BRAIN AND COGNITION 31, 46–62 (1996)

ARTICLE NO. 0024

Developmental Change in Spatial Grouping Activity

among Children with Early Focal Brain Injury:
Evidence from a Modeling Task

JOAN STILES,* CATHERINE STERN,*,† DORIS TRAUNER,‡ AND RUTH NASS§

*Department of Psychology, University of California, San Diego; †San Diego State
University/University of California, San Diego Joint Doctoral Program in Clinical
Psychology; ‡Department of Neurosciences, UCSD Medical School; and
§New York University Medical School

Spatial construction skills were assessed in children with left (LH) or right (RH)
hemisphere focal brain injury and control children. Children copied simple and com-
plex block models which were rated on accuracy and spatial strategy. The accuracy
of simple and complex constructions for 4- to 5-year-old children with LH injury
was indistinguishable from 4-year-old controls. However, although they were able
to produce accurate complex constructions, the processes used by children with
LH injury differed from those of normal children. On both simple and complex
constructions, 4- to 5-year-old children with RH injury showed evidence of develop-
mental delay. For both accuracy and process measures, children with RH injury
performed at a level comparable to normal children at 3 years. A second group of
children with LH and RH injury were tested at 5 to 6 years of age. Both lesion groups
were indistinguishable from 4-year-old controls in terms of accuracy. However, both
children with LH and RH injury used different spatial processes than did controls.
This study emphasizes the importance of distinguishing between products of behav-
ior and the processes which underlie them. It is in the dissociation of products and
process of behavior that the subtle spatial construction deficits in this population
of brain-injured subjects is revealed.  1996 Academic Press, Inc.

Spatial analysis is an important aspect of visuospatial processing. It is

defined as the ability to encode a set of parts or elements from a visually
presented array and to integrate those parts to form a coherent and integrated
whole. In normal development these reciprocal processes are tightly yoked.
That is, the way in which the particular units are encoded influences how a

This research was supported by NICHHD Grant 5-R01-HD25077, NINDS Grant 5-P50-
NS22343, and by NIDCD Grant 5-P50-DC01289. The authors thank the parents and children
for their participation in the study, as well as Lisa Johnson, Trixie Maulion, and Robert Riss-
man for assistance in data transcription. Address reprint requests to Joan Stiles, Department
of Psychology 0109, University of California, San Diego, La Jolla, CA 92093.
46
0278-2626/96 $18.00
Copyright  1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 SPATIAL GROUPING ACTIVITY 47

pattern is integrated and the organized relational structure of the array influ-
ences what is encoded. In studies of adults, it has been well documented
that injury to left hemisphere brain regions interferes selectively with visual
segmenting processes and injury to right hemisphere brain regions interferes
selectively with visual integration processes (e.g. Warrington et al., 1966;
Gainotti & Tiacci, 1970; Swindell et al., 1988; Robertson and Delis, 1986;
Delis et al., 1988).
Recent studies have shown that pre- or perinatal injury to right and left
hemisphere brain regions results in patterns of selective deficits of spatial
analytic processing in children that are similar to those observed in adult
patients (Stiles-Davis et al., 1985, 1988; Stiles & Nass, 1991; Stiles & Thal,
1993). Specifically, children with LH injury have difficulty encoding pattern
features or details, while children with injury to RH areas have difficulty
with the configural aspects of spatial analytic processing. While these profiles
of deficit observed among children are consistent with those reported for
adults, these same studies report that the impairment observed among chil-
dren appears to be less severe and that children appear to be able to compen-
sate for their deficits more readily than adults.
Most studies of spatial deficits following early injury have focused on
children in the school-age period. However, we have reported one study
which examined spatial analytic processing among preschool-aged children
(Stiles & Nass, 1991). In that study, we showed that children as young as
3 years of age with both LH and RH injury are impaired on tasks involving
spontaneous block play. In these tasks, children were given small sets of
blocks and encouraged to build with them. Children with LH injury were
delayed in their constructional activity in that they produced much simpler
constructions than those of normally developing children. Children with RH
injury initially produced simpler constructions, and later produced anoma-
lous disordered heaps of blocks. Although the spontaneous play task pro-
vided clear evidence of specific deficit, the unstructured nature of the task
allowed us to provide only limited descriptions of the children’s behavior.
Furthermore, possible changes in the children’s construction abilities beyond
age 4 were not assessed.
The current study examined in greater detail the performance of 4- to 6-
year-old children on a more highly structured spatial grouping task. In this
study, children were asked to copy a series of six block constructions. Provid-
ing the children with a specific set of models made it possible to analyze
the children’s spatial grouping behaviors more precisely. In the current study,
the children’s construction activities were analyzed in terms of both what
they produced and how they produced them. Models included both simple
and complex constructions, with construction difficulty operationally defined
on the basis of a sample of data from normal children.
Data were analyzed in two ways. In the first study, performance of chil-
dren at 4 to 5 years of age was examined. This age was selected because it
48 STILES ET AL.

is the point at which children from both the LH and RH injury groups in
the earlier spontaneous grouping study began to attempt to generate construc-
tions more complex than single stacks and lines. Careful analysis of how
the children approach the problem of copying the model will provide bet-
ter insight into the nature of the underlying deficits in the groups. The se-
cond study examined the performance of an older group of children with
LH and RH injury. The purpose of these studies was to examine how the
children’s building strategies change over time. Studying behavior of dif-
ferent ages will allow us to begin to address questions about the extent to
which children are able to compensate for their spatial analytic deficits over
time.

STUDY 1: ASSESSMENT OF 4- TO 5-YEAR-OLD CHILDREN

Method
Subjects
The main set of analyses included a total of 32 children, 9 children with left hemisphere
injury (mean age 5 4.6), 7 children with right hemisphere injury (mean age 5 4.5), and 16
normal control children (mean age 5 4.0). A summary of neurological and behavioral findings
from the children with focal brain injury is presented in Table 1. All of the children in the
brain-injured groups had sustained a unilateral focal injury either prenatally or during the first
7 months of life, as documented by a CT or MRI scan. Children were excluded from the study
if they demonstrated evidence of multiple lesions, diffuse damage, or conditions which may
have caused global brain damage, such as a viral infection, encephalitis, or anoxia. The control
group consisted of equal numbers of boys and girls. All of the children in the control group
were tested within 1 month of their fourth birthday.
In a second set of analyses, two additional groups of neurologically normal controls were
introduced to the current investigation: a group of sixteen 36-month-old children and a group
of sixteen 24-month-old children. Each of these groups consisted of equal numbers of boys
and girls and all children were tested within 1 month of their birthday.

Stimulus Materials and Procedure

Each subject was tested individually, seated at a table across from the examiner. The subject
was presented with a set of eight wooden rectangular blocks measuring 7.5 3 5.0 3 2.0 cm.
Using an identical set of blocks, the examiner constructed a model behind a screen, out of
the view of the subject. The examiner lifted the screen and said to the subject, ‘‘Can you
make your blocks look just like mine?’’ Each subject was asked to copy six different block
models. Three of the models were designated as configurally ‘‘simple’’ constructions and three
of the models were designated as configurally ‘‘complex’’ constructions. This designation was
made on the basis of a sample of 16 (8 boys and 8 girls) neurologically normal 42-month-
old children. Constructions designated as ‘‘simple’’ were mastered by at least 75% of the
children in this group. Constructions designated as ‘‘complex’’ were mastered by less than
75% of the children in this group.

Measures of Spatial Construction

Each block construction was assessed on the basis of two separate scoring criteria: accuracy,
or ‘‘product’’ and strategy, or ‘‘process.’’
 TABLE 1
Summary of Neurological and Behavioral Data for 4- to 5-Year-Old Children with RH and LH Injury

Age at test
Subject Sex Lesion site Birth history and lesion onseta,b Seizures in years–months IQc

RH1 F Rt. temporal, parietal Presumed neonatal onset No 3–9 Bayley: 113
RH2 M Rt. frontal, temporal, parietal PPN: blood clot noted in placenta Yes 4–0 Not available
RH3 M Rt. frontal, parietal PPN: hemi noted 6 months No 4–4 WPPSI: 102
RH4 F Rt. parietal Surgery for congenital heart defect Yes 4–7 WPPSI: 84
on Day 1 of life.
RH5 F Rt. subcortical only, temporal, pari- PPN: left hemi No 4–9 WPPSI-R: 94
etal, occipital
RH6 F Rt. frontal, temporal, parietal, occip- Presumed neonatal onset Yes 5–1 WPPSI-R: 66
ital
RH7 M Rt. temporal, parietal Labor induced at 41 weeks No 5–4 WPPSI: 109
PPN: hemi noted at 3 months
Mean age 5 4–6 Mean IQ 5 94.7
LH1 F Lft. infarct to middle and posterior Possible gestational diabetes No 3–11 Not available
cerebral artery PPN: apneic spells Day 1 of life
hemi noted at 3 months
LH2 M Lft. frontal, temporal parietal Presumed neonatal onset No 4–0 WPPSI: 115
LH3 M Lft. frontal and subcortical Presumed neonatal onset No 4–1 WPPSI-R: 99
LH4 F Lft. parietal PPN: hemi noted at 6.5 months No 4–5 WPPSI-R: 116
SPATIAL GROUPING ACTIVITY

LH5 M Lft. parietal PPN: hemi noted at 7 months No 4–7 Bayley: 102
LH6 M Lft. temporal, parietal and subcor- PPN: Cesarean delivery, hemi noted No 4–9 WPPSI: 93
tical at age 1.
LH7 M Lft. temporal Cord around neck No 5–0 WPPSI-R: 93
LH8 F Lft. frontal PPN: neonatal seizures Yes 5–1 WISC-R: 68
LH9 M Lft. temporal, parietal Presumed neonatal onset No 5–2 WPPSI-R: 81
Mean age 5 4–5 Mean IQ 5 95.9
a
PPN, pre- or perinatal.
b
Unless otherwise noted, subject was born following a normal, full-term pregnancy and without birth complications.
49

c
WPPSI-R/WPPSI-III, Wechsler Primary and Preschool Scale of Intelligence—Revised; Wechsler Primary and Preschool Scale of Intelligence—
Third Edition. Bayley, Bayley Infant Intelligence Scales.
50 STILES ET AL.

FIG. 1. Scoring criteria for simple and complex block constructions.

Product. The accuracy of each construction was rated with a score of 0 or 2. A score of
2 indicated an accurate, and a score of 0 indicated an inaccurate construction. The scoring
criteria are described in Fig. 1. The number of constructions which earned full credit for
accuracy, or 2 points, was tallied separately for the three configurally simple constructions
and the three configurally complex constructions, giving each subject two overall product
scores.
Process. The spatial strategy, or process, used by the subjects to generate each block con-
struction was rated on the basis of an established scoring system (see Stiles and Nass, 1991,
and Stiles, 1988). Each construction was described as either a Process I, II, or III. A Process
 SPATIAL GROUPING ACTIVITY 51

I strategy was defined as the use of a simple reiterative relation in a single direction, such as
a stack or a line. A Process II strategy was defined as the use of sequential combinations of
blocks in more than one way or in more than one direction, such as the production of a line
and then a stack. A Process III strategy was defined as the use of a combination of blocks in
more than one way or in more than one direction, where a subject shifted back and forth
between the parts of his or her construction. These strategies differ largely on the basis of
complexity, with each successive level building upon the previous level.

Results

Construction Accuracy
The results of this study comparing normal children with the two lesion
groups, suggested that overall, children with RH injury produce less accurate
block constructions when compared to children with LH injury and 48-
month-old controls. The accuracy of children with LH injury was found to be
indistinguishable from that observed in a 48-month-old control population.
Data from the accuracy measure were analyzed using a two-way mixed-
design ANOVA, with Group (LH injury, RH injury, and 48-month-old
control) as the between-subjects factor and Construction type (simple and
complex) as the within-subjects factors. The dependent variable for the con-
structions was the total number of 2-point constructions (with a maximum
of three) produced by the subject. The dependent variable for the complex
constructions was the total number of 2-point constructions (with a maxi-
mum of three) produced by the subject. Tukey Unequal N Honestly Signifi-
cant Difference (HSD) tests were used for all multiple comparisons (p ,
.05).
The results suggest a significant main effect for Construction Type (F(1,
29) 5 10.9, p , .002) and for Group (F(2, 29) 5 12.2, p , .0001). Although
the Construction Type 3 Group interaction was not statistically significant,
the findings for the simple and complex constructions are presented sepa-
rately for purposes of clarity (see Fig. 2A). Tukey unequal N HSD tests
showed that children with LH injury and 48-month-old controls were sig-
nificantly more accurate in their production of both simple and complex con-
structions than were children with RH injury. There was no difference be-
tween the accuracy of children with LH injury and 48-month-old controls
on either the simple or complex constructions. These findings suggest that
on measures of block construction accuracy, children with LH injury demon-
strate a profile similar to that seen in a group of slightly younger controls.
Children with RH injury demonstrate a profile of accuracy that is signifi-
cantly inferior to the other groups.

Process Strategy
Children’s spatial processing strategies were examined using χ2 analyses
measuring the total number of Process I, Process II, and Process III strategies
52 STILES ET AL.

FIG. 2. (A) The mean accuracy score for simple and complex constructions for 4- to 5-
year-old children with LH and RH injury and 48-month-old controls. ( ) RH (N 5 7), (h)
LH (N 5 9), ( ) 48-month control (N 5 16). (B) The proportion of 4- to 5-year-old children
with LH and RH injury and 48-month-old controls who used Process I, II, or III in the produc-
tion of simple and complex constructions. (m) RH (N 5 7), (●) LH (N 5 9), (h) 48-month
control (N 5 16).

used by each group in the production of block models. Separate analyses

were performed for the simple and complex constructions.
The χ2 analysis of the strategies used for the simple constructions revealed
a marginally significant effect of Group (χ2(4) 5 8.84, p , .06). The results
of the process type analysis for the simple constructions are shown in Fig.
2B. Due to the uneven numbers of subjects in the three subject groups, these
results are expressed as proportions of subjects using each process strategy
and not absolute numbers. Separate χ2s were performed in order to compare
the performance of the two lesion groups to that of the 48-month-old group.
 SPATIAL GROUPING ACTIVITY 53

A marginally significant effect was found between the process types of chil-
dren with LH and RH injury (χ2 (2) 5 5.33, p , .07). The results of these
separate χ2s reveal that the types of processes used by children with LH
injury do not differ from those of the 48-month-old controls. However, a
significant difference between children with RH injury and 48-month-old
controls was revealed, (χ2(2) 5 8.25, p , .01). The observed frequencies
of the types of processes used in the production of the simple constructions
suggested that children with RH injury use different spatial strategies than
48-month-old controls.
For the complex constructions, the χ2 analysis revealed a significant effect
of group (χ2 (4) 5 21.1, p , .0003). Again, separate χ2 analyses were per-
formed in order to compare the performance of the two lesion groups and the
48-month-old controls. Separate χ2 analyses revealed significant differences
between children with LH injury and 48-month-old controls (χ2(2) 5 9.5,
p , .008) and between children with RH injury and 48-month-old controls
(χ2 (2) 5 8.9, p , .05). A significant difference was also found between the
process types used by children with LH and RH injury (χ2 (2) 5 7.6, p ,
.02). These findings suggest that for the complex constructions, children with
both LH and RH injury adopt spatial strategies that not only are unlike those
seen in a group of 48-month-old normal children, but are unlike one ano-
ther.

Construction Accuracy: Follow-Up Analyses

In order to determine whether the differences observed between the chil-
dren with LH injury and the controls reflect delay or deviation from the
normal developmental profile, it is necessary to compare performance with
younger normally developing children. Thus, a second set of analyses was
conducted. These analyses were identical to the first, except that two addi-
tional groups of normally developing children were included in the compari-
son. These data were analyzed using a two-way mixed-design ANOVA, with
Group (LH injury, RH injury, 48-month-old, 36-month-old, and 24-month-
old normal controls) as the between-subjects factor and Construction type
(simple and complex) as the within-subjects factors.
The results suggest a significant main effect for Construction Type (F(4,
58) 5 41.8, p , .0000) and for Group (F(1, 58) 5 42.1, p , .0001). The
Construction Type 3 Group interaction was statistically significant as well
(F(4, 58) 5 6.98, p , .0001). The data are presented in Fig. 3A.
A clear developmental trend in the accuracy of normal children’s block
constructions was revealed. On the simple constructions, Tukey unequal N
HSD tests showed that 48-month-old controls were significantly more accu-
rate in their reproductions than were the 24 month-old controls. The accuracy
of 48-month-old and 36-month-old controls on the simple constructions did
not differ, suggesting that by 36 months of age these constructions are well
54 STILES ET AL.

FIG. 3. (A) The mean accuracy score for simple and complex constructions for 4- to 5-
year-old children with LH and RH injury and 48-, 36- and 24-month-old controls. ( ) RH
(N 5 7), (h) LH (N 5 9), ( ) 48-month control (N 5 16), ( ) 36-month control (N 5 16),
(■) 24-month control (N 5 16). (B) The proportion of 4- to 5-year-old children with LH and
RH injury and 48-, 36-, and 24-month-old controls who used Process I, II, or III in the produc-
tion of simple and complex constructions. (m) RH (N 5 7), (●) LH (N 5 9), (h) 48-month
control (N 5 16), (e) 36-month control (N 5 16), ( ) 24-month control (N 5 16).

mastered by children in both age groups. On the complex constructions, the

performance of 48-month-old controls was significantly better than that of
the children in both the 36- and 24-month-old groups. Children in the 36-
month-old control group were significantly more accurate than were 24-
month-old controls.
On simple constructions, the accuracy of children with LH injury did not
differ from that of 48- or 36-month-old controls. However, children with
LH injury were found to be significantly more accurate than 24-month-old
 SPATIAL GROUPING ACTIVITY 55

controls. On complex constructions, the performance of children with LH

injury and 48-month-old controls did not differ. However, children with LH
injury were shown to be significantly more accurate than both 36- and 24-
month-old controls. These data confirm that the performance of children with
LH injury was comparable to that of normal 48-month-old, and significantly
better than that of 24- and 36-month-old controls.
The performance of children with RH injury on simple constructions sug-
gests that they follow a profile very similar to that exhibited by 36-month-
old children. Children with RH injury were shown to be significantly more
accurate than 24-month-old controls and significantly less accurate than 48-
month-old controls. The performance of children with RH injury did not
differ from that of the 37-month-old control group. These data place the
accuracy performance levels of the RH children at approximately that of
normally developing 36-month-old controls.

Process Strategy: Follow-Up Analyses

As with the product measure, a second set of analyses were conducted
which incorporated two additional groups of controls into the comparisons:
36- and 24-month-old normally developing children. On the simple construc-
tions, an overall χ2, including all groups was significant (χ2 (8) 5 35.7, p ,
.0001). The results of this analysis are presented in Fig. 3B. Examining the
process profiles among the groups of control children, reveals some clear
developmental trends. On the simple constructions, separate χ2 analyses re-
vealed differences among the process profiles of the 48-month-old and the
24-month-old controls (χ2 (2) 5 17.02, p , .0002), as well as between the
36-month-old and the 24-month-old controls (χ2(2) 5 21.5, p , .0001).
There was no difference between the process profiles of the 48-month-old
and the 36-month-old groups. As with the accuracy measure, this finding
may suggest that a ceiling-level performance is reached by 36 months of
age for the simple constructions.
On the simple constructions, children with LH injury demonstrated a sig-
nificantly different profile from the 24-month-old controls (χ2 (2) 5 13.11,
p , .001). There was no difference between the process profiles of the chil-
dren with LH injury when compared to either children in the 48-month-old
or the 36-month-old groups. This finding suggests that by 4 to 5 years of
age, children with LH injury are using processes similar to those observed
in a group of younger neurologically normal children in the production of
simple block constructions.
On the complex constructions, an overall χ2, including both lesion and all
control groups was significant (χ2 (8) 5 66.4, p , .0001). When the perfor-
mance of only the control children was analyzed, significant differences were
seen between the process profiles used by 48-month-old and 36-month-old
controls (χ2 (2) 5 8.3, p , .01), 48-month-old and 24-month-old controls
56 STILES ET AL.

(χ2 (2) 5 38.7, p , .0001) and 36-month-old and 24-month-old controls

(χ2 (2) 5 17.0, p , .0002). These findings suggest that with age, distinct
process profiles emerge and change in normally developing children.
Children with LH injury demonstrated process profiles which were dis-
tinctly different from each of the control groups; 48-month-old controls
(χ2 (2) 5 9.5, p , .008), 36-month-olds (χ2 (2) 5 13.7, p , .001), and 24-
month-olds (χ2 (2) 5 22.7, p , .0001) in the productions of complex con-
structions. Children with RH injury differed from two of the control groups;
48-month-olds (χ2 (2) 5 8.9, p , .01) and 24-month-olds (χ2(2) 5 8.03,
p , .05). The process profile produced by children with RH was not different
from that produced by 36-month-old children. Significant differences were
also obtained in a comparison of the process profiles of the two lesion groups
(χ2 (2) 5 7.6, p , .02).

STUDY 2: ASSESSMENT OF 5- TO 6-YEAR-OLD CHILDREN

Method
Subjects
Subjects for this Study included five children with LH injury (mean age 3 5.9) and six
children with RH injury (mean age 3 5.6). Two of the children in the LH injury group and
two children in the RH injury group contributed data, from an earlier testing point, to the
analyses described in Study 1. All children met the same criteria for inclusion as described
above. A summary of neurological and behavioral findings from this group of children with
focal brain injury is presented in Table 2. The same group of 48-month-old controls as de-
scribed in Study 1 were included in this study.

Stimulus Materials and Procedure and Scoring

The stimulus materials, procedures and scoring were the same as described in Study 1.

Results
Construction Accuracy
The same analyses as previously reported were used to assess the accuracy
of 5-year-old children with LH injury, RH injury and 48-month-old controls
on both simple and complex block constructions. The main effects for both
Group and Construction Type were not significant. Tukey unequal N HSD
tests revealed that there were no differences among the accuracy of the three
groups on simple constructions. The results are presented in Fig. 4. Similarly,
there were no differences among the accuracy of the groups on the complex
constructions. Taken together, these results suggest that by age 5 to 6 years,
children with LH and RH injury were indistinguishable from 48-month-old
controls with respect to the accuracy of their block constructions.
 TABLE 2
Summary of Neurological and Behavioral Data for 5- to 6-Year-Old Children with RH and LH Injury

Age at test
Subject Sex Lesion site Birth history and lesion onseta,b Seizures in years–months IQc

RH4 F Rt. parietal Surgery for congenital heart defect Yes 5–7** WPPSI: 84
on Day 1 of life.
RH8 F Rt. parietal, temporal PPN: seizures on Day 1 of life No 5–7 WPPSI: 110
RH9 F Rt. frontal, parietal PPN: cord wrapped around neck, Yes 5–8 Not available
emergency cesarean section, sei-
zures on Day 1 of life.
RH3 M Rt. frontal, parietal PPN: hemi noted 6 months No 5–9** WPPSI: 102
RH10 F Rt. frontal, temporal, parietal, occip- PPN: seizures in infancy, diagnosis Yes 5–9 WPPSI: 78
ital at 4 months
RH11 M Rt. frontal, temporal, parietal PPN: cord wrapped around neck, No 6–2 WISC-R: 123
hemi noted in early infancy
Mean age 5 5–9 Mean IQ 5 99.4
LH11 F Lft. frontal, temporal, parietal, oc- PPN: Difficult labor, hemi noted at No 5–4 WPPSI: 83
cipital 5 months
LH12 F Lft. MCA Normal delivery, stroke following Yes 5–4 WPPSI-R: 97
head trauma at 6.5 months
LH6 M Lft. temporal, parietal and subcor- PPN: cesarean delivery, hemi noted No 5–9** WPPSI: 93
SPATIAL GROUPING ACTIVITY

tical at age 1.
LH3 M Lft. frontal and subcortical Presumed neonatal onset No 5–5** WPPSI-R: 99
LH10 F Lft. temporal PPN: Hemi noted in early No 6–0 WPPSI: 107
childhood
Mean age 5 5–6 Mean IQ 5 95.8
a
PPN, pre- or perinatal.
b
Unless otherwise noted, subject was born following a normal, full-term pregnancy and without birth complications.
c
WPPSI-R/WPPSI-III, Wechsler Primary and Preschool Scale of Intelligence—Revised; Wechsler Primary and Preschool Scale of Intelligence—
Third Edition. Bayley, Bayley Infant Intelligence Scales.
57

** These subjects contributed data from an earlier testing point to Study 1.

58 STILES ET AL.

FIG. 4. (A) The mean accuracy score for simple and complex constructions for 5- to 6-
year-old children with LH and RH injury and 48-month-old controls. ( ) 5-year RH (N 5
6), (h) 5-year LH (N 5 5), ( ) 48-month control (N 5 16). (B) The proportion of 5- to 6-
year-old children with LH and RH injury and 48-month-old controls who used Process I, II,
or III in the production of simple and complex constructions. (n) 5-year RH (N 5 6), (s)
5-year LH (N 5 5), (h) 48-month control (N 5 16).

Process Strategy
The process profiles of children with LH injury, RH injury and 48-month-
old controls were compared using a χ2 analysis. For the simple constructions,
none of the children with LH or RH injury used Process III strategies, conse-
quently this variable was omitted from the analysis. Using a χ2 analysis with
a contingency correction, no differences were found among the three groups
with respect to the types of process profiles used. For the complex construc-
tions, none of the children with LH or RH injury used Process I, consequently
 SPATIAL GROUPING ACTIVITY 59

this variable was omitted from the analysis and χ2 analyses were conducted
using a contingency correction. In terms of the complex constructions, sig-
nificant differences were revealed among the process profiles of the two le-
sion groups when compared to a 48-month-old control group. Children with
RH injury were shown to adopt a significantly different process profile than
48-month-old controls (χ2 (2) 5 11.35, p , .0008). Children with LH injury
were shown to adopt a marginally different process profile than 48-month-
old controls (χ2(2) 5 3.17, p 5 .07). These findings suggest that despite the
similarity to 48-month-old controls on measures of accuracy, children with
LH and RH injury demonstrate differences on measures of process well into
the school-age years.

Discussion
This study examined the effects of early left and right hemisphere injury
on the performance of 4- to 6-year-old children in a simple block modeling
task. The overall findings are consistent with our earlier report on spontane-
ous block play in 3- and 4-year-old children (Stiles & Nass, 1991; Stiles
1988). The current study elaborates the earlier findings by both extending
the age of test into the early school-age period, and by introducing a more
structured task that allowed us to look in greater detail at both the product
and process of the children’s construction behaviors. Indeed it is in the com-
parison of what the children produced and how they produced it that provided
the best description of deficit and recovery for spatial analytic functioning
following early brain injury.
The data provide quite different performance profiles for children with
early occurring left and right posterior brain injury. By 4 to 5 years of age,
the performance of children with LH injury was indistinguishable from 48-
month-old normal controls on the products measure for both the simple and
the complex constructions. This is consistent with our earlier report on spon-
taneous grouping in which the LH injury group showed evidence of delay
at 3 years, but appeared to catch up by age 4 (Stiles & Nass, 1991). The
process measure assessment of the simple constructions also yielded similar
profiles for the LH injury and normal control groups. However, the process
profiles for the LH injury group on the complex constructions diverged from
those of normal 48-month-old controls. The children with LH injury ap-
peared to use simpler processes to produce these more complex construc-
tions. However, using only the single age comparison group, it was not possi-
ble to determine whether this profile represented deviation from normal
development or simply delay.
The children with RH injury presented quite a different profile from the
children with LH injury and from normal controls. Their performance on
both the product and the process measures was significantly different from
both groups for both the simple and the complex constructions. However,
60 STILES ET AL.

as was the case with the interpretation of data from the group with LH injury,
the single age control group does not allow us to determine whether the
profile is indicative of deviation or delay.
In order to address this issue of deviance versus delay, a second set of
analyses was conducted which included three groups of normally developing
children, ages 24, 36, and 48 months. Data from each of the lesion groups
were compared separately with these three groups. Considering first the data
from the normal children, a clear developmental profile is found. With in-
creasing age the children’s constructions became progressively more accu-
rate, with simple constructions being mastered earlier than complex construc-
tions. The way in which children organized constructions also changed with
age. Older children use more complex processes than younger children, and
all children are more likely to use simpler strategies when making simple
constructions than when making complex ones.
Given the earlier analyses which showed children with LH injury were
as accurate in their productions as 48-month-old normal controls, it was not
surprising to find that for both simple and complex constructions the LH
group was significantly more accurate in their constructions when compared
to the 24- and 36-month-old, but not the 48-month-old controls. A similar
pattern is evident on the process measures for the simple constructions. When
the constructions were simple, children with LH injury produced them accu-
rately and in the same way as the oldest normal children. However, data
from the process measure for the complex constructions showed significant
differences between children with LH injury and all three groups of controls.
Thus, while children with LH injury were able to produce accurate complex
constructions, the way they approached the task differed from normal chil-
dren, in that the children with LH injury used simple procedures to generate
complex constructions. These data are inconsistent with a profile of delay in
that none of the normally developing groups demonstrated this dissociation
between product and process. The impairment is subtle and reflects a devia-
tion from the usual developmental trajectory for spatial analytic processing.
Evidence for subtle, ongoing impairment is seen in the ways in which the
children with LH injury approach the task of solving spatial analytic prob-
lems.
Data comparing the performance of children with RH injury with the
younger normal controls provide a very different profile from the children
with LH injury. On both the product and the process measures, the children
with RH injury showed evidence of marked developmental delay. Perfor-
mance across all measures indicated that the group of children with RH in-
jury performed at a level most comparable to normal children at 36 months
of age. This pattern was consistent for both the simple and the complex
constructions.
Finally, analyses of the data from an older group of children with early
focal lesions showed continued evidence of subtle impairment. Children with
 SPATIAL GROUPING ACTIVITY 61

LH injury continued to present the pattern of dissociation between product

and process observed in the younger group. On the basis of construction
accuracy, children with LH injury performed at a level similar to normal
children. However, the spatial strategies used to generate these constructions
followed a pattern unlike that observed in the normal population. Interest-
ingly the older children with RH injury presented a very similar profile. By
age 5, children with RH injury were able to accurately reproduce both the
simple and the complex constructions. Thus, on the basis of accuracy mea-
sures alone, by at least the early school-age period, children with RH injury
were indistinguishable from normal children at 48 months. However, the
process measure showed that the way in which the children with RH injury
produce the forms differed from normal children. By age 5, the performance
profiles of children with LH and RH injury were very similar. Although both
groups were able to produce the forms, they both used simplified production
strategies.
Consistent with other reports, the results of this study show that early
injury to both left and right posterior brain regions affects children’s perfor-
mance on spatial construction tasks in the preschool period. Children with
RH injury appear to be more affected than children with LH injury, in that
they are initially more delayed, showing a more protracted developmental
trajectory than children with LH injury. During much of the preschool period,
the performance profile of children with RH injury is indicative of develop-
mental delay. However, by the early school age period their profile of perfor-
mance changes to one suggesting deviation from the normal population. Al-
though the children are able to produce accurate copies of the model
constructions, they use very simple procedures to do so. This profile is very
similar to that observed in children with LH injury beginning at an earlier
age. The primary distinguishing factor between the lesion groups in this
study is the timing of the observed changes in behavior.
The similarity of the performance profiles of the lesion groups in the early
school age period was intriguing and somewhat unexpected. Given other
reports of dissociation between the behavioral profiles of children with LH
and RH injury, the similarity of the profiles observed here is striking. How-
ever, it may well be that the current task and methods of assessment are
simply not powerful enough to distinguish subtle performance differences.
In this current study, the children were followed across a period of task mas-
tery. By the end of the developmental period, all of the children were able
to copy the target forms. The only measure of performance difference was
the process measure. Different kinds of spatial analytic impairment might
result in children adopting simplified processing strategies. Thus, while the
process measure provides important information about developmental devia-
tion from a normal developmental profile and subtle prolonged spatial im-
pairment, it is not sensitive to possible differences between the two groups
of early brain injured children.
62 STILES ET AL.

An important message of the current work is that studies which examine

only the products of behavior may be misleading. The processes which un-
derlie a particular outcome are also critical to our understanding of profiles
of impairment and development. In the current study, the patterns of impair-
ment are subtle, and indeed, in many cases not even evident in the products
of the children’s construction activity. It is the developmental dissociation
between levels of performance on the product and process measures that
gives rise to the deviant developmental profile. If the only measure of perfor-
mance in this study had been an assessment of products, the specification
of subtle impairment in this population would have been lost.

REFERENCES
Delis, D. C., Kiefner, M. G., & Fridlund, A. J. 1988. Visuospatial dysfunction following
unilateral brain damage: Dissociations in hierarchical and hemispatial analysis. Journal
of Clinical and Experimental Neuropsychology, 10, 421–431.
Gainotti, G., & Tiacci, C. 1970. Patterns of drawing disability in right and left hemisphere
patients. Neuropsychologia, 8, 379–384.
Robertson, L. C., & Delis, D. C. 1986. ‘Part-whole’ processing in unilateral brain-damaged
patients: Dysfunction of hierarchical organization. Neuropsychologia, 24, 363–370.
Stiles-Davis, J., Sugarman, S., & Nass, R. 1985. The development of spatial relations in four
young children with right-cerebral-hemisphere damage: Evidence for an early spatial con-
structive deficit. Brain and Cognition, 4, 388–412.
Stiles-Davis, J., Janowsky, J., Engel, M., & Nass, R. 1988. Drawing ability in four young
children with congenital unilateral brain lesions. Neuropsychologia, 26, 359–371.
Stiles-Davis, J. 1988. Developmental change in young children’s spatial grouping activity.
Developmental Psychology, 24, 522–531.
Stiles, J., & Nass, R. 1991. Spatial grouping activity in young children with congenital right
or left hemisphere brain injury. Brain and Cognition, 15, 201–222.
Stiles, J., & Thal, D. 1993. Linguistic and spatial cognitive development following early focal
brain injury: Patterns of deficit and recovery. In M. Johnson (Ed.), Brain development
and cognition: A reader. Blackwell Publishers: Oxford. Pp. 643–664.
Swindell, C. S., Holland, A. L., Fromm, D., & Greenhouse, J. B. 1988. Characteristics of
recovery of drawing ability in left and right brain-damaged patients. Brain and Cognition,
7, 16–30.
Warrington, E. K., James, M., & Kinsbourne, M. 1966. Drawing disability in relation to later-
ality of cerebral lesion. Brain, 89, 53–82.