Mood state-Dependent Retrieval - The Effects of Induced Mood on

advertisement
This article was downloaded by: [University College London]
On: 2 February 2011
Access details: Access Details: [subscription number 917204867]
Publisher Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK
The Quarterly Journal of Experimental Psychology Section A
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713683590
Mood state-Dependent Retrieval: The Effects of Induced Mood on Memory
Reconsidered
Pamela M. Kenealy
Online publication date: 22 October 2010
To cite this Article Kenealy, Pamela M.(1997) 'Mood state-Dependent Retrieval: The Effects of Induced Mood on Memory
Reconsidered', The Quarterly Journal of Experimental Psychology Section A, 50: 2, 290 — 317
To link to this Article: DOI: 10.1080/713755711
URL: http://dx.doi.org/10.1080/713755711
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
TH E QUARTE R LY JOU RNAL O F E XPE R IM E NTAL PS YCH OL OG Y, 1997, 50A (2), 290± 317
M ood-state± D epen dent Retrieval: T he E ffects
of Ind uced M ood on M em ory Reconsidered
P amela M . K enealy
Downloaded By: [University College London] At: 02:00 2 February 2011
School of P sychology, R oeha mpton I nstitute London
A n alysis of stu d ies inve stig ating m ood -state ± d ep end en t retrieval iden ti® es m eth od ological
pro blem s th at m ay h ave con trib uted to th e con trov ersy s ur ro u n d ing th e reliability of th e
effectÐ in p ar ticu lar, th e p ossible con fou n d ing of en cod ing an d retrieval in p revious stud ies.
Five exp erim ents are repo r te d investigating th e effects of m ood on lear n ing an d r ec all.
M ood -state ± d ep end en t retrieval w as obser ve d in E xp erim ent 1a (usin g Velten ’ s M ood
Indu ction Pro ced u re); E xp erim en t 1b (u sin g a m usic M IP); an d E xp erim en t 1c (u sin g
Velten’ s M IP at encod ing and a mu sic M IP at retrieval). S u bje cts w h o lear n ed an d r ec alled
in d ifferent m ood s h ad sign i® can tly g reater d ecrem ents in recall th an d id su bjects in the sa m e
m oo ds. E xp erim en ts 2 an d 3 investigated th e effect of observ able retrieval cu es on m ood state± d ep end en t retrieval. In E xp erim ent 2, th e p re sen ce of observ able r etrieval cu es at recall
over ro de sta te-d ep end en t retrieval. In E xp erim en t 3, by m an ipu lating th e p resen ce or
absen ce of obser vable cu es at recall, both th e occ ur ren ce an d th e erasu re of the m ood-state
dep en d enc y w as d em on strate d . M oo d state du rin g lear n ing an d cu ed recall w as also sho w n to
affect p erfo r m anc e in a th ird sessio n u n d er con d iton s of free recall.
T here is consider able co ntroversy su rrounding the state-d ep end ent effects of m ood o n
m em ory (see B laney, 1986; K uiken, 1991). M ood-state± dependent retrieval has im portant
implicatio ns for theories of co ntext effects in m em ory (B add eley, 1982; Bow er, 1981,
D avies & T h om son, 1988; E ich, 1989; Tulving, 1983; Tulving & T hom son, 1971,
1973 ) an d th eories of d epression (B eck, 1976; Blaney, 1986; Ing ram , 1984; Reus,
Weing ar tner & Post, 19 79; Teasdale, 1983), but dem onstrations of the phenom eno n seem
to b e elusive an d unpredictable an d present som e em bar rassm ent for these theories. In a
critical review of the literature on affect an d m em ory, Blaney (1986) has noted that alth ough
the literatu re dem onstrating state-depen den ce involvin g drug-induced states is considerable, there are few relevan t studies involving elated or depressed m ood. O ther reviewers
have noted th at the evid ence for the presence or absen ce of m ood-state dependency rem ains
highly equivocal, w hereas the evid ence fo r m ood cong ruency is stron g (Bad deley, 1990;
Bower & M ayer, 1991; E ich, 1989; E llis & A shbrook, 1991). B ow er (1987) claim s to have
been ``unable to ® gure out an ything system atic that cau ses these con¯ ictin g results’ ’ (p. 451).
Bower confesses to not having carefully inspected all the varying proced ures of the
Requ ests for re prin ts sh ould be sen t to Pam e la K enealy, S chool of Psycho logy, Roeh am pton Institu te L ond on ,
W hiteland s C ollege, West H ill, Lo ndon SW 15 3SN, U.K .
Ó
1997 The E xp erim ental Psych olog y S ociety
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
291
su ccesses an d failures, but co ncludes that ``T he effect seem s a w ill-o’ -the-w isp that
ap pears or not in different experim ents in cap ricious w ays that I do not u nderstand’ ’
(B ow er, 1987, p. 451). K ihlstrom (1991), how ever, sugg ests that no special onus should be
attached to th e m ood m em ory literature sim ply becau se it can not p roduce reliable m ooddependent retrieval. K ihlstrom refers to two oth er classes of experiments, drug-state±
dependent retrieval an d environm ent-dependent m em ory effects, an d concludes that such
effects h ave been dif® cult to ob serve in hum an s under an y circu m stan ces. To recogn iz e
that the effects m ay have lim ited reliability an d be subtle in their operatio n seem s a m ore
useful view, as it m ay per m it a description of these effects w ithin the limits of their
reliability once this has been assessed.
T his pap er prese nts a system atic an alysis of the reported evidence relevan t to the
reliability of laboratory dem o nstratio ns of m ood-state± dependen t retrieval an d highlights
som e of th e problem s that m ay have co ntributed to the controversy surrounding it. Bower
an d M ayer (1991) have concluded that w hat is needed now is som e insightful an alysis of
the reason s, if an y, for th e positive an d neg ative results o n m ood-d ependent retrieval. In
the dom ain of psycholog y, this ® eld is unusual in that th e num ber of pu blished studies
reporting null ® ndings is con siderably g reater th an the num ber reporting positive ® ndings. M ood-state± dependent effects have been reported by Bower, M onteiro, an d G illigan
(1978, Study 3); L eight an d E llis (19 81); S chare, L ishm an , an d Spear (1 984, Study 3);
M ecklenbr au ker an d H ager (1984); an d G age an d Safer (1985 ). E ich an d M etcalfe (1989)
reported m ore pron oun ced m ood-depen den t effects in m em ory for inte rn al (generated
item s) than for exter nal events (read item s); an d E ich, M acau lay, an d Ryan (1994) dem onstrated m ood-dependent m em ory in three studies w here subjects generated au tobiog rap hical even ts in respo nse to neutral nouns. N ull ® nd ings or failu res to replicate have
been reported by Isen et al. (1978); N asby an d Yando (1982); B artlett an d San trock (1979);
Bartlett, Burleson, an d S an trock (1982); Bower et al. (197 8, Studies 1 & 2 ); S chare et al.
(1984, Studies 1 & 2); Bow er et al. (1981, Study 3); Bow er an d M ayer (1985); Wetzler
(1985); H aag a (1989); an d Foa, M cN ally, an d M urdock (1989). To take one exam p le of
the preoccupation w ith null ® ndings in the literature, E llis an d A sh brook (1991) note that
only very limited evidence of state-dependent effects w as obtain ed by L eight an d E llis
(1981, Study 2), an d that in eight possible instances w here state depend enc y could have
been p roduced, the effect w as reliably evident in o nly one.
It seem s therefore, that insightful an alysis of the reaso ns for the positive results o n
m ood-dep end ent retrieval m ay be dif® cult to sup port w ith published studies alo ne. N otw ithstanding this, certain m etho dological consideratio ns have been described w hich are
pertin ent to research con cer ning m ood-state dependency. E llis an d A shbrook (199 1) state
that the m eth odological issues in m ood an d m em ory research can be cate gorized according to three topics: sub ject factors, em otional-state factors, an d factors related to the
co gn itive task. E llis an d A shbrook (1991) provid e a useful review of these issues. H ow ever, there ap pear to b e som e m ore fundam ental p roblem s that can be identi® ed by
reference to stu dies reporting ne gative ® nd ings. T hese problem s concer n the m ethodolog ical rigour of both the design an d th e conduct of these experiments.
T he ® rst p roblem concer ns th e m ood m an ipulation stage of experim entation. T hree of
the stu dies that have reporte d null ® ndin gs do not provide objective m easures of experim entally m anip ulated m ood, leaving open the possibility that su bjects m ay not have been
Downloaded By: [University College London] At: 02:00 2 February 2011
292
KEN EA LY
in the ap propriate m o od state (sam e or different), an d that m ood w as not su ccessfully
m an ipulated (Bower et al., 1978; Isen et al., 1978; N asby & Yando, 1982). Futher m ore,
stu dies of m ood-state dependenc y in general have failed to specify a priori criteria th at
de® ne differen ces between m ood cond ition s. A failure to ® nd differen ces between the
co nditions undertaken in different m ood states m ight therefore simply b e the resu lt of an
insuf® cient difference between the m an ipulated m oods, or variance in th e m ood elicited
under a particular co ndition. E ich an d M etcalfe (1989) have noted that by imp osing a
stringent criterion, the odds of dem on strating m ood-depen den t m em ory w ou ld be
improved. A num ber of studies have utilized an hypnosis technique to induce m oo d
(B ow er, G illigan , & M onteiro, 1981; B ow er & M ayer, 1985; B ow er et al., 1978 ). T he
lim itations of the hypnosis techn iqu e have been w ell docum ente d an d include the effects
of experimenter bias an d dem an d characteristics as w ell as selectiveness in subject populatio n (Bower & M ayer, 1991; G oodw in & W illiam s, 1982; H asher, Rose, Zacks, San ft, &
D oren , 1985; N asby & Yando, 1982). Indeed, it is possible that counter-dem an d m ay
account for som e of the observ ed ® n ding s using this technique, w here the experim enter
inadv ertently presen ts subjects w ith a ``d em an d not to produce dem an d effects’ ’ (Polivy &
D oyle, 1980). W ith these criticism s in m ind, Bower (1 981) has repeated certain experim ents using hap py or sad m usic in the backg round as a m ore unob tru sive for m of m ood
m an ipulation. A lthough noted in Bower (1981), the d ata from these m usic m ood induction experim ents are not reported or referenced. O ther studies of m ood-state± dependen cy have used the Velten technique (as described later on) to induce elated , depressed , or
neutral m ood states (Velten, 1968), bu t they hav e only provided o ne m ood g roup for
co m parison w ith a neutral g roup (e.g. Schare et al., 1984; L eight & E llis, 1981). T he
inclusion of a second, m ore d istinct m ood g roup m ight have per m itted state-dependent
effects to em erge.
A m ore m ajor problem w ith these stud ies how ever, is the possible confounding of
en coding an d retrieval. First, few studies repor t m easures of the level of initial lear ning,
an d they d o not ap pear to have controlled for the fact th at som e subjects m ay learn less
than o thers initially. W hen initial lear ning is not m easured objectively, there is no w ay
to separate the effect of m ood on lea rning from the effect of m ood on retrieva l; the
co nfounding of encoding an d retrieval in m any studies is a serious ¯ aw. Seco ndly,
su bjects in o ne g rou p lear ning less initially than subjects in an other group m ay lead
to the asym m etrical dissociations at recall that have been reported in a num b er of
stu dies, w here state dependency has only b een o bserved for positive m ood states
(B artlett et al., 1982; Bartlett & S an trock, 1979; M ecklenbrau ker & H ager, 1984), a
possible inference b ein g that there w as less initial lear ning in the neg ative m ood-states.
Further sup port for th is con clusion is provid ed by L eight an d E llis (1981) w ho ob served
an asym m etrical m ood-state± dep endent effect using the Velten m ood induction pro cedure. T hey report that the initial lear ning scores of depressed an d neutr al-m ood subjects differed signi® can tly, w ith depressed subjects lear nin g less than neutral-m ood
su bjects. In a pilot study we investigated the effects of alcohol on state-dependent
retrieval (K enealy, 1982). A lcohol affected rate of lear ning w ith subjects in the alcohol
g roup lear ning less in Trial 1 an d 2 than sober subjects. In E xperim ent 1a reported
here, the sam e patter n w as ob served w ith m ood affecting rate of learn ing. O n Trial 1,
elated subjects lear nt signi® can tly m ore (m ean = 12.01) th an depressed subjects (m ean
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
293
= 6.50); in further trials the effects of m ood did not ap pear to be linearly related to
rates of lear ning. D esign ing experiments to ensu re equivalence of in itial lear ning by all
su bjects ap pears to be an im portant factor in dem onstrating sym m etrical state-dependen t lear ning, a point noted by O verto n in 1974 but largely ig nored by contem porary
ex perimentalists in this area. It m ay b e that m any experiments can not be discoun ted o n
these g rounds (e.g. B ow er, personal com m unication 1994); how ever, data o n the rate an d
eq uivalen ce of lear ning have not alw ays been reported, so th e problem can not be entirely
dism issed.
Bow er et al. (1978) ap pear to have provided the clearest dem o nstration of m oodstate± dependent effects, although the phenom enon w as observed o nly under the
en han ced interference of two-list co nditions. Blaney (1986) notes that eviden ce o n
this point is also m ix ed. It is claim ed that m ood-state± dependent ® ndings w ith interference d esigns have been dif® cult to rep licate (Bow er & M ayer, 1985; Wetzler, 1985),
althou gh these attem pts ap pear to have been m odi® ed replications. Blaney (1986) conclu ded that ``E ven this effect, w hen evident, m ight be obtainable o nly w hen exposure
an d recall m oo ds are induced by procedures that are som ew hat sim ilar to o ne an other’ ’
(p. 231). H aag a (1989) pointed out that w hen the m ood induction procedure is the sam e
at lear ning an d recall, there is no w ay to separate the effect of m o od sim ilarity at
lear ning an d recall from th at of induction procedure similarity. H ow ever, in an experim ent using d ifferent induction techniques (th e Velten technique an d an au tobiog rap hical recollection procedure), H aag a con cluded that similarity of m ood indu ction
procedures did not affect recall. N onetheless, m ood-state± dependent retention w as
not dem onstrated.
T he purpose o f this article is not to present a detailed review of the m ood-state±
dependent literature; ad equate accounts m ay be found elsew h ere (e.g. Blaney, 1986;
Peeters & d’ Ydew alle, 1987). H ow ever, it does seem that a num ber of important factors
an d seriou s ¯ aw s in experim ental design have contributed to the equivocal nature of the
® ndings concer nin g m ood-state dependenc y. In sum m ary, all of the studies previously
m entioned have one or m ore of the following problem s: designs that have co nfou nded
en coding an d retrieval by not m easuring (or reporting) initial lear ning scores; a lack of
objective m easures of m ood m an ipulation; the use of the sam e m etho d of m ood ind uction
at encoding an d retrieval, so co nfounding the effect of m ood an d the effect of the
induction procedure. T he evid ence available at present, therefore, does not allow de® nitive statem ents to be m ad e concer ning the generality an d reliability of m ood-state± dependen t retention. A lthough other factors such as those m entioned by E llis an d A shbrook
(1991) m ay clearly be im portant, exp erimental evidence that tak es account of the m ore
fundam ental problem s of d esign inhere nt in the p revious w ork need s to b e provided. T he
ex periments reported here were designed w ith th ese issues in m ind . Inv estigatio n of the
issues required various levels of control an d involved proceed ing throu gh a num ber of
stages of research. In all experim ents, objective m easures of m ood ind uction were provid ed; the ® rst two experiments (1a an d 1b) ad dressed the problem of con fou ndin g
en coding an d retrieval; th e third experim ent (1c) ad dressed the issue of co nfou nding
the effects of m ood an d o f the induction procedure; an d the fourth an d ® fth experiments
investigated som e of the interesting question s co ncer ning the nature of retrieval cues in
m ood-state± dependent retrieval.
294
KEN EA LY
Downloaded By: [University College London] At: 02:00 2 February 2011
E X P E R IM E N T S 1 A , 1 B , 1 C
K ihlstrom (1991) points out that the hypoth esis of m ood-state± dep en dent retrieval is
based on an an alog y w ith state-d ependent m em ory produced by phar m acological substances such as alco hol, barb iturates, am ph etam ines, an d m arijuan a. Som e of th e m ethodological problem s associated w ith previous state-d ependent experim ents have been
noted by O verto n (1974)Ð in particu lar, the issue of asym m etrical or o ne-w ay dissociatio n
of lear ning. In a review of the effects of alcoh ol on m em ory, Ryback (1971) noted th at such
state d ependency tends to be asym m etricalÐ that is, subjects w ho lear n in an intoxicated
state an d recall in a sober state forget m ore than sub jects w ho lear n in a sober state an d
recall in an intoxicated state. O verto n (1966) also observed this asym m etrical dissociatio n
of lear ning. Berger an d Stein (1969) proposed a neural m odel to explain the observatio n
that su bjects going from a drug to a non -drug state forget m ore than do subjects going
from a non -drug to a drug state (refer red to as ``o ne-w ay dissociation of lear ning’ ’ ).
W hether the explanatio n of one-w ay dissociation lies in the structure of the brain or in
the design of the test situatio n (or both) is open to question. It m ay b e that su bjects
``taught’ ’ under the in¯ uence of alcohol have lear ned less at the end of the ``training’ ’ than
the groups n ot taught under alcohol. O verton (1974), in a review of the experimental
m ethods used to study state-dep end ent lear ning, has poin ted out that w e m ust be sure
that an equivalen t am ount of learn ing occurs in all conditions before we claim that the
co ndition ``cau ses’ ’ am nesia an d not sim ply poorer lear ning initially. By an alogy, it is
en tirely possible th at d ifferent m ood states m ay similarly lead to poorer lear ning initially,
an d it m ay be that individual differences w ill also contribute to the variance in initial
lear ning. In any case, w ithout an ob jective m easure of initial lear ning, the effects of m ood
on m em ory at encoding w ill be confounded w ith the effects of m ood at retrieval. Reference
to the p roblem s encountered in d ru g-induced-state± dependent experim en ts (O verto n,
1974 ) suggests the n eed for an im proved design that takes into account the initial lear ning
of all sub jects.
M e th o d
T h e d esign , m aterials, an d pr oced u res for E xp erim ents 1a, 1b, 1c are p resen ted together.
D e sig n
In each exp erim en t a 2 3 4 expe rim en tal d esign w as em p loyed . In E xp erim en t 1a, o n D ay 1, h alf
the sub jects lear n ed th e m em ory task after th e elation versio n of th e Velten m ood indu ction p ro ced u re [G ro up s 1 (E E ) & 2 (E D )], an d h alf th e su bjects lear n ed th e task after the d ep ressio n ver sio n
[G ro u p s 3 (D D ) & 4 (D E )]. O n D ay 2, h alf the su bjects recalled after ind u ction of th e sam e m ood
w ith th e Velten p roc edu re [G ro up s 1 (E E ) & 3 (D D ))], and h alf th e su bjects recalled after ind u ctio n
of a d ifferen t m ood [G ro u p s 2 (E D ) & 4 (D E )]. In E xp erim ents 1b an d 1c, th e d esign w as iden tical,
except th at h ap p y an d sad m usic m ood ind u ction s w ere u sed [G ro up 1 (H H ); G ro u p 2 (H S ); G ro up 3
(S S); G rou p 4 (S H )].
It w as h yp oth esiz ed th at if sim ilarity of m ood state at lear n ing an d recall facilitates rem em berin g,
su bjects lear n ing an d recalling in an ind u ced elated /ha pp y m ood and su bjects lear n ing an d recalling
in an ind u ced dep resse d /sad m ood wo u ld h ave sign i® can tly lower (less) d ecrem en ts in recall th an
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
295
Downloaded By: [University College London] At: 02:00 2 February 2011
su bjects recalling in a d ifferen t ind uced m ood fro m th at in w h ich th ey lear n ed th e m aterial. Form ulating th e m ood -sta te± d ep en d ent retrieval h ypo th esis so as to requ ire a d irect test of th e differen ce in th e decrem en t in recall betw een th e sa m e-m ood an d th e ch an ged-m ood g ro u p s allows for th e
p ossib ility tha t th ere m ay b e sign i® can t sh ifts even for th ose g ro up s lear n ing an d recalling in th e
sam e m ood state becau se of th e u sua l d ecrem en ts th at occu r betw ee n sessio n s (see Wester m an n &
H ager, 1983).
In add itio n, p rov ided a con d ition of equ ivale nce of initial lear n ing w as m et by all su bjects (elated /
h ap py an d de pre sse d /sad ), it w as expected tha t th e state disso ciation d ecrem en ts w ou ld be sym m etrical in na tu reÐ th at is, th e nu m ber of item s forg otten w ould be abou t th e sam e, w h eth er su bjects
lear n ed in an ind u ce d elated m ood an d recalled in an indu ced d epre sse d m ood or lear n ed in an
ind u ced dep resse d m ood an d recalled in an ind uced elated m ood .
M o o d In d u c t i o n s
Velten M ood I nduction P rocedure. A s a nu m ber of stu d ies h ave u sed th e Velten m ood ind u ctio n
p roc edu re to investig ate th e effects of m ood on m em ory, a m od i® ed ver sio n of th e Velten p ro ced u re
w as u sed in th e initial E xp erim ent 1a. T h e m oo d ind u ction p ro ced u res were th e sa m e as th ose u sed
by Teasd ale an d Taylor (1981), a m od i® ed ver sio n of tho se u sed by Velten (1968). For each m ood
ind u ctio n, su bjects read 12 card s bearin g typ ed self-referen t statem en ts. For the elation ind uction ,
the statem en ts u sed w ere: I feel pretty good right now; I feel ha ppy ; I feel cheerful, con® dent; I ca n think
quickly a nd clea rly right now; R ight now, I feel very contented; R ight now, I feel like smiling; I feel a lert,
ha ppy a nd full of energy; I ha ve a feeling of lightness a nd joy; I rea lly like this lighthea rted feeling; I ca n
feel a smile on my fa ce; I feel so good I a lmost feel like la ughing, and I t feels grea t to be a liv e! For th e
d epre ssio n m ood ind uctio n , th e statem ents u sed w ere: I feel unha ppy ; I feel sa d a nd blue; I feel fed up;
I just feel dra ined of energy, worn out; I feel pretty low; Things seem futile, pointless; I feel hopeless; I feel
down-hea rted a nd misera ble; I feel so tired a nd gloomy tha t I would ra ther just sit tha n do a ny thing; I feel
hea vy a nd sluggish; I t seems such a n effort to do much, an d I ’ m fed up with it a ll.
M usic M ood I nduction P rocedure. In E xp erim en ts 1b an d 1c, a m u sic m ood ind u ctio n p ro ced u re
iden tical to th at rep or ted by K enealy (1988) w as use d . M u sic selectio ns w ere record ed o n tap e an d
con sis ted of a ha pp y selection fr om Coppelia (``M azu rka ’ ’ ) by D elibes, an d a sad selection , A da gio in
G minor, by A lb inon i. T h ese tw o selection s ha d be en sho w n differen tially to affect self-rep orted
m ood an d beh aviou r in th e d esired w ay by p reviou s research, an d th e m usic affected sub jects’ m ood s
in th e ab se nce of ins tr u ction s to try to cha ng e th eir m ood s (K en ealy, 1988, Par ro tt & S abini, 1990).
M ood M ea sures. A s a m eas ure of the effect of th e m ood indu ction p ro ce du res, su bje cts w ere
ask ed to rate th eir m ood on thre e 0± 100-line scales. E ach scale w as 10 cm lo ng w ith th e 0 en d labelled
I do not feel a t a ll X , an d the 100 en d labe lled I feel extremely X . O n th e th ree scales ``X ’ ’ w as H a ppy,
A nx ious, an d Despondent in E xp erim en t 1a, an d H a ppy, A nx ious, an d S a d in E xp erim en ts 1b an d 1c.
A T TH I S M O M E N T w as typ ed at th e top of each sh eet to ind icate th at it w as instan tan eou s m ood
that w as to be r ated . T h ese m eau res h ave been u sed in p reviou s stu d ie s to m easu re th e effect of
m ood -in d u ction p ro ced ure s (Teasd ale & Fog arty, 1979; Teasda le & Taylor, 1981; Teasd ale, Taylor, &
Fog ar ty, 19 80).
M ood A ccepta nce C riterion. Becau se the interest of th ese e xp erim en ts w as in ind u cing tw o
d ifferen t m ood states a criterion w as ch osen , before th e exp erim en ts beg an , to d e® ne an ad equ ate
d ifferen ce in m ood betw een th e tw o m ood con d ition s. T h is w as th at th ere sho u ld be at least a 20 p oint diffe ren ce betw een th e ha pp iness an d d esp on den c y r atings im m ed iately after m ood ind u ctio n
296
KEN EA LY
on each d ay. T h is criterion w as u sed by Teasd ale an d Fog ar ty (1979), Teasd ale, Taylor, an d Fogarty
(1980), an d Teasd ale an d Taylor (1981) to d e® n e an ad equa te d ifferen ce in m ood betw een tw o m ood
con d ition s. In add ition , for su bjects in the d ifferen t m oo d gro u p s (E D and D E ) the criterion w as th at
there sh ou ld be at least a 20-p oint d ifferen ce be tw een th e h ap p iness (or desp on d en cy) ratin g o n D ay
1 an d th e h ap p iness (or desp on d en cy) r ating on D ay 2.
Downloaded By: [University College London] At: 02:00 2 February 2011
M e m o ry T a sk
A m em ory task th at cou ld be u sed in a con tro lled labor atory situ atio n w as co n stru cted .
S ub jects w ere ask ed to m em oriz e a sim p le geogr ap h ical m ap, visu ally d isp layed, tog eth er w ith
an au ditorally p resen ted 22-item set of d irection s con cer n ing a p ar ticu lar ro u te. T h e s eries of
d irection s rep resented affectively n eu tr al m aterial th at w ou ld avoid con fou n d ing state-d ep end en t
retrieval w ith m ood co n gr uen cy explan atio n s of th e ® n d ings. S u bjects w ere p resen ted w ith a
sty lized ``m ap’ ’ th at sh ow ed a ro u te com pr ised of 11 se gm ents. T h e en ds of each segm en t
w ere asso ciate d w ith a land m ark (p u b, library ), w h ich w as ind icated by a sim p le line d r aw ing
on th e m ap an d w as also labelled . T he directio n of tu r n of th e ro u te w as also sh ow n at th e en d of each
segm en t. T h e ``m ap ’ ’ w as p resent th rou gh ou t a d escrip tion of th e ro u te, w hich w as read to th e
su bject (``P ro ceed to th e su per m ark et; tu r n left ag ain; con tinue to th e sp or ts cen tre; . . .’ ’ ). T h e
``m ap ’ ’ w as th en rem oved , an d the sub jects were as ked to p rov ide a free recall of th e 22 elem en ts of
the ro u te (landm ark s an d d irection s of tu r n ). T h e resu lting score w as th e nu m ber of elem en ts
recalled , irresp ective of ord er. T h e task w as design ed to p rov ide a n atu ra l an alogu e to th at of
rem em berin g an d giving d irection s for a rou te th ro ug h a city.
I nitia l Lea rning C riterion. A score of on e w as given for each item cor re ctly lear n ed . In ord er to
en sure equ ivalen ce of initial lear n ing, an arb itrary criterion of at least 18 item s cor rect w as ad op ted as
a lear n ing criterio n . E ach su bject w as allowed 4 lear n ing trials to reach th is criterion . If after 4 trials
the criterion w as n ot attained , th en th e largest nu m be r cor re ct in any on e trial wo u ld be taken as th e
su bject’ s p erfo r m anc e score.
P rocedures. O n D ay 1, th e m ood scales were sh ow n and exp lained to su bje cts. T he ove rall
p roc edu re w as th en sum m arized , an d su bjects w ere told tha t th e exp e rim en t w as co n cer ned w ith
m ood an d le ar n ing sty le s. For th e Velten m ood ind u ction p ro ced u re, th e im p ortan ce of tr ying to
becom e fu lly involved in th e m ood ind u ction pro ced u re w as em p h asized . S u bjects w ere told th ey
sh ou ld try as h ard as p os sible to feel th e m ood su ggested by th e s tatem en ts; it w as stated th at th is
m ight take so m e effor t an d th at it w as a cru cial p art of th e exp erim e nt. In E xp erim en t 1a, su bjects
w ere th en ask ed to read th e s tatem en ts silen tly, at th eir ow n spe ed , retur nin g to th e be ginn ing of th e
p ile if they reach ed th e en d. T hey w ere told to sto p read ing after 5 m in.
For th e m usic m ood indu ctio n p ro ced u re, su bjects w ere escorted into a sou n d -p ro of ro om an d
seated com fortably. T h e exp erim enter th en m ad e an excu se of h aving so m eth ing u rg en t to attend to
an d left the ro om , leaving either th e hap p y or th e sad m usic p laying qu ietly in the backg rou n d . T h e
experim en ter retu r n ed to th e ro om after app rox im ately 5± 8 m in.
A ll su bjects w ere th en ask ed to m em oriz e th e geog rap h ical m ap, visu ally d isp layed, toge th er w ith
the au ditorally pre sen ted 22-item set of directio n s con cer n ing a p ar ticu lar ro ute. T h e follow ing
sta nd ard set of instru ctio ns w as given to each sub ject:
``H ere is a picture of a ma p. I would like you to look a t the ma p while I rea d you a series of
instr uctions which will ena ble you to get from one pla ce to a nother on the ma p. I will then ta ke a wa y
the ma p, a nd a sk you to reca ll a s ma ny of the instructions a s you ca n remember.’ ’
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
297
In E xp erim ent 1a, su bjects rated th eir m oo d o n th e th ree scales at th e following p oints in th e
experim en t:
Downloaded By: [University College London] At: 02:00 2 February 2011
1. im m ediate ly after th e 5-m in m ood in ductio n;
2. afte r th e ® rst lear ning trial, in order to verify th at th e subject’ s m ood state had not
chan ged sign i® cantly.
In E xp erim en ts 1b an d 1c, su bjects r ated th eir m ood im m ed iately follow ing th e m usic m ood ind u ction . O n e m easu re of the effect of th e m oo d ind u ction p roced u re w as con sid ered su f® cien t, becau se
the pro ced u re h ad been sh ow n p reviou sly to affect m ood sign i® can tly (K en ealy, 1988).
Su bjects were th en th an ked for their h elp an d ask ed to retu r n at th e sam e tim e o n th e follow ing
d ay, on th e pre text of h aving to com p lete a few m ore lear nin g task s. O n D ay 2 (24 h our s later), after
the app ro p riate m ood ind uction , su bjects rated their m ood on th e thre e scales an d w ere then ask ed to
recall as m any of the instru ctio ns tha t accom p anied th e m ap as th ey cou ld re m em ber from the da y
before, w ithout again refer rin g to th e m ap. T h e n um ber of item s cor rectly n am ed by each su bject w as
taken as their recall score.
E X P E R IM E N T 1 A
M e th o d
T h irty-tw o ® r st- and secon d- year u n derg ra du ate stu d entsÐ 1 4 m ales an d 18 fem alesÐ w ere
recru ited . T he age of th e 24 sub jects m eeting the accep tan ce criterion for th e stud y ran ged from
19 to 39 years, w ith a m ean age of 24.8 year s.
R e s u lts
M ood M ea sures. E ight subjects failed to reach the m ood acceptance criterio n. Table 1
sh ow s the effect of the elatio n an d depression m ood inductions as rated by subjects on the
three 0± 100 m ood scales. T he results show n are the m ean rating s on the three scales from
the two occasions of m easurem ent following m ood induction on D ay 1 an d D ay 2.
Subjects in the sam e m ood groups (E E & D D ) showed n o signi® can t d ifferences in
ratin gs on the three m ood scales, co n® r m ing that they had lear ned (on D ay 1) an d
recalled (on D ay 2) the m em ory task in the sam e m ood state, w hich is crucial to the
ex periment. S ubjects receiving different m ood inductions on D ay 1 an d D ay 2 (E D &
D E ) showed signi® can t differences in ratings o n the ``hap piness’ ’ an d ``despondency’ ’
m ood scales, co n® r m ing that subjects recalling on D ay 2 w ere in a different m oo d state
from the one in w hich they lear ned the m em ory task o n D ay 1. T here were no signi® can t
differences b etween subjects’ ratings o n the an xiety scale on D ay 1 an d D ay 2 in an y of
the four g roups. T he p atter n of results w ith respect to m ood m an ipulation w as id entical
in E xp eriments 1a, 1b, an d 1c, as show n in Table 1.
M ood a nd M emory. M ood induction did not differentially affect lear ning perfor m an ce on D ay 1: there w as no signi® can t difference b etween the lear ning scores of
depressed (m ean = 18.6, S D = 1.16) an d elated (m ean = 19.4, S D = 1.31) subjects o n
D ay 1, t (22) = 1.65, p > .05. T hus, the condition of equivalence of initial learn ing by
su bjects in the different m ood g rou ps had been m et. M ood induction did not differen-
298
KEN EA LY
TA BLE 1
Downloaded By: [University College London] At: 02:00 2 February 2011
E f fe c ts o f H a p p y a n d S a d M o o d In d u c ti o n s
Experiment 1a
Experiment 1b
Experiment 1c
M ood I nduction
M ood I nduction
M ood I nduction
Mea sure*
Da y 1
Da y 2
Da y 1
Da y 2
Da y 1
Da y 2
Group 1
Hap piness
Anxiety
Sad ness
Elatio n
60 .5
21 .7
16 .2
E latio n
59.8 ns
21.3 ns
15.8 ns
Hap py
67.3
12.7
12.0
H ap py
66.6 ns
11.3 ns
11.9 ns
Happ y
63.3
18.5
14.9
H ap py
64.5 ns
16.6 ns
15.1 ns
Group 2
Hap piness
Anxiety
Sad ness
Elatio n
61 .0
15 .0
24 .7
Depression
29 .5 **
11.8 ns
55 .7 **
Hap py
64.7
10.4
16.9
Sad
16.0 **
14.0 ns
61.8 **
Happ y
63.5
12.3
19.4
S ad
19.4 **
14.1 ns
59.2 **
Group 3
Hap piness
Anxiety
Sad ness
D epression
29 .2
9.8
54 .2
Depression
22.7 ns
10.7 ns
57.3 ns
S ad
19.3
11.1
67.9
Sad
18.3 ns
9.7 ns
65.6 ns
Sad
24.2
9.4
62.9
S ad
21.0 ns
10.1 ns
62.7 ns
Group 4
Hap piness
Anxiety
Sad ness
D epression
28 .2
21 .3
52 .0
E latio n
72 .3 **
15.2 ns
19 .2 **
S ad
19.7
10.7
60.1
H ap py
59.9 **
10.0 ns
9.7 **
Sad
25.0
15.4
58.2
H ap py
64.6 **
14.1 ns
15.8 **
* M ean of subjects’ ratin gs o n 0± 100 scales.
** S ign i® can t diffe rence betwen m ood ratings on Day 1 an d Day 2, p < .001.
tially affect overall recall prefor m an ce o n D ay 2 either: there w as no signi® can t difference
between recall scores of depressed (m ean = 15 .3, S D = 3.86) an d elated (m ean = 15.5,
S D = 4.03) subjects o n D ay 2, t(22) = 0.10, p > 0.05.
H owever, an independen t t test an alysis show ed that the decrem ent in recall of the
sam e-state groups (E E + D D ), m ean = 0.42 item s, com pared to the decrem ent in recall of
the ch an ged-state groups (E D + D E ), m ean = 6.8 item s, w as signi® can tly different. T he
chan ged-state groups forgot signi® can tly m ore items than did the sam e-state groups, t(22) =
9.9 79, p < .001. Table 2 show s the lear ning an d recall scores from the four g roups of
su bjects, togeth er w ith the m ean decrem ents in recall for the sam e-state g roups an d the
ch an ged-state g roups.
Further m ore, the dissociatio n decrem ents were about the sam e, w heth er the tran sfer
w as from elation indu ction on D ay 1 to d epression indu ction on D ay 2 [G rou p 2 (E D ):
m ean decrem ent = 7.2, S D = 2.14], or from depressio n induction o n D ay 1 to elatio n
induction on D ay 2 [G rou p 4 (D E ): m ean d ecrem ent = 6.3, S D = 2.26]. T h ere w as no
signi® can t difference in the num b er of item s forgotten by sub jects in these two groups,
t(10) = 0.62, p > .05.
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
299
TA BLE 2
M e a n L e a r n in g a n d R e ca ll S co r e s
Experiment 1a
Da y 1
Group
Mea n
EE
Downloaded By: [University College London] At: 02:00 2 February 2011
M ea n
SD
1.64
18 .8
19.0
1.47
1.7 8
18.5
1.03
0.82
12.2
12.0
2.6 4
1.7 9
SD
19.5
DD
ED
DE
Da y 2
1.38
19.3
18.3
Experiment 1b
Da y 1
Group
HH
SS
HS
SH
E xperiment 1c
Da y 2
Da y 1
Da y 2
M ea n
SD
M ea n
SD
M ea n
SD
Mea n
SD
19.1
18.4
18.1
18.5
1.07
1.51
0.69
0.98
18.1
17.7
10.9
11.3
0.89
1.11
1.21
1.60
19.7
18.5
18.8
18.4
1.25
1.18
1.03
0.97
18.9
17.7
11.5
11.5
1.37
1.16
2.32
1.43
D is cu s s io n
T he results indicated a m ood-state± dependent retrieval effect. Subjects w ho learn ed an d
recalled in the sam e m ood forgot sign i® can tly fewer item s than did subjects w ho lear ned
an d recalled in a different m ood; chan ging su bjects’ m ood state at recall produced signi® can tly g reater decrem ents in recall. In ad dition, the num b er of item s ``fo rgotten’ ’ by
su bjects w as about the sam e, w hether tran sfer w as from elatio n induction on D ay 1 to
depression induction on D ay 2 or from depression induction o n D ay 1 to elation induction on D ay 2, indicating a sym m etrical m ood-state± d ependent retrieval effect.
W ith respect to th e m ood inductions, read ing self-referent statem ents su ccessfully
induced elation an d depression in subjects. M ood induction did not differentially affect
su bjects’ ability to lear n the m em ory task on D ay 1. D epressed subjects did n ot requ ire
m ore trials overall to ach ieve the lear nin g score criterio n, as m ight have been expected
from th e ® ndings of L eight an d E llis (1981); the m ean num ber of trials to reach the
criterion w as 3.9 for depression-indu ction subjects an d 3.8 for elation -induction su bjects;
how ever, o n Trial 1 elated subjects lear ned sig ni® can tly m ore (m ean = 12 .01) th an did
depressed subjects [(m ean = 6.50), t(22) = 2.92, p < .01]. M ood induction w as show n not
to have differen tially affected recall perfor m an ce on D ay 2 either, indicating th at the
observed state-dependent retrieval effect w as due to similarity or dissim ilarity of su bjects’
m ood states at th e time of learn ing an d recall and not to the m ood-induction procedures
affecting sub jects’ ability to perfor m during the learn ing an d recall testing sessio ns. T he
Downloaded By: [University College London] At: 02:00 2 February 2011
300
KEN EA LY
m em ory task proved relatively easy to score, interscorer reliability show ed 99.4% ag reem ent on item s, an d forgetting an d rem em bering of item s w as rand om ly distributed
throughout the task, ap art from a trend for all subjects to rem em ber the ® rst an d last
item s in the task.
It ap pears therefo re in this initial experiment that a design that ensured equivalence of
initial lear ning an d exam ined decrem en ts in recall b etween lear ning an d recall sessions
su ccessfully facilitated the d em onstratio n of m ood-state± dependent re trieval.
It is im portant to investigate how lear ning an d m em ory are affected by the relatively
co m m o n occur rence of tr an sient ¯ uctuatio ns in m oo d, an d the Velten m ood ind uction
procedure allow s for such investigatio ns. H ow ever, a num ber of au thors have pointed
out the disadv an tages of th is procedure. C lark (1983) has noted that a m ajor problem
w ith th e Velten procedure is that 30% to 50% of subjects fail to respo nd to it. In the
ex periment reported here, 25% (n = 8) of subjects failed to respo nd to the procedure
an d did not m eet th e criterion used to de® ne an ad eq uate difference between the m ood
states of elation an d depressio n. In ad dition, the Velten procedure places a stro ng
em phasis on sub jects tryin g hard to get into the desired m ood , an d it is possible that
som e subjects m ay have reported feeling elated or depressed after th e induction procedu res, w hen in fact they have not felt the m oods suggested by the statem ents. A
num ber of au thors have ad dressed the issu e of the co ntribution of dem an d characteristics to the Velten m ood in duction p roced ure (e.g. B uchw ald, Strack & C oyn e, 1981;
Polivy & D oyle, 1980). K enealy (1986) reviewed 46 published experim ents that had
used the Velten p roced ure an d concluded that the ® ndings relatin g to the p rocedure’ s
effectiveness w ere equivocal, as w ere assessm ents of the co ntribution of dem an d
ch aracteristics to the procedure.
Taking these factors into account, it therefo re seem ed adv isable, in order to co ntinue
this line of research, to use an alter native m ethod of m ood induction. T he use of m usic
to induce m ood states ap peared to b e a suitable alter native to w hich subjects m ight
respond m ore easily, an d it has been used in several lab oratory experim ents investigating
the effect of m ood on m em ory (Bower & M ayer, 1991; C lark, 1983; C lark & Teasdale,
1985 ; Parrott & Sabini, 1990). T hus, a m usical procedure w as used as a m ore subtle
m ethod of m ood in duction to investigate the gen erality an d reliab ility of m ood-state±
dependent retrieval.
E X P E R IM E N T 1 B
T his exp erim ent attem pted to rep licate the m ood-state± depen den t retrieval effect
observed in E xperim en t 1a. T he sam e design an d m em ory task were em ployed, but a
m usic m ood-induction procedure (K enealy, 1988) w as used to induce ``hap py’ ’ an d ``sad ’ ’
m oods. T he purpose o f the experiment w as to investigate the generality an d reliability of
the m ood-state± dependen t retrieval effect, using a d ifferent m ethod of m ood induction. It
w as felt th at rep lication of th e previous ® nding w as necessary in ord er to establish
w hether the phenom eno n of m ood-state± dependent retrieval could be dem onstrated
using different m ethods o f m ood in duction .
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
301
M e th o d
T h irty ® rst- an d secon d -year fem ale u n d erg r ad u ate stu d en ts w ere recru ited. T he age of the 28
su bjects m eeting th e accep tan ce criterion for th e stu d y ra ng ed fro m 19 to 26 years, w ith a m ean
age of 22.9 year s. A stereo casse tte d eck con n ected to tw o free-stan d ing sp eaker s, w as u sed to p lay
two selectio ns of m u sic for th e ``ha pp y’ ’ and ``sad ’ ’ m ood ind u ction s, as alread y d escribe d.
Downloaded By: [University College London] At: 02:00 2 February 2011
R e s u lts
M ood a nd M emory. O ne sub ject failed to reach the m ood acceptance criterion, an d
one subject failed to retur n o n D ay 2. M ood ind uction did not differentially affect
lear ning perfor m an ce on D ay 1 ; there w as no signi® can t differe nce between the lear ning
scores of sad (m ean = 18.6, S D = 1.22) an d hap py (m ean = 18.8, S D = 1.01) subjects o n
D ay 1, t(26) = 0.34, p > .05. M ood in du ction d id not differentially affect recall perfor m an ce o n D ay 2 eith er; th ere w as no signi® can t difference b etween recall scores of sad
(m ean = 14.3, S D = 3.73) an d hap py (m ean = 14.7, S D = 3.68) subjects on D ay 2, t(26) =
0.3 0, p > .05.
Independent t test an alysis showed that the decrem ent in recall of the sam e state groups
(H H + S S), m ean = 0.86 item s, com pared to the decrem ent in recall of the chan ged
state g roups (H S + SH ), m ean = 7.29 item s, w as signi® can tly different. T he chan gedstate g roups forgot sign i® can tly m ore item s than did the sam e-state g roups, t(26) =
12.19, p < .001. Table 2 show s the lear ning an d recall scores from the four g roups of
su bjects, togeth er w ith the m ean decrem ents in recall for the sam e-state g roups an d the
ch an ged-state g roups.
A s in the p revious experim ent, th e dissociation decrem ents were about the sam e
w hether the tran sfer w as from hap py induction on D ay 1 to sad induction o n D ay 2
(g rou p 2 : H S; m ean decrem ent = 7.3, S D = 1.60), or from sad induction o n D ay 1 to
hap py induction o n D ay 2 (g roup 4: S H ; m ean decrem ent = 7.6, S D = 1.80). T here w as
no signi® can t difference in the num b er of item s forgotten by su bjects in these two groups,
t(12) = 0.19, p > .05.
D is cu s s io n
T he sym m etrical-m ood-state± dependent retrieval effect ob served in the previous
ex periment w as replicated using a different m ethod of m ood induction. Sub jects lear ning
an d recalling on D ay 2 in d ifferent m ood states had signi® can tly greater decrem ents in
recall th an did subjects lear ning an d recalling in th e sam e m ood states. A ll subjects except
one w ere affected by the procedure. T he m usic procedure proved to be a m ore subtle
m ethod of m ood induction an d easier to ad m in ister than th e m odi® ed Velten (1 968)
technique.
T he laboratory situation provid es a very different context for the individual from the
co ntexts experienced in everyday life, w hich are rarely identical on any two occasio ns.
Indeed, it is possible that the effects observed here w ere d ue to the laborato ry pro cedures of m ood induction rather than to the m ood effects them selves. D em an d ch aracteristics are inherent in an y laboratory experim ent, an d although explicit dem an ds
Downloaded By: [University College London] At: 02:00 2 February 2011
302
KEN EA LY
can be controlled to an extent, there still rem ain im plicit dem and s in such situatio ns to
w hich subjects m ay respon d (for a detailed discussio n of dem an d characteristics see
K enealy, 1988). T he two selections of m usic used in the procedure d escribed here had
been validated previously (Kenealy, 1988) usin g a sim ilar procedure to that described by
Velten (1968), w ith the ad dition of two counter-d em an d g roups identical to those of
Polivy an d D oyle (1980). Polivy an d D oyle hypothesized th at if dem an d characteristics
are responsible for the m ood effects found, subjects in the counter-dem an d groups
would show a reversed em otion effect in the opposite direction. K enealy (1988) foun d
no reversed em otional effects w ith the procedure used in E xperim ents 1a an d 1b.
H owever, in these two experim ents th ere w as no w ay to separate the effect of m ood
sim ilarity at learn ing an d recall from that of in duction procedure sim ilarity (H aag a,
1989 ). A s a solution to the problem of w hether the m ood-state± dependent effects
observed in the p revious experim ents w ere attributable to the procedu res used or to
the m oods them selves, Experim ent 1c attem pted to dem onstrate m ood-state± dependent
retrieval using different m ethods of m ood induction at lear ning an d recall; results m ight
then be m ore convincingly inter preted in ter m s of the effects of m ood rather than of
procedure.
E X P E R IM E N T 1 C
T he issue of con foun din g the effects of m ood an d o f the induction procedure w as
add ressed by using the Velten m ood-indu ction procedure an d the m usic m ood-ind uction
procedure described previously.
M e th o d
T h e design , m em ory task , an d gen era l p ro cedu re were ide ntical to th ose u sed in E xp erim en ts 1a an d
1b. Su bjects received the Velten m oo d-in d u ction p ro cedu re d u ring lear n ing o n D ay 1 an d th e mu sic
m ood -in d u ction pr oc edu re d urin g recall o n D ay 2. Forty -tw o ® rst- an d secon d -year un d erg rad u ate
stu d en tsÐ 20 m ales an d 20 fem alesÐ w ere recr uited as su bjects. T he age of th e 40 su bjects w h o m et
the criterio n ran ged fro m 18 to 38 year s, w ith a m ean age of 22.6 year s.
R e s u lts
M ood a nd M emor y. Two subjects failed to reach the m ood acceptance criterio n.
T here w as a signi® can t difference between the lear ning scores of sad (m ean = 18.5,
S D = 1.05) an d hap py (m ean = 19.3, S D = 1.20) subjects on D ay 1, t(38) = 2.23, p < .05.
H owever, m oo d induction did not differentially affect recall perfor m anc e o n D ay 2Ð
there w as no signi® can t difference between recall scores of sad (m ean = 14.6, S D = 3.65)
an d hap py (m ean = 15.2, S D = 4.03) sub jects on D ay 2, t(38) = 0.49, p > .05.
T he decrem en t in recall of the sam e-state g roups (H H + S S), m ean = 0.80 item s, w as
signi® can tly less than the decrem ent of the chan ged-state groups (H S + SH ), m ean =
7.1 0 item s. T he chan ged -state g roups forgot signi® can tly m ore item s than d id the sam e-
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
303
Downloaded By: [University College London] At: 02:00 2 February 2011
state g roups, t(38) = 14.18, p < .001. Table 2 show s the lear ning an d recall scores from
the four g roup s of subjects, together w ith the m ean decrem ents in recall for th e sam estate g rou ps an d the chan ged-state g rou ps.
A s in the p revious exp eriments, the dissociation d ecrem en ts were about the sam e
w hether the tran sfer w as from hap py induction on D ay 1 to sad induction o n D ay 2
(g rou p 2 : H S; m ean decrem ent = 7.8, S D = 1.89), or from sad induction o n D ay 1 to
hap py induction o n D ay 2 (g roup 4: S H ; m ean decrem ent = 6.9, S D = 1.66). T here w as
no signi® can t difference in the num b er of item s forgotten by su bjects in these two groups,
t(18) = 0.50, p > .05.
D is cu s s io n
T he m ood-state± d ependent hypothesis w as supported, w ith subjects in the chan gedm ood g roups forgetting signi® can tly m ore item s th an subjects in the sam e-m ood g rou ps.
T his effect w as dem onstrated w hen different m ood-in duction procedures w ere u sed at
lear ning an d recall; it therefore seem s m ore likely that the effects on m em ory are attributable to m oo d rather than to th e procedures used to induce m ood.
In the three experiments rep orted h ere, m ood-state± dependent retrieval has ap peared
to b e a reliable phenom eno n. In view of the previous literature in th is area, perhap s the
m ost interesting observation is n ot that the effect has been dem o nstrated, but w hy it has
been dem onstrate d so reliably. T he previous inconsistencies in the literature h ave been
discussed in the introduction, an d a nu m ber of problem s that m ay, at the very least, have
co ntrib uted to the inconsistent nature of ® ndings in this area w ere identi® ed. T he experim ents reported here w ere designed to ad dress these problem s. Speci® cally, objective
assessm ent of m oo d w ith an a priori criterion speci® ed to de® ne differences b etween
m ood conditions con® r m ed that subjects were in the ap pro priate m ood states. A ssessm ent
of m ood has been acknow ledged as necessary (E llis & A shb rook, 1 991) in order to
elim in ate the possibility that a failure to dem o nstrate m ood-state± dependen cy is due to
an insuf® cient difference between th e m an ipu lated m ood s, or variance in th e m oo d
elicited under a particular co nd ition . T he m ood-state± dependent effects observed in
the ® rst two experiments are open to the criticism that the use of the sam e m ethod of
m ood induction at encoding an d retrieval confounded the effect of m ood an d ind uction
procedure. E xperim ent 1c ad dressed this problem . It m ay be that the m em ory task w as a
critical variable in facilitating the state-dependent effect. E llis an d A shbrook (1988; 1 991)
argue that episodic m em ory tasks w ith less m ean ingfu l an d less structured m aterial are
m ore susceptible to m ood effects, although th ey believe that the results argue ag ainst an y
sim ple patter n of m em ory decrem ents across the b oard, reg ardless of the type of m em ory
task em ployed (E llis & A shbrook, 1991, p. 12).
It is clear throug h all this d iscussion that there is a critical m ethodological issue th at
previou s research ap pears to have failed to add ressÐ the issue of exp erimental designs
co nfounding encoding an d retrieval by not m easuring (or reporting) initial lear ning
scores. In th e experim en ts reported here, th e design ensured equivalence of initial lear ning by all g roups of su bjects, an d th e m ood-state± dependent retrieval hypothesis w as
for m ulated so as to requ ire a direct test of the difference in the decrem ent in recall
Downloaded By: [University College London] At: 02:00 2 February 2011
304
KEN EA LY
between th e sam e-m ood an d the chan ged-m ood g roups. T his allowed for the possibility
that th ere m ay b e signi® can t decrem ents for sam e-m ood groups b ecau se of th e usual
decrem ents that occur between sessions (Wester m ann & H ager, 1983). Interestingly, in
E xperiment 1c this w as indeed the case. Sub jects w ho received th e Velten hap py m ood
induction lear ned signi® can tly m ore item s than did sub jects in the Velten sad m oo d
induction. H ow ev er, the m ood-state± depen dent hypothesis w as supported becau se these
su bjects did not forget signi® can tly m ore item s than d id the sad induction g rou p. A m ore
stringent test of the hypoth esis w ould have been to dem an d th at the chan ged -m oo d
g roups should show a signi® can t difference between lear ning an d recall scores; an d
that the sam e-m ood g rou ps should show no signi® can t difference b etween lear ning an d
recall scores. T he ``strict’ ’ m ood-state± d ep endent hypothesis could not have been supported in E xperim ent 1c; further an alyse s revealed that there w ere signi® can t d ifferences
between the lear ning and recall score for subjects in the sam e-m ood g roups, G roup H H :
t(9) = 2 .45, p < .05; G roup SS : t(9) = 2.75, p < .05. We m ust co nclu de, therefore, that the
w ay in w hich th e m ood-state± d ependent hypothesis is stated is critically im portant in
deter m ining w hether or not the effect has been dem o nstr ated. Perhap s the m ost critical
factor co ntributing to the unreliable ® ndings of m ood-state depend enc y is the operational
de® nition of the m ood-state± d ependent retrieval effect.
T he experim ental ® ndings reported here indicate th at the accessibility of retrieval cues
an d therefore the dem o nstratio n of state-dependent retrieval depen ds, in part, on the
ex perien tial context of the in dividual o n different recall occasio ns; on the deg ree of
co rrespondence between the environm ental (lab or atory) contexts for the individ ual o n
different occasions; a nd o n the interaction between the experiential an d enviro nm ental
situations o n each occasion, both of w hich m ak e up the total context. D em o nstratio ns of
the generality an d reliability of m ood-state± dependent retrieval p rovid e a basis of em pirical supp or t from w hich investigatio ns concer ning the nature of retrieval cues in m oodstate± dependent retriev al m ay proceed.
E ich, Weing artner, Stillman , an d G illin (1975) have stated that one of the keys to
understanding hu m an state-dependen t lear ning is the elu cidation o f the nature an d cau se
of retrieval failure. T hey note that to the extent that successful retrieval depends on the
restoratio n of the o rigin al encoding state, decrem ents in recall perform an ce associated
w ith differen t state condition s (e.g. lear n ``hap py’ ’ Ð recall ``sad’ ’ ) m ust re¯ ect a failure o n
the part of the retrieval m echan ism to ® nd relevan t infor m ation that is otherw ise accessible for recall under sam e-state conditions (e.g. lear n ``hap py’ ’ Ð recall ``hap py’ ’ ). A s the
deleterious effect of a state chan ge on recall perfor m an ce re¯ ects a problem in retrieval
rather than in the storage of infor m ation, an d as all acts of recall can be considered as
being m ediated by retrieval cu es (Tulving & M ad ig an , 1970), E ich et al. (1975) argue that
the crux of the problem would seem to lie in the process thro ugh w hich the cues essential
for recall of available in for m ation are subjectively generated. If this is the case, then the
provision of pow erful retrieval cues at th e time of recall should effectively ne gate an y
difference in recall brought about by state m an ipulatio ns.
In their study of the effects of m arijuan a on m em ory E ich et al. (1975) discovered that
state-dependent retrieval w as restricted to the n om inally non-cued test of free recall an d
w as not ap parent w h en recall w as prom pted w ith ap prop riate extralist retrieval cues
(category labels of w ord lists). T hey suggested that a chan ge in the experiential contexts
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
305
of su bjects m ay interfere w ith th e accessibility of retrieval cues that presum ably m ediate
free recall but no t cu ed recall. E ich (1980), after reviewing studies of drug-state± dependen t retrieval, concluded that of the variables considered to play key roles in the occurrence of the state-dependent retrieval phenom eno n, the presence or absence of discretely
identi® able retrieval cues in the cognitive enviro nm ent of the rem em berer w as revealed to
m atter m ost. E ich (1980) dem o nstrated th at, w ith very few exceptions, failures to dem onstrate d rug-state± d ependent retrieval are co n® ned to conditions in w hich retrieval is
tested in the presence of explicit rem inders, or ``ob servable’ ’ cuesÐ for exam ple, category
nam e cues (``list’’ cu es), or literal copies of the item s or events to be rem em bered (``copy’ ’
cu es)Ð an d successful dem o nstrations are con® ned to co nditions in w hich testing takes
place in the absence of ob servable cues bu t in the presence of ``invisible’ ’ cues such as
sim ilar experiential co ntexts.
T he experim ents previou sly reported here dem o nstrated m ood-state± dependent
retrieval u nder conditions of free recallÐ th at is, in th e absence of an y ``observable’ ’
cu es related to the m em ory task. K ihlstrom (1991) sug gests that w hen rich, in for m ative
cu es are provided by the retrieval enviro nm ent or can easily be generated by the
individual, the effect of state cues w ill be relatively weak. T h e next experim ent, therefore, investigated the assum ption th at sub jects lear ning an d recalling in the sam e m ood
states, as in the previous experim ents, utilized ``invisible’ ’ cues during recall, an d that a
ch an ge of m ood state interfered w ith the accessib ility of these retrieval cues, lead ing to
signi® can tly poorer recall by sub jects w ho lear ned an d recalled in different m ood states.
If this assum ption is correct, then the provision of ``ob servable’ ’ retrieval cues at the
time of recall should effectively neg ate an y difference in recall resulting from a chan ge
of m ood state. W ithin E ich’ s ter m inology, this represents a failure to dem onstrate statedependent retrieval, as it is a cue-dependent phenom enon; an d the question here w as
w hether m ood-state± d ependent retrieval is cue-depen dent in the sam e m an ner as drugstate± dependent retrieval ap pears to be (E ich , 1980). E ich an d B irn bau m (1987) have
also sug gested that the state-dep end ent m em ory effects produced by psychoactive drugs
are m ediated by their affective properties. T he results of a study by Bartlett an d
San trock (1979), exam in ing m ood-state± depen dent retrieval in young children under
co nditions o f free recall an d cued recall, m irrored the ® ndings of E ich et al. (1975) an d
are co nsistent w ith such reasoning. It w as sug gested by these au thors th at a ch an ge in
m ood red uced the children’ s ab ility to generate ap p ropriate retrieval cues in a freerecall test, an d that d rugs are not necessary to create this effect. T he purpose of
E xperiment 2 w as to investigate the effects of induced m ood on m em ory under conditions of cued recall.
E X P E R IM E N T 2
T he m usic m ood-induction procedure w as used to in duce hap py an d sad m ood states in
su bjects. T he design, m em ory task, an d general procedure were identical to that of
E xperiment 1, except that subjects w ere presented w ith an ob servable ``copy’ ’ cu e at
the tim e of recall on D ay 2. It w as predicted that providing sub jects w ith a discretely
Downloaded By: [University College London] At: 02:00 2 February 2011
306
KEN EA LY
identi® able retrieval cue at time of recall would ``over ride’ ’ the deleterious effects of a
ch an ge of m ood state at recall. W ith respect to the issue of w hat constitutes an effective
cu e for the recovery of inform ation rendered inaccessible for recall as a conseq uen ce of a
ch an ge of m ood state, the encoding speci® city hypothesis (Thom so n & Tulving, 1 970)
holds that only those cues present at encoding w ill b e effective retrieval cues. It w as
therefore decided to use the b asic visual outline of the m ap p resented to subjects in the
initial m em ory task, w itho ut the nam es an d visual representation s of item s, as an observable ``copy’ ’ cu e at the time of recall on D ay 2. It w as hypothesiz ed that providing su bjects
w ith a visual ``cu e’ ’ at tim e of recall w ould facilitate rem em berin g reg ardless of m ood
state; subjects learnin g an d recalling m aterial in different m ood states w ould not have
signi® can tly g reater decrem ents in recall than would sub jects lear ning an d recalling in the
sam e m ood states.
M e th o d
S u b j e c ts
Twen ty-® ve ® rs t- an d secon d- year fem ale u nd erg ra du ate s tu d ents w ere r ecr uited as su bjects.
T h e age of the 24 su bjects w h o p articipated in th e exp erim en t ran ged fr om 20 to 25 years, w ith a
m ean age of 21.5 years.
Pro ce d u re
W h en th ey re tu r n ed on D ay 2, after th e ap p rop r iate m ood ind u ction , sub jects w ere sh ow n th e
basic visu al ou tline of the m ap th at th ey h ad lear n t the d ay be fore, w ith n am es an d visu al rep resen tation s of item s exclud ed. T h ey wer e ask ed to recall as m an y of th e instr u ction s th at accom p an ied th e
m ap as the y could rem e m ber fro m th e d ay before. T h e m ap cu e w as p resen t for 1 m in an d w as th en
rem oved be fore sub jects’ cu ed r ecall. T he nu m ber of item s cor rectly n am ed by each su bject w as
taken as their recall score.
R e s u lts
M ood M ea sures. O ne su bject failed to reach the m ood acceptance criterio n an d did
not retur n o n D ay 2. Table 3 show s the effect of th e hap py an d sad m ood ind uctions as
rated by su bjects on the th ree 0± 100 m ood scales. T he patter n of results w ith respect to
m ood m an ipu lation w as identical to that found in E xperiment 1.
M ood a nd M emory. M ood induction did not differentially affect lear ning perfor m an ce o n D ay 1Ð there w as no sign i® can t difference betwen the lear ning scores of sad
(m ean = 18.7, S D = 0.78) an d hap py (m ean = 18.5, S D = 1.15) subjects on D ay 1, t(22) =
0.2 1, p > .05. H ow ever, m ood induction d id differen tially affect recall perfor m an ce o n
D ay 2Ð there w as a signi® can t d ifference between recall scores of sad (m ean = 14.8, S D =
2.8 3) and h ap py (m ean = 17.8, S D= 1.14) su bjects o n D ay 2, t(22) = 3.40, p < .01.
Independent t test an alysis sh ow ed that the decrem ents in recall of the sam e-state
g roups (H H + SS), m ean = 1.67 item s, com pared to the decrem ents in recall of the
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
307
ch an ged-state g roups (H S + SH ), m ean = 3.08 item s, w ere not signi® cantly different. A s
predicted , cueing overrode the state-dependent effect on recall, t(22) = 1.378, p > .05.
H owever, the decrem ent in recall of subjects w ho lear ned in a hap py m ood an d recalled in
a sad m ood (H S), m ean = 5.67 item s, w as signi® can tly g reater than that of the other three
g roups [com parison w ith sam e-state group (SS ), m ean = 2.5 item s, t(10) = 2 .65, p < .05].
Table 4 show s th e lear ning an d recall scores from the four g roup s of sub jects, tog ether
w ith the m ean decrem ents in recall for the sam e-state g roups an d the chan ged-state
g roups.
T here w as a signi® can t difference in the dissociation decrem ents of subjects in the two
g roups w ho recalled in a m ood state different from the o ne in w h ich th ey had lear ned.
Subjects w ho lear ned in a hap py m ood an d recalled in a sad m ood (g rou p 2: H S; m ean
decrem ent = 5.7 , S D = 1.63) forgot signi® can tly m ore item s than d id sub jects w h o
lear ned in a sad m ood an d recalled in a hap py m ood (g roup 4 : SH ; m ean decrem ent =
0.8 , S D = 0.41, t(10) = 7.35, p < .001).
TA BLE 3
E f fe c ts o f H a p p y a n d S a d M o o d In d u c ti o n s
Experiment 2
Experiment 3
Mood I nduction
M ood I nduction
M ea sure*
Da y 1
Da y 2
Da y 1
Da y 2
Group 1
Hap piness
Anxiety
S ad ness
Happ y
66 .0
12 .6
15 .3
H ap py
67.3 ns
13.3 ns
16.7 ns
Hap py
62.6
15.6
19.2
Happ y
61.6 ns
13.2 ns
21.4 ns
Group 2
Hap piness
Anxiety
S ad ness
Happ y
64 .2
11 .5
14 .8
S ad
17.5 **
9.2 ns
65.2 **
Hap py
68.6
10.4
17.6
Sad
20.0 **
11.6 ns
64.1 **
Group 3
Hap piness
Anxiety
S ad ness
Sad
21 .8
10 .3
66 .5
S ad
19.5 ns
12.5 ns
69.2 ns
Sad
18.8
14.0
68.8
Sad
17.6 ns
13.4 ns
63.9 ns
Group 4
Hap piness
Anxiety
S ad ness
Sad
18 .2
11 .3
65 .8
H ap py
58.3 **
11.8 ns
14.2 **
Sad
17.2
10.8
66.6
Happ y
66.8 **
11.4 ns
18.6 **
* M ean of subjects’ ratin gs o n 0± 100 scales.
** S ign i® can t differen ce betwee n m ood ratin gs o n Day 1 and D ay 2,
p < .001.
308
KEN EA LY
Downloaded By: [University College London] At: 02:00 2 February 2011
D is cu s s io n
M ood-state± dependent retrieval w as not dem on strated under all the con ditions of cued
recall. C han ging m ood state at recall produced an asym m etrical dissociation of m aterial
from m em ory. Sub jects lear nin g in a hap py m ood an d recalling in a sad m oo d had
signi® can t decrem en ts in recall, even though they w ere provided w ith a visual cue at
time of recall. H ow ever, presentation of a visual cue at tim e of recall provided su bjects
lear ning in a sad m ood an d recalling in a hap py m ood w ith an effective cu e that enabled
them to retrieve m aterial from m em ory desp ite a chan ge of m ood state at recall; these
su bjects did not have signi® can t decrem ents in recall.
T he observed asym m etrical dissociatio n can not be explained by a difference in the
initial learn ing scores of sub jects in the different g roups, as m ood o n D ay 1 did not affect
lear ning scores; there w as no signi® can t difference in th e lear ning scores of hap py an d sad
su bjects.
M ood-state± depen den t re trieval, then, ap peared to be cue-depen dent under certain
co nditions of m ood state. Providin g a visual cue at the tim e of recall enabled hap py-m ood
su bjects to rem em ber m aterial that they lear ned w hile in a sad m ood but ap peared not to
en able sad-m ood subjects to rem em ber m aterial lear nt w hilst they w ere in a hap py m ood.
It m ay be that subjects in a sad m ood are particu larly suspicious of experimental ``aids’ ’ ,
or lack the m otivation to utilize such cues effectively. It has been suggested that depression m ight not fundam entally im pair the resources required for good perfor m an ce o n
unintentional learn ing tasks (H ertel & Rud e, 1991), b ut that depression m ay affect the
initiation of cognitive processes, such as spontaneous use of strate gies an d the m otivatio n
to u tilize extern al cues effectively (H ertel & H ardin, 1990).
A n alter native explana tion for these ® ndings is that the ``cue’ ’ m ay have been so
obvious that subjects w ere able to w ork out from the visual representatio n of the m ap
those item s that contained directional com ponents (``tur n right’ ’ ; ``turn left’ ’ , etc.). In
order to deter m in e w hether this m ight account for the high recall scores of som e su bjects
on D ay 2, the seven item s containing only directional infor m ation were rem oved from the
data, w hich were then re-an alysed. E ssentially the sam e patter n concerning d ifferences
em erged. T he visual outline of the m ap did not therefore constitute a recall test w here
su bjects w ere m erely presented w ith the ``cor rect’ ’ item s, as the data rem ain essentially
the sam e w hen th ese item s are discoun ted; the ``cue’ ’ also ap peared to have facilitated
recall of the non-directional item s.
It seem s intuitively likely th at sad -m ood subjects, being in a neg ative m ood state,
m ight m ore read ily w ish to utilize an ap propriate retrieval cue in order to recall m aterial
from a hap pier, m ore positive time, becau se an effective cue m ay not o nly b ring to m in d
the m aterial encoded w ith the cue but also th e m em ory of the particular m ood state in
w hich it w as encod ed. H owever, the ® ndings observed in this experim ent are not consisten t w ith this sup position. In the event, b efore com ing to an y ® r m conclusio ns reg arding the issue, it w as d ecided to replicate this experiment in part in order to investigate the
reliability of the ® ndings.
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
309
TA BLE 4
M e a n L e a r n in g a n d R e ca ll S co r e s
Experiment 2
Downloaded By: [University College London] At: 02:00 2 February 2011
Group
HH
SS
HS
SH
Experiment 3
Da y 1
Da y 2
Da y 1
Lea rning
Cued
R eca ll
Lea rning
Da y 2
Free
Reca ll ( 1)
Cued
R eca ll
Free
R eca ll ( 2)
Mea n
SD
M ea n
SD
M ea n
SD
M ea n
SD
M ea n
SD
M ea n
SD
18.0
18.5
19.2
18.8
0.63
0.55
1.33
0.98
17.2
16.0
13.5
18.1
0.7 5
2.6 8
2.5 9
1.2 1
18.6
18.2
18.8
18.4
0.89
1.09
0.83
1.52
18.4
17.4
14.2
13.2
1.51
1.14
2.16
1.92
18.0
16.8
18.2
17.0
1.58
1.64
0.84
1.22
17.4
16.6
14.0
16.8
1.14
0.89
1.82
0.84
E X P E R IM E N T 3
In the p revious experiment the conditions of retrieval w ith resp ect to the p resence or
absence of exp licit (``observable’ ’ ) cues were held co nstantÐ all sub jects retrieved m aterial
in a cu ed recall condition w ith the provision of an ob servable ``copy’ ’ cue. In ad dition to
investigating th e reliability of the previous experimental ® ndings, the present experim ent
co ncer ns the effects of varying the retrieval cond ition s in a w ithin-subject design, w hich
would allow for an investigatio n of the nature of retrieval cues an d their transfer effects.
L eight an d E llis (1981), usin g th e Velten m ood-ind uction procedure, have previously
investigated th e role of depressed m o od induction on the tran sfe r of coding strategies.
T hey found that th e m ood state induced on D ay 1 recall deter m ined subseq uent recall of
a new list 24 hours later, indicating the tran sfer of coding strategies. T hey su ggest that
train ing under a depressed m ood resulted in the ad option of inef® cient codin g strate gies,
w hich w ere tran sferred to seco nd-list lear ning regard less of m ood state o n the secon d
day; an d that neutral m oo d in training enabled sub jects to ad opt a m ore ef® cient org an izational strategy, w hich persisted into the tran sfer task even w hen tran sfer followed a
depressed m ood induction . T he interest in the previous experiment reported h ere w as in
retrieval processes rather than encod ing strate gies (as an initial lear ning criterion w as
imposed in these experiments), but it seem s likely that if subjects are presented w ith an
effective retrieval cue at th e tim e of recall (cued-recall) that is cap able of neg ating the
effects of a chan ge of m o od state, utilization of such a cu e m ay tran sfer to later recall
co nditions w here a cue is not explicitly presen ted (free-recall) an d lead to effective
retrieval of m aterial reg ard less of m ood state.
E ich (1980) has noted that by m an ipulating the presence or absence of exp licit list cues
(e.g. cate go ry nam es) in the cognitive env ironm en t of the rem em berer, it is possible to
dem o nstrate both the occurren ce an d the erasure of state dependence w ithin one an d the
sam e experiencing individual. E ich et al. (1975), studying the effects of m arijuan a states,
have dem o nstrated state-d ependent retrieval und er conditions of free recall an d erasu re of
the state-dependent effect under conditions of cued recall, w ith the p resence of observable cues. Similar interactions have been obser ved by S w an so n an d K insbour ne (1976),
Downloaded By: [University College London] At: 02:00 2 February 2011
310
KEN EA LY
w hose experim ent focused on the state-dep end ent effects of m ethylphenidate (Ritalin) in
hyperactive children, an d by Weing artner (1978) in a study o f eserine-p roduced statedependent retrieval in ad ults. H avin g dem onstrated m ood-state± dep end ent retrieval in a
free-recall situation followed by erasure of th e effect in a cued-recall situatio n, the question
of interest, then, is w hether m aterial rem ains m ood-state± dependent in a later free-recall
situation, or w heth er the previous experience of an effective cue w ill enable the in dividual
effectively to retrieve infor m atio n again at a later tim e, reg ardless of m ood state.
W ithin everyday lear ning an d rem em b ering ter m s, this m ay be seen as an alogous to a
situation w here an ind ividual in a sad or depressed m oo d w ho co nsistently rem em bers
events an d ep isodes that occurred previously in a similar m ood state is presented w ith a
cu e (e.g a photog rap h, or som e other m em orabilia) w hich effectively enables them to
recall incid ents that occur red in hap pier m o od states. A lthough it is acknow ledged th at
the experiments here refer to a highly speci® c set of item s, this an alogy suggests that it
m ay then be possible for such an individu al to self-gen erate ap propriate retrieval cues,
sh ou ld they once m ore ® nd them selves in a sad or d epressed m ood, w hich w ill enable
them effectively to recall incidents from a hap pier tim e.
T he purpose of E xperim ent 3, then, w as to investigate the cue-dependent n ature of
m ood-state± dependent retrieval using a w ithin-subjects design. A ll m ood-induced subjects w ere asked to recall under three conditions: free recall, followed by cued recall (w ith
presentation of a visual cue), followed by free recall again.
T he general hypothesis investigated w as w hether m aterial that ap pears inaccessible to
su bjects w h o free-call in a d ifferent m ood state to that in w hich they lear n the m aterial
(state-dependent retrieval) b ecom es accessible under conditions of cued recall, reg ardless
of a chan ge o f m ood state at the tim e of recall (erasure of state-dep endent retrieval) an d
rem ains accessible in a later free-recall situ atio n, through the tran sfer of ef® cient retrieval
strategies.
M e th o d
S u b j e c ts
Twen ty-® ve ® r st- an d secon d -y ear fem ale u nd erg rad u ate stu d en ts w ere recru ited . T h e age of th e
20 su bjects w h o p articipated in th e exp erim ent r an ged fro m 20 to 36 years, w ith a m ean age of 23.3
year s.
Pro ce d u re
T h e sam e m usic m ood -in d uction pro ced ur e w as u sed to ind uce h ap py an d sad m ood state s in
su bjects. T h e d esign , m em ory task , an d gen eral p ro ced u re were iden tical to th e p reviou s exp erim en ts, w ith th e exc eption of th e recall sessio n on D ay 2.
O n D ay 2, after th e ap p ro p riate m ood ind uction , sub jects w ere ask ed to recall as m any of th e
instru ction s th at accom p an ied th e m ap as th ey cou ld rem em ber fro m th e d ay befor e, w ithout ag ain
refer rin g to th e m ap. Im m ediately after th is free-re call sessio n the y w ere sh ow n th e visua l ou tline of
the m ap (n am es an d visu al rep resen tatio ns of item s w ere exclud ed) an d were ask ed ag ain to recall as
m an y of th e instru ction s th at accom p an ied th e m ap as th ey cou ld rem em ber fr om th e d ay befor e
(cu ed recall). T h e d ur atio n of m ap cueing w as 1 m inÐ th e sam e as in E xp erim en t 2. A fter an interval
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
311
of ap prox im ately 10 m in, d u ring w hich tim e th e ap p ro p riate m u sic w as p laying qu ietly in the backgr ou n d , su bjects w ere again ask ed to free-re call as m an y of th e instru ctio ns as they cou ld rem em ber
w ithou t ag ain refer rin g to the m ap. T h e nu m ber of ite m s cor rectly n am ed by each su bject in each
con d ition w as taken as th eir recall score for th at p articu lar co n dition .
R e s u lts
Downloaded By: [University College London] At: 02:00 2 February 2011
M ood M ea sures. Two sub jects did not m eet the m ood accep tan ce criterion, an d three
su bjects failed to return on D ay 2. Table 3 show s the effect of the hap py an d sad m ood
inductions. T he patter n of results w ith respect to m ood m an ipulation w as identical to that
found in E xperiments 1 an d 2.
M ood a nd M emory. Table 4 show s the m ean lear ning an d recall scores for sub jects in
the three conditions of recall. M ood induction did not differen tially affect lear nin g
perfor m an ce on D ay 1Ð
th ere w as no sign i® can t difference between the lear ning scores
of hap py (m ean = 18.7, S D = 0.82) an d sad (m ean = 18.3, S D = 1.25) subjects on D ay 1,
t(18) = 0.84, p > .05.
A percentage score w as derived for each subject. T he num ber of item s cor rectly
recalled by each su bject on D ay 2 w as expressed as a p ercentag e of the num ber of item s
lear nt by that subject on D ay 1. T he percentages in Figure 1 refer to the m ean percentage
(unw eighted ) scores of subjects in each g ro up. Sub jects w ho learn ed an d recalled in the
sam e m ood states had essentially similar recall scores in all three conditions, an d the recall
scores on D ay 2 ran ged from 15 item s to 20 item s. T he m ean percentage of item s lear nt
on D ay 1 that w ere cor rectly recalled by subjects w ho lear ned in a sad m ood an d recalled
in a hap py m ood on D ay 2 w as 71.6% o f item s in the ® rst free-recall session; this w as
improved to a m ean of 92.2% in the cued-recall sessio n an d m aintained at a m ean of
91.4% cor rect in the second free-recall sessio n. H ow ever, the m ean percentage of item s
lear nt that w ere correctly recalled by sub jects w ho lear ned in a hap py m ood an d recalled
in a sad m ood on D ay 2 w as 75.2% of item s in the ® rst free-recall sessio n, w hich
improved to a m ean of 96.8% correct in the cued-recall sessio n, bu t w hich w as not
m aintained in th e second free-recall session, w here the m ean percentage recall w as
only 74.2% of the item s lear nt on D ay 1.
A dopting the previous hypothesis-driven an alysis for Ex periment 3 w as in suf® cient
becau se of the possibility of car ry-over effects from the ® rst recall con dition to the cued
recall an d from the cued recall to the second free-recall co ndition. T his rationale led to a
ch an ge o f ap proach an d statistical an alysis of data from this experiment. T he signi® can ce of
the patter n of results show n in F igure 1 w as tested in a three-way an alysis of variance, w ith
repeated m easures o n one factor; w ith m ood o n D ay 1 (Factor A ), m ood on D ay 2 (Factor
B), an d condition of recall (Factor C ). T he repeated-m easures Factor C had three conditions (levels) of re call: free recall (1) vs. cued recall vs. free recall (2). T he dependent
variables were the num bers of item s cor rectly recalled in each of the three conditions o n
D ay 2 as a percentage of th e num b er of item s lear nt by each subject o n D ay 1 (recall scores).
T here w ere no signi® can t m ain effects of m ood o n D ay 1 or on D ay 2. H ow ever, there
w as a signi® can t effect of recall condition o n D ay 2, F(2, 32) = 20.22, p < .001, a
Downloaded By: [University College London] At: 02:00 2 February 2011
312
KEN EA LY
F IG . 1 .
M o o d -s t a t e - d e p e n d e n t e f f e c t s u n d e r t h r e e c o n d i t i o n s o f r e c a l l o n D a y 2 .
signi® can t interactio n between m ood on D ay 1 an d m ood on D ay 2, F(1, 16 ) = 25.30, p <
.00 1; a signi® cant interaction between m ood on D ay 1 an d recall con ditions on D ay 2,
F(2, 32) = 9.63, p < .01, a sig ni® can t interaction between m ood on D ay 2 an d recall
co nditions, F(2, 32) = 7.10, P < .01, an d, ® nally, a highly signi® can t three-w ay interactio n
between m ood on D ay 1, m ood o n D ay 2, an d recall condition on D ay 2, F(2, 32) = 31.08,
p < .001.
In ord er to explore th e three-w ay interaction (A 3 B 3 C ) in g reater detail, tests of
sim ple m ain effects were used (K irk, 19 68, p. 291). T hese post h oc com pariso ns of m ean s
co ntrasted sam e-m ood an d chan ged-m ood sub jects sep ar ately for each m em ory test.
B etween-groups A na lyses. T he b etween-g roups an alyses revealed that the difference
on D ay 2 b etween recall o f subjects in g roup H H (hap py on D ay 1, h ap py o n D ay 2)
an d subjects in G roup H S (hap py on D ay 1, sad on D ay 2) w as signi® can t in the ® rst
free-recall co ndition, F(1, 3) = 35.01, p < .01, w as n ot sign i® can t in the cued-recall
co ndition, F(1, 3) < 1, an d w as signi® can t in the second free-recall co ndition, F(1, 3) =
49.2, p < .01.
Tur ning to su bjects in the g roup SS (sad o n D ay 1, sad on D ay 2) an d g roup S H (sad
on D ay 1, hap py on D ay 2), the d ifference o n D ay 2 between subjects’ recall in th ese two
g roups w as signi® can t in the ® rst free-recall condition, F(1, 3) = 36.2, p < .01, an d w as
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
313
Downloaded By: [University College London] At: 02:00 2 February 2011
not signi® can t in the cued-recall co nd ition, F(1, 3) < 1, neither w as the difference signi® can t in the second free-recall condition, F(1, 3) < 1.
Within-groups A na lyses. T he w ithin-g roups an alyse s exam ined the recall of su bjects
w ithin the four g ro up s across the th ree conditions of recall o n D ay 2. T his revealed that the
recall of su bjects in different m ood states on D ay 2 differed sign i® can tly across conditions:
G roup H S, F(2, 32) = 71.3 5, p < .001, an d g roup SH , F(2, 32) = 59.59, p < .001; w hereas
recall of subjects in th e sam e m ood states on D ay 2 did not differ across co nditions: g rou p
H H , F(2, 32) = 3.01, n.s., an d g roup S S, F(2, 32) = 2.05, n .s.
To explore the signi® can t effects in g reater detail, a series of N ew m an -K euls’ com p arison s were co nducted on th e m ean s of each co ndition for G roups H S an d SH separately,
w ith p set at .05. T hese revealed that, for g roup H S, cued recall w as superior to each of
the free-recall co nditions, ps < .05, w hich them selves did not differ, p > .05. For g roup
SH , cued recall w as superior to the ® rst free-recall condition , p < .05, but did not d iffer
from the second free-recall co ndition, p > .05, an d this second free-recall co ndition w as
su perior to the ® rst free-recall con dition, p < .05 (see Figure 1).
D is cu s s io n
T he results showed that listen ing to m usic successfully induced hap piness an d sad ness in
su bjects. M ood induction did not differentially affect subjects’ ab ility to lear n the task o n
D ay 1Ð there w as no signi® can t difference between the lear ning scores of hap py induction an d sad induction subjects o n D ay 1.
T he p rediction of this exp eriment w as that the ® rst free-recall perfor m an ce of
su bjects would show a state-dependent retrieval effect an d that cued-recall perfor m an ce
would not. T he data from the between-g roups an alyses support this prediction ; a statedependent retrieval effect w as observed in the ® rst free-recall cond ition, w ith su bjects
w ho lear ned an d recalled in the sam e m ood states recalling signi® can tly m ore item s
than subjects w h o lear ned an d recalled in different m ood states. T his state-dependent
effect w as n ot ap parent in the cued-recall co ndition, w here there w ere no signi® can t
differences between the recall scores of subjects w ho recalled in the sam e m ood states
an d subjects w h o recalled in different m ood states (to those of D ay 1), indicating that
the provision of a ``copy’ ’ cu e at tim e of recall en abled all subjects to retrieve m aterial
effectively reg ardless of a chan ge of m ood state. T he ® nding of E xperim ent 2, that the
cu e w as effective fo r sub jects w ho lear ned in a sad m ood an d recalled in a hap py m ood
but not for subjects w ho lear ned in a hap py m ood an d recalled in a sad m ood w as not
replicated h ere; indeed, Table 4 sh ow s that not only w as the cu e m ore effective for
su bjects in g roup H S in this experiment com pared w ith subjects in the previous
ex periment, b ut also that su bjects in this group h ad higher m ean recall scores in the
cu ed condition than d id subjects in g roup SH . T he data m ay be accounted for by the
fact th at subjects were required to free-recall in this experiment before being presented
w ith a retrieval cue, an d this initial recall m ay have enabled them to use the cue m ore
effectively in the cued-recall session.
W ith respect to the tran sfer of effective retrieval strategies, the question investigated
w as w heth er the state-depen dent retrieval effect erased in the cued-recall condition w ould
Downloaded By: [University College London] At: 02:00 2 February 2011
314
KEN EA LY
reap pear in a second free-re call condition, or w hether subjects would b e able to utilize the
effective retrieval cue provided in the cu ed-recall condition to ne gate the deleterious
effect of a chan ge of m ood state. T he b etween -g roup s an alysis indicates the re-em ergence
of the state-dependent effect for su bjects w ho w ere recalling in a sad m ood after having
lear ned in a h ap py m ood Ð their percentage recall w as signi® can tly lower than th at of
su bjects w ho had lear ned an d recalled in a hap py m ood state. H ow ever, the percentage
recall scores of subjects w ho learn ed in a sad m ood an d recalled in a hap py m ood were not
signi® can tly d ifferent from those of su bjects w ho learned an d recalled in the sam e sad
m ood states.
T hus it ap p eare d that providin g an observable retrieval cue to su bjects in a sad m ood
en abled them to recall m aterial lear nt w hilst in a previously hap py m ood, but su bjects
ap pear to b e unable to tran sfer this ability to situations w here the cue is n ot explicitly
presented. T he w ithin-group an alyses su pport this explanatio n in that subjects w ho
lear ned in hap py m oo d an d recalled in a sad m ood showed signi® can tly better recall in
the cued-recall co ndition com p ared to the ® rst free-recall condition, but that this
improvem ent w as not su stained in the second free-recall condition, w here percentage
recall reverted to the poor perfor m an ce ob served in the ® rst free-recall co nditionÐ that
is, recall w as ag ain affected by a chan ge of m ood state from the initial lear ning m ood state.
By com parison , sub jects w ho learn ed in a sad m ood an d recalled in a hap py m ood show ed
signi® can tly su perior recall in the cued condition com pared to the ® rst free-recall condition, an d th is improvem ent in recall w as m aintained in the second free-recall co ndition,
w here p recentage recall scores were signi® can tly higher than those of the ® r st free-recall
co ndition. For subjects in this group, the provisio n of an observable ``copy’ ’ cue enabled
them effectively to override the deleterious effects of a chan ge of m ood state an d to
m aintain this effective retrieval of m aterial in a further free-recall condition.
In co nclusion, the som ew hat unreliable natu re of the effectiveness of ob servable
``copy’ ’ cues to over ride the deleterious effects of a chan ge of m oo d state w ar ran ts further
investig ation; cues w ere show n to b e effective in this experiment, but not und er all
co nditions in E xperiment 2. T he ® n dings reported here suggest that m aking accessible
to subjects in a sad m ood m aterial that they previously encoded in a hap py m ood does not
en able them to self-generate ap propriate retrieval cu es to access that sam e m aterial at a
later time. B y an alogy, the results m ay also sug gest that once som eo ne in a hap py m ood is
able to retrieve m aterial encoded in a sad m ood, the accessibility of this m aterial is
retained or tran sfer red to subseq uent occasio ns of recall, reg ardless of th e dissim ilarity
between encoding an d retrieval m ood states.
R EFER EN CES
Badd eley, A.D. (1982). D om ains of recolle ction . P sychologica l R eview, 89, 708± 729.
Badd eley, A.D. (199 0). Huma n memory: theor y a nd practice (p p. 390± 397). H ove, UK : L aw rence E rlbau m
Associates L td .
Bartlett, J.C., Burleson, G., & S an trock, J.W. (1982). Emotiona l m ood an d mem ory in you ng children.
J ourna l of Experimenta l Child P sychology, 34, 59± 76.
Bartlett, J.C., & S an trock, J.W. (1979). Affect-d ependent episodic m emory in you ng children. Child
Development, 50, 513± 51 8.
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
315
Beck, A.T. (1976). Cognitive thera py a nd emotiona l disorders. Inter nation al Univer sities Press, N ew Yor k.
B erg er, B .D., & S te in , L . (1 96 9). A n an alys is of th e lear ning de® cits prod uce d b y scop o lam ine.
P sychopha rma cologia , 14 , 27 1± 283.
Blane y, P.H. (1986). Affect an d m emo ry: A review. P sychologica l B ulletin, 99, 229± 246.
Bower, G.H. (1981). M oo d an d M em or y. American P sychologist, 36, 129± 14 8.
Bower, G.H . (1987). Com m entary on mood an d memory. B eha viour R esea rch a nd Thera py, 25, 443± 45 5.
Bower, G.H ., Gillig an , S.G., & M on teiro, K .P. (19 81). S elec tivity of lear ning cau sed by affective states.
J ourna l of Experimenta l P sychology: Genera l, 110 , 451± 47 3.
B ow er, G.H ., & M aye r, J.D. (198 5). Failu re to replicate m oo d-dependen t retrieval. B ulletin of the
P sychonomic S ociety, 23, 39± 4 2.
Bower, G.H ., & M ayer, J.D. (1991) . In search of m oo d-dependent retr ieval. In D. Ku iken (E d.), Mood
a nd memory: Theor y, resea rch a nd a pplica tions (pp. 133± 168). L on don : Sage P ublicatio ns.
Bower, G.H ., M on teiro, K .P., & Gillig an , S.G. (19 78). E motion al m ood as a context of lear ning an d
recall. J ourna l of Verba l Lea rning & Verba l B eha viour, 17, 573± 585.
B uchwald , A.M ., S trac k, S ., & Co yn e, J.C. (198 1). D em an d char ac te ristics an d th e Velten M oo d
Inductio n P roced ure. J ourna l of Consulting a nd Clinica l P sychology, 49, 478± 479 .
Clark, D.M . (1983). O n th e inductio n of depressed m ood in th e labor atory : Evalu ation an d com parison
of th e Velten an d M usical P roc edures. Adva nces in B eha viour Resea rch a nd Thera py, 5, 27± 49.
Clark, D.M ., & Teasd ale, J.D. (198 5). Co nstrain ts o n th e effe cts of m ood o n m em or y. J ourna l of
P ersona lity a nd S ocia l P sychology, 48, 1595± 16 08 .
Davies, G.M ., & T hom so n, D.M (1988). Con text in con text. In G.M . D avies & D.M . Th om son (E ds.),
Memory in context: Context in memory (p p. 33 5± 345), Chichester, U K: John W iley.
Eich, E. (1980). T he cu e-dependent natur e of state- dependent retrieval. Memory & Cognition, 8, 157± 17 3.
Eich, E. (1989). T heor etical issues in state dependent m em ory. In H.L . Roed iger, III, & F.I.M . Craik
(E ds.), Va rieties of memory a nd consciousness. Essa ys in honour of E ndel Tulving (pp. 331± 354). Hillsdale,
NJ: L awr ence Erlbau m Associates, Inc.
Eich, E., & Bir nbaum , I.M . (1987). O n th e relationship betwee n th e dissociative an d effective prop erties
of drugs. In G.M . D avies & D. T hom so n (E ds.). M emory in context: Context in memory (p p. 81± 93).
Chichester, U K: John W iley.
Eich, E., M acau lay, D., & Ryan, L . (1994). M oo d-dependent m em or y for even ts of the perso nal past.
J ourna l of Experimenta l P sychology: Genera l, 123 , 201± 21 5.
Eich, E., & M etcalfe, J. (19 89 ). M ood -dependent m emor y for inter nal ver sus exte rnal even ts. J ourna l of
Experimenta l P sychology: Lea rning, M emory a nd Cognition, 15, 443± 455.
Eich, E., Weingar tn er, H., Stillm an, R.C., & Gillin, J.C. (1975). S tate -dependent accessib ility of retrieval
cues in th e retention of a cate go rized list. J ourna l of Verba l Lea rning & Verba l B eha viour, 14, 408± 41 7.
Ellis, H .C., & Ashbrook, P.W. (1988). Resou rce allocati on mod el of th e effec ts of depressed m ood states
on memory. In K . F iedler & J. Forg as (E ds.), Affect, cognition a nd socia l beha viour. Toro nto, Can ad a:
Hog refe.
Ellis, H .C., & Ashbrook, P.W. (1991). T he ``state’ ’ of mood an d mem or y research: A selective review. In
D. Ku ike n (E d.), M ood a nd memory: theory, resea rch a nd a pplica tions (pp. 1± 21). L o ndon : S age
Pu blic ation s.
Foa, E.B., M cNally, R., & M urdock, T.B. (1989 ). Anxious m ood an d m em or y. B eha viour R esea rch a nd
Thera py, 27 , 141± 147.
G ag e, D.F., & S afer, M .A. (1985 ). H em isphere differe nces in th e m ood -state- dependent effe ct for
reco g n iti o n o f em o tio na l f ac es. J ourna l of E x perimenta l P sy chol ogy : L ea rning, M emor y, a nd
Cognition, 11, 752± 763.
Goodw in, A.M ., & W illiams, J.M .G. (198 2). M ood induction s researchÐ its im plicatio ns fo r clinical
depression. B eha viour R esea rch a nd Thera py, 20, 373± 38 2.
Haag a, D. (1989). M ood -state -dependent retentio n using identical or non -identical mood induction s at
lear ning an d recall. B ritish J ourna l of Clinica l P sychology, 28, 75 ± 83.
Hasher, L ., Rose, K .C., Zac ks, R.T., S an ft, H ., & Dor en, B. (198 5). M ood recall an d sele ctivity in
nor mal college students. J ourna l of Experimenta l P sychology: Genera l, 114 , 104± 118.
Downloaded By: [University College London] At: 02:00 2 February 2011
316
KEN EA LY
Hertel, P.T., & Hardin, T.S. (1990). Rememberin g with an d with ou t awaren ess in a depressed m oo d:
Evide nce of de® cits in initiative. J ourna l of Experimenta l P sychology: Genera l, 119 , 45± 59.
Hertel, P.T., & Rude, S.S. (1991) . D epressive de® cits in m emory: Focusing atten tio n improves subsequent recall. J ourna l of E xperimenta l P sychology: Genera l, 120 , 301 ± 309.
Ing ram , R.E. (1984). Infor mation processing an d fee dback: Effects of mood infor mation favor ability on
the cogn itive processing of personally relavant infor m atio n. Cognitiv e Thera py a nd R esea rch, 8, 371± 38 6.
Isen, A.M ., S halker, T.E ., Clark, M ., & K ar p, L . (1978) . Affect, ac cessibility of m aterial in mem or y, an d
behaviour: A cogn itive loop ? J ourna l of P ersona lity a nd S ocia l P sychology, 36, 1± 12.
Ken ealy, P.M . (19 82). State- dependent recall: T he effects of alcoh ol and induced mood on m emor y.
B ulletin of the B ritish P sychologica l S ociety, 35, A.83.
Ken ealy, P.M . (1986) . T he Velten M ood Inductio n Proced ure: A m eth od ological review. M otiva tion a nd
Emotion, 10, 315± 33 5.
Ken ealy, P.M . (1988). Valid ation of a M usic M ood Inductio n P roc edure: S om e preliminary ® ndings.
Cognition a nd Emotion, 2, 41± 48.
K ihlstrom , J.F. (1991) . O n what does m ood -dependent m em ory depend? In D. Ku iken (E d.), M ood a nd
memory: theor y, resea rch a nd a pplica tions (p p. 23± 32 ). L on don : Sage Pu blic ation s.
K irk, R.E. (1968). Experimenta l design: P rocedures for the beha vioura l sciences. Belm on t, CA: B rook s/Cole
Pu blishing Com pan y.
Ku ike n, D. (E d.). (1991) . M ood a nd memory: Theory, resea rch a nd a pplica tions. Lo ndon : S age P ublication s.
L eigh t, K .A., & Ellis, H .C. (1981 ). Emotio nal mood states, strategies, an d state-d ependency in m emor y.
J ourna l of Verba l Lea rning a nd Verba l B eha viour, 20, 251± 266.
M ecklen br au ker, S., & Hager, W. (198 4). Effec ts of mood on mem or y: Exp erim ental tests of a m oo dstate- dependent retrieval hypoth esis and a m ood -co ng ruity hyp oth esis. P sychologica l R esea rch, 46,
335± 376.
Nasby, W., & Yan do, R. (1982 ). Selective encod ing and retrieval of affectiv ely valen t info rm ation : Tw o
cogn itive con sequences of children’s m ood states. J ourna l of P ersona lity a nd S ocia l P sychology, 43,
1244± 1253.
O ver to n, D.A. (1966) . S tate-d ependent lear ning prod uced by depressan t an d atr op hine-like drugs.
P sychopha rma cologia , 10 , 6± 31.
O ver ton , D.A. (1974) . Exp erimental m eth od s for th e study o f state-d ependent lear ning. Federa tion
P roceedings, 33, 1800± 18 13.
Par rott, W.G., & Sabini, J. (1990). M o od an d mem ory under natur al con dition s: Evid ence for m ood
incong ruent recall. J ourna l of P ersona lity a nd S ocia l P sychology, 59, 321± 33 6.
Peete rs, R., & d’ Ydew alle, G. (1987). In¯ uences of emotion al states upon m emor y: Th e state of th e art.
Communica tion & Cognition, 20, 171± 19 0.
Polivy, J., & Doyle, C. (1980) . S hor t repor ts: L abor atory induction of mood states th rough reading selfrefe rent m o o d state m ents: A ffe ctiv e chan g es o r dem an d ch ar ac te ristic s? J ourna l of A bnor ma l
P sychology, 89, 286± 290 .
Reus, V.L ., Weingartn er, H ., & P ost, R.M . (1979). Clinical im plication s of state-d ependent lear ning.
American J ourna l of P sychia try, 136 , 927± 931 .
Ryb ack, R .S. (1971). T he con tinuum an d speci® city of th e effects of alcoh ol on memory. Qua rterly
J ourna l of S tudies of Alcohol, 32 , 995± 998.
Sch ar e, M .L ., L isman, S.A ., & Sp ear, N.E. (1984). T he effec ts of mo od variati on on state -dependent
retention . Cognitive Thera py a nd Resea rch, 8, 387± 408.
Sw an so n, J.M ., & K insbou rne, M . (1976). Stim ulant- related state-d ependent lear ning in hyp erac tive
children. S cience, 192 , 1354 ± 1357.
Teas dale, J.D. (1983 ). Negative th inkin g in depression: Cause, effe ct or reciprocal relationship? Adva nces
in B eha v iour R esea rch & Thera py, 5, 3± 25.
Teas dale, J.D., & Fog ar ty, S.J. (1979). Differen tial effects of induce d m ood on retrieval of pleasan t an d
unpleas an t even ts from episodic m emory. J ourna l of Abnorma l P sychology, 88, 248 ± 257.
Teas dale, J.D., & Taylor, R. (1981 ). Induced m ood an d th e ac cessibility of mem ories: An effect of m ood
state or of mood induction proc edure? B ritish J ourna l of Clinica l P sychology, 20, 39± 48.
Downloaded By: [University College London] At: 02:00 2 February 2011
M O O D -S T A T E ± D E P E N D E N T R E T R I E V A L
317
Teas dale, J.D., Taylor, R., & Fog ar ty, S.J. (1980). Effects of induced elation -depression o n th e accessibility of mem or ies of hap py an d unhappy exp eriences. B eha viour R esea rch & Thera py, 18, 339± 34 6.
T hom son , D.M ., & Tu lvin g, E. (1970) . Associativ e encod ing an d retrieval: Weak an d stro ng cues.
J ourna l of Experimenta l P sychology, 86, 255± 262.
Tu lvin g, E. (1983). Elements of episodic memory. O xford: Clarendon P ress.
Tu lvin g, E., & M ad igan , S.A . (1970) . M emor y an d ver bal lear ning. Annua l R eview of P sychology, 21,
437± 484.
Tu lvin g, E., & T hom so n, D.M . (19 71). Retrieval processes in recogn ition m emor y: Effects of associative
con text. J ourna l of Experimenta l P sychology, 87, 116 ± 124.
Tu lvin g, E., & T hom son, D.M . (1973) . Encod ing speci® city an d retrieval processes in episod ic m emor y.
P sychologica l R eview, 80, 353± 373 .
Velten , E. (1968). A labor atory tas k for induction of mood state s. B eha viour R esea rch & Thera py, 6,
473± 482.
Weingar tn er, H . (197 8). Hum an state- dependent lear ning. In B.T. Ho., D.W. R ichards, & D.L. Chute
(E ds.). Drug discrimina tion a nd sta te-dependent lea rning (p p. 361± 380 ). L on do n: A cade mic P ress.
Wester mann , R ., & Hager, W. (1983) . O n severe te st of tren d hyp oth eses in psychology. The P sychologica l
Record, 33 , 201± 211.
Wetzler, S. (1985). M ood -state-dependent retrieval: A failu re to replicate. P sychologica l R eports, 56,
759± 765.
Origina l ma nuscript receiv ed 28 October 1994
Accepted revision receiv ed 25 J uly 1996
Download