Psychological Research (2006) 70: 26–31
DOI 10.1007/s00426-004-0185-6
O R I GI N A L A R T IC L E
Ruth K. Raanaas Æ Svein Magnussen
Serial position effects in implicit memory for multiple-digit numbers
Received: 19 December 2003 / Accepted: 27 May 2004 / Published online: 3 September 2004
Ó Springer-Verlag 2004
Abstract Serial position effects in implicit and explicit
memory were investigated in a short-term memory task.
A study list composed of four, spatially distributed,
two-digit numbers was presented, followed by an item
recognition task (explicit test) and an implicit memory
task in which participants were asked to verify a simple
addition equation where the presented answer was either
primed or not primed by one of the number pairs in the
study list. Similar serial position effects were observed in
explicit and implicit memory, with faster response times
for correct decisions on the first than on the later list
positions. The presence of a primacy effect but no recency effect is consistent with previous studies of explicit
memory with visual presentation. The results suggest
that similar principles of temporal information processing govern priming and episodic short-term memory.
Introduction
In explicit memory tasks having a list format, serial position effects are almost inevitably observed. Primacy
and/or recency effects have been reported for a wide
variety of semantic and episodic tasks in both short-term
and long-term memory (e.g., Crowder, 1993; Helstrup &
Magnussen, 2001; Maylor, 2002; Nairne, 1991; Pinto &
Baddeley, 1991; Roediger & Crowder, 1976). In contrast
to the large body of research on serial position functions
in explicit memory, little is known about serial position
effects in implicit memory, defined as manifestations of
memory that occur in the absence of intentions to reR. K. Raanaas (&) Æ S. Magnussen
Department of Psychology, University Of Oslo,
PO Box 1094, 0317 Blindern, Oslo, Norway
E-mail: r.k.raanaas@psykologi.uio.no
S. Magnussen
Centre for Advanced Study,
the Norwegian Academy of Science and Letters,
Oslo, Norway
collect (McDermott, 2000). The few experiments published on serial position effects in implicit memory have
given conflicting results. In experiments using word stem
completion, word fragment completion, or word pair
association tasks, primacy effects are either found
(Brooks, 1999; Gershberg & Shimamura, 1994; Kuster,
1998; Sloman, Hayman, Otha, Law, & Tulving, 1988) or
not found (Brooks, 1994; Phaf & Wolters, 1996; Rybash
& Osborne, 1991), and recency effects are either found
(Brooks, 1999; Gershberg & Shimamura, 1994; Kuster,
1998; McKenzie & Humphreys, 1991; Rybash & Osborne, 1991; Sloman et al., 1988) or not found (Brooks,
1994; Phaf & Wolters, 1996).
Serial position effects may be absent from implicit
memory for (at least) two reasons. First, implicit memory may reflect the operation of a different memory
system from that of explicit memory, governed by different principles of organization. The distinction between explicit and implicit memory is frequently linked
to the concept of parallel memory systems, with priming
proposed as a separate system of implicit memory
(Foster & Jelicic, 1999; Tulving & Schacter, 1990;
Jacoby & Dallas, 1981; Verfaellie, Gabrieli, Vaidya, &
Croce, 1996; Tulving, 1999). However, implicit and explicit memory may also be different expressions of a
common system (Poldrack, Desmond, Glover, & Gabrieli, 1998; Roediger, Buckner, & McDermott, 1999).
According to Craik (2003) the distinction between implicit and explicit memory is the awareness or lack of
awareness that previous events and learning play a part
in present perception and performance. Craik (2003)
claims that: ‘‘In most cases such awareness is neither
necessary nor helpful in performing sensori-motor and
cognitive tasks; performance runs off efficiently, automatically and unconsciously’’ (p. 334). Accordingly,
implicit memory may affect and in many cases may be
responsible for performance in traditional explicit
memory tests (Craik, 2003; Roediger, 2003). Studies of
serial position functions in implicit memory may contribute to this debate on the following logic. If serial
position effects are consequences of encoding processes,
27
as traditionally believed (e.g., Atkinson & Shiffrin, 1968;
Craik & Lockhart, 1972; Glenberg et al., 1980), they
should turn up in memory research independent of the
retrieval format being implicit or explicit, provided the
different formats reflect a common memory system.
Serial position effects may also turn up if the different
formats reflect separate memory systems, provided the
two memory systems are governed by similar principles
of organization. However, the presence of serial position
effects in one retrieval format and the absence of such
effects in another format would count in favor of the
operation of separate memory systems.
The second reason is purely experimental. Raanaas
and Magnussen (submitted) identified several factors
that might contribute to the fragility of serial position
effects in implicit memory in the previous studies. First,
experimental designs that pair a single stimulus with a
single test session allow only a single measurement at
each list position and make the results sensitive to itemspecific effects. Second, participants may have recognized the real purpose of the experiment and explicit
processing may have interfered with implicit processing.
As words are generally easy to recognize, they may be
relatively unsuitable as stimulus material in studies of
serial position effects in priming, unless the study occurs
incidentally as in Raanaas and Magnussen’s (submitted)
study, where the study list is disguised in a Stroop-like
task. Third, the exclusive use of an accuracy indicator of
memory performance may mask memory effects that
might otherwise be revealed by the more sensitive
response-time indicator. Experiments on perceptual
short-term memory have shown that under certain
conditions response times may demonstrate memory
effects where accuracy does not (Magnussen, Idås, &
Holst Myhre, 1998; Magnussen, 2000).
In the present study, serial position effects in implicit
memory have been measured in a priming task with
multiple digit numbers as material, as numbers are easily
perceivable and verbally coded, but evidently not so easy
to remember (Nordby, Raanaas, & Magnussen, 2002;
Raanaas, Nordby, & Magnussen, 2002). A primed
addition verification task was used as an implicit test, the
effect of priming in the production and verification of
arithmetic problems being previously well documented
(e.g., Arbuthnott, 1996; Campbell, 1991; Campbell &
Tarling, 1996; Delazer & Girelli, 2000; Delazer, Ewen, &
Benke, 1997), camouflaged as a distracter task to the
participants. A study list composed of four digit pairs
was followed by an explicit recognition test, announced
to the participants as the real purpose of the experiment,
consisting of a previously presented or a novel digit pair,
and the participants had to decide if it was present or not
in the study list (explicit test). Between the memory test
and the presentation of the next study list a simple
addition equation (e.g., 21+14=35) was presented for
the participants to verify, the presented answer of which
was either primed or not primed by one of the digit pairs
in the previously studied list (implicit task). The use
of two-digit numbers gives the sufficient number of
combinations for each list position to be tested several
times for each participant. Accuracy and response times
were recorded.
Serial position effects in explicit short-term memory
for numbers have previously been thoroughly documented (Conrad & Hull; 1968; Frick, 1984; Nordby
et al., 2002; Raanaas et al., 2002). In the present
experiment we measure these effects in explicit and
implicit memory concurrently.
Materials and methods
The participants memorized a study list composed of
four two-digit numbers (e.g., 22 56 39 20), immediately
followed by a recognition test. The recognition test was
followed by an implicit memory task explained to the
participants as a distracter task between trials. In the
implicit test participants were shown a simple addition
equation and asked to verify if the statement was true or
false; in half of the ‘‘true’’ trials the presented answer
corresponded to one of the digit pairs in the study list.
The design was a simple one-variable within-subject
design with the variable serial position having four levels
(positions), the experiment was completely balanced
with respect to true and false arithmetic equations,
priming and non-priming condition, and serial position.
Participants
A total of 80 volunteer students participated, 40 women
who averaged 19.2 years of age (range 16–26) and 40
men whose average age was 18.6 (range 16–27). Participants were native Norwegian speakers with normal or
corrected-to-normal vision and were paid a small sum
for their participation. All signed an informed consent.
Apparatus and stimulus material
A standard IBM computer (128 MB RAM) was used.
The program was written in SuperLab (produced by
Cedrus, San Pedro, CA, USA) and run on Microsoft
Windows. The numbers were presented on a Targa 21’’
screen. Responses were made on a RB 410 Response
Box (produced by Cedrus). The digits displayed on the
PC screen were 10 mm tall ‘Arial’ (font size 60), black on
a light gray background, providing high legibility. One
typed space was used between the pairs in the stimulus
numbers. The stimulus lists were composed of randomly
selected two-digit numbers between 11 and 99, with the
restrictions that identical or immediately adjacent
number pairs did not appear in the same number, and
identical number pairs did not appear in successive
numbers. In the implicit memory test the addition
equations were either on the format: A one- or two- digit
number plus a one-digit number where addition gave an
‘extra decade’ (e.g. 42+9=51), or a two-digit number
28
plus a two-digit number where no extra decades were
created; i.e., simple addition could be computed on the
units and decades separately (e.g., 21+17=38). The
equations were presented in a horizontal format, and
the largest addend was always placed left. None of the
addends were part of the previous study list. The
incorrect sums were all wrong by a number between 2
and 15. Half the trials were primed and half were not
primed. Among both the primed and the non-primed
trials half of the addition equations were correct and
half were incorrect. Among the primed (correct and
incorrect equations) there were two tests on each of the
four serial positions.
Eight versions of the stimulus list were designed
(varied between participants), all composed of randomly
selected numbers, and unique and randomly selected
sequences of test/distracter trials, correct/incorrect
addition equations, and serial positions. Males and
females were separately but randomly allocated to the
eight different stimulus lists: Ten persons in each group,
five men, and five women. In the explicit task half the
trials consisted of numbers appearing in the study list
and half the trials were novel numbers. Numbers that
were presented in the recognition tests were not represented in the implicit memory test. Except for this
restriction there were no links between the explicit and
the implicit tests. Each version comprised a total of 32
trials.
Procedure
The participants were tested individually. The experiment was introduced to the participant by a written
instruction on the monitor, telling the participant that
the purpose of the experiment was to test the short-term
memory for numbers, and that addition equations were
inserted between the memory tests to make the task
more difficult. To both tasks they should respond as fast
as possible. Two practice trials were presented to
familiarize the participants with the experimental
Fig. 1 Recognition of two-digit
numbers. a Percentage of
correct responses are plotted as
a function of serial position in
the stimulus list for previously
presented numbers. b Response
times for correct recognitions
are plotted as a function of
serial position in the stimulus
list. The mean response time to
the correctly rejected new
numbers is shown for
comparison
procedure. In each trial, the prompt ‘‘memorize the
number’’ was presented on the screen for 2,000 ms,
followed by an interval of 1,000 ms, before an 8-digit
number, grouped in pairs (e.g., 26 56 43 80) was presented for 4,000 ms. A mask composed of eight ‘x’-s, i.e.,
XX XX XX XX, was presented for 3,000 ms followed
by another interval of 1,000 ms. The explicit recognition
test was prompted by the signal ‘‘check your memory’’
for 2,000 ms before a two-digit number was presented
on the screen, which was either presented or not presented in the stimulus list. The participant’s task was to
press the ‘‘yes’’ or ‘‘no’’ button on the Response Box as
fast as possible. An interval of 2,000 ms was inserted
before the implicit memory test was prompted by the
signal ‘‘compute’’ followed by a short interval of
1,000 ms before the verification task was presented (e.g.,
77+12=89 in the non-primed condition) and left on the
screen until the participants responded. The participant’s task was to solve whether it was correct (press
‘‘yes’’) or incorrect (press ‘‘no’’) as fast as possible. The
next trial was started after an interval of 2,000 ms.
Results and discussion
The results for the eight different stimulus lists were
pooled. In the explicit condition the percentage of correctly recognized items and response time to the correctly recognized items were analyzed, with response
times ± 3 standard deviations of the mean removed as
outliers (3.3%); individual means (n=4) were calculated
at each list position. In the implicit condition the percentage of correct responses to the correct arithmetic
equations, and response time to the correctly verified
correct arithmetic equations were analyzed. Again,
response time scores ± 3 standard deviations from the
mean were removed (3.1%) and individual means (n=2)
were calculated at each serial position.
Figure 1 shows the results of the explicit memory test,
with the left panel (a) displaying the percentage of correctly recognized items. There is a uniformly high level of
29
recognition performance, with no serial position effects,
F(3,237) = .546, p = .651. In Fig. 1b response times to
the correctly recognized items are plotted as a function of
serial position, with the mean response time to the correctly rejected novel items represented by a horizontal
dotted line. The average response time to novel numbers
was 952.4 ms and to previously presented numbers it was
995.1 ms, giving a significant difference, t(79) = 2.543,
p = .013. The overall effect of serial position for the
previously presented numbers was also significant,
F(3,234) = 4.714, p = .003 reflecting a primacy effect,
but no recency effect, which was confirmed by pair-wise
comparisons of all serial positions shown in Table 1.
Figure 2a shows the proportion of correct responses
for the correct arithmetic statements as a function of
serial position in the primed condition, with the results
from the non-primed condition represented by common
reference indicated by a horizontal dotted line. Not
unexpectedly, the level of performance is very high with
an average of 89.7 and 86.9% correct responses in the
primed and non-primed conditions, giving no significant
overall effect of priming, t(79) = 1.501, p = .137, and
there is no serial position effect for the primed condition,
F(3,237) = 1.047, p = .372. However, the previous
presentation of the number corresponding to the correct
statements helped the speed with which the decisions
were made. Figure 2b shows the corresponding response
times for the primed condition, with the results from the
non-primed condition represented as a dotted line.
There is a significant overall effect of priming with faster
decisions in the primed (2,419.6 ms) compared to the
non-primed (2,549.4 ms) decisions, t(79) = 2.983,
p = .004. There is also an overall significant serial position effect for the primed condition, F(3,216) = 2,765,
p < .05, reflecting a primacy effect, but no recency effect, as confirmed by pair-wise comparisons between all
positions presented in Table 2.
The explicit and implicit tests were not designed to be
directly comparable. The experimental purpose of the
explicit memory test was to distract the participants and
conceal the real purpose of the experiment, and postexperimental questioning confirmed the success of this
strategy. In addition, the two tests involve different
cognitive operations, possibly of different complexity, as
evidenced by the fact that response times were about 2.5
times higher in the implicit test than in the explicit test.
However, despite the differences between the explicit and
Table 1 Pair-wise comparisons of all serial position scores for
speed of correct recognition in explicit memory
Table 2 Pair-wise comparisons of all serial position scores for
speed of verification of an addition equation primed by the study
list
Paired comparisons
Degrees of
freedom
t-value
Significance
(2-tailed)
Paired comparisons
Degrees
of freedom
t-value
Significance
(2-tailed)
Positions
Positions
Positions
Positions
Positions
Positions
79
79
78
79
78
78
3.083
1.959
3.673
1.244
.227
1.354
.003
.054
.0001
.271
.821
.180
Positions
Positions
Positions
Positions
Positions
Positions
72
75
75
76
76
79
2.456
2.731
2.177
.491
.897
.418
.016
.008
.033
.625
.372
.677
1
1
1
2
2
3
and
and
and
and
and
and
2
3
4
3
4
4
Fig. 2 Priming of two digit numbers. a The percentage of correctly
verified correct addition equations are plotted as a function of
serial position of the sum number in the stimulus list. The reference
for priming is the mean percentage of correctly verified correct
addition equations based on the use of new numbers. b Response
1
1
1
2
2
3
and
and
and
and
and
and
2
3
4
3
4
4
times for correctly verified correct addition equations where the
presented answer is primed by the study list as a function of the list
position, with response times for correctly verified correct addition
equations not primed by the study list as a common reference value
for all positions
30
implicit tests, parallel serial position effects were
observed, and in both types of test the effects turned up in
the speed but not in the accuracy of the memory performance. The absence of serial position effects in the
accuracy of performance may partly be due to the high
level of performance in this experiment and thus ceiling
effects, but other studies of short-term memory have
also shown that response time may be a more sensitive
indicator of memory processing than accuracy (Magnussen et al., 1998; Magnussen, 2000; Raanaas & Magnussen, submitted). The fact that the deflection in the RT
serial position curve was more pronounced in the implicit
test than in the explicit test might be related to the different response time levels in the two tasks.
Somewhat surprisingly, the results of Fig. 1b showed
that new items were recognized faster than studied items.
An analysis of the accuracy data indicated a similar
‘‘study cost.’’ This ‘‘new’’ item advantage may be
related to the fact that the ability to reject negative
probes is affected by the similarity of the probes to the
memorized items (e.g., Lamberts, Brockdorff, & Heit,
2003; Lively & Sanford, 1972; Reynolds & Goldstein,
1974). Other studies have found that participants
respond faster to new dissimilar words than to old
words (Lamberts et al., 2003). In the present experiment
‘‘new’’ number pairs differed from the studied ones at
both the unit and decade positions.
The results of the present paper thus confirm the
presence of serial position effects in implicit memory. The
serial position patterns in both types of test are, furthermore, consistent with a number of previous findings
that primacy but no recency is observed with visual
presentation compared with auditory presentation,
where both primacy and recency are typically observed
(LeCompte, 1992; Nordby et al., 2002; Raanaas et al.,
2002), a phenomenon termed the modality effect (Conrad
& Hull, 1968). The results suggest that equivalent information stored in explicit and implicit memory is organized in a similar fashion, and would thus appear most
consistent with the idea of explicit and implicit memory
as expressions of a common memory system.
Acknowledgements This study was supported by the Norwegian
Research Council (MH). We thank David Geary and Irving Koch
for constructive comments on a previous version of the paper.
References
Arbuthnott, K. D. (1996). To repeat or not to repeat: Repetition
facilitation and inhibition in sequential retrieval. Journal of
Experimental Psychology: General, 125, 261–283.
Atkinson, R. C., & Shiffrin R. M. (1968). Human memory: A
proposed system and its control processes. In K. W. Spence &
J. T. Spence (Eds.), The psychology of learning and motivation:
Advances in research and theory (Vol.2, pp. 89–195). San Diego:
Academic.
Brooks, B. M. (1994). A comparison of serial position effects in
implicit and explicit word-stem completion. Psychonomic Bulletin and Review, 1, 264–268.
Brooks, B. M. (1999). Primacy and recency in primed free association and associative cued recall. Psychonomic Bulletin and
Review, 6, 479–485.
Campbell, J. I. D. (1991). Conditions of error priming in number
fact retrieval. Memory and Cognition, 19, 197–209.
Campbell J. I. D., & Tarling, D. P. M. (1996). Production, verification and error priming in cognitive arithmetic. Memory and
Cognition, 24, 156–172.
Conrad, R., & Hull, A. J. (1968). Input modality and the serial
position curve in short-term memory. Psychonomic Science, 10,
135–136.
Craik, F. I. M. (2003). Commentary. In J. S. Bowers & C. J.
Marsolek (Eds.), Rethinking implicit memory (pp. 327–336).
New York: Oxford University Press.
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A
framework for memory research. Journal of Verbal Learning
and Verbal Behavior, 11, 671–684.
Crowder, R. G. (1993). Short-term memory: Where do we stand?
Memory and Cognition, 21, 142–145.
Delazer, M., & Girelli, L. (2000). Priming arithmetic reasoning in
an amnesic patient. Brain and Cognition, 43, 138–143.
Delazer, M., Ewen, P., & Benke, T. (1997). Priming arithmetic facts
in amnesic patients. Neuropsychologia, 35, 623–634.
Foster, J. K., & Jelicic, M. (1999). Memory: Systems, process or
function? New York: Oxford University Press.
Frick, R. W. (1984). Using both an auditory and a visual shortterm store to increase digit span. Memory and Cognition, 12,
507–514.
Gershberg, F. B., & Shimamura, A. P. (1994). Serial position effects
in implicit and explicit tests of memory. Journal of Experimental
Psychology: Learning, Memory and Cognition, 20, 1370–1378.
Glenberg, A. M., Bradley, M. M., Stevenson, J. A., Kraus, T. A.,
Tkachuk, M. J., Gretz, A. L., Fish, J. H. & Turp, B. M. (1980).
A two-process account of long-term serial position effects.
Journal of Experimental Psychology: Human Learning and
Memory, 6, 355–369.
Helstrup, T, & Magnussen, S. (2001). The mental representation of
familiar, long-distance journeys. Journal of Environmental
Psychology, 21, 411–421.
Jacoby, L. L., & Dallas, M. (1981). On the relationship between
autobiographical memory and perceptual learning. Journal of
Experimental Psychology: General, 110, 306–340.
Kuster, J. P. (1998). Serial position effects in the implicit memory
task of word stem completion. Dissertation Abstracts International: Section B: The Sciences and Engineering, 59, 1396.
Lamberts, K., Brockdorff, N., & Heit, E. (2003). Feature-sampling
and random walk models of individual-stimulus recognition.
Journal of Experimental Psychology: General, 132, 351–378.
LeCompte, D. C. (1992). In search of strong visual recency effects.
Memory and Cognition, 20, 263–272.
Lively, B. L., & Sanford, B. J. (1972). The use of category information in a memory-search task. Journal of Experimental
Psychology, 93, 379–385.
Magnussen, S. (2000). Low-level memory processes in vision.
Trends in Neurosciences, 23, 247–251.
Magnussen, S., Idås, E., & Holst Myhre, S. (1998). Representation
of orientation and spatial frequency in perception and memory:
A choice reaction time analysis. Journal of Experimental Psychology: Human perception and performance, 24, 707–718.
Maylor, E. A. (2002). Serial position effects in semantic memory:
Reconstructing the order of verses of hymns. Psychonomic
Bulletin and Review, 9, 816–820.
McDermott, K. B. (2000). Implicit memory. In A. E. Kazdin (Ed.),
The encyclopedia of psychology (pp. 231–234). New York:
Oxford University Press.
McKenzie, W. A., & Humphreys, M. S. (1991). Recency effects in
direct and indirect memory tasks. Memory and Cognition, 19,
321–331.
Nairne, J. S. (1991). Positional uncertainty in long-term memory.
Memory and Cognition, 19, 332–340.
31
Nordby, K., Raanaas, R. K., & Magnussen, S. (2002). The expanding telephone number. I. Keying briefly presented multiple-digit numbers. Behaviour and Information Technology, 21,
27–38.
Phaf, R. H., & Wolters, G. (1996). Elaboration effects in implicit
and explicit memory tests. Psychological Research, 58, 284–
293.
Pinto, A. D. C., & Baddeley, A. D. (1991). Where did you park
your car? Analysis of naturalistic long-term recency effect.
European Journal of Psychology, 3, 297–313.
Poldrack, R. A., Desmond, J. E., Glover, G. H. & Gabrieli, J. D. E.
(1998). The neural basis of visual skill learning: An fMRI study
of mirror reading. Cerebral Cortex, 8, 1–10.
Raanaas, R. K., & Magnussen, S. Serial position effects in implicit
memory. Manuscript submitted for publication.
Raanaas, R. K., Nordby, K., & Magnussen, S. (2002). The expanding telephone number. II. Age variations in immediate
memory for multiple-digit numbers. Behaviour and Information
Technology, 21, 39–45.
Reynolds, J. H., & Goldstein, J. A. (1974). The effects of category
membership on memory scanning for words. American Journal
of Psychology, 87, 487–495.
Roediger, H. L. III. (2003). Reconsidering implicit memory. In J. S.
Bower & C. J. Marsolek (Eds.), Rethinking implicit memory (pp.
3–18). New York: Oxford University Press.
Roediger, H. L. III., & Crowder, R. G. (1976). A serial position
effect in recall of United States presidents. Bulletin of the Psychonomic Society, 8, 275–278.
Roediger, H. L. III., Buckner, R., & McDermott, K. B. (1999).
Components of processing. In J. K. Foster & M. Jelicic (Eds.),
Memory: Systems, process, or function? (pp 31–65). New York:
Oxford University Press.
Rybash, J. M., & Osborne, J. L. (1991). Implicit memory, the serial
position effect, and test awareness. Bulletin of the Psychonomic
Society, 29, 327–330.
Sloman, S. A., Hayman, C. A. G., Otha, N., Law, J., & Tulving, E.
(1988). Forgetting in primed fragment completion. Journal of
Experimental Psychology: Learning, Memory and Cognition, 14,
223–239.
Tulving, E. (1999). Study of memory: process and systems. In J. K.
Foster, & M. Jelicic (Eds.), Memory: Systems, process or function? (pp. 11–30). New York: Oxford University Press.
Tulving, E., & Schacter, D. L. (1990). Priming and human memory
systems. Science, 247, 301–306.
Verfaellie, M., Gabrieli, J. D. E., Vaidya, C. J., & Croce, P. (1996).
Implicit memory for pictures in amnesia: Role of etiology and
priming task. Neuropsychology, 10, 517–537.