Investigation of age-related differences in an adapted Hayling task

Archives of Gerontology and Geriatrics 59 (2014) 599–606

Contents lists available at ScienceDirect

Archives of Gerontology and Geriatrics journal homepage: www.elsevier.com/locate/archger

Investigation of age-related differences in an adapted Hayling task Isabelle Tournier a,b,*, Virginie Postal a, Ste´phanie Mathey a a b

Laboratoire de Psychologie, Sante´ et Qualite´ de Vie, Universite´ Bordeaux Segalen, France INSIDE, University of Luxembourg, Luxembourg

A R T I C L E I N F O

A B S T R A C T

Article history: Received 6 December 2013 Received in revised form 28 July 2014 Accepted 29 July 2014 Available online 6 August 2014

The Hayling task is traditionally used to assess activation and inhibitory processes efficiency among various populations, such as elderly adults. However, the classical design of the task may also involve the influence of strategy use and efficiency of sentence processing in the possible differences between individuals. Therefore, the present study investigated activation and inhibitory processes in aging with two formats of an adapted Hayling task designed to reduce the involvement of these alternative factors. Thirty young adults (M = 20.7 years) and 31 older adults (M = 69.6 years) performed an adapted Hayling task including a switching block (i.e., unblocked design) in addition to the classical task (i.e., blocked design), and the selection of the response between two propositions. The results obtained with the classical blocked design showed age-related deficits in the suppression sections of the task but also in the initiation ones. These findings can be explained by a co-impairment of both inhibition and activation processes in aging. The results of the unblocked Hayling task, in which strategy use would be reduced, confirmed this age-related decline in both activation and inhibition processes. Moreover, significant correlations between the unblocked design and the Trail Making Test revealed that flexibility is equally involved in the completion of both sections of this design. Finally, the use of a forced-response choice offers a format that is easy to administer to people with normal or pathological aging. This seems particularly relevant for these populations in whom the production of an unrelated word often poses problems. ß 2014 Elsevier Ireland Ltd. All rights reserved.

Keywords: Aging Activation Inhibition Hayling task

1. Introduction The Hayling task was created by Burgess and Shallice (1996) to explore the initiation and suppression of expected semantic responses in patients suffering from frontal lobe lesions. It comprises two sections of 15 sentences in which the last word is omitted but is strongly cued by the semantic context of the sentence. As a result, the missing word is automatically activated during reading (e.g., ‘‘The child is crying for his. . . mother’’). In section A (response initiation), the participants are required to provide an appropriate word to complete each sentence. In this response initiation section, the context and spreading activation in semantic memory cause the automatic activation of the missing word. On the contrary, in section B (response suppression), the

* Corresponding author at: Integrative Research Unit on Social and Individual Development, University of Luxembourg, Route de Diekirch BP2, L-7220 Walferdange, Luxembourg. Tel.: +352 46 66 44 9298; fax: +352 46 66 44 39298. E-mail addresses: [email protected], [email protected] (I. Tournier). http://dx.doi.org/10.1016/j.archger.2014.07.016 0167-4943/ß 2014 Elsevier Ireland Ltd. All rights reserved.

participants are required to provide a word that makes no sense in the context to complete each sentence. Therefore, the production of an unrelated word in the response suppression section involves inhibiting the word automatically activated by the semantic context. The Hayling task is frequently used to assess semantic inhibition performances in various populations such as the elderly (Be´langer & Belleville, 2009), depressed (Gohier et al., 2009), or alcohol-dependent adults (Noe¨l et al., 2013). However, findings are not always congruent, especially in the aging domain. In the initiation section, several studies have shown an impaired performance (Bielak, Mansueti, Strauss, & Dixon, 2006; Collette, Schmidt, Scherrer, Adam, & Salmon, 2009; Lin, Chan, Zheng, Yang, & Wang, 2007), whereas others did not show any age difference (Andre´s & Van der Linden, 2000; Belleville, Rouleau, & Van der Linden, 2006; Borella, Ludwig, Fagot, & de Ribaupierre, 2011). Regarding the suppression section, most of the studies suggest an impaired performance in older adults (Andre´s & Van der Linden, 2000; Belleville et al., 2006; Bielak et al., 2006; Collette et al., 2009). However, others failed to observe any age effect or even showed longer response times in young adults than in older ones (Borella, Delaloye, Lecerf, Renaud, & de Ribaupierre, 2009; Borella et al.,

600

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

2011). Thus, most of the results suggest a decline in the suppression section during aging, whereas the data about modification in the initiation section are less consensual. Variations in populations studied (older adults vs. young adults, older adults vs. very old adults) or in paradigms used (sentences read or heard, format) could explain these discrepant findings. Moreover, as presented in the next section, the current Hayling task format is thought to have some limitations. 1.1. Adapted formats of the Hayling task A limitation of the traditional Hayling task is the possibility to use non-inhibitory processes to perform the suppression section, such as anticipating the answer prior to hearing the sentence or not paying attention to the sentence (Burgess & Shallice, 1996). Although Burgess and Shallice (1996) considered the possible use of strategies as the consequence of good executive functioning, their utilization reduces the involvement of inhibition (Andre´s & Van der Linden, 2004). Moreover, the efficiency of strategy use tends to decline with aging (Lemaire, 2010) and could penalize elderly adults. To counteract the use of strategies, Be´langer and Belleville (2009) proposed an adapted version of the task with an unblocked design in which the initiation and suppression sections were presented in a randomized order. A cue was provided upon completion of the sentence presentation to instruct participants about the task to be done (i.e., give an appropriate or inappropriate word). The authors’ goal was to ensure that participants process the sentence completely and do not select their response in advance, which is possible in the blocked design. With this adapted Hayling task, Be´langer and Belleville tested young adults, older adults, patients with mild cognitive impairment (MCI), and patients with a diagnosis of Alzheimer’s disease. Their results showed that older adults had longer response times and made more errors than younger adults on the sentences associated with the suppression instruction only. Interestingly, MCI patients showed poorer performance than older adults on suppression instruction, whereas a previous study using the classic Hayling task indicated normal performance in this population (Belleville, Chertkow, & Gauthier, 2007). The authors conceded that their new procedure increased the complexity of the task by requiring the task’s goal to be maintained in working memory and by switching between initiation and suppression instructions. However, they considered that working memory and switching were equally involved in both sections and could not explain the impairments of MCI patients in the suppression section. Although these results are very promising, the absence of blocked sections, such as in the classical Hayling task, restricts comparisons between the two designs. Consequently, it would be interesting to propose an adapted Hayling task combining the classical presentation (i.e., blocked design) and the alternated presentation (i.e., unblocked design). Another limitation of the Hayling task mentioned by Andre´s and Van der Linden (2004) concerns the processing occurring during the suppression section. The very action of producing an inappropriate word requires the inhibition of the appropriate word and the complex search of a semantically unrelated word. The complexity of the suppression section is supported by neuroimaging data using positron emission tomography, showing an increased activity in middle and inferior frontal areas generally associated with complex processes such as planning or updating (Collette et al., 2001). The difficulty of producing an alternative response in the Hayling task is due to the fact that the response is not provided in the stimulus display (such as in the Stroop task) and requires being internally generated (Castner et al., 2007). The internal generation of an alternative response may be avoided by proposing a response choice. For example, by asking the

participant to choose the inappropriate word between words on the computer screen, the alternative response is provided in the stimulus display. Delaloye et al. (2009) proposed this modification and used a task adapted from the Hayling task where two words appeared on the screen after the participants had heard the sentence. In the first part, participants chose as quickly as possible the word that completed the sentence correctly, and in the second part, they chose as quickly as possible the word that completed the sentence incorrectly. The fitting word and the competitor were comparable regarding lexical frequency, grammatical class, gender, number, and number of letters. Results obtained by Delaloye et al. (2009) with this adapted Hayling task suggest a decrease in inhibition with aging. These adapted Hayling tasks have been proposed to improve the assessment of inhibition processes. However, other cognitive processes such as speed of processing and working memory might also influence Hayling task performance. 1.2. Cognitive processes associated with completion of the Hayling task Processing speed and working memory decline has been regularly hypothesized as explanations of cognitive changes observed with aging (e.g., Hasher, Zacks, & May, 1999; Salthouse, 1991). These assumptions are central in the literature on cognitive aging, given that decline in processing speed (e.g., Finkel, Reynolds, McArdle, & Pedersen, 2007; Lemke & Zimprich, 2005) and working memory (e.g., Hertzog, Dixon, Hultsch, & MacDonald, 2003; Salthouse, 1994a) with aging is regularly reported. Although these two variables have been rarely adjusted statistically in aging studies using the Hayling task, a few data indicate that they influence performance (Andre´s & Van der Linden, 2000; Collette et al., 2009). Andre´s and Van der Linden (2000) reported that after the statistical control of processing speed, the positive correlation between age and the suppression section became non-significant for response latencies but remained significant regarding error score. Moreover, Collette et al. (2009) observed that the age effect disappeared in the initiation section but not in the suppression one when span size and processing speed were taken as confounding covariates. Thus, it seems important to take into account processing speed and working memory in case of age difference when assessing age-related changes in the Hayling task. Moreover, it seems reasonable to assume that processing speed and working memory might also impact the processing of sentences during the task. Indeed, previous studies suggest an involvement of age-related slowing in the increase in reading time (Postal & Mathey, 2007) and listening difficulties (Tun, McCoy, & Wingfield, 2009) during aging. Likewise, older adults with high working memory span were faster readers than lower-span older adults (Kemtes & Kemper, 1997). Most of the Hayling task studies to date have used a blocked presentation of the sentences, the latter being read aloud by the experimenter or being recorded. One study allowed the participants to read the sentences at their own pace, but reading and response times were not distinguished and the study failed to find any age effect on inhibition (Borella et al., 2011). The fact that the participants did not control the presentation rate of sentences in most of the studies could have two different negative effects: (a) a too short a time could disturb the correct processing of the sentence, especially in older adults (i.e., less inhibition necessary because weakened activation), and (b) too long a presentation time could induce supplementary cognitive processes not controlled by the experimenter, especially in young adults (i.e., use of strategies limiting the need for inhibition). Consequently, even if the sentences in the Hayling Task are relatively simple and strongly constrained regarding context, it would be interesting if sentence presentation was self-paced,

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

thereby allowing participants to control their speed of reading. Moreover, asking participants to read the sentences aloud limits the use of a strategy that consists in avoiding paying attention to them (Burgess & Shallice, 1996). Finally, although Be´langer and Belleville (2009) consider that the use of an unblocked design may add a switching component, the latter should be equally involved in the completion of both sections. The Trail Making Test (Reitan, 1958) is one of the most commonly used tests in neuropsychological practice to assesses flexibility (Kortte, Horner, & Windham, 2002; Shibuya-Tayoshi et al., 2007). It is composed of two parts: Part A involves connecting circles with numbers in numerical order (i.e., 1-2-3-. . .-24-25) whereas part B requires alternately connecting circles with numbers and circles with letters (i.e. 1-A-2-B-. . ..-L-13). Part B, which assesses flexibility after the adjustment of Part A (Crowe, 1998; Kortte et al., 2002), is linked to slower performances with aging (Seo et al., 2006; Soukup, Ingram, Grady, & Schiess, 1998; Tun, O’Kane, & Wingfield, 2002). An investigation of the link between the Trail Making Test and the two formats of the Hayling task (i.e., blocked and unblocked designs) would give insights into the involvement of cognitive flexibility in each part of the task. 1.3. The present study The main aim of the present study was to investigate activation and inhibitory processes underlying age-related differences in two formats of the Hayling task. To do this, we compared the age effect in the classical design of the task (i.e., blocked design; Part A and B) and a switching design (i.e., unblocked design) used by Be´langer and Belleville (2009). We predicted that performance on the initiation and suppression sections should decline with aging in both designs. However, a smaller effect was expected in the unblocked than in the blocked design owing to non-possible use of anticipatory strategies that should especially impede young adults. According to age-related changes in processing speed and working memory regularly observed and to previous findings in the Hayling task, we decided to statistically adjust processing speed and working memory in the event of age differences. The secondary aim was to study the relationship between the unblocked design and cognitive flexibility. To do this, correlations between each design of the Hayling task (i.e., blocked and unblocked) and a flexibility measure, the Trail Making Test, were investigated. We predicted a significant correlation of the Trail Making Test with the unblocked design but not with the blocked design, owing to the necessity to switch between the initiation and suppression instructions in the unblocked design. The presence of moderate-to-strong correlations between the unblocked design and the Trail Making Test should indicate that this adapted Hayling task assesses not only activation and inhibition but also flexibility. 2. Method 2.1. Participants Thirty young (university undergraduates, range 18–25 years, M = 20.68, SD = 1.89) and 31 older (community dwelling and senior

601

university learners, range 63–80 years, M = 69.61, SD = 7.35) adults took part in this study. To screen for cognitive impairment, we included only participants with a Mini Mental State Examination (MMSE, GRECO version, Kalafat, Hugonot-Diener, & Poitrenaud, 2003) score of 26 or more (M = 28.93, SD = 1.06). The proportion of women in each age group, 73% in young adults and 65% in older adults, did not differ significantly, x2(2, N = 61) = .58 et p > .10. The two groups of participants were matched on the number of years of education (young adults, M = 13.60, SD = 1.00; and older adults, M = 12.84, SD = 2.73), t(59) = 1.43, p > .10. The level of vocabulary measured by part B of the Mill-Hill test (Deltour, 1993) was greater in the older group than in the younger group (respectively, M = 40.3, SD = 2.04 vs. M = 35.7, SD = 3), t(59) = 6.62, p < .001. 2.2. Materials and procedure 2.2.1. The Hayling switching task We used a computerized adaptation of the Hayling task to measure activation and inhibition in a semantic context. Firstly, the participants completed the traditional Hayling task (blocked design) composed of the initiation (Part A) and suppression (Part B) sections presented separately and one after another. Next, new sentences were presented but the initiation and suppression instructions were randomly presented (unblocked design). Consequently, participants could not anticipate the production of a related or unrelated word. Within each design, they read a sentence in French in which the last word was missing. All sentences were created to be highly constrained at the semantic level so that the missing words were automatically activated during reading (‘‘Les pompiers ont e´teint le . . . feu’’ [‘‘The firefighters extinguished the . . .. fire’’]). For this, 34 highly constrained sentences were used from the study of Belleville et al. (2006). Seventy new sentences were created with the help of sentence completion norms for 744 incomplete French sentences (Robichon, Besson, & Faı¨ta, 1996). All the 104 sentences were submitted to a pre-test and were read by 23 young university students aged from 18 to 29 years. For each sentence, participants were asked to give the first word that came to mind, and to estimate on a three-point Likert scale the ease of finding this word. The objective was to estimate the strength of the semantic activation and consequently the fact that sentences were highly constrained semantically. The 88 sentences completed with the same word by 100% of participants and with the best ease-ofretrieval scores were selected for the experimental task, and the next 10 sentences were used for practice. To offer two response options, competitors of the correct words were selected in a French lexical database (New, Pallier, Ferrand, & Matos, 2001). The correct and the competitor words were matched on lexical frequency, gender, grammatical class, number of letters, and number of syllables. In addition, the number of characters in the sentences was matched across the two parts. Characteristics of the materials according to each section are presented in Table 1. The sentence reading was self-administered: participants pressed the space bar when they had finished reading the sentence. In the blocked design, the first part (part A) was the initiation section, containing 5 practice items and 15 test items.

Table 1 Characteristics of materials according to each section of the Hayling switching task. Section

Target frequency

Distractor frequency

Number of letters

Number of syllables

Number of characters per sentence

Blocked design

Initiation Suppression

70.73 71.49

72.06 70.18

5.40 6.27

1.40 1.67

39.00 39.87

Unblocked design

Initiation Suppression

66.71 69.62

71.39 68.63

5.61 6.50

1.46 1.75

35.32 37.28

602

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

Participants pressed a button on a keyboard as quickly and accurately as possible to indicate which word of the two correctly completed the sentence. A green border surrounded the two words to remind participants during the task about the instructions in which to give the correct word (green = correct word). The suppression section (Part B) was proposed in the second part. Five new practice items and 15 test items were presented. Participants pressed a button on a keyboard as quickly and accurately as possible to indicate which word of the two did not correctly complete the sentence. They were informed of the normality of incongruence of the completed sentence in this part. A red border surrounded the two words to remind them during the task about the instructions in which to give the incorrect word (red = incorrect word). After the blocked presentation, the initiation and suppression sections were presented randomly in an unblocked design, with 10 practice items (i.e., the same as in activation and inhibition sections) and 58 new test items. The order was to be very attentive to the color of the frame bordering the two proposed words in order to know the relevant action: to select the correct word if the frame was green, or to select the incorrect word if the frame was red. Participants had to choose as quickly as possible the correct or incorrect word according to the instructions. Frame presentation was pseudo-randomized with 8 series of one, 10 series of two and 10 series of three items without color change (i.e., without alternating between orders) appearing in a mixed order. In each part, a fixation cross preceded presentation of the sentence. The participants had to read the sentence aloud naturally and to press the space bar immediately after reading it. The sentence was then immediately replaced by two words on the screen. They had to press the key button on the left of the keyboard to select the word presented on the left of the screen, and to press the key button on the right of the keyboard to select the word on the right of the screen. The distribution of the answer on the left and the right hand side was pseudo-randomized and no more than three successive responses on the same side were possible. Lastly, for each part, visual feedback (i.e., a blue cross) was provided in case of error or too long a response time. Four versions of the task were created to counterbalance the order (to give the correct word vs. the incorrect word) and the distribution of the response to choose (on the right vs. on the left). As dependent variables, we measured response latencies in milliseconds of correct word responses and number of errors, as well as sentence reading time in milliseconds. 2.2.2. Other cognitive measures The Boxes task. A computerized adaptation of the Boxes task (Salthouse, 1994b) was used to measure processing speed. A series of 100 squares with a missing side were presented on the computer screen. Participants indicated the missing side by pushing a button on a computer keyboard, as quickly and precisely as possible. Buttons for responses were indicated in green and were arranged to form a square, in order to avoid difficulties in memorizing response keys. A series of 20 squares was given for practice. The mean response time in milliseconds for correct responses was scored. The Operation Span Task. A computerized version of the Operation Span Task (Turner & Engle, 1989) was used to measure working memory. A mathematical operation and a to-be-remembered word were presented on a computer screen (e.g., 2  5 + 7 = 19 Boat). First, participants had to say aloud if the operation was true or not, and then had to read aloud the written word. The next operation-word pair was presented as soon as the participants read the word (see Emery, Hale, & Myerson, 2008 for the same procedure). The presentation rate of the pairs was directed by the experimenter to prevent differences in use of

strategies by the participants. At the end of each list, the cue ‘‘Recall’’ was presented and participants recalled the words in the current trial. The task was composed of 15 trials with two to six operation-word pairs, with 3 trials for each length. Trials were presented in increasing length and the task was finished after three consecutive errors. Three supplementary trials were used for practice (respective lengths of 2, 3 and 4 items). We verified that participants correctly performed at least 80% of the mathematical operations, in order to be sure that they were paying attention to the process. The dependent variable was span size. Trail Making Test. The Trail Making Test (Reitan, 1958) was used to measure cognitive flexibility. Participants were instructed to complete the task as quickly as possible and without mistakes. In the event of a mistake, the instruction was to return to the previous point and continue the task. The switching Part B of this task could not be used directly as a measure of flexibility because visual processing and motor speed have also an influence on this part and several calculations (i.e., Part B Part A; Part B/Part A; (Part B Part A)/Part A) have been proposed to control for this involvement. However, ratios (i.e., Part B/Part A; (Part B Part A)/ Part A) are assumed to have a curvilinear relationship with cognitive impairments (Corrigan & Hinkeldey, 1987; Golden, Osmon, Moses, & Berg, 1981). Consequently, the dependent variable was the difference in response times (in seconds) between part B and part A (Part B minus Part A, TMT B-A), to adjust the involvement of psychomotor speed and visual scanning (Hester, Kinsella, Ong, & McGregor, 2005). 3. Results The data were checked for outliers. These were defined as a value at least 1.5 standard deviations above or below the mean group. No more than 2% of data were concerned so we used truncation to deal with outliers. We changed the value of extreme scores to that of the next highest/lowest non-outlier. Consequently the relative order of the data was maintained and the highest or lowest scores remained the highest or lowest scores (Osborne & Overbay, 2004). After truncation, all the variables showed acceptable skewness and kurtosis. First, to detect possible age differences on processing speed and working memory, we conducted a t-test on Boxes and Operation Span Tasks. Next, a 2 (Age: young adults, older adults)  2 (Section: initiation, suppression)  2 (Design: blocked, unblocked) analysis of variance (ANOVA) was conducted. In the event of age difference on processing speed and working memory measures, the influence of these two variables on possible age effect on Hayling performance was assessed by covariance analyses. Third, correlations between each part of the Hayling task and a flexibility measure, the Trail Making Test, were conducted. 3.1. Measures of processing speed and working memory Older adults showed slower response times than young adults (respectively 811 ms and 552 ms) on processing speed, t(59) = 11.33, p < .001. Concerning working memory, the mean scores of older adults were smaller than those of young adults but not significantly different, respectively 3.19 and 3.55, t(59) = 1.51 and p > .10. Hence, only the measure of processing speed has been statistically controlled thereafter. 3.2. The Hayling switching task A 2 (Age)  2 (Section)  2 (Design) ANOVA was conducted on mean response times. Next, ANCOVAs with processing speed as a covariate were conducted on mean response times. Mean response times and estimated mean response times are presented in Table 2.

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606 Table 2 Mean response times and standard errors (in parentheses) according to age, section, design, and estimated mean response times of ANCOVA. Blocked design

Unblocked design

Initiation

Suppression

Initiation

Suppression

733 (32) 1074 (31)

780 (29) 1213 (28)

785 (40) 1387 (40)

897 (38) 1553 (38)

Speed of processing (ANCOVA) Young adults 792 (46) 823 (42) Older adults 1017 (45) 1171 (41)

828 (58) 1345 (57)

899 (56) 1551 (55)

Young adults Older adults

Given the low rate of errors (i.e., <7%), the participants were dichotomized for each condition into those who did not make any errors versus those who did, and Chi-square tests were used to analyze error rates. Analyses of response times revealed a significant main effect of age, F(1,59) = 173.74, p < .001 and h2p = 0.75, with slower response times for older than young adults (1307 ms vs. 799 ms). This age effect was reduced but remained significant when processing speed was used as covariate, F(1,58) = 40.56, p < .001 and h2p = 0.41. The main effect of section was significant, F(1,59) = 149.51, p < .001 and h2p = 0.72, with slower response times for the suppression section than the initiation section (1111 ms vs. 995 ms). The main effect of design was significant, F(1,59) = 56.56, p < .001 and h2p = 0.49, with slower response times for the unblocked design than the blocked design (1155 ms vs. 950 ms). The interaction between age and section was significant, F(1,59) = 14.90, p < .001 and h2p = 0.20, and the difference between young and older adults was greater in the suppression section (544 ms) than in the initiation section (471 ms). ANOVAs and ANCOVAs were conducted separately for each section (i.e., initiation versus suppression) on response time to investigate the possible influence of processing speed on age effects. As suggested by the significant interaction between age and section, the age effect size was smaller in the initiation sections, F(1,59) = 131.53, p < .001 and h2p = 0.69, than in the suppression sections, F(1,59) = 202.83, p < .001 and h2p = 0.77. This age effect was reduced after using processing speed as covariate but the reduction was larger for the initiation sections, F(1,58) = 26.39, p < .001 and h2p = 0.31, than the suppression sections, F(1,58) = 53.39, p < .001 and h2p = 0.48. The interaction between age and design was significant, F(1,59) = 19.61, p < .001 and h2p = 0.25, the difference between young and older adults being greater in the unblocked design (+629 ms) than in the blocked design (+388 ms). ANOVAs and ANCOVAs were conducted separately on response time for each design (i.e., blocked versus unblocked) to investigate the possible influence of processing speed on age effects. As suggested by the significant interaction between age and design, the age effect size was smaller in the blocked design, F(1,59) = 92.97, p < .001 and h2p = 0.61, than in the unblocked design, F(1,59) = 138.77, p < .001 and h2p = 0.70. However, the reduction in the age effect after using processing speed as covariate was larger for the blocked design, F(1,58) = 16.52, p < .001 and h2p = 0.22, than for the unblocked design, F(1,58) = 37.25, p < .001 and h2p = 0.39. The interaction between section and design was significant, F(1,59) = 4.83, p = .03 and h2p = 0.08, the difference between initiation and suppression section being greater in the unblocked (+139 ms) than in the blocked design (+93 ms). The interaction between age, section, and design, was not significant, F < 1. Given the importance of comparing the age effect in the blocked and unblocked designs, 2 (Age)  2 (Section) ANOVAs were conducted separately for each design on response times. Regarding the age effect, the comparison of partial eta-squared revealed a

603

smaller effect in the blocked design, F(1,59) = 92.97, p < .001 and h2p = 0.61, than in the unblocked design, F(1,59) = 138.78, p < .001 and h2p = 0.70. When processing speed was used as covariate, the age effect in both designs was reduced but remained significant when processing speed was used as covariate, respectively F(1,58) = 16.52, p < .001 and h2p = 0.22, and F(1,58) = 37.23, p < .001 and h2p = 0.39. Regarding the interaction between age and section, the comparison of partial eta squared revealed a larger effect in the blocked design, F(1,59) = 9.72, p < .01 and h2p = 0.14, than in the unblocked design, F(1,59) = 3.96, p = .05 and h2p = 0.06. After using processing speed as covariate, the effect was reduced but remained significant for the blocked design, F(1,58) = 5.34, p < .05 and h2p = 0.08, and increased slightly for the unblocked design, F(1,58) = 8.28, p < .01 and h2p = 0.12. Regarding errors (Table 3), the Chi-squared test revealed no age difference, x2(1, N = 244) = 0.20 and p > .10, with 24 young adults and 22 older adults making at least one error during the whole task. More participants committed errors in the suppression than in the initiation sections, respectively 31 and 15, x2(1, N = 244) = 6.86 and p = .009. Moreover, more participants committed errors in the unblocked that in the blocked design, respectively 37 and 9, x2(1, N = 244) = 21.00 and p < .001. Age comparisons for each section revealed no age differences for initiation sections, x2(1, N = 122) = 0.04 and p > .10, or for suppression sections, x2(1, N = 122) = 0.53 and p > .10. Likewise, age comparisons for each design revealed no age differences for the blocked design, x2(1, N = 122) = 0.16 and p > .10, or for the unblocked design, x2(1, N = 122) = 0.10 and p > .10. Finally, section comparisons for each design indicated only a tendencial effect in the blocked design, with more participants committing errors in the suppression than in the initiation section (respectively 7 and 2, x2(1, N = 122) = 3.00 and p = .08), whereas the effect reached significance in the unblocked design, with more participants committing errors in the suppression than the in initiation section (respectively 24 and 13, x2(1, N = 122) = 4.69 and p = .03). 3.3. Correlations between the Hayling switching task and the Trail Making Test The links between each part of the Hayling task and the Trail Making Test were investigated to assess the involvement of cognitive flexibility in the unblocked design. A significant age effect was observed on the Trail Making Test performance. Time to complete Part B minus time to complete Part A (TMT B-A) was longer for older adults than young ones, respectively 40.1 s (SD = 16.4) and 28.4 s (SD = 13.45), t(59) = 3.04, p = .003. Bivariate correlations were performed separately for young and older adults. In young adults, all correlations between TMT B-A and sections were non-significant. In older adults, correlations between TMT B-A and the initiation and suppression sections were non-significant in the blocked design but were significant in the unblocked design, respectively r = .44, p = .01 and r = .49, p < .01. 4. Discussion The aims of the present study were to further knowledge about the age effect in the Hayling task and to provide a better Table 3 Number of participants with at least one error according to age, section, and design. Blocked

Young adults (N = 30) Older adults (N = 31)

Unblocked

Initiation

Suppression

Initiation

Suppression

0 2

5 2

7 6

12 12

604

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

understanding of the cognitive processes involved in the task. Although the Hayling task is frequently used to assess activation and inhibition processes in various populations, it appeared relevant to adjust the initial task to improve its sensitivity to impairments of activation and inhibition processes. To do this, we compared the performance of young and older adults on the initiation and suppression sections presented with a classical design (i.e., blocked design, Burgess & Shallice, 1996) and those obtained with a switching design (i.e., unblocked design, see Be´langer & Belleville, 2009). Moreover, modifications were made with regard to sentence presentation and response selection. The present results reveal an age effect on the initiation and suppression sections for both the blocked (i.e., classical task design) and unblocked designs (i.e., switching design), even after statistical adjustment of processing speed. Older participants were slower than young ones in the initiation sections, so the interaction between age and sections suggests a decline in inhibition with aging. These findings are consistent with those of previous studies showing slower response times in older adults than in young ones on both the initiation (Bielak et al., 2006; Collette et al., 2009; Lin et al., 2007) and suppression sections (Andre´s & Van der Linden, 2000; Belleville et al., 2006; Bielak et al., 2006; Collette et al., 2009). The detection of an age effect on initiation and suppression sections is an argument in favor of the several changes made in the current protocol. One of them involves choosing the answer between two displayed words (Delaloye et al., 2009) while the original Hayling design (Burgess & Shallice, 1996) requires the participant to produce the response word himself. The objective was to avoid the complex production of an alternative response that involves much more than inhibitory processes (Andre´s & Van der Linden, 2000) and could bias the purity of the inhibition measure. This pre-selection of the word response may explain why we have observed an age effect on response time but not on error whereas other authors have observed an age-related change on both of them (e.g., Be´langer & Belleville, 2009). With the former production format, errors could be more due to the difficulty to produce completely unrelated responses than the difficulty to inhibit the prepotent response. Another change made in order to improve the purity and precision of the scores was to use an aloud and self-administrated reading of sentences rather the traditional blocked presentation. The aim of this procedure modification was twofold. First, it should limit the possibility of an excessive presentation time that would increase the risk of supplementary cognitive processes not controlled by the experimenter (see Collette et al., 2001). Second, it should also avoid the possibility of a too short presentation time that may disturb the correct processing of the sentences. This last risk could be hypothesized especially in older adults as a consequence of a reduction of processing speed in aging (Postal & Mathey, 2007). As suggested by previous data (Andre´s & Van der Linden, 2000; Collette et al., 2009), processing speed is involved during completion of the Hayling task. Indeed, a reduction in age effect size both on the initiation and suppression sections after the statistical adjustment of processing speed was observed in the present study, although the effects remained significant. It could be argued that the increase in response times with age in the initiation section is only due to a decrease in the older adults’ motor speed necessary to press the response keys. However, this age effect remained significant after the statistical adjustment of the processing speed task, which also requires pressing response keys. This result suggests that the observed age differences are not only due to slowing with age or difficulty with the task procedure. In previous Hayling task studies, the statistical adjustment of processing speed made the age effect disappear on the initiation section (Collette et al., 2009) and considerably reduced it on the suppression section (Andre´s & Van der Linden, 2000; Collette et al., 2009). However, unlike Collette

et al. (2009), we did not have the opportunity to study the impact of working memory on age-related performance due to any significant age difference on our working memory task despite a lower score for elderly adults. This absent of significant age effect is surprising but not isolated in the literature (Waters & Caplan, 2003). The experimenter observations during the working memory task reveal that the calculation part was very difficult for a majority of young adults whereas almost all of the older participants performed the calculations easily and quickly. This difference is probably due to a cohort effect (i.e., different educational background and differences in access to higher education between young and older adults), which may have impeded the performance of young adults to retain words because a lot of time and cognitive resources were allocated to calculation. For future studies, the use of ‘‘older’’ young participants, of less educated older adults or/and of various working memory tasks will prevent this possible bias. However, it remains clear that the cognitive variables known to decline with aging need to be adjusted, which could explain the discrepancies between findings in studies investigating the Hayling task. A switching design was used to avoid the use of potential strategies (Be´langer & Belleville, 2009). Since previous studies showed a decline in strategy use with aging (e.g., Lemaire, 2010), the possibility of using strategies during the Hayling task could augment the age-related decline on this task. However, the current task design does not allow to assess directly the use of strategy. Although the unblocked design appears more complex than the blocked one because the participants cannot anticipate the task they will have to perform in the latter, the significant correlations in older adults with the Trail Making Test indicate that flexibility is involved to the same extent in the initiation and suppression sections of this unblocked design (respectively r = .44, p = .01, and r = .49, p < .01). Interestingly, this result supports the assumption of Be´langer and Belleville (2009) that switching is equally involved in the completion of both sections. Together with previous knowledge about the Hayling task, the significant correlations between the unblocked design and the measure of cognitive flexibility give insights into the cognitive components involved in this adapted Hayling task. As a consequence of this flexibility involvement, a concurrent measure of TMT performance is to recommend adjusting the results to this switching design for a more accurate assessment of inhibition processes. Executive functions are generally referred to as processing related to goal-directed behavior or control of complex cognition, and are especially involved during non-routine situations (McCabe, Roediger, McDaniel, Balota, & Hambrick, 2010). However, there is debate about which cognitive processes should be considered as belonging to executive functions. A current relative consensus is to consider inhibition, switching and updating as three major executive functions (Miyake et al., 2000). Another finding is that the various executive functions appear to be not completely independent by sharing some underlying commonality (Collette et al., 2005; Miyake et al., 2000; but see Adrover-Roig, Sese´, Barcelo´, & Palmer, 2012). The distinction observed with a population of young adults by Miyake et al. (2000) has been broadly replicated with a more age-heterogeneous group composed of participants aged between 20 and 81 (Fisk & Sharp, 2004). Moreover, a new component of efficiency of access to long-term memory was observed and linked to the verbal fluency tasks added by the authors (see Adrover-Roig et al., 2012). Consequently, the activation of knowledge in semantic memory is also a relevant component of executive functioning to investigate. Several of these executive functions are directly required by the present Hayling switching task. Although the debate between a unitary or a multidimensional vision of executive functions is still relevant in the literature (Chan, Shum, Toulopoulou, & Chen, 2008), several

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

authors assume that executive functions reflect both unified and diversified constructs (Friedman et al., 2008; Banich, 2009; McCabe et al., 2010). Therefore, consistent with this latter view, it can be considered as an advantage to assess executive functions in a single task rather than in several different tasks to reduce variability due to the task characteristics. As shown in Fig. 1, both the initiation and suppression sections of the blocked and unblocked designs rely on activation of knowledge in semantic memory. The initiation section of the blocked design relies only on the activation component whereas the suppression section of the blocked designs and the initiation and suppression sections of the unblocked design also involve specific components. In particular, the suppression section of the blocked design involves the inhibition component in order to eliminate the appropriate word and give the unrelated word. Finally, a flexibility component is involved in the unblocked design because it implies that the participant rapidly switches between the tasks (Monsell, 2003). This flexibility component occurs in addition to an activation component for sentences associated with initiation, and activation and inhibition components for sentences associated with suppression. The fact that correlations between the Trail Making Test and the unblocked design were significant only for older adults seems to be explained by their lowest performance on this task. Indeed, participants with low flexibility skills would have more difficulties to perform switching in the unblocked design because cognitive resource will become insufficient to switch between initiation and suppression orders. This age difference suggests that flexibility skills of young adults were sufficient and were only slightly solicited in the unblocked design on the contrary of older adults. This interpretation is also congruent with the largest age effect observed in the unblocked design compared to the blocked design (.14 versus .06). Finally, future studies should be conducted to investigate whether this Hayling switching task has research and clinical value, e.g. for discriminating between healthy aging, MCI, and the earlier stages of Alzheimer’s disease. A first step has been made in this direction by Be´langer and Belleville (2009) who were able to show an increase in the suppression condition of response partially related to the target for persons with MCI. Moreover, people who showed cognitive decline or progression to AD at the 24-month follow-up produced more responses that reflected partial inhibition and longer response time in the inhibition condition than persons with stable MCI (Be´langer & Belleville, 2009). However, an alternative explanation is that the use of a switched design overloaded the cognitive processes of persons with MCI in whom a deficit of divided attention has also been reported (Belleville et al., 2007). Indeed, the involvement of flexibility processes in an unblocked design may have overloaded the cognitive system of persons with MCI and, as a consequence, impeded their ability to

Fig. 1. Activation, inhibition, and flexibility processes in the blocked and unblocked designs of the Hayling switching task.

605

deal efficiently with the controlled task of semantic inhibition. If the inhibition of partially related responses is the key to the MCI difficulties, MCI effect would not appear in the blocked design of the present task that requires participants to choose the answer between two displayed words because they do not have to produce alternative answers themselves. Moreover, the fact that the responses are chosen rather than being produced could also improve understanding of the task by individuals with pathological cognitive aging, and might offer an efficient procedure to assess activation and inhibition simultaneously in this population.

5. Conclusion The results obtained with this modified Hayling task show that older adults have slower response times than young adults in the initiation and suppression sections, thus confirming some agerelated decline in semantic activation and inhibition. Since previous studies often failed to find simultaneously an age effect in the initiation and suppression sections or the persistence of these effects after the adjustment of processing speed, it is possible that the modifications in the procedure regarding sentence processing and response production make the task more sensitive to interindividual differences on semantic activation and inhibition processes. Therefore, future studies should be conducted to investigate whether the modified format of the Hayling switching task we used has clinical value, e.g. for discriminating between healthy aging, MCI, and the early stage of Alzheimer’s disease.

Conflict of interest statement The authors declare that there is no conflict of interest. Acknowledgments We thank OAREIL (office aquitain de recherche, d’e´tudes, d’information et de liaison sur les proble`mes des personnes aˆge´es), its director Mr. Jean-Jacques Amyot, and its members. References Adrover-Roig, D., Sese´, A., Barcelo´, F., & Palmer, A. (2012). A latent variable approach to executive control in healthy ageing. Brain and Cognition, 78, 284–299. Andre´s, P., & Van der Linden, M. (2000). Age-related differences in supervisory attentional system functions. Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 55, 373–380. Andre´s, P., & Van der Linden, M. (2004). Les capacite´s d’inhibition: une fonction frontale? Revue Europe´enne de Psychologie Applique´e, 54, 137–142. Banich, M. T. (2009). Executive function: The search for an integrated account. Psychological Science, 18, 89–94. Be´langer, S., & Belleville, S. (2009). Semantic inhibition impairment in mild cognitive impairment: A distinctive feature of upcoming cognitive decline? Neuropsychology, 23, 592–606. Belleville, S., Chertkow, H., & Gauthier, S. (2007). Working memory and control of attention in persons with Alzheimer’s disease and mild cognitive impairment. Neuropsychology, 21, 458–469. Belleville, S., Rouleau, N., & Van der Linden, M. (2006). Use of the Hayling task to measure inhibition of prepotent responses in normal aging and Alzheimer’s disease. Brain and Cognition, 62, 113–119. Bielak, A. A. M., Mansueti, L., Strauss, E., & Dixon, R. A. (2006). Performance on the Hayling and Brixton tests in older adults: Norms and correlates. Archives of Clinical Neuropsychology, 21, 141–149. Borella, E., Delaloye, C., Lecerf, T., Renaud, O., & de Ribaupierre, A. (2009). Do age differences between young and older adults in inhibitory tasks depend on the degree of activation of information? European Journal of Cognitive Psychology, 21, 445–472. Borella, E., Ludwig, C., Fagot, D., & de Ribaupierre, A. (2011). The effect of age and individual differences in attentional control: A sample case using the Hayling test. Archives of Gerontology and Geriatrics, 53, e75–e80. Burgess, P. W., & Shallice, T. (1996). Response suppression, initiation and strategy use following frontal lobe lesions. Neuropsychologia, 34, 263–273.

606

I. Tournier et al. / Archives of Gerontology and Geriatrics 59 (2014) 599–606

Castner, J. E., Copland, D. A., Silburn, P. A., Coyne, T. J., Sinclair, F., & Chenery, H. J. (2007). Lexical-semantic inhibitory mechanisms in Parkinson’s disease as a function of subthalamic stimulation. Neuropsychologia, 45, 3167–3177. Chan, R. C. K., Shum, D., Toulopoulou, T., & Chen, E. Y. H. (2008). Assessment of executive functions: Review of instruments and identification of critical issues. Archives of Clinical Neuropsychology, 23, 201–216. Collette, F., Schmidt, C., Scherrer, C., Adam, S., & Salmon, E. (2009). Specificity of inhibitory deficits in normal aging and Alzheimer’s disease. Neurobiology of Aging, , 875–889. Collette, F., Van der Linden, M., Delfiore, G., Degueldre, C., Luxen, A., & Salmon, E. (2001). The functional anatomy of inhibition processes investigated with the Hayling task. NeuroImage, 14, 258–267. Collette, F., Van Der Linden, M., Laureys, S., Delfiore, G., Degueldre, C., Luxen, A., et al. (2005). Exploring the unity and diversity of the neural substrates of executive functioning. Human Brain Mapping, 25, 409–423. Corrigan, J. D., & Hinkeldey, N. S. (1987). Relationships between Parts A and B of the Trail Making test. Journal of Clinical Psychology, 43, 402–409. Crowe, S. F. (1998). The differential contribution of mental tracking, cognitive flexibility, visual search, and motor speed to performance on Parts A and B of the Trail Making Test. Journal of Clinical Psychology, 54, 585–591. Delaloye, C., Moy, G., Baudois, S., De Bilbao, F., Remund, C. D., Hofer, F., et al. (2009). The contribution of aging to the understanding of the dimensionality of executive functions. Archives of Gerontology and Geriatrics, 49, e51–e59. Deltour, J. J. (1993). Echelle de vocabulaire de Mill Hill de J.C. Raven. Braine-le-Chaˆteau: Editions l’Application des Techniques Modernes. Emery, L., Hale, S., & Myerson, J. (2008). Age differences in proactive interference, working memory, and abstract reasoning. Psychology and Aging, 23, 634–645. Finkel, D., Reynolds, C. A., McArdle, J. J., & Pedersen, N. L. (2007). Age changes in processing speed as a leading indicator of cognitive aging. Psychology and Aging, 22, 558–568. Fisk, J. E., & Sharp, C. A. (2004). Age-related impairment in executive functioning: Updating, inhibition, shifting, and access. Journal of Clinical and Experimental Neuropsychology, 26, 874–890. Friedman, N. P., Miyake, A., Young, S. E., DeFries, J. C., Corley, R. P., & Hewitt, J. K. (2008). Individual differences in executive functions are almost entirely genetic in origin. Journal of Experimental Psychology: General, 137, 201–225. Gohier, B., Ferracci, L., Surguladze, S., Lawrence, E. J., El Hage, W., Kefi, W., et al. (2009). Cognitive inhibition and working memory in unipolar depression. Journal of Affective Disorders, 116, 100–105. Golden, C. J., Osmon, D. C., Moses, J. A., & Berg, R. A. (1981). Interpretation of the HalsteadReitan neuropsychological test battery. New York: Grune & Stratton. Hasher, L., Zacks, R. T., & May, C. P. (1999). Inhibitory control, circadian arousal, and age. In D. Gopher & A. Koriat (Eds.), Attention and performance XVII: Cognitive regulation of performance: Interaction of theory and application. (pp. 653–675). Cambridge, MA: MIT Press. Hertzog, C., Dixon, R. A., Hultsch, D. F., & MacDonald, S. W. S. (2003). Latent change models of adult cognition: Are changes in processing speed and working memory associated with changes in episodic memory? Psychology and Aging, 18, 755–769. Hester, R. L., Kinsella, G. J., Ong, B., & McGregor, J. (2005). Demographic influences on baseline and derived scores from the Trail Making Test in healthy older Australian adults. Clinical Neuropsychologist, 19, 45–54. Kalafat, M., Hugonot-Diener, L., & Poitrenaud, J. (2003). Standardisation et e´talonnage franc¸ais du Mini Mental State (MMS) version GRECO. Revue de Neuropsychologie, 13, 209–236. Kemtes, K. A., & Kemper, S. (1997). Younger and older adults’ on-line processing of syntactically ambiguous sentences. Psychology and Aging, 12, 362–371.

Kortte, K. B., Horner, M. D., & Windham, W. K. (2002). The Trail Making Test, Part B: Cognitive flexibility or ability to maintain set? Applied Neuropsychology, 9, 106–109. Lemaire, P. (2010). Cognitive strategy variations during aging. Current Directions in Psychological Sciences, 19, 363–369. Lemke, U., & Zimprich, D. (2005). Longitudinal changes in memory performance and processing speed in old age. Aging, Neuropsychology, and Cognition, 12, 57–77. Lin, H., Chan, R. C. K., Zheng, L., Yang, T., & Wang, Y. (2007). Executive functioning in healthy elderly Chinese people. Archives of Clinical Neuropsychology: The Official Journal of the National Academy of Neuropsychologists, 22, 501–511. McCabe, D. P., Roediger, H. L., McDaniel, M. A., Balota, D. A., & Hambrick, D. Z. (2010). The relationship between working memory capacity and executive functioning: Evidence for a common executive attention construct. Neuropsychology, 24, 222–243. Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex frontal lobe tasks: A latent variable analysis. Cognitive Psychology, 41, 49– 100. Monsell, S. (2003). Task switching. Trends in Cognitive Sciences, 7, 134–140. New, B., Pallier, C., Ferrand, L., & Matos, R. (2001). Une base de donne´es lexicales du franc¸ais contemporain sur Internet: Lexique. L’Anne´e Psychologique, 101, 447–462. Noe¨l, X., Van der Linden, M., Brevers, D., Campanella, S., Verbanck, P., Hanak, C., et al. (2013). Separating intentional inhibition of prepotent responses and resistance to proactive interference in alcohol-dependent individuals. Drug and Alcohol Dependence, 128, 200–205. Osborne, J. W., & Overbay, A. (2004). The power of outliers (and why researchers should always check for them). Practical Assessment Research Evaluation, 9, 1–8. Postal, V., & Mathey, S. (2007). Diffe´rences lie´es a` l’aˆge lors de la lecture de phrases: e´tude des processus d’activation et d’inhibition. Revue Europe´enne de Psychologie Applique´e, 57, 91–100. Reitan, R. M. (1958). Validity of the Trail Making test as an indicator of organic brain damage. Perceptual and Motor Skills, 8, 271–276. Robichon, F., Besson, M., & Faı¨ta, F. (1996). Norme de comple´tion pour 744 contextes linguistiques franc¸ais de diffe´rents formats. Revue Canadienne de Psychologie Expe´rimentale, 50, 205–233. Salthouse, T. A. (1991). Mediation of adult age differences in cognition by reductions in working memory and speed of processing. Psychological Science, 2, 179–183. Salthouse, T. A. (1994a). The aging of working memory. Neuropsychology, 8, 535–543. Salthouse, T. A. (1994b). The nature of the influence of speed on adult age differences in cognition. Developmental Psychology, 30, 240–259. Seo, E. H., Lee, D. Y., Kim, K. W., Lee, J. H., Jhoo, J. H., Youn, J. C., et al. (2006). A normative study of the Trail Making test in Korean elders. International Journal of Geriatric Psychiatry, 21, 844–852. Shibuya-Tayoshi, S., Sumitani, S., Kikuchi, K., Tanaka, T., Tayoshi, S. Y., Ueno, S.-I., et al. (2007). Activation of the prefrontal cortex during the Trail-Making Test detected with multichannel near-infrared spectroscopy. Psychiatry and Clinical Neurosciences, 61, 616–621. Soukup, V. M., Ingram, F., Grady, J. J., & Schiess, M. C. (1998). Trail-making test: Issues in normative data selection. Applied Neuropsychology, 5, 65–73. Tun, P. A., O’Kane, G., & Wingfield, A. (2002). Distraction by competing speech in young and older adult listeners. Psychology and Aging, 17, 453–467. Tun, P. A., McCoy, S., & Wingfield, A. (2009). Aging, hearing acuity, and the attentional costs of effortful listening. Psychology and Aging, 24, 761–766. Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28, 127–154. Waters, G. S., & Caplan, D. (2003). The reliability and stability of verbal working memory measures. Behavior Research Methods, Instruments, & Computers, 35, 550–564.