Sie sind auf Seite 1von 19

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR I7, 649_667 ( 1978)

On Interpretingthe Effectsof Repetition:


Solvinga ProblemVersusRemembering
a Solution
Lnnny L. Jecosy
M acM uster Ll niuersitt,
When a problem is repeated. the later presentation of the problem sometimes results in the
subjectrespondingby remembenngthe solution rather than by going through.theoperations that
would otherwise be necessaryto solve the problem. The meansof obtaining the solution is shown
to influence subsequent retention performance: retention of the solution suffers if it has been
obtained by remembenng rather than by solving the problem. The distinction betweensolving a
problem and remembenng a solution is used in an account of the etTectof spacing repetitions and
other standard memory phenomena. The relevance of the distinction to tasks such as word
perception is also discussed.

Supposethat you are askedto find the sum petitionsof the problem by severalintervening
o f 3 7+ 1 5 + 1 2 . A f t e r h a v i n g o b t a i n e d r h i s s u mproblems of the same form.
your are immediatelypresentedwith the same
This exampleof addition is the basisof the
problem. The type of processingthat you do analysisof the eflectof reperitionon memory
will differ drastically on the repeated pre- that is presentedin this paper. The task of
sentation. On the first encounter you un- memorizinga list of words can be comparedto
doubtedly went throu_eh the process of the task of solving a seriesof problems"The
addition to obtain the sum: on the second presentationof a word [or memory constitutes
encounter.the sum is readilyavailableand can a problem: the subject must find operarions
be _uivenwithout going back through the that will render that word memorable after
operationsof adding the numbers. lndeed. a some delay. For example. the subject mav
full repetition of the processingactivity may image the referent of the word in order to
be difficult, if not impossible,to accomplish enhancememorv. As with math problems.it is
without some delay. which is probably the unlikely that a repetitionof a word resultsin a
rationalefor the commonly prescribedroutine full repetitionof the processing.If one haslust
of checking an addition by adding the num- imaged their own dog in order to make the
bers in reverseorder rather than simply re- word "dog" more memorable.imagin_etheir
adding them in the same order. To make it dog a second time as a consequenceof the
possibleto repeatthe full processof addition, word being repeatedis unlikely to require a
it is probably sufficient to separatethe re- iull repetition of the processesthat were
necessaryfor the original imaging.In general.
it seemsthat one can retrievethe product of
The author is grateful to Lee Brooks and F. l. M. Craik
their prior memorizing activiry without fully
for comments and suggestionson an earlier draft oi this
repeatin,ethat memorizing activiry.
paper.This researchwas supported by Grant 4028 t from
The meansby which a solution to a problem
the National Research Council ol Canada. Address
is
obtained
will influencesubsequentrerention
repnnt requests to Larrv L. Jacoby. Deparrment oi
of the probl em and i ts sol uti on. Thi s cl ai m
P s y c h o l o g y .M c M a s t e r U n i v e r s i t y . H a m i l t o n . O n r a r i o .
Canada L8S 4Kl.
has been used in recommending"discovery"
649

ri' ft \31""]ll?)];.il'i
.oor,,*n,"ltt,

l l r i g h t s o i r c p r o d u c t r o n i n a n y . t' b r m r e s c r r e d .

650

LARRY L. JACOBY

learning as compared to "reception" learning is effortlesslyretrieved.


(Bruner, 1966);the suggestionis that working
The experiments that are to be reported
through a problem to its solution enhances provide a clear demonstrationoi the memorv
memory as comparedto a situation wherethe consequencesof solvin_sa problem versus
solution is provided. Little is known about rememberi nga sol uti on. Much of the subsehow these effectswork. However, one possi- quent discussionwill centeraround the effects
bility is that solving a problem results in a of spacingrepetitions.However. the contrast
"richer" memory of that problem and its betweensolving a problem and remembering
solution. In the math example, the further a sol uti on i s appl i cabl eover a much w i der
operationsthat are performedwhen addition rangeof si tuati onsthan i s usual l vconsi dered
is required may result in a more extensive w hen di scussi ngthe mernori zi ngof a l i st of
memory of the problem by including substeps w ords. One potenti al appl i cati on thar i s of
leading to its solution. This additional infor- current interest involves word identitrcation.
mation could provide a further basis for A pronunciation for a word can be consubsequentrecognition of the problem and structedby goin_e
throu_eha seriesoI rulesthat
increasethe number of potentialcuesfor later deal with letter to sound correspondences.
As
recall"A secondexplanation of the retention i n the math exampl e" how ever. thi s conadvantageof solving a problem as compared structive activity is likely to be bvpassedor
to reading or effortlessly remembering the minimized when the conditions are such as to
solution appealsto the role of consciousness allow the subject to easily remember a proin determining subsequentretention. In the nuniciation that he has encountered premath example.addin-ea seriesof numbers to viously.Thus, the contrast with which we are
obtain a solution invlovesconsciousness
in a dealingis relevantto many tasksin addition to
way " that "effortless" rememberin-e of the those of solving math problems or memorizsolution does not. When adding the numbers. ing word lists" Potential applications of the
it seems necessaryto monitor one's own distinction between solving a problem and
processingwhile an effortlessretrieval of a rememberinga solution are describedin the
solution seems "automatic". The involve- -eeneraldiscussion.
ment of consciousnessmay enhance subThe _generaldiscussion also includes a
sequentretention performance.
reviewof severalexperiments
that can be used
This analysisis relevantto the spacingeffect to support the claim that an advantage in
t hat i s w e l l d o c u me n te di n th e me mory l i tera- subsequentretention is gained by constructt ur e (H i n tz m a n . 1 9 7 4 1T. h e a r-e u menti s that ing rather than rememberinga solution. The
the proccssingof the first presentationof a ar*qumentthat the effectof spacinurepetitions
wor d m a k e s a v a i l a b l e a n a p p ropri ate en- resuitsfrom a changein the mode of obtainins
coding and thereby trivializesthe processing a solution, or achieving an encoding. is exassociatedwith the second presentation of panded and contrasted with other explanathe word. As the spacing of repetitions is tions. This ar_qumentis then extended to
increased.the amount of processingof the accountfor vari ousmemorv phenomcnathat
repeatedword that is required to attain an have previ ousl v been di scusscd i n rnuch
appropriateencodi n g i ncreases:
consequently, narrow er contexts.The di sti rl ctronbctncen
one should expectretentionto be enhancedas sol vi ng a probl em und rcmcnrhrcri rl;.r
g sol ua function of the spacing of repetitions.As ti on i s shoun ttl have crl nsi dcrubl chcuri sti c
ar-euedwith referenceto solving a problem. val ue:thi s di sti ncti oncan bc uscd to suggest
working with an item to derive an encoding experi mentsthat w ' oul d not ari se from the
produces subsequentretention that exceeds more tradi ti onalexpl anati onsof the phenomthat produced when an appropriateencoding ena that are reviewed.

R E T E N T I O N E F F E C T SO F S O L V I N G V S R E M E } I B E R I N C

Ex p rn ry rN r I

651

relatedwords. For pairs that were to be read.


the ri ght-hand member oi the pai r w as preSubjectsen_ea_eed
in a task that is similar to
sented i ntact. For pai rs that requi red a
that of solvinga crosswordpuzzle.A cue word
response to be constructed, two or more
was presentedalon_ewith a few lettersand a
letters were deleted from the interior of the
series of blanks representing the missing
right-handmemberof the pair: the subjectwas
letters of a word that was related to the cue
to
say the word that could be formed by
word (e.g..foot s- -e). The subject'stask was
restoring the missing letters.
to report the word that could be producedby
The experimentwas designedso that each
f illing th e b l a n k s(s h o ei n th e a b o v eexampl e).
of
si x condi ti onsw ererepresented
by l l i tems
I n s o me i n s ta n c e s th
. e ta s k o f s o l vi ng the
mi
xed
i
n
72-i
tem
a
si
ngl
e
l
i
st.
One condi ti on
puzzlewas trivialized by precedingthe prob(R ) consi sted of the 12 i tems that w ere
lem w i th i ts s o l u ti o n : th e p ri ma ry mani pupresentedonly onceand in which the response
lation in the first experimentwas to vary the
had only to be reutl by the subject.A second
spacin_u
of the puzzle and its solution. The
(C ) consi stedof the l 2 i tems that
pr oc e s s i n gre q u i re d to o b ta i n th e sol uti on condi ti on
were presentedonly once but for which the
and, consequently,later memory were exresponsehad to be c'orr.srrurc'tetl.ln
two oi the
pectedto be _greater
when presentationof the
remaining
pair
conditions,
was
presented
each
solution was separated from the puzzle by
twice
with
the
response
being
read
both
times
intervening items rather than immediately
(R R ): i n one of these R R condi ti ons. the
preceding the puzzle in the list. Retention
performance was assessedby means of an secondpresentationimmediatelyfollowed the
unexpectedtest of cued recall: the cue word first, and in the other it lollowed with a lag of
from each of the puzzles (e._s.,foot) was 20 i tems.In the fi nal tw o condi ti onsthe i tem
presentedas a cue for recall of the solution was to be read the first time and constructed
the second (R C ): agai n. i n one of the R C
words.
Co mp a ri s o n sa m o n g c o n d i ti o n s w ere de- condi ti onsthe repeti ti onw as i mmedi ateand
s igne dto p ro v i d e i n fo rma ti o na b o u t the pro- i n the other al ter 20 i nterveni ngi tems.
E i ghteensubj ectsw ere pai d S 2.00i hourto
cessingcarried out to solve the puzzle. For
participate. Testine was conducted in indiexample, in the first experiment.cued recall
af t err e a d i n gth e s o l u ti o na n d th e n s ol vi ngthe vidual sessions.
Muteriuls. Seventy-two pairs of related
puzzle was compared to cued recall after
words
were selectedfrom the Connecticut
hav in g re a d th e s o l u ti o n tw i c e . When the
free-association
norms. In selecting pairs.
solution word immediately precededpresenneither
the
most
frequent
associationto a cue
tation of the puzzle,solvingthe puzzlewas not
word
nor
a
bizarre
association
was selected.
expectedto entail any more effort or produce
The
intent
was
pairs
to
select
such
that the
any better recall than would result flrom
response
word
could
be
solved
in
the
consimply readingthe solution word for a second
di
ti
ons
requi
ri
ng
constructi
on
w
i
thout
the
t im e. Wi th g re a te rs e p a ra ti o no f a p uzzl eand
sol
uti
on
bei
ng
too
obvi
ous.
The
response
it s s o l u ti o n . h o w e v e r. th e re q u i r ement of
solvin_ethe puzzle was expectedto produce membersof pairs varied in lengthfrom four to
hi-eherretentionthan would be producedby a eight letters.When constructionof a response
was required.the first letter and the last letter
s ec on dre a d i n eo f th e s o l u ti o n w o rd .
of the responsewerealwayspresented.For the
longer response words, up to four letters
Methotls
including the first and last letter oi the word
Desig4n
untl suhjecrs.Subjectseither read or
were provided: two or more letters were
c ons t ru c te dth e ri g h t-h a n dm e m b e rof pai rsof deletedfrom eachresponseword that required

652

LARRY L. JACOBY

construction.Deletedletterswere replacedby the data that are of pri marv i nterest. In


blank s s o th a t th e n u m b e r o f l e tte rs i n the analvzing the cued-recalldata comine from
word that was to be constructedwas obvious. condi ti ons that have been requi red to conSix lists were formed by rotating pairs struct a response.one has the option oi
t hr ough p re s e n ta ti o n c o n d i ti o n s s o that conditionalizing cued-recall on successful
across lists each presentationcondition was constructionof the response.The rationalefor
representedby the same pairs. Within a list. such condi ti onal i zi n_s
i s: If a subj ect w as
the order o[ pairs was such that each pre- unabl e to construct a parti cul ar response
sentationcondition was representedby rrpairs duri ng presentati on
of the l i st then he w as not
beforeany presentationcondition was repre- exposed to that response:consequently.the
sentedbv rr+ I pairs.
subject cannot be expected to recall the
Proc'etlure.
The list of pairs was preparedas response on the later test of cued recall.
a stack of note cards rvith each note card Despitethis consideration.the probability of
c ont aini n go n e re l a te dp a i r. A ti m i n g devi ce cued-recall was not conditionalized in the
was usedto pacesubjectsthrough this stackof analysesthat are to be reported"The decision
note cardsat a rate of 6 seconds/card"
Subjects not to condi ti onal i zethe probabi l i tyof cuedwere informed that we were interestedin how recall was motivated by concern for a potenlong it t o o k th e m to s o l v ep ro b l e mso f th e type tial confoundin_e
that could result irom item
they might encounter in a crossword puzzle. selectionproblems.Cued-recallcan obviously
They were to turn a note card when signaled not be conditionalized when subjects only
to do so by the timin_edevice.If the ri_eht-hand read the solution to a problem; conditionalmember of the pair on the note card was not ' izing for the problemsthat requiredconstrucintact. they were to say a word that contained ti on of a sol uti on may resul t i n sel ecti vel y
the provided letters and whose remainin,s droppin_e-outthe harder pairs so that the
letters would fit in the blanks: they were comparison of the "read" and "construct"
further informed that the responsethey ,eave conditionsis confoundedwith the difficulty of
had to be related to the cue word that was the pairs on which their performance is
providedon the.card.As soon as they thought assessed.
Although they will not be reported.
they knew the answer. they were to push a analyses were also carried-out with conbutton that was in front of them and say the ditionalized scores.In
the result of
_eeneral,
s olut ion a l o u d " If th e ri _ e h t-h a nme
d mber of conditionalizing scores was to make effects
the pair on a card was intact"they wereto push largerthat were also presentand significantin
the button and read the response aloud. the analysesof unconditionalizedscores.[n
Subjectswere told that their reactiontimes to no i nstancedi d the resul ts of an anal ysi s
"read" items were to serve as a baselinefor of conditionalized
scores conflict with contheir reactiontime to responsesthat had to be clusionsthat are to be drawn from an analysis
construced.In reality.reactiontimes were not of unconditionalizedscores.
recorded:the reaction-timetask was simply
The level of significancefor all statistical
used to provide a cover story for subjects.
testsw as set at p < .05.
After subjects had worked their way
through the deck of notecards.they were_eiven Resrr/rsund Discussiorr
an unexpectedtest of cued-recall;the left-hand
Subjectswere generally successfulin conmember of each pair was provided,written in structing the appropriate response:77" of
,,
a random order on a sheet,as a cue for recallof the responseswere correct in the condition in
the right-handmemberof eachpair.The cued- which the pairs to be constructed were prerecall test was subject-paced.
sented only once. When the items had been
Anulvses.The test of cued-recallprovided read 20 items earlier (Read-Consrruct.

653

RETENTION EFFECTS OF SOLVING VS REMEMBERING

spaced) the rate of successfulconstruction


was 909/.,significantly higher: F(1, 171:13.60.
M S . -.0 1 1 . Wh e n th e i te m w a s to be constructed immediately after having been read
(Read-Construct, immediate) the rate of
successwas 99i(, significantly higher than
the delayed condition: F(1, 17): 23.94,
MS. :.gtg3.
The argument made earlier was that constructing a responseas a solution to a problem
should produce retention greater than that
produced by simply reading the response.
Further, the retention advantage that would
result from solving a problem should depend
on the processinginvolved in constructing the
solution. Immediately precedinga problem by
presentation of its solution should trivialize
the problem to such an extent that the processesinvolved in solving the problem should
not differ appreciably from those that are
required to simply read the solution a second
time; consequently, one should expect no
advantage to result from construeting the
solution as compared to a second reading of
the solution. When presentationof the solution is widely separated from that of the
problem, however, solving of the problem.
should be nontrivial and give rise to retention
that exceedsthat coming from reading the
solution for a second time. The cued-recall
data are presented in Figure I and provide
support for each of the above predictions.
When a pair was presentedonly once,constructionof a solution resultedin substantially
highercued-recallthan did simply readingthe
solution word in a pair, F(1, 171:55.92,
MS.:.02. For pairs that were repeated.the
effectof spacing repetitions was much greater
in the Read-Construct condition than in the
Read-Read condition, F(1, 17\:22.00,
M S . :.9 1 . W h e n re a d i n g o f th e sol uti on
immediately preceded presentation of the
problem, cued recall in the Read-Construct
condition did not exceedthat in the ReadRead condition; however, with spaced
presentation, the Read-Construct condition produced substantially higher cued

-ts

RC

85
=

f, 75
Irl

a6s
5
555
do5

s
rlt

Pss
L

25

IMMEOIATE

SPACED

ONCE
PRESENTED

Frc. J. Probability of cued-recall as a function of


reading (R) vs construction (C) in E.rperiment L

recall than did the Read*Read condition.


Comparisons with once-presented items
reveal that reading the solution immediately
prior to being required to construct the
solution produces lower performance than
resultsfrom constructingthe solution without
havin_epreviously read it, F(1. L7)- 15.91.
MS .:.91. W i th spaced presentati ons.the
retention advantage conferred by a prior
readingof the solution in the Read-Construct
condition is approximately equal to that
gained in the Read-Readcondition. That is.
the differencebetween Read and Read-Read
is approximately equal to that between
Construct and Read-Construct: the prior
readingof the responseenhancesrecall in both
instances.
It was once generally believed that the
important condition for learning was to lead
the subject, by whatever means. to make a
correctresponse.This beliefin the importance
of making the correct responsehas motivated

654

LARRY L. JACOBY

educational practice. It is not unusual for a


teacher to present a problem along with its
solution and then require the classto "parrot"
that solution.Within the Skinneriantradition,
programmed instruction was designed to
ensure that a correct responsewas made. An
inserted question often occurs almost immediately after the text that provides the
answer to that question; in addition, prompts
suchas rhyming cuesor a portion of the letters
comprising the response are provided to
further ensurethat the correct responsewill be
given. One point to be made by the present
study is that the processesinvolved in solving
a problem determine retention of the solution.
If the problem is trivialized by presentingthe
solution immediately prior to the problem or
by simpy requiring the person to read the
solution, retention performance will suffer.
It might be argued that the retention advantage _eained by constructing rather than
reading or remembering a solution is due to
differencesin study time; it takes longer to
construct a solution than to read one, and this
differencein effectivestudy time is responsible
for effects in subsequent retention. First, it
probably did not take twice as long to construct a solution as to read a solution. However, reading the solution twice produced
substantially lower recall than did constructing a solution only once; recall of oncepresented items that required construction
was higher than that in the Read-Read condition. Further, arguments about differences
in efectiue study time are meaninglessunless
we have someidea of what constituteseffective
study, and of the variations in processingthat
are responsible for differencesin the effectiveness of study. Other data (e.g., Craik &
Tulving, 1975) can be used to suggest that
differences in time per se are irrelevant to
differencesin retention that are produced by
manipulating orienting tasks.
ExpenrueNr 2
The results of the first experiment could
be summarized by the statementthat increas-

ing the effort required to solve a problem


enhances later retention performance. The
second experiment provides further evidence
on the role of effort by directly varying the
difficulty of the problems themselves.In one
condition, the crossword puzzle problems
were extremely easy to solve. Puzzlesfor that
condition were constructed by deleting a
single interior letter from the solution word
(e.g.,check m-ney); the result for most pairs
was to make the problem so easythat it seems
possible to just read the solution word.
Puzzles for a second condition were made
more difficult by deleting two interior letters
from the solution word (e.g.,lance s5r). As
can be seen from the examples, deleting a
secondletter appearsto produce a substantial
increasein the difficulty of the problem. The
more difficult problems were expectedto yield
higher retention performance"
The second experiment also differed from
the first in that the effectof intermediate levels
of spacing were investigated in the second
experiment. In the current memory literature
there is some disagreementwhether there is a
dichotomous effect of immediate vs spaced
repetitions or a true continuous effect of
spacing repetitions; that is, some studies find
differencesonly between massed and nonzero
levels of spacing while other studies find
differencesamong nonzero levels of spacing
(see Hintzrnan, 1974, for a review). In the
present situation, this observation can be
translated into a speculationabout the role of
short-term memory" A prior presentation of
the solution to a problem might reduce later
retention only if that solution is still in shortterm memory when the problem is encountered" If so, one would expect a difference
betweenimmediate and widely spacedrepetitions but would not expect increases in
spacing outside the range of short-term to
influence later retention.
The interaction of spacing with problem
difficulty is also of interest. With massed
presentationof the solution and problem, the
two levels of problem difficulty should yield

RETENTION EFFECTSOF SOLVING VS RE},IEMBERING

655

equivalentlevelsof later recall:for both types


of pr ob l e m s ,th e ta s k o f p ro v i d i n g a s ol uti on
should be trivial. At greaterlevelsoi spacing.
however,the more difficult problems should
producehigher retentionthan the easierones.

these l 2 condi ti onsrequi red repeti ti onsoi a


pai r, a l i st w as 200 pai rs l ong. For the
repeti ti oncondi ti ons.presentati ons
of a pai r
were separatedby 0, 10,20 or 40 intervening
pairs.Twelve lists were constructedby rotati ng pai rs through condi ti ons so that across
lists each condition was representedb,v-the
Method
same pairs:six of theselists were presentedto
Desigttttntl subjec'rs.
The secondexperiment tw o subj ectsw hi l e the remai ni ngsi x l i stsw ' ere
employed the same crosswordpuzzletask as presentedto only one subject.
did the first experiment.However. all repetiA final cued-recalltest was constructedin
tion conditions in the second experiment the samemanner as describedfor Experiment
involved first readingthe responsememberof
l. The procedure was also identical to that
a pair and then later encounteringthat pair as describedfor the first experiment.
a problem that required the previously read
,,lncrlr'si.s.
As in Experiment l. the cuedresponseas a solution (the Read-Construct recall data that will be reported were not
arrangementin Experiment 1). Eight repeti- conditionalized on the subject correctly
tion conditions were produced by lactorially solvin_sthe corresponding crossword puzzle
combining two levels of solution difficulty probl em. A gai n. condi ti onal i zeddata w ere
( E as yv s H a rd ) w i th fo u r l e v e l so f s p aci ngof al soanal yzed,but the resul tsof thoseanal yses
presentations(0, 10, 20, or 40 intervening do not al ter any concl usi onsdraw n on the
pair s ) .In fo u r a d d i ti o n a l c o n d i ti o n s.a pai r basi sof the uncondi ti onal i z' ed
data.
was presentedonly once. To produce these
Significancelevel for all tests was set at
four conditions, the two levels of problem p < . 0 5 .
dif f ic ul tyw e rec o mb i n e dw i th th e s o luti onto
the problem being either read or constructed.
Problem difficulty was a pseudovariable,in- Re.su/rsuntl Di sctL.ssiorr
serted for purposes of analyses.when the
s olut io n s to th e o n c e -p re s e n te dp r obl ems
were read. All conditions were represented
within-subjects.
T he su b j e c tsw e re l 8 s tu d e n tse n rol l ed at
M c M aste r U n i v e rs i tyw h o w e rep a i d 52.00rhr
t o par ti c i p a tei n th e e x p e ri me n t.
iVIuteriuls untl procetlure.The materials
comprised 120pairs of relatedwords selected
from the Connecticut free-association
norms
using the same criteria as in Experiment 1.
Solution words varied from four to six letters
in length. Easy problems were produced by
replacing one interior letter of the solution
word with a blank; difrcult problems were
produced by replacingtwo interior lettersof
the solution word with blanks.
To constructa list, 10pairs wereassignedto
each of the 12 conditions described in the
design and subjects section. Since eight of

D i fferencesi n the probabi l i ty of an unsuccessfulattempt at solving the crossword


problems verified that the "hard" problems
w ere i ndeed more di ffi cul t than w ere the
" easy" probl ems: the probabi l i ty of bei ng
unabl e to sol ve a probl em i n the oncepresentedcondi ti onsw as .12 for hard problems and .02 for easyproblems.Prior reading
of the sol uti on faci l i tated sol vi ng of the
probl ems w hen readi ng of the sol uti on i mmediatelyprecededpresentationof the problem (0-spacing); the probability of bein_s
unabl eto sol vea probl emunderthoseci rcumstanceswas quite low (.005Ifor both the easy
and the difficult problems. When 40 items
intervenedbetweenreading the solution and
presentati on
of the probl em.the probabi l i tyof
being unable to solvea difficult problem (.04)
was still lower than that in the once-presented
condition where the solution was not read

656

LARRY L. JACOBY

70
J

60

oi

t*t*'

,i

q 50
(\

5 40
=

.24

fl

F<
ti 20
n-

to

20

40

o^rcE
PRESNTED

SPAC/IVG
F r c . 2 . P r o b a b i l i t y o f c u e d - r e c a l la s a f u n c t i o n o f
construction difficulty and spacing.

pr ior t o p re s e n ta ti o no f th e p ro b l e m (.12).
W it h t he e a s y p ro b l e ms . i n c o n tra st. the
pr obabil i tyo f b e i n gu n a b l eto s o l v ea p ro btem
when 40 items intervenedbetweenreadingthe
s olut iona n d p re s e n ta ti o n
o f th e p ro b l e m(.02)
was equa l to th a t i n th e c o rre s p o n d i n gonce_
pr es ent e dc o n d i ti o n .
The cued-recall results lrom the second
ex per im e n t a re d i s p l a y e d i n F i g u re 2. A
portion of thoseresultssimply replicateeffects
found in the first experiment. Among the
once-presenteditems, being required to cons t r uc t a s o l u ti o n p ro d u c e d s u b s ta nti al l y
hi_eher
cued-recallthan did reading the solution.F(1. l7)-80.76",VS":.02. Within the
conditions that required construction. reading t he s o l u ti o n i mme d i a te l yp ri o r to s ol vi ng
a problem that required that solution (0spacin_e)lowered later cued-recall as compared to the corresponding once-presented
c ondit ion s th a t s o l v e d th e p ro b l e m w i t hout
pr ev ious l y re a d i n g th e s o l u ti o n . F( l .
171: 32. 64 ,MS . :.0 2 .

It was earlier suggestedthat the resultsof


the first experiment reflect the influence of
short-term memory. The suggestionwas that
pri or readi ng of the sol uti on w i l l depress
later cued-recallonly if the solution resides
in short-term memory after the problem
requi ri ng that sol uti on i s presented.Thi s
posi ti on l eads ro the predi cti on that i ncreasesin spacingbeyond the range of shortterrn memory should have no effecton later
cued recall; that is, one should find an
immediatevs spacedeffectbut should find no
diflerencesamone -sreater levels of spacin_e.
The resultsof the secondexperimentrevealed
a quite large effectof spacin_spresentations.
f(3. 5 I ;: 33.98, MS . - 02. C ontrary ro expectations, however, the effect oi spacing
presentationsremained significant when the
O-spacingconditions were dropped from the
a n a l y s i sF, ( 2 , 3 4 \ : 3 . 9 3 , M S . - . 0 2 . I t d o e sn o r
seem reasonableto argue that the effectsof
spacin_e
within the rangeof lG40 intervening
items are due to differencesin the probability
of the solution residingin short-termmemory
durin_ethe presentationof the problem: the
l evel sof spaci ngi nvol ved are al l outsi de of
what is usually consideredto be the range
of short-term memory. Some factor thit
operatesover a _ereater
rangethan doesshortterm memory is apparently responsiblefor
the spacing effect observed in the present
experiments.
A l though the more di ffi cul t probl emsw ere
expected to produce higher retention than
werethe easyproblems.resultslrom the oncepresenteditems reveal no effect of problem
difficultyon later cued-recall.However.effects
of problem difficulty are observedwhen one
examines the repeated items. Among the
repeated items. the more difficult problems
producedhighercued-recallthan did the easy
probl ems,F(1. 17\:9.76 ful S ,:.007. E xami nation of the data presented in Figure 2
suggeststhat problem difficulty interactswith
the spacingof presentations.
At 0-spacin_e.
the
two levels of problem difficulty produced
essentially equivalent levels of cued-recall

RETENTION EFFECTSOF SOLVING VS RE}TEMBERING

while the more difficult problems produced


higher performance than did the easy problems at the greater levels of spacing. The
interaction of spacing and problem difficulty,
however, was not statistically significant,
F< 1.
A significantinteraction was found rvhenan
analysiswas carried out on the two levelsof
problem difficulty at 40-spacingand the two
once presenteditem conditions that required
c o n s tru c ti o n o f a re s p o n s e ,f(1, l 7)-5.53.
M S .:.0 2 8 " E x a mi n a ti o n o f th i s i nteracti on
shows that problem difficulty had an effect
with the repeated items but not with once
presenteditems. Further, prior reading of the
solution enhancedrecall relative to the oncepresenteditems only for the difficult problems:
at the longest-spacing,the level of cued recall
produced by easy items is approximately
equal to that produced by the corresponding
once-presenteditems.
Theseresultscan easilybe interpretedin the
same terms as was Experiment 1. Presenting
the solution of a problem prior to the presentationof that problem providesthe subject
with two meansof _eenerating
a response:The
subjectcan either rememberthe solution that
he was given previously or he can use the
information provided by the problem to construct a solution (mixes of the two means of
-eeneratinga response are, of course. also
possible).When a presentationof the solution
immediately precedesa presentationof the
problem, the subjectalmost certainly remembers rather than constructsthe solution. and
later retention performancesuffers"The effect
of spacingof presentationsfor both levelsof
problem difficulty can be interpreted as being
due to a correspondin_e
increasein the likelihood that a solution to the problem must be
constructed rather than remembered.
When a solution was not presentedprior to
the presentationof a problem as was the case
with once-presentedpairs. the subject had no
option but to construct a solution. It appears
that the only important factor for later retention was that construction be required: the

657

difficultyof the problemdid not influencelater


cued-recallperformance.This lack of an effect
of problem difficulty may simply result from
problem difficulty having been manipulated
over too narrow of a range; however. the
manipulation was sufficientto produce substantiallymore unsuccessful
attemptsto solve
the difficult problems as compared to the
easierones.Perhapsthe most surprisingresult
is the lar_se
advantagein cued-recallproduced
by an easy construction as compared to
readin-ethe solution to a problem. As shown
by the example provided earlier,the deletion
of a singleletterappearsto make the problems
so easy that one can just read the solution:
however,solvingproblemsthat were eventhis
easy produced subsequent recall that was
double that produced by actually readin_ethe
solution. Additional researchis required to
determine whether or not a continuous effect
of problem difficulty can be obtained. If the
effects prove to be dichotomous, as is suggestedby the resultsof the presentexperiment.
it may be necessaryto invoke the concept ol
consciousnessto explain the effectof problem
difficulty. To enhancelater retention. it may
only be necessaryto disrupt the flow of
processingso that some minimal amount of
consciousconstruction is required.
The effect of problem difficulty found with
repeated items remains to be explained. In
these cases,problem difficulty may have had
its effectby influencing the easeof remembering the solution. Even at the longer spacings.
subjects may have sometimes remembered
rather than constructed the solution. This
rememberin_s
of the solution is more likely
with the easyproblems where only one letter
of the solution word is deletedthan with the
hard problems where two lettersof the solution are deleted.That is. becausethere are
more letters and therefore a more restrictive
context. the easy problems provide a better
cue for recall of the previously_eivensolution
than do the hard problems: consequently.
constructon of the solution is required more
often for the hard problems with resulting

658

LARRY L. JACOBY

better retention. Evidence that the prior tion and that of the problem. An implication
reading of the solution does influencesolving of emphasizing retrieval is that when rethe problem even at the longest spacing is membering of the solution is enhanced by
provided by both experiments"In both experi- providing more effectivecues. as in the easy
ments,the probability of being unablero solve construction as compared to the hard cona problem was lower when the solution had struction conditions, subsequent retention
been read previously. This reduction in the performance will suffer even when presenprobability of being unablero solvea problem tation of the solution does not immediately
is presumably due to the solution being at precedethat of the problem. The presentation
least partially rememberedin some instances of lesseffectivecuesfor retrieval makes it more
rather than being solely constructed.
likely that the subject will have to solve the
The possibility of rememberingrather than problem rather than remember the solution.
constructinga solution, even when the solu- and subsequentretention benefits.
tion does not immediately precede presenThe above account of the results claims
tation of the problem, castsa new light on the that rememberinga solution always leads to
role of short-term memory in producing the poorer later remembering of that solution
effectof spacingrepetitions.Greeno (1967)has than doesconstructionof the solution.Such a
emphasizedthe role of short-term memory in position is too extreme in that rememberin_s
producing the spacing effect by arguing that sometimesinvolves construction. As one
exa subject might learn nothing from the pre- ample, Lindsay and Norman (197'7) argue
sentation of an item if that item currently convincin_elyt hat construction or reconstrucresidesin short-terrn memory. This is said to tion is involved when we answer a question
be becausethe subject will not select a new about where we were on some specifieddata in
"code" for an item that residesin short-term the distant past.
Perhapsa distinction needsto
memory during its repetition; memory over be drawn between effortful and
effortless
the long term is described as requirin_ethe retrieval (e.g.,Gotz & Jacoby, 1974).
Effortful
selectionof an appropriatecode.Similarly,in retrieval involves many of the same processes
the present paper. it was suggested that as does construction and acts the sameway as
presentation of a problem may have little construction to enhance later retention.
In
effecton subsequentretention if the solution contrast, effortlessrememberingof a solution.
to that problem currently residesin short- regardlessof the spacingof the solution
and
term memory so that solving the problem is problem,is much like readingthe
solution and
trivialized.On the basis of the resultsof the doesrelativelylittle to enhancelater
retention
present experiment, however, it appears that performance"Further theorizing
at this point
effortlessrememberin_erather than residence is by necessity speculative" However,
one
in short-terrn memory is the important factor advantage offered by the procedures
emfor subsequent retention. Discussions of ployed in the present experiments is that
the
short-term memory have usually emphasized task is one that can be further analyzed to
limited-capacity notions so that it is the yield information about the processes
in
number of intervening items that is seen as which subjectsengageto deal with problem.
a
determinin_ewhether or not a particular item
The main questions left unansweredin the
will still reside in short-terrn memory when it above account are: What is
involved in the
is repeated.Implicating easeo[ remembering, construction of a solution and why
does
in contrast, emphasizesthe importance o[ the engaging in construction enhance
later recues provided for retrieval of an earlier pre- tention performance?These questions will
be
sentedsolution as well as the number of items considered in the
discussion.
Before
_seneral
interveningbetweenpresentationof the solu- considering those questions, however. the

RETENTION EFFECTSOF SOLVING VS RE}IEMBERINC

effectobtained herewill be compared


spacin_s
with that obtained in more typical memory
experiments,and the applicability of current
theoriesoi the spacingeffectto the resultsof
the presentexperimentswill be discussed.
Ge x rn e l D rs c u s s ro N
Whereasit is possiblethat the spacingeffect
found here has a totally different basis than
does the spacingeffectfound in more typical
memory experiments(e._e..
Melton. 1967),it
seemsmore likely that the two are closely
related.In order to memorizea word, a subject
must engagein some seriesof operations:for
example, finding relations among words or
ima*uingthe words.As with the math problems
considered earlier and the crossword puzzle
problems used in the presentexperiments.it
seemsunlikely that thesememorizin_s
operations are fully repeated when the second
presentationof a word immediatelyfollowsits
first presentation.In the remainder of this
paper, I procecd as if the spacing effectfound
here and the spacing effect iound in more
t y pi c a l m e m o ry e x p e ri m e n tsh a v ea common
basis.If this common basisis accepted,it is of
int e re s t to s e e h o w v a ri o u s th e ori esof the
spacingeffectfare in attemptingto accountfor
the resultsof the presentexperiments.
One explanation of the spacing effect has
appealed to differencesin the frequency of
rehearsal as a function of the spacing of
repetitions. The claim is that an item is
rehearsed durin_e the interval intervening
between its presentations; consequently,
spacedrepetitionsof an item result in more
rehearsalof the repeateditem than do massed
r epe ti ti o n s(R u n d u s .1 9 7 1 ).
T h i s g r earernumber of rehearsals is used to explain the
retention advanta_ue
of spacedrepetitionsby
assumingthat lon_e-term
memory of an item is
a direct function of the number of rehearsals
t hat i te m h a s re c e i v e dA. l th o u g h i t may appl y
in other situations,the frequencyof rehearsal
explanation cannot account ior the spacing
effect obtained in the present experimenrs.

659

First. the incidental learnine procedure


employed here made it uniikely that subjects
would rehearsean item during intervalsoutsi deof i ts presentati on.
More i mportantl y.the
differential rehearsalexplanation cannot account for the debilitatin-eeffectof reading the
solution to a problem immediately prior to
solving the problem. [t is not reasonableto
claim that the prior reading of the solution
resultedin the solution being rehearsedless
than it would have been had the solution nor
been read prior to presentationof the problem.
The encodi ng vari abi l i ty hypothesi shas
provided a secondexplanationof the effectof
spacingrepetitions.By this hypothesis.there
are severaldifferentwaysa to-be-remembered
word can be encoded:the more differentwavs
a word is encodedthe better will be retention
since each different encoding provides an
additional access route to the word in
memory. It is further assumedthat an increase
in spacin_e
makes it more likely that repetitions o[ an item will be encoded differently.
Thus, the effect of spacing of repetitions is
attri buted to an i ncreasei n the number of
encodingsof the repeateditem (Melton. 1961
Madi gan. 1969).A vari ant of the encodi ng
vari abi l i ty hypothesi sassumesthat an i tem
becomesconditioned to contextual elements
that are active during the presentationof the
item. The spacingeffectis then explainedon
the basisof differences
in the similarity of these
contextual elementsas a function of spacinu
(Anderson& Bower. 1972:Glenberg, 1971).
Thereseemsto be no way that anything like
encodin_evariability could have operated to
produce the spacing effect observed in the
presentexperiments.The encodingvariability
hypothesis appears irrelevant when one
abandonsthe procedureof presentinga list of
words to be memorizedand instead presents
a seriesoi problems that are to be solved.
Notions discussedearlier. however, do provi de a meansof rei nterpreti ngdata that have
been presentedas supporti ng the encodi nu
vari abi l tty expl anati onof the spaci ngeffecr.

660

LARRY L. JACOBY

(e .g ., M a di gan. 1969) the subj ect rememberi ngthe sol uti on to a


S e v e ra l i n v e s ti _ s a to rs
have demonstratedthat the effectof spacin_u probl em rather than constructi nsthat sol u repetitions can be reduced by varyin_uthe ti on. Thi s rememberi ngof the sol uti on pro c o n te x t i n w h i c h th e re p e a ted w ord i s duces poorer retenti on so the l ocus of the
presented.If the context biases a different spacing effect is in the re_gistrationof rhe
interpretation for each presentation of the secondpresentati on.
The concl usi onthat the
repeated word (e.*e..fever-CHILL. snow- re-gistrationof the later presentationis defiCH IL L ) th e s p a c i n ge ffe c ti s fl a t ter than i t i s ci enti s compel l edby the fi ndi ngi n the presenr
wh e n th e c o n te x tb i a s e sth e s a m emeani ngfor experi mentsof an ubsol uredebi l i tati ngeffect
. h i s b i a s i n g of di fferent of repeti ti on w hen readi ne the sol uti o n
ea c h p re s e n ta ti o n T
int e rp re ta ti o n sb y m a n i p u l a ti ng conte.\t i s i mmedi atel yprecededpresentati onoi a probass u m e dto mi m i c w h a t h a p p e nsi n ordi nary l em. Further.the i nfl uenceoi havi ng read the
c ir c u m s ta n c ew
s h e n re p e ti ti o n so f an i tem are sol uti on i s not seen as bei ng opti onal : i t i s
wid e l ys p a c e di:t i s c l a i me dth a t b oth mani pu- nearl y i mpossi bl e to be uni nl l uenced
by
lat i o n si n c re a s eth e n u mb e ro f a c c essrouresto havi ng j ust read the sol uti on w hen one i s
t he re p e a te di te m. A n a l te rn a t i vei nterpre- s o l v i n ga p r o b l e m .
t at i o n . h o w e v e r"i s th a t th e c h a n gei n context
A l thou-ehI agreew i th cl ai ms of the habi essentiallyproduces different problems that tuati on hypothesi s.
w har i s habi tuati on'That
l
ar e to b e s o l v e d .C h a n g i n g c o ntext i s anal - i s,w hat processes
are i nvol vedi n habi tuati on'l
ogo u sto fi rs ta s k i n ga p e rs o nto a dd 37 and l 5 It may be possi bl eto descri behabi tuari onby
and th e n a s k i n g th e m to mu l ti p l y 37 and 15. appealin_u
to notions that have beenusedhere
A lt h o u g h th e n u m b e rsre m a i nth e samei n the to expl ai n the effectof spaci ng reperi ti on s.
two problems.the answerto the first problem P erhaps a habi tuated sti mul us i s one for
c an n o t b e c a rri e do v e r to tri v i a l i zethe sol vi n_s which an encodingcan be rememberedrather
of t h e s e c o n dp ro b l e m "S i mi l a rl y .operari ons than constructed.Thi s vi evr oi habi tuati on
c arri e do u t to e n c o d ea n i te m i n one context contrasts with a view' recently proposed b_rm a y n o t p ro v i d e a n e n c o d i n _th
e at i s appro- W a g n e rl l 9 7 6 l . W a g n e rc l a i m st h a r n ' h e na n
priate to the item repeated in a different event i s al ready representedi n short-rern - l
c on te x t.T h e ma n i p u l a ti o no f c o n text resul ts memory.furtheroccurrences
of that eventare
in m o re h rl l p ro c e s s i n go f l a te r p resentati ons renderedl esseffecti vethan they w oul d otherof t h e re p e a te di te m . a n d c o n s e quentl y.en- w i se be. A si mi l ar h,vpothesi sabour the
han c e sre te n ti o n .
i mportanceof short-termmemory w as tested
A th i rd e x p l a n a ti o no f th e s p a ci ngeffecti s i n E xperi mentI of the presenti nvesti gati on.
s im i l a r to th e a c c o u n t o ffe re d here. B y a There i t w as concl udedthat the sol uti on
to a
h a b i t u a t i o nh y p o t h e s i (sH i n t z m a n .1 9 7 4 )t.h e probl emdi d not have to resi dei n
short-term
spacing effect is due ro the deficient regis- memorv to i nfl uencethe sol vi ngof
the probt r at i o n o f l a te r p re s e n ta ti o nw
s h e n repeti ti ons l em: al l that appearednecessary
w as thar tl te
of a n i te m a re m a s s e d H
: i n tz m a n . B l ock.and sol uti onto rhe probl emcoul d be " effortl essl 1 , "
S um m e rs (1 9 7 3 \ p ro v i d e e v i d e n cethat the r e t r i e v e d S
. i m i i a r l y f o r h a b i t u a r i o n .i t m a \
enc o d i n c o f l a te r p re s e n ta ti o n si s defi ci ent. onl y be i mportant that a pri or encodi ng
oi
This deficient regisrrarion is described as an eventi s retri evedso an encodi n_e
neednot
bei n gd u e to h a b i tu a ti o na n d i s c onsi deredto be constructed.Thi s assumesthat i t i s
the
be o u ts i d eo f th e s u b j e c t' sv o l u n tary control . necessi ty
of constructi onthat _si ves
ri seto the
I n o u tl i n e .th e h a b i tu a ti o nh y p o t hesi sagrees ori enti n_eresponseobserved
i n studi es of
wit h th e n o ti o n sd e s c ri b e di n th e i n troducti on h a b i t u a t i o n .
to explain the spacing effect. There it w,ils
A seri es of experi mentsbv W augh and
s u_ u g e s tethda t a m a s s e dre p e ti ti o nresul tsi n N o r m a n ( l 9 6 t t ) m a y b e r e l e r a n t r o
under-

RETENTION EFFECTSOF SOLVING VS REMEMBERING

standing the effectsof spacing repetitions and


the processesunderlying habituation. Waugh
and Norman were interestedin specifyingthe
nature of an event that would displace an
earlier event from short-term memory. The
results of their experiments revealed that a
new and unpredictable event would displace
an earlier event; however, a repetition of a
recently presented event would not. If we
identify short-term memory with consciousness,it appearsthat the processingof a massed
repetition is automatic in that it does not
heavily involve consciousness.Combining
this piece with arguments made earlier we
arrive at the following picture: Presentationof
an event whose solution or encoding can be
easily remembered does not give rise to an
orienting responseor heavily involve conpresentationof such an event will
sciousness;
also have little impact on later retention.The
necessityof construction, in contrast, givds
rise to an orienting response,involves consciousnessto a,qreaterdegree,and producesa
substantial effect on later retention performance. The spacing of repetitions has its effect
by determinin_e
whether a solution or encoding can be rememberedor must be contructed.
The Generulitvof' Elfects of Constructiotl
Effectscan be found usingmanipulationsin
addition to those employed in the present
experiment and, therefore. the speculation
about differentmodesof solvinga problem or
responding becomesmore interesting.Before
-eoingon to deal with some negativeeffectsof
remembering a solution or encoding, one
positive effectwill be cited. A consistenrfinding reported in many reaction-timestudiesis
that the responseto an eventthat is repeatedis
quicker than the responseto an event that
occurred earlier but was not the last one to
occur. Bertelson (1963) has proposed rhat
when a stimulus is presented a subject first
checksmemory to seeif the presentedstimulus
is the same as the one that immediately
precededit. If the stimulus is the same, the
subject makes the same responseas he did

661

previously:if it is not the same.the subjecrhas


to retrievea responsethat is appropriate to the
presentedstimulus.The retrieval of a response
takes additional time so responding is more
rapid when the retrieval is not necessary;that
is, when the subject can simply _eivethe same
response as was _eivento the immediately
preceding stimulus. Bertelson's distinction
between repeating a responsevs retrieving a
response is essentially the same as the distinction that has been drawn here betrveen
remembering a solution vs constructin*sa
solution. Repeatinga solution is more efficient
than is constructingone in that repetitionof a
solution can be done fasterand. perhaps.with
less involvement of consciousness.Further.
there is some evidence(Keele, 1969)that can
be interpretedas showing that the repetition
effect found in reaction-time studies.like the
effectsfound in Experiment 2, are not limited
to short-term memory.
Slamecka(Note 1) has reportedresultsthat
are similar to those found here with the oncepresenteditems.Slameckafound that generating a responseto an item (e.g.,a rhyme or an
associate of the presented item) produced
better later retention than did reading the
same response.One factor that differentiates
reading a response from contructing a response is that the task of constructing a
responseis a more difficultone.Severalexperiments have shown that a difficult inital task is
associated with high levels oi retention.
lllustrations of the relation berweenthe difficulty o[ an initial retrieual and subsequent
retention level have been provided by Gotz
and Jacoby (197a) and Whirten and Bjork
(1977) among others. A parallel series of
demonstrationshas relatedthe difficulty of an
initial decision to subsequentrerention level.
For example, in one experiment by Jacoby.
(in preparation)subjects
Craik, and Be_eg
were
required to specify which word in a pair
referred to the lar_qerobject: later retention
was higher when membersof a pair were close
in size(flea-ant)rather than highly discrepant
in size (flea-elephant).Aubel and Franks

662

LARRYL.

(1978)havedemonstratedthat the difficultyof


comprehension influences later retention. It
was found that requiring additional effort
toward comprehension of a sentence enhanced recall so long as the sentence was
eventually understood.
There are some reasons to suggest that
repeatingan item a large number of times has
effectsthat parallel those of massing repetitions of an item. Dependent upon the similarity of problems and other factors that
contribute to interference, it is possible to
remember the solution to a problem that has
been solved rather than it being necessaryto
construct the solution when the problem is
again encountered"That interferenceis important can be illustrated by asking the reader to
find the sum of 37 + 15+ tz,the math problem
that was presentedin the introduction of this
paper. It is likely that the solution to that
problem was easily rememberedrather than it
being necessary to again go through the
operationsof addition to solve the problem.
This is true even though the presentationsof
the problem are widely separated. If interference had been increased by requiring the
reader to solve a number of other math
problemsprior to repeatingthe one problem,
however" it would be necessaryto again -eo
through the operationsof addition to obtain
the solution for the secondpresentationof the
problem. Interferencecan apparently be offset
by increasingthe number of presentationsof
the repeated problem.
Many of the characteristicsthat have been
used to describe"automatic" responding that
results from repetition (e.g., Norman. 1976:
Shiffrin & Schneider. 1,977)are the same as
those used here to describe responding to
massed repetitions. An item that has been
'represented a large number of times is
spondedto more rapidly, and does not appear
to tax the limited capacity processorthat can
be identified with consciousness.Effects on
memory of presentinga well-learneditem or a
problem for which the solution can be effortlessly remembered might be expected to

JACOBY

parallel those effectsthat are obtained when


massedrepetitionsof an item are presented.A
commonplace example is the difficulty in
remembering whether or not we turned off a
light switch. By the argument given here,
memory for turning off the switch is poor due
to the automatization of that activity through
a large number of repetitions.
The experiments in the present investigation were concernedonly with the effectsof
repeating an item or a problem. However.
similar effectsmay be obtained as a result of
extended practice with a rask. Through practice, perfonnance of a task usually becomes
more efficient; the task is accomplished
smoothly, rapidly, and with less effort. This
greater efficiencymay be gained at the expense
of memory for the individual encounterswith
the task. Perceptionof words and other events
can be considered as skilled tasks, and thus
amenableto this analysis.Kolers (1975)has
described the results of his experiments on
readin-e transformed text in these terms.
Kolers found that students that are unpracticed in reading inverted text remember
sentencesread in inverted text better than
sentencesread in normal text. However. after
extendedpractice in reading inverted text, the
memory advantage for inverted sentences
largely disappears.That is. increasedskill is
associatedwith poorer retention"One finding
that is particularly relevant to the present
investigationhas to do with manipulation of
the transformation performed on the text. In
one experiment (Kolers, 1973) transformations varied in the amount of difficulty they
produced for reading. This differencein difficulty, however, was not mirrored in later
retention performance: the effect oi reading
transformedtext appearedto be all-or-none in
that reading transformed text produced better
memory than did reading normal text but
there were no differencesamong the various
transformations. This lack of a difference
among transformations parallels the lack of
an effect of problem difficulty found in
Experiment 2 of the present investigation.

RETENTION EFFECTS OF SOLVING VS REMEITBERING

663

than constructing the solution may have made


it unnecessaryfor the subject to deal with
semantic relationships between the cue and
solution word. Construction, in contrast, requires more processing of meaning and this
more meaningful processing produces a
higher level of retention (Craik & Lockhart,
1972\.
One factor that is ignored by the levelof-processing and distinctiveness notions is
the affective consequences of repetition.
Generally,we do not enjoy sitting through the
samemovie twice, a joke heard for the second
time is less funny, repeatedly producing the
solution for the same problem is boring. The
idea is that the necessity of construction
involvesconsciousnessand engendersarousal
in a way that effortlessrememberin,edoes not;
it is' this involvement of consciousnessand
heightened arousal that is responsible for
dillerence in subsequent levels of retention.
The consequencesof repetition for arousal
and consciousness were briefly described
earlier in conjuction with the discussion of
habituation. There is also evidenceto suggest
that arousal influencesretention. One illustration of the effectsof arousal is the memory of
heanng about President Kennedy's assassination; people can typically recall in great
detail the circumstancesin which they heard
the news. Brown and Kulick (1977) have
describedthese vivid memories as "flashbulb
Why Should ConstructionEnhanceRetention? memories" and go on to speculatethat there
One interpretation of the effectsdiscussed may be some biological value associatedwith
here assumesthat a task is made easier or keeping an exact record of the circumstances
suppolls "automatic'" processingby deleting surrounding a significant event.In this vein, it
someoperations.Retention sufferssincedelet- seems quite reasonable to argue that reing these operations detracts from the dis- membering the solution to a problem has
tinctivenessof the encoding of the event and biological value; a great deal of efficiency is
reduces the number of potential bases of gainedif the solution to a difficult problem can
retrieval. That is, there is less learned about be rememberedrather than the problem being
the event to individualize it; there are fewer solved anew each time it is encountered.
operationsto be recognizedand consequently,
A second example of the effectsof heightretrieval suffers(Lockhart, Craik, & Jacoby, ened arousal comes from studies of animal
1976:Jacoby& Craik, 1978).In the caseof the learning.There it is found that the occurrence
crossword puzzle problems employed in the of a "surprising" stimulus is remembered
present experiments, remembering rather longer and producesmore learning than does

Proactive inhibition observed in studies of


verbal learning may result from subjects becoming more skilled at the task of memorizing
lists of words. As a function of paractice,fewer
trials are required to reach a performance
criterion on a list. That is, learning-to-learn
occurs; coinciding with this increase in
learning-to-learnis poorer retention for later
lists in the series. In contrast to the interferencetheory of forgetting usually employed
to interpret this proactive inhibition (e.9.,
Postman & Underwood, 1973),the present
position emphasizes the influence of prior
practice on the encoding of events at input.
A similar argument regarding proactive inhibition has been made by Warr (1964).
The discussion here has obviously gone
rather far afield in pointing out effectsthat
may be relatedto thoseobtained in the present
experiment.However, it does seem clear that
requiring construction of a response influencesthe subsequentlevel of retention in a
variety of situations. The mapping-out of
similarities among those situations is likely to
be useful. For example, it may be reasonable
to talk about remembering vs constructing a
procedure to deal with a particular task in
much the same terms as are used for talking
about rememberingvs constructing a solution
to a specific problem.

664

LARRY L" JACOBY

an expected stimulus (e._e.,Kamin, 1969: probl em,i r seemsqui te


l i kel y that both l * ei Wagner" t9761: these effectsmay be due to of-processi ng
and arousal are i nvol red i rr
heightenedarousal. Until recently. a more deternri ni ngretenti on
so that the task i s not
commonplaceeffectthat might be attributed to choose betw een
them but i s rather to
to diflerencesin arousal was supported by determi new hat
contri buti oni s made by each.
only anecdotal evidence;however" Kintsch Despiteits difficulty,
I t'eelthat the task is not
and Bates (1977)have provided more tradi- an impossible
one: we are currently carryin_e
tional evidenceby showin_e
that studentshave out experi ments
that w e hopew i l l separarerhe
excellent retention for jokes inserted in a cffectsof
arousal from those that have been
lecture: indeed. memory of jokes often sur- attributed to
differencesin level-oi-processing
passesthat of content material.The superior or distinctiveness.
memory of jokes may be due to the
_greater
successof jokes in capturing the interest of
SulrvrrRy nro CoxclUSroNS
students.Variablessuch as interestor arousal
are likely to have effectsin a lar_ee
number of
The present experiments clearly demonsituations including memory for prose, dis- strate that
solving a problem enhances
c us s io n s(K e e n a n .M a c W h i n n e y ,& Mayhew . subsequent
retention as comparecl to re_
1977),and so forth; however"little has been memberi ng
a sol uti on"W hy sol vi nga probl em
done to incorporatetheseeffectsinto theories should yield
this retention advanra_se
is an
of memory.
important question for future rcsearch.HowThc notions of distinctivenessand arousal ever, application
of the distinction betwecn
differ liom one anothbr in much the sameway solvin,ea problem
and rememberin_q
a solu_
asdo the notions of or_eanization
and stren_sth ti on need not w ai t on an expl anati onoi the
( c f .J aco b y ,Ba rtz ,& E v a n s "l 9 7 g ).An account retention
differerices"Even without such an
in terms of distinctiveness
or level-ofprocess- expl anati on"the di sti ncti onhas consi derabl e
ing attributeseffectsto differences
in the extent hcuristic value. By emphasizingthe effects
of
to rvhich an item is elaboratedlbr encoding. rememberi n_s
a sol uti on. the di sti ncti on enLike the or_uanizationtheories.thc claim is courages
the appl i cati onof a l ar_ge
l i terature
t hat en h a n c e me n ot f re te n ti o n re q ui resthat concerned w i th
the condi ti ons that l bster
more aspects of an event be appreciated: memorv.
That literature can be irsed to
particularly useful for rerenrion is finding sug_uest
mani pul ati ons that w i l l ai d i n the
r elat io n s h i pas mo n gi te msi n a l i s t"T h e noti on anal ysi s of
standard memory phenomena.
of arousal"in contrast"_sivesrise to what is The interpretation
of the cflect of spacing
c s s enti a l l ya s tre n g th th e o ry o f me mory. In repetitions
offeredhereprovidesone examole.
this instance,however.differencesin stren_r{th That i nterpretati on
cl ai ms that the poo,
are seenas beingdue to differencesin arousal retcntion
atier massedrepetitionsresultslitrrn
rather than to differencesin number of repeti- the encodi ng
of l ater presentati ons
of an i tem
tronsas is usuallyassumed.If one is to ar_gue
in bein_eremembcred rathcr than construcred.
terms oi biological value. a strengthening The menrory
literaturecan be usedto _senerate
effect that results from arousal or surpnse si tuati ons
other than massedrepeti ti crnthat
seemsat leastas valid as onc that arisesfrom will ioster
easy rememberingof a prior enan event bein_srepeated.
coding or solution. For example,when repeti_
The notions of level-of-processing
and dis- tions of a word are separatedin a list that is
to
tinctivenesshave been criticized for being be learned,
rememberin_g
oi the prior encodin_s
vague (e._e.,Baddeley, 1978). The idea of tbr a
repeated word should be more likely
ar ous ali s o b v i o u s l ya t l e a s ta s v a g u ea s that of when repetitions
are separatedby sorne unIevelof processing. To further compound the rel ated
acti vi ty (e._s.,
addi ng numbers)rather

RETENTION EFFECTS OF SOLVING VS REMEMBERINC

than by the learning of other words. Consequently, the effect of spacing repetitions
should interact with the nature of the activity
intervening between repetitions;the effectsof
spacing should be less pronounced when the
intervening material is distinct from the items
that are to be remembered.Similar lines of
argument can be used to propose manipulations that will amplify or reduce proacrive
inhibition. As suggestedearlier, proactive
inhibition may in part result from a subject
rememberingrather than constructing a procedure for dealing with a particular task or
class of situations.If so, manipulations that
interferewith this rememberingof procedures
should reduce proactive inhibition.
The distinction berweensolving a problem
and rememberinga solution may also help to
clarify the notion of automaticity. Previous
authors have emphasizedextendedpracticeas
a necessary precondition for automaticity
(e.g., Norman, 1976: Shiffrin & Schneider,
1977).Similarly, in discussinglan_euageprocessing, Schank (1976\ has suggested that
repeated encounters with a _eivenclass of
situationare instrumentalin the evolutionof a
"script" that will guide the processing of
further situationsof the same kind. The view
taken here, in contrast,equatesautomaticity
with the rememberingof a solution or encoding; the rememberingof a solution eliminates
the necessityof carrying out the computations
that would otherwisebe required to arrive at
the solution. so performance appears automatic. This emphasis on remembering a
solution can be used to suggestthat factors in
addition to extendedpracticedetermineauromaticity. Rememberingof the solution and.
consequently,automaticity will also be influenced by the length of the delay since the
last encounter with the task or event, the
nature of the activity intervening since that
prior encounter, the similarity of the current
situationto the previousone,and so forth. The
implication is that automaticity is situationspecific:a responsethat is automatic in one
situation will not be automatic in a situation

665

that is lessfavorableto remembering.Further,


the number of repetitions required to produce
automaticity will depend on the values of
other variables that influence remembering.
Another potential application of the distinction between solving a problem and remembering a solution involves reading.
Programs of reading instructions have vacillated between employing "look-say" and
"phonics" methods of instruction. The
phonics method is designedto provide a serof
rules so that one is able to construct a
pronunciation by dealing with parts of a word
while the look-say method instructs the
learnerto rememberthe pronunciation for the
word as a whole. A question of continuing
concern relatesto the skilled reader:Does the
skilled reader remember or construct a pronunciation for a word? By the view taken here.
both rememberingand construction are likely
to be involved. When a word is presentedin an
unfamiliar context, for example,construction
of a pronunciation may be necessary.However,if that word is then repeatedafter a short
duration, it seemsunlikely that it is necessary
to fully repeat the prior construction to arrive
at a pronunication;rather, the pronunciation
can be remembered.Mixing of thesemodes of
word indentification is also possible. The
readermay engagein some constructionwith
the effect that the construction yields a sufficient number of additional cues to allow
rememberingof the pronunciation;rendering
further construction unnecessary. In connected discourse,repetitions occur with substantial frequency. Further understanding of
the effectsof these repetitions on processing
appearsessentialfor a realistic theory of word
identification. Similar arguments can be
applied to other aspectsof reading. For example, when an argument is first encountered
in a paper, comprehension of the ar_qument
may require a great deal of construction:
however, depending on the conditions for
memory, the argument may be remembered
rather than constructed when it is encountered again later in the paper.

666

LARRY L. JACOBY

In conclusion, to understand the effectsof


repetition we must specifyhow the processing
of the repeated event is altered by its prior
presentation"It is incorrect to conclude that
becausean event is repeatedthe processingof
that event is also repeated.Rather, repetition
of an event can result in the solution being
rememberedwithout the necessityof engaging
in the activities that would otherwise be
required to obtain that solution. The meansby
which a solution is obtained influencessubsequent retention performance; subsequent
retention suffers when the solution is rememberedrather than being constructed.The
reason for this retention advantage of construction is not clear:however,arousaland the
necessary involvement of consciousnessin
construction may play some role. The distinction between solving a problem and remembering a solution is potenrially useful for
understanding several phenomena including
the effects of spacing repetitions, proactive
inhibition, and automaticity. The distinction
also appearsimportant for an analysisof tasks
such as word identification.

Cnetr. F. I. M.. & Locrner,r. R" S. Levelsof processing:


A framework for memory research. Journal of
Verbal Learning and Verbal Behavior. lgjL- ll.
67t-684.

Cnex. F. I. M.. & Turvwc. E. Depth of processingand


the retentionof wordsin episodicmemory.Journal
of Experimental Ps-vchology:General. 1975. 104,
26V294.
GrsNaERc.A. M. Influencesof retrievalprocesses
on the
spacingeffectin free recall.Journalof E-rperimenral
Psychology:
HumanLearningandMemory. lgjj.3,
282-294.
Gorz, A." & Jecony. L. L. Encoding and retneval
processsin long-term retention. Journal of
E-rperiment
al Psychology" 1914. 102,29 l-Zgi. .
GnrsNo,J. G" Paired-associate
learningwith short-term
retention:Mathematicalanalysisand data regarding identification of parameters. Journal of
Mathematical Psychology.1967.4, 430-172.
Hnnzuru. D. L. Theoreticalimplicationsof the spacing
effect.In R. L. Solso (Ed.). Theoriesin cognirive
psychology:TheLoyolaSymposium.
Hillsdale.N. J.:
l:urence ErlbaumAssociates.1974.
HrNrzunN.D. L."Br-ocr.R.A..& Suvvrns.J.J. Modality
tags and memory ior repetitions:Locus of the
spacingeflect.J ournulof Ver baI Leurningund Ver bal
Behauior,1973 12, 229-238.
"
hcosv. L. L.. Bmrz" W. H". & EvnNs.J. D. A functional
approach to levels of processing.Journal of
E-rperimental Psvchology: Human Learning and
Memorl41978.4. 331-346.
Jrcosv. L" L.. & Cnlx. F. I. M. Effectsof elaborationof
processingat encodingand retrieval: Trace distinctiveness
and recoveryof initial conrext.In L. S.
Cermak and F. I" M. Craik (Eds.). Levetsof
processingand human memory. Hiilsdale. N.J.:
LaurenceErlbaum Associates.1978.
RereneNcrs
K.lulN, L" J. Predictability.surprise.attentionand conAvornsox. J. R.. & Bown. C. H. Recognitionand
ditioning"In B. A. Campbelland R. M. Church
retrieval processesin free recall. psychological
(Eds")" Punishment and aversive behavior.
Review,1972.Tt, 9j-123.
EnglewoodCliffs. N.J.: prentice-Hall.1969.
Aussr. P. M.. & Fnexrs. J. J. The effectsof efforttoward Krerr. S" W. Repetition effect:
A memory-dependent
comprehensionon recall. Memory and Cognition.
process.Journal of Experimenrulpsychology.1969.
1978.6,2V25.
n,243-249.
Blpoelsy, A" D" The trouble with levels:A reexam_ KrsNnN. J. M.. MncWsrNxw.
8.. & Mnnrrw. D.
ination of Craik and Lockhart,s framework for
Pragmaticsin memory: A study of natural conmemory research.psvchologicalReview. 197g.g5,
versation. Journal of Verbal Learning and Verbal
I3yt52.
Behavior.1977.16,54F580.
BnraLsox, P. S-R relationshipsand reacriontimes to
Ktvrscn" W." & Blrrs. E. Recognirionmemory for
new versusrepeatedsignalsin a serial task.Journal
statementsflrom a classroom lecture. Journal of
of ExperimentalPsychology,1963.65, 47g-4g4.
Experimental Psychology: Human Learning and
BnowN, R., & Kuucr, l. Flashbulb memories.
Memory, 1977"3, l5G-159.
Cognilion,1977,5, 73-99.
KlrtN. K.. & Surz. E. Specifyingthe mechanisms
in a
BnuNrn" J. S. Some elementsof discovery.In L. S.
levels-of-processing
approachto memory.Journal
Shulman & E. R. Kiesler (Eds.). Learning by
of E-rperimennl Psychology:Human Learnins and
discovery.Chicago:Rand-McNally. 1966.
Memory, 1,976.
2, 67l-61-9.

RETENTION EFFECTS OF SOLVING VS REMEMBERING

Kouns. P. A. Rememberingoprations.Memory and


Cognition.1973.l, 347-355.
of automatized
KoreRs. P. A. Memorial consequences
encoding. Journal of Experimental Psychology:
HumanLearningand Memory, 1975.l, 689-701.
Ltxpsnv. P. H.. & NonneN. D. A. Human information
processing.
New York: AcademicPress.1977.
Locrunnr. R. S..Cnrrr. F.l. M.. &Jncosv. L. L. Depth
recognitionand recall:Someaspects
of processing.
of a generalmemory system.In J. Brown (Ed.),
Recalland recognilron.London: Wiley. 1976.
Mrorcex. S. A. Intraserialrepetitionand coding processesin lree recall.Journalof VerbalLearningand
VerbalBehavior.1969.8, 828-835.
Mrrrox. A. W. Repetitionand retrievalfrom memory.
Science.1967.158,532.
NewYork: Wiley.
Nonurx. D. A. Memoryandattenrron.
1976.
Posruex. L.. & UxoERwooD.B. J. Critical issuesin
interference
theory.MemoryandCognition.1973.l,
llH0.
in freerecall.
RuNous.D. Analysisof rehearsalprocesss
Journal of E-rperimental Psychology. 1971, E9,
63-77.
Scnrur. R. C. The role of memoryin languageprocessing. In C. N. Cofer (Ed.), The structureof human

667

memory.SanFrancisco.Calif.: W. H. Freemanand
Company.1976.
ScHrrnrx. R. M.. & ScHxrpR. W. Controlled and
automatic human information processing: II.
Perceptuallearning. automatic attending. and a
Review,1977. U,
general theory. Ps.vchological
t27-t90.
WlcNm,. A. R. Primingin STM: An informationproor retrievalcessingmechanismfor self-generated
generatd depression
in performance.In T. J. Tighe
and R. N. Leaton(Eds.).Habiruailon:Perspectives
, animal behavior. and
from child development
hysiology.Hillsdale.N.J.: LaurenceErlbaum
neurop
Associates.1976.
Wnnn. P. B. The relative importanceof proactiveinhibition and degreeof learning in retention of
paired-associate items. British Journal of
Psychology.1964.55, 1F30.
Wlucs. N. C.. & NonunN. D. A. The measureof
interferencein primary memory. Journal of Verhat
Learningand VerbalBehavior,1968..7.611-626.
Wxnrsx. W. 8.. & Blonr. R. A. Learninglrom tests:
Effectsof spacing.Journal of Verbal Learning and
VerbalBehavior.1917.16.465478.
(Received
August8, 1978).

Das könnte Ihnen auch gefallen