cingulate and the lateral prefrontal cortices in performing two tasks simultaneously or successively
Cerebral Cortex
Training-induced functional activation changes in dual-task processing: An FMRI study
Cerebral Cortex
Role of prefrontal and anterior cingulate regions in decision-making processes shared by memory and nonmemory tasks
Cerebral Cortex
Aging and the psychological refractory period: Task-coordination strategies in young and old adults
Psychology and Aging
Cerebral control of unimanual and bimanual movements: An H2(15)O PET study
Neuroreport
Age differences in dual-task interference are localized to response generation processes
Psychology and Aging
The changing role of the speed of processing construct in the cognitive psychology of human aging
Cited by (36)
Age-related differences in the role of the prefrontal cortex in sensory-motor training gains: A tDCS study
2021, NeuropsychologiaCitation Excerpt :Further to this, in a functional magnetic resonance imaging (fMRI) study directly comparing the neural substrates in dual-task interference between older and young adults, Hartley et al. (2011) suggested that both age groups showed similar underlying neural substrates involving a medial prefrontal and lateral frontal-parietal network. However, older adults showed additional recruitment of the frontopolar prefrontal cortex of which increased activation was observed to be correlated with successful reduction in dual-task interference (Hartley et al., 2011). Moreover, as is the case noted above, there is great heterogeneity across tDCS studies on cognition in older adults due to differing stimulation parameters (e.g., stimulation intensity, duration, single session versus repeated sessions, brain regions target, and/or cognitive tasks used, Prehn and Flöel, 2015).
Aging of the frontal lobe
2019, Handbook of Clinical NeurologyHow attention modulates encoding of dynamic stimuli in older adults
2018, Behavioural Brain ResearchCitation Excerpt :Indeed, divided attention during encoding activates the left ventrolateral prefrontal cortex (vlPFC) [13–15], thought to support cognitive control over memory [16,17] and allocation of attentional resources to two streams of incoming information [18]. Unlike younger adults, older adults show diminished ability to modulate neural activity of frontal and parietal regions in a dual task as compared to a single task [6,7]. Hence, for older adults, these regions are anticipated to show further decrease in neural modulation when comparing low to high attentional loading during encoding.
Putting age-related task activation into large-scale brain networks: A meta-analysis of 114 fMRI studies on healthy aging
2015, Neuroscience and Biobehavioral ReviewsShedding light on the effect of priority instructions during dual-task performance in younger and older adults: A fNIRS study
2015, Brain and CognitionCitation Excerpt :It seems that PRP paradigms mostly activate the right hemisphere, while increased priority demands or simultaneous DT paradigms produce bilateral activation. When comparing the age-related brain activation associated with short versus long SOA in a PRP paradigm, Hartley et al. (2011) found that older adults activation patterns were not as extensive as younger adults in the basal ganglia, the left occipital cortex and the left orbitofrontal cortex, but that activations were greater in older than younger adults in the right anterior prefrontal cortex (BA 10), the left (BA 39) and the right (BA 19) middle occipital gyri and the right cuneus (BA 19). In contrast, Erickson et al. (2007) used a DT paradigm that required simultaneous completion of color and letter discrimination tasks without any task priority instructions.