Non-dominant hand use increases completion time on part B of the Trail Making Test but not on part A

Non-dominant hand use increases completion time on part B of the Trail Making Test but not on part A Behav Res (2018) 50:1074–1087 DOI 10.3758/s13428-017-0927-1 BRIEF COMMUNICATION Non-dominant hand use increases completion time on part B of the Trail Making Test but not on part A 1 1 Laura Klaming & Björn N. S. Vlaskamp Published online: 13 July 2017 The Author(s) 2017. This article is an open access publication . . Abstract The Trail Making Test (TMT) is used in neuropsy- Keywords Trail Making Test B/A ratio Executive chological clinical practice to assess aspects of attention and functioning Non-dominant hand use executive function. The test consists of two parts (A and B) and requires drawing a trail between elements. Many patients are assessed with their non-dominant hand because of motor The Trail Making Test (TMT) is a frequently used neuropsy- dysfunction that prevents them from using their dominant chological test to assess aspects of attention and executive hand. Since drawing with the non-dominant hand is not an functions (Bowie & Harvey, 2006; Lezak, Howieson, & automatic task for many people, we explored the effect of Loring, 2004; Tombaugh, 2004;Wagner, Helmrich, hand use on TMT performance. The TMT was administered Dahmen, Lieb, & Tadic, 2011). The TMT consists of drawing digitally in order to analyze new outcome measures in addi- a trail between elements that are quasi-randomly scattered on tion to total completion time. In a sample of 82 healthy par- paper and has two parts. Part A involves making a trail be- ticipants, we found that non-dominant hand use increased tween numbers in ascending order. Part B consists of 13 num- completion times on the TMT B but not on the TMT A. The bers and 12 letters, which the patient is instructed to connect in average completion time increased by almost 5 seconds, an alternating pattern. The patient is asked to complete the which may be clinically relevant. A substantial number of trails as quickly as possible, and completion time is measured participants who performed the TMT with their non- as the main outcome (Bowie & Harvey, 2006; Lezak et al., dominant hand had a B/A ratio score of 2.5 or higher. In 2004). The clinical interpretation of performance on the TMT clinical practice, an abnormally high B/A ratio score may be is based on part A mainly reflecting visual search and motor falsely attributed to cognitive dysfunction. With our digitized speed skills, and part B also requiring higher-order cognitive pen data, we further explored the causes of the reduced functions such as cognitive flexibility and task switching TMT B performance by using new outcome measures, includ- (Lezak et al., 2004). The completion time is longer for ing individual element completion times and interelement var- part B than for part A, and the B–A difference as well as the iability. These measures indicated selective interference be- B/A ratio are considered to reveal deficits in executive func- tween non-dominant hand use and executive functions. Both tion (Arbuthnott & Frank, 2000;Bowie &Harvey, 2006; non-dominant hand use and performance of the TMT B seem Corrigan & Hinkeldey, 1987; Gaudino, Geisler, & Squires, to draw on the same, limited higher-order cognitive resources. 1995; Kortte, Horner, & Windham, 2002; Lezak et al., 2004; Reitan & Wolfson, 1995; Yochim, Baldo, Nelson, & Delis, 2007). Research has consistently shown that performance on the TMT is strongly influenced by age and education, with * Laura Klaming completion times on both parts of the TMT increasing with laura.klaming@philips.com growing age and fewer years of education (Amodio et al., 2002; Fromm-Auch & Yeudall, 1983; Robins Wahlin, Bäckman, Wahlin, & Winblad, 1996; Salthouse & Fristoe, Department of Brain, Behavior and Cognition, Philips Research, 1995; Tombaugh, 2004). High Tech Campus 34, 5656 AE Eindhoven, The Netherlands Behav Res (2018) 50:1074–1087 1075 Another factor that may influence TMT performance in a Imaging research has corroborated the finding that clinical setting is the use of the non-dominant hand. This is an non-dominant hand use interferes with cognitive process- important factor, because patients who suffer from motor dys- ing and shown that performing a simple motor task (se- function that prevents them from using their dominant hand quential finger movements) with the non-dominant hand may be required to perform the TMT with their non-dominant results in greater cortical activity than performing a sim- hand. Currently no norm scores are available for completion ple motor task with the dominant hand and in similar times on the TMT with the non-dominant hand, and little is cortical activity as performing a complex motor task known about how hand use affects TMT performance, which (random finger movements) with the dominant hand makes it difficult to interpret completion times and derived (Mattay et al., 1998). Similarly, another study (Jäncke scores like the B–A difference and the B/A ratio. An increase et al., 1998) has shown that performing a sequential in completion time or derived score may be falsely attributed movement with the non-dominant hand (in right-handed to non-dominant hand use, which may result in an underesti- subjects) results in greater right hemisphere activation mation of cognitive problems; alternatively, an increase in than left hemisphere activationduringperformanceof completion time or the derived scores may be falsely attribut- the same movement with the dominant (right) hand. ed to a cognitive problem rather than hand use, which would These findings indicate that motor movements with the result in an overestimation of cognitive problems. non-dominant hand are less familiar and automatic, and Hand use may impact TMT performance in several ways. therefore consume more cognitive resources. First, research has shown that performance on a motor task is Taken together, these findings suggest that performing a generally slower with the non-dominant hand, which is as- motor task with the non-dominant hand while simultaneously sumed to be mainly attributable to lower movement accuracy performing a cognitive task increases cognitive load, which and a greater need for corrective movements (Annett, Annett, compromises performance on both or one of the two tasks. On Hudson, & Turner, 1979). Research has furthermore shown the basis of the findings of prior research, we therefore hy- that healthy individuals are markedly slower when performing pothesized that performing the TMT A with the non-dominant a standard neuropsychological test involving fine motor skills, hand would not or would only marginally slow down comple- such as copying or cancellation (Cramond, Clark, & Smith, tion times, since there would be little competition for the same 1989) or name writing (Fromm-Auch & Yeudall, 1983), with cognitive resources because the TMT A mainly reflects visual the non-dominant hand. search and motor speed skills. We furthermore hypothesized Second, and more problematically, there is evidence for an that performing the TMT B with the non-dominant hand interference effect between simultaneous performance of motor would increase completion times, since performance of the and cognitive tasks. Studies that have employed a dual-task par- TMT B requires a substantial contribution of higher-order cognitive functions, and therefore both the TMT B and use adigm have consistently shown that simultaneously performing a motor and a cognitive task increases cognitive load and results of the non-dominant hand would compete for the same limited in an interference effect, with performance on both tasks deteri- cognitive resources. orating (Baddeley & Della Salla, 1996; Hausdorff, Yogev, To our knowledge, three studies have explored the effect of Springer, Simon, & Giladi, 2005; Lindenberger, Marsiske, & dominant versus non-dominant hand use on completion times Baltes, 2000; Siu, Chou, Mayr, van Donkelaar, & Woollacott, for the TMT (LoSasso, Rapport, Axelrod, & Reeder, 1998; 2008; Theill, Martin, Schumacher, Bridenbaugh, & Kressig, Toyokura, Ishida, Watanabe, Okada, & Yamazaki, 2003; 2012). The interference effect may be stronger when the motor Toyokura, Sawatari, Nishimura, & Ishida, 2003). LoSasso task involves use of the non-dominant hand. For example, and colleagues compared completion times on the original healthy individuals were found to perform the recall part of the TMT and on a parallel version of the TMT for 40 right- Rey Osterrieth Complex Figure Test significantly worse when handed and 40 left-handed individuals who performed the they had used their non-dominant hand as opposed to their dom- tests with their both dominant and their non-dominant hand. inant hand when copying the figure (Yamashita, 2010). Completion times were found to be slightly longer for the non- According to the author, this finding is due to drawing the figure dominant hand for both the original and the parallel TMT B. with the non-dominant hand leading to fewer cognitive resources This intermanual difference was not significant, however, and being allocated to the performance of the cognitive task—i.e. the was considered clinically irrelevant (LoSasso et al., 1998). copying of the figure. This interpretation is supported by a study Toyokura and colleagues (Toyokura, Ishida, et al. 2003; that found slowed performance on a cognitive test that requires Toyokura, Sawatari, et al., 2003) explored differences in com- executive function—i.e. random number generation—when par- pletion times for the Japanese version of the TMT, which ticipants simultaneously performed a motor speed task—i.e. the consists of numbers (part A) and numbers and Japanese kana grooved pegboard task—with their non-dominant hand. This letters (part B). In both studies, no intermanual difference in dual-task interference effect was not found for the dominant completion times was found (Toyokura, Ishida, et al., 2003; hand (Strenge & Niederberger, 2008). Toyokura, Sawatari, et al., 2003). 1076 Behav Res (2018) 50:1074–1087 Although these three studies have provided interesting in- et al., 2003) and to be able to explore other measures in addi- sights, a number of important limitations make it difficult to tion to total completion time. To our knowledge, this study conclude at this time that there is not a clinically relevant was the first to look at additional outcome measures, besides difference between performing the TMT with the dominant total completion time, that could provide valuable information versus the non-dominant hand. First, completion times were about cognitive functioning and that are not available when measured manually with a stopwatch in all three studies. administering the TMT without pen digitization. Moreover, Manual measurement of completion times may not always this study was the first to look at digitized trails of the tradi- be precise, which may have introduced additional variance tional paper–pencil TMT. unrelated to the participants in these studies. Second, the exact administration procedure of the TMTwas not explained in any of these three studies, and it is therefore unclear what the begin Method and end times of the measurement were and—importantly— to what degree errors have affected completion times. Participants Differences in administration account for the large variability in TMT completion times reported in different studies, which A total of 117 healthy right-handed individuals participated in is problematic and is considered to be an important limitation the study. Handedness was determined with the Edinburgh of the TMT (Soukup, Ingram, Grady, & Schiess, 1998; Handedness Inventory. One participant was found to have a Woods, Wyma, Herron, & Yund, 2015). Treating TMTs with tendency toward left-handedness (score of –.48) and was and without errors as synonymous cognitive tests hampers therefore excluded from further analyses. The data of 11 par- interpretation. Self-correction of an erroneous movement sub- ticipants were not included because of technical issues. Of the stantially increases the total completion time, which may thus remaining 105 participants, 23 (21.9%) made a mistake during reflect additional motoric and cognitive processes. Errors that either the TMT A or the TMT B or both, and were therefore are pointed out by the examiner and then corrected by the excluded from further analyses. Participants who made mis- individual also introduce examiner timing into the completion takes were excluded in order to obtain a pure measure of total time. Therefore, trails with errors are different from trails with- completion time, since as we described above, the correction out errors, which makes the inclusion of TMTs with errors in of errors has a substantial impact on total completion time and research problematic. It is unclear how many participants changes the test. A mistake was defined as any path that de- made errors on the TMT in the three studies described above viated from the correct path—for example, if a participant and whether these individuals were included in the analyses. went from 18 to 20 rather than from 18 to 19. The numbers An important limitation of the Toyokura et al. studies of participants who made mistakes were similar for both con- (Toyokura, Ishida, et al., 2003; Toyokura, Sawatari, et al., ditions (ten in the dominant hand condition and 13 in the non- 2003) is that the Japanese version of the TMTwas used, which dominant hand condition; see below for a description of the is not directly comparable to the original TMT. The ratio be- conditions). The remaining 82 participants (28 women, 54 tween the TMT A and the TMT B is much higher for the men) were all right-handed as measured with the Edinburgh original version of the TMT than for the Japanese version Handedness Inventory (M =.80, SD = .15). They ranged in (Toyokura, Sawatari, et al., 2003), indicating that part B is age from 20 to 65 years of age (M =36 years, SD =12). Of the more difficult than part A (Tombaugh, 2004), which makes participants, 57 (69.5%) had a university degree ranging from it impossible to draw any valid conclusions about intermanual a BSc to a PhD, whereas the remaining 25 had a vocational differences in completion times for the original TMT based on degree as their highest. All participants had normal or the two Toyokura et al. studies. corrected-to-normal vision and were employed at Philips The aim of the present study was to investigate the differ- Research. Individuals were recruited with flyers and were ences in dominant and non-dominant hand use on the TMT in not compensated for their participation. a sample of healthy individuals. More specifically, we tested the hypothesis that non-dominant hand use would increase Materials completion times on the TMT B but would increase comple- tion times less or not at all on the TMT A. This hypothesis is Participants received the traditional paper–pencil TMT. based on the assumption that non-dominant hand use requires Underneath the paper TMT, a Wacom Intuos Pro tablet digi- additional cognitive resources, which interferes with perfor- tized the movements (see also Fig. 8 in the Results). Every mance on the TMT B. In the present experiment, the TMTwas new sheet of paper was aligned to markers on the tablet to administered in the traditional paper–pencil way, but was re- assure that all TMT’s were performed at the exact same loca- corded digitally in an unobtrusive way in order to overcome tion on the tablet. The position of the paper was fixed by some of the shortcomings of previous research (LoSasso et al., taping it to the tablet. We administered the traditional paper– 1998; Toyokura, Ishida, et al., 2003; Toyokura, Sawatari, pencil test to be able to draw conclusions about the TMT as it Behav Res (2018) 50:1074–1087 1077 is currently used in clinical practice. Using a tablet with a a new iteration of the same procedure was executed. This was screen or a computerized TMT to track movements would repeated until the newly calculate threshold was identical to not be suitable for this purpose because it would change the the previous one. physical properties of the test. The Wacom Intuos Pro tablet sampled the pen position at Segment-by-segment and element-by-element analysis 133 Hz. Pen positions were recorded with Movalyzer (devel- Aside from total completion time, we also performed more oped by Neuroscript). Pen pressure was not calibrated and detailed analyses of the trails. For this purpose, we fully auto- therefore not used in the analyses. matically detected the order in which the elements were com- pleted and extracted features per completion. Automatic de- Procedure tection of completion paths was conducted by comparing the spatial pattern within the area closest to an element (or Participants gave written informed consent prior to participa- Voronoi cell) with two simple model patterns. One of those tion. Participants were randomly assigned to complete the model patterns represented completion of the element by TMT either with their dominant hand (N = 41) or with their connecting the element with the pen entry and exit of the non-dominant hand (N = 41). The TMT was administered in Voronoi cell with two lines (one from the entry into the the standard manner, with part A preceding part B. For both Voronoi cell to the element and one from the element to the parts, the standard practice test with eight items was adminis- point of exit out of the Voronoi cell); the other model repre- tered prior to the test. sented no completion with a single line connecting entry and Participants performed the TMT individually in a room exit of the area. The decision whether an element was com- with a research assistant present. The study was conducted pleted or not was made on the basis of the similarity of the data by two different research assistants who were trained to use to the two models (using a two-dimensional least squares the same instructions for all participants. Administration of the method). If both models made similar predictions (e.g., when test took approximately 10 minutes. an element was completed in a straight path), the decision was made on the basis of whether there had been a local drop in Conventional and additional digital parameters pen velocity close to the element (within a radius of 0.5 cm for the TMT from the center of the elements). All classifications were then inspected manually; only eight of a total of 4,100 (i.e., 82 Both conventional and additional digital parameters for the times 25 TMT As + 82 times 25 TMT Bs) classifications were TMT were measured and included in the analyses. incorrect and had to be corrected manually. Conventional parameters include the total completion time, An important aspect of TMT performance we were inter- measured both with a stopwatch and digitally as described ested in was the time needed to move from one element to the below. Additional digital parameters for the TMT include a next—i.e. the element completion time. This was defined as segment-by-segment and element-by-element analysis, the total time required for finding, planning, processing, and interelement variability, and a separation of layout-related pro- executing the movement to the next element. The assumption cesses from executive processes on the TMT B. is that this starts as soon as the preceding element is completed and ends when the target element is completed. We defined Completion time Completion times were determined from the moment of completion as the moment the pen moves with the raw pen position data. They were calculated as the differ- subthreshold velocity as it is approaching the target element. ence in time between the moment the pen touched the tablet at The velocity threshold was calculated in the same way as the first element and the moment the pen stopped at the last described above. element of the TMT. The pen stop was defined as the velocity With this algorithm for splitting the TMT trail and calcu- of the pen dropping below threshold velocity within 1 cm of lating element completion times, we first explored whether the center of the last element. Velocity was calculated as the any difference between dominant and non-dominant hand instantaneous velocity using a second-order polynomial fit to use was related to performance on specific elements of the interpolate within a window of five samples centered on the TMT. We analyzed the data in the same way as Poreh, sample of interest. Threshold velocity was adaptively deter- Miller, Dines, and Levin (2012), whousedacomputer- mined through an iterative process for each individual assisted version of the TMT. The participants performed the TMT trial to account for noise in the tablet as well as human TMT on paper as usual, and with every element completion motor noise. First, a threshold was set arbitrarily. Next, a new the experimenter clicked with a mouse on a button that repre- threshold was calculated as five times the standard deviation sented the respective element. Poreh et al. divided the TMT A above the mean of the velocity of all samples below the and TMT B into five segments and calculated the time needed predefined threshold. If the resulting value was lower than to complete each of those segments. They found that the last the predefined threshold, it was set as the new threshold and part of the TMT B is particularly related to executive 1078 Behav Res (2018) 50:1074–1087 functioning and hypothesized that this may be related to low indication about an individual’s consistency across a task or search demands, since most elements are cancelled by then. In across multiple tasks or sessions, and can therefore provide our study, this translated to the prediction that a difference additional information about the individual’s cognitive func- between the two groups might be particularly evident in the tioning. The reason for the growing interest in IIV in the field last segment of the TMT B. of neuropsychology is that this measure is considered an in- We also inspected the trails in further detail on an element- formative measure because it is more highly correlated with by-element basis to identify any differences between domi- cognitive dysfunction than is the overall reaction time when nant hand and non-dominant hand use. Aside from differences patients are engaged in cognitively demanding tasks involving in individual element completion times related to the hypoth- working memory and set switching (MacDonald et al., 2009; esis of Poreh et al. (2012), using the left hand might also MacDonald et al., 2006; Strauss et al., 2007;West etal., introduce an effect on individual element completion times 2002). In a clinical context, IIV may provide more insight into because the hand covers different parts of the TMT, which the cognitive status of patients and allow for more accurate may make it either easier or harder to find certain elements. interpretations of test outcomes (Schretlen et al., 2003; The element-by-element analysis allowed for an investigation Tanner-Eggen et al., 2015). of hand bias of the TMT. Separation of layout-related processes from executive pro- Inter-element variability In the scientific literature and in cesses on TMT B The TMT B introduces a set-switching clinical neuropsychology, there is growing interest in intra- task, which causes an increment in completion times relative individual variability (IIV; e.g., MacDonald, Li, & Bäckman, to the TMT A, which is assumed to be due to increased exec- 2009; MacDonald, Nyberg, & Bäckman, 2006; Schretlen, utive demands. However, performing the TMT also involves Munro, Anthony, & Pearlson, 2003; Strauss, Bielak, Bunce, other (cognitive) tasks, such as visual search and moving the Hunter, & Hultsch, 2007; Tanner-Eggen, Balzer, Perrig, & pen from one element to the next, which are mostly related to Gutbrod, 2015; West, Murphy, Armilio, Craik, & Stuss, the layout of the TMT. Even though the TMT A and B were 2002). The majority of research on IIV has focused on vari- not designed to differ on these tasks, they do (Gaudino et al., ability in reaction time tasks, where IIV refers to changes in 1995), which leaves open the possibility that an increase in the reaction time data within an individual on a particular task completion time on the TMT B with the non-dominant hand rather than in the mean reaction time. Reaction time tasks could be due to interference with layout-related processes consist of many trials, and because multiple measurements rather than with executive functions. are collected per individual, IIV can be calculated. With the Here we sought evidence that non-dominant hand use in- standard paper–pencil TMT there is no way to get insight into terferes with executive processes on the TMT B by explicitly the variability in performance within an individual participant, separating the contribution of executive processes from the because only the overall completion time is measured. contribution of layout-related processes to the element com- However, with completion times per element derived from pletion times. For executive tasks, we assumed that the pro- the digital pen recordings a measure very similar to the IIV cessing time required for the completion of each element of can be calculated, a measure we call the inter-element the TMT B was roughly equal (for sake of the argument, we variability (IEV). We derived this measure from the digital ignore that there might be slight differences between the ele- pen recordings as follows. First, a distribution was compiled ments in terms of set-switching; e.g., it may be easier to go of the completion times per participant for all elements on the from letters to numbers than vice versa, it may be easier to TMT except for the first and last elements. Next, the IEV was keep the first letters of the alphabet in working memory than calculated as the difference between the 10% and 90% cuts later letters, etc.). This means that if non-dominant hand use through the distribution, as is typically done to calculate IIV. were to mainly affect executive functions, the same amount of Because the completion times are determined by the partici- additional processing would be expected for every element pant but also by the element characteristics (e.g., the time completed with the non-dominant hand relative to the process- needed to find visual information is dependent on its visual ing time for every element completed with the dominant hand. eccentricity and on the distance to neighboring information; For tasks related to the layout of the TMT (such as visual see, e.g., Vlaskamp, Over, & Hooge, 2005), the element com- search and motor tasks), we assumed that a different amount pletion times needed to be normalized prior to further process- of processing time would be required for each element, de- ing. Normalization was done by dividing each individual ele- pending on its spatial configuration: Some distances between ment completion time by the median completion time for each subsequent elements are very long whereas others are short; element across participants within a condition. These normal- some elements are in more cluttered areas than others; and ized completion times were then used to calculate the IEV. sometimes elements are positioned in between subsequent Currently in clinical neuropsychology, the mean comple- elements. These factors are known to have a large impact on tion time is a more typical outcome measure. IIV gives an tasks such as visual search (Vlaskamp & Hooge, 2006; Behav Res (2018) 50:1074–1087 1079 Vlaskamp, Over, & Hooge, 2005) and motor processes. On the basis of this assumption, any interference of non-dominant hand use with layout-related processes would be particularly noticeable on elements that already required long processing times with the dominant hand. In short, an across-the-board increment in element completion times in the non-dominant hand condition would be related to executive processes, but systematic increments with longer element completion times in the non-dominant hand condition would be related to layout-related processes. Statistical analyses All statistical analyses were performed using SPSS (version 23). These statistical analyses included independent-samples t Fig. 1 Completion times on the TMTA (left panel) and B (right panel) as tests, chi-square tests, and general linear models. Correlation a function of age. Each dot represents the completion time of one analyses (Pearson and point-biserial) were used to determine participant—black dots for participants in the dominant hand condition, the impact of potential covariates. and yellow dots for participants in the non-dominant hand condition. The lines are regression lines Results 51.05 s (SD = 11.28) for the non-dominant hand condition. The difference between the dominant and non-dominant hand We found no differences in age [t(80) = –0.365, p =.716], conditions in completion times on the TMT B was on average handedness [t(80) = –0.603, p = .548], gender [χ (1, N =82) = 4.88 s, which is considerably higher than the difference of 0.868, p = .352], and education level [χ (1, N =82) =1.439, p 1.9 s found in previous research (LoSasso et al., 1998). = .230] between the two conditions. The mean completion As expected, a general linear model (GLM) with age as a times for the entire sample for the TMT A were 30.55 s (SD covariate and condition as a fixed factor revealed a main effect = 8.59) measured with a stopwatch and 26.41 s (SD =7.00) of age (p = .047) and no difference in completion times be- measured digitally. This difference was statistically significant tween conditions for the TMT A [F(1, 79) = 0.141, p =.708]. [t(81) = –10.89, p < .0001]. The mean completion times for A GLM including TMT B completion time as the dependent the entire sample for the TMT B were 52.82 s (SD =13.22) variable, age and TMT A as covariates, and condition as a measured with a stopwatch and 48.61 s (SD =12.46)mea- fixed factor revealed an interaction effect of condition and sured digitally, which was also a statistically significant dif- TMT A (p = .001). As can be seen in Fig. 3, the TMT A ference [t(81) = –8.859, p <.0001]. completion time is a good predictor for the TMT B completion In line with previous research (e.g., Tombaugh, 2004), age was found to correlate with both TMT A (r =.223, p =.044) and TMT B (r = .251, p = .023) completion times, which increased with increasing age (see Fig. 1). Neither education nor gender correlated with TMTA (education: r = –.158, p = pb .157; gender: r =.01, p = .926) or TMT B (education: r = pb pb .021, p = .853; gender: r = –.104, p = .351) completion pb times. No differences emerged between men and women in completion times on TMT A [t(80) = –0.093, p =.926] or TMT B [t(80) = 0.938, p =.351]. The effect of hand use on TMT A and TMT B completion times As can be seen in Fig. 2, the mean completion times for the TMT A were 26.06 s (SD = 7.33) for the dominant hand condition and 26.76 s (SD = 6.73) for the non-dominant hand Fig. 2 Mean completion times for TMT A and B for the dominant (black condition. The mean completion times for the TMT B were dots) and non-dominant (yellow dots) hand conditions. Error bars indicate the standard error of the mean 46.17 s (SD = 13.21) for the dominant hand condition and 1080 Behav Res (2018) 50:1074–1087 Fig. 3 Relation between TMT A and B completion times for the dominant (black) and non-dominant (yellow) hand conditions. The dots Fig. 4 Average completion times for the TMT A and B after applying a indicate individual completion times—black dots for participants in the median split on the TMT A times (vertex up means slow TMT A, vertex dominant hand condition, yellow dots for participants in the non-domi- down means fast TMT A). Black lines and symbols represent the nant hand condition. The solid lines are regression lines. The red dashed dominant hand condition, yellow lines and symbols the non-dominant line indicates a B/A ratio of 3, and the black dashed line indicates a B/A hand condition. The error bars are standard errors of the mean ratio of 1 and there were also no differences in age [t(39) = –1.396, p = .172], handedness [t(39) = –0.273, p = .786], gender time in the dominant hand condition (correlation: r =.79, p < 2 2 [χ (1, N =41) = 1.336, p = .248], and education level [χ (1, .0001; slope: ß = 1.42), but not in the non-dominant hand N =41) =0.042, p = .837] between the fast and slow TMT A condition (correlation: r =.229, p = .15; slope: ß=0.38). groups in the non-dominant hand condition. In the dominant Given the significant interaction effect of condition and hand condition, a difference in education level did emerge TMT A in the model, the TMT A completion time and condi- [χ (1, N =41) =5.159, p = .023] between the fast and slow tion were mean-centered for better interpretability of the mod- TMT A groups, with a larger proportion of participants with a el. The GLM revealed a main effect of TMT A completion higher education in the fast TMT A group, but it seems un- time (p < .0001), a trend for age (p = .068), and a trend for likely that this would explain the difference in completion condition [F(1, 77) = 3.757, p = .056]. As can also be seen in times on the TMT B. In the fast TMT A group, the difference Fig. 3, several participants have a B/A ratio score close to or between the dominant and non-dominant hand conditions in above 3, which is considered a cutoff score for set-switching completion times on the TMT B was on average 11.04 s, impairment in clinical practice (Arbuthnott & Frank, 2000). which is considerably higher than the difference of 1.9 s found Of the ten participants with the highest B/A ratio scores (all > in previous research (LoSasso et al., 1998). 2.5), eight were in the non-dominant hand condition and two were in the dominant hand condition. We reanalyzed the data after distributing the participants in TMT B/A ratio a slow TMT A group and a fast TMT A group, based on a median split, to better understand the interaction effect of con- TMT B completion times for participants who performed the dition and TMT A. The slow TMT A group included 41 par- TMT A fast with their non-dominant hand were markedly ticipants (dominant hand, N = 21; non-dominant hand, N = different from those for the other groups. Was this also 20), and the fast TMT A group included 41 participants (dom- reflected in the B/A ratio? The mean B/A ratio score for the inant hand, N = 20; non-dominant hand, N =21).As can be dominant hand condition was 1.8 (SD =0.38), as compared to seen in Fig. 4, we found a significant difference between the 2.02 (SD = 0.68) in the non-dominant hand condition. A two- dominant and non-dominant hand conditions in TMT B com- tailed independent t-test showed that this difference was not pletion times for participants who had a fast completion time statistically significant [t(80) = –1.758, p =.084].When di- on the TMT A [t(39) = –4.125, p < .0001], but not for partic- viding the sample into a fast TMT A and a slow TMT A group ipants who had a slow completion time on the TMTA [t(39) = based on the median split, the difference in B/A ratio scores 0.461, p = .648]. This difference cannot be explained by de- between the dominant and non-dominant hand conditions was mographic factors, since there were no differences in age significant in the fast TMT A group [t(39) = –2.717, p =.01], [t(39) = 0.738, p = .465], handedness [t(39) = 1.226, p but not in the slow TMTA group [t(39) = 0.635, p = .529] (see = .227], and gender [χ (1, N =41) =0.01, p = .92] between the Fig. 5). In the non-dominant hand fast TMT A group, seven participants (35%) had a B/A ratio score higher than 2.5, as fast and slow TMT A groups in the dominant hand condition, Behav Res (2018) 50:1074–1087 1081 Fig. 5 Mean B/A ratios for the fast and slow TMT A groups in the Fig. 6 Mean completion times on each of the five segments for the dominant (black dots) and non-dominant (yellow dots) hand conditions. dominant (black symbols) and non-dominant (yellow symbols) hand The error bars are standard errors of the mean. The red dashed line indi- conditions. Round symbols refer to the TMT A, square symbols to the cates a B/A ratio of 3, and the black dashed line indicates a B/A ratio of 1, TMT B. Error bars are standard errors of the mean which means no additional time cost for switching sets on the TMT B relative to the TMT A compared to only two participants (9.5%) in the dominant =4.614, p = .005] for the TMT A. On the TMT B, a mixed hand fast TMT A group, one participant (4.8%) in the non- ANOVA showed a significant main effect of segment dominant hand slow TMT A group, and no participants in the [F(3.534, 54.104) = 5.492, p = .001] and no interaction effect dominant hand slow TMT A group. Put differently, seven between condition and segment [F(3.534, 17.802) = 1.807, (70%) of the participants with a B/A ratio score higher than p = .136]. The difference between the two conditions showed 2.5 were in the non-dominant hand fast TMT A group. atrend[F(1, 80) = 3.264, p =.075]. These findings suggest that the difference between the Segment-by-segment and element-by-element analysis dominant and non-dominant hand conditions on the TMT B of TMT completion times was due to a general slowing across all segments of the TMT B rather than to a slowing on a specific segment of the To have a more complete understanding of why the differ- test. In the dominant hand condition, participants showed a ences between the dominant (right) and non-dominant (left) pattern on the TMT B similar to that found in previous re- hands occurred on the TMT B, we analyzed completion times search—i.e. they were fast on the first segment, then slowed over TMT segments as defined by Poreh et al. (2012). Both down on the second, third, and fourth segments, and acceler- the TMT A and B were divided into five segments, each ated on the last segment (Poreh et al., 2012). In the non- consisting of five elements—i.e. Segment 1 consisted of dominant hand condition, participants showed a similar pat- Elements 1 to 5 (A) and Elements 1 to 3 (B); Segment 2 tern but were particularly slow on the third segment, although consisted of Elements 6 to 10 (A) and Elements C to E (B); the interaction between condition and segment was not signif- Segment 3 consisted of Elements 11 to 15 (A) and Elements 6 icant, as we mentioned above. to 8 (B); Segment 4 consisted of Elements 16 to 20 (A) and To gain an even more detailed understanding, we explored Elements H to J (B); and Segment 5 consisted of Elements 21 the completion times of the individual elements. In Fig. 7,the to 25 (A) and Elements 11 to 13 (B). Prior research has shown mean completion times per element are plotted for the TMTA that on the TMTA, participants are fastest on the first segment and B in both conditions. On both parts of the TMT, some and slowest on the third segment, and on the TMT B, partic- elements were completed faster than others, which may indi- ipants are fastest on the first segment, slower on Segments 3 cate that these elements have different physical properties or and 4, and then faster again on Segment 5 (Poreh et al., 2012). require different cognitive processes. Moreover, it can be seen Figure 6 shows the mean completion times per segment for that on both parts of the TMT, some elements are completed both the TMT A and B for the dominant and non-dominant faster with one hand than with the other. hand conditions. A mixed analysis of variance (ANOVA) with Figure 8 shows the elements with the biggest differences in condition as between-subjects variable and segment as a completion times between the two conditions. The black ele- within-subjects variable showed a significant interaction ef- ments are the ones on which the dominant (right) hand was faster, and the orange elements are the ones on which the non- fect between condition and segment [F(2.819, 20.426) 1082 Behav Res (2018) 50:1074–1087 As was pointed out by LoSasso and colleagues (1998), the hand may block some elements from view and thereby affect TMT performance. This suggestion has not yet been support- ed by research, but our data seem to confirm that the hand may block some elements from view and thereby affect TMT per- formance. The locations of the elements with different com- pletion times between the two conditions appear to be system- atic to some extent because some elements can be viewed freely with the one hand, whereas they are blocked from view when using the other hand. For some of the elements there is virtually no difference between the two conditions. The dif- ferences are smallest for the elements in green in Fig. 8.For the TMT A, these are Elements 15 (Segment 1), 16 (Segment 2), and 20 and 21 (Segment 3), and for the TMT B, these are Elements 9 and 10 (Segment 2) and 20 and 21 (Segment 4). Fig. 7 Mean completion times for each element of the TMT A (upper panel) and B (lower panel) for the dominant (black symbols) and non- dominant (yellow symbols) hand conditions. Error bars are standard Inter-element variability errors of the mean. Element numbers indicate the order number and do not refer to the content of the elements As we described above, we found that non-dominant hand use affects TMT B but not TMTA completion times. To find further dominant (left) hand was faster. As can be seen in Fig. 8,for support for the hypothesis that this is due to an interference both TMT A and B, all orange elements are on the right of the effect between non-dominant hand use and performance of a preceding element, whereas five of the eight black elements task that has high executive demands, we explored performance are on the left of the preceding element. On the TMT B, the variability in addition to total completion times by using IEV. elements that were completed fastest with the non-dominant In Fig. 9, the mean IEV is shown for the four groups (left) hand—relative to the dominant (right) hand—were (fast/slow TMT A, dominant/non-dominant hand) on both Elements 8 (Segment 1) and 16, 18, and 19 (Segment 4). the TMT A and the TMT B. IEV increases from the TMT A These are all situated to the right of the preceding element, to the TMT B. This was expected, because part B of the except for Element 16. Elements 5 (Segment 1), 11 and 12 TMT requires more executive resources than part A. In addi- (Segment 3), and 17 (Segment 4) were completed faster with tion, it is known from reaction time data that variability in- the dominant (right) hand, and Elements 5, 11, and 12 are creases as reaction time increases (Wagenmakers & Brown, clearly to the left of the preceding element. 2007). As can be seen, all slopes are roughly similar, except Fig. 8 Completion times per element for the TMT A (left) and B (right). green elements had no difference in completion times between the dom- Black elements were completed faster with the dominant (right) hand; inant (right) and non- dominant (left) hand orange elements were completed faster with the non-dominant (left) hand; Behav Res (2018) 50:1074–1087 1083 Fig. 9 Inter-element variability (IEV) as a function of mean Fig. 10 Median completion times for each element of the TMT B with TMT duration. The four lines represent TMT A and B performance for the non-dominant hand, plotted against the median completion time for each of the groups. The letters in the symbols indicate the TMT version. the same element of the TMT B with the dominant hand. Triangles with The change in IEV from TMT A to TMT B is strikingly different for the the vertex facing down are participants with a fast TMT A, and triangles non-dominant hand fast TMT A group with the vertex facing up are participants with a slow TMT A. The solid line is a least-squares linear fit to the data of participants with a fast TMT A, and the fat dashed line is a least-squares linear fit to the data of for the slope of the participants in the non-dominant hand participants with a slow TMTA. The thin dashed line has a slope of 1 and condition who were fast on the TMT A. The slope of the indicates equal performance in the two conditions non-dominant hand fast TMT A group is steeper than the slopes of the other three groups [independent samples t-test: motor processing. If, on the other hand, non-dominant hand t(78) = –3.217, p = .002]. This indicates that for this group, use mainly affected executive functions, the lines would shift IEV increased more from TMT A to TMT B than in the other upward relative to the dashed line, but the slope would remain three groups, which suggests that the executive load from 1. This was based on the notion that all elements would have TMT A to TMT B increased more relative to the other groups. similar completion times in terms of executive processing. As This finding provides additional support for our hypothesis canbe seeninFig. 10, the data are most in line with the latter that non-dominant hand use increases the executive demands hypothesis. Both the fast and slow TMTA groups have slopes of the TMT. smaller than 1. In the left part of Fig. 10 the lines are above the dashed line, showing that the non-dominant hand condition Separation of layout-related processes from executive was relatively slow on elements with short completion processes on the TMT B times—i.e. elements that have low visual search and motor demands. In the slow TMT A group, this is averaged out by Figure 10 shows the median completion times for each ele- faster completion of elements with long completion times. In ment of the TMT B in the non-dominant hand condition, the fast TMTA group, however, the non-dominant hand group plotted against the median completion times for each element is on average slower on elements of the TMT B, independent of the TMT B in the dominant hand condition. The plotted of their completion times with the dominant hand. The results times are median times because these are more robust to ex- of this analysis show that the reduced performance with the treme values than the mean. The two lines show the separate non-dominant relative to the dominant hand is not due to regression lines for participants with a fast TMT A and a slow processes related to the layout of the TMT, but they lend TMT A. The thin dashed line has a slope of 1 and indicates further support to our hypothesis that non-dominant hand equal performance in both conditions. By comparing the two use mainly affects executive functions and therefore interferes regression lines to this line, we can infer whether non- with TMT B performance. dominant hand use primarily affected layout-related or exec- utive processes. If non-dominant hand use primarily affected layout-related processes, the fitted regression lines would have a slope greater than 1, because completion times in the non- Discussion and conclusion dominant hand condition would go up for elements that re- quired more time to process. This was based on the notion that The study has shown that use of the non-dominant hand af- completion times with the dominant hand would also increase fects TMT performance. As we hypothesized, hand use was for elements that had higher demands in terms of visual and found to increase the completion time on the TMT B but not 1084 Behav Res (2018) 50:1074–1087 on the TMT A. This effect was selectively present in a sub- TMT (e.g., Tombaugh, 2004). The difference in mean com- group of participants—i.e. individuals in the non-dominant pletion time for TMT B between age groups 35 to 44 and 45 to hand condition who performed the TMT A fast. As a conse- 54 is about 5 s. An increase in 5 s among individuals between quence, for this group, non-dominant hand use also affected 35 and 44 is equal to at least a 10% drop in percentile when the B/A ratio; of all participants with a ratio higher than 2.5, scoring in the 30% percentile or better (Tombaugh, 2004). 70% were in this group. This finding highlights the impor- When looking specifically at people who were fast on the tance of a detailed exploration of the data, since participants TMT A, the effect of using the non-dominant hand becomes can show substantial differences in their behavior during a even more pronounced. Using the non-dominant hand in- cognitive test like the TMT. On the basis of detailed analyses creased completion time by 11 s in this subgroup. An 11 s of the completion times for individual elements of the increase in completion time on the TMT B is close to the TMT B—in particular, IEV and an analysis of layout-related difference in completion time between the age groups 25 to processes versus executive processes—we found evidence for 34 and 45 to 54—i.e. age groups that are 20 years apart. An our hypothesis that this decrease in performance on the increase in 11 s among individuals between 35 and 44 is equal TMT B is related to non-dominant hand use affecting execu- to at least a 20% drop in percentile when scoring in the 30% tive functions, thereby interfering with TMT B performance. percentile or better (Tombaugh, 2004). As we described above, in contrast to the TMT A, which Furthermore, in our sample of healthy individuals, three mainly reflects visual search and motor speed skills, comple- participants scored on or above the B/A ratio cutoff score of tion of the TMT B also requires higher-order cognitive re- 3 (eight participants had a B/A ratio score higher than 2.5) sources. Based on our findings, non-dominant hand use seems when they performed the TMT with their non-dominant hand. to compete for the same limited cognitive resources, which This was due mostly to a particularly fast completion time on results in a decrease in completion time on the TMT B. We the TMT A and a slow completion time on the TMT B. It discuss the outcomes and their clinical relevance in more de- seems, therefore, that an abnormal B/A ratio score can be tail below. due to hand use and is consequently not a reliable indicator of cognitive deficits if the TMT is performed with the non- Clinical relevance dominant hand. It is important to know how non-dominant hand use affects Digital parameters TMT performance, since patients who are unable to use their dominant hand may perform the test with their non-dominant In the present study, TMT performance was recorded digitally. hand. An alternative to administering the written TMT to this The importance of digital measurement of cognitive function clinical group would be to use the oral TMT (Ricker, Axelrod, has been highlighted by others (Bauer et al., 2012; Poreh et al., &Houtler, 1996). It is, however, important to note that the oral 2012; Salthouse & Fristoe, 1995; Schatz & Browndyke, 2002; TMT has been argued not to be an analogue of the written Woods et al., 2015) because measurements can be done more TMT, but rather a complementary task, because it measures a accurately and in a more standardized way. Moreover, a digital different underlying cognitive construct (Mrazik, Millis, & TMT allows for the recording of additional measures that may Drane, 2010). Moreover, in clinical practice the use of the provide relevant information that is missed in the current pa- written TMT with the non-dominant hand seems to be more per–pencil version of the test, such as segment-by-segment common than the use of the oral TMT, possibly because psy- and element-by-element analysis of the TMT. Research in this chometric and normative data for the oral TMT are sparse area has shown that more detailed analyses of additional pa- (Mrazik et al., 2010). rameters can provide valuable information (Poreh et al., 2012; Since there is currently limited knowledge about how non- Salthouse & Fristoe, 1995; Woods et al., 2015), which is con- dominant hand use affects performance, the present study pro- firmed by the findings of our study. vides insights that are highly relevant by clearly showing that Even though our findings show that a slowing in perfor- TMT completion times and derived scores like the B/A ratio mance on the TMT B with the no-ndominant hand is not due need to be interpreted with caution if a patient uses his non- to layout-related processes, the element-by-element analysis dominant hand to avoid false attribution of increased comple- revealed that some elements were completed faster than tion time and derived scores to cognitive deficits. As the re- others. It has been hypothesized before that slowing with the sults show, an abnormal test performance may be caused by left relative to the right hand (and vice versa) on specific using the non-dominant hand, which in our study resulted in a elements of the TMT is related to the position of the hand mean difference of almost 5 s on the TMT B, which is higher and the fact that the hand obstructs the view of certain ele- than a difference of 1.9 s found in previous research (LoSasso ments (e.g., LoSasso et al., 1998). This hypothesis is in line et al., 1998). A difference of 5 s seems clinically relevant with our data. Generally, elements that are located to the right when comparing it to existing norm scores for the of the preceding element were completed faster with the left Behav Res (2018) 50:1074–1087 1085 hand and elements that are located left to the preceding ele- contrast, people in the non-dominant group who were relative- ment were completed faster with the right hand. It has further- ly slow on the TMT A used the available resources less to more been hypothesized that faster completion of the last seg- enhance their motoric performance, which left room for exec- ment of TMT B is related to a decrease in visual scanning utive processing when performing the TMT B. This kept the needs and may therefore be a more pure measure of executive difference between the TMT A and B in completion time and functioning (Poreh et al., 2012). Our findings confirm that IEV within a normal range. healthy individuals are faster on the last segment of TMT B. Besides exploring IEV, we performed a detailed analysis of However, since we did not find a difference between the dom- the completion times for individual elements of the TMT B inant and non-dominant hand condition in completion time on separating the contribution of executive processes from the the last segment of TMT B, our findings do not corroborate contribution of layout-related processes to the element com- the hypothesis that faster completion of the last segment is due pletion times. This analysis clearly showed that the non- to a decrease in visual scanning needs and a purer measure of dominant hand condition was on average slower on elements executive functioning. of the TMT B independent of the time required for layout- By looking at the total completion times for TMT A and B, related processes. This finding provides additional support for we found support for our hypothesis that non-dominant hand our hypothesis by showing that non-dominant hand use main- use interferes with performance of the TMT B but not the ly affects executive functions rather than layout-related pro- TMT A because completion of TMT B and non-dominant cesses and therefore interferes with TMT B performance. hand use draw on the same limited cognitive resources. We As the detailed analyses demonstrate, digital measurement performed two additional analyses that were possible because clearly provides the opportunity for exploring the specific un- we measured TMT B performance digitally. First, we deter- derlying processes that contribute to a more complete under- mined the IEV on the basis of individual element completion standing of how non-dominant hand use affects TMT B com- times. IEV is analogous to IIV in computerized reaction time pletion. In general, we strongly believe that even though at tasks used in experimental psychology, and it could be an present clinical neuropsychological assessments are conduct- interesting new outcome measure of a computerized TMT. ed in a paper–pencil-based format, it is likely that in the com- As we described above, there is growing interest in perfor- ing years neuropsychological tests will be performed on a mance variability as an additional outcome measure, since it digital medium to an increasing extent. It is, however, impor- is more highly correlated with cognitive dysfunction than is tant to note that although digital neuropsychological assess- the overall reaction time when patients are engaged in cogni- ment offers various benefits, there are a number of important tively demanding tasks involving working memory and set- issues to consider, such as the need to establish the psycho- switching (MacDonald et al., 2006; Strauss et al., 2007; West metric properties of new digital measures (Bauer et al., 2012; Schatz & Browndyke, 2002), the need to understand potential et al., 2002). As expected, we found that IEV was higher on the TMT B than the TMT A. Moreover, we found that three of technological complications and limitations (Bauer et al., the subgroups showed consistent behavior across the TMT A 2012; Cernich, Brennana, Barker, & Bleiberg, 2007), as well and B, since their IEV increased equally from the TMT A to as the need to provide methodological detail regarding the TMT B. The non-dominant hand fast TMT A group, how- computer-based assessment measures to enable replication, ever, showed a larger increase in IEVon the TMT B compared which will eventually contribute to confidence in the system to the other three groups. and method (Schatz & Browndyke, 2002). We believe that We believe the non-dominant hand underperformance on this study contributes to the growing body of research on the TMT B is due to motor control and performance of a task digital measurement of cognitive function by demonstrating that requires executive functions tapping into the same cogni- the added value of digital measurement of the TMT. tive resources. Non-dominant hand use requires more re- sources than dominant hand use because the latter is more automatic. Since only limited resources are available, non- Limitations of the study dominant hand use can reduce the resources available to per- form a task that requires executive functions. As we described In line with previous research, we found a correlation between above, the TMT B has higher executive demands than the completion time and age (Amodio et al., 2002; Fromm-Auch TMT A, and therefore successful completion of the TMT B & Yeudall, 1983; Robins Wahlin et al., 1996; Salthouse & requires a larger share of the available resources. Participants Fristoe, 1995; Tombaugh, 2004). However, in contrast to pre- who were fast on the TMT A used the available resources to vious research, we did not find a correlation between comple- enhance their motoric performance. This worked well on the tion time and education level. A possible explanation for this TMT A and made them relatively fast. However, on the finding is that the education level was relatively high in the TMT B this left too few resources for executive processing, present study. It is therefore possible that the findings of this increasing completion time and IEV disproportionally. In study will not generalize to other parts of the population. 1086 Behav Res (2018) 50:1074–1087 Annett, J., Annett, M., Hudson, P. T. W., & Turner, A. (1979). The control Additionally, the sample included only healthy partici- of movement in the preferred and non-preferred hands. Quarterly pants. On the basis of the notion that use of the non- Journal of Experimental Psychology, 31, 641–652. dominant hand while performing a cognitively demanding Arbuthnott, K., & Frank, J. (2000). Trail Making Test, part B as a measure task interferes with its performance because it relies on of executive control: Validation using a set-switching paradigm. Journal of Clinical and Experimental Neuropsychology, 22, 518– shared cognitive resources, it seems likely that intermanual differences on the TMT will be even more pronounced in Baddeley, A., & Della Salla, S. (1996). Working memory and executive people with cognitive deficits. More research on the underly- control. Philosophical Transactions of the Royal Society B, 351, ing mechanisms for individuals whose cognitive functioning 1397–1403. Bauer, R. M., Iverson, G. L., Cernich, A. N., Binder, L. M., Ruff, R. M., is affected as the result of trauma or disease will therefore be & Naugle, R. I. (2012). Computerized neuropsychological assess- necessary. ment devices: Joint position paper of the American Academy of A third limitation of the study is that the sample consisted Clinical Neuropsychology and the National Academy of only of right-handed individuals. Since our element-by- Neuropsychology. Archives of Clinical Neuropsychology, 27, 362– element analysis suggests that the TMT may be biased for 373. Bowie, C. R., & Harvey, P. D. (2006). Administration and interpretation the left or the right hand, it will be important to replicate this of the Trail Making Test. Nature Protocols, 1, 2277–2281. study including also left-handed participants. Cernich, A. N., Brennana, D. M., Barker, L. M., & Bleiberg, J. (2007). Sources of error in computerized neuropsychological assessment. Archives of Clinical Neuropsychology, 22S, S39–S48. Corrigan, J. D., & Hinkeldey, N. S. (1987). Relationships between parts A Conclusion and B of the Trail Making Test. Journal of Clinical Psychology, 43, 402–409. The present study has shown that use of the non-dominant Cramond, H., Clark, M., & Smith, M. (1989). The effect of using the hand affects performance on the TMT. Performing part B of dominant or nondominant hand on the Rivermead Perceptual Assessment Battery. Clinical Rehabilitation, 3, 215–221. the TMT with the non-dominant hand increases completion Fromm-Auch, D., & Yeudall, L. T. (1983). Normative data for the time, since both using the non-dominant hand and the cogni- Halstead–Reitan neuropsychological tests. Journal of Clinical tive task itself draw on the same cognitive resources. Our Neuropsychology, 5, 221–238. study hints at important clinical consequences of using the Gaudino, E. A., Geisler, M. W., & Squires, N. K. (1995). Construct non-dominant hand. A B/A ratio score close to or higher than validity in the Trail Making Test: What makes part B harder? Journal of Clinical and Experimental Neuropsychology, 17, 529– 3 could be falsely attributed to cognitive dysfunction, whereas at least in some cases a high B/A ratio score may be due to Hausdorff, J. M., Yogev, G., Springer, S., Simon, E. S., & Giladi, N. performing the test with the non-dominant hand. (2005). Walking is more like catching than tapping: Gait in the This study demonstrates the importance of a more detailed elderly as a cognitive complex task. Experimental Brain Research, 164, 541–548. analysis of TMT performance that is possible when it is mea- Jäncke, L., Peters, M., Schlaug, G., Posse, S., Steinmetz, H., & Müller- sured digitally. A more detailed analysis of the different compo- Gärtner, H.-W. (1998). Differential magnetic resonance signal nents of the TMT can be used to better interpret specific out- change in human sensorimotor cortex to finger movements of dif- comes and may eventually be used to improve the reliability of ferent rate of the dominant and subdominant hand. Cognitive Brain Research, 6, 279–284. the TMT. The present study therefore adds to the growing body Kortte, K. B., Horner, M. D., & Windham, W. K. (2002). The Trail of research on the benefits of digital cognitive testing. Making Test, part B: Cognitive flexibility or ability to maintain set? Applied Neuropsychology, 9, 106–109. Acknowledgements The authors would like to thank Julie de Kok and Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Daisy van Minde in assisting in the data collection for the study. Neuropsychological assessment. Oxford: Oxford University Press. Lindenberger, U., Marsiske, M., & Baltes, P. B. (2000). Memorizing Open Access This article is distributed under the terms of the Creative while walking: Increase in dual-task costs from young adulthood Commons Attribution 4.0 International License (http:// to old age. Psychology and Aging, 15, 417–436. creativecommons.org/licenses/by/4.0/), which permits unrestricted use, LoSasso, G. L., Rapport, L. J., Axelrod, B. N., & Reeder, K. P. (1998). distribution, and reproduction in any medium, provided you give appro- Intermanual and alternate-form equivalence on the Trail Making priate credit to the original author(s) and the source, provide a link to the Test. Journal of Clinical and Experimental Neuropsychology, 20, Creative Commons license, and indicate if changes were made. 107–110. MacDonald, S. W. S., Li, S.-C., & Bäckman, L. (2009). Neural under- pinnings of within-person variability in cognitive functioning. Psychology and Aging, 24, 792–808. References MacDonald, S. W., Nyberg, L., & Bäckman, L. (2006). Intra-individual variability in behavior: Links to brain structure, neurotransmission and neuronal activity. Trends in Neurosciences, 29, 474–480. Amodio, P., Wening, H., Del Piccolo, F., Mapelli, D., Montagnese, S., Pellegrini, A., … Umiltà, C. (2002). Variability of trail making test, Mattay, V. S., Callicott, J. H., Bertolino, A., Santha, A. K. S., van Horn, J. symbol digit test and line trait test in normal people: A normative D., Tallent,K.A., … Weinberger, D. R. (1998). Hemispheric control study taking into account age-dependent decline and sociobiological of motor function: A whole brain echo planar fMRI study. Psychiatry Research: Neuroimaging Section, 83,7–22. variables. Aging Clinical and Experimental Research, 14,117–131. Behav Res (2018) 50:1074–1087 1087 Mrazik, M., Millis, S., & Drane, D. L. (2010). The oral Trail Making Test: Theill, N., Martin, M., Schumacher, V., Bridenbaugh, S. A., & Kressig, R. W. (2012). Simultaneously measuring gait and cognitive perfor- Effects of age and concurrent validity. Archives of Clinical Neuropsychology, 25, 236–243. mance in cognitively healthy and cognitively impaired older adults: Poreh, A. M., Miller, A., Dines, P., & Levin, J. (2012). Decomposition of The Basel motor–cognition dual-task paradigm. Journalofthe the Trail Making Test—Reliability and validity of a computer American Geriatric Society, 59, 1012–1018. assisted method for data collection. Archives of Assessment Tombaugh, T. N. (2004). Trail Making Test A and B: Normative data Psychology, 2, 57–72. stratified by age and education. Archives of Clinical Reitan, R. M., & Wolfson, D. (1995). Category test and trail making test Neuropsychology, 19, 203–214. as measures of frontal lobe functions. Clinical Neuropsychologist, 9, Toyokura, M., Ishida, A., Watanabe, F., Okada, N., & Yamazaki, M. 50–56. (2003). Intermanual difference in the Japanese Trail Making Test Ricker, J. H., Axelrod, B. N., & Houtler, B. D. (1996). Clinical validation and its mirror version: Intra-subject comparison of the task- of the oral Trail Making Test. Neuropsychiatry, Neuropsychology, completion time, cognitive time, and motor time. Disability and and Behavioral Neurology, 9, 50–53. Rehabilitation, 25, 1339–1343. Robins Wahlin, T.-B., Bäckman, L., Wahlin, A., & Winblad, B. (1996). Toyokura, M., Sawatari, M., Nishimura, Y., & Ishida, A. (2003). Trail Making Test performance in a community-based sample of Nondominant hand performance of the Japanese Trail Making Test healthy very old adults: Effects of age on completion time, but not and its mirror version. Archives of Physical Medicine and on accuracy. Archives of Gerontology and Geriatrics, 22, 87–102. Rehabilitation, 84, 691–693. Salthouse, T. A., & Fristoe, N. M. (1995). Process analysis of adult age Vlaskamp, B. N., & Hooge, I. T. C. (2006). Crowding degrades saccadic effects on a computer-administered Trail Making Test. search performance. Vision Research, 46, 417–425. Neuropsychology, 9, 518–528. Vlaskamp, B. N., Over, E. A., & Hooge, I. T. C. (2005). Saccadic search Schatz, P., & Browndyke, J. (2002). Applications of computer-based performance: The effect of element spacing. Experimental Brain neuropsychological assessment. The Journal of Head Trauma Research, 167, 246–259. Rehabilitation, 17, 395–410. Wagenmakers, E.-J., & Brown, S. (2007). On the linear relation between Schretlen, D. J., Munro, C. A., Anthony, J. C., & Pearlson, G. D. (2003). the mean and the standard deviation of a response time distribution. Examining the range of normal intraindividual variability in neuro- Psychological Review, 114, 830–841. doi:10.1037/0033-295X.114. psychological test performance. Journal of the International 3.830 Neuropsychological Society, 9, 864–870. Wagner, S., Helmrich, I., Dahmen, N., Lieb, K., & Tadic, A. (2011). Siu, K.-C., Chou, L.-S., Mayr, U., van Donkelaar, P., & Woollacott, M. H. Reliability of three alternate forms of the Trail Making Tests A (2008). Does inability to allocate attention contribute to balance and B. Archives of Clinical Neuropsychology, 26, 314–321. constraints during gait in older adults? Journals of Gerontology, West, R., Murphy, K. J., Armilio, M. L., Craik, F. I. M., & Stuss, D. T. 63A, 1364–1369. doi:10.1093/gerona/63.12.1364 (2002). Lapses of intention and performance variability reveal age- Soukup, V. M., Ingram, F., Grady, J. J., & Schiess, M. C. (1998). Trail related increases in fluctuations of executive control. Brain and Making Test: Issues in normative data selection. Applied Cognition, 49, 402–419. Neuropsychology, 5, 65–73. Woods, D. L., Wyma, J. M., Herron, T. J., & Yund, E. W. (2015). The Strauss, E., Bielak, A. A., Bunce, D., Hunter, M. A., & Hultsch, D. F. effects of aging, malingering, and traumatic brain injury on comput- (2007). Within-person variability in response speed as an indicator erized trail-making test performance. PLoS ONE, 10, e0124345. doi: of cognitive impairment in older adults. Aging, Neuropsychology, 10.1371/journal.pone.0124345 and Cognition, 14, 608–630. Yamashita, H. (2010). Right- and left-hand performance on the Rey– Strenge, H., & Niederberger, U. (2008). Unidirectional interference in use Osterrieth complex figure: A preliminary study in non-clinical sam- of nondominant hand during concurrent grooved pegboard and ran- ple of right handed people. Archives of Clinical Neuropsychology, dom number generation tasks. Perceptual and Motor Skills, 106, 25, 314–317. 763–774. Tanner-Eggen, C., Balzer, C., Perrig, W. J., & Gutbrod, K. (2015). The Yochim, B., Baldo, J., Nelson, A., & Delis, D. C. (2007). D-KEFS Trail neuropsychological assessment of cognitive deficits considering Making Test performance in patients with lateral prefrontal cortex measures of performance variability. Archives of Clinical lesions. Journal of the International Neuropsychological Society, Neuropsychology, 30, 217–227. 13, 704–709. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Behavior Research Methods Springer Journals

Non-dominant hand use increases completion time on part B of the Trail Making Test but not on part A

Free
14 pages

Loading next page...
 
/lp/springer_journal/non-dominant-hand-use-increases-completion-time-on-part-b-of-the-trail-hR5tnr74qZ
Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Psychology; Cognitive Psychology
eISSN
1554-3528
D.O.I.
10.3758/s13428-017-0927-1
Publisher site
See Article on Publisher Site

Abstract

Behav Res (2018) 50:1074–1087 DOI 10.3758/s13428-017-0927-1 BRIEF COMMUNICATION Non-dominant hand use increases completion time on part B of the Trail Making Test but not on part A 1 1 Laura Klaming & Björn N. S. Vlaskamp Published online: 13 July 2017 The Author(s) 2017. This article is an open access publication . . Abstract The Trail Making Test (TMT) is used in neuropsy- Keywords Trail Making Test B/A ratio Executive chological clinical practice to assess aspects of attention and functioning Non-dominant hand use executive function. The test consists of two parts (A and B) and requires drawing a trail between elements. Many patients are assessed with their non-dominant hand because of motor The Trail Making Test (TMT) is a frequently used neuropsy- dysfunction that prevents them from using their dominant chological test to assess aspects of attention and executive hand. Since drawing with the non-dominant hand is not an functions (Bowie & Harvey, 2006; Lezak, Howieson, & automatic task for many people, we explored the effect of Loring, 2004; Tombaugh, 2004;Wagner, Helmrich, hand use on TMT performance. The TMT was administered Dahmen, Lieb, & Tadic, 2011). The TMT consists of drawing digitally in order to analyze new outcome measures in addi- a trail between elements that are quasi-randomly scattered on tion to total completion time. In a sample of 82 healthy par- paper and has two parts. Part A involves making a trail be- ticipants, we found that non-dominant hand use increased tween numbers in ascending order. Part B consists of 13 num- completion times on the TMT B but not on the TMT A. The bers and 12 letters, which the patient is instructed to connect in average completion time increased by almost 5 seconds, an alternating pattern. The patient is asked to complete the which may be clinically relevant. A substantial number of trails as quickly as possible, and completion time is measured participants who performed the TMT with their non- as the main outcome (Bowie & Harvey, 2006; Lezak et al., dominant hand had a B/A ratio score of 2.5 or higher. In 2004). The clinical interpretation of performance on the TMT clinical practice, an abnormally high B/A ratio score may be is based on part A mainly reflecting visual search and motor falsely attributed to cognitive dysfunction. With our digitized speed skills, and part B also requiring higher-order cognitive pen data, we further explored the causes of the reduced functions such as cognitive flexibility and task switching TMT B performance by using new outcome measures, includ- (Lezak et al., 2004). The completion time is longer for ing individual element completion times and interelement var- part B than for part A, and the B–A difference as well as the iability. These measures indicated selective interference be- B/A ratio are considered to reveal deficits in executive func- tween non-dominant hand use and executive functions. Both tion (Arbuthnott & Frank, 2000;Bowie &Harvey, 2006; non-dominant hand use and performance of the TMT B seem Corrigan & Hinkeldey, 1987; Gaudino, Geisler, & Squires, to draw on the same, limited higher-order cognitive resources. 1995; Kortte, Horner, & Windham, 2002; Lezak et al., 2004; Reitan & Wolfson, 1995; Yochim, Baldo, Nelson, & Delis, 2007). Research has consistently shown that performance on the TMT is strongly influenced by age and education, with * Laura Klaming completion times on both parts of the TMT increasing with laura.klaming@philips.com growing age and fewer years of education (Amodio et al., 2002; Fromm-Auch & Yeudall, 1983; Robins Wahlin, Bäckman, Wahlin, & Winblad, 1996; Salthouse & Fristoe, Department of Brain, Behavior and Cognition, Philips Research, 1995; Tombaugh, 2004). High Tech Campus 34, 5656 AE Eindhoven, The Netherlands Behav Res (2018) 50:1074–1087 1075 Another factor that may influence TMT performance in a Imaging research has corroborated the finding that clinical setting is the use of the non-dominant hand. This is an non-dominant hand use interferes with cognitive process- important factor, because patients who suffer from motor dys- ing and shown that performing a simple motor task (se- function that prevents them from using their dominant hand quential finger movements) with the non-dominant hand may be required to perform the TMT with their non-dominant results in greater cortical activity than performing a sim- hand. Currently no norm scores are available for completion ple motor task with the dominant hand and in similar times on the TMT with the non-dominant hand, and little is cortical activity as performing a complex motor task known about how hand use affects TMT performance, which (random finger movements) with the dominant hand makes it difficult to interpret completion times and derived (Mattay et al., 1998). Similarly, another study (Jäncke scores like the B–A difference and the B/A ratio. An increase et al., 1998) has shown that performing a sequential in completion time or derived score may be falsely attributed movement with the non-dominant hand (in right-handed to non-dominant hand use, which may result in an underesti- subjects) results in greater right hemisphere activation mation of cognitive problems; alternatively, an increase in than left hemisphere activationduringperformanceof completion time or the derived scores may be falsely attribut- the same movement with the dominant (right) hand. ed to a cognitive problem rather than hand use, which would These findings indicate that motor movements with the result in an overestimation of cognitive problems. non-dominant hand are less familiar and automatic, and Hand use may impact TMT performance in several ways. therefore consume more cognitive resources. First, research has shown that performance on a motor task is Taken together, these findings suggest that performing a generally slower with the non-dominant hand, which is as- motor task with the non-dominant hand while simultaneously sumed to be mainly attributable to lower movement accuracy performing a cognitive task increases cognitive load, which and a greater need for corrective movements (Annett, Annett, compromises performance on both or one of the two tasks. On Hudson, & Turner, 1979). Research has furthermore shown the basis of the findings of prior research, we therefore hy- that healthy individuals are markedly slower when performing pothesized that performing the TMT A with the non-dominant a standard neuropsychological test involving fine motor skills, hand would not or would only marginally slow down comple- such as copying or cancellation (Cramond, Clark, & Smith, tion times, since there would be little competition for the same 1989) or name writing (Fromm-Auch & Yeudall, 1983), with cognitive resources because the TMT A mainly reflects visual the non-dominant hand. search and motor speed skills. We furthermore hypothesized Second, and more problematically, there is evidence for an that performing the TMT B with the non-dominant hand interference effect between simultaneous performance of motor would increase completion times, since performance of the and cognitive tasks. Studies that have employed a dual-task par- TMT B requires a substantial contribution of higher-order cognitive functions, and therefore both the TMT B and use adigm have consistently shown that simultaneously performing a motor and a cognitive task increases cognitive load and results of the non-dominant hand would compete for the same limited in an interference effect, with performance on both tasks deteri- cognitive resources. orating (Baddeley & Della Salla, 1996; Hausdorff, Yogev, To our knowledge, three studies have explored the effect of Springer, Simon, & Giladi, 2005; Lindenberger, Marsiske, & dominant versus non-dominant hand use on completion times Baltes, 2000; Siu, Chou, Mayr, van Donkelaar, & Woollacott, for the TMT (LoSasso, Rapport, Axelrod, & Reeder, 1998; 2008; Theill, Martin, Schumacher, Bridenbaugh, & Kressig, Toyokura, Ishida, Watanabe, Okada, & Yamazaki, 2003; 2012). The interference effect may be stronger when the motor Toyokura, Sawatari, Nishimura, & Ishida, 2003). LoSasso task involves use of the non-dominant hand. For example, and colleagues compared completion times on the original healthy individuals were found to perform the recall part of the TMT and on a parallel version of the TMT for 40 right- Rey Osterrieth Complex Figure Test significantly worse when handed and 40 left-handed individuals who performed the they had used their non-dominant hand as opposed to their dom- tests with their both dominant and their non-dominant hand. inant hand when copying the figure (Yamashita, 2010). Completion times were found to be slightly longer for the non- According to the author, this finding is due to drawing the figure dominant hand for both the original and the parallel TMT B. with the non-dominant hand leading to fewer cognitive resources This intermanual difference was not significant, however, and being allocated to the performance of the cognitive task—i.e. the was considered clinically irrelevant (LoSasso et al., 1998). copying of the figure. This interpretation is supported by a study Toyokura and colleagues (Toyokura, Ishida, et al. 2003; that found slowed performance on a cognitive test that requires Toyokura, Sawatari, et al., 2003) explored differences in com- executive function—i.e. random number generation—when par- pletion times for the Japanese version of the TMT, which ticipants simultaneously performed a motor speed task—i.e. the consists of numbers (part A) and numbers and Japanese kana grooved pegboard task—with their non-dominant hand. This letters (part B). In both studies, no intermanual difference in dual-task interference effect was not found for the dominant completion times was found (Toyokura, Ishida, et al., 2003; hand (Strenge & Niederberger, 2008). Toyokura, Sawatari, et al., 2003). 1076 Behav Res (2018) 50:1074–1087 Although these three studies have provided interesting in- et al., 2003) and to be able to explore other measures in addi- sights, a number of important limitations make it difficult to tion to total completion time. To our knowledge, this study conclude at this time that there is not a clinically relevant was the first to look at additional outcome measures, besides difference between performing the TMT with the dominant total completion time, that could provide valuable information versus the non-dominant hand. First, completion times were about cognitive functioning and that are not available when measured manually with a stopwatch in all three studies. administering the TMT without pen digitization. Moreover, Manual measurement of completion times may not always this study was the first to look at digitized trails of the tradi- be precise, which may have introduced additional variance tional paper–pencil TMT. unrelated to the participants in these studies. Second, the exact administration procedure of the TMTwas not explained in any of these three studies, and it is therefore unclear what the begin Method and end times of the measurement were and—importantly— to what degree errors have affected completion times. Participants Differences in administration account for the large variability in TMT completion times reported in different studies, which A total of 117 healthy right-handed individuals participated in is problematic and is considered to be an important limitation the study. Handedness was determined with the Edinburgh of the TMT (Soukup, Ingram, Grady, & Schiess, 1998; Handedness Inventory. One participant was found to have a Woods, Wyma, Herron, & Yund, 2015). Treating TMTs with tendency toward left-handedness (score of –.48) and was and without errors as synonymous cognitive tests hampers therefore excluded from further analyses. The data of 11 par- interpretation. Self-correction of an erroneous movement sub- ticipants were not included because of technical issues. Of the stantially increases the total completion time, which may thus remaining 105 participants, 23 (21.9%) made a mistake during reflect additional motoric and cognitive processes. Errors that either the TMT A or the TMT B or both, and were therefore are pointed out by the examiner and then corrected by the excluded from further analyses. Participants who made mis- individual also introduce examiner timing into the completion takes were excluded in order to obtain a pure measure of total time. Therefore, trails with errors are different from trails with- completion time, since as we described above, the correction out errors, which makes the inclusion of TMTs with errors in of errors has a substantial impact on total completion time and research problematic. It is unclear how many participants changes the test. A mistake was defined as any path that de- made errors on the TMT in the three studies described above viated from the correct path—for example, if a participant and whether these individuals were included in the analyses. went from 18 to 20 rather than from 18 to 19. The numbers An important limitation of the Toyokura et al. studies of participants who made mistakes were similar for both con- (Toyokura, Ishida, et al., 2003; Toyokura, Sawatari, et al., ditions (ten in the dominant hand condition and 13 in the non- 2003) is that the Japanese version of the TMTwas used, which dominant hand condition; see below for a description of the is not directly comparable to the original TMT. The ratio be- conditions). The remaining 82 participants (28 women, 54 tween the TMT A and the TMT B is much higher for the men) were all right-handed as measured with the Edinburgh original version of the TMT than for the Japanese version Handedness Inventory (M =.80, SD = .15). They ranged in (Toyokura, Sawatari, et al., 2003), indicating that part B is age from 20 to 65 years of age (M =36 years, SD =12). Of the more difficult than part A (Tombaugh, 2004), which makes participants, 57 (69.5%) had a university degree ranging from it impossible to draw any valid conclusions about intermanual a BSc to a PhD, whereas the remaining 25 had a vocational differences in completion times for the original TMT based on degree as their highest. All participants had normal or the two Toyokura et al. studies. corrected-to-normal vision and were employed at Philips The aim of the present study was to investigate the differ- Research. Individuals were recruited with flyers and were ences in dominant and non-dominant hand use on the TMT in not compensated for their participation. a sample of healthy individuals. More specifically, we tested the hypothesis that non-dominant hand use would increase Materials completion times on the TMT B but would increase comple- tion times less or not at all on the TMT A. This hypothesis is Participants received the traditional paper–pencil TMT. based on the assumption that non-dominant hand use requires Underneath the paper TMT, a Wacom Intuos Pro tablet digi- additional cognitive resources, which interferes with perfor- tized the movements (see also Fig. 8 in the Results). Every mance on the TMT B. In the present experiment, the TMTwas new sheet of paper was aligned to markers on the tablet to administered in the traditional paper–pencil way, but was re- assure that all TMT’s were performed at the exact same loca- corded digitally in an unobtrusive way in order to overcome tion on the tablet. The position of the paper was fixed by some of the shortcomings of previous research (LoSasso et al., taping it to the tablet. We administered the traditional paper– 1998; Toyokura, Ishida, et al., 2003; Toyokura, Sawatari, pencil test to be able to draw conclusions about the TMT as it Behav Res (2018) 50:1074–1087 1077 is currently used in clinical practice. Using a tablet with a a new iteration of the same procedure was executed. This was screen or a computerized TMT to track movements would repeated until the newly calculate threshold was identical to not be suitable for this purpose because it would change the the previous one. physical properties of the test. The Wacom Intuos Pro tablet sampled the pen position at Segment-by-segment and element-by-element analysis 133 Hz. Pen positions were recorded with Movalyzer (devel- Aside from total completion time, we also performed more oped by Neuroscript). Pen pressure was not calibrated and detailed analyses of the trails. For this purpose, we fully auto- therefore not used in the analyses. matically detected the order in which the elements were com- pleted and extracted features per completion. Automatic de- Procedure tection of completion paths was conducted by comparing the spatial pattern within the area closest to an element (or Participants gave written informed consent prior to participa- Voronoi cell) with two simple model patterns. One of those tion. Participants were randomly assigned to complete the model patterns represented completion of the element by TMT either with their dominant hand (N = 41) or with their connecting the element with the pen entry and exit of the non-dominant hand (N = 41). The TMT was administered in Voronoi cell with two lines (one from the entry into the the standard manner, with part A preceding part B. For both Voronoi cell to the element and one from the element to the parts, the standard practice test with eight items was adminis- point of exit out of the Voronoi cell); the other model repre- tered prior to the test. sented no completion with a single line connecting entry and Participants performed the TMT individually in a room exit of the area. The decision whether an element was com- with a research assistant present. The study was conducted pleted or not was made on the basis of the similarity of the data by two different research assistants who were trained to use to the two models (using a two-dimensional least squares the same instructions for all participants. Administration of the method). If both models made similar predictions (e.g., when test took approximately 10 minutes. an element was completed in a straight path), the decision was made on the basis of whether there had been a local drop in Conventional and additional digital parameters pen velocity close to the element (within a radius of 0.5 cm for the TMT from the center of the elements). All classifications were then inspected manually; only eight of a total of 4,100 (i.e., 82 Both conventional and additional digital parameters for the times 25 TMT As + 82 times 25 TMT Bs) classifications were TMT were measured and included in the analyses. incorrect and had to be corrected manually. Conventional parameters include the total completion time, An important aspect of TMT performance we were inter- measured both with a stopwatch and digitally as described ested in was the time needed to move from one element to the below. Additional digital parameters for the TMT include a next—i.e. the element completion time. This was defined as segment-by-segment and element-by-element analysis, the total time required for finding, planning, processing, and interelement variability, and a separation of layout-related pro- executing the movement to the next element. The assumption cesses from executive processes on the TMT B. is that this starts as soon as the preceding element is completed and ends when the target element is completed. We defined Completion time Completion times were determined from the moment of completion as the moment the pen moves with the raw pen position data. They were calculated as the differ- subthreshold velocity as it is approaching the target element. ence in time between the moment the pen touched the tablet at The velocity threshold was calculated in the same way as the first element and the moment the pen stopped at the last described above. element of the TMT. The pen stop was defined as the velocity With this algorithm for splitting the TMT trail and calcu- of the pen dropping below threshold velocity within 1 cm of lating element completion times, we first explored whether the center of the last element. Velocity was calculated as the any difference between dominant and non-dominant hand instantaneous velocity using a second-order polynomial fit to use was related to performance on specific elements of the interpolate within a window of five samples centered on the TMT. We analyzed the data in the same way as Poreh, sample of interest. Threshold velocity was adaptively deter- Miller, Dines, and Levin (2012), whousedacomputer- mined through an iterative process for each individual assisted version of the TMT. The participants performed the TMT trial to account for noise in the tablet as well as human TMT on paper as usual, and with every element completion motor noise. First, a threshold was set arbitrarily. Next, a new the experimenter clicked with a mouse on a button that repre- threshold was calculated as five times the standard deviation sented the respective element. Poreh et al. divided the TMT A above the mean of the velocity of all samples below the and TMT B into five segments and calculated the time needed predefined threshold. If the resulting value was lower than to complete each of those segments. They found that the last the predefined threshold, it was set as the new threshold and part of the TMT B is particularly related to executive 1078 Behav Res (2018) 50:1074–1087 functioning and hypothesized that this may be related to low indication about an individual’s consistency across a task or search demands, since most elements are cancelled by then. In across multiple tasks or sessions, and can therefore provide our study, this translated to the prediction that a difference additional information about the individual’s cognitive func- between the two groups might be particularly evident in the tioning. The reason for the growing interest in IIV in the field last segment of the TMT B. of neuropsychology is that this measure is considered an in- We also inspected the trails in further detail on an element- formative measure because it is more highly correlated with by-element basis to identify any differences between domi- cognitive dysfunction than is the overall reaction time when nant hand and non-dominant hand use. Aside from differences patients are engaged in cognitively demanding tasks involving in individual element completion times related to the hypoth- working memory and set switching (MacDonald et al., 2009; esis of Poreh et al. (2012), using the left hand might also MacDonald et al., 2006; Strauss et al., 2007;West etal., introduce an effect on individual element completion times 2002). In a clinical context, IIV may provide more insight into because the hand covers different parts of the TMT, which the cognitive status of patients and allow for more accurate may make it either easier or harder to find certain elements. interpretations of test outcomes (Schretlen et al., 2003; The element-by-element analysis allowed for an investigation Tanner-Eggen et al., 2015). of hand bias of the TMT. Separation of layout-related processes from executive pro- Inter-element variability In the scientific literature and in cesses on TMT B The TMT B introduces a set-switching clinical neuropsychology, there is growing interest in intra- task, which causes an increment in completion times relative individual variability (IIV; e.g., MacDonald, Li, & Bäckman, to the TMT A, which is assumed to be due to increased exec- 2009; MacDonald, Nyberg, & Bäckman, 2006; Schretlen, utive demands. However, performing the TMT also involves Munro, Anthony, & Pearlson, 2003; Strauss, Bielak, Bunce, other (cognitive) tasks, such as visual search and moving the Hunter, & Hultsch, 2007; Tanner-Eggen, Balzer, Perrig, & pen from one element to the next, which are mostly related to Gutbrod, 2015; West, Murphy, Armilio, Craik, & Stuss, the layout of the TMT. Even though the TMT A and B were 2002). The majority of research on IIV has focused on vari- not designed to differ on these tasks, they do (Gaudino et al., ability in reaction time tasks, where IIV refers to changes in 1995), which leaves open the possibility that an increase in the reaction time data within an individual on a particular task completion time on the TMT B with the non-dominant hand rather than in the mean reaction time. Reaction time tasks could be due to interference with layout-related processes consist of many trials, and because multiple measurements rather than with executive functions. are collected per individual, IIV can be calculated. With the Here we sought evidence that non-dominant hand use in- standard paper–pencil TMT there is no way to get insight into terferes with executive processes on the TMT B by explicitly the variability in performance within an individual participant, separating the contribution of executive processes from the because only the overall completion time is measured. contribution of layout-related processes to the element com- However, with completion times per element derived from pletion times. For executive tasks, we assumed that the pro- the digital pen recordings a measure very similar to the IIV cessing time required for the completion of each element of can be calculated, a measure we call the inter-element the TMT B was roughly equal (for sake of the argument, we variability (IEV). We derived this measure from the digital ignore that there might be slight differences between the ele- pen recordings as follows. First, a distribution was compiled ments in terms of set-switching; e.g., it may be easier to go of the completion times per participant for all elements on the from letters to numbers than vice versa, it may be easier to TMT except for the first and last elements. Next, the IEV was keep the first letters of the alphabet in working memory than calculated as the difference between the 10% and 90% cuts later letters, etc.). This means that if non-dominant hand use through the distribution, as is typically done to calculate IIV. were to mainly affect executive functions, the same amount of Because the completion times are determined by the partici- additional processing would be expected for every element pant but also by the element characteristics (e.g., the time completed with the non-dominant hand relative to the process- needed to find visual information is dependent on its visual ing time for every element completed with the dominant hand. eccentricity and on the distance to neighboring information; For tasks related to the layout of the TMT (such as visual see, e.g., Vlaskamp, Over, & Hooge, 2005), the element com- search and motor tasks), we assumed that a different amount pletion times needed to be normalized prior to further process- of processing time would be required for each element, de- ing. Normalization was done by dividing each individual ele- pending on its spatial configuration: Some distances between ment completion time by the median completion time for each subsequent elements are very long whereas others are short; element across participants within a condition. These normal- some elements are in more cluttered areas than others; and ized completion times were then used to calculate the IEV. sometimes elements are positioned in between subsequent Currently in clinical neuropsychology, the mean comple- elements. These factors are known to have a large impact on tion time is a more typical outcome measure. IIV gives an tasks such as visual search (Vlaskamp & Hooge, 2006; Behav Res (2018) 50:1074–1087 1079 Vlaskamp, Over, & Hooge, 2005) and motor processes. On the basis of this assumption, any interference of non-dominant hand use with layout-related processes would be particularly noticeable on elements that already required long processing times with the dominant hand. In short, an across-the-board increment in element completion times in the non-dominant hand condition would be related to executive processes, but systematic increments with longer element completion times in the non-dominant hand condition would be related to layout-related processes. Statistical analyses All statistical analyses were performed using SPSS (version 23). These statistical analyses included independent-samples t Fig. 1 Completion times on the TMTA (left panel) and B (right panel) as tests, chi-square tests, and general linear models. Correlation a function of age. Each dot represents the completion time of one analyses (Pearson and point-biserial) were used to determine participant—black dots for participants in the dominant hand condition, the impact of potential covariates. and yellow dots for participants in the non-dominant hand condition. The lines are regression lines Results 51.05 s (SD = 11.28) for the non-dominant hand condition. The difference between the dominant and non-dominant hand We found no differences in age [t(80) = –0.365, p =.716], conditions in completion times on the TMT B was on average handedness [t(80) = –0.603, p = .548], gender [χ (1, N =82) = 4.88 s, which is considerably higher than the difference of 0.868, p = .352], and education level [χ (1, N =82) =1.439, p 1.9 s found in previous research (LoSasso et al., 1998). = .230] between the two conditions. The mean completion As expected, a general linear model (GLM) with age as a times for the entire sample for the TMT A were 30.55 s (SD covariate and condition as a fixed factor revealed a main effect = 8.59) measured with a stopwatch and 26.41 s (SD =7.00) of age (p = .047) and no difference in completion times be- measured digitally. This difference was statistically significant tween conditions for the TMT A [F(1, 79) = 0.141, p =.708]. [t(81) = –10.89, p < .0001]. The mean completion times for A GLM including TMT B completion time as the dependent the entire sample for the TMT B were 52.82 s (SD =13.22) variable, age and TMT A as covariates, and condition as a measured with a stopwatch and 48.61 s (SD =12.46)mea- fixed factor revealed an interaction effect of condition and sured digitally, which was also a statistically significant dif- TMT A (p = .001). As can be seen in Fig. 3, the TMT A ference [t(81) = –8.859, p <.0001]. completion time is a good predictor for the TMT B completion In line with previous research (e.g., Tombaugh, 2004), age was found to correlate with both TMT A (r =.223, p =.044) and TMT B (r = .251, p = .023) completion times, which increased with increasing age (see Fig. 1). Neither education nor gender correlated with TMTA (education: r = –.158, p = pb .157; gender: r =.01, p = .926) or TMT B (education: r = pb pb .021, p = .853; gender: r = –.104, p = .351) completion pb times. No differences emerged between men and women in completion times on TMT A [t(80) = –0.093, p =.926] or TMT B [t(80) = 0.938, p =.351]. The effect of hand use on TMT A and TMT B completion times As can be seen in Fig. 2, the mean completion times for the TMT A were 26.06 s (SD = 7.33) for the dominant hand condition and 26.76 s (SD = 6.73) for the non-dominant hand Fig. 2 Mean completion times for TMT A and B for the dominant (black condition. The mean completion times for the TMT B were dots) and non-dominant (yellow dots) hand conditions. Error bars indicate the standard error of the mean 46.17 s (SD = 13.21) for the dominant hand condition and 1080 Behav Res (2018) 50:1074–1087 Fig. 3 Relation between TMT A and B completion times for the dominant (black) and non-dominant (yellow) hand conditions. The dots Fig. 4 Average completion times for the TMT A and B after applying a indicate individual completion times—black dots for participants in the median split on the TMT A times (vertex up means slow TMT A, vertex dominant hand condition, yellow dots for participants in the non-domi- down means fast TMT A). Black lines and symbols represent the nant hand condition. The solid lines are regression lines. The red dashed dominant hand condition, yellow lines and symbols the non-dominant line indicates a B/A ratio of 3, and the black dashed line indicates a B/A hand condition. The error bars are standard errors of the mean ratio of 1 and there were also no differences in age [t(39) = –1.396, p = .172], handedness [t(39) = –0.273, p = .786], gender time in the dominant hand condition (correlation: r =.79, p < 2 2 [χ (1, N =41) = 1.336, p = .248], and education level [χ (1, .0001; slope: ß = 1.42), but not in the non-dominant hand N =41) =0.042, p = .837] between the fast and slow TMT A condition (correlation: r =.229, p = .15; slope: ß=0.38). groups in the non-dominant hand condition. In the dominant Given the significant interaction effect of condition and hand condition, a difference in education level did emerge TMT A in the model, the TMT A completion time and condi- [χ (1, N =41) =5.159, p = .023] between the fast and slow tion were mean-centered for better interpretability of the mod- TMT A groups, with a larger proportion of participants with a el. The GLM revealed a main effect of TMT A completion higher education in the fast TMT A group, but it seems un- time (p < .0001), a trend for age (p = .068), and a trend for likely that this would explain the difference in completion condition [F(1, 77) = 3.757, p = .056]. As can also be seen in times on the TMT B. In the fast TMT A group, the difference Fig. 3, several participants have a B/A ratio score close to or between the dominant and non-dominant hand conditions in above 3, which is considered a cutoff score for set-switching completion times on the TMT B was on average 11.04 s, impairment in clinical practice (Arbuthnott & Frank, 2000). which is considerably higher than the difference of 1.9 s found Of the ten participants with the highest B/A ratio scores (all > in previous research (LoSasso et al., 1998). 2.5), eight were in the non-dominant hand condition and two were in the dominant hand condition. We reanalyzed the data after distributing the participants in TMT B/A ratio a slow TMT A group and a fast TMT A group, based on a median split, to better understand the interaction effect of con- TMT B completion times for participants who performed the dition and TMT A. The slow TMT A group included 41 par- TMT A fast with their non-dominant hand were markedly ticipants (dominant hand, N = 21; non-dominant hand, N = different from those for the other groups. Was this also 20), and the fast TMT A group included 41 participants (dom- reflected in the B/A ratio? The mean B/A ratio score for the inant hand, N = 20; non-dominant hand, N =21).As can be dominant hand condition was 1.8 (SD =0.38), as compared to seen in Fig. 4, we found a significant difference between the 2.02 (SD = 0.68) in the non-dominant hand condition. A two- dominant and non-dominant hand conditions in TMT B com- tailed independent t-test showed that this difference was not pletion times for participants who had a fast completion time statistically significant [t(80) = –1.758, p =.084].When di- on the TMT A [t(39) = –4.125, p < .0001], but not for partic- viding the sample into a fast TMT A and a slow TMT A group ipants who had a slow completion time on the TMTA [t(39) = based on the median split, the difference in B/A ratio scores 0.461, p = .648]. This difference cannot be explained by de- between the dominant and non-dominant hand conditions was mographic factors, since there were no differences in age significant in the fast TMT A group [t(39) = –2.717, p =.01], [t(39) = 0.738, p = .465], handedness [t(39) = 1.226, p but not in the slow TMTA group [t(39) = 0.635, p = .529] (see = .227], and gender [χ (1, N =41) =0.01, p = .92] between the Fig. 5). In the non-dominant hand fast TMT A group, seven participants (35%) had a B/A ratio score higher than 2.5, as fast and slow TMT A groups in the dominant hand condition, Behav Res (2018) 50:1074–1087 1081 Fig. 5 Mean B/A ratios for the fast and slow TMT A groups in the Fig. 6 Mean completion times on each of the five segments for the dominant (black dots) and non-dominant (yellow dots) hand conditions. dominant (black symbols) and non-dominant (yellow symbols) hand The error bars are standard errors of the mean. The red dashed line indi- conditions. Round symbols refer to the TMT A, square symbols to the cates a B/A ratio of 3, and the black dashed line indicates a B/A ratio of 1, TMT B. Error bars are standard errors of the mean which means no additional time cost for switching sets on the TMT B relative to the TMT A compared to only two participants (9.5%) in the dominant =4.614, p = .005] for the TMT A. On the TMT B, a mixed hand fast TMT A group, one participant (4.8%) in the non- ANOVA showed a significant main effect of segment dominant hand slow TMT A group, and no participants in the [F(3.534, 54.104) = 5.492, p = .001] and no interaction effect dominant hand slow TMT A group. Put differently, seven between condition and segment [F(3.534, 17.802) = 1.807, (70%) of the participants with a B/A ratio score higher than p = .136]. The difference between the two conditions showed 2.5 were in the non-dominant hand fast TMT A group. atrend[F(1, 80) = 3.264, p =.075]. These findings suggest that the difference between the Segment-by-segment and element-by-element analysis dominant and non-dominant hand conditions on the TMT B of TMT completion times was due to a general slowing across all segments of the TMT B rather than to a slowing on a specific segment of the To have a more complete understanding of why the differ- test. In the dominant hand condition, participants showed a ences between the dominant (right) and non-dominant (left) pattern on the TMT B similar to that found in previous re- hands occurred on the TMT B, we analyzed completion times search—i.e. they were fast on the first segment, then slowed over TMT segments as defined by Poreh et al. (2012). Both down on the second, third, and fourth segments, and acceler- the TMT A and B were divided into five segments, each ated on the last segment (Poreh et al., 2012). In the non- consisting of five elements—i.e. Segment 1 consisted of dominant hand condition, participants showed a similar pat- Elements 1 to 5 (A) and Elements 1 to 3 (B); Segment 2 tern but were particularly slow on the third segment, although consisted of Elements 6 to 10 (A) and Elements C to E (B); the interaction between condition and segment was not signif- Segment 3 consisted of Elements 11 to 15 (A) and Elements 6 icant, as we mentioned above. to 8 (B); Segment 4 consisted of Elements 16 to 20 (A) and To gain an even more detailed understanding, we explored Elements H to J (B); and Segment 5 consisted of Elements 21 the completion times of the individual elements. In Fig. 7,the to 25 (A) and Elements 11 to 13 (B). Prior research has shown mean completion times per element are plotted for the TMTA that on the TMTA, participants are fastest on the first segment and B in both conditions. On both parts of the TMT, some and slowest on the third segment, and on the TMT B, partic- elements were completed faster than others, which may indi- ipants are fastest on the first segment, slower on Segments 3 cate that these elements have different physical properties or and 4, and then faster again on Segment 5 (Poreh et al., 2012). require different cognitive processes. Moreover, it can be seen Figure 6 shows the mean completion times per segment for that on both parts of the TMT, some elements are completed both the TMT A and B for the dominant and non-dominant faster with one hand than with the other. hand conditions. A mixed analysis of variance (ANOVA) with Figure 8 shows the elements with the biggest differences in condition as between-subjects variable and segment as a completion times between the two conditions. The black ele- within-subjects variable showed a significant interaction ef- ments are the ones on which the dominant (right) hand was faster, and the orange elements are the ones on which the non- fect between condition and segment [F(2.819, 20.426) 1082 Behav Res (2018) 50:1074–1087 As was pointed out by LoSasso and colleagues (1998), the hand may block some elements from view and thereby affect TMT performance. This suggestion has not yet been support- ed by research, but our data seem to confirm that the hand may block some elements from view and thereby affect TMT per- formance. The locations of the elements with different com- pletion times between the two conditions appear to be system- atic to some extent because some elements can be viewed freely with the one hand, whereas they are blocked from view when using the other hand. For some of the elements there is virtually no difference between the two conditions. The dif- ferences are smallest for the elements in green in Fig. 8.For the TMT A, these are Elements 15 (Segment 1), 16 (Segment 2), and 20 and 21 (Segment 3), and for the TMT B, these are Elements 9 and 10 (Segment 2) and 20 and 21 (Segment 4). Fig. 7 Mean completion times for each element of the TMT A (upper panel) and B (lower panel) for the dominant (black symbols) and non- dominant (yellow symbols) hand conditions. Error bars are standard Inter-element variability errors of the mean. Element numbers indicate the order number and do not refer to the content of the elements As we described above, we found that non-dominant hand use affects TMT B but not TMTA completion times. To find further dominant (left) hand was faster. As can be seen in Fig. 8,for support for the hypothesis that this is due to an interference both TMT A and B, all orange elements are on the right of the effect between non-dominant hand use and performance of a preceding element, whereas five of the eight black elements task that has high executive demands, we explored performance are on the left of the preceding element. On the TMT B, the variability in addition to total completion times by using IEV. elements that were completed fastest with the non-dominant In Fig. 9, the mean IEV is shown for the four groups (left) hand—relative to the dominant (right) hand—were (fast/slow TMT A, dominant/non-dominant hand) on both Elements 8 (Segment 1) and 16, 18, and 19 (Segment 4). the TMT A and the TMT B. IEV increases from the TMT A These are all situated to the right of the preceding element, to the TMT B. This was expected, because part B of the except for Element 16. Elements 5 (Segment 1), 11 and 12 TMT requires more executive resources than part A. In addi- (Segment 3), and 17 (Segment 4) were completed faster with tion, it is known from reaction time data that variability in- the dominant (right) hand, and Elements 5, 11, and 12 are creases as reaction time increases (Wagenmakers & Brown, clearly to the left of the preceding element. 2007). As can be seen, all slopes are roughly similar, except Fig. 8 Completion times per element for the TMT A (left) and B (right). green elements had no difference in completion times between the dom- Black elements were completed faster with the dominant (right) hand; inant (right) and non- dominant (left) hand orange elements were completed faster with the non-dominant (left) hand; Behav Res (2018) 50:1074–1087 1083 Fig. 9 Inter-element variability (IEV) as a function of mean Fig. 10 Median completion times for each element of the TMT B with TMT duration. The four lines represent TMT A and B performance for the non-dominant hand, plotted against the median completion time for each of the groups. The letters in the symbols indicate the TMT version. the same element of the TMT B with the dominant hand. Triangles with The change in IEV from TMT A to TMT B is strikingly different for the the vertex facing down are participants with a fast TMT A, and triangles non-dominant hand fast TMT A group with the vertex facing up are participants with a slow TMT A. The solid line is a least-squares linear fit to the data of participants with a fast TMT A, and the fat dashed line is a least-squares linear fit to the data of for the slope of the participants in the non-dominant hand participants with a slow TMTA. The thin dashed line has a slope of 1 and condition who were fast on the TMT A. The slope of the indicates equal performance in the two conditions non-dominant hand fast TMT A group is steeper than the slopes of the other three groups [independent samples t-test: motor processing. If, on the other hand, non-dominant hand t(78) = –3.217, p = .002]. This indicates that for this group, use mainly affected executive functions, the lines would shift IEV increased more from TMT A to TMT B than in the other upward relative to the dashed line, but the slope would remain three groups, which suggests that the executive load from 1. This was based on the notion that all elements would have TMT A to TMT B increased more relative to the other groups. similar completion times in terms of executive processing. As This finding provides additional support for our hypothesis canbe seeninFig. 10, the data are most in line with the latter that non-dominant hand use increases the executive demands hypothesis. Both the fast and slow TMTA groups have slopes of the TMT. smaller than 1. In the left part of Fig. 10 the lines are above the dashed line, showing that the non-dominant hand condition Separation of layout-related processes from executive was relatively slow on elements with short completion processes on the TMT B times—i.e. elements that have low visual search and motor demands. In the slow TMT A group, this is averaged out by Figure 10 shows the median completion times for each ele- faster completion of elements with long completion times. In ment of the TMT B in the non-dominant hand condition, the fast TMTA group, however, the non-dominant hand group plotted against the median completion times for each element is on average slower on elements of the TMT B, independent of the TMT B in the dominant hand condition. The plotted of their completion times with the dominant hand. The results times are median times because these are more robust to ex- of this analysis show that the reduced performance with the treme values than the mean. The two lines show the separate non-dominant relative to the dominant hand is not due to regression lines for participants with a fast TMT A and a slow processes related to the layout of the TMT, but they lend TMT A. The thin dashed line has a slope of 1 and indicates further support to our hypothesis that non-dominant hand equal performance in both conditions. By comparing the two use mainly affects executive functions and therefore interferes regression lines to this line, we can infer whether non- with TMT B performance. dominant hand use primarily affected layout-related or exec- utive processes. If non-dominant hand use primarily affected layout-related processes, the fitted regression lines would have a slope greater than 1, because completion times in the non- Discussion and conclusion dominant hand condition would go up for elements that re- quired more time to process. This was based on the notion that The study has shown that use of the non-dominant hand af- completion times with the dominant hand would also increase fects TMT performance. As we hypothesized, hand use was for elements that had higher demands in terms of visual and found to increase the completion time on the TMT B but not 1084 Behav Res (2018) 50:1074–1087 on the TMT A. This effect was selectively present in a sub- TMT (e.g., Tombaugh, 2004). The difference in mean com- group of participants—i.e. individuals in the non-dominant pletion time for TMT B between age groups 35 to 44 and 45 to hand condition who performed the TMT A fast. As a conse- 54 is about 5 s. An increase in 5 s among individuals between quence, for this group, non-dominant hand use also affected 35 and 44 is equal to at least a 10% drop in percentile when the B/A ratio; of all participants with a ratio higher than 2.5, scoring in the 30% percentile or better (Tombaugh, 2004). 70% were in this group. This finding highlights the impor- When looking specifically at people who were fast on the tance of a detailed exploration of the data, since participants TMT A, the effect of using the non-dominant hand becomes can show substantial differences in their behavior during a even more pronounced. Using the non-dominant hand in- cognitive test like the TMT. On the basis of detailed analyses creased completion time by 11 s in this subgroup. An 11 s of the completion times for individual elements of the increase in completion time on the TMT B is close to the TMT B—in particular, IEV and an analysis of layout-related difference in completion time between the age groups 25 to processes versus executive processes—we found evidence for 34 and 45 to 54—i.e. age groups that are 20 years apart. An our hypothesis that this decrease in performance on the increase in 11 s among individuals between 35 and 44 is equal TMT B is related to non-dominant hand use affecting execu- to at least a 20% drop in percentile when scoring in the 30% tive functions, thereby interfering with TMT B performance. percentile or better (Tombaugh, 2004). As we described above, in contrast to the TMT A, which Furthermore, in our sample of healthy individuals, three mainly reflects visual search and motor speed skills, comple- participants scored on or above the B/A ratio cutoff score of tion of the TMT B also requires higher-order cognitive re- 3 (eight participants had a B/A ratio score higher than 2.5) sources. Based on our findings, non-dominant hand use seems when they performed the TMT with their non-dominant hand. to compete for the same limited cognitive resources, which This was due mostly to a particularly fast completion time on results in a decrease in completion time on the TMT B. We the TMT A and a slow completion time on the TMT B. It discuss the outcomes and their clinical relevance in more de- seems, therefore, that an abnormal B/A ratio score can be tail below. due to hand use and is consequently not a reliable indicator of cognitive deficits if the TMT is performed with the non- Clinical relevance dominant hand. It is important to know how non-dominant hand use affects Digital parameters TMT performance, since patients who are unable to use their dominant hand may perform the test with their non-dominant In the present study, TMT performance was recorded digitally. hand. An alternative to administering the written TMT to this The importance of digital measurement of cognitive function clinical group would be to use the oral TMT (Ricker, Axelrod, has been highlighted by others (Bauer et al., 2012; Poreh et al., &Houtler, 1996). It is, however, important to note that the oral 2012; Salthouse & Fristoe, 1995; Schatz & Browndyke, 2002; TMT has been argued not to be an analogue of the written Woods et al., 2015) because measurements can be done more TMT, but rather a complementary task, because it measures a accurately and in a more standardized way. Moreover, a digital different underlying cognitive construct (Mrazik, Millis, & TMT allows for the recording of additional measures that may Drane, 2010). Moreover, in clinical practice the use of the provide relevant information that is missed in the current pa- written TMT with the non-dominant hand seems to be more per–pencil version of the test, such as segment-by-segment common than the use of the oral TMT, possibly because psy- and element-by-element analysis of the TMT. Research in this chometric and normative data for the oral TMT are sparse area has shown that more detailed analyses of additional pa- (Mrazik et al., 2010). rameters can provide valuable information (Poreh et al., 2012; Since there is currently limited knowledge about how non- Salthouse & Fristoe, 1995; Woods et al., 2015), which is con- dominant hand use affects performance, the present study pro- firmed by the findings of our study. vides insights that are highly relevant by clearly showing that Even though our findings show that a slowing in perfor- TMT completion times and derived scores like the B/A ratio mance on the TMT B with the no-ndominant hand is not due need to be interpreted with caution if a patient uses his non- to layout-related processes, the element-by-element analysis dominant hand to avoid false attribution of increased comple- revealed that some elements were completed faster than tion time and derived scores to cognitive deficits. As the re- others. It has been hypothesized before that slowing with the sults show, an abnormal test performance may be caused by left relative to the right hand (and vice versa) on specific using the non-dominant hand, which in our study resulted in a elements of the TMT is related to the position of the hand mean difference of almost 5 s on the TMT B, which is higher and the fact that the hand obstructs the view of certain ele- than a difference of 1.9 s found in previous research (LoSasso ments (e.g., LoSasso et al., 1998). This hypothesis is in line et al., 1998). A difference of 5 s seems clinically relevant with our data. Generally, elements that are located to the right when comparing it to existing norm scores for the of the preceding element were completed faster with the left Behav Res (2018) 50:1074–1087 1085 hand and elements that are located left to the preceding ele- contrast, people in the non-dominant group who were relative- ment were completed faster with the right hand. It has further- ly slow on the TMT A used the available resources less to more been hypothesized that faster completion of the last seg- enhance their motoric performance, which left room for exec- ment of TMT B is related to a decrease in visual scanning utive processing when performing the TMT B. This kept the needs and may therefore be a more pure measure of executive difference between the TMT A and B in completion time and functioning (Poreh et al., 2012). Our findings confirm that IEV within a normal range. healthy individuals are faster on the last segment of TMT B. Besides exploring IEV, we performed a detailed analysis of However, since we did not find a difference between the dom- the completion times for individual elements of the TMT B inant and non-dominant hand condition in completion time on separating the contribution of executive processes from the the last segment of TMT B, our findings do not corroborate contribution of layout-related processes to the element com- the hypothesis that faster completion of the last segment is due pletion times. This analysis clearly showed that the non- to a decrease in visual scanning needs and a purer measure of dominant hand condition was on average slower on elements executive functioning. of the TMT B independent of the time required for layout- By looking at the total completion times for TMT A and B, related processes. This finding provides additional support for we found support for our hypothesis that non-dominant hand our hypothesis by showing that non-dominant hand use main- use interferes with performance of the TMT B but not the ly affects executive functions rather than layout-related pro- TMT A because completion of TMT B and non-dominant cesses and therefore interferes with TMT B performance. hand use draw on the same limited cognitive resources. We As the detailed analyses demonstrate, digital measurement performed two additional analyses that were possible because clearly provides the opportunity for exploring the specific un- we measured TMT B performance digitally. First, we deter- derlying processes that contribute to a more complete under- mined the IEV on the basis of individual element completion standing of how non-dominant hand use affects TMT B com- times. IEV is analogous to IIV in computerized reaction time pletion. In general, we strongly believe that even though at tasks used in experimental psychology, and it could be an present clinical neuropsychological assessments are conduct- interesting new outcome measure of a computerized TMT. ed in a paper–pencil-based format, it is likely that in the com- As we described above, there is growing interest in perfor- ing years neuropsychological tests will be performed on a mance variability as an additional outcome measure, since it digital medium to an increasing extent. It is, however, impor- is more highly correlated with cognitive dysfunction than is tant to note that although digital neuropsychological assess- the overall reaction time when patients are engaged in cogni- ment offers various benefits, there are a number of important tively demanding tasks involving working memory and set- issues to consider, such as the need to establish the psycho- switching (MacDonald et al., 2006; Strauss et al., 2007; West metric properties of new digital measures (Bauer et al., 2012; Schatz & Browndyke, 2002), the need to understand potential et al., 2002). As expected, we found that IEV was higher on the TMT B than the TMT A. Moreover, we found that three of technological complications and limitations (Bauer et al., the subgroups showed consistent behavior across the TMT A 2012; Cernich, Brennana, Barker, & Bleiberg, 2007), as well and B, since their IEV increased equally from the TMT A to as the need to provide methodological detail regarding the TMT B. The non-dominant hand fast TMT A group, how- computer-based assessment measures to enable replication, ever, showed a larger increase in IEVon the TMT B compared which will eventually contribute to confidence in the system to the other three groups. and method (Schatz & Browndyke, 2002). We believe that We believe the non-dominant hand underperformance on this study contributes to the growing body of research on the TMT B is due to motor control and performance of a task digital measurement of cognitive function by demonstrating that requires executive functions tapping into the same cogni- the added value of digital measurement of the TMT. tive resources. Non-dominant hand use requires more re- sources than dominant hand use because the latter is more automatic. Since only limited resources are available, non- Limitations of the study dominant hand use can reduce the resources available to per- form a task that requires executive functions. As we described In line with previous research, we found a correlation between above, the TMT B has higher executive demands than the completion time and age (Amodio et al., 2002; Fromm-Auch TMT A, and therefore successful completion of the TMT B & Yeudall, 1983; Robins Wahlin et al., 1996; Salthouse & requires a larger share of the available resources. Participants Fristoe, 1995; Tombaugh, 2004). However, in contrast to pre- who were fast on the TMT A used the available resources to vious research, we did not find a correlation between comple- enhance their motoric performance. This worked well on the tion time and education level. A possible explanation for this TMT A and made them relatively fast. However, on the finding is that the education level was relatively high in the TMT B this left too few resources for executive processing, present study. It is therefore possible that the findings of this increasing completion time and IEV disproportionally. In study will not generalize to other parts of the population. 1086 Behav Res (2018) 50:1074–1087 Annett, J., Annett, M., Hudson, P. T. W., & Turner, A. (1979). The control Additionally, the sample included only healthy partici- of movement in the preferred and non-preferred hands. Quarterly pants. On the basis of the notion that use of the non- Journal of Experimental Psychology, 31, 641–652. dominant hand while performing a cognitively demanding Arbuthnott, K., & Frank, J. (2000). Trail Making Test, part B as a measure task interferes with its performance because it relies on of executive control: Validation using a set-switching paradigm. Journal of Clinical and Experimental Neuropsychology, 22, 518– shared cognitive resources, it seems likely that intermanual differences on the TMT will be even more pronounced in Baddeley, A., & Della Salla, S. (1996). Working memory and executive people with cognitive deficits. More research on the underly- control. Philosophical Transactions of the Royal Society B, 351, ing mechanisms for individuals whose cognitive functioning 1397–1403. Bauer, R. M., Iverson, G. L., Cernich, A. N., Binder, L. M., Ruff, R. M., is affected as the result of trauma or disease will therefore be & Naugle, R. I. (2012). Computerized neuropsychological assess- necessary. ment devices: Joint position paper of the American Academy of A third limitation of the study is that the sample consisted Clinical Neuropsychology and the National Academy of only of right-handed individuals. Since our element-by- Neuropsychology. Archives of Clinical Neuropsychology, 27, 362– element analysis suggests that the TMT may be biased for 373. Bowie, C. R., & Harvey, P. D. (2006). Administration and interpretation the left or the right hand, it will be important to replicate this of the Trail Making Test. Nature Protocols, 1, 2277–2281. study including also left-handed participants. Cernich, A. N., Brennana, D. M., Barker, L. M., & Bleiberg, J. (2007). Sources of error in computerized neuropsychological assessment. Archives of Clinical Neuropsychology, 22S, S39–S48. Corrigan, J. D., & Hinkeldey, N. S. (1987). Relationships between parts A Conclusion and B of the Trail Making Test. Journal of Clinical Psychology, 43, 402–409. The present study has shown that use of the non-dominant Cramond, H., Clark, M., & Smith, M. (1989). The effect of using the hand affects performance on the TMT. Performing part B of dominant or nondominant hand on the Rivermead Perceptual Assessment Battery. Clinical Rehabilitation, 3, 215–221. the TMT with the non-dominant hand increases completion Fromm-Auch, D., & Yeudall, L. T. (1983). Normative data for the time, since both using the non-dominant hand and the cogni- Halstead–Reitan neuropsychological tests. Journal of Clinical tive task itself draw on the same cognitive resources. Our Neuropsychology, 5, 221–238. study hints at important clinical consequences of using the Gaudino, E. A., Geisler, M. W., & Squires, N. K. (1995). Construct non-dominant hand. A B/A ratio score close to or higher than validity in the Trail Making Test: What makes part B harder? Journal of Clinical and Experimental Neuropsychology, 17, 529– 3 could be falsely attributed to cognitive dysfunction, whereas at least in some cases a high B/A ratio score may be due to Hausdorff, J. M., Yogev, G., Springer, S., Simon, E. S., & Giladi, N. performing the test with the non-dominant hand. (2005). Walking is more like catching than tapping: Gait in the This study demonstrates the importance of a more detailed elderly as a cognitive complex task. Experimental Brain Research, 164, 541–548. analysis of TMT performance that is possible when it is mea- Jäncke, L., Peters, M., Schlaug, G., Posse, S., Steinmetz, H., & Müller- sured digitally. A more detailed analysis of the different compo- Gärtner, H.-W. (1998). Differential magnetic resonance signal nents of the TMT can be used to better interpret specific out- change in human sensorimotor cortex to finger movements of dif- comes and may eventually be used to improve the reliability of ferent rate of the dominant and subdominant hand. Cognitive Brain Research, 6, 279–284. the TMT. The present study therefore adds to the growing body Kortte, K. B., Horner, M. D., & Windham, W. K. (2002). The Trail of research on the benefits of digital cognitive testing. Making Test, part B: Cognitive flexibility or ability to maintain set? Applied Neuropsychology, 9, 106–109. Acknowledgements The authors would like to thank Julie de Kok and Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Daisy van Minde in assisting in the data collection for the study. Neuropsychological assessment. Oxford: Oxford University Press. Lindenberger, U., Marsiske, M., & Baltes, P. B. (2000). Memorizing Open Access This article is distributed under the terms of the Creative while walking: Increase in dual-task costs from young adulthood Commons Attribution 4.0 International License (http:// to old age. Psychology and Aging, 15, 417–436. creativecommons.org/licenses/by/4.0/), which permits unrestricted use, LoSasso, G. L., Rapport, L. J., Axelrod, B. N., & Reeder, K. P. (1998). distribution, and reproduction in any medium, provided you give appro- Intermanual and alternate-form equivalence on the Trail Making priate credit to the original author(s) and the source, provide a link to the Test. Journal of Clinical and Experimental Neuropsychology, 20, Creative Commons license, and indicate if changes were made. 107–110. MacDonald, S. W. S., Li, S.-C., & Bäckman, L. (2009). Neural under- pinnings of within-person variability in cognitive functioning. Psychology and Aging, 24, 792–808. References MacDonald, S. W., Nyberg, L., & Bäckman, L. (2006). Intra-individual variability in behavior: Links to brain structure, neurotransmission and neuronal activity. Trends in Neurosciences, 29, 474–480. Amodio, P., Wening, H., Del Piccolo, F., Mapelli, D., Montagnese, S., Pellegrini, A., … Umiltà, C. (2002). Variability of trail making test, Mattay, V. S., Callicott, J. H., Bertolino, A., Santha, A. K. S., van Horn, J. symbol digit test and line trait test in normal people: A normative D., Tallent,K.A., … Weinberger, D. R. (1998). Hemispheric control study taking into account age-dependent decline and sociobiological of motor function: A whole brain echo planar fMRI study. Psychiatry Research: Neuroimaging Section, 83,7–22. variables. Aging Clinical and Experimental Research, 14,117–131. Behav Res (2018) 50:1074–1087 1087 Mrazik, M., Millis, S., & Drane, D. L. (2010). The oral Trail Making Test: Theill, N., Martin, M., Schumacher, V., Bridenbaugh, S. A., & Kressig, R. W. (2012). Simultaneously measuring gait and cognitive perfor- Effects of age and concurrent validity. Archives of Clinical Neuropsychology, 25, 236–243. mance in cognitively healthy and cognitively impaired older adults: Poreh, A. M., Miller, A., Dines, P., & Levin, J. (2012). Decomposition of The Basel motor–cognition dual-task paradigm. Journalofthe the Trail Making Test—Reliability and validity of a computer American Geriatric Society, 59, 1012–1018. assisted method for data collection. Archives of Assessment Tombaugh, T. N. (2004). Trail Making Test A and B: Normative data Psychology, 2, 57–72. stratified by age and education. Archives of Clinical Reitan, R. M., & Wolfson, D. (1995). Category test and trail making test Neuropsychology, 19, 203–214. as measures of frontal lobe functions. Clinical Neuropsychologist, 9, Toyokura, M., Ishida, A., Watanabe, F., Okada, N., & Yamazaki, M. 50–56. (2003). Intermanual difference in the Japanese Trail Making Test Ricker, J. H., Axelrod, B. N., & Houtler, B. D. (1996). Clinical validation and its mirror version: Intra-subject comparison of the task- of the oral Trail Making Test. Neuropsychiatry, Neuropsychology, completion time, cognitive time, and motor time. Disability and and Behavioral Neurology, 9, 50–53. Rehabilitation, 25, 1339–1343. Robins Wahlin, T.-B., Bäckman, L., Wahlin, A., & Winblad, B. (1996). Toyokura, M., Sawatari, M., Nishimura, Y., & Ishida, A. (2003). Trail Making Test performance in a community-based sample of Nondominant hand performance of the Japanese Trail Making Test healthy very old adults: Effects of age on completion time, but not and its mirror version. Archives of Physical Medicine and on accuracy. Archives of Gerontology and Geriatrics, 22, 87–102. Rehabilitation, 84, 691–693. Salthouse, T. A., & Fristoe, N. M. (1995). Process analysis of adult age Vlaskamp, B. N., & Hooge, I. T. C. (2006). Crowding degrades saccadic effects on a computer-administered Trail Making Test. search performance. Vision Research, 46, 417–425. Neuropsychology, 9, 518–528. Vlaskamp, B. N., Over, E. A., & Hooge, I. T. C. (2005). Saccadic search Schatz, P., & Browndyke, J. (2002). Applications of computer-based performance: The effect of element spacing. Experimental Brain neuropsychological assessment. The Journal of Head Trauma Research, 167, 246–259. Rehabilitation, 17, 395–410. Wagenmakers, E.-J., & Brown, S. (2007). On the linear relation between Schretlen, D. J., Munro, C. A., Anthony, J. C., & Pearlson, G. D. (2003). the mean and the standard deviation of a response time distribution. Examining the range of normal intraindividual variability in neuro- Psychological Review, 114, 830–841. doi:10.1037/0033-295X.114. psychological test performance. Journal of the International 3.830 Neuropsychological Society, 9, 864–870. Wagner, S., Helmrich, I., Dahmen, N., Lieb, K., & Tadic, A. (2011). Siu, K.-C., Chou, L.-S., Mayr, U., van Donkelaar, P., & Woollacott, M. H. Reliability of three alternate forms of the Trail Making Tests A (2008). Does inability to allocate attention contribute to balance and B. Archives of Clinical Neuropsychology, 26, 314–321. constraints during gait in older adults? Journals of Gerontology, West, R., Murphy, K. J., Armilio, M. L., Craik, F. I. M., & Stuss, D. T. 63A, 1364–1369. doi:10.1093/gerona/63.12.1364 (2002). Lapses of intention and performance variability reveal age- Soukup, V. M., Ingram, F., Grady, J. J., & Schiess, M. C. (1998). Trail related increases in fluctuations of executive control. Brain and Making Test: Issues in normative data selection. Applied Cognition, 49, 402–419. Neuropsychology, 5, 65–73. Woods, D. L., Wyma, J. M., Herron, T. J., & Yund, E. W. (2015). The Strauss, E., Bielak, A. A., Bunce, D., Hunter, M. A., & Hultsch, D. F. effects of aging, malingering, and traumatic brain injury on comput- (2007). Within-person variability in response speed as an indicator erized trail-making test performance. PLoS ONE, 10, e0124345. doi: of cognitive impairment in older adults. Aging, Neuropsychology, 10.1371/journal.pone.0124345 and Cognition, 14, 608–630. Yamashita, H. (2010). Right- and left-hand performance on the Rey– Strenge, H., & Niederberger, U. (2008). Unidirectional interference in use Osterrieth complex figure: A preliminary study in non-clinical sam- of nondominant hand during concurrent grooved pegboard and ran- ple of right handed people. Archives of Clinical Neuropsychology, dom number generation tasks. Perceptual and Motor Skills, 106, 25, 314–317. 763–774. Tanner-Eggen, C., Balzer, C., Perrig, W. J., & Gutbrod, K. (2015). The Yochim, B., Baldo, J., Nelson, A., & Delis, D. C. (2007). D-KEFS Trail neuropsychological assessment of cognitive deficits considering Making Test performance in patients with lateral prefrontal cortex measures of performance variability. Archives of Clinical lesions. Journal of the International Neuropsychological Society, Neuropsychology, 30, 217–227. 13, 704–709.

Journal

Behavior Research MethodsSpringer Journals

Published: Jul 13, 2017

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off