Acta Psychologica 148 (2014) 181–187

Contents lists available at ScienceDirect

Acta Psychologica

j ourna l homepage: www.e lsev ie r .com/ locate /actpsy

Influences of grasp selection in typically developing children

Kate Wilmut ⁎, Maia ByrneFaculty of Health Life Sciences, Oxford Brookes University, Oxford OX3 0BP, United Kingdom

⁎ Corresponding author at: Faculty of Health andUniversity, Oxford OX3 0BP, United Kingdom. Tel.: +44 1

E-mail address: k.wilmut@brookes.ac.uk (K. Wilmut).

http://dx.doi.org/10.1016/j.actpsy.2014.02.0050001-6918/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 December 2012Received in revised form 11 February 2014Accepted 12 February 2014Available online 4 March 2014

Psychinfo codes:23302820

Keywords:Grip selectionMotor planningDevelopmentEnd-state-comfort

When reaching towards an object, adults favour grasps which, following the intended action, end in a comfort-able position even when this requires them to start in an uncomfortable position (the end-state-comfort effect).However, this strategy is not consistently used by children who instead seem to favour a minimal pre-contactrotation of the hand, even when this results in an uncomfortable end position. In terms of multiple movements,the strategies used for grip selection are unclear; adults may still grasp for end-state-comfort given their propen-sity to plan to the end of amovement; however, childrenwho are less able to concatenatemovementmay tend tostart-state-comfort movements. The current study considered grip selection in children ranging from 4 to12 years and in a group of adults. Participants were asked to rotate a disc so that an arrow pointed towards aspecific target(s), the number of sequences in a movement was increased from one to three. Planning for end-state-comfort was seen in all participants and a clear developmental trajectory was identified whereby therelative comfort of an end position could be directly predicted by age in months. Adults and 10–12-year-oldsfavoured an end-state-comfort strategy whereas the younger children gave equal weighting to end-state-comfort, start-state-comfort and no initial rotation strategies. All groups were able to end a movementcomfortably when it was composed of three steps; however, the proportion of movements relying on an end-state-comfort strategy decreased as sequence length increase whereas the proportion of start-state-comfortand no initial rotation strategies increased. The current data support the concept that a mechanism for planninggrasps may be based on motor experience.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

The seemingly simple actions we perform in everyday life require agreat deal of planning and consideration regarding their outcome.Movements are constrained by many factors, for example, the positionof our body at the start of a movement and the biomechanicalconstraints that prevent us from carrying out a movement. When plan-ning a movement, prediction plays a very important role. In this senseprediction refers to our ability to estimate future states of the body interms of the resulting position in space given a specificmotor command(Wolpert, Diedrichsen, & Flanagan, 2011; Wolpert & Flanagan, 2001).Therefore, allowing us to choose a motor command which results inan optimal end position minimising fatigue and maximising comfort(Haggard, 1998; Rosenbaum, Vaughan, Barnes, & Jorgensen, 1992).One example of this kind of prediction in motor planning is the end-state-comfort effect (Rosenbaum et al., 1990) whereby an uncomfort-able start position is chosen in order that the end result is comfortable.The most commonly used example of this effect is that of an upturnedglass. In order to right the glass to its natural position and fill it up

Life Sciences, Oxford Brookes865 483 781.

with water from a jug, most adults will approach the glass with theirthumb pointing downwards and their elbow pointing up so that theend position is comfortable (Fischman, 1997). The end-state-comforteffect has been replicated in numerous studies focusing on adultpopulations (Fischman, 1997; Rosenbaum, Cohen, Meulenbroak, &Vaughan, 2006; Rosenbaum, Vaughan, Jorgensen, Barnes, & Stewart,1993; Rosenbaum et al., 1990).

Studies considering the developmental time course of the end-state-comfort effect have produced mixed results. Using the exampleof the upturned glass Adalbjornsson, Fischman, and Rudisill (2008)found that 20% of 2–3-year-olds and 35% of 5–6-year-olds demonstrat-ed an end-state-comfort effect. Manoel and Moreira (2005) used thedowel placing task developed by Rosenbaum et al. (1990) and demon-strated that approximately 50% of 2.5- to 6-year-olds select grips whichend in comfort on the majority of their trials. Using the same taskWeigelt and Schack (2010) showed that the number of children usinga grasp resulting in end-state-comfort increased as age increased,with 18% of 3-year-olds selecting grasps for comfort, 45% of 4-year-olds and 67% of 5-year-olds. Smyth and Mason (1997) consideredslightly older children and their results indicated that the ability toselect grasps for end-state-comfort develops rapidly but not fullybetween the ages of 4 and 8 years of age, with 4–5-year-olds selectinggrasps for end-state-comfort on approximately 58% of trials and 7–8-year-olds on approximately 67% of trials.

182 K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

These findings have been interpreted as evidence that planning forend-state-comfort is not apparent during childhood and ergo thatchildren lack planning skills. The inference here is that when childrenuse a grasp which results in end-state-comfort, planning is used andwhen a grasp fails to result in end-state-comfort, planning is not used.This raises an important issue, initially raised by van Swieten et al.(2010), of whether grip selection tasks are driven by executive planningor motor planning (or both). van Swieten et al. (2010) argued that theend-state-comfort effect cannot rely on executive planning as adultsdo not consistently select grasps which end in comfortable positionswhich would suggest, if executive planning is the driving force, thatexecutive function fails on some trials. Instead van Swieten et al.(2010) suggested that all grip selection relies on motor planning andthat each grasp is selected on the basis of the most efficient movement.Grip selection may be biased by end-state-comfort, if the cost of anuncomfortable end position outweighs an initial awkward movement.However, if that initial awkward movement, for whatever reason,comes at a cost over and above a comfortable end position, then agrasp will be planned for start-state-comfort. van Swieten et al.(2010) used a dowel rotation task that was set up in such a way thatparticipants could either maximally rotate the hand at the start of amovement and finish in comfort or start a movement with minimalrotation and end in an uncomfortable position. The adults showed aclear bias to grasp for end-state-comfort and to some extent this wasmirrored in the 9–14-year-olds. However, the 5–8-year-olds showed amuch stronger bias towards initialminimal rotation. It seems, therefore,that the young children choose to start the movement with a neutral,well practiced movement even though it results in an uncomfortableend position demonstrating a bias of minimal pre-contact rotation inyoung children.

The studies discussed so far have considered a single movement butin reality we rarely perform discreet actions, more oftenwe link togeth-er multiple actions, for example grasping a key, rotating this to put intoa lock and then turning the key to lock the door. Ideally, a movementsuch as this would not require readjustment of grasp. Rosenbaumet al. (1992) showed that grasp was influenced by the perception ofcomfort of a given end position of an impending movement. They con-cluded that the pointingmovement (the secondmovement) influencedthe planning of the first movement (grasp choice) and that grasp choicewas reliant on bothmovements. Haggard (1998) considered the degreeof change in grasp selection when asked to make one to five sequentialmovements. He found that adults used a different grasp for threesequence movements when the sequences differed by end target only,showing that the graspwas selected on the basis of the final target rath-er than the first or second. This effect dissipated when the movementsequence increased to four orfive targets (Haggard, 1998). This suggeststhat the final target in a sequence can bias grip selection, but only formovement sequences with three or less targets. To our knowledgesequential movements for grasp selection have not been considered inchildren; however, some research has considered sequential pointing.Data in this type of task suggest that children are less likely than adultsto concatenate movements into an over-arching action (Badan, Hauert,&Mounoud, 2000;Wilmut,Wann, & Brown, 2006) but rather plan eachmovement individually. This finding may carry across into grasp selec-tion, if it does, then we would expect to see grasps for multiplemovements being biased by factors relating to the first target in thesequence rather than the second or third.

The current study aims to look at different strategies for selectinggrasps for one, two and three movement sequences. van Swieten et al.(2010) only considered two strategies or biases, end-state-comfortand minimal rotation. In their paper pre-contact minimal rotation wasdefined as a rotation of the hand up to 135°. This is still quite a largepre-contact movement and could incorporate biases which are notsolely driven by minimal rotation. For this reason the current studywill classify a minimal rotation differently. The strategies to be consid-ered are: 1) end-state-comfort (ESC e.g. Rosenbaum et al., 1990);

2) no initial rotation, previous studies have shown that pre-contactminimal rotation can bias grasp selection (e.g. van Swieten et al.,2010); here we consider whether a no pre-contact initial rotation(0° rotation of the hand) biases grasp selection; 3) start-state-comfort,a grasp could be selected on the basis of a comfortable start position(e.g. Adalbjornsson et al., 2008); 4) previous movement, repeating anaction can be easier than generating a new action (e.g. de Lussanet,Smeets, & Brenner, 2002; Van Bergen, van Swieten, Williams, & Mon-Williams, 2007) and so the grasp used may simply be the same as thatpreviously used; and 5) another type of grasp which does not fit intoany of the above strategies. This was included as the set-up allowedgrasps which did not meet the criteria for any on the aforementionedstrategies. Based on previous studies we would expect that adultswould predominately adopt an end-state-comfort strategy (Rosenbaumet al., 1990) but the use of this would decrease as age decreases(van Swieten et al., 2010). Instead children are expected to show a pro-pensity towards one of the other strategies. Based on the van Swietenet al. (2010) study we may expect children to be biased by the no initialrotation strategy; however, this study allows us to consider additionalbiases and how these influence grasp selection. Bias towards thesestrategies will be considered developmentally and across the length ofthe movement sequence. Given that Haggard (1998) found that adultswere able to select a grasp based on the final element of a sequence wewould expect the propensity of selecting grasps for end-state-comfortto carry into both two and three movement sequences. Given thatchildren tend towards planning a single movement at a time we mightexpect that any bias towards selecting grasps for end-state-comfortseen in a one sequence movement would decrease as movementsequence increases. However, whether this is uniform across all agesandwhat bias replaces end-state-comfort for grasp selection is unknown.

2. Method

2.1. Participants

Twenty typical adults withmean age: 27 years, 3 months (10malesand 10 females) were recruited from Oxford Brookes University to takepart in this study. Sixty children from three age groups (all N=20with10 girls and 10 boys per age group) also took part in the study. Themeanage of the 10–12-year-olds was 11 years, 0 month, of the 7–9-year-oldswas 8 years, 6 months and of the 4–6-year-olds was 5 years, 5 months.All children were recruited from local schools and research group net-works at Oxford Brookes University. All movements were made usingthe preferred hand and all participants were right handed. To ensurethat noparticipants hadmovement difficulties individualmovement as-sessmentswere carried out. Adults were assessed on the age band threemanual dexterity component of the Movement Assessment Battery forChildren 2nd Edition (MABC-2; Henderson, Sugden, & Barnett, 2007).Although this assessment is not designed for an adult population agood performance, alongside self report of a lack of motor difficultiescan be taken as an indication of a typical motor ability. All adults scoredabove the 15th percentile. The children all completed the appropriateage band on the test component of the MABC-2 and all fell above the15th percentile. Participant information is given in Table 1.

2.2. Materials

Participants were seated in front of a wooden octagonmounted on aback board; the octagonwas surrounded by 8 different colours. The sizeof the octagon to be grasped was varied depending on the size of thehand. The octagon could be rotated about its centre so that an arrow,initially pointing directly upwards, could be turned to point to one ofthe 8 colours on the board (see Fig. 1 for an illustration of the set-up).A start node placed 0.3 times total arm length from the octagon wasprovided and participants were instructed to grasp this between their

Table 1Details of the five different age groups, including age, gender and movement assessment scores.

Children Adults

4–6 7–9 10–12

N 20 20 20 20Mean age 5 years 5 months 8 years 6 months 11 years 0 months 27 years 3 monthsGender ratio (M:F) 10:10 10:10 10:10 10:10MABC percentile score 64.4 60.6 56.8 N/AMD percentile score 51.5 61.8 54.6 57.2

183K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

thumb and index finger prior to the start of each trial. All movementswere video recorded.

2.3. Procedure

For each participant a baseline or neutral grasp was determined byasking the participant to grasp the octagon in a comfortable or neutralposition, ensuring that each finger lay on a different side. A colour or asequence of colours was named and participants were instructed tograsp hold of the octagon and rotate it so that the arrow pointed tothe colour(s) in the order they were listed. Each trial began with theparticipant grasping the start node between their thumb and indexfinger. They were explicitly told that they should think about the wayin which they grasped the octagon to ensure that they could completethe movement; however, the issue of comfort was not raised. Rotationof the octagon could be done either in a clockwise or anti-clockwisedirection or a combination of these directions for multiple coloursequences. Participants were instructed to ensure that their fingersremained in contact with the octagon for the entirety of the movementand that they could not alter their grasp during the movement.Sequences of one, two or three colours were used and all participantscompleted the shortest movements first moving up through thesequence lengths. The multiple colour sequences all included a move-ment to green which necessitated a 180° rotation. One sequencemovement also included rotations to green but also to colours requiringa smaller rotation (see Appendix A for the colour sequences). Therewasnomovement sequence that was not achieved at least once without thechild moving his or her fingers/hand. The presentation of sequenceorder in a blocked fashion rather than random allows both the childrenand adults to progress from the simplest task (one movement) to themost difficult (three movements). The disadvantage of this is that anybetween-group differences may be affected by learning. However, theadvantage of this is it optimises the performance of the youngest chil-dren. There is no ideal trial ordering for testing children but the authorsbelieve that a fixed presentation order is most appropriate for youngchildren.

Fig. 1.A. Schematic illustration of the octagon and surrounding colours, including codingnumbewas the largest size used. The diameters of the octagons were 13 cm, 12.5 cm, 10.5 cm, 8 cm a

2.4. Judgements of end-state-comfort

In order to determine whether an end position was comfortableor not, the level of comfort for each possible end position was judgedby an independent cohort of adults (15 right handers and 15 lefthanders aged between 18 and 30 years of age). These ‘raters of comfort’were asked to place their thumb on each side of the octagon followingboth a clockwise and an anti-clockwise rotation and rate each of thosehand positions on a 1–6 scale: very comfortable (6); slightly com-fortable (5); neither comfortable nor uncomfortable (4); slightlyuncomfortable (3); very uncomfortable (2); or impossible (1). Averagejudgements of each positionwere calculated separately for right and lefthanders and a positionwas taken as comfortable if the average score fellat or below 4 (slightly comfortable to very comfortable) and thestandard deviation was below 1. This produced four positions forright handers and four positions for left handers that were deemedcomfortable/uncomfortable. A similar method was used by Rosenbaumet al. (1992).

2.5. Data analysis

Grasp positions were determined by watching the video recordings.For each grasp the positioning of the thumb was recorded (sides of theoctagon were numbered 1–8 and so each thumb position was definedby one of these values, see Fig. 1). In order to ensure reliability of codinghand position in this way 25% of participants were coded by a secondcoder. Inter-rater reliability analyses were carried out on the codedposition of the thumb by the two raters. Cohen's kappa showedexcellent agreement between the two coders for adults [kappa = 0.99p b .001], 10–12-year-olds [kappa = 0.96 p b .001], 7–9-year-olds[kappa = 0.95 p b .001] and 4–6-year-olds [kappa = 0.96 p b .001].

For eachmovement a comfort rating based on the endposition of thethumb was derived from the independent ratings of comfort. Thisresulted in an end comfort score for each participant across one, twoand three movement sequences. This value fell between 1 and 6 and ahigh value indicated a propensity to use grips ending in a comfortable

rs thatwereused to code each grasp. B. Photographof the actual set-up; the octagon shownnd 6.5 cm.

184 K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

position. This value will be referred to as end comfort rating. In additionto this measure the strategy or bias used to select grasps for eachmovement was considered. Grasps were classified as following one offive different strategies: end-state-comfort (ESC) strategy, a movementwas classified as ending in end-state-comfort if the end comfort ratingwas above 4; no initial rotation strategy, the initial pre-contact rotationof the hand from the start node to the octagon could be determinedbased on the placement of the thumb,1 and movements whereparticipants did not end in a comfortable position and did not rotatetheir wrist from start node to the octagon were classified as no initialrotation; start-state-comfort grasp strategy, all participants were askedto grasp the octagon in a way they considered neutral, and this wastaken as their baseline grasp,2 and movements where participantsdid not end in a comfortable position and used a baseline grasp wereclassified as start-state-comfort strategy (SSC); previous movementstrategy, movements that did not end in comfort but that showed thesame grasp as that used for the previous movement were classified asprevious movement strategy; unnamed strategy, movements that didnot end in comfort, that showed some initial rotation and did not usea baseline grasp or the same grasp used on the previous trial wereclassified as unnamed strategy. Each trial was classified as biased byonly one of the strategies described above. For each participant thepercentage of trials where each strategy was used was calculated forone, two and three movement sequences.

3. Results

3.1. End comfort rating

The end comfort rating was considered using a 2 × 3 factorialANOVA (age group × sequence length); these data can be found inFig. 2. The main effect of sequence length and age group was found.The main effect of sequence length, F(2,152) = 86.05 p b .001 partialη2 = .53 was due to more comfortable end positions for one sequencemovements compared to two sequencemovements andmore comfort-able end positions for two sequence movements compared to threesequence movements (one sequence b two sequence b three sequence,all p b 0.05 following Bonferroni correction). The main effect of agegroup, F(3,76)= 19.46p b .001 partial η2= .43was due to a significant-ly higher average comfort rating (i.e. more comfortable) for the adultsand 10–12-year-olds compared to the 7–9 year-olds, who in turn hada higher end comfort rating compared the 4–6 years (adults = 10–12 years b 7–9 years b 4–6 years, all p b 0.05 following Bonferroni cor-rection). The interaction between age group and sequence length wasalso significant, F(6,152) = 2.24 p = 0.042 partial η2 = .08. This inter-action was explored using simple main effects with Bonferroni correc-tion. The main effect of sequence length was found for each age group[adults: F(2,75) = 16.80 p b .001 partial η2 = .31, 10–12 years:F(2,75) = 16.66 p b .001 partial η2 = .31, 7–9 years: F(2,75) = 15.76p b .001 partial η2 = .30, 4–6 years: F(2,75) = 27.60 p b .001 partialη2 = .42]. For the adults this was due to a higher comfort (more com-fortable) rating for one movement sequences compared to two move-ment sequences and for two movement sequences compared to threemovement sequences (one movement b two movements b threemovements). For the child groups, this was due to a higher comfort rat-ing (more comfortable) for the one movement sequence compared tothe two or three movement sequence (one movement b two move-ments= threemovements). The end comfort ratings were also consid-ered for each age group across each sequence length for grasps ending

1 Given the orientation of the thumb at the start node, placing the thumb on side ‘6’ or‘5’ for a left hander, required approximately no initial rotation.

2 For all participants this grasp involved some initial rotation of the wrist from startnode to octagon and was classified as a comfortable grasp based on the independentratings of comfort; thus using this strategy would suggest start-state-comfort.

in comfort only. There were no differences of age group or sequencelength.

In order to more fully explore the developmental trajectory ofend comfort a regression analysis was carried out for each sequencelength considering whether age in months predicted end comfort.The adult group was excluded from this analysis as it was felt thata significant result when including this cohort may solely be drivenby the adults rather than a true reflection of the developmentaltime course in childhood. Age predicted the comfort of the end posi-tion for all three sequence lengths: onemovement sequence, F(1, 59)= 11.53 p =.001, R2 = .17, β = 24.9; two movement sequenceF(1,59) = 31.49 p b .001, R2 = .35, β= 34.62; and three movementsequence F(1, 59) = 16.75 p b .001, R2 = .22, β = 28.02. As age inmonths increases end positions become more comfortable.

3.2. Strategies of grasp selection

The percentage values of trials showing each strategy (end-state-comfort, no initial rotation, SSC grasp, previous movementand unnamed) are depicted in Fig. 3A and were analysed using a 4× 3 × 4 ANOVA (age group × sequence length × strategy type).The main effect of strategy was found [F(4,304) = 139.20 p b .001partial η2 =.65]; post hoc tests indicated that this was due to theESC strategy being used more often than the other strategies, SSCgrasp being used more than no initial rotation or unnamed and pre-vious movement strategy being used least (ESC N SSC grasp N no initialrotation =unnamed N previous movement, p b 0.05 using Bonferronicorrection). In addition, interactions between strategy and group[F(12,304) = 4.80 p b .001 partial η2 = .16] and sequence and strategy[F(8,608)= 29.49 p b .001 partial η2 = .28] were found. Given that ESCand SSC were the two most common strategies Fig. 3B shows the per-centage of trials where an ESC strategy was chosen over a SSC strategy.

In order to explore the significant two-way interactions simplemaineffects with Bonferroni correction were used. For the group by strategyinteraction an effect of strategy was found for each age group [adults:F(4,73) = 107.47 p b .001 partial η2 = .86, 10–12 years: F(4,73) =70.58 p b .001 partial η2= .80, 7–9 years: F(4,73)= 73.21 p b .001 par-tial η2= .80, 4–6 years: F(4,73)= 54.05p b .001partial η2= .75; Pillai'strace reported]. Adults and 10–12-year-olds used the ESC strategymoreoften than the other four strategies (ESC N no initial rotation = SSCgrasp = previous movement = unnamed); 7–9-year-olds used anESC strategy and SSC strategy equally as often, but these were usedmore than the other three strategies (ESC = SSC grasp N no initialrotation = previous movement = unnamed); and the 4–6-year-oldsused the ESC strategy, the SSC grasp strategy and the no rotationstrategy equally as often and these were used more than the previousmovement or unnamed strategy (ESC = no initial rotation =SSCN previousmovement= unnamed). For the interactionbetween sequenceand strategy an effect of sequence was found for all strategies apart fromSSC grasp and previous movement [ESC: F(2,75) = 61.20 p b .001partial η2 = .62, no initial rotation: F(2,75) = 25.21 p b .001 partialη2 = .42, unnamed: F(2,75) = 7.63 p = .001 partial η2 = .17; Pillai'strace reported]. For the ESC strategy the effect was due to a reductionin the use of the strategy from one movement compared to two move-ments and from two movements compared to three movements (onemovement N two movements N three movements). For the no initialrotation, and unnamed strategies both showed an increase in usagefrom one movement compared to the two or three movementsequence (one movement N two movements = three movements).

4. Discussion

End-state-comfort has always been treated as a categorical variable,a movement either ends comfortably or it does not. The measurementof a variable in this way lacks precision and could plausibly occludeage group effects, whereby children progressively end movements

1.0

1.5

2.0

2.5

3.0

3.5

1 2 3

End

com

fort

Sequence length

End Comfort Rating

10-11yrs 7-9yrs

4-6yrs AdultsVery comfortable

Slightly comfortable

Neither comfortablenor uncomfortable

Fig. 2. An illustration of the end comfort rating for the four age groups. Error bars represent standard error. The classification of comfort is illustrated on the graph.

185K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

in more comfortable positions, but these would not be defined as ‘com-fortable’. The current study quantified comfort of the end position byusing independent judgements of each hand position. These datademonstrate that 10–12-year-olds end movements with the same de-gree of comfort as adults, but that younger children, 7–9-year-oldsand 4–6-year-olds show movements ending in less comfortablepositions. This supports previous studies which have suggested thatthe ability to select grasps for end-state-comfort develops rapidly

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

One movement Two movements

A) Percentage of moves using ea

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

One movement Two movements

B) End-State-Comfort vs. Start

Fig. 3. A. A graph showing the percentage of moves for which the different strategy types werebaseline grasp (start-state-comfort) by light grey bars, previousmovement by barswith diagonand start-state-comfort. End-state-comfort is depicted by black bars and start-state-comfort (b

between 4–5 years and 6–8 years, but that it still had not reached anadult level by 8 years of age (Smyth & Mason, 1997). More specifically,however, the current data suggest that the ability to select grasps forend-state-comfort is developed by 10–12 years of age and that thereis a clear developmental trajectory, with age in months directlypredicting the end comfort rating. An important point to note here isthat when considering the end comfort of comfortable movementsonly, there was no effect of age. Therefore, the developmental trend

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

Three movements

ch strategy type

End-State-Comfort

No Initial Rotation

Start-state-comfort

Previous Mvmt

Unnamed

Adu

lts

10-1

2yrs

7-9y

rs

4-6y

rs

Three movements

-State-Comfort

Start-state-comfort

End-State-Comfort

adopted. End-state-comfort is depicted by black bars, no initial rotation by dark grey bars,al lines and unnamed by bars with black dots. B. A graph comparing just end-state-comfortaseline grasp) by light grey bars.

186 K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

described above is driven by a reduction in the number of uncomfortablemovements made as age increases rather than a distinct difference be-tween the comfortable movements.

When young children fail to select a grasp for end-state-comfort thatis not to say they are not planning a movement, but rather that they arenot planning for end comfort. van Swieten et al. (2010) demonstratedthat younger children favoured a minimal rotation movement overa movement ending in end-state-comfort. The current study alsoconsidered strategies used for selecting grasps; five possible strategiesmotivated by previous literature were considered: end-state-comfortstrategy (Rosenbaum et al., 1990), no initial rotation strategy (vanSwieten et al., 2010), start-state-comfort grasp strategy (Adalbjornssonet al., 2008), previous movement strategy (de Lussanet et al., 2002;Van Bergen et al., 2007) and unnamed strategy. The adults and 10–12-year-olds used the ESC strategy more often than any other strategy andno consistent difference between the other strategies was seen.The 7–9-year-olds used the ESC strategy equally as often as the start-state-comfort strategy; other strategies were used albeit less often andno one more consistently than any other. Finally, the 4–6-year-oldsused the ESC strategy, the start-state-comfort strategy and the no initialrotation strategy equally as often andmore than the other strategies. Thispattern of results demonstrates a clear developmental shift towardsending movements in comfort over and above other strategies, but alsodemonstrates a clear usage of start-state-comfort strategies (no initialrotation and start-state-comfort) in younger children. van Swietenet al. (2010) found that younger children tend towards using a minimalrotation strategy (an initial rotation of up to 135°); in the current paperwe have partialled this strategy down further. Both the no initial rotationstrategy and the start-state-comfort strategy would have been classifiedasminimal rotation strategies in the van Swieten et al. study, but herewehave shown a distinct developmental difference between the two, with7–9-year-olds moving away from no initial rotation and towardsselecting grasps biased by the most natural way to grasp the octagon(SSC grasp).

How children and adults select grips for a singlemovement is impor-tant in understanding movement planning and motor learning;however, this study took this one-step further and also considered theselection of grips for multiple movements. When considering sequencelength one consistent result was found; movements tended towardsending uncomfortably, the use of the end-state-comfort strategy forselecting grasps decreased and the use of alternative strategiesincreased as sequence length increased. When considering the ratingof comfort, the adults and 10–12-year-olds show a clear increase acrosseach sequence length, with three sequence movements ending withless comfort than the two sequence movements and the two sequencemovements ending with less comfort than the three sequence move-ments. To some extent this compliments the findings of Haggard(1998) who demonstrated that adults plan for the final element of amovement in movement sequences up to and including three move-ments. This is demonstrated here by the propensity of adults to planfor end-state-comfort (i.e. plan according to the end target) in all ofthe movements they made even though this decreased as sequencelength increased. In addition, the fact that the children were able toplan for end-state-comfort in sequential movements (again eventhough this decreased as sequence length increased) which suggeststhat they are able to concatenate sequential movements into oneover-arching action. This is seemingly opposed to previous studieslooking at the concatenation of pointing kinematics (Badan et al.,2000; Wilmut et al., 2006). However, these are very different tasks,one considering placement of the hand during a grasp, the other precisecontrol of the hand duringmovement and so the ability to link togetheror concatenate action in one and not the other is not untenable.

In terms of strategy use over the different sequence lengths, the useof the end-state-comfort strategy decreased as sequence lengthincreased and the use of the no initial rotation and unnamed strategy in-creased as sequence length increased. Although the start-state-comfort

strategy and previous movement strategy were used this was invariantacross sequence length. This may mean that it is the no initial rotationstrategy and the unnamed strategy which ‘replace’ the end-state-comfort strategy when planning for end-state-comfort is no longer theoptimal strategy. The change in strategy use as sequence length in-creases suggests that there is a shift in priority as an action becomes lon-ger or more complicated. Ending a movement in comfort becomes lessimportant as that movement gets longer. van Swieten et al. (2010) sug-gested that the mechanism behind the selection of grasps, and indeedmotor planning, reflects a previous motor history, with successful ac-tions being used again and less successful actions rejected. They usethis theory to explain their developmental shift from aminimal rotationbias to an end-state-comfort bias, with the younger children using aminimal rotation strategy as it includes movements that they have pre-viously executed and are familiar with. In contrast, older children aremore aware of the costs of theminimal rotation strategy (i.e. not endinga movement in comfort) and so favour an end-state-comfort strategy.The developmental changes and change across sequence length in thecurrent study can be explained in a similar way. All of the age groupsare less familiar with the two and three sequence movements and soall participants tend towards a less efficient movement (a greater pro-portion of movements ending uncomfortably) and adopt a start-state-comfort strategy as familiarity of this movement outweighs the costsof ending the movement in an uncomfortable position. Furthermore,this mechanismmay explain the type of inefficient strategies used acrossthe age groups (by inefficient we are referring tomovements not endingin end-state-comfort). The 4–6-year-olds use a no initial rotationstrategy and a start-state-comfort strategy, suggesting that thesemovements are equally as familiar and have equal costs, i.e. they bothend in an uncomfortable position. In contrast, the 7–9-year-olds tendto use the start-state-comfort strategy as their inefficient strategy.This shift may be due to emerging costs with the no initial rotationstrategy compared to the start-state-comfort strategy which comewith experience; the start-state-comfort grasp strategy results in amore comfortable start-state-position compared to the no initialrotation strategy. In the oldest children we see that the inefficientmovement strategies are all used equally as often, reflecting an under-standing that all of these have the same cost, i.e. not ending in comfort,thus making all of the less efficient strategies equally as efficient(or inefficient).

Within a wider context this paper has considered what influencesbehaviour when the behaviour is the grasping of an object. It hasbeen demonstrated that end-state-comfort has a lessened effect onbehaviour as movement sequences increase and that when the biasfor end-state-comfort is strong for a single movement (as in adults)the bias is greater in the two and three sequence movement. Therefore,factors that influence our behaviour in a simple context may be reducedin more complex behaviours but are still present and important. Inchildren, this means that a bias for end-state-comfort which is low fora single movement becomes lower for the two and three sequencemovements. This may mean that in everyday life children are graspingobjects in such a way that they end themovement in an uncomfortableposition and possibly that they are unable to continue with the move-ment without first re-adjusting their grasp. Many other factors not con-sidered in the current study could influence grasping behaviour andfuture studies could directly compare these. In addition, by using differ-ent sequences studies could consider the point atwhich a child is unableto complete a movement given his or her starting point.

This study has provided more evidence to suggest that selectinggrips for end-state-comfort develops rapidly during childhood and isseemingly at an adult level of control by 10–12 years of age. The strate-gies used to select grasps change during this period, with childrenmoving away from easier initial movements, to start-state-comfortmovements to end-state-comfort movements as they grow older. Themove away from easier initial movements/start-state-comfort may bedue to a greater understanding of the cost involved in ending a

187K. Wilmut, M. Byrne / Acta Psychologica 148 (2014) 181–187

movement uncomfortably as children get older. Furthermore, a reduc-tion in end comfort and use of the end-state-comfort strategy wasseen for all groups as sequence length increased; however, all groupswere able to end movements comfortably for three-sequencemovements suggesting that all the children could concatenate move-ment even though they may have chosen to plan for the start of thesequence rather than the end. Whether the developmental trajectoryseen is in line with the development of aspects of working memory,given the role of working memory in motor planning (Rosenbaum,Chapman,Weigelt, Weiss, & van derWel, 2012), is yet to be discovered.

Acknowledgements

This study was funded by an ESRC grant awarded to Kate Wilmut(RES-061-25-0472). We would like to thank all of the participantswho took part in this study and Ian Wilmut who made the equipment.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.actpsy.2014.02.005.

References

Adalbjornsson, C. F., Fischman, M. G., & Rudisill, M. E. (2008). The end-state-comfort effectin young children. Research Quarterly for Exercise and Sport, 79(1), 36–41.

Badan, M., Hauert, C. -A., & Mounoud, P. (2000). Sequential pointing in children andadults. Journal of Experimental Child Psychology, 75, 43–69.

de Lussanet, M. H. E., Smeets, J. B. J., & Brenner, E. (2002). The relation between taskhistory and movement strategy. Behavioural Brain Research, 129, 51–59.

Fischman, M. G. (1997). End-state-comfort in object manipulation [abstract]. ResearchQuarterly for Exercise and Sport, 68, A-60 (Suppl.).

Haggard, P. (1998). Planning of action sequences. Acta Psychologica, 99, 201–215.

Henderson, S., Sugden, D., & Barnett, A. (2007). Movement Assessment Battery for Children(2nd ed.). Oxford: Pearson.

Manoel, E. J., & Moreira, C. R. P. (2005). Planning manipulative hand movements: Doyoung children show the end-state-comfort effect? Journal of Human MovementStudies, 49, 93–114.

Rosenbaum, D. A., Chapman, K. M., Weigelt, M., Weiss, D. J., & van der Wel, R. (2012).Cognition, action and object manipulation. Psychological Bulletin, 138(5), 924–946.

Rosenbaum, D. A., Cohen, R. G., Meulenbroak, R. G., & Vaughan, J. (2006). Plans forgrasping objects. In M. Latash, & F. Lestienne (Eds.), Motor control and learning overthe lifespan (pp. 9–25). New York: Springer.

Rosenbaum, D. A., Marchak, F., Barnes, H. J., Vaughan, J., Slotta, J.D., & Jorgensen, M. J.(1990). Constraints for action selection: Overhand versus underhand grip. In M.Jeannerod (Ed.), Attention and performance XIII (pp. 321–342). Hillsdale, NJ: LawrenceErlbaum Associates.

Rosenbaum, D. A., Vaughan, J., Barnes, H. J., & Jorgensen, M. J. (1992). Time course ofmovement planning: Selection of hand grips for object manipulation. Journal ofExperimental Psychology: Learning, Memory, and Cognition, 18, 1058–1073.

Rosenbaum, D. A., Vaughan, J., Jorgensen, M. J., Barnes, H. J., & Stewart, E. (1993). Plans forobject manipulation. In D. E. Meyer, & S. Kornblum (Eds.), Attention and performanceXIV—A silver jubilee: Synergies in experimental psychology, artificial intelligence andcognitive neuroscience (pp. 803–820). Cambridge, MA: MIT Press, Bradford Books.

Smyth, M. M., & Mason, U. C. (1997). Planning and execution of action in children withand without developmental coordination disorder. Journal of Child Psychology andPsychiatry, 38, 1023–1037.

Van Bergen, E., van Swieten, L. M., Williams, J. H. G., & Mon-Williams, M. (2007). Theeffect of orientation on prehension movement time. Experimental Brain Research,178, 180–193.

van Swieten, L. M., van Bergen, E., Williams, J. H. G., Wilson, A.D., Plumb, M. S., Kent, S. W.,et al. (2010). A test of motor (not executive) planning in developmental coordinationdisorder and autism. Journal of Experimental Psychology: Human Perception andPerformance, 36(2), 493–499.

Weigelt, M., & Schack, T. (2010). The development of end-state comfort planning inpre-school children. Experimental Psychology, 57(6), 476–482.

Wilmut, K., Wann, J. P., & Brown, J. H. (2006). Problems in the coupling of eye and hand inthe sequential movements of children with developmental coordination disorder.Child: Care, Health and Development, 32(6), 665–678.

Wolpert, D.M., Diedrichsen, J., & Flanagan, J. R. (2011). Principles of sensorimotor learning.Nature Reviews Neuroscience, 12, 739–751.

Wolpert, D.M., & Flanagan, J. R. (2001). Motor prediction. Current Biology, 11(18),R729–R732.