The Nine Box Maze Test: A measure of spatial memory development in children

Brain and Cognition 52 (2003) 144–154 www.elsevier.com/locate/b&c

The Nine Box Maze Test: A measure of spatial memory development in children Linda M. Pentland,a,b,* Vicki A. Anderson,a,b Sherelle Dye,a and Stephen J. Woodc a Department of Psychology, University of Melbourne, Victoria, Australia, 3010 Murdoch ChildrenÕs Research Institute, Royal ChildrenÕs Hospital, Flemington Rd, Parkville, Victoria, Australia, 3052 Cognitive Neuropsychiatry Research & Academic Unit, Department of Psychiatry, University of Melbourne, Victoria, Australia, 3010 b

c

Accepted 21 April 2003

Abstract This study investigates the development of visuo-spatial memory in school-aged children, as measured by the Nine Box Maze Test Child Version (NBMT-CV). This task, originally developed for adults by Abrahams, Pickering, Polkey, and Morris (1997), utilises an allocentric framework to assess the complexities of spatial memory. Sixty children participated in this study (aged 5–12 years), which also involved administration of traditional Ônon-verbalÕ memory tests. Results indicate that visuo-spatial memory develops across childhood and that the NBMT-CV taps distinct skills compared to other Ônon-verbalÕ memory tasks. The theoretical, assessment and developmental issues raised by these findings are discussed. Ó 2003 Elsevier Science (USA). All rights reserved. Keywords: Allocentric; Hippocampus; Cognitive map

1. Introduction Non-verbal memory is a process that relates to the encoding and retrieval of spatial representations. As such, spatial memory includes storage of information about objects and their location. The capacity to store information in spatial ‘‘maps’’ allows for novel environments to become familiar, and thus represents an important skill at all stages of the lifespan. This familiarity in turn supports independent navigation, which is central to social behavior. Although much research has been directed towards investigating the development of non-verbal memory, theoretical and methodological issues have led to confusion around the nature and development of these functionally significant skills. Although neuropsychological tasks provide a sensitive means of assessing verbal memory processes, nonverbal memory measures are often criticized as not tapping relevant cognitive processes (Barr, 1997). Traditional non-verbal memory tests, such as the Spatial *

Corresponding author. Fax: +61-3-9345-5544. E-mail address: [email protected] (L.M. Pentland).

Learning Test (SLT) (adapted form LÕhermitte & Signoret, 1972) and Complex Figure of Rey (CFR) (Rey, 1941) have been administered to healthy children and yielded contradictory results (Anderson & Lajoie, 1996). The disparity in the development of non-verbal memory suggested by these two tasks may reflect a range of factors, including variable task difficulty and the absence of a rationale to link these measures to theories of spatial function. In addition, Luciana and Nelson (1998) emphasize that current non-verbal memory tests are multi-factorial, which clouds the interpretation of endpoint scores. Whilst the studies that have applied traditional non-verbal memory tasks to healthy children are useful in guiding interpretation of test scores, it is difficult to draw conclusions about cognitive development. Animal lesion studies provide a theoretical basis to guide expectations about the nature and development of non-verbal memory. Specifically, these studies suggest that there are two dissociable although interconnected visual processing systems (Ungerleider & Mishkin, 1982). The ventral pathway, often referred to as the ‘‘what’’ system (Barr, 1997), is responsible for encoding and storing object properties, such as shape and color.

0278-2626/03/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0278-2626(03)00079-4

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

The second is a dorsal or ‘‘where’’ system (Barr, 1997), which processes spatial properties such as location and size. On these grounds, non-verbal memory is more precisely defined in terms of a synthesis of visual and spatial (visuo-spatial) information. One major limitation of conceptualizing visuo-spatial processing in terms of ventral and dorsal streams is that this separation does not allow for the integration of such information, which is presumably fundamental to place learning. There are a variety of reference systems, by which objects and locations can be associated, which include egocentric, proximal and distal associations (Jagaroo, 1999). Distal association is the most adaptive level of encoding as it is independent of personal perspective (egocentric) and juxtaposed environmental cues (proximal). Distal or allocentric associations utilize remote cues or landmarks to form a coordinate reference system to create a map that allows the location of objects to be identified from any given starting point. It is this level of association that is thought to be mediated by the hippocampus (Jarrard, 1993; Morris, Garrud, Rawlins, & OÕKeefe, 1982; Nadel & MacDonald, 1980; Olton & Papas, 1979; Rudy, Stadler-Morris, & Albert, 1987). In humans there is emerging evidence that the right hippocampus supports allocentric memory (Abrahams et al., 1999; Abrahams et al., 1997; Maguire, Frackowiak, & Frith, 1996; Maguire, Frackowiak, & Frith, 1997). Whereas Maguire et al. (1996, 1997) demonstrated this relationship in healthy subjects using functional imaging techniques, Abrahams and respective colleagues (1999, 1997) studied a clinical population using a task, the Nine Box Maze Test (NBMT), which could easily be incorporated into standard neuropsychological assessment. This task is strongly grounded in OÕKeefe and NadelÕs (1978) cognitive mapping theory and includes separate measures of object-based and location (spatial) function. In accordance with Olton, Becker, and Handelmann (1979), their design also distinguished between information that was held constant across trials (reference memory) and that which varied (working memory). Results indicated that the location or spatial memory measures were the most sensitive indicator of right hippocampal damage. Reference and working measures generally did not differ across groups. Although innovative, the design did not consider the role of strategy, which Luciana and Nelson (1998) identified as a central consideration in interpreting overall task performance. In addition, subjects were not asked to associate object and location measures, which may have further extended the ceiling of the NBMT. Abrahams et al.Õs (1997) study raises the possibility that the same task could be modified for children and potentially used to address questions of visuo-spatial memory development. Although a number of studies have applied experimental allocentric tasks to healthy

145

children, there is no consensus in terms of the age at which children are expected to develop the capacity to form distal or allocentric associations. Estimates range from infancy (Huttenlocher, Newcombe, & Sandberg, 1994) through mid childhood (Overman, Pate, Moore, & Peuster, 1996) to maturity at age 9 (Lehnung et al., 1998; Piaget & Inhelder, 1967). Of note, none of these studies found significant gender differences in the development of these abilities. The significant disparity around the age at which a child develops the capacity to utilize distal cues, may be partially attributed to clouding of stages of development. That is, a lack of clarity around whether a skill is emerging, developing or mature (Dennis, 1989). In addition, most studies report the mean age at which milestones are attained. Whilst this dependent variable provides an approximate indicator of developmental stage, significant individual variability is often overlooked. One possible synthesis of the current literature is that rudimentary skills may emerge in infancy, continue to develop in early-mid childhood and mature around age 9. The primary aim of this study was to modify the NBMT developed by Abrahams et al. (1997) and investigate performance in a sample of healthy children aged 5–12. The NBMT shows promise as a useful measure of visuo-spatial memory development, given Abrahams et al.Õs (1997) finding that this task was sensitive to lesions in the right hippocampus (a part of the brain thought to be central to visuo-spatial memory), in adults. The basic design of Abrahams et al. (1997) was extended to include an associative component, whereby children were asked to explicitly associate object and location. Inclusion of this more difficult component extended the ceiling of the measure and improved the likelihood that the task would be sensitive to protracted development. In addition, the role of strategic behavior was investigated. Specifically, it was hypothesized that young children would demonstrate basic allocentric abilities, in-keeping with Huttenlocher et al. (1994). It was expected that these abilities would develop rapidly throughout early childhood and mature by the age of 9, consistent with the conservative estimate of Lehnung et al. (1998). It was expected that the object/location associative condition would be the slowest to mature, given the more difficult nature of this measure. No gender differences were anticipated. In terms of reference and working memory, it was hypothesized that there would be a significant correlation between reference memory and strategy. In the absence of awareness that the two objects and locations are held constant across trails, performance on reference and working memory measures was not expected to differ. Finally, it was expected that performance measures obtained from the Nine Box Maze Test—Child Version (NBMT-CV)

146

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

would be statistically separable from those of traditional psychometric tests, given its unique allocentric demands.

2. Method 2.1. Participants Sixty children (30 boys, 30 girls), aged 5.2–12.5 years, participated in the study. Children were selected from a range of state primary schools representing a broad socio-economic group. Children were randomly selected from those enrolled in age-appropriate classes. Parents were then sent information on the study together with consent forms and a questionnaire requesting demographic and developmental information. Children with a documented physical, sensory or cognitive impairment, and those currently enrolled in specialist education classes, were excluded. In this instance, and where parents did not consent, the next child on the roll was contacted. Participants were evenly distributed across Grades Prep (the first year of education in the Australian school system), 3 and 6, with equal numbers of boys and girls in each age group. As presented in Table 1, Grade Prep students were aged 5–6 years, Grade 3 children 8–9 years and those in Grade 6 were 11–12 years old. All participants spoke fluent English and were right handed. Socio-economic status (SES) was determined using the Scale of Occupational Prestige (Daniel, 1983), which rates parental occupation from 1 (high) to 7 (low). As reported in Table 1, SES was normally distributed across the sample and there were no significant differences among the three age groups (F ð2; 57Þ ¼ 1:59, p > :05). 2.2. Measures The NBMT-CV represented the core of the assessment battery. In addition, a range of established psychometric tests were employed for comparative purposes. These included four neuropsychological

measures of visuo-spatial function and two auditoryverbal tasks. 2.2.1. The Nine Box Maze Test—Child Version Abrahams et al.Õs (1997) design was adapted to include child-appropriate vocabulary, style and length of instructions and familiar yet engaging materials. A new scoring procedure, based on points received for correctly recognized or recalled objects and locations, was developed. In addition, participants were asked to link object and location information. Materials utilized in the NBMT-CV consisted of a toy car, spoon, apple, lollipop, ball, book, felt marker, toothbrush, teddy bear and cup; nine identical cylindrical containers or ‘‘bins’’ (17  15 cm) with detachable lids; a square table (74  74 cm); four matching chairs and an object recognition booklet, containing seven A4 size photographs of five and nine-item arrays of the objects. The NBMT-CV incorporated three stages: (a) Object Familiarization, (b) Five Box Maze (5BM), and (c) Nine Box Maze (9BM). In the Object Familiarization stage the examiner instructed the participant ‘‘IÕm going to show you some things that I want you to remember.’’ Ten objects were then presented individually for approximately 10 s each in a fixed order. To encourage attention to the object the examiner asked ‘‘What is this?’’ and ‘‘Would you play with it?’’ as each item was presented. After a brief delay, filled by Digit Span, the examiner asked, ‘‘What things did I show you before?’’ and recorded the childÕs responses. For the 5BM, five unlidded bins were placed in a circle on the table. The position of the bins and chairs around the table were not explicitly marked, but were known to the examiner. This 5BM, although not part of Abrahams et al.Õs (1997) adult version, was included to ensure the task was simple enough to tap emerging allocentric memory function. For each trial, the childÕs starting position, the position they moved to, the bins used and the objects hidden, were determined using a quasi-random sequence that was fixed for all participants. The child commenced the 5BM seated in Position

Table 1 Demographic characteristics of the sample Grade Prep (n ¼ 20)

Grade 3 (n ¼ 20)

Grade 6 (n ¼ 20)

Total (n ¼ 60)

Age M SD Range

5.82 0.42 5.18–6.58

8.89 0.21 8.58–9.33

11.83 0.42 11.17–12.50

8.85 2.50 5.18–12.50

SES M SD Range

3.52 1.13 1.70–5.90

3.76 0.87 1.90–5.30

3.76 0.84 2.30–6.00

3.68 0.95 1.70–6.00

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

1. The examiner informed the child ‘‘I am going to put two things into two separate bins, then get you to change seats and tell me which things and which bins were used.’’ On each trial the objects were placed into the bins in full view of the child. The lids were then placed on each bin before the child was asked to move to a designated position. Once the child was reseated, the examiner asked, ‘‘Which things were hidden?’’ If the child accurately recalled any objects, confirmatory feedback was given (e.g., ‘‘Well done’’) and a point was scored for each object freely recalled, as detailed on the record form. If the child could not freely recall one or more objects, they were asked to identify the remaining object/s from the first page of the object recognition booklet, with a photograph of the ball, spoon, apple, felt marker and book placed in a spatial array. If correct, confirmatory feedback was given and a single point was scored for each object recognized. If the child was still unable to remember one or more objects, the examiner told them which items were hidden. Participants who did not require this recognition trial were credited the object recognition points. The examiner then requested the child to ‘‘Point to the bins with things in them.’’ Children were given feedback as to whether the bin they identified was correct. If they did not identify all bins correctly the examiner provided this information. The child was then asked ‘‘Which things were hidden in which bins?’’ After the participant responded, immediate feedback was given and the lids were taken off the bins to reveal the items inside. A further point was scored for each correct association. Where participants gave more than one response, the accuracy of their first choice only was recorded. In situations where a participant simultaneously pointed to the bins when naming the hidden objects correctly, the above procedure was condensed and the participant received the maximum score for that trial. If an error was made however, the child was asked to separately nominate the objects and the locations used and appropriate scores were given. If the child successfully identified both objects, both locations, and linked this information correctly, they proceeded directly to the 9BM. If not, the child attempted a second five box trial. If required, the next page of the object recognition booklet was used, depicting the same five objects but in a different array. The instructions and procedure were otherwise repeated as described above. A third trial was provided if required. However, if a participant was still unsuccessful the test was discontinued as the child had not displayed the capacity to cope with the more demanding 9BM. The objects and locations utilized in the 5BM varied across trials. The instructions and procedure for the 9BM were similar to the 5BM however four objects were hidden within nine bins. The object recognition booklet con-

147

tained photographs of nine objects (the car, spoon, apple, ball, book, felt marker, toothbrush, teddy, and cup) in four different arrays. During this stage all participants completed four trials, regardless of the success of their responses. Throughout all trials two objects (the car and cup) and two locations (Bins 3 and 8) remained constant, providing measures of reference memory. The association between these objects and locations varied across trials. The remaining two objects and locations changed over each trial, providing measures of working memory. After the final trial, participants were asked, ‘‘What did you do to help you remember what things were hidden and where they were?’’ These self-reports were coded into five levels: 1, no strategy; 2, watching closely; 3, object based; 4, location based; 5, both object and location based. All measures derived from the NBMTCV are reported in Table 2. 2.2.2. Traditional memory measures Six psychometric memory measures were administered. Dependent variables for each measure are reported in Table 2, together with proposed cognitive demands. 1. Spatial Learning Test (SLT) (Anderson & Lajoie, 1996; adapted from LÕhermitte & Signoret, 1972). The SLT involves learning picture locations within a spatial array. Although the SLT is often referred to as a measure of visuo-spatial memory, verbal strategies can also be applied to the task. Trials to criterion represents the number of attempts the child required to correctly reproduce the spatial array over two consecutive trials. 2. Visual Reproduction (VR) (Wechsler Memory ScaleRevised, Wechsler, 1987). This measure requires recall of geometric shapes, both immediately and after a delay. The proportion of information retained was also calculated to allow for expected variation in immediate recall between age groups. 3. Complex Figure of Rey (CFR) (Rey, 1941). This task involves asking the child to copy a complex design, which taps visuo-spatial constructional ability. Incidental recall of the design was also tested after a 3-min delay. The accuracy of the childÕs initial copy and delayed recall were scored according to TaylorÕs criteria (cited in Spreen & Strauss, 1998). Waber and HolmesÕ (1985) scoring procedure was used to code the organization of the initial copy. 4. Block Span (BS) (Anderson & Lajoie, 1996; adapted from Milner, 1971). BS was included as a measure of processing capacity in the visuo-spatial domain. 5. Digit Span Forward (DS) (from the Digit Span subtest of the Wechsler Intelligence Scale for Children-III, Wechsler, 1991). The ÔForwardÕ condition was administered as a verbal comparator for BS.

148

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

Table 2 Description of variables Task

Variable

Range

Cognitive demands

Five Box Maze (5BM)

Total of objects recalled, objects recognized, locations recalled and objects and locations associated correctly across 3 trials

0–24

Composite measure of allocentric memory function

Nine Box Maze (9BM)

Total of all measures (as per Five Box Maze) across 4 trials

0–64

Object familiarization Total number of objects recalled across trials Number of reference objects recalled Number of working objects recalled Total number of objects recognized across trials Total number of locations recalled Number of reference locations recalled Number of working locations recalled Total number of correct object/location associations Strategy

0–10 0–16 0–8 0–8 0–16 0–16 0–8 0–8 0–16 1–5

Composite measure of allocentric memory function Processing capacity Object recall Object recall Object recall Object encoding Location recall Location recall Location recall Associative recall Executive function

Block Span (BS)

Maximum length correct

2–7

Processing capacity

Digit Span (DS)

Maximum length correct

3–9

Processing capacity

Spatial Learning Test (SLT)

Number of trials to criterion Number of locations recalled on delay

2–10 0–9

Associative learning Associative recall

Visual Reproduction (VR)

Immediate recall Delayed recall Proportion retained

0–41 0–41 0–100%

Object encoding Object recall Object recall

Complex Figure of Rey (CFR)

Copy score Delayed score Proportion retained Organizational level

0–36 0–36 0–100% 1–5

Visuo-spatial function Object recall Object recall Executive function

Paired Recall (PR)

Immediate recall (proportion correct) Delayed recall (proportion retained)

0–100% 0–100%

Associative learning Associative recall

6. Paired Recall (PR) (Test of Memory and Learning, Reynolds & Bigler, 1993). As verbal paired-associate tasks are known to be sensitive to dominant hippocampal function (Saling et al., 1993), this auditoryverbal task was included as a comparison measure for NBMT-CV. As per standard administration, participants aged 5–8 years were administered a 6 item list while those aged 9 years and above completed an 8 item list. A final trial was administered after a 15 min filled delay; to provide a measure of delayed recall. Given that the number of words varied across age groups, proportional indices were calculated for immediate and delayed recall.

mediate); Object Familiarization; DS; Object Familiarization (delay); 5BM; 9BM; SLT (delay); PR (immediate); CFR (copy); BS; CFR (delay); PR (delay).

2.3. Procedure

3.1. Nine Box Maze Test—Child Version

The Ethics Committees of the Department of Education Victoria and University of Melbourne approved the study. Families agreeing to participate returned a completed consent form and background questionnaire. Children meeting the selection criteria were individually assessed in a single session of approximately 40 min at their school in a quiet environment. All data was collected by one author (S.D.). Tasks were administered in the following fixed order: SLT (immediate); VR (im-

Results for all measures of the NBMT-CV are presented in Table 3. As predicted, childrenÕs performance on Object Familiarization increased across the age range, (F ð2; 57Þ ¼ 6:45, p < :01), with TukeyÕs post hoc analyses revealing children aged 8–9 and 11–12 years performed similarly and recalled significantly more items than younger subjects. Children generally performed the 5BM without difficulty and there were no significant differences across

3. Results In all parametric analyses, the distribution of the variables was examined. If assumptions were not met, the data was appropriately transformed (except where otherwise stated). The effect of gender was explored statistically, but did not reach significance for any variable.

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

149

Table 3 Performance on the NBMT-CV Measure

5–6 years M þ

Object familiarization (max ¼ 10)

8–9 years SD

M

11–12 years SD

M

SD

5.70

1.13

6.70

0.92

6.95

1.40

5BM Total (max ¼ 24)

23.50

1.24

23.60

1.23

23.70

0.98

9BM—Total scores Object recall (max ¼ 16) Object recognition (max ¼ 16) Location recall (max ¼ 16) Object and location (max ¼ 16) Total (max ¼ 64)

11.70 14.00 11.70 9.75 47.35

1.81 1.08 2.89 1.20 4.86

13.80 14.70 14.05 11.90 54.55

1.47 1.03 1.70 2.51 5.09

14.50 15.25 14.75 13.00 57.50

1.47 0.91 1.29 2.38 4.55

9BM—Reference memory Reference objects (max ¼ 8) Reference locations (max ¼ 8)þ

6.45 6.05

1.43 1.91

6.95 7.25

1.36 0.91

7.35 7.30

0.81 0.98

9BM—Working memory Working objects (max ¼ 8) Working locations (max ¼ 8)

5.25 5.65

1.25 1.60

6.85 6.75

0.88 1.37

7.15 7.45

0.99 0.83

+

p < :05. p < :01. ** p < :001. *

groups (F ð2; 57Þ ¼ 0:15, p > :05) (see Table 3). A ceiling effect was observed for the 5BM, as even young children performed close to the maximum and 53 students (88.3%) achieved the maximum score. Every child progressed to the 9BM. Table 3 also details the performance of each age group on the 9BM. In terms of summary scores (across all trials), abilities increased with age, as indicated by the significant main effects for object free recall (F ð2; 57Þ ¼ 18:85, p < :001); object recognition (F ð2; 57Þ ¼ 7:72, p < :01); location recall (F ð2; 57Þ ¼ 12:17, p < :001); object and location associations (F ð2; 57Þ ¼ 10:26, p < :001) and total performance (F ð2; 56Þ ¼ 23:29, p < :001). Post hoc analyses revealed that 5–6 year olds performed consistently more poorly than older students. The specific pattern of differences was that for all measures, except object recognition, 5–6 year olds performed more poorly than children in the older age groups, who were not significantly different from each other. This pattern suggests a developmental spurt in these abilities at around age 7. However, inspection of the means suggests that older students were not achieving maximal scores. Significant differences on object recognition were only apparent between the 5–6 and 11–12 years groups, suggesting a gradual development across the studied age range. In addition to summary scores, performance across trials was examined for each age group using a repeated measures design. Fig. 1 illustrates a similar pattern of performance across trials for each age group. This is confirmed statistically by the significant main effects for trial and age (F ð3; 171Þ ¼ 9:94, p < :001; and F ð2; 57Þ ¼ 23:94; p < :001, respectively), and the absence of a sig-

nificant trial  age group interaction (F ð6; 171Þ ¼ 0:96, p > :05). Repeated measures analysis was performed to investigate the hypothesis that the association of object and location developed more slowly than recall of either of these measures separately. This analysis revealed that there was a significant effect for recall type (F ð2; 114Þ ¼ 26:92; p < :001), with contrasts indicating that the separate variables for object and location recall differed significantly from the associative measure. There was no significant interaction between recall type and age (F ð4; 114Þ ¼ 0:16, p > :05). This pattern of results indicates that the capacity to associate object and location develops more slowly than the ability to recall this information separately. Repeated measures analysis of variance was employed to investigate the difference between reference and working memory for objects and locations, detailed in Table 3. Two separate analyses were performed. In the first analysis the difference between the number of objects recalled in the reference and working memory conditions represented the dependent variable. The main effect for condition was significant (F ð1; 57Þ ¼ 5:02, p < :05). Inspection of the means and contrasts confirm that subjects recalled more objects in the reference, as opposed to working, memory condition. The main effect of age was also significant (F ð2; 57Þ ¼ 18:85, p < :01). There was, however, no significant age X condition interaction (F ð2; 57Þ ¼ 2:48, p > :05). In the second analysis the dependent variable was the difference between the number of locations recalled in working and reference memory conditions. The main effect for this measure was not significant

150

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

(q ¼ 0:29, p < :05). ANCOVA indicated that the main effects for age were no longer significant when strategy was covaried for this measure (F ð2; 56Þ ¼ 1:21, p > :05). The correlation between strategy and location reference memory did not reach significance. In terms of working memory measures, strategy was significantly correlated with the both object (q ¼ 0:32, p < :05) and location recall (q ¼ 0:41, p < :01). ANCOVA revealed that the main effect for age remained significant when strategy was held constant for object and location working memory (F ð2; 56Þ ¼ 14:12, p < :01 and F ð2; 56Þ ¼ 4:26, p < :05, respectively). Fig. 1. Mean performance across 9BM trials by age group.

(F ð1; 57Þ ¼ 1:41, p > :05), indicating that the variability or stability of locations used did not impact on performance. Strategies used in the 9BM are summarised in Table 4. While most 5–6 year olds stated they had not employed a strategy, and the majority in the 11–12 year age group used more complex strategies, children aged 8–9 years gave more varied responses. In this interim age group, 50% of students reported they had either used no strategy or had watched closely (Levels 1 and 2), while 30% reported implementing an object and location based strategy (Level 5). Non-parametric Kruskal– Wallis ANOVA revealed significant main effects of age for strategy (v2 ð2Þ ¼ 20:80, p < :001). Further, strategy type significantly correlated with 9BM total scores (q ¼ :51, p < :001), the number of objects and locations recalled (q ¼ :39, p < :01; q ¼ :38; p < :01, respectively) as well as the number of object/location associations (q ¼ :49, p < :001). ANCOVA indicated that after statistically controlling for strategy type, age effects remained significant for the 9BM total score (F ð2; 56Þ ¼ 11:77, p < :001), the number of objects recalled (F ð2; 56Þ ¼ 12:10, p < :001), locations recalled (F ð2; 56Þ ¼ 5:85, p < :01) and object/location associations (F ð2; 56Þ ¼ 3:49, p < :05). The relationship between strategy score and reference and working memory measures was also explored. Spearman correlation revealed a modest relationship between strategy and object reference memory

3.2. The relationship between the NBMT-CV and psychometric tasks Partial correlations were adopted as a means of exploring the relationship between measures of the NBMT-CV and other psychometric tasks administered. All NBMT-CV measures were included in the analysis together with selected variables listed in Table 2. Whilst all derived variables were analyzed, not all are reported as there was a high degree of inter-correlation between many measures derived from the same task. The correlation matrix between the NBMT-CV (9BM) and traditional measures is reported in Table 5. Inspection of this table reveals a variable pattern of correlation between the 9BM and other measures. Of note, there is little correlation between the 9BM and percentage retained on VR and PR. Given that these latter measures are independent of the amount encoded originally, these data raise the possibility that the 9BM is more sensitive to encoding rather than recall of information. The stronger relationship between the 9BM, BS and VR (immediate) supports this interpretation. To explore this further the relationship between BS and the 9BM was investigated in more detail. The relationship between measures of processing capacity and learning was initially examined. ANOVA

Table 4 Type of strategy employed across age groups (N ¼ 60) 5–6 years 1. 2. 3. 4. 5.

No strategy 17 Watching closely 1 Object based 1 Location based 0 Object and 1 location based Total 20

8–9 years

11–12 years

Total

9 1 1 3 6

2 1 4 4 9

28 3 6 7 16

20

20

60

Fig. 2. Mean performance across SLT trials by age group.

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

151

Table 5 Correlations between scores on the NBMT-CV (9BM), Block Span (BS), Spatial Learning (SLT), Visual Reproduction (VR) Rey Figure (CFR) and Paired Recall (PR) for total students (N ¼ 60) Measure 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

9BM trial 1 9BM total 9BM objects total 9BM location total 9BM obj/loc total 9BM strategy BS SLT trials to criterion VR immediate VR percent recall CFR organization CFR percent recall PR percent recall

1 —

2 .72 —

3 

4 

.63 .74 —

5 

.57 .81 .37 —

6 

.57 .91 .50 .74 —

7 

.37 .49 .37 .40 .47 —

8 

.56 .69 .54 .57 .60 .48 —

9

).36 ).27þ ).28þ ).08 ).29þ ).30þ ).35 

—

10 

.54 .58 .56 .47 .47 .52 .75 ).29+ —

).19 ).00 .15 ).07 ).06 .13 ).12 ).08 ).12 —

11

12 

.45 .59 .54 .44 .47 .53 .65 ).32þ .71 .14 —

13 

.36 .42 .46 .29þ .32þ .37 .48 ).23 .53 .18 .64 —

.25 .15 .10 .15 .12 .06 .25 ).27þ .20 ).02 .33 .20 —

+

p < :05. * p < :01. ** p < :001.

revealed the main effect of age was significant for BS and DS, (F ð2; 57Þ ¼ 29:18, p < :001 and F ð2; 57Þ ¼ 10:92, p < :001, respectively). Collectively these results suggest that childrenÕs capacity to register information increases with age. As reported in Table 5, BS significantly correlated with 9BM total scores (r ¼ :69, p < :001), Trial 1 scores (r ¼ :56, p < :001), the number of objects recalled (r ¼ :54, p < :001), and locations recalled (r ¼ :57, p < :001). After covarying for BS, significant age effects were not detected for the Trial 1 score and for the number of locations recalled (F ð2; 56Þ ¼ 1:70, p > :05; and F ð2; 56Þ ¼ 1:81, p > :05, respectively), suggesting that these variables and BS utilized similar encoding and visual skills. As children appeared to predominantly use encoding to solve the first trial of the 9BM, an additional measure separating this trial from the total score (i.e., the total of Trials 2–4) was computed as a cumulative memory score. A significant age effect was identified for this cumulative score (summing trails 2–4) after statistically controlling for BS (F ð2; 56Þ ¼ 3:40, p < :05) which suggested that the 9BM cannot be mediated purely at an immediate or encoding level and requires consolidation of information across trials. The relationship between the 9BM and SLT was examined as both involve consolidation of material across multiple trials. Given this task similarity, a stronger relationship may have been expected than the modest correlations reported in Table 5. However, this weak correlation possibly reflects the Ôlow ceilingÕ apparent on the SLT, as even young children performed well. Although of interest, it was difficult to statistically compare the learning curves for these respective tasks, as data for the SLT was heavily skewed and could not be corrected by transformation. Fig. 2 plots the performance over trials for each age group. The relationship between the 9BM and other measures was further explored by principal components

Table 6 Varimax rotated principal components analysis employing summary variables Factor 1 a

Factor 2

9BM total recall 9BM trial 1 9BM trials 2–4 BS VR (immediate) VR (delay) CFR% recalled SLT (trials to criterion)

0.95 0.60a 0.94a 0.63a 0.47 0.47 0.27 )0.01

0.25 0.44 0.11 0.59a 0.76a 0.77a 0.65a )0.67a

Eigenvalue % of variance

4.73 59.06

0.98 12.20

a

Factor loadings > 0:50.

analysis (PCA) with varimax rotation. Two factors accounting for 71.30% of the variance were extracted, as presented in Table 6. Variables loading highly on Factor 1 (accounting for 59.06% of the variance) included the 9BM measures and BS, while Factor 2 (explaining 12.20% of the variance) included SLT, VR, and CFR variables, as well as BS. This factor structure suggests that the 9BM measures are tapping distinct processes to traditional non-verbal memory tests, with the exception of a commonality in general processing resources (as indicated by duel loading of BS). The separateness of these factors possibly represents the unique allocentric demands of the 9BM.

4. Discussion 4.1. The developmental trajectory This study applied the NBMT-CV, a measure of visuo-spatial memory adapted from Abrahams et al.

152

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

(1997), to a sample of healthy children aged 5–12. Results suggest that the NBMT-CV is a suitable measure of visuo-spatial memory for children, as evidenced by differential performance across age groups. Specifically, visuo-spatial memory, as measured by the NBMT-CV, has emerged by age 5, undergoes a developmental spurt around age 7 and then remains relatively stable between ages 8–12. It remains to be determined whether further development occurs during adolescence and what skill level represents maturity. In terms of emerging abilities, even the youngest children (aged 5–6) in this sample were able to master the 5BM, which was simplified only in terms of cognitive load but not task demands. This finding together with Huttenlocher et al. (1994) and Overman et al. (1996) suggests that the basic foundations of distal place learning are present in the young child, and subsequent development represents quantitative rather than qualitative changes. In the absence of the 5BM it would have been difficult to determine why 5–6 year olds struggled on the 9BM, possibly resulting in an underestimation of their visuo-spatial memory capacity. Performance measures of the more demanding 9BM consistently revealed that 5–6 year old subjects performed more poorly than those aged 8–9 and 11–12, who were indistinguishable from each other. Thus there is strong evidence of a spurt in visuo-spatial memory skills around age 7, which Overman et al. (1996) hypothesized reflects maturation of the hippocampus. This is a provocative hypothesis, particularly given the strong theoretical grounds to argue for hippocampal involvement in distal place learning. However, studies investigating the structural development of the hippocampus do not suggest a critical period of development around age 7 (Benes, Turtle, Khan, & Farol, 1994; Benes, 1998; Kretschmann, Kammradt, Krauthausen, Sauer, & Wingert, 1986; Utsunomiya, Takano, Okazaki, & Mitsudome, 1999). Alternatively, the improved performance at around age 7 may reflect utilization of more efficient strategies. Results of the present study suggest that strategy becomes more sophisticated with age, with a transition occurring within the 8–9 year old group. Although a correlation between strategy and performance was expected, only a weak relationship was identified. Whilst it is possible that performance on the NBMT-CV and strategy are not strongly related, this runs counter to the extensive literature on the association between executive functions and performance on memory tasks (Lehnung et al., 1998; Luciana & Nelson, 1998). Conversely it may be that the measure of strategy employed in this study is not a sensitive reflection of the approaches actually executed by subjects. Therefore, although this study and that of Overman et al. (1996) concur that there is a spurt in visuo-spatial memory around age 7, the underlying neuro-cognitive basis for this development remains a question for future investigation.

It was hypothesized that visuo-spatial memory development would be mature by around age 9. Whilst the performance of the 8–9 year old group was comparable to 11–12 year old subjects on NBMT-CV measures, both groups performed well below the ceiling of the task. Thus, in the absence of an adult or adolescent comparison group, it is difficult to determine if the abilities displayed by the older children in this sample have reached maturity. Abrahams et al. (1997) included some data on the performance of adult controls on the NBMT, which suggest a lower error rate for their control group than that observed in this study. Although this possibly indicates further development between child and adulthood, it is, difficult to directly compare Abrahams et al.Õs (1997) results and the present study, as there were some methodological differences including scoring procedure. Overman et al.Õs (1996) study also included an adult comparison group for some measures. Their results indicate 5–12 year olds perform comparably to adults on simpler measures, but not on more demanding allocentric tasks. Thus there is some indirect evidence to suggest that higher-level aspects of visuospatial memory undergo further development into adolescence and adulthood. As implied by Overman et al.Õs (1996) results, some aspects of visuo-spatial memory may mature at differential rates. In the present study this was supported by the finding that the capacity to associate object and location developed more slowly than recall of each measure in isolation. Although conjecture, this differential pattern of development may reflect unique neuro-anatomical correlates for the associative task, possibly involving the hippocampus. This would be in keeping with an associative theory of hippocampal function (Cohen et al., 1999; Squire & Zola-Morgan, 1991). The possibility of a differential pattern of development for reference and working conditions was also examined. There was no difference in the number of locations recalled in reference and working memory conditions. Subjects did, however, recall more objects in the reference as opposed to working memory condition. Of note, 5–6 year olds performed well on object recall in the reference condition. This pattern of results is not dissimilar to that observed by Abrahams et al. (1997) with their control group making few errors on the reference object condition. Thus it is possible that the reference condition for objects may be inherently easier than other measures, which raises the possibility that reference memory can be interpreted in terms of cognitive load. Although the significant relationship between strategy and reference memory for objects supports this stance, it cannot be directly established from the current measure of strategy that a high score equates to an awareness of objects being held constant across trials. To avoid this ambiguity, it may be more efficacious to

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

simply ask children were they aware that two objects and locations were used in each trial. As expected gender did not differentially effect performance. This finding is consistent with the majority of recent studies (Aliotti & Rajabiun, 1991; Anderson & Lajoie, 1996; Linn & Petersen, 1985). The absence of a relationship between gender and end-score performance measures, does not preclude the possibility that boys and girls may have approached the task differently, or recruited different brain structures, as found by Gron, Wunderlich, Spitzer, Tomczak, and Riepe (2000) in an adult functional imaging study. 4.2. What is the NBMT-CV measuring? The NBMT-CV has a different theoretical rationale and methodology to traditional measures of non-verbal memory. This difference was demonstrated statistically through PCA, which revealed that all NBMT-CV measures clustered onto one factor. This factor structure possibly reflects the unique allocentric or distal cue demands of the task, which is consistent with AbrahamsÕ et al. (1997) interpretation of the cognitive requirements of the task. This stance assumes that children have encoded the object and location using external (distal) cues. That is, cues other than their own position or coincident (proximal) information is used to code the information. The current study provides indirect evidence that the NBMT-CV taps allocentric demands, as having the subject move between trials makes it difficult to use an egocentric strategy. The design of this study also did not include direct or proximal cues, making it less likely that the task was mediated by these coincident cues. Although an indirect conclusion, the results of this study suggest that the NBMT-CV taps distal or allocentric processes, which may be sensitive to right hippocampal function. In addition to allocentric demands, there is evidence to suggest that the NBMT-CV draws on attentional capacity, given that BS shares variance with the measures of the NBMT-CV. A similar relationship was also found between processing capacity and other memory tasks. This non-specific result is consistent with the well-established interplay between processing capacity and new learning (Luciana & Nelson, 1998). For the NBMT-CV this relationship was particularly strong for the first trial, where group differences did not persist when BS was covaried. Group differences did, however, persist for all other trials. This suggests that processing capacity is a stronger determinant of performance on the first, as opposed to subsequent, trials. That is, the first trial may tap an encoding process, while subsequent trials are related to consolidation. The persistence of age effects after processing capacity had been controlled also suggests that the demands of the NBMT-CV cannot be met simply by holding information in an immediate memory store.

153

The NBMT-CV yields a range of measures and information not provided by traditional psychometric tests. At present, the SLT and CFR are often employed as measures of non-verbal memory. Although the SLT is arguably the most similar to the NBMT-CV, in that it requires the association of object and location information, only a weak correlation was identified and no common variance on PCA. There are however differences in the nature of the task as SLT can be mediated by verbal and non-verbal strategies and the information can be associated using a simple egocentric framework. The lack of relationship between these measures also provides indirect evident to suggest that the NBMT-CV cannot be mediated by an egocentric frame of reference. A slightly stronger relationship between the NBMT-CV and CFR was observed, which possibly reflects the more complex nature of the CFR. However, the NBMT-CV is a more diverse measure of visuo-spatial memory function as it allows for fractionation of object and location, as well as examination of the effect of increased complexity. 4.3. Future directions The results of this study generate several avenues for future research. In terms of the issues raised for healthy children, the developmental path through adolescents into adulthood remains unclear. At a methodological level, several simple modifications could assist in further clarifying the demands of the task. In the first instance, the measure of strategy needs to be revised. In addition to free recall of strategy it would be useful to ask children directly whether they were aware that two objects and locations were constant throughout. This would clarify what is being tapped in reference and working memory conditions, as at present this appears to be ambiguous. It would also be of interest to incorporate a final trial, where subjects recalled objects and locations without altering their position. This would provide data on the impact of changing perspective, which is assumed to invoke allocentric or distal processes. Although further work must be undertaken before the NBMT-CV could be considered a useful clinical measure, the results of this study highlight the potential of this task to provide developmental and normative information. Specifically, the current study did not find evidence of a floor effect, which suggests that the NBMT-CV has the potential to detect impaired or delayed visuo-spatial memory throughout childhood. As the original version of this task has already been found to be sensitive to non-dominant hippocampal dysfunction (Abrahams et al., 1997), it is seems likely that the same relationship may be found for children. Thus the NBMT-CV has the potential to provide a developmentally appropriate measure of visuo-spatial memory with the sensitivity to non-dominant hippocampal function that has eluded many traditional tasks.

154

L.M. Pentland et al. / Brain and Cognition 52 (2003) 144–154

References Abrahams, S., Morris, R. G., Polkey, C. E., Jarosz, J. M., Cox, T. C., Graves, M., & Pickering, A. (1999). Hippocampal involvement in spatial and working memory: A structural MRI analysis of patients with unilateral mesial temporal lobe sclerosis. Brain and Cognition, 41(1), 39–65. Abrahams, S., Pickering, A., Polkey, C. E., & Morris, R. G. (1997). Spatial memory deficits in patients with unilateral damage to the right hippocampal formation. Neuropsychologia, 35(1), 11–24. Aliotti, N. C., & Rajabiun, D. A. (1991). Visual memory development in preschool children. Perceptual and Motor Skills, 73(3 Pt 1), 792–794. Anderson, V. A., & Lajoie, G. (1996). Development of memory and learning skills in school-aged children: A neuropsychological perspective. Applied Neuropsychology, 3/4, 128–139. Barr, W. B. (1997). Examining the right temporal lobeÕs role in nonverbal memory. Brain and Cognition, 35(1), 26–41. Benes, F., Turtle, M., Khan, Y., & Farol, P. (1994). Myelination of a key relay zone in the hippocampal formation occurs in the human brain during childhood, adolescence, and adulthood. Archives of General Psychiatry, 51(6), 477–484. Benes, F. M. (1998). Brain development, VII. Human brain growth spans decades. American Journal of Psychiatry, 155(11), 1489. Cohen, N. J., Ryan, J., Hunt, C., Romine, L., Wszalek, T., & Nash, C. (1999). Hippocampal system and declarative (relational) memory: Summarizing the data from functional neuroimaging studies. Hippocampus, 9(1), 83–98. Daniel, A. (1983). Power, privilege and prestige: Occupations in Australia. Melbourne: Longman-Cheshire. Dennis, M. (1989). Language and the young damaged brain. In T. Boll, & B. Bryant (Eds.), Clinical neuropsychology and brain function: Research, measurement and practice (pp. 89–123). Washington: American Psychological Association. Gron, G., Wunderlich, A. P., Spitzer, M., Tomczak, R., & Riepe, M. W. (2000). Brain activation during human navigation: Genderdifferent neural networks as substrate of performance. Nature Neuroscience, 3(4), 404–408. Huttenlocher, J., Newcombe, N., & Sandberg, E. H. (1994). The coding of spatial location in young children. Cognitive Psychology, 27(2), 115–147. Jagaroo, V. (1999). Towards an analytic framework for the visuospatial domain: Spatial reference frames, cognitive operations, and neural systems. Journal of Clinical and Experimental Neuropsychology, 21(1), 134–146. Jarrard, L. E. (1993). On the role of the hippocampus in learning and memory in the rat. Behavioral and Neural Biology, 60(1), 9–26. Kretschmann, H. J., Kammradt, G., Krauthausen, I., Sauer, B., & Wingert, F. (1986). Growth of the hippocampal formation in man. Bibliotheca Anatomica, 28, 27–52. LÕhermitte, J., & Signoret, J. L. (1972). Analyse neuropsychologique et differenciation des syndromes amnesque. Revue Neurologique, 74, 20–38. Lehnung, M., Leplow, B., Friege, L., Herzog, A., Ferstl, R., & Mehdorn, M. (1998). Development of spatial memory and spatial orientation in preschoolers and primary school children. British Journal of Psychology, 89(Pt 3), 463–480. Linn, M. C., & Petersen, A. C. (1985). Emergence and characterization of sex differences in spatial ability: A meta-analysis. Child Development, 56(6), 1479–1498. Luciana, M., & Nelson, C. A. (1998). The functional emergence of prefrontally-guided working memory systems in four- to eightyear-old children. Neuropsychologia, 36(3), 273–293.

Maguire, E. A., Frackowiak, R. S., & Frith, C. D. (1996). Learning to find your way: A role for the human hippocampal formation. Proceedings of the Royal Society London, Series B, 263(1377), 1745–1750. Maguire, E. A., Frackowiak, R. S. J., & Frith, C. D. (1997). Recalling routes around London: Activation of the right hippocampus in taxi drivers. Journal of Neuroscience, 17(18), 7103–7110. Milner, B. (1971). Interhemispheric differences in localization of psychological processes in man. British Medical Journal, 27, 272–277. Morris, R. G., Garrud, P., Rawlins, J. N., & OÕKeefe, J. (1982). Place navigation impaired in rats with hippocampal lesions. Nature, 297(5868), 681–683. Nadel, L., & MacDonald, L. (1980). Hippocampus: Cognitive map or working memory. Behavioral Neural Biology, 29(3), 405– 409. OÕKeefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Claredon Press. Olton, D. S., Becker, J. T., & Handelmann, G. E. (1979). Hippocampus, space, and memory. The Behavioural and Brain Sciences, 2, 313–365. Olton, D. S., & Papas, B. C. (1979). Spatial memory and hippocampal function. Neuropsychologia, 17(6), 669–682. Overman, W. H., Pate, B. J., Moore, K., & Peuster, A. (1996). Ontogeny of place learning in children as measured in the radial arm maze, Morris search task, and open field task. Behavioral Neuroscience, 110(6), 1205–1228. Piaget, J., & Inhelder, B. (1967). The childÕs conception of space. (Langdon, F.J., Lunzer, J.L., Trans.). New York: W.W. Norton. Rey, A. (1941). LÕexamen psychologique dans les cas dÕencephalopathie traumatique. Archives de Psychologie, 28, 286–340. Reynolds, C. R., & Bigler, E. D. (1993). Test of memory and learning. New York: Psychological Assessment Resources. Rudy, J. W., Stadler-Morris, S., & Albert, P. (1987). Ontogeny of spatial navigation behaviors in the rat: Dissociation of ‘‘proximal’’and ‘‘distal’’-cue-based behaviors. Behavioral Neuroscience, 101(1), 62–73. Saling, M. M., Berkovic, S. F., OÕShea, M. F., Kalnins, R. M., Darby, D. G., & Bladin, P. F. (1993). Lateralization of verbal memory and unilateral hippocampal sclerosis: Evidence of taskspecific effects. Journal of Clinical and Experimental Neuropsychology, 15(4), 608–618. Spreen, O., & Strauss, E. (1998). A compendium of neuropsychological tests (2nd ed.). New York: Oxford University Press. Squire, L. R., & Zola-Morgan, S. (1991). The medial temporal lobe memory system. Science, 253(5026), 1380–1386. Ungerleider, L. G., & Mishkin, M. (1982). Two cortical visual systems. In D. J. Ingle, M. H. Goodale, & R. J. W. Mansfield (Eds.), Analysis of visual behavior (pp. 549–586). Cambridge, MA: MIT Press. Utsunomiya, H., Takano, K., Okazaki, M., & Mitsudome, A. (1999). Development of the temporal lobe in infants and children: Analysis by MR-based volumetry. American Journal of Neuroradiology, 20(4), 717–723. Waber, D. P., & Holmes, J. M. (1985). Assessing childrenÕs copy productions of the Rey–Osterrieth Complex Figure. Journal of Clinical and Experimental Neuropsychology, 7, 264–280. Wechsler, D. (1987). Manual for the Wechsler Memory Scale—Revised. New York: Psychological Corporation. Wechsler, D. (1991). Manual for the Wechsler Intelligence Scale for children (third ed.). San Antonio, TX: Psychological Corporation.