Académique Documents
Professionnel Documents
Culture Documents
a
University Sapienza, Rome, Italy
b
University of North Carolina, Chapel Hill, USA
Published online: 30 Mar 2009.
To cite this article: Pietro Spataro , Neil Mulligan & Clelia Rossi-Arnaud (2010)
Effects of divided attention in the word-fragment completion task with unique and
multiple solutions, European Journal of Cognitive Psychology, 22:1, 18-45, DOI:
10.1080/09541440802685979
Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed
in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan, sub-
licensing, systematic supply, or distribution in any form to anyone is expressly
forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
EUROPEAN JOURNAL OF COGNITIVE PSYCHOLOGY
2010, 22 (1), 1845
Neil Mulligan
University of North Carolina, Chapel Hill, USA
Clelia Rossi-Arnaud
University Sapienza, Rome, Italy
Over the last 30 years, a great deal of research has been devoted to the study
of implicit memory (Mulligan & Brown, 2003; Roediger & McDermott, 1993;
1999; Prull, 2004; Vaidya et al., 1997). The present experiments were designed
to test these accounts of implicit memory by investigating the role of divided
attention at encoding on the word fragment completion task with unique and
multiple solutions.
The Transfer-Appropriate-Processing framework focuses on the type of
processing accomplished on the stimuli. The TAP theory predicts that the
efficiency of memory performance will depend on the extent to which
encoding processes are reengaged at the time of retrieval, and makes a
distinction between two classes of cognitive processes: perceptual processes,
which rely on the analysis of surface-level features, and conceptual processes,
based on the analysis of meaning. Many implicit tasks can be classified as
perceptual, because they involve the identification or completion of
degraded cues, like word fragments or incomplete pictures. In contrast,
implicit tests like semantic association or the generation of category
exemplars can be conceived as primarily conceptual, because they require
the retrieval of semantic information. The TAP view argues that dissocia-
tions are to be expected between perceptual and conceptual implicit tests.
Substantial support for this distinction has been found, with conceptual
priming generally sensitive to manipulations of semantic encoding and
insensitive to perceptual study manipulations, and perceptual priming
typically exhibiting the opposite pattern (for review, see Mulligan, 2003;
Roediger & McDermott, 1993).
Despite its substantial success, the perceptualconceptual distinction has
been recently challenged by another distinction, that between identification
and production priming (Gabrieli et al., 1999). Identification tasks (e.g.,
perceptual identification, lexical decision, etc.) involve search processes that
converge on a single appropriate response or representation in memory
(Prull, 2004): Participants are required to identify target items, verify their
attributes, or classify them into semantic categories. Identification priming
can involve entire stimuli presented normally (such as in tasks of word
naming), or degraded or masked forms of the stimuli (such as in word/
picture fragment completion or in masked identification): in all of these
cases, participants attempt to identify the cue or some feature of it
(Fleischman & Gabrieli, 1998; Gabrieli et al., 1999). In contrast, production
20 SPATARO, MULLIGAN, ROSSI-ARNAUD
correct. If the target word has been previously primed, the outcome of the
lexical search will always be the same representation that was activated in the
study phase. On the contrary, when the fragments are constructed to allow
multiple solutions, the WFC task is thought to become a production task
(Nyberg, Nilsson, & Olofsson, 1994; Olofsson, 1995). This is because in that
condition the correct identification of the pattern of letters and missing
spaces will not guide participants to unique responses; instead, the cue will
match with a large number of lexical entries, any one of which could be used
to make a plausible response. This production stage should be necessarily
followed by a selection process that chooses one single entry: In this
framework, priming is hypothesised to increase the chance that the studied
word will be selected from among the generated alternatives. The same
distinction has been claimed by Barnhardt (2004, 2005) in relation to the
word-stem completion task (WSC). The typical WSC task uses stems that can
be completed with at least 10 words different from the target one, making it a
clear example of production test; however, when the stems are selected to have
only one or few solutions, then the test can be better described as an
identification one.
It should be noted that the identificationproduction distinction, as well as
that between perceptual and conceptual priming, is not absolute nor
unambiguous. Both Fleischman and Gabrieli (1998) and Prull (2004)
outlined that virtually no task is process pure: Depending on experimental
conditions and stimulus materials, any given implicit test can be characterised
as being the result of a mixture of processes. However, the critical point here is
that the latter claim does not prevent us from defining implicit tasks
according to predominant processes. For example, Fleischman and Gabrieli
(1998) suggested that, although both tests require the production of a word,
the WFC with unique solutions may depend relatively more than WSC on the
identification of pattern of letters than on productive processes. In a similar
vein, we propose that identification processes are predominant in the
WFC with unique solutions, whereas production/selection processes are
more critical in the WFC with multiple solutions. That is, if one sees
the identification-production distinction as endpoints of a continuum,
22 SPATARO, MULLIGAN, ROSSI-ARNAUD
Weldon, Stadler, & Riegler, 1992). The expectancy that perceptual implicit
tasks like the WFC should be unaffected by decreases of attentional resources
at encoding has been confirmed in four previous experiments that used
unique-solution fragments (Clarys et al., 2000; Mulligan, 1998; Mulligan &
Hartman, 1996; Parkin et al., 1990).
It should be outlined that the design of the present experiments has the
virtue of contrasting production and identification priming within a common
task. Prior studies examined identification and production priming across
different types of tasks (e.g., category verification vs. category generation),
which typically varied on a number of dimensions in addition to the critical
dimension of number of legitimate responses (e.g., type of response, type of
dependent measure, etc.). In the present case, the use of multiple-solution
versus unique fragments allows us to isolate the effects of retrieval
competition while holding constant the type of response, type of dependent
measure, general nature of retrieval cues, etc.
EXPERIMENT 1
A short-term memory load paradigm was used to divide attention during the
encoding phase (e.g., Clarys et al., 2000; Mulligan, 1998). Up to date, the
consequences of DA on the WFC task have always been studied using
fragments with unique solutions (e.g., Mulligan & Hartman, 1996; Parkin
et al., 1990), thus making the WFC an identification task. On the contrary, in
the present experiment fragments were constructed to have either one single
appropriate response or multiple possible solutions: In the latter case WFC is
thought to become a production task.
The manipulation of solution set size has been used in other studies to
assess the effects of test delay and number of presentations of the target word
at encoding (Nyberg et al., 1994; Olofsson, 1995). Surprisingly, results
showed little differences between fragments with unique and multiple
solutions: for example, Olofsson (1995) reported that an increase of the
retention interval up to 24 hours did not produce differential forgetting for
the two types of fragments.
24 SPATARO, MULLIGAN, ROSSI-ARNAUD
attention and fragment type should provide strong support for the soundness
of the distinction between identification and production priming.
An attentional load of seven was used to assure that the strength of the
manipulation was sufficient to detect subtle effects. The degree of difficulty of
the distractor task can be an important determinant of DA effects in implicit
memory (e.g., Mulligan, 1997; Wolters & Prinsen, 1997). In particular,
Mulligan (1997) applied the short-term memory load paradigm to implicit
and explicit versions of the category-exemplar production task. The
attentional load varied between 0, 1, 3, and 5 items: Results showed that a
mild division of attention (loads of three elements) reduced category-cued
recall but not conceptual priming, whereas a stronger division of attention
(loads of five elements) diminished performance on both tasks and entirely
eliminated conceptual priming. Thus, explicit and implicit memory tests
differ in their sensitivity to the consequences of DA: Comparatively stronger
manipulations are needed to impair implicit rather than explicit tasks (see
also Berry, Henson, & Shanks, 2006; MacDonald & MacLeod, 1998). On the
other hand, it seems unreasonable to employ sequences with more than seven
elements, as participants have trouble retaining more items, yielding no
functional increase in the loads (Mulligan, 1998).
Method
Participants. Thirty undergraduate students (16 females and 14 males)
of the University ‘‘Sapienza’’ of Rome participated voluntarily (mean age
24.8 years; mean education16.8 years).
placed at the beginning and at the end of every study list, as primacy and
recency buffers. Mean overall frequencies of the three master lists were
respectively: 102.2, 101.5, and 102.3 occurrences per million according to the
CoLFIS Vocabulary (Laudanna, Thornton, Brown, Burani, & Marconi,
1995).
Two master lists were assigned to FA (0-load) and DA (7-load) conditions;
the words in the third list were used as baseline items during the test phase. Thus,
all study lists contained 52 words (four buffers and 48 targets). Counter-
balancing was applied so that all master lists were rotated through the three
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
different roles (FA, DA, and baseline) an equal number of times. With the
exception of buffer items, all study trials were randomised for all participants,
producing a mixed list with regard to the attentional manipulation.
The memory test consisted of 80 word fragments. Fragments with
multiple solutions could be completed with a mean of 4.7 words (from a
minimum of 3 to a maximum of 7). All fragments consisted of 3 letters
(when the words were 6 letters in length) or 4 letters (when the words were 7
or 8 letters in length), with underscores in the place of missing letters. To
reduce unwanted variability in the construction of the fragments, three
different fragment patterns were allowed for each length of the solution word
(see Table 1). Eight fragments were used as practice items at the beginning of
the task; for the remaining cues, 48 fragments corresponded to the studied
words (respectively, 24 unique and 24 multiple fragments) and 24 corre-
sponded to new words (again, 12 unique and 12 multiple fragments).
Presentation of the fragments was randomly ordered by the software
SuperLab (Version 4.0). Fragments and words were written with the font
Times New Roman, 72 point, bold.
TABLE 1
Method of fragment construction for each word length
Word length Fragment Example
(2000). Each study trial was composed of four events: (1) First, a fixation
point (a cross in the middle of the computer screen) was displayed for
500 ms; (2) next, the digit-letter sequence (for the 7-load condition) or a string
of seven Xs (for the 0-load condition) was presented for 3000 ms: for the
7-load trials participants were told to read the string aloud and retain it in
memory until the recall signal was shown, whereas in the 0-load trials they
had only to answer ‘‘PASS’’; (3) then, the study word was presented for
2500 ms: instructions required to read the word aloud but no reference was
made of a subsequent memory test; (4) finally, the recall signal (three
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
that the manipulation of the number of solutions achieved its primary aim:
When target words were unprimed, the probability to be selected among
multiple competitors was greatly reduced.
The same conclusion can be gained trough the analysis of the raw
percentages of correct completions for studied words: A repeated ANOVA
with fragment type as within-subjects factor showed that the main effect of
fragment type was significant, F(1, 29)64.887, MSE0.410. Again the
proportion of correct completions was higher for unique than multiple
fragments (M0.343 vs. M0.181). Thus, the manipulation of the number
of solutions of the fragments had strong effects both on unstudied and
studied items.
For each type of fragment (unique vs. multiple solution), the proportion of
correct completions with unstudied words was subtracted from the proportion
of correct completions with studied words, thus yielding priming values which
are shown in Table 2. A repeated 22 ANOVA with fragment type and
attentional load (0-load vs. 7-load) as within-subjects factors revealed: (1) no
effects of fragment type or attentional load: Priming levels were roughly equal
for unique and multiple fragments (M0.115 vs. M0.108), F(1, 29)B1, and
equal in the FA and DA conditions (M0.117 vs. M0.107), F(1, 29)B1; (2)
no interaction between divided attention and fragment type, F(1, 29)B1. The
same pattern of results held when the data from the six subjects claiming
intentional retrieval from the study phase were removed from the analyses,
F(1, 23)B1 in all cases.
When using the short-term memory load paradigm, it is important to
ascertain if performance in the secondary task was adequate to ensure a
sufficiently strong DA manipulation (Mulligan, 1997, 1998; Wolters & Prinsen,
1997). The mean value of recalled elements in the 7-load condition was
5.47 (SD0.89; minimum3.33, maximum6.67). When the analysis was
limited to the participants whose scores were above the median (Mn5.63;
N15), the effect of attentional load was still nonsignificant, F(1, 14)B1.
Given the low number of unaware participants, no statistical analysis
was performed on this variable; however, it seems important to note that
overall priming was greater for aware (M0.125) than unaware participants
(M0.083).
28 SPATARO, MULLIGAN, ROSSI-ARNAUD
TABLE 2
Percentages and priming scores on the WFC task in Experiment 1 as a function of
attentional condition and fragment type
Attentional condition
Unique solution
Percentage studied 0.344 0.341
Percentage nonstudied 0.227 0.227
Priming 0.117 0.114
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
Multiple solution
Percentage studied 0.189 0.172
Percentage nonstudied 0.072 0.072
Priming 0.117 0.100
the present experiment, we do not think that explicit retrieval has unduly
contaminated the results for several reasons. First, only a few participants
claimed intentional retrieval. Second, we did not require engagement in deep
or intentional encoding at study. Finally, data from the present experiment
suggest that test-awareness does not mediate the effect of DA: Indeed, if
awareness was necessary to obtain a reduction of the WFC performance,
then we should have found a significant effect in the first experiment,
because almost all participants stated an awareness of the relationship
between study and test phases. Thus, it appears that test-awareness was not
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
EXPERIMENT 2
Although the data of Experiment 1 are relevant to the identification
production framework, it should be noted that they provide only negative
evidence: The lack of differences between the two types of fragments might
be partially explained by the general absence of effect of our attentional
manipulation. A more complete analysis requires some positive proof: In
particular, a stronger disconfirmation would come from the demonstration
that multiple- and unique-solution fragments lead to the same amount of
impairment in a condition in which DA significantly reduces priming levels.
To have the maximum possibility of obtaining reliable effects of the atten-
tional manipulation, two factors were modified in Experiment 2: the ex-
posure time of the target words and the type of presentation of the attention
conditions during the encoding phase (mixed vs. blocked).
The importance of the interaction between study time and divided
attention has been highlighted by Ganor-Stern, Seamon, and Carrasco
(1998): Using the possibleimpossible object decision paradigm, they
showed that a DA manipulation (involving a flanking digits procedure)
significantly impaired perceptual priming only when the encoding time of
the drawings was reduced from 5 to 3 s.
In relation to the type of experimental design, compelling data has been
obtained both with implicit and explicit memory. Analysing the levels-of-
processing effects on the WFC task, Challis and Brodbeck (1992) and
Thapar and Greene (1994) demonstrated that a semantic encoding condition
produced more priming relative to a physical condition when the presenta-
tion of the two types of trials was blocked, but not when they were randomly
intermixed in the same list. Among the possible explanations, the authors
proposed that participants might truncate lexical processing in the blocked
list; that is, when participants are repeatedly requested to answer questions
about the presence/absence of single letters, they might stop processing the
30 SPATARO, MULLIGAN, ROSSI-ARNAUD
words as lexical units. Powerful effects of the type of design were also
obtained by Begg and colleagues (Begg & Roe, 1988; Begg & Snider, 1987) in
the context of a read versus generate manipulation: They found an
advantage of the generate condition on a recognition test only when the
encoding task was manipulated within rather than between subjects. They
concluded that, when the two tasks were mixed in the same list, the
generation of some stimuli inhibited lexical processing of read words,
making participants ‘‘lazy readers’’.
To verify the importance of these two factors, Experiment 1 was
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
replicated with two major changes: During the encoding phase, words
were shown for 1 s (instead of 2.5 s; for a similar procedure see Gabrieli
et al., 1999), and the presentation of the two attentional conditions
(07 loads) was blocked. In addition, a recognition test followed the indirect
WFC task. To reduce the number of aware subjects, we adopted some of the
expedients recommended by Bowers and Schacter (1990): During the
encoding phase, target words for the WFC were intermixed with other
words whose memory was verified only in the subsequent recognition test;
similarly, at the time of retrieval, fragments corresponding to the studied
words were intermixed with a higher number of filler fragments (in this
second experiment the ratio of target to filler fragments was equal to 2:3,
whereas in the first experiment it was 2:1).
The selection of the words from Experiment 1 was aimed at reducing as far
as possible the difference between the baseline percentages of completion for
unique- and multiple-solution fragments. The results of Experiment 1 showed
that priming was equal for the two types of fragments; however, Vaidya et al.
(1997) suggested that more elaboration at encoding is necessary to obtain full
priming on implicit production tasks. Thus, one would expect an asymmetry
in the effects of study: A single brief encounter with the target words should
enhance priming more for unique than multiple-solution fragments (that is,
more for the identification than the production version of the WFC). The
failure to confirm this expectancy in Experiment 1 could be accounted for by
the high difference in the baseline levels of completion: Removing this
discrepancy will allow us to address this issue more thoroughly.
The main theoretical predictions were the same as in Experiment 1.
Support for the identificationproduction hypothesis would be again
provided by the achievement of a reliable interaction between fragment
type and divided attention: A significant DA effect was expected with
multiple-solutions fragments, but not with unique-solutions fragments. On
the contrary, according to the TAP theory, the two modifications should not
change the resilience of the WFC task, because lexical access continued to be
warranted by the overt articulation of the target words in the encoding
phase.
DIVIDED ATTENTION AND FRAGMENT COMPLETION 31
Method
Participants. Thirty-two undergraduate students (18 females and 14
males) of the University ‘‘Sapienza’’ of Rome participated voluntarily (mean
age24.2 years; mean education16.8 years).
Design and materials. A 22 design was again used, with fragment type
and attentional load as within-subjects variables. Study materials consisted of
48 words used in Experiment 1. All words were from 6 to 8 letters in length,
and were divided in four sublists of 12 items. Within each sublist, half of the
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
words were associated with unique fragments; the other half were associated
with nonunique fragments. Thirty pilot participants produced baseline
completion proportions of between 0.180 and 0.200 for the four lists.
Frequencies of the sublists were comparable to those in Experiment 1,
ranging between 105.1 and 110.5 occurrences per million according to the
CoLFIS vocabulary (Laudanna et al., 1995). Sixteen additional words were
selected: Two words were placed at the beginning and at the end of each study
list, as primacy and recency buffers; the remaining 12 words were presented
during the encoding phase along with the target words for the WFC (equally
divided between the 07 load conditions), but were tested only in the
successive recognition task.
The four sublists were counterbalanced across the attentional conditions,
so that they were assigned to the 0-load and 7-load conditions an equal
number of times. Thus, a complete study list contained a total of 40 words:
four buffers, 24 words to be tested on both the WFC and recognition tasks,
and 12 words tested only on the recognition test. Presentation of the
attentional conditions was blocked, and the order of the blocks was
counterbalanced across participants.
The WFC task consisted of 60 word fragments: 24 fragments corre-
sponded to the studied words (12 unique and 12 multiple); the other
36 corresponded to new words. Filler fragments were always constructed to
have unique solutions and were not scored. To further disguise the real
aim of the experiment, the first 15 fragments at the beginning of the test
could be completed only with new words. Multiple fragments were the same
employed in Experiment 1: They could be completed with a mean of 4.2
words (range37). Finally, the recognition test consisted of 72 words: 36
words corresponded to studied items, while the other 36 words corresponded
to new items (not presented in the WFC test).
Procedure. Procedures for the encoding phase and the WFC task were
similar to those used in Experiment 1, except that words were presented for
1000 ms instead of 2500 ms. In addition, one half of the participants were
presented first with all 0-load trials and then with all 7-load trials; the other
32 SPATARO, MULLIGAN, ROSSI-ARNAUD
half performed the study phase in the reverse order. In the 7-load condition
participants were again requested to read the words aloud; however, the
experimenter stated that words had no role in this phase, and were presented
only to distract them from string recall.
Following the WFC and the posttest questionnaire, participants were
involved in casual conversations for 5 min. Lastly, the recognition task was
explained. Participants were given two sheets of paper that contained
72 words arranged in random order; they were instructed to circle only the
words studied during the encoding phase.
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
TABLE 3
Percentages and priming scores on the WFC task in Experiment 2 as a function of
attentional condition and fragment type
Attentional condition
Unique solution
Percentage studied 0.354 0.260
Percentage nonstudied 0.199 0.199
Priming 0.155 0.061
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
Multiple solution
Percentage studied 0.266 0.187
Percentage nonstudied 0.167 0.167
Priming 0.099 0.020
proportion of hits was higher for the words studied in the FA (M0.491)
than in the DA condition (M0.338)], thus replicating the impairment
usually found with explicit memory measures.
In summary, Experiment 2 demonstrated that, at least in particular
experimental conditions, DA at encoding can reduce priming on the indirect
task of word fragment completion. This result contradicts a strong version of
the TAP theory, but is congruent with other recent research showing that
perceptual implicit memory is not completely invulnerable to attentional
manipulations (Crabb & Dark, 1999, 2003; Rajaram, Srinivas, & Travers,
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
2001).
However, even in this second experiment the extent of DA impairment
was equivalent for unique- and multiple-solution fragments: Indeed, the
difference between priming levels in the FA and DA conditions was greater
for unique-solution fragments (0.094 vs. 0.079). This failure, coupled with
the findings of Experiment 1, provides additional evidence against the
hypothesis that production priming, relative to identification priming,
requires greater amounts of attention at encoding: Importantly, this
conclusion holds both when the effect of divided attention on unique-
solution fragments (e.g., on identification priming) was not -significant
(Experiment 1), but also when it was substantial (Experiment 2).
Contrary to Experiment 1, the effect of the manipulation of the number
of fragment solutions on priming approached significance in the major
analysis, and was fully significant when the sample was restricted to
unintentional participants. This result is congruent with the idea that a
brief study episode will have greater consequences on unique- rather than on
multiple-solution fragments, provided that initial baseline levels are equated.
At this point some speculations are in order. First, it should be outlined
that, according to some authors (Weldon, 1991, 1993), the WFC task is quite
sensitive to conceptually driven processing, especially when the fragments
are exposed for long periods of time, as in the present experiments (12 s).
Weldon, for example, suggested that the WFC task is affected both by
perceptual and conceptual information, but the first is recruited more
quickly. When its mixed nature is taken into account, the present findings
could be explained by supposing that the DA manipulation affected the
conceptual component of the WFC. However, this interpretation is partially
undermined by two problems. First, the vast majority of the studies argue
against the conceptual nature of the WFC task (see the introduction). For
example, MacLeod and Masson (1997) found that the generateread
manipulation produced different patterns of results on two perceptual
implicit tests: Read words resulted in more priming than generated words in
the WFC, whereas equivalent priming was found in the word identification
task. The authors concluded that word identification involved an initial
interpretive encoding that included records of conceptual as well as
DIVIDED ATTENTION AND FRAGMENT COMPLETION 35
perceptual analyses, whereas the WFC was based solely on the results of
perceptual operations. Similarly, Blum and Yonelinas (2001; see also
Mitchell & Bruss, 2003) reported that unintentional participants demon-
strated cross-modality priming in the stem completion and the word
identification tasks, but not in the WFC: They suggested that the visual
word-fragment completion test had unique properties, because it relied only
on visual (perceptual) memory processes, whereas the other tasks relied also
on phonological or semantic processes. Furthermore, if the WFC is partly
conceptual, why did previous researches fail to find a DA effect? As
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
GENERAL DISCUSSION
Two experiments investigated the effects of divided attention on identifica-
tion and production versions of the word fragment completion task.
Multiple-solution fragments were used to test the hypothesis that production
priming (unlike the standard identification version of the WFC, with
unique-solution fragments) should imply a greater involvement of cognitive
resources during encoding (Gabrieli et al., 1999). In the first experiment,
target words were studied for 2.5 s and the presentation of attentional
conditions was randomly mixed; in the second experiment study time was
reduced to 1 s and presentation of the FA/DA conditions was blocked. The
latter modifications were done in order to maximise the probability of
36 SPATARO, MULLIGAN, ROSSI-ARNAUD
Divided attention
Although the analysis of the factors that mediate the effect of divided
attention on the WFC task was not the primary aim of the present research,
the findings of Experiment 2, being in contrast with previous conclusions of
other authors (Clarys et al., 2000; Mulligan, 1998; Mulligan & Hartman,
1996; Parkin et al., 1990), demand some comments.
First, we note that, despite this apparent incongruence, our current results
are in line with other recent research showing that implicit perceptual
memory is not completely immune to attentional manipulations. The WFC
task itself was found to be impaired by a selective-attention paradigm
(Rajaram et al., 2001; Stone et al., 1998): for example, Rajaram and
colleagues (2001) reported that the request to direct attention towards an
attribute different from the word identity, as the print colour, was sufficient
to significantly reduce the magnitude of completion priming, when compared
DIVIDED ATTENTION AND FRAGMENT COMPLETION 39
impaired by a modified version of the Stroop task, in which the attention was
manipulated across distinct objects rather than within the same words (e.g.,
participants were requested to do colour judgements about Xs flanking the
words). All these results point to the conclusion that perceptual implicit
memory depends on attentional requirements beyond those needed for the
simple identification of the words.
Furthermore, since we asked participants to read the words during the
encoding phase, our results imply that the correct lexical access to the words
is not sufficient to ensure a complete immunity against attentional
manipulations. However, even this result can be made congruent with the
current literature. Mulligan and Hornstein (2000) had their participants
studying words printed in different colours: The instructions were to read the
word (full attention condition), to identify the colour (colour-name
condition), or to do both tasks (both condition). The results showed that
performance in the perceptual identification task was significantly disrupted
by the requirement to respond to the words’ colour: Critically, the
impairment was reliable even in the both condition, despite the instruction
to read aloud the words. Thus, in those experiments lexical access was not
sufficient to equate priming to the levels reached in the read-only condition.
In summary, although the results of Experiment 2 are apparently in contrast
with a strong version of the TAP theory (because the WFC is thought to be a
data-driven task), they still agree with much literature in demonstrating that:
(1) Perceptual implicit memory relies on attention at encoding, even if the
extent of this reliance is lower when compared with explicit or conceptual
implicit memory; (2) lexical access is not sufficient to prevent the effects of
divided attention.
As concerns the mechanisms that produced the DA impairment, it is
possible that in the blocked DA condition, the continuous repetition of the
secondary task (e.g., the accurate recall of the strings of letters and digits),
coupled with the short study time of target words and the type of
instructions (which discouraged extensive elaborations of the words),
induced participants to treat it as the primary concern, thus diverting
attentional resources from the lexical processing of words to string retrieval.
In this condition, reading the words (and thus accessing the lexical
40 SPATARO, MULLIGAN, ROSSI-ARNAUD
representations) did not ensure the same level of lexical processing that was
achieved in the full attention condition. Using Begg and Roe’s (1988); see
also Begg & Snider, 1987; Begg, Snider, Foley, & Goddard, 1989) definitions,
in the blocked DA condition participants became ‘‘lazy readers’’: The
repeated execution of the short-term memory task led to a shallow
elaboration of the words, probably aimed only at their identification,
whereas in the full attention condition the extent of lexical processing was
more accurate. This strategy was not followed in the mixed list, because the
constant alternation of the 07 load conditions oriented participants to
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
perform the same standard level of lexical elaboration on all the target
words. As suggested by Richardson-Klavehn and Gardiner (1998), attention
to the study stimuli as lexical units is critical for repetition priming, and is
required in addition to lexical access, which can occur automatically. Our
explanation fits well with this framework, because it implies that in the
blocked DA condition lexical access was present, but lexical processing (that
is, the words’ elaboration as lexical units) was very superficial. The account
is also congruent with the idea that the blocked condition allows for a more
rigorous control of both processing strategies and allocation of attention
(Bentin, Moscovitch, & Nirhod, 1998).
In conclusion, by proposing a lexical deficit in the DA blocked condition,
our hypothesis complies with other evidence showing that lexical processing
has a critical importance for repetition priming: for example, Weldon (1991)
demonstrated that the study of anagrams produced significant priming in
fragment completion and perceptual identification only when participants
used a simple rule to turn the anagrams into words. In the same vein,
Nelson, Keelean, and Negrao (1989) found that instructions to search at the
word level enhanced WFC performance, whereas instructions to generate
letters to fill missing spaces had no effect. These and other data sustain the
idea that lexical processing has a primary role in perceptual implicit tasks.
SUMMARY
Two experiments studied the difference between identification and produc-
tion priming using the word-fragment completion task with unique
(identification version) or multiple-solution (production version) fragments.
The purpose was to test the hypothesis that production priming, given the
presence of competition, should require greater amounts of attentional
resources during the encoding of target words (Gabrieli et al., 1999).
According to this framework, when the fragments are constructed to allow a
competition between target words and alternative solutions, the biased
selection of primed stimuli will occur only if their encoding has been
sufficiently elaborated; this, in turn, should imply more attention-demanding
DIVIDED ATTENTION AND FRAGMENT COMPLETION 41
adults.
REFERENCES
Baques, J., Saiz, D., & Bowers, J. S. (2004). Effects of working memory load on long-term word
priming. Memory, 12(3), 301313.
Barnhardt, T. M. (2004). Different involuntary mechanisms underlie priming and LOP effects in
stem completion tests. Memory, 12, 614636.
Barnhardt, T. M. (2005). Number of solutions effects in stem decision: Support for the distinction
between identification and production processes in priming. Memory, 13(7), 725748.
Barnhardt, T. M., & Geraci, L. (2008). Are awareness questionnaires valid? Investigating the use of
post-test questionnaires for assessing awareness in implicit memory tests. Memory and
Cognition, 36(1), 5364.
Begg, I., & Roe, H. (1988). On the inhibition of reading by generating. Canadian Journal of
Experimental Psychology, 42(3), 325336.
Begg, I., & Snider, A. (1987). The generation effect: Evidence for generalized inhibition. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 13(4), 553563.
Begg, I., Snider, A., Foley, F., & Goddard, R. (1989). The generation effect is no artifact:
Generating makes words distinctive. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 15(5), 977989.
Bentin, S., Moscovitch, M., & Nirhod, O. (1998). Levels of processing and selective attention
effects on encoding in memory. Acta Psychologica, 98, 311341.
Berry, C. J., Henson, R. N. A., & Shanks, D. R. (2006). On the relationship between repetition
priming and recognition memory: Insights from a computational model. Journal of Memory
and Language, 55(4), 515533.
Blum, D., & Yonelinas, A. P. (2001). Transfer across modality in perceptual implicit memory.
Psychonomic Bulletin and Review, 8(1), 147154.
Bowers, J. S., & Schacter, D. L. (1990). Implicit memory and test awareness. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 16, 404416.
Brown, A. S., Jones, T. C., & Mitchell, D. B. (1996). Single and multiple test repetition priming in
implicit memory. Memory, 4, 159173.
Challis, B. H., & Brodbeck, D. R. (1992). Level of processing affects priming in word
fragment completion. Journal of Experimental Psychology: Learning, Memory, and Cognition,
18, 595607.
42 SPATARO, MULLIGAN, ROSSI-ARNAUD
Clarys, D., Isingrini, M., & Haerty, A. (2000). Effects of attentional load and ageing on word-stem
and word fragment implicit memory tasks. European Journal of Cognitive Psychology, 12(3),
395412.
Crabb, B. T., & Dark, V. J. (1999). Perceptual implicit memory requires attentional encoding.
Memory and Cognition, 27(2), 267275.
Crabb, B. T., & Dark, V. J. (2003). Perceptual implicit memory relies on intentional, load-sensitive
processing at encoding. Memory and Cognition, 31(7), 9971008.
Craik, F., Moscovitch, M., & McDowd, J. (1994). Contributions of surface and conceptual
information to performance on implicit and explicit tasks. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 20(4), 864875.
Fleischman, D. A., & Gabrieli, J. D. E. (1998). Repetition priming in normal aging and Alzheimer’s
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013
disease: A review of findings and theories. Psychology and Aging, 13(1), 88119.
Fleischman, D. A., Monti, L. A., Dwornik, L. M., Moro, T. T., Bennett, D. A., & Gabrieli, J. D. E.
(2001). Intact identification priming and impaired production priming and in Alzheimer’s
disease. Journal of the International Neuropsychological Society, 7(7), 785794.
Gabrieli, J. D., Vaidya, C. J., Stone, M., Francis, W. S., Thompson-Schill, S. L., Fleischman, D. A.,
et al. (1999). Convergent behavioral and neuropsychological evidence for a distinction between
identification and production forms of repetition priming. Journal of Experimental Psychology:
General, 128(4), 479498.
Ganor-Stern, D., Seamon, J. G., & Carrasco, M. (1998). The role of attention and study time
in explicit and implicit memory for unfamiliar visual stimuli. Memory and Cognition, 26(6),
11871195.
Geraci, L. (2006). A test of the frontal lobe functioning hypothesis of age deficits in production
priming. Neuropsychology, 20(5), 539548.
Horton, K. D., Wilson, D. E., Vonk, J., Kirby, S. L., & Nielsen, T. (2005). Measuring automatic
retrieval: A comparison of implicit memory, process dissociation, and speeded response
procedures. Acta Psychologica, 119(3), 235263.
Isingrini, M., Vazou, F., & Leroy, P. (1995). Dissociation of implicit and explicit memory tests:
Effect of age and divided attention on category exemplar generation and cued recall. Memory
and Cognition, 23, 462467.
Kinoshita, S. (2001). The role of involuntary aware memory in the implicit stem and fragment
completion tasks: A selective review. Psychonomic Bulletin and Review, 8(1), 5869.
Laudanna, A., Thornton, A. M., Brown, G., Burani, C., & Marconi, L. (1995). Un corpus
dell’italiano scritto contemporaneo dalla parte del ricevente. In S. Bolasco, L. Lebart, & A.
Salem (Eds.), III Giornate internazionali di Analisi Statistica dei Dati Testuali (pp. 103109).
Rome, Italy: Cisu.
Li, L., Miller, E. K., & Desimone, R. (1993). The representation of stimulus familiarity in anterior
inferior temporal cortex. Journal of Neurophysiology, 69, 19181929.
Light, L. L., & Prull, M. W. (1995). Aging, divided attention, and repetition priming. Swiss Journal
of Psychology, 54, 87101.
Light, L. L., Prull, M. W., & Kennison, R. F. (2000). Divided attention, aging, and priming in
exemplar generation and category verification. Memory and Cognition, 28, 856872.
MacDonald, P. A., & MacLeod, C. M. (1998). The influence of attention at encoding on direct and
indirect remembering. Acta Psychologica, 98(23), 291310.
MacLeod, C. M., & Masson, M. E. J. (1997). Priming patterns are different in masked
word identification and word fragment completion. Journal of Memory and Language, 36(4),
461483.
Mace, J. L. (2003a). Involuntary aware memory enhances priming on a conceptual implicit
memory task. American Journal of Psychology, 116(2), 281290.
Mace, J. L. (2003b). Study-test awareness can enhance priming on an implicit memory task:
Evidence from a word completion task. American Journal of Psychology, 116(2), 257279.
DIVIDED ATTENTION AND FRAGMENT COMPLETION 43
divided attention and number of visual objects. Psychological Research, 66(3), 157165.
Mulligan, N. W. (2003). Effects of cross-modal and intramodal division of attention on perceptual
implicit memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29(2),
262276.
Mulligan, N. W., & Brown, A. S. (2003). Attention and implicit memory. In L. Jiménez (Ed.),
Attention and implicit learning: Advances in consciousness research (pp. 297334). Amsterdam:
John Benjamins.
Mulligan, N. W., Duke, M., & Cooper, A. W. (2007). The effects of divided attention on auditory
priming. Memory and Cognition, 35(6), 12451254.
Mulligan, N. W., & Hartman, M. (1996). Divided attention and indirect memory tests. Memory and
Cognition, 24(4), 453465.
Mulligan, N. W., & Hornstein, S. (2000). Attention and perceptual priming in the perceptual
identification task. Journal of Experimental Psychology: Learning, Memory, and Cognition,
26(3), 626637.
Mulligan, N. W., & Peterson, D. (2008). Attention and implicit memory in the category verification
and lexical decision tasks. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 34(3), 662679.
Mulligan, N. W., & Stone, M. (1999). Attention and conceptual priming: Limits on the effects of
divided attention in the category-exemplar production task. Journal of Memory and Language,
41(2), 253280.
Musen, G., & Treisman, A. (1990). Implicit and explicit memory for visual patterns. Journal of
Experimental Psychology: Learning, Memory and Cognition, 16(1), 127137.
Nelson, D. L., Keelean, P. D., & Negrao, M. (1989). Word-fragment cuing: The lexical search
hypothesis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(3),
388397.
Nyberg, L., Nilsson, L. G., & Olofsson, U. (1994). Repetition effects on word fragment
completion: The role of competition among responses. Scandinavian Journal of Psychology,
35(1), 5666.
Nyberg, L., Winocur, G., & Moscovitch, M. (1997). Correlation between frontal lobe functions and
explicit and implicit stem completion in healthy elderly. Neuropsychology, 11(1), 138146.
Olofsson, U. (1995). Retention interval, response competition and word familiarity effects in
primed fragment completion. European Journal of Cognitive Psychology, 7(2), 131143.
Parkin, A. J., Reid, T. K., & Russo, R. (1990). On the differential nature of implicit and explicit
memory. Memory and Cognition, 18(5), 507514.
Parkin, A. J., & Russo, R. (1990). Implicit and explicit memory and the automatic/effortful
distinction. European Journal of Cognitive Psychology, 2(1), 7180.
Pashler, H. E. (1994). Dual-task interference in simple tasks: Data and theory. Psychological
Bulletin, 116(2), 220244.
Pashler, H. E. (1998). The psychology of attention. Cambridge, MA: MIT Press.
44 SPATARO, MULLIGAN, ROSSI-ARNAUD
Szymanski, K. F., & MacLeod, C. M. (1996). Manipulation of attention at study affects an explicit
but not an implicit test of memory. Consciousness and Cognition, 5(12), 165175.
Tenpenny, P. L., & Shoben, E. J. (1992). Component processes and the utility of the conceptually-
driven/data-driven distinction. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 18(1), 2542.
Thapar, A., & Greene, R. L. (1994). Effects of level of processing on implicit and explicit tasks.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(3), 671679.
Vaidya, C. J., Gabrieli, J. D. E., Keane, M. M., Monti, L. A., Gutierrez-Rivas, H., & Zarella, M. M.
(1997). Evidence for multiple mechanisms of conceptual priming on implicit memory tests.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 23(6), 13241343.
Weldon, M. S. (1991). The time course of perceptual and conceptual contributions to word
Downloaded by [Universita Studi la Sapienza] at 02:26 05 September 2013