475

British Journal of Psychology (2003), 94, 475485

2003 The British Psychological Society
www.bps.org.uk

Adults versus childrens performance on the

Stroop task: Interference and facilitation

Barlow C. Wright* and Amanda Wanley

Neurosciences Research Institute, Aston University, Birmingham, UK

Stroop tasks (Stroop, 1935) present stimuli having two dimensions, and participants
respond to one dimension whilst ignoring the other. The two dimensions are made
congruent, incongruent or neutral with respect to one another. Many claim that Stroop
interference is higher in children than in adults. However, taking interference as the
difference between the incongruent condition and the neutral condition, with inter-
ference within the congruent condition itself termed the incongruity effect then
surprisingly few studies directly address this issue. Also, there is recent debate as to
whether the facilitation effect (the contrast between the congruent and neutral
conditions) is just the opposite of interference. The present investigation (N = 31)
concerned a direct comparison between children and adults. Although the incongruity
effect reduced with age, interference did not. However, facilitation was far higher in
children than in adults. The groups opposite facilitation/interference asymmetry was
used to examine the recent claim that these derive from inadvertent word-reading and
suppression of semantic activation, respectively; rather than solely from the latter.

In the Stroop task (Stroop, 1935), participants are presented with stimuli having two
dimensions, and are required to attend to one dimension whilst ignoring the other
(Everatt et al., 1999; Long & Lyman, 1987; Milliken & Tipper, 1998). In the original task
(Stroop, 1935), the to-be-ignored (i.e. distractor or non-focal) dimension was a colour-
word and the to-be-attended (i.e. target or focal) dimension was the colour that the word
was written in. The general nding is that it takes longer to name colours in the context
of conicting colour words than when presented on their own. Stroops original
explanation for this was that, as a consequence of learning to read, associations between
written colour-words and their meanings become more direct than between perceptual
colour identication and the meaning of those percepts. The result is asymmetrical
inuences (Posner & Snyder, 1975; Shiffrin & Schneider, 1977), with word-reading
exerting greater interference on colour-naming than vice versa.

* Requests for reprints should be addressed to Barlow Wright, Neurosciences Research Institute, Aston University, Aston
Triangle, Birmingham, B4 7ET, UK (e-mail: [email protected]).
476 Barlow C. Wright and Amanda Wanley
Various aspects of this explanation have been expanded on over the years, but
almost all agree that the word is at least partially resolved during some stage of the
processing of the colour, resulting in any inadvertent activation of the meaning of
the word interacting detrimentally with the meaning of the colour-patch (i.e.
conicting semantic activationCohen, Dunbar, & McClelland, 1990; Dalrymple-
Alford & Budayr, 1966; Logan, 1980; Shiffrin & Schneider, 1977). The various
accounts differ in terms of the point(s) at which they assert that `meaning is
available (see Milliken & Tipper, 1998, for discussion), and how/why the semantic
effects emerge (Glaser & Glaser, 1989; Long & Lyman, 1987; Posner & Snyder, 1975;
Sugg & McDonald, 1994).
We note that most conclusions have been reached on the basis of adult research only
(see Milliken & Tipper, 1998); and this is despite reading prociency, identied by
Stroop as an important factor, being a skill that develops with age. Everatt et al. (1999)
argue that level of reading development, which is linked to age from around 4 years to at
least the mid-teens, does indeed inuence the amount of interference. On the additional
assumption that word-reading is obligatory as well as automatic from around 7 years
(MacLeod, 1991), we would expect children to show greater interference than adults
from the word dimension. Against this assertion, Jerger et al. (1993), from their own
ndings and a review of a number of other studies, conclude that Stroop interference
does not alter with age; although it may alter with certain other group differences, such
as in deaf versus hearing children. However, in contrast to these conclusions, most of
the relevant investigations have concluded that interference is lower in adults than in
children (Carter, Mintun, & Cohen, 1995; Comalli, Wapner, & Werner, 1962; Guttentag
& Haith, 1978; Vurpillot & Ball, 1979). Similar ndings have been reported for tasks
using a non-focal dimension other than a word (Balaban, Snidman, & Kagan, 1997;
Girelli, Lucangeli, & Butterworth, 2000). Thus, interference may be related to atten-
tional differences, with the word dimension amplifying the effect because of reading
being highly over-learned (Guttentag & Haith, 1978).
Arguably, the most inuential and widely cited developmental/life-span study has
been that of Comalli et al. in 1962 (e.g. see Everatt et al., 1999; Girelli et al., 2000; Jerger
et al., 1993; Schadler & Thissen, 1981). Indeed, in the most inuential review article to
date, MacLeod (1991, p. 185) writes `The overall picture of development, then, can be
summarized fairly concisely by the Comalli et al. (1962) result. Comalli et al. tested 11
participant groups aged from 7 to 80 years on tasks which included a neutral and an
incongruent condition, and reported a substantial decrease in interference with age.
Additionally, the 7- and 8-year-old groups and the 6580 year-olds showed much longer
latencies than the other groups. Comalli et al. noted that the 7-year-olds tended to
physically point to each successive stimulus before naming the focal dimension, and also
that the eldest group tended to produce additional verbalizations before producing
colour labels. We note that these tendencies, rather than interference per se, may
underlay the much longer naming times for the youngest and oldest groups. In line with
this possibility, Baumler (1969) found that, rather than increasing from 65 years,
interference systematically decreases from 50 up to 80 years. It should be noted that
this study included participants aged between 50 and 64 years, whereas the Comalli
et al. study did not. The contrasting ndings of Comalli et al. (1962) and Baumler (1969),
and the absence of more recent studies into age-related changes in Stroop interference,
leaves this issue unresolved.
Interference as measured by Comalli et al. (1962) was indexed by total naming time
for the incongruent condition. Whilst this measure does include interference, it also
 Adults versus childrens Stroop performance 477
includes other possible factors, such as distractability, visual scanning, general reading
uency or time to initiate motor sequences (Everatt et al., 1999). Developmental studies
still tend to use this index of interference (Balaban et al., 1997). However, some
studies (e.g. Carter et al., 1995) are more in line with adult research (e.g. MacLeod &
MacDonald, 2000), which additionally advocate another measure of interference; less
inuenced by these other factors and calculated by subtracting the naming time for the
neutral condition from that for the incongruent condition (cf. Dalrymple-Alford &
Budayr, 1966; Posner & Snyder, 1975). To aid in clarity, we hereafter refer to the
interference effect stemming from the impact of the non-focal dimension on the focal
dimension in the incongruent condition as the `incongruity effect and reserve the term
interference specically to refer to the detriment in performance in the incongruent
condition relative to the neutral condition (e.g. Glaser & Glaser, 1989; MacLeod &
MacDonald, 2000).
In Comalli et al.s study, performance in the neutral condition from each age group to
the next roughly shadowed performance in the incongruent condition. If we exclude
their 7- and 8-year-olds and their 65 + group, interference differences with age appear to
be much reduced. On this issue, most of the studies we could nd that calculated or
analysed interference over and above the incongruity effect support the view that there
is no reliable decrease in interference with age from around 9 to 18 years (e.g. Merrill,
Sperber, & McCauley, 1981; Visser, Das-Smaal, & Kwakman, 1996). However, this
conclusion is not without challenge. For example, Carter et al. (1995) found inter-
ference to decrease with age, and Schadler and Thissen (1981) found conicting
evidence indicating that interference both increases and decreases with age. What is
required is a study focusing on this issue, which uses more recent practices in
presentation techniques.
If interference results from greater semantic discord between non-focal and focal
dimension in the incongruent condition, relative to a neutral baseline, then it should
also be possible to aid (facilitate) the naming of the focal dimension by employing a non-
focal dimension congruent with it rather than incongruent with it (e.g. a word indexing
the same colour as the colour patch to be namedDalrymple-Alford & Budayr, 1966).
This facilitation effect is typically much smaller than interference (Glaser & Glaser,
1989). Sugg and McDonald (1994) account for this asymmetry by asserting that,
although a single construct of `suppression of semantic activation may underpin
both facilitation and interference (Posner & Snyder, 1975), the two stimulus dimensions
are nevertheless processed along differing pathways, with the pathway for words faster,
stronger or more automatic than that for colour patches (MacLeod & Dunbar, 1988;
MacLeod & MacDonald, 2000). This would predict that the cost of conict might be
greater than the benet of congruence; as a result of it taking longer to inhibit the
inappropriate word response compared to the advantage of congruency (the latter
would be limited because of colour patch naming and would then begin to approach
word-reading speed itself).
However, MacLeod and MacDonald (2000) argue that single-process accounts (e.g. in
terms of inhibition/suppression of semantic activation) fail to explain some pertinent
ndings on the balance between interference and facilitation. It seems that some tasks
affect facilitation but not interference (Dunbar & MacLeod, 1984; Glaser & Glaser, 1989;
Jerger et al., 1993), whilst others affect interference without affecting facilitation
(Hagenaar & Van der Heijden, 1986; MacLeod & Dunbar, 1988; Merikle & Gorewich,
1979; Tzelgov, Henik, & Berger, 1992). Tzelgov et al. (1992) and MacLeod and
MacDonald (2000) contend that this dissociability suggests that interference and
478 Barlow C. Wright and Amanda Wanley
facilitation are not caused by the same process (e.g. semantic priming) simply operating
in reverse.
The primary purpose of the present study was investigation of whether Stroop
interference does or does not decrease with age. Unlike much previous developmental
research, we used a computer presentation paradigm. Such paradigms are now standard
in Stroop research, partly because they offer more precise measurement and partly
because they permit measurement of responses on a trial-by-trial basis. However, to our
knowledge, this methodology has yet to be applied to a study focusing mainly on the
question of change in interference with age. A second aim was to investigate whether
either the interferencefacilitation asymmetry or the facilitation effect itself alters with
age, as this aspect of Stroop research has received very little attention up until now; and
to determine whether childadult differences might help resolve the issue of whether
interference and facilitation have a common source or different sources. Our nal aim
was investigation of whether the incongruity effect, and speed of responding more
generally, are or are not lengthened in children.

Method
Participants
A total of 31 participants took part in this experiment. Of these, 11 (4 males and 7
females) were children between 9 and 13 years of age (M = 11.23, SD = 1.36 years) and
20 (7 males and 13 females) were adults (M = 29.35, SD = 5.45 years). All participants
reported having no problems related to anomalous colour perception, and had normal
or corrected-to-normal visual acuity.

Materials
The Stroop task was presented on an HP 9067 computer with Proview 780 display.
Naming times were collected on a trial-by-trial basis and responses collected using a
headset microphone. Errors were noted by a researcher seated behind the participant.

Design
The experiment used a mixed factorial design with two main factors. The rst being
condition (with three levels corresponding to the incongruent, neutral and congruent
conditions) and the second being age (with two levelschildren vs. adults). Measures
of interference and facilitation were calculated from the condition times.
The colours used for the congruent and incongruent conditions were good
exemplars of red, green, blue and yellow. To offer a control for the presence of the
words themselves contributing to interference in the incongruent condition, the neutral
condition substituted the words `car, `sheep, `plug and `jigsaw in place of the
respective colour words. For children, all these words are not only well-known but
were also expected to be similar in relative frequency to each other; and so would not
render any obvious naming time disadvantage compared to the adults performance.
The words were also similar to the colour-words in syllabic and phonological
structure, with the deliberate exception of onset (Rayner & Posnansky, 1978).
However, they were not directly associated with any specic colour, and were also
not semantically related to each other (Carter et al., 1995; Marcel, 1983).
 Adults versus childrens Stroop performance 479
In order to minimize fatigue and to reduce biases towards or away from any
particular colour patch, the stimuli were presented on a white background of fairly
low intensity. Colour patches were rectangular in shape, with on-screen dimen-
sions of 2.2 cm high 3.5 cm wide. For each stimulus, a word was printed in black
immediately below the colour patch and centred relative to it; in a font equivalent
to Times New Roman 32 point and with the distance between the word and
colour patch roughly equal to half the height of the print used for the word. The
on-screen width of each stimulus (i.e. both dimensions) together with participants
seated approximately 60 cm from the screen, resulted in stimuli subtending a
horizontal visual angle of around 38 and 200 and a vertical visual angle of 48
and 12 0.
The experiment was divided into six blocks of trials, two blocks per condition. There
were 48 trials per block. The blocks were given in a random order, with the constraint
that one block of each of the three conditions was always presented before a second
block of any given condition was presented. Order of trials in each block was
randomized prior to testing. Two parallel forms of the experiment were made; thus,
in total there were four blocks of 48 trials per condition.

Procedure
Participants were tested in a quiet laboratory. They were briefed on the nature of the
task and given verbal instructions regarding what responses to make and how to make
them. Participants were informed that they would see colour patches with words
written beneath. They had to name the colour patch as fast as they could, ignoring the
word. The experimenter gave a short demonstration of the task and the required
responses. At this point, participants were informed of the importance of keeping errors
low, but also of responding to colour patches as quickly as they were able. Responses
were recorded using a headset microphone, with its input sensitivity adjusted to suit the
participant being tested.
Before beginning the test sessions, participants were given practice on each of
the three conditions. This came in the form of blocks of 12 trials, followed by a
break. A trial began with a variable delay of between 500 and 800 milliseconds
(ms). The delay on a given trial was followed by presentation of the colour patch
simultaneously with the word. Participants named the colour patch, ignoring the
word. As soon as the response for the current trial was registered, the next trial
was automatically begun. Breaks between practice blocks were used to give
feedback to participants as necessary. During these, participants indicated whether
they had given any incorrect or non-uent responses. These were briey discussed
along with any issues noted by the experimenter, with a corrective course of action
suggested if appropriate (e.g. `try to say only colour words). Participants were
permitted to repeat any or all of the practice blocks if unsure of any respect of the
procedure.
During testing, each participant took both parallel forms of the task in random order,
separated by a break of between 2 and 10 minutes. Additionally, within each task, there
was a break between each condition and each block, with participants beginning the
next condition or block only when they were ready. Each parallel form of the
experiment took between 15 and 20 minutes to complete, and together with practice
sessions, brieng and debrieng, the entire experiment took between 45 and 60
minutes.
480 Barlow C. Wright and Amanda Wanley

Results
For each condition, participants gave a total of 192 responses, across four blocks of 48
trials. No more than two error responses were observed in any block for any participant,
and error rates were similar in all three conditions. For these reasons, although mean
error data are presented (see bottom half of Table 1), no further analysis of errors will be
made here. Also, although it is possible to manipulate speed versus accuracy (Lee, Wee,
Tzeng, & Hung, 1992), typically performance on Stroop tasks is not characterized by
speedaccuracy trade-offs. Rather than one improving at the expense of the other,
participants giving faster responses also tend to make fewer errors (Girelli et al., 2000;
Jerger et al., 1993; MacLeod & Dunbar, 1988).
Response times were measured to a resolution of 1 ms. In order to reduce possible
outlier effects, the median naming time was calculated on a block-by-block basis. For
each condition, the mean of the four block-times was taken as a participants naming
time, calculated to the nearest millisecond. Table 1 (top half) summarizes the mean
naming times according to condition and age group. Adults tended to respond more
quickly than children, with this the case in all three conditions. The usual prole of
systematically increasing mean naming times with condition was observed for both
groups. A two-way (3 2) mixed model ANOVA was carried out to assess the statistical
signicances of the trends shown in Table 1. The factors were condition and age,
respectively; with repeated measures on the rst factor. This statistically conrmed
the effect of condition (F(2, 58) = 240.23, p = .001) and also the effect of age
(F(1, 29) = 12.536, p = .001). The interaction between condition and age was also
statistically signicant (F(2, 58) = 16.361, p = .001).

Table 1. Table of condition means for adults and children

Adults Children

Response times (milliseconds)

Congruent 582.40 (11.81) 613.73 (15.92)
Neutral 623.35 (14.54) 732.73 (19.61)
Incongruent 711.90 (16.91) 814.36 (22.80)
Interference 88.55 (4.57) 82.00 (6.17)
Facilitation 40.95 (9.12) 118.64 (12.30)
Errors (%)
Congruent 1.380 1.373
Neutral 1.510 1.800
Incongruent 2.240 2.651

Note. Number of participants = 11 children and 20 adults. Values in parentheses represent standard
errors.

As noted earlier, differences in Stroop effect according to age may have been due to a
general tendency to faster verbal or motor responses with age, rather than to differences
in interference itself. On this point, the statistically signicant interaction between age
and condition could be taken as a direct indication that, over and above any general
effect, Stroop interference and facilitation both increase with age. However, it should be
borne in mind that the interaction actually indicates that the differences between
the three pair-wise combinations of conditions (congruent vs. neutral, neutral vs.
 Adults versus childrens Stroop performance 481
incongruent and congruent vs. incongruent, when taken together) did increase with
age.
In order to permit more direct analyses of how interference and facilitation change
with age, we calculated these directly for each participant (Carter et al., 1995).
Interference and facilitation effects are displayed beneath the condition times of
Table 1. The interference effects for the two age groups were quite similar, even
though the age difference between the groups was around 16 years. The facilitation
effects for the two groups, however, were markedly different, with children showing a
facilitation effect some three times the magnitude of adults. As determining whether
Stroop interference changes with age was a main research question, a one-way ANOVA
was rst carried out on the interference data alone. This conrmed that interference did
not alter signicantly with age (F < 1). An analogous analysis on the facilitation data
conrmed that, unlike interference, facilitation did signicantly alter with age
(F(1, 29) = 25.74, p = .001).
In order to seek conrmation of whether the amount of facilitation relative to
interference altered with age group, a two-way ANOVA was then carried out, with age
group and Stroop-effect-type as factors. This indicated that there was an overall
difference between age groups (F(1, 29) = 13.063, p = .001), although no overall
difference between interference compared with facilitation was found (F < 1). Of
particular interest, the interaction between age and Stroop-effect-type was statistically
signicant (F(1, 29) =35.576, p = .001). This interaction statistically conrmed that the
prole of greater facilitation than interference in the children was reversed in the adults,
and this was driven by facilitation differences between the groups but not by
interference differences.

Discussion
Response times on a colour-word Stroop task systematically increased from the
congruent through the neutral and nally to the incongruent condition; with adults
faster than children in each condition. However, although the incongruity effect was
smaller in adults than in children, the interference effect did not differ between these
groups. Unlike interference though, the facilitation effect did decrease with age. We
now discuss these ndings, along with some implications for the issue of whether
interference and facilitation derive from the same source or distinct sources.
Our nding that mean response times for each of the three conditions were faster in
adults than in children is in line with previous ndings that the incongruity effect
decreases with age from around 7 years onwards (Comalli et al., 1962; Guttentag &
Haith, 1978). Vurpillot and Ball (1979) argue that one reason for children showing a
different incongruity effect to adults is that they do not yet have an adult concept of
what sameness means. However, children almost certainly do have an adult-like concept
of sameness by 11 years (Merrill et al., 1981). An alternative explanation is that partially
or wholly activated phonological-to-semantic representations of the non-focal dimen-
sion cause greater conict with the intended response (Logan, 1980; Rayner &
Posnansky, 1978; Rieger & Gauggel, 1999); and children take more time than adults
to suppress, inhibit or otherwise recover from such conicts (Balaban et al., 1997).
However, in one key respect there was no inappropriate response in the congruent
condition, and so no need to inhibit an inappropriate response; and yet children
responded more slowly in this condition as well as in the other two conditions (see
MacLeod & MacDonald, 2000, for discussion in relation to adults). Thus, development of
482 Barlow C. Wright and Amanda Wanley
inhibitory ability may not represent a complete explanation of why the children were
slower to respond than adults, across all conditions. Another cause might be that adults
are generally faster at planning and execution of motor sequences (Van Selst & Jolicoeur,
1997); and this was one reason for us considering interference as distinct from
incongruity.
Our analysis of interference showed this to be virtually identical in the child and
adult groups (Visser et al., 1996). Whilst very little previous research directly addresses
this developmental issue, interference has been found not to differ across other group
contrasts such as good versus poor comprehenders (Merrill et al., 1981) and rst
language versus second language (Lee et al., 1992). In contrast to interference, we found
facilitation to be greater in children than in adults (see Carter et al., 1995; Jerger et al.,
1993 for similar ndings regarding differences in facilitation for other groups). Glaser
and Glaser (1989) found that with the standard Stroop task with adults, facilitation is
much reduced compared with interference. Earlier, we noted that experimental
manipulations can affect facilitation without affecting interference (and vice versa),
which could be taken as suggesting two distinct processes (Tzelgov et al., 1992).
Following this view, MacLeod and MacDonald (2000) argue that facilitation is caused by
participants inadvertently reading the words whilst trying to focus attention on the
colour patches; with interference caused by difculty in suppressing the semantic
activation associated with correctly reading the words. They also argue that, whilst this
dual process explanation accounts both for facilitation and interference effects under a
myriad of experimental paradigms, a single-process account in terms of suppression of
semantic activation (e.g. Cohen et al., 1990; Posner & Snyder, 1975) would not sufce.
On the basis of MacLeod and MacDonalds (2000) `inadvertent word-reading
hypothesis, we might predict that children have greater difculty remaining focused
on the task at hand, which is to name the colour patch; and so will tend, more than
adults, inadvertently to read the words. In the congruent condition, if the participant
does not recover and outputs a word on a few occasions instead of the colour patch,
then the error will not be detected; possibly not even by the participant him/herself
(MacLeod & MacDonald, 2000). Of course, as word-reading proceeds faster than colour
naming, any increased tendency to read the words will speed up the congruent
condition relative to the neutral condition; contributing to a larger facilitation effect
in children than in adults.
MacLeod and MacDonald (2000) argue that the interference effect should not be
explained in the same way, but instead should be explained in terms of semantic
activation plus suppression. Regarding children, it does seem reasonable to expect that
these mechanisms are well developed perhaps as early as 7 years (Comalli et al., 1962;
Everatt et al., 1999). Although children might respond more slowly than adults, the
difference between the neutral and incongruent condition will be driven partly by the
difference in semantic activation caused by the inherent structure of these respective
conditions (Glaser & Glaser, 1989). As the latter differential can be considered to be
xed in so far as the same task is given to children and adults, we should predict that the
performance difference (i.e. the interference effect) would also be the same in children
as adults. Regarding the suppression component, partially activated semantic represen-
tations, stemming from partially resolved phonological representations of the non-focal
dimension, will need to be suppressed if the appropriate response is to be given. This
incurs a cost (e.g. in terms of timeRieger & Gauggel, 1999). Such a cost is also present
in the neutral condition, but in this case the uninvited semantic activation is easier to
suppress, and in any event would not represent a competing response in the same sense
 Adults versus childrens Stroop performance 483
as in the incongruent condition. Thus, as with the semantic component to interference,
any difference between adults and children regarding the suppression component
would lead to less efcient performance, but would do so in the incongruent and neutral
conditions to the same extent.
We note that as well as MacLeod and MacDonalds (2000) theory, a single-process
account can also explain our ndings both for interference and facilitation. The
explanation of interference is essentially the same as the MacLeod and MacDonald
(2000) explanation. In the congruent condition, however, should there be any residual
semantic activation from the ignored word, this would add to activation of the colour
patch, which would pass its threshold for generating or initiating the correct response
sooner (Logan, 1980). Additionally, children more than adults would tend to fail in
suppressing semantic activation from the word in all three conditions. This would affect
the neutral and incongruent conditions to the same extent (slowing both), but would
slow the congruent condition far less, because any residual semantic activation would
have little, if any, negative effect on that condition. Thus, either an explanation in terms
of two separate processes or an explanation in terms of only a single semantically based
process can account for our results.
However, although an explanation based on applying MacLeod and MacDonalds
theory to children does correctly predict our results both for facilitation and inter-
ference, this account is not parsimonious. Specically, no reason is given for why
inadvertent word-reading should feature only in the congruent condition, and suppres-
sion of semantic activation only in the remaining two conditions. Similarly, any account
purely in terms of suppression of semantic activation needs to acknowledge that, at least
on some trials, inadvertent reading of the word would probably occur. We therefore
favour a synthesis couched in terms of both processes applying to all three conditions.
As espoused earlier, both accounts are broadly in agreement regarding the neutral
and incongruent conditions. In the congruent condition, inadvertent word-reading
responses and congruent semantic activation would interact in a way they could not do
in the other two conditions. Specically, on congruent trials where engagement of
word-reading was begun but then successfully disengaged (i.e. the response was not
based on the word), there might nevertheless be some residual semantic activation from
the word that means that activation of the colour patch does not have to start from the
normal resting level (Cohen et al., 1990). Also, any failure to suppress inadvertent word-
reading or semantic activation from the word dimension, would be benecial to
performance only in the congruent condition. Thus, as long as we hold that children
are more prone to inadvertently attending to the word dimension than adults are (Girelli
et al., 2000; Posner & Rothbart, 1992; Rayner & Posnansky, 1978), either the inadvertent
word-reading component alone or the suppression of semantic activation component
alone can explain the relative size of the facilitation and interference effect we found
between children and adults. However, both together offer a more plausible account of
our ndings of greater facilitation in children than adults, but identical interference.

Conclusions
Stroop interference is said to increase up to around 7 years of age, and then decrease to
around 65 years. We tested the second part of this claim, that young adolescents show a
larger interference effect than do adults. Taking `interference as the absolute level
of performance in the incongruent condition, a phenomenon for which the term
484 Barlow C. Wright and Amanda Wanley
`incongruity effect is suggested, then our ndings support the view that it is decreased
in adults relative to young adolescents. Taking interference as the disadvantage of the
incongruent condition relative to the neutral condition, however, children and adults
actually exhibit the same level of interference as one another.
Although interference is identical in children and adults, children show greater
facilitation than interference, with the reverse asymmetry in adults. MacLeod and
MacDonalds (2000) inadvertent word-reading hypothesis can account for these oppo-
site proles; but does not explain why inadvertent reading would not apply to the
neutral and incongruent conditions nor why semantic processing would not feature in
the congruent condition. We note that an account solely in terms of suppression of
semantic activation can also predict our ndings. However, this account does not
explain why inadvertently reading responses would not play a part particularly in the
congruent condition. In the absence of these answers at this time, we favour an
explanation in terms of both processes, rather than either alone. However, further
research is needed to determine which of them occurs more often, and whether either
features more in children than in adults.

References
Balaban, M. T., Snidman, N., & Kagan, J. (1997). Attention, emotion, and reactivity in infancy and
early childhood. In P. J. Lang & R. F. Simons (Eds.), Attention and orienting: Sensory and
motivational processes (pp. 369 391). Hillsdale, NJ: Erlbaum.
Baumler, G. (1969). Decrease in achievement capacity as a result of age with particular reference
to the Stroop-interference tendency. Psychologische Beitrage, 11, 3468.
Carter, C. S., Mintun, M., & Cohen, J. D. (1995). Interference and facilitation effects during
selective attention: An H2150 PET study of Stroop task performance. NeuroImage, 2,
264 272.
Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the control of automatic processes: A
parallel distributed processing account of the Stroop effect. Psychological Review, 97,
332 361.
Comalli, P. E., Jr., Wapner, S., & Werner, H. (1962). Interference effects of Stroop colour-word test
in childhood, adulthood, and ageing. Journal of Genetic Psychology, 100, 4753.
Dalrymple-Alford, E. C., & Budayr, B. (1966). Examination of some aspects of the Stroop color-
word test. Perceptual and Motor Skills, 23, 1211 1214.
Dunbar, K., & MacLeod, C. M. (1984). A horse race of a different color: Stroop interference
patterns with transformed words. Journal of Experimental Psychology: Human, Perception
and Performance, 10(5), 622639.
Everatt, J., McCorquodale, B., Smith, J., Culverwell, F., Wilks, A., Evans, D., Kay, M., & Baker, D.
(1999). Associations between reading ability and visual processes. In J. Everatt (Ed.), Reading
and dyslexia: Visual and attentional processes (pp. 139). London: Routledge.
Girelli, L., Lucangeli, D., & Butterworth, B. (2000). The development of automaticity in accessing
number magnitude. Journal of Experimental Child Psychology, 76(2), 104122.
Glaser, W. R., & Glaser, M. O. (1989). Context effects in Stroop-like word and picture processing.
Journal of Experimental Psychology: General, 118, 1342.
Guttentag, R. E., & Haith, M. M. (1978). Automatic processing as a function of age and reading
ability. Child Development, 49, 707716.
Hagenaar, R., & Van der Heijden, A. H. (1986). Target-noise separation in visual selective attention.
Acta Psychologica, 62(2), 161176.
Jerger, S., Stout, G., Kent, M., Albritton, E., Loiselle, L., Blondeau, J., & Jorgenson, S. (1993).
Auditory Stroop effects in children with hearing impairment. Journal of Speech & Hearing
Research, 36, 1083 1096.
 Adults versus childrens Stroop performance 485
Lee, W. L., Wee, G. C., Tzeng, O. J. L., & Hung, D. L. (1992). A study of interlingual and intralingual
Stroop effect in three different scripts: Logograph, syllabary, and alphabet. Cognitive
Processing in Bilinguals, 427442.
Logan, G. D. (1980). Attention and automaticity in Stroop and priming tasks: Theory and data.
Cognitive Psychology, 11, 523553.
Long, G. M., & Lyman, B. J. (1987). Foveal and parafoveal processing of asynchronous Stroop
stimuli. British Journal of Psychology, 78(2), 151162.
MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review.
Psychological Bulletin, 109, 163203.
MacLeod, C. M., & Dunbar, K. (1988). Training and Stroop-like interference: Evidence for a
continuum of automaticity. Journal of Experimental Psychology: Learning, Memory and
Cognition, 14, 126135.
MacLeod, C. M., & MacDonald, P. A. (2000). Interdimensional interference in the Stroop effect:
Uncovering the cognitive and neural anatomy of attention. Trends in Cognitive Sciences,
4(10), 383391.
Marcel, A. J. (1983). Conscious and unconscious perception: Experiments on visual masking and
word recognition. Cognitive Psychology, 15, 197237.
Merikle, P. M., & Gorewich, N. J. (1979). Spatial selectivity in vision: Field size depends upon noise
size. Bulletin of the Psychonomic Society, 14, 343346.
Merrill, E. C., Sperber, R. D., & McCauley, C. (1981). Differences in semantic encoding as a
function of reading comprehension skill. Memory & Cognition, 9, 618624.
Milliken, B., & Tipper, S. P. (1998). Attention and inhibition. In H. Pashler (Ed.), Attention
(pp. 191222). Hove, Sussex: Psychology Press.
Posner, M. L., & Rothbart, M. K. (1992). Attentional mechanisms and conscious experience. In
A. D. Milner & M. D. Rugg (Eds.), The neuropsychology of consciousness (pp. 9111). San
Diego, CA: Academic Press.
Posner, M. L., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.),
Information processing and cognition: The Loyola symposium (pp. 5585). Hillsdale, NJ:
Erlbaum.
Rayner, K., & Posnansky, C. (1978). Stages of processing in word identication. Journal of
Experimental Psychology, General, 107(1), 6480.
Rieger, M., & Gauggel, S. (1999). Inhibitory after-effects in the stop signal paradigm. British
Journal of Psychology, 90, 509518.
Schadler, M., & Thissen, D. M. (1981). The development of automatic word recognition and
reading skill. Memory and Cognition, 9(2), 132141.
Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing.
II. Perceptual learning, automatic attending, and a general theory. Psychological Review,
84(2), 127163.
Stroop, J. P. (1935). Studies of interference in serial verbal reactions. Journal of Experimental
Psychology, 18, 643662.
Sugg, M. J., & McDonald, J. M. (1994). Time course of inhibition in colour-response and word-
response versions of the Stroop task. Journal of Experimental Psychology, 20, 647675.
Tzelgov, J., Henik, A., & Berger, J. (1992). Controlling Stroop effects by manipulating expectations
for color words. Memory and Cognition, 20, 727735.
Van Selst, M., & Jolicoeur, P. (1997). Decision and response in dual-task interference. Cognitive
Psychology, 33, 266307.
Visser, M., Das-Smaal, E., & Kwakman, H. (1996). Impulsivity and negative priming: Evidence for
diminished cognitive inhibition in impulsive children. British Journal of Psychology, 87(1),
131140.
Vurpillot, E., & Ball, W. A. (1979). The concept of identity and childrens selective attention. In
G. A. Hale & M. Lewis (Eds.), Attention and cognitive development (pp. 2342). New York:
Plenum.
Received 21 September 2001; revised version received 10 January 2003