This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation

Age-Related Differences in Dual-Task Visual Search

Are Performance Gains Retained?

^a Georgia Institute of Technology, Atlanta

Peter J. Batsakes, School of Psychology, Georgia Institute of Technology, Atlanta, GA 30332-0170 E-mail: pbatsakes{at}prodigy.net.

Decision Editor: Toni C. Antonucci, PhD

	Abstract

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
Older and young adults practiced a verbal/spatial dual task and were tested for retention performance 1 month later. Participants first practiced each component task separately to individually determine component processing time. Thus, age-related differences in single-task detection sensitivity were minimized prior to performing the dual task. Participants practiced the dual task for two 1.5-hour sessions. Following the retention interval, they were retested on the single-task components and on the dual task. Correct detection as well as signal detection parameters were examined. Older adults demonstrated decreased sensitivity as well as a more conservative response bias during acquisition. Retention performance for the single tasks replicated previous retention studies, demonstrating age-related performance declines when stimulus-specific learning is assessed. Dual-task retention capability declined for both older and young adults equally when detection accuracy, but not perceptual sensitivity, was measured. Response bias changed differentially for older and young adults across the retention interval.

Amajor goal of the present study was to add to our general understanding of age-related differences in the retention of newly acquired skills. Questions of knowledge and skill acquisition have been an important focus of research within the area of cognitive aging. Yet, the field has less vigorously addressed questions concerned with the maintenance of newly acquired knowledge and skill across time. Skilled performance deteriorates with disuse for both older and young adults; however, a broader understanding of age-related decline is important from both theoretical and practical perspectives. In this study we investigate age-related differences in the retention characteristics of trained dual-task search performance. The present evaluation of retention capability assesses the replicability of findings from previous studies and extends those findings to a more complex set of tasks than have been previously investigated.

Previous research involving age-related differences in skilled search has drawn a distinction between the retention of stimulus-specific and task-specific learning (Fisk, Cooper, Hertzog, and Anderson-Garlach 1995; Fisk, Hertzog, Lee, Rogers, and Anderson-Garlach 1994). This distinction draws on formal theories of skill acquisition in complex tasks (for a review, see Schneider and Detweiler 1987, Schneider and Detweiler 1988). Stimulus-specific learning refers to the developments, that occur in the processing of the trained stimuli across extended practice. For example, both young and older adults become skilled at searching through feature differentiation and, with very extended practice, young adults develop an automatic attention response (AAR) in which target stimuli seemingly "attract" and distractor stimuli seemingly "repel" attention (Rogers and Fisk 1991). Regardless of the learning mechanisms supporting search performance, however, older adults generally perform more poorly when stimulus-specific retention capability is assessed (e.g., Fisk et al. 1994, Fisk et al. 1995; Rogers, Gilbert, and Fisk 1994).

Task-specific skill, on the other hand, refers to the general strategies that participants form without regard to the particular stimuli involved (see Fisk et al. 1994; Kramer, Larish, Weber, and Bardell 1999). This type of skill is specific to the task domain at hand, yet, is not tied to the stimuli used during the training session. Task-specific strategies that develop for search detection tasks include more efficient search of the display, rehearsal of memory set items, execution of responses, and so forth. Extended training leads to the development of more efficient search strategies for both young and older adults, and retention capability is generally equivalent when task-specific learning is assessed (Fisk et al. 1994; Fisk et al. 1995).

The present study is an attempt to further our understanding of age-related differences and similarities in the retention of acquired skill within the search-detection domain. We chose to examine dual-task search performance for three reasons. First, dual-task manipulations represent a robust way of increasing the difficulty and complexity of processing requirements (see e.g., McDowd and Craik 1988; Salthouse 1982; Schneider and Detweiler 1988; Sperling 1984). Second, such a procedure has been useful in evaluating a variety of age-related issues in cognition (for reviews, see Hartley 1992; Kramer and Larish 1996) and should lend itself to the analysis of age-related retention capability. Finally, performance of dual tasks has been conceptualized as requiring attentional and task control beyond the learning of the specific stimuli used in the task (for a review, see Schneider and Detweiler 1988). Studies involving extended practice within a dual-task framework have demonstrated that this aspect of dual-task performance improves dramatically with experience (Damos, Bittner, Kennedy, and Harbeson 1981; Kramer and Larish 1996; Schneider and Fisk 1982, Schneider and Fisk 1984), at least for young adults.

	Age-Related Dual-Task Performance

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
An examination of the aging and dual-task literature certainly tells us that age-related dual-task performance differences exist (Lorsbach and Simpson 1988; McDowd and Craik 1988; Salthouse, Rogan, and Prill 1984; see also Somberg and Salthouse 1982). However, there is little agreement among researchers as to the mechanisms underlying these performance differences. One class of theories deals, in one form or another, with the quantitative increase in the complexity of dual-task over single-task processing requirements. According to this class of theories, older adults experience a general slowing of cognitive processes with age (Birren 1974; Salthouse 1985). The result is a disproportionate performance decrement for older adults as the number of cognitive operations necessary for task performance increases. This performance decrement can be thought of as resulting from an enduring resource-based characteristic of older adults, as much responsible for initial performance as for any trained level of performance (e.g., Salthouse, Hambrick, Lukas, and Dell 1996).

However, other research suggests that multiple-task processing itself possesses characteristics, apart from the overall complexity of the task, that prove disproportionately difficult for older adults. Most important for the present investigation, this line of evidence suggests that these age-related differences are skill based, or at least not the result of enduring age-related characteristics. For instance, some researchers have found larger age-related performance decrements in less, rather than more, complex task combinations (e.g., Korteling 1993; Rogers, Bertus, and Gilbert 1994), suggesting that task-control processes play an important role in understanding age-related differences in performance. For example, Korteling 1991 showed that older adults initially lack the attentional flexibility required for the switching of attention between task components. More recently, evidence has been collected showing that training can modify the strategic approach to the dual task, resulting in changes in age-related patterns of performance (Baron and Mattila 1989; Kramer, Larish, and Strayer 1995; Sit and Fisk 1999) and in a reduction in initial age-related performance differences (Kramer et al. 1999). Collectively, this body of research suggests that strategic differences are present in age-related dual-task performance and that practice can produce more efficient performance strategies for both young and older adults. Thus, there is evidence to suggest that examining age-related retention of dual-task performance will add to our understanding of skill-based retention performance beyond retention for the specific stimuli used during training.

	Overview of Present Study

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
This study was designed to address two major questions regarding aging and dual-task processing. First, we wanted to determine the extent to which group differences in dual-task performance were associated with declines in processing speed with age. Using a multiple-frame stimulus presentation format (for a review, see Shiffrin 1988), age-related differences in single-task search were reduced through the systematic manipulation of stimulus presentation rates on the basis of individual performance. We could, therefore, consider initial dual-task performance differences to more greatly reflect strategic and skill-based influences. In addition, task speed during dual-task processing was varied to manipulate attentional demands (see e.g., Shiffrin 1988). This type of manipulation involves a resource considered important in the aging literature (Salthouse 1989), in dual-task situations in particular (Schneider and Fisk 1982), and clearly moderates performance in general (Schneider and Shiffrin 1977). Finally, the manipulation allowed us to evaluate the relationship between varying demands on attention and performance for both age groups.

The second question concerns the measurement of age-related differences in the retention characteristics of skilled performance for this class of tasks. Results from the small number of age-related, dual-task training studies suggest that performance will improve with dual-task practice for both age groups, although the dual-task performance decrement may remain despite practice (e.g., Rogers, Bertus, and Gilbert 1994). The present experiment, therefore, was conducted to broaden our understanding of age-related differences in the retention of skilled search associated with dual-task performance. Signal detection theory (Green and Swets 1966) was used to separate stimulus-specific learning (sensitivity) from more strategic (response criterion) retention characteristics for both single and dual-task performance. We predict a general, age-related decline in sensitivity over the course of the retention interval (a measure of stimulus-specific learning). Also, if the dual-task data are consistent with previous single-task retention data, then the learning associated with efficient dual-task search performance should decline equivalently for young and older adults.

	Methods

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
Participants
Twenty-four young adults aged 17 to 37 years (M = 21, SD = 3.7) and 24 older adults aged 64 to 82 years (M = 71.6, SD = 4.71) participated in the experiment. The young adults were compensated with a combination of money (up to $100) and course credit. The older adults were paid volunteers from the community. Medication screening eliminated participants taking medications having more than minimal effects on attention. Participants self-rated their general health on a scale of 1 (excellent) to 5 (poor). These ratings (M = 1.63, SD = .77, for the young adults, and M = 1.83, SD = .984, for the older adults) indicated that both groups thought their general health to be in the good to excellent range. All participants were required to pass a Snellen Acuity Test at the level of 20/40 for far and near vision (corrected or uncorrected). Ability tests were administered to participants during the training and retention phases of the experiment (see Table 1 for a summary of results).

Task and Stimuli
The dual task was adapted from Detweiler and Lundy 1995. Two consistently mapped visual search tasks, semantic-category search and pattern search, were adopted as component tasks. A multiple-frame procedure presented 60 successive frames (displays) of stimuli to each participant to form a single trial (see Schneider and Shiffrin 1977, for a thorough discussion of multiple-frame procedures). On each frame, both a word and a pattern were displayed for a fixed duration. Frame-duration time (i.e., the time from the onset of one frame to the onset of the next frame) was determined individually for each participant. This adaptive training procedure is described in the single-task training section.

For the category search task, eight semantic categories were chosen from the Battig and Montague 1969 taxonomic category norms. Eight exemplars were selected from each of the following categories: articles of clothing, building parts, colors, earth formations, furniture, four-footed animals, human body parts, and weapons. Category exemplars were from 3 to 5 letters in length and were chosen on the basis of prototypicality, high frequency ratings (Battig and Montague 1969), and lack of visual distinctiveness. Words were displayed in black lower case lettering on a light gray background. All letters were 1-cm high and ranged from 2.5 cm in length for a three-letter word to 3.4 cm in length for a five-letter word.

The pattern search task consisted of a 5 x 5 grid of squares composed of black lines on a light blue background. Eight different spatial patterns were constructed by blacking out two of the nine squares within a smaller 3 x 3 grid. These spatial patterns formed the target and distractor stimuli. On each frame, one of these spatial patterns appeared within the larger 5 x 5 grid. There were nine locations within the display in which the target or distractor pattern could appear. The choice of locations containing the pattern on any given frame was determined through random selection of a location adjacent to the previous location of the trial. The target pattern could appear anywhere within the display grid. The word stimulus for the category search task was positioned above the spatial pattern task in the display, with 0.4 cm separating the two. The visual angle subtended by the entire dual-task display was 00.00 5.75° in height and 4.15° in width. During single-task trials, the displays for both search tasks were present. The stimuli within the unattended task consisted of only distractors and never any previously practiced consistently mapped (CM) targets. See Fig. 1 for a diagram of single-task and dual-task trials.

View larger version (18K): [in this window] [in a new window]   Figure 1. Diagram of single- and dual-task trials. (See text for an explanation of the procedures.)

Procedure
Single-task training.
The single-task training session consisted of 18 blocks of four trials. The word search and the pattern search task were alternated between blocks beginning with the word search task on Block 1. Each participant, therefore, received nine blocks and 36 trials of training on each single-task component. Each trial consisted of the following sequence of events. Participants pressed a key to initiate the trial. This was followed by a 3-s presentation of the target category (either a semantic-category or spatial pattern). A 1-s "ready" signal was then displayed, followed by the 60 frames of the multiple-frame trial. Participants were instructed to monitor the relevant display and respond to targets for the word search task by pressing the W key with their right index finger and to targets in the pattern search task by pressing the P key with their left index finger. On any given trial, between 6 and 8 of the 60 total frames contained targets and at least 1,000 ms elapsed between the presentation of any two targets. Over the course of any single-task block, there were between 24 and 32 target frames, averaging 504 target frames per session for each participant. This permutation of target-frame occurrence was enforced to maintain active participation throughout a given trial. Participants were instructed that they had a window of up to 1,000 ms to respond to any target before it was recorded as a miss. Targets never appeared on Frame 1 or on Frame 60 of a trial. A correct detection was recorded for a target if a response was made within 200 ms to 1,000 ms after the presentation of the target. A false alarm was recorded if a response was made but no target was present. At the end of each trial, a display was presented reporting the participant's number of correct detections, misses, and false alarms to encourage monitoring of individual performance.

Determining frame duration.
An adaptive procedure was used during single-task training to determine, individually, task speeds used during dual-task trials (Whaley and Fisk 1993). For each block of single-task trials, an A' score (see below for discussion of A') was calculated for each participant. If A' was greater than .96 on any individual block, the frame time was decreased by 25 ms on the subsequent block of that task. If, however, a participant's A' was less than .94, the frame time was increased by 25 ms. Frame times remained constant if A' fell within the .94 to .96 range. The initial frame time was set at 400 ms for both tasks and for both age groups.

Dual-task training.
Two dual-task training sessions followed the single-task training session. Each session consisted of 18 blocks of four trials and each block contained between 24 and 32 target frames (between 12 and 16 target frames per task component), averaging 504 target frames per session, with an equal number of word and pattern targets appearing within each block of trials. There were three individually determined processing rates used during dual-task training. These different task speeds were created as a function of the ending averaged frame times of the two component tasks from single-task training ([ending frame time word task + ending frame time pattern task]/2). Processing time per frame for the first six blocks of each dual-task session were fixed at twice this ending average frame time (the slow speed condition). The next six blocks of each session used a display time fixed at 1.5 times the ending average (the medium speed condition), and the last six blocks were fixed at the given participant's ending averaged frame time from the single-task session (the fast speed condition). For instance, if a participant's ending average frame duration from single-task training was 300 ms, the slow, medium, and fast task speeds would be 600 ms, 450 ms, and 300 ms, respectively. Over the two sessions of dual-task training, participants received 12 total blocks of practice on each of the three prescribed task speeds.

For dual-task trials, participants were instructed to divide attention equally between both search displays and respond to targets in each display by pressing the appropriate key. Targets could appear on any frame with the following constraints: only one target could appear on any given frame (word or pattern), there would be at least 1,000 ms between the presentation of any two targets, and no target could appear on either Frame 1 or Frame 60 of a trial. At the end of the second dual-task training session, additional single-task data was collected to determine a baseline for perceptual sensitivity measures observed during single-task retention testing.

Retention testing.
Retention testing occurred 4 weeks after dual-task training and consisted of testing both single task and dual tasks. Each participant received a total of six blocks of single-task retention testing with either two or four single-task blocks occurring before and either two or four after the nine blocks of dual-task trials. For the single-task retention testing, frame times were fixed at the ending frame times from the preretention single-task performance measure. For dual-task trials, processing time (frame speed) alternated every block of trials so that each of the three task speeds appeared every third block. The order of frame speed (fast, medium, or slow) was counterbalanced across participants through the use of a partial Latin square and was replicated across age groups.

	Training Results

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
Overview of Analyses
In addition to considering correct detections to evaluate performance, we calculated signal-detection measures. Signal detection, as a model of perceptual processing, seemed appropriate, because it allows independent assessment of perceptual sensitivity and response bias (Green and Swets 1966). A' was chosen as the detection sensitivity index (Craig 1979) and is a measure of the area under the receiver operating characteristic curve ranging from 0.5 (chance detection) to 1.0 (perfect detection). The A' measure is a more distribution-free measure of detection sensitivity than d' (the traditional measure of a performer's sensory efficiency) and seemed a more appropriate measure when false alarms become low as they do in some of the present conditions.

We use as a measure of response bias B''_D (Donaldson 1992; See, Warm, Dember, and Howe 1997). B''_D ranges from -1.00 (liberal) to 1.00 (conservative), with a value of 0.00 representing no bias. In particular, B''_D seemed more appropriate than other parametric and nonparametric measures of response bias in that it maintains measurement effectiveness over the full range of sensitivity from chance levels to perfect performance (See et al. 1997).

Single-Task Training Determination of Task Speed
Perceptual sensitivity.
Initial perceptual sensitivity (A') scores were .955 and .868 for the young and older adults, respectively, averaged over both task components. By the end of single-task training, detection sensitivity was roughly equivalent for the younger and older adults (.989 and .971, respectively). A mixed design analysis of variance (ANOVA) was performed on these data. Age (young, older) was the between-subjects factor. Young adults' perceptual sensitivity was higher, F(1,44) = 45.47, p < .0001, MSE = 3.46, both young and old adults improved detection performance with practice, F(8,352) = 54.7, p < .0001, MSE = 0.898; however, older adults benefited more overall from training, F(8,352) = 13.80, p < .0001, MSE = 0.898. There was no effect of component task (word search, pattern search) or any interactions involving this factor. Over the course of single-task training, average frame times decreased to 229 ms for the young participants on the word search task component and to 213 ms on the pattern search task component. Frame times decreased to 301 ms and 303 ms, respectively, for the older adults. Thus, as expected, older adults required longer display times, but initial age-related differences in detection performance were greatly reduced through the adaptive-training procedure.

Initial Dual-Task Costs
Although age differences in single-task detection sensitivity were minimal, older adults experienced a disproportionate decline in detection sensitivity during initial dual-task performance, as exhibited by a significant Age x Test Time (end of single-task training, beginning of dual-task training) interaction, F(1,46) = 15.29, p = .0003, MSE = 2.83. However, both young and older adults strategically responded to initial dual-task processing demands by protecting word search performance at the expense of pattern search sensitivity, as indicated by a significant Test Time x Task Component interaction, F(1,46) = 8.07, p = .0063, MSE = 0.003. Age did not moderate this interaction.

Dual-Task Training
Correct detections.
The overall correct detections data from dual-task training are presented in Table 2 . An analysis of these data revealed a significant main effect of age, F(1,46) = 116.49, p < .0001, MSE = 50.34, as older adults performed more poorly across dual-task training than their younger counterparts. Task speed (slow, medium, or fast) moderated performance, F(2,92) = 54.76, p < .0001, MSE = 3.65. Task speed, however, did not interact with age. Both young and older adults improved detection performance over the course of dual-task training, F(11,506) = 53.56, p < .0001, MSE = 1.54, and initial performance differences were reduced with practice, F(11,506) = 4.98, p < .0001, MSE = 1.54.

Pattern search performance, in general, was not as proficient as word search performance for either young or older adults. However, the young adults detected a greater number of targets than their older counterparts, F(1,46) = 92.08, p < .0001, MSE = 52.04, and task speed moderated overall performance, F(2,92) = 72.53, p < .0001, MSE = 4.39. Unlike performance on the word search component, older adults were penalized more by display rate than their younger counterparts, F(2,92) = 3.15, p < .05, MSE = 4.39. The age-related effect of processing time, however, was eliminated (see below) by the end of dual-task practice.

Once again, detection performance improved over the course of dual-task training for both young and older adults F(11,506) = 26.50, p < .0001, MSE = 2.10, and older adults improved slightly more with practice, F(11,506) = 3.09, p = .0005, MSE = 2.10. Analysis of final level performance on the pattern search component revealed a significant main effect of age, F(1,46) = 45.93, p < .0001, MSE = 5.79, and a significant main effect of Task Speed F(2,92) = 22.25, p < .0001, MSE = 1.72. At the end of dual-task practice, however, age and task speed did not interact.

Perceptual sensitivity.
Overall perceptual-sensitivity measures for each block of dual-task training are presented in Table 2 . An analysis of these data revealed a significant main effect of age, F(1,46) = 87.04, p < .0001, MSE = 0.030, as young adults performed at a higher level of sensory efficiency than their older counterparts. Task speed moderated overall performance, F(2,92) = 45.58, p < .0001, MSE = 1.72. Age, however, did not interact with task speed. Both young and older adults improved over the course of dual-task training, F(11,506) = 50.66, p < .0001, MSE = 0.77, and initial age-related differences were reduced with practice, F(11,506) = 5.00, p < .0001, MSE = 0.77.

Further analysis revealed a main effect of task component, F(1,46) = 118.85, p = .0001, MSE = 0.03, and a Task Component x Age interaction, F(1,46) = 19.89, p = .0001, MSE = 0.03. Both young and older adults demonstrated higher detection sensitivity for the word search task. However, the difference between word and pattern search performance was greater for older adults, accounting for the interaction. The analysis also uncovered an interaction between task speed and task component, F(2,92) = 37.19, p = .0001, MSE = 0.004, and a Task Speed x Task Component x Age 3-way interaction, F(2, 92) = 5.61, p = .005, MSE = 0.004.

A separate analysis on the word search task component revealed a main effect of task speed, F(2,92) = 12.81, p = .0001, MSE = 0.002; however, the Age x Task Speed interaction was not significant. An analysis on the pattern search task component revealed a main effect of task speed, F(2,92) = 49.18, p = .0001, MSE = 0.007, and a Task Speed x Age interaction, F(2,92) = 5.31, p = .006, MSE = 0.007. Analysis of final level pattern search performance, however, revealed no interaction between age and task speed. These data, therefore, mirror the correct detections data, further suggesting that the greater effects of processing speed for older adults were reduced by the end of dual-task training.

Response bias.
Criterion measures (B''_D) for each block of dual-task training are presented in Table 2 . These measures were calculated from correct detections and false alarms from each block of training. Bias is reported in terms of responding "target present." These data revealed that older participants were more conservative in responding during dual-task performance than their younger counterparts, F(1,46) = 31.55, p < .0001, MSE = 0.29. Faster task speeds resulted in more conservative responding, F(2,92) = 11.64, p < .0001, MSE = 0.039. As with the overall perceptual sensitivity data, however, task speed and age did not interact. Additional analyses revealed that both young and older adults were more conservative in responding to the pattern search task component, F(1,46) = 80.93, p = .0001, MSE = 0.09. However, there was no significant Age x Task Component interaction.

End of Training Single-Task Performance
Overall, the young adults attained an average A' of .986 over the six blocks of single-task training given at the end of the third training session. The older adults attained an average A' of .965. Although there was still a significant main effect of age, F(1,46) = 15.67, p = .0003, MSE = 1.95, there is no clear indication that either young or older adults significantly improved stimulus discriminability over the course of dual-task training. This explanation could possibly account for the improvement observed in the present dual-task data. However, it is clear from these results, as well as the results of previous dual-task search training studies (i.e, Schneider and Fisk 1982), that stimulus discriminability is not the sole arbiter of improved dual-task performance. As with the dual-task data, the older adults were slightly more conservative than their younger counterparts in terms of criterion to respond (B''_D) averaging .70 over the single-task performance measure. The young adults averaged .66. This difference, however, was not significant, F(1,46) = 1.64, p > .20.

Summary of Dual-Task Training Data
Age-related differences in speed of perceptual processing were minimized through individual determination of exposure rate to the stimuli during single-task practice. However, older adults were still disproportionately affected by the requirement to perform both search tasks concurrently. When measured in correct detections, younger adults experienced a 26% decline in performance from single- to dual-task processing. Older adults, on the other hand, experienced a 55% decline in performance. Another significant finding from the dual-task data was the older adults' adoption of a more conservative response criterion during dual-task performance. Although older adults are often more conservative than their younger counterparts, this difference was not significant during single-task conditions, suggesting that older adults adopted a general strategy of reduced responding during dual-task performance possibly as a result of their reduced ability to divide attention sufficiently over both task displays. Finally, as predicted, practice improved both older and young adults' dual-task performance.

One problem with the present dual-task training data was the difficulty of interpreting Task Speed x Practice Block interactions. This difficulty resulted from our blocked training procedure (participants received six sequential blocks of practice on each of the three task speeds during each dual-task training session). Thus, we have concentrated our analysis on final level dual-task performance because of its importance in interpreting the dual-task retention data.

Another potential problem with the present data concerns the possibility of ceiling effects for the young adults at the end of single-task training (making difficult our interpretation of the age-related costs associated with initial dual-task processing). We evaluated this possibility by conducting a set of parallel analyses on the single-task and dual-task data using only the bottom one-third of young adult performers at the end of single-task training compared with the older adults.

Using this select sample of young adult participants, differences in perceptual sensitivity between young and older adults at the end of single-task training were minimal (.978 for the young adults, and .971 for the older adults). As pointed out by an anonymous reviewer, the A' value of .978 still seems rather close to the maximum of 1.0. However, it should be noted that A' is designed to be very sensitive at extreme levels of accuracy (Craig 1979). Thus, the value of .978 represents a substantial difference of 0.5 d' units from overall young adult performance (.989) and 2.4 d' units from "near perfect" performance.

During initial dual-task training, this sample of young adults declined in overall perceptual sensitivity (A' = .914 on the first block of dual-task performance), whereas older adults declined a significantly greater amount (A' = .805), F(1,30) = 11.76, p = .0018, MSE = 2.7. Thus, even when using the lowest performing young adults, there was still a large age-related cost associated with initial dual-task performance.

A similar parallel analysis of perceptual sensitivity measures was conducted across the entire dual-task training period using the bottom one-third of younger adult performers from the first block of dual-task training. These younger adults had a mean A' during initial dual-task performance of .844. Although still better than the older adults' average (A' = .805), these young adults were performing lower than the overall group average of young adults on the first block of dual-task training (A' = .908). The subsequent Age x Practice Block interaction, however, was not significant, F(11,330) = 1.52, p = .12, suggesting that the Age x Practice Block interaction in previous analyses occurred because younger adults may have initially been closer to single-task performance levels during dual-task training. Dual-task performance improved for both young and older adults; however, the age-related performance decrement was not reduced significantly with practice.

	Retention Results

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
The results from the training phase of this study indicated that we can meaningfully evaluate age-related characteristics of single- and dual-task retention performance. Evaluation of the single-task data allows estimates of retention performance for the individual task components. If the present single-task data are consistent with previous data evaluating age-related retention capability in visual search, then more confidence can be placed in the reliability of the dual-task assessment (in the sense that the participants at least behave in the single tasks as did participants in previous studies). We therefore take a quick look at the single-task data prior to the main focus of the study, which is the dual-task data.

Single-Task Retention
Correct detection, perceptual sensitivity, and response criterion measures from the end of the single-task performance measure and the beginning of single-task retention are presented in Table 3 . These data were, indeed, consistent with previously reported reaction-time data for visual-search retention performance (Fisk et al. 1994, Fisk et al. 1995). Analysis of performance, as measured by perceptual sensitivity (A') indicated an overall decline in detection performance from training to retention, F(1,46) = 19.27, p = .0002, MSE = .001. This decline was greatest for the older adults, F(1,46) = 17.74, p = .0002, MSE = 0.001. There essentially was no decrease in overall perceptual sensitivity for the young participants over the course of the retention interval (A' declined from .982 to .981). The older participants, however, experienced a significant decline in perceptual sensitivity with a change in A' from .974 to .927 (about the equivalent of 1 d' unit). However, sensitivity at retention was still superior to initial single-task performance levels. Analysis of the response-criterion data suggested that both the older and young adults became more conservative in their responding over the course of the retention interval, F(1,46) = 5.34, p = .03, MSE = 0.04. The main effect of age, however, was not significant.

View larger version (18K): [in this window] [in a new window]   Figure 2. Mean perceptual sensitivity as measured by A' at the end of dual-task training and the beginning of retention testing for the young adults (A) and for the older adults (B) as a function of task speed. Also depicted is response criterion (B''D) at the end of dual-task training and the beginning of dual-task retention testing for the young (C) and older (D) adults.

When we looked at initial retention performance for the word search task component only, we discovered that there was very little difference between young and older adults in response criterion (no main effect of age, p > .10). However, when we looked at pattern search task performance, we discovered a significant main effect of age, F(1,46) = 14.75, p = .0004, MSE = 0.016, and a significant Age x Test Time interaction, F(1,46) = 4.96, p = .03, MSE = 0.009. Whereas older adults became more liberal in responding for both task components, young adults actually became more conservative during retention performance for the pattern search task component, accounting for the Age x Test Time interaction.

Summary of Retention Data
Older adults, as expected, experienced a greater decline in perceptual sensitivity than their younger counterparts, as indexed both by single-task and dual-task performance measures. Although it is possible that ceiling effects could have masked greater stimulus-specific (sensitivity) decline for the younger adults during retention performance, this explanation would be unlikely to account for the present data. Disproportionate declines in stimulus-specific learning after similar retention intervals have been observed across a wide variety of search/detection tasks (Fisk et al. 1994, Fisk et al. 1995; Batsakes and Fisk 1997). Thus, an understanding of the single-task retention data is rather clear. An understanding of the dual-task data, however, is more complex. Whereas dual-task sensitivity declined more for older adults over the retention interval, skilled performance, at least as measured by correct detections, declined equivalently for both young and older adults. Furthermore, evidence that learning associated with skilled dual-task processing remained impressively intact for older adults comes from the lack of any disproportionate decline in performance for one component task over another. Thus, older adults retained the flexibility in distributing attention over both task components gained during dual-task training. Finally, the surprising finding was the increase in response rate during dual-task performance for older adults after the retention interval, resulting in increased false alarms.

	Discussion

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 
The purpose of this study was to (a) examine age-related differences in dual-task performance when task speed (processing rate; Schneider and Shiffrin 1977) was calibrated for each individual participant and (b) evaluate the replicability of retention capability found in previous single-task, visual search experiments and extend those findings to the retention of dual-task performance. The present data are interesting for several reasons. First, the data set is unique because of the extended interval for assessing retention of dual-task performance. Second, even though age-related differences were observed, the level of retention performance shown in this study suggests that both younger and older adults are capable of retaining an impressive level of skill for relatively complex tasks. And third, the pattern in the present retention data replicate age-related findings for less demanding search-detection tasks, suggesting that a consistent relationship may emerge for predicting age-related retention capabilities.

The first question concerned whether there are age-related differences in dual-task performance. Of course, numerous studies before this one have documented age-related, dual-task performance decrements (for reviews, see Hartley 1992; Kramer and Larish 1996). In the present study, however, we found greater age-related costs associated with dual-task processing, even when individual differences in single-task performance were minimized through the adaptive single-task training procedure. When transferred to the dual-task, older adults' performance was more disrupted than younger adults' performance whether indexed by measures of perceptual sensitivity or correct detections. Given that display processing time was individually determined for each participant, this main effect of age in dual-task performance is surprising if it is assumed that processing rate (from a general slowing perspective) is the sole arbiter of age-related differences in dual-task performance. Also surprising is the finding that the speed manipulation did not interact with age by the end of only two sessions of dual-task practice, even though task speed did moderate overall performance. Thus, something beyond a single resource must be accounting for the age-related, dual-task performance differences.

An anonymous reviewer suggested that a nonlinear function representing the relationship between age groups in processing speed might still account for the present dual-task data. However, even this explanation seems unlikely given the lack of an effect of task speed after practice and the complex pattern of results exhibited in the retention data (see below). An alternative explanation is that dual-task performance is moderated by age-related declines in the processes necessary for the management and coordination of multiple component tasks (Kramer et al. 1995, Kramer et al. 1999). Improvement in dual-task performance with practice is often associated with increased efficiency in these task control structures, especially when stability in single-task performance can be demonstrated. In addition, the dual-task training data demonstrated that, in at least this class of dual-tasks, older adults seem more conservative in their willingness to respond. Thus, the findings from the training phase of the study add to the growing evidence that, to fully understand age-related differences in dual-task performance, the understanding of task control parameters such as executive processes (Kramer et al. 1999), task integration (Korteling 1991, Korteling 1993), and response control (Schneider and Detweiler 1988) must be considered.

The second question concerned the replicability of retention capability found in previous single-task, visual search experiments. In the current single-task data, as in those previous studies (see Fisk et al. 1994, Fisk et al. 1995), older adults showed more decline in performance than young adults after a 1-month delay. The previous studies had used reaction time as the dependent variable. Here we demonstrated the same pattern of results using signal detection measures. The present procedure localized the single-task findings to perceptual sensitivity but not response bias, consistent with the conceptual model proposed by Fisk and colleagues 1995. Here we found more dramatic performance declines for the older adults compared with younger adults for correct detections, which were caused by a decline in the ability to distinguish targets from distractors (perceptual sensitivity) but not by a differential age-related change in response bias. Not performing the search task over a 1-month period did produce more conservative responding, but the bias for the single-task did not change differentially as a function of age.

The second question was also informed by the present dual-task retention data. When performance was assessed by means of correct detections, we did not find a differential age-related effect over the retention interval. Performance declined, but in a manner similar for both young and older adults. When the ability to differentiate targets from distractors was assessed (A', a measure related to the stimuli and not the strategy), then, as with single-task performance, older adults demonstrated more of a decline compared with their younger counterparts. Most intriguing is the finding that the retention interval differentially affected older and younger adults' response criterion. Older adults' response criterion became more liberal compared with their response criterion during training. The older adults were thus able to maximize correct detections at the expense of a greater number of false alarms.

The retention results suggest that the nature of learning during dual-task training can be specific to the trained stimuli (see also Schneider and Fisk 1984); however, the learning is also strongly related to task control factors. We have referred to this latter type of learning as task-specific learning to be consistent with previous studies (e.g., Fisk et al. 1994, Fisk et al. 1995). Our present findings and learning dichotomy is in line with the general attention literature (for reviews, see Gopher 1982; Schneider and Detweiler 1988; Shiffrin 1988). Although perceptual sensitivity did decline for older adults during dual-task retention performance, it was clear that skill at dividing attention over the two task displays remained remarkably intact for both age groups. Instead, perceptual sensitivity declined because of an ability to distinguish targets from distractors. Older adults responded by becoming more willing to respond to nontargets (overall target detections remained the same as during training, only the number of false alarms increased over the retention interval). This pattern of data suggests a general shift in strategy among the older participants. This strategy shift was reflected in a more liberal response bias for the older adults.

One possible explanation for older adults' shift in response criterion during dual-task retention testing is that they adjusted control parameters during dual-task retention testing as a function of their perceived sensitivity to the trained stimuli. Such an explanation stresses the importance of understanding age-related differences in strategic and metacognitive processes during performance of complex tasks (see Hertzog and Hultsch 2000). For instance, in another study (also adopting a dual-task procedure), Baron and Mattila 1989 demonstrated that when speed of responding was highly stressed, older adults were willing to lower accuracy to maximize performance. This led to an overall reduction in the age-related dual-task decrement often observed among cognitive aging researchers. Thus, understanding task control adjustments in response to sensitivity declines among older adults presents a fruitful arena for further research, especially when assessing complex task performance.

	Acknowledgments

 
This research was conducted as partial fulfillment of the requirements for the Masters of Science degree by Peter J. Batsakes. The work was supported by National Institutes of Health (National Institute on Aging) Grants R01AG07654 and P50AG11715. We thank C. Hertzog, W. Rogers, and W. Schneider for critical review and comments on an earlier version of this article.

Received for publication January 13, 1999. Accepted for publication March 9, 2000.

	References

TOP Abstract Age-Related Dual-Task... Overview of Present Study Methods Training Results Retention Results Discussion References

 

• Baron A., Mattila W., 1989. Response slowing of older adults: Effects of time-limit contingencies on single- and dual-task performance. Psychology and Aging 4:66-72. [Medline]
• Batsakes, P. J., & Fisk, A. D. (1997). Aging and retention of newly acquired skill: Does performance predict retention of underlying learning? In Proceedings of the Human Factors and Ergonomics Society 42^nd Annual Meeting. Santa Monica, CA: Human Factors and Ergonomics Society.
• Battig, W. F., & Montague, W. E. (1969). Category norms for verbal items in 56 categories: A replication and extension of the Connecticut category norms. Journal of Experimental Psychology Monographs, 80, (3, Whole No. 3).
• Birren J. E., 1974. Translations in gerontology—from lab to life: Psychophysiology and speed of response. American Psychologist 29:808-815. [Medline]
• Craig A., 1979. Nonparametric measures of sensory efficiency for sustained monitoring tasks. Human Factors 21:69-77. [Medline]
• Damos D. L., Bittner A. C., Kennedy R. S., Harbeson M. M., 1981. Effects of extended practice on dual-task tracking performance. Human Factors 23:627-631.
• Detweiler M. C., Lundy D. H., 1995. Effects of single- and dual-task practice on acquiring dual-task skill. Human Factors 37:193-211.
• Donaldson W., 1992. Measuring recognition memory. Journal of Experimental Psychology: General 121:275-277.
• Fisk A. D., Cooper B. P., Hertzog C., Anderson-Garlach M. M., 1995. Age-related retention of skilled memory search: Examination of associative learning, interference, and task-specific skills. Journal of Gerontology: Psychological Sciences 50B:P150-P161. [Abstract]
• Fisk A. D., Hertzog C. K., Lee M. D., Rogers W. A., Anderson-Garlach M., 1994. Long-term retention of skilled visual search: Do young adults retain more than old adults?. Psychology and Aging 9:206-215. [Medline]
• Gopher D., 1982. A selective attention test as a predictor of success in flight training. Human Factors 24:173-184. [Medline]
• Green D. M., Swets J. A., 1966. Signal detection theory and psychophysics Wiley, New York.
• Hartley A. A., 1992. Attention. Craik F. I. M., Salthouse T. A., , ed.The handbook of cognition and aging 3-49. Erlbaum, Hillsdale, NJ.
• Hertzog C., Hultsch D. F., 2000. Metacognition in adulthood and old age. Craik F. I. M., Salthouse T. A., , ed.The handbook of cognition and aging II 417-466. Erlbaum, Mahwah, NJ.
• Korteling J. E., 1991. Effects of skill integration and perceptual competition on age-related differences in dual-task performance. Human Factors 33:35-44. [Medline]
• Korteling J. E., 1993. Effects of age and task similarity on dual-task performance. Human Factors 35:99-114. [Medline]
• Kramer A. F., Larish J. L., 1996. Aging and dual-task performance. Rogers W., Fisk A. D., Walker N., , ed.Aging and skilled performance: Advances in theory and application 83-112. Erlbaum, Mahwah, NJ.
• Kramer A. F., Larish J. L., Strayer D. L., 1995. Training for attentional control in dual-task settings: A comparison of young and old adults. Journal of Experimental Psychology: Applied 1:50-76.
• Kramer A. F., Larish J. L., Weber T. A., Bardell L., 1999. Training for executive control: Task coordination strategies and aging. Gopher D., Koriat A., , ed.Attention and performance XVII. Cognitive regulation of performance: Interaction of theory and application 617-652. MIT Press, Cambridge, MA.
• Lorsbach T. C., Simpson G. B., 1988. Dual-task performance as a function of adult age and task complexity. Psychology and Aging 3:210-212. [Medline]
• McDowd J. M., Craik F. I. M., 1988. Effects of aging and task difficulty on divided attention performance. Journal of Experimental Psychology: Human Perception and Performance 14:267-280.
• Rogers W. A., Bertus E. L., Gilbert D. K., 1994. Dual-task assessment of age differences in automatic process development. Psychology and Aging 9:398-413. [Medline]
• Rogers W. A., Fisk A. D., 1991. Are age differences in consistent-mapping visual search due to feature learning or attention training?. Psychology and Aging 6:542-550. [Medline]
• Rogers, W. A., Gilbert, D. K., & Fisk, A. D. (1994, April). Long-term retention of general skill and stimulus-specific abilities in associative learning: Age-related differences. Paper presented at the fifth Cognitive Aging Conference, Atlanta, GA.
• Salthouse T. A., 1982. Adult cognition: An experimental psychology of aging Springer-Verlag, New York.
• Salthouse T. A., 1985. A theory of cognitive aging North Holland, Amsterdam.
• Salthouse T. A., 1989. The role of processing resources in cognitive aging. Howe M. L., Brainerd C. L., , ed.Cognitive development in adulthood 185-239. Springer-Verlag, New York.
• Salthouse T. A., Hambrick D. Z., Lukas K. E., Dell T. C., 1996. Determinants of adult age differences on synthetic work performance. Journal of Experimental Psychology: Applied 2:305-329.
• Salthouse T. A., Rogan J. D., Prill K. A., 1984. Division of attention: Age differences on a visually presented memory task. Memory and Cognition 12:613-620. [Medline]
• Schneider W., 1988. Micro experimental laboratory: An integrated system for IBM PC compatibles. Behavior Research Methods Instrumentation and Computers 20:206-217.
• Schneider W., Detweiler M. C., 1987Vol. 21,. A connectionist/control architecture for working memory. Bower G. H., , ed.The psychology of learning and motivation 53-119. Academic Press, Orlando, FL.
• Schneider W., Detweiler M. C., 1988. The role of practice in dual-task performance: Toward workload modeling in a connectionist/control architecture. Human Factors 30:539-566. [Medline]
• Schneider W., Fisk A. D., 1982. Concurrent automatic and controlled visual search: Can processing occur without resource cost?. Journal of Experimental Psychology: Learning Memory and Cognition 8:261-278.
• Schneider W., Fisk A. D., 1984. Automatic category search and its transfer. Journal of Experimental Psychology: Learning Memory and Cognition 10:1-15. [Medline]
• Schneider W., Shiffrin R. M., 1977. Controlled and automatic human information processing: I. Detection, search, attention. Psychological Review 84:1-66.
• See J. E., Warm J. S., Dember W. N., Howe S. R., 1997. Vigilance and signal detection theory: An empirical evaluation of five measures of response bias. Human Factors 39:14-29.
• Shiffrin R. M., 1988. Attention. Atkinson R. C., Herrstein R. J., Luce R. D., , ed.Steven's handbook of experimental psychology 739-811. Wiley, New York.
• Sit R. A., Fisk A. D., 1999. Age-related performance in a multiple-task environment. Human Factors 41:26-34. [Abstract/Free Full Text]
• Somberg B. L., Salthouse T. A., 1982. Divided attention abilities in young and old adults. Journal of Experimental Psychology: Human Perception and Performance 8:651-653.
• Sperling G., 1984. A unified theory of attention and signal detection. Parasuraman R., Davies D. R., , ed.Varieties of attention 103-181. Academic Press, New York.
• Whaley C. J., Fisk A. D., 1993. Effects of part-task training on memory set unitization and retention of memory-dependent skilled search. Human Factors 35:639-652.

This Article Abstract Full Text (PDF) Services Download to citation manager PubMed PubMed Citation