The Role of Memory in the Tower of London Task
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ME MORY, 1999, 7 (2), 209±231 The Role of Memory in the Tower of London Task L.H. Phillips, V. W ynn, K.J. Gilhooly, S. Della Sala, & R.H. Logie University of Aberdeen, UK The Tower of London (TOL) task is widely used as a neuropsychological test of planning. Relatively little is known of the cognitive com ponents of the task, and in particular the role of memory in performance. The current studies on normal adults looked at the role of verbal and spatial working memory in the TOL. The effects of verbal and visuospatial dual-task manipulations on TOL performance were examined in an experiment with 36 participants. Both verbal and visuospatial executive secondary tasks caused poorer performance on the TOL; however, concurrent articulatory suppression enhanced performance. The results suggest that executive and spatial com ponents are important in the task, and raise questions about the role of preplanning in the TOL. INTRODUCTION There is currently much debate about how to describe and assess executive control processes of cognition (e.g. Rabbitt, 1997 ; Reitan & W olfson, 1994). This interest has arisen largely from the study of patients with lesions to the frontal lobes of the brain, who exhibit failures of executive control, such as failures of planning . Several ``frontal lobe tests’ ’ have been developed, which have face validity as measures of executive functions, and are proposed to be selectively sensitive to frontal lobe lesions. How ever, it is dangerous to assume that poor performance on any such ``frontal test’ ’ indicates poor executive function, because there are always multiple possible reasons for low scores (Phillips, 1997), and the construct validity of such tests is unkno wn (Kafer & Hunter, 1997). Moreover, these frontal tests lack specificity, i.e. patients with lesions in other areas of the brain also show poor performance (Anderson, Bigler, & Blatter, 1995 ; Grafman, Jonas, & Salazar, 1990). Frontal lobe tests are increasingly used to measure executive function in clinical and non-clinical populations so it is important that the cognitive processes involved in the tests are more accurately specified. Requests for reprints should be sent to Louise Phillips, Psychology Department, University of Aberdeen, Aberdeen AB24 2UB, Scotland, UK. Email: louise.phillips@abdn.ac.uk These studies were funded by Economic and Social Research Council (UK), grant number R000236119. Thanks to Jim Urquhart for writing the programs described in this paper and Geoff Ward for advice in developing the Tower of London task. Ó 1999 Psychology Press Ltd
210 PHILLIPS ET AL. One such test which has been widely used to assess executive function is the Tower of London (TOL) task, a variant on the Tow er of Hanoi task (Shallice, 1982 , 1988), see Fig. 1. In the TOL, coloured disks must be m oved one-by-one from an initial state to m atch a goal state. Instructions are given to plan the whole sequence of m oves that m ust be carried out m entally, before executing the sequence. The task has been studied in a variety of patient groups and also in norm al populations (e.g. Allamanno et al., 1987 ; Joyce & Robbins, 1991 ; Kafer & Hunter, 1997 ; Morris, Ahm ed, Syed, & Toone, 1993; Owen et al., 1990 ; Owen, Doyon, Petrides, & Evans, 1996 ; W ard & Allport, 1997 ; W atts, MacLeod, & M orris, 1988). The TOL is used as a m easure of planning ability: i.e. poor performance is interpreted as inability to plan efficiently (see e.g. Morice & Delahunty , 1996 ; Owen, 1997 ; Shallice, 1982 ; Temple, Carney, & Mullarkey, 1996). Frontal lobe patients perform poorly on the TOL; for example, Owen et al. (1990) found that m ove times were slower in frontal lobe patients, and argued that this was due to inefficient preplanning necessitating extra on-line planning during the execution phase. Accuracy of plans is also assessed in term s of the num ber of moves m ade in excess of the m inimum possible to solve the trials, and frontal patients make a higher number of excess moves than other patient groups (Owen et al., 1990; Shallice, 1982). However, FIG. 1. A n exam p le o f a T ower o f L on d on tr ial.
MEMORY AND THE TOWER OF LONDON TASK 211 the term ``planning ’ ’ is used loosely in neuropsychological literature (Goel & Grafman, 1995 ; McCarthy & W arrington, 1990), and little is known of the com ponent cognitive processes involved in m aking and executing plans (Kafer & Hunter, 1997; Owen et al., 1990 ; W ard & Allport, 1997). Kafer and Hunter (1997), using factor analytic techniques, argue that the TOL has poor validity as a measure of planning, and em phasise the need to gain further inform ation about the cognitive processes involved in this test. The current version of the TOL task involved com puterised presentation and responses made using a lightpen. In order to investigate TOL perform ance in a non-clinical popul ation, a wider range of difficulty was introduced than on the standard neuropsychological version, with five disks, as used by W ard and Allport (1997), rather than three disks. The TOL trials differed in the m inimum num ber of m oves to solution, and num ber of indirect or counterintuitive moves (moves that do not put a disk in its final goal position, but are essential for solution; W ard & Allport, 1997). Trials with indirect moves engender conflict between the overall goal (m ove disks to their final positions) and the subgoal (move disk away from its final position). Such goal±subgoal conflicts are known to cause particular difficulties for frontal lobe patients (Goel & Grafm an, 1995 ; M orris, M iotto, Feigenbaum , Bullock, & Polkey, 1997), and to increase time spent planning on the TOL (W ard & Allport, 1997). Working Memory in the TOL Task The role of memory in planning is often highlighted (e.g. Cohen, 1996; Hayes- Roth & Hayes-Roth, 1979; Owen et al., 1990). Cohen (1996) argues that working m em ory is important in form ulating, retaining, and im plementing plans, as well as revising them on-line. In the TOL, it seems plausible that the setting up, maintenance, and execution of a m ultistage plan will make considerable dem ands on working memory resources. A contrasting argument is proposed by W ard and Allport (1997). They describe a study in which the memory load of the TOL task was decreased by allowing on-screen m ovem ent of the disks during planning. Decreasing the m em ory load in this way did not affect time spent planning, which W ard and Allport argue is evidence that working memory resources do not lim it perform ance on the task. How ever, allowing people to plan by m oving disks is boun d to include som e increase in tim e to make a plan due to the motor responses required, so perhaps this is balanced out by a correspondin g decrease in the actual time spent thinking about a plan. Also, W ard and Allport do not report the effect of their working m em ory m anipulation on the number of excess moves made, the variable that is taken to indicate efficiency of planning in neuropsychological studies. The presentation and response requirements of the TOL are visual and spatial, but it is not clear whether the planning requirem ents load verbal or visuospatial memory resources. Do people carry out the task by silently
212 PHILLIPS ET AL. verbalising a plan of action (M orris et al., 1993), or by visualising a sequence of movem ents (Joyce & Robbin s, 1991 ; Owen et al., 1990 ; W elsh, Cicerello, Cuneo, & Brennan, 1995)? There is evidence from experimental m anipulations and neuropsychological patients that verbal and visuospatial memory are separable aspects of the memory system (for a review see Della Sala & Logie, 1993). M any authors have highlighted the role of visuospatial memory in the TOL; and especially visuospatial W M (Joyce & Robbin s, 1991 ; Morice & Delahunty, 1996 ; Owen et al., 1996). Temple et al. (1996) suggest that it is difficult to verbalise the processes involved in the TOL task, implying that visuospatial rehearsal m ust take place instead. W elsh et al. (1995) exam ined Tower of Hanoi performance in a non-clinical popul ation. Interview data were taken after each trial, and participants reported that they ``m entally visualised the m ovement of rings on the pegs’ ’ , which W elsh et al. interpret to indicate the use of visuo- spatial rehearsal in the task. A contrasting argument is put forward by M orris et al. (1993), who m aintain that active verbal rehearsal is involved in the TOL because patient and brain activation studies suggest left rather than right hem isphere involvem ent. Patient deficits tend to be found only after left frontal lobe lesions (Glosser & Goodglass, 1990 ; Shallice, 1982 ; although Owen et al., 1990 found TOL deficits after right frontal lesions too). Also, in brain activation studies of normal individuals during TOL perform ance, left frontal lobe activation was significant while right frontal activation was not (M orris et al., 1993 ; Owen et al., 1996). Although it is an oversim plification to associate left hemisphere activity exclusively with language, Morris et al. argue that the lateralisation during TOL performance is great enoug h to suggest verbal planning and rehearsal. Relationships have been reported between measures of verbal W M and the TOL in schizophrenics (M orice & Delahu nty, 1996). Dual Task Effects on TOL Performance This experiment exam ines the role of different com ponents of working memory in TOL perform ance. The experimental paradigm draws on the influential three- com ponent model of working memory (W M ) of Baddeley and Hitch (1974). The model com prises a ``central executive’ ’ responsible for such cognitive control functions as planning (Baddeley, 1986), a verbal buffer, and a visuospatial buffer. The focus of research based on this model has been on experimental manipulations, and in particular the effects of a range of dual tasks on performance on memory and reasoning tasks (Baddeley, 1996 ; Gilhooly, Logie, W etherick, & W ynn, 1993; Logie, Gilhooly, & W ynn, 1994). The model has proved useful in exploring the involvem ent of different com ponents of cognition in tasks of reasoning (Logie et al., 1994), and has support from both neuropsychological and cognitive evidence (e.g. Della Sala & Logie, 1993 ;
MEMORY AND THE TOWER OF LONDON TASK 213 Smith & Jonides, 1997). By using a range of dual tasks, the involvem ent of the different com ponents of the W M m odel can be addressed. Two important distinctions are m ade: verbal vs visuospatial processes, and executive vs non- executive processes. In the current study, four different types of secondary task (described in more detail later) are exam ined in com bination with the TOL task: articulatory suppression to load the phonol ogical loop, random num ber generation to load the phonological loop and central executive, pattern tapping to load the visuospatial scratchpad, and random tapping to load the scratchpad and central executive. By com paring the effects of these different secondary tasks on TOL performance, conclusions can be draw n about the role of the verbal buffer, visuospatial buffer, and executive com ponents of memory in the TOL task. 1. Articulatory suppression: Articulatory suppression, repeatedly saying a phrase or sequence, loads the phonol ogical loop com ponent of working memory by preventing silent verbal rehearsal (Baddeley, 1986). It has been widely used to exam ine the role of verbal rehearsal in cognitive tasks, and is known to interfere with verbal short-term m em ory (Baddeley, Lewis, & Vallar, 1984). Articulatory suppression does not interfere with all tasks (e.g. accuracy of conditional reasoning, Evans & Brooks, 1981) supporting the argument that interference effects are caused by the presence of verbal rehearsal rather than general resource dem ands. In relation to the TOL, Shallice (1982) found that articulatory suppression did not interfere with accuracy of TOL perform ance, and concluded that the task was not carried out using verbal rehearsal. How ever, perform ance on the articulation task was not measured, so it is unclear how dual task trade-off factors m ight have influenced perform ance. In the current study the effects of articulatory suppression on TOL perform ance and vice versa are examined in detail, to explore the role of verbal rehearsal in the task. 2. Random number generation: In relation to the Baddeley and Hitch m odel of working memory, articulating a sequence of numbers that is as random as possible loads the central executive component of working memory (and the phonol ogical loop). Producing random outpu t dem ands repeated inhibition of stereotyped autom atic sequences (Baddeley, 1986), and thus places considerable dem ands on the executive processes. Random num ber generation has been found to interfere with executive and reasoning tasks, such as figural fluency, a non- verbal ``frontal lobe’ ’ test (Phillips, 1997), and syllogistic reasoning (Gilhooly et al., 1993). However, it is im portant to note that random number generation does not interfere with perform ance on all cognitive tasks: for exam ple random generation did not significantly disrupt a m ental size com parison task (Pearson & Logie, 1998). If random num ber generation interferes with TOL perform ance, this would indicate that executive processing is involved in the task. How ever, executive function is unlikely to be a single entity but rather a fractionated system of control processes (Rabbitt, 1997), and it is of interest to speculate
214 PHILLIPS ET AL. whether control of verbal and visuospatial processing m ight activate rather different executive resources. If this is the case, and the TOL m akes demands predominantly on visuospatial executive processing, then there m ight be relatively little interference between random number generation and TOL. 3. Spatial pattern tapping: This secondary task requires participants to tap out a sim ple repeated spatial pattern. It is proposed to load the visuospatial scratchpad com ponent of working memory (Logie & M archetti, 1991 ) and has been shown to interfere with tasks such as the ``manikin’ ’ test which demands mental rotation (Farm er, Berman, & Fletcher, 1986). Pattern tapping does not interfere with tasks better solved by verbal strategies, such as syllogistic reasoning (Gilhooly et al., 1993), or verbal (AB) reasoning (Farm er et al., 1986). If spatial rehearsal processes are essential in the TOL (Joyce & Robbins, 1991 ; Owen et al., 1990; W elsh et al., 1995), then pattern tapping should interfere with TOL perform ance. 4. Random tapping : In order to investigate the role of non-verbal executive processes in cognitive tasks, tapping a sequence of locations as randomly as possible has been used (Baddeley, 1996). This task loads both central executive resources, and the visuospatial scratchpad. There is evidence that random tapping interferes significantly with performanc e on a range of verbal fluency tasks which are often used to test executive function (Phillips, Crawford, & Chapm an, 1998). Again, it can be noted that random tapping does not interfere with all tasks: e.g. short-term m em ory for colours rem ains intact (Pearson, 1996). Agreem ent that visuospatial working m em ory is involved in the TOL suggests that random tapping should interfere quite substantially with TOL performance. By com paring perform ance across the secondary tasks, conclusions about the involvement of different cognitive components in the TOL can be reached. For example com paring the effects of articulatory suppression (repeating the num bers from one to ten) and random number generation (trying to generate as random a sequence as possible using the numbers from one to ten) will allow inferences about the involvem ent of executive processes over and above the shared com ponents of the secondary tasks such as retrieval of digit names, verbal output, and maintaining the timing of outpu t at a regular pace. Also, com paring the effects of pattern tapping and random tapping will allow inferences about the executive com ponents of generating randomness over and abov e shared com pone nts suc h as co-ordina ting m otor respo nses and maintaining timing of taps. It is predicted that secondary tasks with a high executive load should affect TOL perform ance. Fu rther, the pattern of interference with articulatory suppression or spatial tapping will be determ ined by the extent to which TOL performance is dependent on verbal memory (M orris et al., 1993 ) or visuospatial memory (Owen, 1997). Possible trade-offs between prim ary and secondary tasks will also be investigated.
MEMORY AND THE TOWER OF LONDON TASK 215 METHOD Participants The participants were 36 youn g adults, 12 male and 24 fem ale, with a mean age of 20, and an age range of 18 to 25. They were recruited through posters or the subject panel of the Aberdeen Psycholog y Departm ent, and m ost were undergraduate students. Materials The TOL task was carried out on a PC equipped with a 17 ¢ ¢ monitor and lightpen. All com puter programs were written in Visual Basic 3.0. A wooden board, 280m m ´ 280mm , was used for the random tapping secondary task. This had a 135m m ´ 135m m opening containing nine 25mm square sprung switches arranged in three rows of three. The board was adjusted for the pattern tapping task by fixing a cross-shaped piece of black cardboard over the centre switches, leaving four switches arranged in a square. An open-fronted box was used to obscure the tapping board from view during performance of the task, to ensure there were no visual distractions from the secondary tasks. The switches were attached to another com puter for recording the location and latency of keypress responses. The intervals between responses for the verbal secondary tasks were recorded using a directional m icrophone attached to the second com puter via a voice key. The actual verbal responses for the random generation secondary task were recorded on a tape recorder. A digital metronom e set to beat once per second was used to help participants regulate the rate of responses. Procedure Tower of London Task There were eight 5-disk Tower of London trials in each condition. The com plexity of trials differed in terms of the number of moves required for solution (ranging from 3±11), and the number of indirect moves (ranging from 0±6), i.e. m oves that were essential to the optimum solution, but did not place a 1 disk into its final goal position (Goel & Grafman, 1995; W ard & Allport, 1997 ) . The trials were arranged in approxim ately ascending order of difficulty. See Fig. 1 Ward and Allport (1997) report that the number of ``chunks of indirect moves’ ’ is a better predictor of planning time in the TOL than the number of minimum moves or number of indirect moves. However, a program that computed all minimum move paths for TOL trials revealed that for most trials there is more than one minimum-move solution path, and these paths differ in terms of Ward and Allport’ s classification of ``chunks of indirect moves’ ’ but usually not in the ``number of indirect moves’ ’ (see Appendix), supporting the use of the less ambiguous latter measure.
216 PHILLIPS ET AL. 1 for an exam ple of the screen layout. In the Appendix, the num ber of moves required for each of the trials is listed. It can be noted that these are matched across sets (with the exception of trial 8, where the trials in set A and B could be solved in 10 moves, and the trial in set C in 11 moves: despite this there was no significant difference between the sets in terms of the average number of moves taken to solution, because relatively few individuals solved these trials in 10 moves). The participant was seated in front of the com puter m onitor with a practice TOL trial on screen. Instructions were given to m ove the coloured disks on the bottom set of three pegs to exactly m atch the positions of those on the first set. Participants were told that there were two parts to each trial: first mentally planning the moves to make the bottom set of disks m atch those of the goal set in the fewest possible moves, and second using the lightpen to move the disks on the bottom set of pegs as quickly as possible. The need to preplan before beginning to move the actual disks was emphasised. W hen satisfied that the final state of the disks m atched the goal state, participants selected a box marked ``accept’ ’ to m ove on to the next trial. The com puter recorded for each trial: preplan tim es (from appearance of disks to first movem ent with lightpen), tim e taken to make each m ove (com bined into a m ean move time per trial), and num ber of moves m ade to solution (converted to an excess m oves measure by subtracting the minimum moves in which trials could be solved). All participants carried out the Tower of Londo n (TOL) task in a control (task alone) condition and in two dual conditions ( either verbal random generation and articulatory suppression or spatial random generation and spatial tapping). The presentation order of the three task conditions was alternated between participants, and the three sets of problems were used equally for each condition. The secondary tasks were also given in single-task conditions, with no TOL task, in order to exam ine the trade-off between primary and secondary tasks. For all tasks (including control conditions), a m etronom e beat w as played throughout. TOL perform ance was assessed (in all control and dual-task conditions) in terms of plan times, move tim es, and num ber of excess m oves. Perform ance was assessed initially across all eight trials. TOL measures were analysed by separate repeated measures ANOVAs on plan tim es, m ove times, and excess moves, com paring scores in control and dual-task conditions. Post-hoc com parisons between the different conditions were made using the LSD test. Analyses were also carried out to exam ine whether the type of TOL trial (num ber of moves to solution, and num ber of indirect or goal m oves) was influential in determ ining the effects of dual tasks. Trials were analysed in pairs, m atched on the m inimum num ber of m oves to solution, and num ber of indirect moves required: three sets of two trials were com pared. Overall performance levels on each pair of trials indicated similarity in terms of planning times and excess m oves.
MEMORY AND THE TOWER OF LONDON TASK 217 Trials with no indirect moves: trials 1 & 2 had no indirect m oves, and three/ five minim um moves to solution. Trials with two indirect moves: trials 4 & 5 had seven moves to solution and mostly required two indirect moves (the exception being trial B4 with three indirect m oves) Trials with four indirect moves: trials 6 & 7 required nine m oves to solution, and had four indirect moves. Repeated m easures ANOVAs were carried out to examine the effects of dual- task condition and trial type on TOL perform ance in both verbal and visuospatial conditions. Dependent measures (planning time, and m oving time) were averaged across the two trials at each level of difficulty. Num ber of excess moves was calculated, but with only two trials at each difficulty level, the resulting distribu tions had large numbers of zero values, du e to m any participants m aking no excess moves on individual trials. Accuracy of perform ance was therefore recoded as a dichotomous variable (0 = no excess moves made across the two trials, 1 = one or more excess moves made across the two trials), and com parisons of the effects of secondary tasks on accuracy m ade by carrying out separate Cochran’ s Q tests at each level of difficulty. Secondary Tasks Articulatory suppression: The participant was told to say the digits 1 to 9 aloud repeatedly in time to the m etronom e beat. Verbal random generation: The participant was instructed to say aloud the digits 1 to 9 in as random an order as possible, in time to the m etronom e beat. Spatial pattern tapping: The participant was instructed to tap the four switches of the tapping board continuo usly in a clockw ise direction in time to the metronome beat, with the non-preferred hand. Spatial random generation: Participants were asked to tap each of the nine switches in as random an order as possible with the non-preferred hand, at the rate of one per second to the beat of the metronome. It is im portant to analyse perform ance on the secondary tasks, in order to investigate possible dual-task trade-offs between primary and secondary tasks. Secondary task performance was assessed in terms of mean inter-response interval and intra-individual variability in response times. The two executive secondary tasks were also assessed in term s of Evans’ RNG index of the relative frequency of digram com binations, on which higher scores indicate less efficient randomisation (Evans, 1978). In order to standardise the number of responses be ing analysed, secon da ry task pe rform anc e m easures were calculated separately for both the first and last 100 taps or vocalisations during TOL perform ance. The latter 100 responses would have occurred during more
218 PHILLIPS ET AL. dem anding TOL trials with a large number of indirect moves. Performance on secon dary tasks was examined by separate ANOVAs on each secondary task com paring control responses with the first 100 responses during TOL and last 100 responses during TOL. RESULTS TOL Task Performance Preplanning Time Counterintuitively, the time spent preplanning on the TOL was significantly reduced when the task was performed with any of the four secondary tasks. (see Fig. 2, first row). Indeed, under articulatory suppression, participants planned for only one third of the tim e spent during control TOL performance. Verbal secondary tasks caused considerably faster preplan tim es, F(2,34) = 38.2, P < .001, but there was no difference between articulatory suppression and random num ber generation in their effects on TOL preplan tim es. Visuospatial secon dary tasks also caused faster planning times, F(2,34) = 5.73, P < .01, but again there was no difference between pattern tapp ing and random tapping in their effects on TOL preplan times. Disk Moving Time The considerably faster preplan tim es during secondary task perform ance might suggest that participants would be slower at m oving disks, because extra time would have to be spent planning during the execution of moves. However, this was not the case (see Fig. 2). Analysis revealed that there was no effect of visuospatial secondary tasks on TOL move times, F(2,34) = 2.29, n.s. The verbal conditions did differ in their effects on tim e spent moving , F(2,34) = 18.8, P < .0001 , and post-hoc tests revealed that articulatory suppression differed from both control and random generation conditions, such that m oves were m ade more quickly during articulatory suppression. Number of Excess Moves Made There was a significant effect of verbal secondary tasks on the number of moves made in the TOL task, F(2,34) = 4.30, P < .05 (see Fig. 2, third row). Post-hoc tests revealed that random num ber generation resulted in more moves being m ade than either articulatory suppression or control conditions. There was also an effect of visuospatial secondary tasks on the number of moves made in the TOL task, F (2, 34) = 3.76, P < .05, with ra n d o m t a p p in g ca usin g significa n t ly m o re m o ves t o b e m a d e t h an in sin gle-t a sk co n d itio n s. P a tt er n ta p p in g d id n ot q u ite sign ifica n t ly affect T O L a ccu ra cy as co m p a red t o
FIG. 2. P r eplan tim es (top ), m o ve tim es (m id dle), and nu m b er of m o ves tak en (b otto m ) on the T O L t r ia ls in co n tr o l a n d seco nd a r y t ask co nd it io n s. Secon d a r y ta sk a b b r evia t io n s: R G = ra nd om gen era tion o f n u m b ers, A S = a r ticu lato ry su pp ression , R T = r a ndo m ta ppin g, P T = pa ttern tap ping. 219
220 PHILLIPS ET AL. co n t ro l p erfo r m a nce (th e L SD t est revea led th e sign ifican ce o f d ifferen ce b etween p a tt ern t a p p in g a n d con t ro l t o b e P = .06). Trial Type and the Effects of Dual Tasks on TOL The next set of analyses examined the effects of trial type (in terms of number of indirect m oves) and secondary task on TOL perform ance. In relation to the verbal secondary tasks, there were no differences between control, articulatory suppression, and random generation conditions in term s of the accuracy of performance, at any of the three levels of difficulty (Cochran’ s Q values of 4.67, 4.24, and 4.00 for zero, two, and four indirect-m ove trials respectively). Next, analysis of the effects of visuospatial secondary tasks on TOL excess m oves was considered. There was no effect of the secondary tasks on accuracy at any level of difficulty (Cochran’ s Q values of .17, 3.17, and .15, for zero, two, and four indirect-m ove trials respectively). There was a significant effect of dual-task condition on planning times, such that both articulatory suppression and random generation caused considerably less tim e to be spent preplanning , F (2, 34) = 5.21, P = .01. T h ere was a lso a n effect o f t ria l d ifficulty, F (2, 34) = 4.78, P < .05, with t h e t ria ls req u irin g fo u r in d irect m o ves resu ltin g in m u ch lo n ger p la n n in g t im es t h an th o se with n o in d irect m o ves. F u rt h er, th e in tera ctio n b etween d u a l-t a sk co n d itio n a n d tria l t yp e wa s sign ifica n t , F (4, 68) = 5.93, P < .001, a s sh o wn in F ig. 3. In FIG. 3. T he effects o f du a l task s a nd t ria l typ e o n T O L plan ning tim e.
MEMORY AND THE TOWER OF LONDON TASK 221 t h e co n tro l co nd itio n , t he m o re d em a n d in g t ria ls resu lted in co nsid era b ly lo n ger p la n n in g t im es (in crea sin g fro m 7 seco n d s t o 19 seco n d s), h o wever, in b o th du a l-t a sk co n d itio n s t h ere wa s n o effect o f t ria l d ifficu lty o n p la n n in g t im e, with a p p ro xim a t ely 6 seco n d s p er t ria l sp en t p la n n in g. A sim ila r p a tt ern wa s seen in rela t io n t o t h e sp a tia l seco n d a ry t a sk s. D u a l-t a sk co n d itio n a ffect ed th e t im e sp en t p la n n in g, with m u ch less t im e sp en t p la n n in g d urin g con cu rren t p a t t ern o r ran d om ta p p in g, F (2, 34) = 8.11, P < .01. T h ere was a lso a n effect o f t rial t yp e, F (2, 34) = 9.78, P < .001, with fou r-in d irect -m o ve t ria ls p la n n ed m o re slo wly t ha n t wo - o r n o -ind irect - m o ve tria ls. T h e in t era ctio n b etween d u al-t a sk co n d itio n a nd t ria l t yp e wa s a lso sign ifican t , F (4, 68) = 2.55, P < .05, a s sh o wn in F ig. 3. Sim ila r t o t h e verb al t ask s, in t h e co n tro l co n d itio n p la n n in g t im e in crea sed d ra m a t ica lly fro m n o -in d irect-m o ve t ria ls (7 seco n d s) t o fo u r-in d irect-m o ve t ria ls (22 seco n d s), wh ereas p la n n in g t im es d u rin g seco n d a ry ta sk s d id no t in crea se a s t ria l t yp e differed, rem a in in g a t a rou n d 10 seco n d s. In relation to the tim e spent moving on the TOL trials, there was an effect of du al-task con dition for verba l seconda ry tasks, F (2, 34) = 14.0, w it h a rticu la t o ry su p p ressio n ca u sin g fa ster m o ve t im es th a n eith er ra nd o m gen era t io n o r co n t ro l co n d itio ns. T h e effect o f t ria l t yp e on m o ve t im es wa s a lso sign ifica n t, F (2, 34) = 5.12, P < .05, with fou r-in d irect -m o ve trials b ein g p erfo rm ed m o re qu ick ly t h a n t wo -in d irect -m o ve tria ls. T h er e wa s n o in t era ctio n bet ween d u a l-t a sk con d itio n a n d t ria l t yp e, F (4, 68) = 1.44. T h ere wa s n o effect o f t h e sp a tial seco n da ry t ask s o n T O L m o ve t im es, F (2, 34) = 2.66. T ria l t yp e d id affect m o ve t im es, F (2, 34) = 5.65, P < .01, with lo n gest m o ve t im es o n t h e t wo -in d irect -m o ve t ria ls (as in t h e verb al co n d itio n s d escrib ed ea rlier). F in ally, t h ere was no in t er action between trial t yp e an d d u a l-t ask co nd itio n in d eterm in ing m o ve t im es, F (4, 68 ) = .516. Trade-off Between Primary and Secondary Tasks It is extrem ely im portant, when using dual-task m ethods, to exam ine the perform ance on both primary and secondary tasks in order to establish whether perform ance trade-offs are occurring. W e therefore report the results from analyses of inter-response intervals, variability, and Evans RNG scores for secondary task data. Analysis of Inter-response-Intervals (IRIs) from Secondary Tasks The data on IRIs are plotted in Fig. 4. Concurrently perform ing the Tower of London caused slower IRIs com pared to control performance in: the articulatory suppression task, F (2, 34) = 6.87, P < .01; ran d o m gen era t ion o f n u m b ers, F (2, 34) = 4.19, P < .05; a n d p a tt ern t a p p in g, F (2, 34) = 6.99, P < .01. T h ere wa s n o sign ifica n t effect o f co ncu rren t T O L perfo r m a nce on t h e IR Is
FIG. 4. M ea n in ter-r espo n se-interva ls (to p gr a phs) fo r contr ol a nd fir st / last 100 r espo nses o f the second ar y ta sk s d u rin g T O L ; a n d st an dar d d eviat io ns (b o ttom gr ap hs) of con tr ol an d fir st / last 100 I R I s du rin g T O L . Seco nd ar y ta sk a b br evia tion s: R G = r a ndo m genera tion o f n um bers, A S = ar ticula tory su p p ression , R T = ra nd om ta pp in g, P T = p attern ta ppin g. 222
MEMORY AND THE TOWER OF LONDON TASK 223 fo r ra n do m t ap p in g resp on ses, F (2, 34) = .67, n .s. T hese resu lts su ggest t h a t seco n d a ry t a sk perfo rm a n ce wa s gen era lly slo wed wh ile co n cu rren tly ca rryin g o u t t h e la t t er T o wer o f L o n d o n t ria ls, excep t in th e ca se o f ra nd o m ta p p in g. Intra-individual Variability of IRIs A more sensitive index of dual-task costs is often the variability of IRIs in the secondary task. This measures how accurately individuals can maintain a steady rhythm of verbal or tapping responses. The variability of the control, and first and last 100 IRIs during TOL for the four secondary tasks are plotted in Fig. 4, second row. Control performance showed less variable IRIs than dual-task perform ance in: articulatory suppression, F (2, 34) = 7.92, P < .01; ra nd o m n u m b er gen er a t io n , F (2, 34 ) = 3.95, P < .05; p a t t er n t a p p in g, F (2, 34) = 10. 5, P < .001; a n d ra n d o m t ap p in g, F (2, 34) = 13.01 , P < .0001. Seco n d a ry t a sk resp o n ses were co n sid era b ly m o re va ria b le d u rin g T O L a cro ss all o f th e t a sk s. Randomness of Number Generation and Tapping The final analyses of secondary task perform ance com pared the randomness of verbal output and keypresses using the Evans’ (1978) RNG index in control and concurrent TOL perform ance conditions (see Fig. 5). Verbal random num ber generation was no t significantly less random when carried out concurrently w ith the TO L task, althou gh the com parison app roached significance, F (2, 34) = 3.06, P = .06. In co n tra st, t ap p in g wa s sign ifica n t ly less ra n d o m d u rin g T O L , F (2, 34) = 6.33, P < .01. DISCUSSION Ca rrying out any of the secondary tasks caused participants to reduce substantially the am ount of time spent preplanning on the TOL task. This effect was consistent in both verbal and visuospatial, and high and low executive-loading secondary tasks. In control con ditions, planning tim e increased dramatically as the number of m oves and indirect moves required to solve trials increased. In contrast, there was no increase in planning time in corresponden ce to trial difficulty under any of the dual-task conditions. This suggests that accurate preplanning for com plex TOL trials was not carried out during secon dary task conditions. These results were contrary to expectations, and suggest that either people ca nnot plan wh ile carryin g o u t an y o t h er co gn itive a ctivit y; o r t ha t it is t o o effo rt fu l a n d a versive t o d o b o t h t h in gs a t o n ce. The TOL is usually seen as a task measuring the ability to plan: so preventing planning activity would be though t to cause poorer accuracy and slower
224 PHILLIPS ET AL. FIG. 5. E van s’ m ea su r e o f r an do m n ess (R N G ind ices) in contr o l con dition s a nd dur ing fir st / last 100 in ter-r esp on se interva ls con cur r ently with th e T O L . H igh er ind ex = less r a ndo m pr o du ctio n. execution of the task. How ever, none of the secondary tasks caused slower execution times. This was particularly surprising in the case of the spatial secon dary tasks, in which the motor com ponent of the non-preferred hand m ight have interfered with TOL execution by the preferred hand. Further, articulatory suppression caused people to execute the TOL move sequence m ore quickly than in the control condition. Both random tapping and random number generation caused more m oves to be m ade on the TOL, with the effect of pattern tapping on excess moves approaching significance. This suggests involvem ent of executive com ponents of working m emory in effective performance on the TOL, along with spatial memory. It could be argued that the detrimental effects of random tapping and random num ber generation on the num ber of excess moves m ade might be due to low preplanning times. This would suggest that the locus of the executive com ponent in the TOL is generating a mental plan of m oves. How ever, this explanation can be questioned. The condition that caused the greatest reduction in preplanning tim e, concurrent articulatory suppression, did not cause any change in the number of excess movesÐ indeed it resulted in a trend towards fewer excess m o ves. T h is su ggest s t h at t h e in crea ses in excess m o ves d u rin g o th er seco n d a ry t ask co n ditio n s were n o t a d irect resu lt o f decrea sed t im e sp en t p la nn in g. F ur t her, p la n n in g t im es wer e o n ly su b sta n t ia lly red u ced b y
MEMORY AND THE TOWER OF LONDON TASK 225 seco n d a ry t a sks d u rin g t h e m o st d em a n d ing T O L t ria ls, yet th ere wa s n o evid en ce t h a t t h e effects o f execu t ive seco n d a ry t a sk s o n excess m o ves wer e stro nger o n t h ese t ria ls. A ltern a t ively, execu tive seco n d a ry t a sk s m ay a ffect excess m o ves by redu cin g o n -lin e p la n n in g d u rin g T O L execu t io n . T h is fit s with t h e m o d el o f ``o p po r t un istic p la n n in g’ ’ p ro po sed b y H a yes-R o t h a n d H a yes-R o th (1979), in wh ich p la n n in g is la rgely an a ctivit y ca rried o u t in sm a ll b u rsts d u ring co gn itive a ctivit y, rat h er t h a n a sep a rat e p h a se in wh ich en t ire p la n s a re m a d e o f p erfo rm an ce b efo re execu t io n . Although non-executive dual tasks dram atically decreased the am ount of tim e participants spent preplanning their moves, this did not have a strong effect on the accuracy of TOL perform ance. This concurs with findings (W ynn et al., 1997 ) that varying initial preplanning tim es on the TOL task did not affect the accuracy with which people performed. In one study, some participants were instructed to plan as usual in the TOL task, while others were not told to plan. Those who were not told to plan spent considerably less time preplanning than those following ``plan’ ’ instructions, yet the two groups did not differ in term s of the num ber of excess moves made. The use of the TOL as a measure of planning can be questioned in light of the current evidence. Although som e elite individuals may be a ble t o p rep la n a la rge n u m b er o f m o ves in th e T O L t a sk if fo rced t o d o so b y t a sk in stru ctio n s (W a rd & A llp o rt , 1997), it seem s u n lik ely th a t m o st in d ivid u als will d o so if given t h e u su a l in stru ctio n s used in n eu ro p sych o lo gical a d m in ist rat io n o f t h e T O L : i.e. bein g t o ld t o p rep lan , b u t n o t p r even ted fro m o n -lin e p la n n in g d urin g t h e t ask . O n a verage, p a rt icip a n t s a re lik ely to u se a stra t egy o f o n -line p la n n in g b ecau se it is less d em a n d in g, yet effect ive in so lvin g t h e ta sk . P o o r p erform an ce o n t he T O L b y a n y in d ivid u a l o r clin ica l gr o u p sh o u ld n o t b e in terp reted as d u e t o d ysfu n ctio n al p la nn in g wit ho u t co nvin cin g su p p o rtin g evid en ce: a n u m b er o f o t h er co gn itive d eficit s co u ld a lso ca u se p o o r p erfo rm a n ce. In general, concurrent articulatory suppression appears to have had very beneficial effects on TOL performance: preplan tim es were a third that of the control condition, move times were significantly faster, and the num ber of moves made was fewer than in the control condition (although this difference was not significant). This apparently counterintuitive result concurs with other findings (e.g. Brandim onte & Gerbino, 1993 ; Hitch, Brandimonte, & W alker, 19 95) which sug gest that interfering with verbal articulation im proves perform ance on some visuospatial tasks. Brandim onte and Gerbino (1993) argue that articulatory suppression discourages the use of inefficient verbal strategies on visuospatial tasks, and promotes the use of more appropriate visuospatial strategies. In the current experim ent, random generation of numbers did not improve perform ance, suggesting that only relatively non-demanding secondary tasks can promote the use of visuospatial strategies. Another possibility is that the rhythm created by articulation might have improved perform ance by encoouraging faster m ovement. However, the advantage was
226 PHILLIPS ET AL. not found for any of the other secondary tasks, which also required responses at the same rate. The results suggest that performance on tasks involving spatial planning can be im proved by chanting one to nine during performance, and it would be of interest to look in future research at whether this effect extended to other tasks such as route planning. Perhaps instructions requesting preplanning result in verbally rehearsed plans being made (hence the very short plan tim es during articulatory suppression). However, during execution, these verbal plans have to be translated into motor movem ents, and evaluated on-line in relation to a visual display; so visuospatial memory is likely to be more important during execution of the plans. During execution, it might be difficult to translate a verbal plan into spatial movem ents, and to relate the plan to any unexpected interm ediate visual states (for exam ple, where a plan was inaccurate). Visuospatial on-line planning and evaluation is therefore likely to be m ore effective. Although the purpose of this paper was not to test models of working memory per se, t h e fin d in g t ha t a verb a l secon d ary t a sk can im pr o ve perfo r m a nce o n a sp at ia l p la n n in g t a sk h a s so m e im p lica t io n s fo r different m o d els o f wo rk in g m em o r y. In rela tio n t o t h e Ba dd eley a n d H itch m od el, t his result ca n b e in t erp reted in t erm s o f p reven t in g verb a l reh ea rsa l in t he a rticu la t o ry lo o p d u ring p la nn in g, wh ich wo u ld en co u ra ge t h e u se o f a n o p t im a l sp at ia l stra t egy in vo lvin g t h e visu osp at ia l scrat ch pa d . A lso , m o d els o f work in g m em o ry t h at p ro p o se in d ep en d en t lim ited -ca p acity verba l a n d visu o sp at ial reso u rces (Sh a h & M iya k e, 1996), co u ld a cco m m o d a t e t h ese fin d in gs in a sim ilar m a n n er. H o wever, th o se wh o p u t fo rward a sin gle lim ited -ca p acity view o f wo rk in g m em o r y (e.g. Sa lth o u se, 1990; Sw an so n , 1996 ; T u rn er & E n gle, 1989 ) m igh t h a ve d ifficulties in exp la in in g t h ese fin d in gs. Analysis of the secon dary task performance suggests that the TOL task was prioritised to the detrim ent of the secondary tasks. Articulatory suppression, random generation of num bers, and pattern tapping were all perform ed more slowly during the latter TOL trials. This might have been due to fatigue, although if so it is not clear why random tapping was not similarly im peded. Slowed responses could have been due to general problem s in monitoring and regulating speed of output, especially because in all of the secondary tasks response latencies were m ore variable during TOL performance as com pared to control conditions. The randomness of responses produced in both num ber and tapping conditions was lower during TOL, although this was not significant for num ber generation. The secondary task data suggest some degree of general, perhaps attentional, interference with TOL. It is possible that the interference between the secondary tasks and TOL could have been mostly at a peripheral level, for exam ple in terms of perceptual or motor response interference rather than related to higher-level memory or planning . However, findings from another experim ent suggest that this is not the case. Using a similar method to that outlined here, the effects of dual tasks on a
MEMORY AND THE TOWER OF LONDON TASK 227 TOL m otor task were investigated. The TOL motor task has been used to examine the contribution of sensori-motor factors in the TOL in a num ber of neuropsychological studies (e.g. Owen et al., 1996), and consists of trials matched in terms of start and goal states to TOL trials, but with each individual move demonstrated by com puter, so that the participant merely has to copy the moves made using a lightpen. This task therefore has the same perceptual and motor demands as the TOL task itself, but without the planning and memory com ponents. In contrast to the large effects of TOL on secondary task perform ance reported here (see Fig. 4), carrying out the TOL motor task did not significantly affect latency or variability of responses in any of the four secondary tasks. This suggests that the effects of TOL on the secondary tasks is largely due to the m em ory and planning dem ands of the task. It has been proposed that the TOL invok es visuospatial m em ory processes (Joyce & Robbins, 1991 ; Owen et al., 1990 ; W elsh et al., 1995). A contrary theory proposes that TOL taps active verbal processes because it is dependent on the left frontal lobe function (M orris et al., 1993). The current results suggest that visuospatial m em ory has a stronger role in TOL performance than verbal memory. Pattern tapping, a task proposed to load visuospatial rehearsal processes, caused greater interference with TOL perform ance than articulatory suppression, a task proposed to load verbal rehearsal. It is perhaps worth noting that the current version of the TOL task, with up to 11 move trials, is likely to be more demanding of m em ory than the neuropsychological version which tends to have a maxim um trial length of five m inimum m oves. The suggestion that the TOL is dependent on spatial, rather than verbal, memory processes is not necessarily contrary to findings that: TOL deficits are more com m on after left than right frontal lobe lesions (Glosser & Goodglass, 1990 ; Shallice, 1982), and brain activation during TOL in a norm al subjects is highest in the left frontal lobe (M orris et al., 1993; Owen et al., 1996). An alternative explanation for the neuropsycho logical findings is that the left frontal lobes are involved in visuospatial memory. Suppor t for this idea can be found in results reported by Owen et al. (1996). Owen et al. looked at brain activation in norm al adults during the TOL and a task of visuospatial memory. Both the TOL task and the visuospatial memory task activated the left frontal lobe, but com parisons revealed greater left frontal lobe activation during performance of the visuospatial m emory task as com pared to the TOL. This suggests that much of the frontal involvem ent of the TOL task may be defined in terms of visuospatial memory dem ands. In conclusion, these results indicate that the Tower of Londo n task does require executive processing, because both verbal and spatial concurrent executive tasks proved detrimental to perform ance. However, the nature of the executive processing in the task may relate m ore to the execution and monitoring of on-line planning than the ability to form effective preplans of large sequences of moves. Further, visuospatial rather than verbal m em ory is
228 PHILLIPS ET AL. involved in the task, as indicated by poorer performance during a concurrent spatial task and enhanced performance during a concurrent verbal task. Manuscript received 14 January 1998 M anuscript accepted 27 July 1998 REFERENCES Allamanno, N., Della Sala, S., Laiacona, M., Pasetti, C., & Spinnler, H. (1987). Problem solving ability in aging and dementia: Normative data on a non-verbal test. Ita lia n J our nal of N eur ological S ciences, 8, 111±120. Anderson, C.V., Bigler, E.D., & Blatter, D.D . (1995). Frontal lobe lesions, diffuse damage, and neuropsychological functioning in traumatic brain-injured patients. J our nal of C linica l a nd E xper iment al N europsychology, 17, 900±908. Baddeley, A. (1996). Exploring the central executive. Q ua r terly J our na l of E x per imenta l P sychology, 49, 5±28. Baddeley, A.D . (1986). W or king memory. O xfor d: O xfor d U niver sity P r ess. Baddeley, A.D., & Hitch, G. (1974). W orking memory. In G.A . Bower (Ed.), Recent a dva nces in lea rning a nd motiva tion (Vol. 8, pp. 647±667). N ew Y o r k : A cad em ic P r ess. Baddeley, A.D., Lewis, V.J., & Vallar, C. (1984). Explaining the articulatory loop. Q uar terly J our nal of E xper imental P sychology, 36, 233±252. Brandimonte, M .A., & Gerbino, W. (1993). W hen imagery fails: Effects of verbal recoding on accessibility of visual memories. In C. Cornoldi, R.H. Logie, M.A. Brandimonte, G. Kaufmann, & D. Reisberg (Eds.), S tr etching the ima gina tion: Representations of tr ansfor mations in menta l ima gery. (p p . 31±76). N ew Y or k: O xfo rd U niver sity P r ess. Cohen, G. (1996). M emor y in the r ea l wor ld ( 2nd E dn.) . H ove, U K : P sycho logy P r ess. Della Sala, S., & Logie, R. (1993). When working memory does not work: The role of working memory in neuropsychology. In F. Boller & J. Grafman (Eds.), H andbook of N eur opsychology (Vo l. 8, p p. 9±71). A m st erd a m : E lsevier . Evans, F.J. (1978). Monitoring attention deployment by random number generation: An index to measure subjective randomness. Bulletin of the P sychonomic S ociety, 12, 35±38. Evans, J.St.B.T., & Brooks, P.G . (1981). Competing with reasoning: A test of the working memory hypothesis. Cur r ent P sychologica l R esear ch, 1, 139±147. Farmer, E.W ., Berman, J.V.F., & Fletcher, Y.L. (1986). Evidence for a visuo-spatial scratch-pad in working memory. Q ua r terly J our na l of E x per imental P sychology, 38, 675±688. Gilhooly, K.J., Logie, R.H., W etherick, N.E., & Wynn, V. (1993). Working memory and strategies in syllogistic reasoning tasks. M emor y & C ognition, 21, 115±124. Glosser, G., & Goodglass, H. (1990). Disorders in executive control functions among aphasic and other brain damaged patients. J our na l of C linica l a nd E x per imental N europsychology, 12, 485± 501. Goel, V., & Grafman, J. (1995). Are the frontal lobes implicated in `planning’ functions? Interpreting data from the Tower of Hanoi. N eur opsychologia, 33, 623±642. Grafman, J., Jonas, B., & Salazar, B. (1990). W isconsin Card Sorting Test performance based on location and size of neuroanatomical lesion in Vietnam veterans with penetrating head injury. P erceptua l and M otor S kills, 71, 1120±1122. Hayes-Roth, B., & Hayes-Roth, F. (1979). A cognitive model of planning. C ognitive S cience, 3, 275±310. Hitch, G.J., Brandimonte, M.A., & Walker, P. (1995). Two types of representation in visual memory: Evidence from the effects of stimulus contrast on image combination. Memory & Cognition , 23 , 147±154.
MEMORY AND THE TOWER OF LONDON TASK 229 Joyce, E.M., & Robbins, T.W . (1991). Frontal lobe function in Korsakoff and non-Korsakoff alcoholics: Planning and spatial working memory. Neuropsychologia , 29, 709±723. Kafer, K.L., & Hunter, M . (1997). On testing the validity of planning/problem-solving tasks in a normal population. Journal of the International Neuropsychologica l Society, 3, 108±119. Logie, R.H., Gilhooly, K.J., & W ynn, V. (1994). Counting on working memory in arithmetic problem solving. Memory & Cognition, 22, 395±410. Logie, R.H., & M archetti, C. (1991). Visuo-spatial working memory: Visual, spatial or central executive? In R.H. Logie & M . Denis (Eds.), Mental images in human cognition (pp. 105±115). Amsterdam: Elsevier. McCarthy, R.A., & W arrington, E.K. (1990). Cognitive neuropsycholog y. London: Academic Press. Morice, R., & Delahunty, A. (1996). Frontal/executive impairments in schizophrenia. Schizo- phrenia Bulletin, 22 , 125±137. Morris, R., Miotto, E.C., Feigenbaum, J.D., Bullock, P., & Polkey, C.E. (1997). The effect of goal± subgoal conflicts on planning ability after frontal- and temporal-lobe lesions in humans. Neuropsychologia , 35, 1147±1157. Morris, R.G., Ahmed, S., Syed, G.M., & Toone, B.K. (1993). Neural correlates of planning ability: Frontal lobe activation during the Tower of London test. Neuropsychologia , 31 , 1367±1378. Owen, A.M . (1997). Cognitive planning in humans: Neuropsychological, neuroanatomical and neuropharmacological perspectives. Progress in Neurobiology, 53 , 431±450. Owen, A.M ., Downes, J.J., Sahakian, B.J., Polkey, C.E., & Robbins, T.W. (1990). Planning and spatial working memory following frontal lobe lesions in man. Neuropsychologia , 28, 1021±1034. Owen, A.M., Doyon, J., Petrides, M ., & Evans, A.C. (1996). Planning and spatial working memory: A positron emission tomography study in humans. European Journal of Neuroscience, 8, 353±364. Pearson, D.G . (1996). Working memory and strategy use during mental comparison tasks. Unpublished PhD, Aberdeen University, Aberdeen. Pearson, D.G., & Logie, R.H. (1998). Strategy use and working memory during mental size comparisons. Manuscript in preparation. Phillips, L. (1997). Do `frontal tests’ measure executive function? Issues of assessment and evidence from fluency tests. In P.M.A . Rabbitt (Ed.), Methodology of frontal and executive function (pp. 191±214). Hove: Psychology Press. Phillips, L.H., Crawford, J.R., & Chapman, P. (1998). Cognitive control processes in verbal fluency . M anuscript submitted for publication. Rabbit, P.M.A . (1997). Methodology of frontal and executive function. Hove, UK: Psychology Press. Reitan, R.M ., & W olfson, D. (1994). A selective and critical review of neuropsychological deficits and the frontal lobes. Neuropsycholog y Review, 4, 161±198. Salthouse, T.A. (1990). W orking memory as a processing resource in cognitive aging. Develop- mental Review, 10, 101±124. Shah, P., & M iyake, A. (1996). The separability of working memory resources for spatial thinking and language processing: An individual differences approach. Journal of Experimental Psychology , 125 , 4±27. Shallice, T. (1982). Specific impairments of planning. Philosophical Transactions of the Royal Society of London B, 298 , 199±209. Shallice, T. (1988). From neuropsycholog y to mental structure. Cambridge: Cambridge University Press. Sm ith, E.E., & Jonides, J. (1997). Working memory: A view from neuroimaging. Cognitive Psychology , 33, 5±42. Swanson, H.L. (1996). Individual and age-related differences in children’ s working memory. Memory & Cognition, 24, 70±82. Temple, C.M ., Carney, R.A., & Mullarkey, S. (1996). Frontal lobe function and executive skills in children with Turner’ s syndrome. Developmenta l Neuropsycholog y, 12 , 343±363.
230 PHILLIPS ET AL. Turner, M.L., & Engle, R.W . (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28 , 27±154. Ward, G., & Allport, A. (1997). Planning and problem-solving using the 5-disc Tower of London task. Quarterly Journal of Experimental Psychology, 50, 49±78. Watts, F.N., M acLeod, A.K., & Morris, L. (1988). Associations between phenomenal and objective aspects of concentration problems in depressed patients. British Journal of Psychology, 79, 241± 250. Welsh, M., Cicerello, A., Cuneo, R., & Brennan, M. (1995). Error and temporal patterns in Tower of Hanoi performance: Cognitive mechanisms and individual differences. Journal of General Psychology , 122 , 69±81. Wynn, V., Phillips, L., Gilhooly, K., Logie, R., & Della Sala, S. (1997). Planning and the Tower of London: An analysis of ` think-aloud’ protocols. Paper presented at the British Psychological Society Cognitive Section XIV Annual Conference, Bristol, UK.
MEMORY AND THE TOWER OF LONDON TASK 231 APPENDIX Information on the Tower of London Trials Used Minimum Indirect Solution Trial moves moves paths Start array of disks Goal array of disks A1 3 0 1 m4 m3 m2 m1 r1 r2 l1 l2 m1 r1 A2 5 0 1 l5 l3 l2 l4 l1 m1 r1 m3 m2 r2 A3 5 1 1 r2 m2 m1 l1 r1 l4 l1 l3 l2 r1 A4 7 2 1 m1 m2 lr r1 m3 r3 r1 m2 m1 r2 A5 7 2 2 r1 r3 m1 l2 m2 l2 r1 l1 m1 r2 A6 9 4 5 r2 r1 l2 l1 m1 r1 m1 l1 r2 l2 A7 9 4 3 r1 l1 m2 r2 m1 r2 r4 r5 r3 r1 A8 10 5 1 l1 l2 l4 l5 l3 r3 l1 l1 r2 r1 B1 3 0 1 m2 m1 r1 l1 m3 l3 m1 l4 l1 l2 B2 5 0 1 r1 m1 m2 m3 m4 l2 r2 r1 l3 l1 B3 5 1 1 r3 m1 r2 r1 r4 l2 l1 l4 r1 l3 B4 7 3 2 m1 r3 m2 r1 r2 m1 r3 l1 l2 m2 B5 7 2 1 l2 r3 r1 r2 l1 m1 l1 l2 l3 l4 B6 9 4 1 r2 r1 r4 r5 r3 l1 r4 r2 r3 r1 B7 9 4 3 m4 m2 m1 m3 m5 m1 m2 m3 r1 l1 B8 10 5 1 m5 m1 m3 m4 m2 m1 l2 l1 r2 r1 C1 3 0 1 r2 r1 l2 l1 r3 m2 r1 m3 l1 m1 C2 5 0 1 l2 m3 m2 r1 m1 l2 l1 r1 l3 l4 C3 5 1 1 l1 m3 l2 m1 m2 m3 r1 m2 m1 m4 C4 7 2 1 r4 l1 r3 r2 r1 m2 m1 r3 r4 r1 C5 7 2 2 l3 l2 m1 l1 m2 m1 r2 r3 r4 r1 C6 9 4 4 m1 r1 m3 m2 l1 r1 m1 m3 r2 m2 C7 9 4 3 l2 l3 r1 l1 l4 l2 r2 l1 l3 r1 C8 11 6 4 r2 r4 r3 r5 r1 r3 r1 r5 r2 r4 Trial: letter represents set (A, B, or C), number indicates trial order. Minimum moves: Minimum number of moves in which a trial can be solved. Indirect moves: The number of moves in a minimum solution path which does not place a disk into its target position. Solution paths: The number of different solution paths with which the trial can be solved in the minimum moves possible. Minimum moves, indirect moves and solution path measures were established using a computer program that calculates minimum move paths for TOL trials. Positions: Disks are reported in the same order of colours for each trial. Positions refer to peg, then height. The three pegs are referred to as l (left), m (middle), and r (right). The positions rise from 1 to 5, with 1 at the bottom and 5 at the top.
You can also read