Implicit and Explicit

Memory and Learning

John F. Kihlstrom
University of California, er!eley
Jennifer "orfman
#orth$estern University
Lillian %ar!
University of California,
er!eley

Note: An edited version of this article appeared in M. Velmans & S.
Schneider (Eds.), The Blackwell Companion to Consciousness. !ford,
".#.$ Blackwell, %&&'.

(earnin) and memor* are ine!trica+l* intertwined. The capacit* for
learnin) pres,pposes an a+ilit* to retain the knowled)e ac-,ired
thro,)h e!perience, while memor* stores the +ack)ro,nd knowled)e
a)ainst which new learnin) takes place. .,rin) the dark *ears of
radical +ehaviorism, when the concept of memor* was deemed too
mentalistic to +e a proper s,+/ect of scientific st,d*, research on
h,man memor* took the form of research on ver+al learnin)
(Anderson, %&&&0 Schwart1 & 2eis+er), 3443).

Explicit and Implicit Memory
5n the earliest *ears of scientific ps*cholo)*, research foc,sed on
immediate conscio,s e!perience, in the form of sensations and
percepts anal*1ed first +* ps*choph*sicists like 6e+er and 7echner
and then +* str,ct,ralists like 6,ndt and Titchener. 6,ndt +elieved
that 8hi)her8 mental processes were not amena+le to e!perimental
st,d*. B,t 9ermann von E++in)ha,s proved him wron) in 3::;$ +*
co,ntin) repetitions and calc,latin) savin)s in relearnin), E++in)ha,s
invented the ver+al<learnin) paradi)m that has dominated the
scientific st,d* of memor* ever since (T,lvin) & =raik, %&&&).

%rinciples of Conscio&s 'ecollection
7or most of the cent,r* followin) E++in)ha,s, the ps*cholo)* of
memor* was concerned with conscio,s recollection with the a+ilit*
to recall or reco)ni1e events that had occ,rred in the past. 7rom this
research has emer)ed a small set of principles that lar)el* )overn how
h,man memor* operates (#ihlstrom, 344>).
(tage )nalysis$ Memories are analo)o,s to +ooks in a li+rar*, or the
information contained within them$ mental representations of events
are encoded as memor* traces, which are retained in memor* stora)e
and retrieved in the co,rse of on)oin) e!perience, tho,)ht, and
action.
Ela*oration$ Memor* for an event is a f,nction of the de)ree to which
that event is related to pre<e!istin) knowled)e at the time of
encodin).
+rgani,ation$ Memor* is also a f,nction of the de)ree to which
events are related to each other.
-ime."ependency$ Memor* fades with time, mostl* +* virt,e of
interference amon) competin) memor* traces.
)vaila*ility vs. )ccessi*ility$ Encoded memories, availa+le in
memor* stora)e, ma* not +e accessi+le when retrieval is attempted.
C&e."ependency$ The pro+a+ilit* of retrievin) an event is a f,nction
of the informational val,e of c,es provided at the time of retrieval.
Encoding (pecificity$ 2etrieval is most effective when c,es present
at the time of retrieval match those processed at the time of encodin).
(chematic %rocessing$ Events that are relevant to c,rrentl* active
+eliefs, e!pectations, and attit,des are remem+ered +etter than those
that are irrelevant0 events that are incon)r,ent with these mental
schemata are remem+ered +etter than those that are con)r,ent.
'econstr&ction$ Memor* reflects a mi! of information contained in
the memor* trace and knowled)e derived from other so,rces0 in the
final anal*sis, memories are +eliefs, and remem+erin) an event is
more like writin) a stor* from fra)mentar* notes, than readin) it from
a +ook.
Interpersonal$ 2emem+erin) is an act of interpersonal
comm,nication as well as of information retrieval, and so memories
are shaped +* the social conte!t in which remem+erin) occ,rs.

-axonomy of Memory and Kno$ledge
These principles appl* to so<called 8lon)<term8 memor* (6illiam
?ames called it secondary memory). B,t the domain of memor* also
incl,des modalit*<specific sensory registers (e.)., iconic and echoic
memor*), and primary memory (also known as short-term or working
memor*), which ma* operate on somewhat different principles. 7or
e!ample, the time<dependenc* characteristic of for)ettin) from the
sensor* re)isters is prod,ced +* deca* and displacement rather than
interference. @rimar* memor* is sometimes viewed as a separate
memor* s*stem from secondar* memor*0 in other theories, primar*
memor* is identified with representations stored in lon)<term memor*
that are c,rrentl* in a state of activation. @rimar* or workin) memor*
is closel* identified with conscio,sness.
The knowled)e stored in lon)<term memor* comes in two +road
forms. Declarative knowledge constit,tes o,r f,nd of fact,al
knowled)e, and can +e represented +* sentence<like propositions.
Procedural knowledge consists of o,r co)nitive repertoire of r,les and
skills, and can +e represented +* 8if<then8 str,ct,res known as
prod,ctions. 6ithin the domain of declarative knowled)e, we can
distin),ish episodic memor*, or a,to+io)raphical memor* for events
that have occ,rred in o,r personal past, and semantic memor*, a sort
of impersonal mental dictionar*. @roced,ral knowled)e can +e f,rther
classified into motoric and percept,al<co)nitive skills. The declarative<
proced,ral distinction has its immediate ori)ins in comp,ter science
and artificial intelli)ence, +,t can +e traced +ack to 2*les distinction
(in The Concept of Mind, 34A4) +etween 8knowin) that8 and 8knowin)
how8, and Ber)sons assertion (in Matter and Memory3433) that
8the past s,rvives in two forms8 << as recollections and as ha+its.
Episodic memor* is what most people mean +* 8memor*8, as opposed
to 8knowled)e8.

%riming Effects in )mnesia and #ormal Memory
7or most of its histor*, the scientific st,d* of episodic memor* was
concerned mostl* with conscio,s recollection, to the e!tent that it was
concerned with conscio,sness at all, and the notion of ,nconscio,s
memor* was rele)ated mostl* to the 7re,dian fantas*land. B,t
+e)innin) in the 34>&s, research +e)an to s,))est that the notion of
,nconscio,s memories was valid after all if not in the 7re,dian form.
f partic,lar interest were st,dies of patients with the amnesic
syndrome associated with +ilateral dama)e to the hippocamp,s and
related str,ct,res in the medial temporal lo+e, or to the mammillar*
+odies and related str,ct,res in the diencephalon. 5n 34>:, 6arrin)ton
and 6eiskrant1 reported an e!periment in which amnesic patients
were asked to st,d* a list of familiar words. =ompared to control
s,+/ects, the patients performed ver* poorl* on standard tests of recall
and reco)nition. 9owever, when the* were presented with three<letter
stems or fra)ments, and asked simpl* to complete the c,es with the
first word that came to mind, amnesics and controls were e-,all* likel*
to complete the c,es with items from the st,died list.
This is a priming effect, in which the processin) of one item infl,ences
the processin) of another item. 5n positive primin), the prime
facilitates processin) of the tar)et0 in ne)ative primin), the prime
inhi+its processin) of the tar)et. 5n this instance, the primin) effect
indicates that the st,died items were encoded in memor*, retained in
stora)e, and infl,enced performance on the completion test. The fact
that e-,ivalent levels of primin) occ,rred in ne,rolo)icall* intact
s,+/ects, who remem+ered the primin) episode normall*, and amnesic
patients, who had ver* poor memor*, indicates that primin) can +e
dissociated from conscio,s recollection. n the +asis of evidence like
this, Schacter distin),ished +etween two e!pressions of episodic
memor*$ e!plicit and implicit (Schacter, 34:'). E!plicit memor* refers
to conscio,s recollection of a past event, as e!emplified +*
performance on recall and reco)nition tests. B* contrast, implicit
memor* refers to an* effect of an event on s,+se-,ent e!perience,
tho,)ht, or action. @rimin) is, of co,rse, /,st s,ch an effect. The
dissociation +etween primin) and recall in amnesic patients indicates
implicit memor* can persist in the a+sence of e!plicit memor*.
Spared primin) in amnesic patients has now +een confirmed in a host
of st,dies, and has +een e!tended to a wide variet* of other forms of
amnesia$
antero)rade and retro)rade amnesia occ,rrin) as a conse-,ence of
electroconv&lsive therapy (E=T) for depression0
antero)rade amnesia prod,ced +* general anesthesia administered
to s,r)ical patients (see also the article on 8Anesthesia8, +* #ihlstrom
and =ork, elsewhere in this vol,me)0
antero)rade amnesia associated with conscio&s sedation in
o,tpatient s,r)er* (see also 8Anesthesia8)0
memor* disorders o+served in dementia, incl,din) Al1heimers
disease, as well as those enco,ntered in normal aging0
h*pnotic and posthypnotic amnesia followin) appropriate
s,))estions to h*pnoti1a+le s,+/ects0
8f,nctional8 or 8ps*cho)enic8 amnesias enco,ntered in )en,ine cases
of dissociative disorder, incl,din) dissociative amnesia, dissociative
f,),e, and the interpersonalit* amnesia of dissociative identit*
disorder (also known as m,ltiple personalit* disorder).
5n each of these cases, the memor* disorder primaril* impairs e!plicit
memor* and spares implicit, which is either wholl* or relativel* intact.
5t is in this sense that implicit memor* persists in the a+sence of
e!plicit memor*. 9owever, implicit memor* can +e o+served in
individ,als with normal memor* f,nctions as well . 7or e!ample,
normal s,+/ects show si)nificant savin)s in relearnin) for items that
the* can neither recall nor reco)ni1e. And while ela+oration is an
important determinant of e!plicit memor*, 8depth of processin)8 has
relativel* little impact on man* primin) effects. 5n nonamnesic
individ,als implicit memor* ma* +e said to +e independent of e!plicit
memor*, in that primin) does not depend on whether the prime is
conscio,sl* remem+ered.

-he /oca*&lary of Implicit Memory
5n )eneral, dissociations +etween e!plicit and implicit memor* come in
several forms (2ichardson<#lavehn, Bardiner, & ?ava, 344>). 5n
population dissociations, a condition like amnesia or a)in) affects
e!plicit memor*, +,t not implicit memor*. 5n functional dissociations,
an e!perimental varia+le (like depth of processin)) affects e!plicit
memor* +,t not implicit memor*. f co,rse, the dissociations can also
)o the other wa*$ shiftin) from a,ditor* presentation at time of st,d*
to vis,al presentation at time of test can have a +i) effect on implicit
memor*, +,t relativel* little effect on e!plicit memor*. 5n single
dissociations, a sin)le varia+le effects one e!pression of memor*,
e!plicit or implicit, +,t not the other. 5n double dissociations, a sin)le
varia+le has opposite effects on e!plicit and implicit memor*. .o,+le
dissociations are the 89ol* Brail8 of co)nitive ne,rops*cholo)*,
+eca,se the* provide compellin) evidence that two f,nctions, s,ch as
e!plicit and implicit memor*, are mediated +* separate co)nitive
mod,les or +rain s*stems. B,t the* are also e!ceedin)l* rare. Man*
ostensi+le do,+le dissociations are more like twin dissociations, in
which one varia+le affects e!plicit +,t not implicit memor*, while
another varia+le affects implicit +,t not e!plicit memor*.
5mplicit memor* is ,s,all* tested with a primin) task, +,t primin)
comes in a n,m+er of different forms. Most research has foc,sed on
repetition priming, in which the tar)et of the primin) test is a
recapit,lation, in whole or in part, of the prime itself. 7or e!ample,
s,+/ects mi)ht st,d* a word like doctor and then +e asked to complete
the stem doc- or the fra)ment d-c-o- with the first word that comes to
mind, to identif* the word doctor when presented a)ainst a nois*
+ack)ro,nd, or to decide whether the letter strin) doctor is a le)al
word. B,t semantic priming effects can also +e o+served when
s,+/ects who have st,died a word like doctor are asked to )ive free
associations to c,es like nurse, or to )enerate instances of cate)ories
like occupations. 2epetition primin) can +e mediated +* a perception-
based representation that is limited to the ph*sical attri+,tes of the
prime and its confi),ration in space and time, +,t semantic primin)
re-,ires a meaning-based representation that incl,des information
a+o,t the semantic and concept,al feat,res of the prime. Semantic
primin) can +e st,died with the same tasks normall* ,sed to meas,re
repetition primin), s,ch as percept,al identification and le!ical
decision, provided that the tar)et does not recapit,late the prime.
E!plicit and implicit memor* are sometimes referred to as
8declarative8 and 8proced,ral8 memor*, or 8declarative8 and
8nondeclarative8 memor* (S-,ire, #nowlton, & M,sen, 344C),
respectivel*. The declarative<proced,ral distinction was initiall* +ased
on the view that preserved learnin) in amnesia was limited to
proced,ral knowled)e s,ch as co)nitive and motor skills, and an
interpretation of primin) as proced,ral (if-then) in nat,re. B,t amnesic
patients can ac-,ire new declarative knowled)e as well, provided that
the* do not have to remem+er the circ,mstances in which the*
learned it0 and semantic primin) is ar),a+l* mediated +* semantic
memor*, which is a component of declarative knowled)e. 6hile some
implicit e!pressions of memor* ma* +e mediated +* proced,ral or
nondeclarative knowled)e, the declarative<nondeclarative distinction
risks conf,sin) the interpretation of e!plicit memories as
representations that can +e conscio,sl* 8declared8 with the
propositional format in which declarative knowled)e is represented.
Similarl*, tests of e!plicit and implicit memor* are sometimes referred
to as 8direct8 and 8indirect8. That is to sa*, recall tests memor*
directl*, while savin)s or primin) test memor* indirectl*. 5t sho,ld +e
,nderstood, tho,)h, that the direct<indirect distinction applies to
memor* tests and not to e!pressions of memor*. 5n principle, primin)
co,ld +e ,sed to assess conscio,sl* accessi+le memories that the
s,+/ect declines to report, m,ch as ps*choph*siolo)ical meas,res are
,sed in forensic lie<detection. Alon) the same lines, e!plicit and
implicit memor* are sometimes referred to as 8intentional8 and
8incidental8 respectivel*. That is to sa*, in recall tests s,+/ects are
instr,cted to intentionall* remem+er some past event, while primin)
occ,rs incidentall* when the s,+/ect is performin) some non<memor*
task. The intentional<incidental distinction reminds ,s that there are
two aspects of conscio,sness relevant to memor*$ conscio,s
awareness and conscio,s control (B,tler & Berr*, %&&3). A conscio,s
memor* mi)ht well emer)e, ,nintentionall* and inadvertentl*, in the
co,rse of a primin) test a sit,ation that is sometimes referred to as
8invol,ntar* e!plicit memor*8. 5nvol,ntar* conscio,s recollection has
+een a topic of literar* disc,ssion at least since the time of @ro,st
(Salaman, 34'&), and of scientific investi)ation since the time of
Balton (=rovit1, 34'&), +,t it sho,ld +e distin),ished from implicit
memories that are inaccessi+le to conscio,s recollection in the first
place.
5n the final anal*sis, +oth the 8direct<indirect8 and 8intentional<
incidental8 dichotomies fail to capt,re the essence of the e!plicit<
implicit distinction which is that e!plicit memor* is conscio,s
recollection, and implicit memor* is ,nconscio,s memor*, of the past.
B,t if e!plicit memor* is ,nconscio,s memor*, wh* not simpl* call it
thatD The answer is more likel* to +e fo,nd in sociolo)* than
ps*cholo)*, as those who wo,ld make a science of ,nconscio,s mental
life have so,)ht to avoid the taint of 7re,dian ps*choanal*sis. Even
witho,t the specter of 7re,d loomin) over their sho,lders, the topic of
conscio,sness still makes some ps*cholo)ists nervo,s (7lana)an,
344%). Still, what makes implicit memor* interestin) is not that
implicit tests provide indirect, possi+l* s,rreptitio,s, assessments of
what a person remem+ers0 nor that implicit e!pressions of memor*
occ,r invol,ntaril*. 6hat makes implicit memor* interestin) is that it
represents the d*namic infl,ence of memor* in the a+sence of
conscio,s recollection.

-heories of Implicit Memory
Based on the 8mod,larit*8 view pop,lar in co)nitive ne,roscience, a
n,m+er of theorists have s,))ested that e!plicit and implicit memor*
reflect the performance of separate memor* s*stems in the +rain
(Schacter & T,lvin), 344A). 7or e!ample, S-,ire has identified e!plicit
memor* with a medial temporal<lo+e memor* s*stem incl,din) the
hippocamp,s and related str,ct,res (S-,ire & Eola<Mor)an, 3443).
.ama)e to this s*stem will impair e!plicit memor* +,t spare implicit
memor*, which is mediated +* other +rain s*stems, pres,ma+l*
cortical in nat,re. At the other end, T,lvin) and Schacter have
proposed that repetition primin) is mediated +* a set of perceptual
representation systems that store representations of the ph*sical
str,ct,re of the prime, +,t not its meanin) (T,lvin) & Schacter, 344&).
7or e!ample, a visual word form system associated with the
e!trastriate corte! mediates vis,al stem<completion, while an auditory
word-form system mediates a,ditor* percept,al identification.
Semantic primin), in t,rn, is held to +e mediated +* a separate
semantic memor* s*stem.
Another take on the m,ltiple<s*stems view is provided +* Bowers and
Marsolek (Bowers & Marsolek, %&&C). 5nstead of invokin) m,ltiple
memor* s*stems, the* propose that implicit memor* is a +*prod,ct of
+rain s*stems that are devoted to percept,al pattern reco)nition,
concept,al processin), and motor +ehavior, rather than memor* per
se. n their view, implicit memor* is a +*prod,ct of the learnin)
capa+ilit* of these s*stems. These s*stems have memor*, in that the*
are capa+le of encodin) and reco)ni1in) information, +,t the* are not
memor* s*stems. Altho,)h Bowers and Marsoleks approach is +ased
on contemporar* theories of o+/ect reco)nition, ps*cholin),istics, and
concept formation, it has its deeper roots in a proposal +* Ewald
9erin), the 34
th
<cent,r* sensor* ph*siolo)ist, that memor* is 8a
,niversal f,nction of all or)ani1ed matter8 (9erin), 3:'&F3::&, p. >C).
9erin)s ideas, in t,rn, were promoted +* Sam,el B,tler, a,thor of
Erewhon (3:'%) and The Way of ll !lesh (34&C), in a )ro,nd<
+reakin) +ook on "nconscious Memory (3::&) that act,all* predated
E++in)ha,s. "nconscio,s memor*, on 9erin)s and B,tlers view,
ma* +e likened to the 8memor*8 of a paper clip which, when once
+ent, is easier to +end a)ain in the same direction. @aper clips do not
have memor* s*stems, +,t the* do have a ph*sical str,ct,re that
allow them to retain traces of stim,lation. Bowers and Marsolek do not
have m,ch to sa* a+o,t e!plicit memor*, which pres,ma+l* is
mediated +* a dedicated +rain s*stem.
B* contrast with the m,ltiple<s*stems view, other theories hold that
e!plicit and implicit e!pressions of memor* are the prod,cts of a sin)le
memor* s*stem. 7or e!ample, Mandlers activation view ar),es that
primin) in all its forms is mediated +* the a,tomatic activation and
inte)ration, at the time of encodin), of pre<e!istin) knowled)e
str,ct,res correspondin) to the prime0 e!plicit memor*, +* contrast,
re-,ires effortf,l ela+oration to esta+lish new relations amon)
activated str,ct,res (Mandler, 34:&). B,t activation, inte)ration, and
ela+oration all take place within a sin)le memor* s*stem. 2oedi)ers
transfer-appropriate processing view (2oedi)er & Mc.ermott, 344C)
holds that most implicit memor* tasks, s,ch as repetition primin), are
8percept,all* driven8, in that the* re-,ire access onl* to s,rface
feat,res of an o+/ect0 +* contrast, e!plicit memor* tasks are
8concept,all* driven8, in that the* re-,ire access to semantic or
conte!t,al information associated with the st,died item. 5n this view,
dissociations occ,r +eca,se e!plicit memor* depends on 8top<down8 or
8s*m+olic8 processin), while implicit memor* depends on 8+ottom<,p8
or 8data<driven8 processin).
Get a third sin)le<s*stems view invokes ?aco+*s process dissociation
framework to e!plain dissociations +etween e!plicit and implicit
memor* (?aco+*, 3443). 5n this view, e!plicit memor* is lar)el* a
prod,ct of conscio,s, controlled, effortf,l, deli+erate processin), while
implicit memor* is lar)el* a prod,ct of ,nconscio,s, a,tomatic,
effortless, invol,ntar* processin). ?aco+* has also introd,ced a
method, the process dissociation procedure (@.@), which meas,res the
relative contri+,tions of a,tomatic and controlled processin) to an*
task +* pittin) them a)ainst each other in the 8method of opposition8
(M@). A t*pical res,lt of the @.@ is to confirm that the performance of
normal s,+/ects on a memor* task is mediated +* a mi! of controlled
and a,tomatic processes, while the performance of amnesic patients is
lar)el* s,pported +* a,tomatic processes.

-esting the -heories
Each of these views has its stren)ths and weaknesses, not least
+eca,se the* evolved in different research conte!ts. M,ltiple<s*stems
theories are +ased lar)el* on work with ne,rolo)ical patients, while
sin)le<s*stem theories emer)ed mostl* from work on ne,rolo)icall*
intact s,+/ects. The m,ltiple<s*stems views +ask in the reflected )lor*
of co)nitive ne,roscience, +,t are +edeviled +* the temptation to
invoke a new memor* s*stem to e!plain ever* new dissociation
revealed +* research. The activation view )ives a pla,si+le acco,nt of
primin) res,lts, +,t finds it diffic,lt to e!plain how activation co,ld
persist for da*s or months as it is sometimes o+served to do. The
transfer<appropriate processin) view can e!plain dissociations not onl*
+etween e!plicit and implicit memor*, +,t also those that occ,r
+etween two e!plicit or two implicit memor* tasks (one percept,al, the
other concept,al in nat,re), +,t has some diffic,lt* e!plainin)
dissociations +etween semantic primin) and e!plicit memor*, +oth of
which are, in its terms, concept,all* driven. A f,rther -,estion is
whether it is appropriate to term e!plicit memor* as concept,all*
driven in the first place.
The @.@ view, for its part, offers a wa* to reconcile sin)le<s*stem and
m,ltiple<s*stem views, on the ass,mption that a,tomatic and
controlled processes are +ased on separate processin) mod,les that
operate on a sin)le memor* store. At the ver* least, it has provided an
increasin)l* pop,lar techni-,e for meas,rin) the contri+,tions of
a,tomatic and controlled processes to task performance, and offers a
wa* to reconcile sin)le<s*stem and m,ltiple<s*stem views. 9owever,
the mathematics of the @.@ re-,ires the tro,+lin) ass,mption that
these processes are independent of each other. An alternative view,
also consistent with a sin)le<s*stem view of memor*, descri+es
a,tomatic processes as em+edded in, and th,s red,ndant with,
controlled ones. 7or e!ample, Mandler has proposed that the
a,tomatic activation and inte)ration of stored information, which is the
+asis for implicit memor*, precedes the effortf,l ela+oration of these
knowled)e str,ct,res, which is the +asis for e!plicit memor*. 5n s,ch a
s*stem, e!plicit memor* entails implicit memor*, even if the reverse is
not the case.
ne area where the vario,s theories make competin) predictions is
with respect to implicit memor* for novel, ,nfamiliar information.
Activation theories wo,ld seem to s,))est that this is not possi+le,
+eca,se there is +* definition no pre<e!istin) knowled)e
str,ct,re stored in memor* to +e activated, or modified, +* percept,al
inp,t. B* contrast, the other theories are, at least in principle, open to
the ac-,isition of new information. 5n fact, there is considera+le
evidence for primin) of novel nonver+al items s,ch as dot patterns and
novel o+/ects tho,)h not, apparentl*, for line drawin)s of
8impossi+le8 o+/ects that cannot e!ist in three<dimensional space
(m,ch like the drawin)s of the Swiss artist M.=. Escher). Altho,)h
interpretation of these findin)s remains somewhat controversial,
primin) for novel stim,li wo,ld seem to +e inconsistent with
activationFmodification views of implicit memor*. 5nstead, the* appear
to s,pport the m,ltiple<s*stems view that repetition primin), at least,
is the prod,ct of percept,al representation s*stems that encode and
preserve str,ct,ral descriptions of stim,l,s events. @rimin) does not
occ,r for impossi+le o+/ects +eca,se the percept,al representation
s*stem cannot form a str,ct,ral description of o+/ects that cannot
e!ist in three<dimensional space.
The sit,ation with respect to primin) for ver+al materials, s,ch as
words, is more complicated. Earl* res,lts, which showed primin) for
words like candy and number (which have pre<e!istin) representations
in semantic memor*) +,t not for pse,dowords like canber and numdy
(which do not) are consistent with the activation view of implicit
memor*. Bowers fo,nd primin) for words (like kite), nonwords that
followed the r,les of En)lish ortho)raph* (like kers) and for ille)al
nonwords (like #yks), a)ain contradictin) the activation view. 9owever,
as Bowers himself noted, the primin) he o+tained for ille)al nonwords
ma* have +een contaminated +* e!plicit memor*, which softens the
+low somewhat. n the other hand, .orfman fo,nd primin) for
pse,dowords made ,p of familiar morphemes (like genvive) and
familiar s*lla+les (like fasney), +,t not for pse,dos*lla+ic pse,dowords
(like erktofe) made ,p of elements that are neither morphemes nor
s*lla+les in En)lish. These res,lts are consistent with the view that
primin) of novel (and familiar) words res,lts from the activation and
inte)ration of pre<e!istin) s,+le!ical components stored in memor*$
@rimin) cannot not occ,r where there are no s,ch components to +e
activated.
The theoretical de+ate contin,es +ack<and<forth, +,t theoretical
development is hampered +* the fact that e!perimental research on
implicit memor* is narrowl* foc,sed on a sin)le e!perimental
paradi)m namel*, repetition primin). 5t has +een estimated that
some :&H of implicit memor* tests are percept,al in nat,re, involvin)
variants on repetition primin). Viewed in this li)ht, it is not s,rprisin)
to find theorists proposin) that implicit memor* is the prod,ct of a
percept,al representation s*stem, or of percept,all* +ased processin).
B,t if implicit memor* incl,des semantic primin), as well as repetition
primin), s,ch theories are too limited to acco,nt for the entire
phenomenon. 2epetition primin) ma* +e independent of depth of
processin) << tho,)h a more acc,rate statement wo,ld +e that it is
onl* relatively independent0 +,t this is ,nlikel* to +e the case for
semantic primin). 2epetition primin) ma* +e modalit* specific
tho,)h not hyperspecific0 +,t a)ain, this is ,nlikel* to +e the case for
semantic primin). 2esearch on implicit memor* m,st move +e*ond
repetition primin) if we are ever to determine its tr,e nat,re.

Interactions et$een Explicit and Implicit Memory
win) lar)el* to the he)emon* of co)nitive ne,roscience, the most
pop,lar theor* of implicit memor* remains some version of the
m,ltiple<s*stems view. 7or e!ample, Schacter and T,lvin) have
proposed that while repetition primin) is mediated +* a percept,al
representation s*stem, semantic primin) is mediated +* a semantic
memor* s*stem and spared proced,ral learnin) (see 8E!plicit and
5mplicit (earnin)8, +elow) +* a proced,ral memor* s*stem (Schacter &
T,lvin), 344A). Even so, claims for a strict separation of these memor*
s*stems sho,ld not +e made too stron)l*. 5f these vario,s memor*
mod,les were tr,l* independent of each other, we wo,ld e!pect to see
ne,rolo)ical cases where e!plicit memor* is spared and implicit
memor* impaired. The reverse, of co,rse, is what is commonl*
o+served in amnesia. 5n fact, onl* one case has +een reported in which
implicit memor* is impaired and e!plicit memor* intact << and that one
is ,ncertain. The patient, known as M.S., who has an e!tensive
scotoma secondar* to +rain s,r)er*, performed poorl* on a vis,al test
of repetition primin), +,t normall* on test of reco)nition a reversal
of the ,s,al findin) in amnesic patients. 9owever, M.S. also showed
normal performance on a test of concept,al primin), so it can hardl*
+e said that he lacks implicit memor*.
6hatever their ,nderl*in) +asis, the interaction +etween e!plicit and
implicit memor* can also +e o+served in other wa*s. S,+/ects who
conscio,sl* reco)ni1e that the items on a percept,al<identification test
come from a previo,sl* st,died wordlist ma* develop a mental set that
act,all* enhances their primin) performance which is wh*
researchers in this area take care to assess 8test awareness8 in their
s,+/ects, and wh* ?aco+*s 8process dissociation8 proced,re has
+ecome so pop,lar. Amnesic patients are not a+le to take advanta)e of
e!plicit memor*, of co,rse, +,t that does not mean that conscio,s
recollection cannot infl,ence primin) in other circ,mstances.
Moreover, there is considera+le evidence that s,+/ects can take
strate)ic advanta)e of implicit memor* to enhance their performance
on tests of e!plicit memor*. Altho,)h free recall epitomi1es conscio,s
recollection, +oth Mandler and ?aco+* have ar),ed that reco)nition
/,d)ments can +e mediated +* either conscio,s recollection of the test
item, or +* a feelin) of familiarit* that mi)ht +e +ased on primin). 5f
so, then when implicit memor* is spared, s,+/ects can strate)icall*
capitali1e on the primin)<+ased feelin) of familiarit* to enhance their
performance on reco)nition tests. 6e know that, as a r,le, reco)nition
is s,perior to recall in normal s,+/ects and this is also tr,e for
ne,rolo)ical patients with the amnesic s*ndrome, depressed patients
receivin) E=T, and demented patients s,fferin) from Al1heimers
disease, and normal s,+/ects with posth*pnotic amnesia. 5n addition,
st,dies of recollective e!perience indicate that amnesic reco)nition is
t*picall* accompanied +* int,itive feelin)s of familiarit*, rather than
f,ll<fled)ed remem+erin).
Accordin)l*, it seems reasona+le to s,))est that s,ccessf,l reco)nition
in amnesia can +e mediated +* spared implicit memor*. This claim has
+een vi)oro,sl* de+ated +* S-,ire and his collea),es, who insist that
primin) is inaccessi+le to conscio,s awareness, and so cannot serve as
a +asis for reco)nition. .espite methodolo)ical iss,es c,ttin) this wa*
and that, st,dies emplo*in) the process<dissociation proced,re clearl*
indicate that amnesic reco)nition is mediated +* a primin)<+ased
feelin) of familiarit* as theor* s,))ests the* mi)ht +e, and as the
s,+/ects themselves sa* the* are. 5t ma* +e that recollection and
familiarit* are )overned +* separate memor* s*stems0 +,t a)ainst a
f,rther proliferation of memor* s*stems, it ma* +e more parsimonio,s
to concl,de that e!plicit and implicit memor* interact after all.

Explicit and Implicit Learning
Traditionall*, learnin) has +een defined as a relativel* permanent
chan)e in +ehavior that occ,rs as a res,lt of e!perience. Earl*
investi)ators << @avlov and Thorndike, 6atson and Skinner <<
constr,ed learnin) as conditionin) << the formation of associations
+etween environmental stim,li and an or)anisms responses to them
(Bower & 9il)ard, 34:30 Schwart1 & 2eis+er), 3443). 9owever, the
co)nitive revol,tion in ps*cholo)* has led to a reconstr,al of learnin)
as a relativel* permanent chan)e in knowledge that occ,rs as a res,lt
of e!perience declarative and proced,ral knowled)e that the
or)anism will s,+se-,entl* ,se for its own p,rposes in predictin) and
controllin) environmental events. Th,s, in classical conditionin) the
or)anism forms e!pectations concernin) the likel* conse-,ences of
events, and in instr,mental conditionin) the or)anism forms
e!pectations concernin) the likel* conse-,ences of its own +ehaviors.
9ow this knowled)e translates into +ehavior is another matter.
5n addition to classical and instr,mental conditionin), researchers have
st,died perceptual learning, involvin) lon)<lastin) chan)es in
perception or percept,al<motor coordination, as in the case of prism
adaptation to inverted or distorted ima)es0 and conceptual learning,
+* which individ,als ind,ce a+stract concepts from enco,nters with
specific instances not to mention lan),a)e learnin), and especiall*
the learnin) of a second lan),a)e. Altho,)h these forms of learnin)
seem to +e mediated +* direct e!perience, Band,ra has descri+ed
social learning, also known as vicario,s or o+servational learnin), in
which the individ,al )ains knowled)e +* o+servin) other people. Social
learnin) comes in two +road forms$ +* e!ample, thro,)h imitation and
modelin), and +* precept, thro,)h sponsored teachin).
At least in the case of h,mans (and certainl* other primates, pro+a+l*
other mammals, perhaps other verte+rates, and ma*+e some
inverte+rates) this co)nitive emphasis on individ,als ac-,irin)
knowled)e to help them predict and control events in the world implied
that learnin) was a conscio,s activit*. This is also tr,e of percept,al
learnin), which occ,rs more rapidl* with active than passive
movements of the o+server. And its emphasis on the role of o+servin),
modelin), and teachin) not to mention the fact that civili1ation has
created instit,tions to s,pport these activities << marks social learnin),
too, as a conscio,s act of mind. Ievertheless, it is also tr,e that some
or)anisms, like the sea moll,sk aplysia, can learn even tho,)h the*
pro+a+l* do not have eno,)h ne,rons, m,ch less a cere+ral corte!, to
s,pport conscio,sness. Even in h,mans, who have a capacit* for
conscio,sness, it has lon) +een evident that some learnin) can take
place ,nconscio,sl* (Adams, 34;'0 2a1ran, 34>3).

Implicit Learning
5n point of fact, the concept of implicit learning was introd,ced into the
ps*cholo)ical literat,re well +efore that of implicit memor*. 5n a
pioneerin) series of e!periments p,+lished in 34>', 2e+er asked
s,+/ects to memori1e lists of letter strin)s, each of which had +een
)enerated +* a Markov<process artificial )rammar a set of r,les that
specified what letters co,ld appear in the strin), and in what order.
ver trials, the s,+/ects fo,nd it easier to memori1e )rammatical
strin)s, compared to random strin)s, indicatin) that their learnin) was
e!ploitin) the )rammatical str,ct,re. Moreover, when presented with
new strin)s, s,+/ects were a+le to distin),ish +etween )rammatical
and non)rammatical strin)s at levels si)nificantl* +etter than chance,
indicatin) that the* had ac-,ired some knowled)e of the )rammar. Get
when -,eried, the s,+/ects were ,na+le to specif* the )rammatical
r,le itself. The* had learned the )rammar, and this knowled)e had
),ided their +ehavior, +,t the* were not aware that the* had learned
an*thin), and the* were not aware of what the* had learned (2e+er,
344C).
At ro,)hl* the same time, ne,rops*cholo)ists noticed that, over trials,
amnesic patients improved their performance on s,ch tasks as ma1e
learnin), p,rs,it<rotor learnin), and mirror<reversed learnin). =learl*,
then, amnesic patients had the capacit* to ac-,ire new skills, +,t the*
did not reco)ni1e the tasks, nor did the* remem+er the learnin)
e!periences0 moreover, the* seemed to have no conscio,s awareness
of their newl* ac-,ired knowled)e. (ater st,dies showed that amnesic
patients co,ld learn artificial )rammars, /,st as ne,rolo)icall* intact
individ,als do.
B* analo)* with memor*, we can define e!plicit learnin) as a relativel*
permanent chan)e in knowled)e or +ehavior that is accompanied +*
conscio,s awareness of what has +een learned. 5mplicit learnin), then,
refers to a relativel* permanent chan)e in knowled)e or +ehavior in
the a+sence of conscio,s awareness of what has +een learned.
Sometimes evidence for implicit learnin) is taken as evidence for
implicit memor*, +,t implicit memor* is more narrowl* restricted to
the learnin) episode itself, while implicit learnin) covers the knowled)e
ac-,ired in that episode. 5n a famo,s case p,+lished in 3433,
=laparede descri+ed an amnesic patient who for)ot an episode in
which he pricked her hand with a pin while )reetin) her, +,t who was
conscio,sl* aware that 8Sometimes people hide pins in their hands8
(#ihlstrom, 344;). This patient was conscio,s of what she had learned,
+,t displa*ed source amnesia (also known as cr*ptomnesia, or
,nconscio,s pla)iarism), a concept more closel* related to implicit
memor*. 5mplicit learnin) )oes +e*ond the formation of simple
associations, as in classical or instr,mental conditionin), and involves
the ac-,isition of knowled)e of some comple!it*, at some level of
a+straction.

/arieties of Implicit Learning
5mplicit learnin) has +een st,died in a wide variet* of e!perimental
paradi)ms, in addition to artificial )rammars and motor learnin) (for
comprehensive reviews, see Berr* & .ienes, 344C0 7rensch & 2,n)er,
%&&C0 Se)er, 344A0 Stadler & 7rensch, 344:)$
Concepts$ 5n a paradi)m somewhat similar to artificial )rammar
learnin), s,+/ects learn to identif* instances of novel concepts, s,ch as
patterns of dots that var* aro,nd a protot*pe, witho,t +ein) a+le to
descri+e the definin) or characteristic feat,res of the concepts
themselves.
Covariation "etection$ S,+/ects learn the association +etween two
feat,res, s,ch as hair len)th and personalit*, +,t cannot identif* the
+asis for their predictions.
(e0&ence Learning$ S,+/ects learn the se-,ence in which certain
stim,li will occ,r for e!ample, the appearance of a tar)et in a
partic,lar location on a comp,ter screen witho,t +ein) a+le to
specif* the se-,ence itself.
"ynamic (ystems$ S,+/ects learn to control the o,tp,t of a comple!
s*stem +* manip,latin) an inp,t varia+le, witho,t +ein) a+le to
specif* the relationship +etween the two.
5n each of these cases, s,+/ects demonstrate, +* performance
meas,res s,ch as acc,rac* or response latenc* in /,d)ment, or
prediction or control of +ehavior, that the* have ac-,ired knowled)e
from e!perience0 *et the* are ,na+le to provide an acc,rate acco,nt
of the methods +* which the* achieve these res,lts. The* have learned
somethin) new, +,t the* do not know what the* know.

1hat is Learned in Implicit Learning2
+servations of preserved learnin) capacit* in amnesic patients led
=ohen and S-,ire to propose that amnesia impaired declarative
memor* (8knowin) that$), +,t spared proced,ral memor* (8knowin)
how8 =ohen & S-,ire, 34:&). @ercept,al and motor skills, s,ch as
covariation detection, se-,ence learnin), and motor learnin), can
certainl* +e represented as s*stems of if-then prod,ctions, +,t it is not
at all clear that all forms of implicit learnin) are proced,ral in nat,re.
7or e!ample, there is evidence that amnesic patients can ac-,ire new
concept,al knowled)e in amnesia knowled)e that, ordinaril*, wo,ld
+e represented in the propositional format characteristic of declarative
memor*.
5n this respect, artificial )rammar learnin) is an interestin) case. The
str,ct,re of a Markov<process finite<state )rammar lends itself easil*
to translation into a prod,ction s*stem$ %f the first letter is a @, then
the ne!t letter m,st +e a T or a V0 %f the first letter is a T, then the
ne!t letter m,st +e an S or an J0 and so on. n the other hand,
s,+/ects memori1in) )rammatical strin)s mi)ht simpl* a+stract what a
8protot*pical8 )rammatical strin) looks like. 6hen makin)
)rammaticalit* /,d)ments, s,+/ects co,ld then compare test items to
this stored protot*pe or, perhaps, to the specific instances stored in
memor* d,rin) the memori1ation phase of the e!periment. 5n either
case, the ,nconscio,s knowled)e ac-,ired thro,)h implicit learnin)
wo,ld more closel* resem+le declarative than proced,ral knowled)e
*et another reason not to ,se 8declarative8 instead of 8e!plicit8 to la+el
conscio,s memor*.
5n principle, +oth protot*pes and r,le s*stems are a+stract knowled)e
representations that )o +e*ond the specific instances enco,ntered in
the st,d* set. Accordin)l*, an important -,estion concerns the de)ree
to which implicit learnin) is )enerali1a+le +e*ond the specific. That is,
can s,+/ects appl* a )rammar learned from strin)s of @s, Vs, and Ts to
test strin)s composed of (s, Bs, and GsD 2e+er reported that this was
the case, tho,)h s,+se-,ent research has often fo,nd that transfer is
s,+stantiall* de)raded. St,dies of transfer in other domains have also
*ielded mi!ed res,lts. f co,rse, the de)ree of transfer will depend on
the de)ree of initial learnin). 5n the artificial )rammar e!periments,
classification performance t*picall* ran)es +etween >&<:&H correct,
with the more fre-,ent o,tcomes at the lower end of this ran)e.
Viewed a)ainst a +ase rate of ;&H correct, a performance at >;H
ma* +e statisticall* si)nificant, +,t ma* not leave a lot of room to
show incomplete transfer.

Is Implicit Learning 'eally Unconscio&s2
5mplicit learnin) is distinct from mere incidental learning, where
knowled)e is ac-,ired in the a+sence of instr,ctions or intention to
learn, +,t the person is conscio,s of what he or she has learned
(E*senck, 34:%). The critical feat,re of implicit learnin) is that it is
,nconscio,s, in the sense that the s,+/ects are ,naware of what the*
have learned. .oc,mentin) dissociations +etween e!plicit and implicit
learnin), then, is a somewhat trick* +,siness. Man* st,dies do not
)ive a )reat deal of detail a+o,t the methods +* which s,+/ects
conscio,s knowled)e was assessed, +,t it is pro+a+l* not eno,)h
merel* to ask s,+/ects in the artificial<)rammar e!periments to
descri+e the r,le that )overns the letter strin)s, and co,nt them as
8,nconscio,s8 when the* fail to do so. 5n the first place, ,nless the
test stim,li are ver* caref,ll* constr,cted, even partial awareness of
the r,le that the first letter m,st +e either a P or a T, for e!ample,
ma* +e eno,)h to permit s,+/ects to discriminate +etween
)rammatical and ,n)rammatical strin)s at +etter than chance levels.
Altho,)h investi)ators of e!plicit and implicit memor* have developed
ri)oro,s standards for matchin) e!plicit and implicit tasks, similar
standards are )enerall* lackin) in st,dies of implicit learnin).
The ar),ment that implicit learnin) is reall* ,nconscio,s is sometimes
+olstered +* the fact that amnesic patients show preserved implicit
learnin). f co,rse, amnesics also for)et the learnin) episode as well,
conf,sin) implicit learnin) with implicit memor*. 5n this re)ard, it is
somewhat disconcertin) to note that s,+/ects can show si)nificant
implicit 8learnin)8 even in the a+sence of an* learnin) e!perienceK
That is to sa*, in some e!perimental proced,res involvin) classification
performance, it is possi+le for s,+/ects to int,it the str,ct,re of the
tar)et cate)or* from test instances, even when the* were denied an
opport,nit* to learn the cate)or* d,rin) a prior st,d* phase.
9owever, implicit learnin) is not alwa*s, or entirel*, spared in amnesic
patients. 7or e!ample, amnesics show normal levels of percept,al
learnin) in a vis,al search task, +,t impaired learnin) when conte!t,al
c,es are added to the proced,re. Even intact implicit learnin) +*
amnesic patients does not mean that e!plicit and implicit learnin)
m,st +e mediated +* different +rain s*stems. .issociations +etween
reco)nition and concept learnin) can +e sim,lated in a comp,tational
model of e!emplar memor* that has onl* one s*stem for storin)
memor*, with different thresholds for reco)nition and classification.
5t is sometimes claimed that implicit learnin), precisel* +eca,se it is
a,tomatic and ,nconscio,s, is a ver* powerf,l (as well as more
primitive) form of learnin) more powerf,l than conscio,s forms of
learnin) that emer)ed more recentl* in evol,tionar* histor* (2e+er,
344C). 6hile it does seem ama1in) that s,+/ects can pick ,p
knowled)e of somethin) as comple! as an artificial )rammar or a
d*namic s*stem a,tomaticall*, and appl* it ,nconscio,sl*, claims for
the s,periorit* of ,nconscio,s processin) sometimes seem to reflect a
2omantic notion of the ,nconscio,s that )oes +ack to von 9artmann
(9artmann, 3:>:F34C3, p. A&), who wrote that the ,nconscio,s 8can
reall* o,tdo all the performances of conscio,s reason8. "nfort,natel*,
enth,siasts of implicit learnin) have not alwa*s compared implicit
learnin) to conscio,s, deli+erate knowled)e ac-,isition. 9ow well
wo,ld s,+/ects perform if we act,all* showed them the finite<state
)rammar, or if we )ave them feed+ack a+o,t their classification
performanceD 6hat if we simpl* told s,+/ects the se-,ence of
-,adrants in which the tar)et wo,ld appearD

-he Implicit and the Unconscio&s
To)ether with the concept of a,tomaticit*, research on implicit learnin)
and memor* constit,ted ps*cholo)*s first steps toward a revival of
interest in ,nconscio,s mental life (#ihlstrom, 34:'). Altho,)h the
ps*cholo)ical ,nconscio,s s,ffered m,ch in the %&
th
cent,r* from taint
+* 7re,dian ps*choanal*sis the concepts and methods emplo*ed to
st,d* implicit learnin) and memor* have now +een e!tended to other
domains, s,ch as perception and even thinkin) and +e*ond
co)nition to emotion and motivation (#ihlstrom, 3444). 5n this wa*,
the st,d* of implicit learnin) and memor* offer a new, non<7re,dian
perspective on ,nconscio,s mental life and, in t,rn, on
conscio,sness itself.

'eferences
Adams, ?. #. (34;'). (a+orator* st,dies of +ehavior witho,t
awareness. Psychological &ulletin' (), C:C<A&;.
Anderson, ?. 2. (%&&&). *earning and memory+ n integrated approach
,-nd ed./. Iew Gork, IG, "S$ ?ohn 6ile* & Sons, 5nc.
Berr*, .. =., & .ienes, E. (344C). %mplicit learning+ Theoretical and
empirical issues. 9ove, ".#.$ Erl+a,m.
Bower, B. 9., & 9il)ard, E. 2. (34:3). Theories of learning (;th ed.).
En)lewood =liffs, I.?.$ @rentice<9all.
Bowers, ?. S., & Marsolek, =. ?. (Eds.). (%&&C). 0ethinking implicit
memory. !ford$ !ford "niversit* @ress.
B,tler, (. T., & Berr*, .. =. (%&&3). 5mplicit memor*$ 5ntention and
awareness revisited. Trends in Cognitive 1ciences' (, 34%<34'.
=ohen, I. ?., & S-,ire, (. 2. (34:&). @reserved learnin) and retention
of pattern anal*1in) skill in amnesia$ .issociation of knowin) how and
knowin) that. 1cience' -23, %&'<%3&.
=rovit1, 9. 7. (34'&). 4alton5s walk6 methods for the analysis of
thinking' intelligence and creativity. Iew Gork$ 9arper & 2ow.
E*senck, M. 6. (34:%). 5ncidental learnin) and orientin) tasks. 5n =.
2. @,ff (Ed.), 7andbook of research methods in human memory and
cognition (pp. 34'<%C:). Iew Gork$ Academic.
7lana)an, . (344%). Conscousness reconsidered. =am+rid)e, Ma.$
M5T @ress.
7rensch, @. A., & 2,n)er, .. (%&&C). 5mplicit (earnin). Current
Directions in Psychological 1cience' 2-(3), 3C<3'.
9artmann, E. v. (3:>:F34C3). Philosophy of the unconscious+
1peculative results according to the inductive method of physical
science. (ondon$ 2o,tled)e and #e)an @a,l.
9erin), E. (3:'&F3::&). n memor*. 5n S. B,tler (Ed.), "nconscious
memory (Iew Ed., 343& ed., pp. >C<:>). (ondon$ 7ifield.
?aco+*, (. (. (3443). A process dissociaton framework$ Separatin)
a,tomatic from intentional ,ses of memor*. 8ournal of Memory 9
*anguage, in press.
#ihlstrom, ?. 7. (34:'). The co)nitive ,nconscio,s. 1cience'
-:;(A:%3), 3AA;<3A;%.
#ihlstrom, ?. 7. (344;). Memor* and conscio,sness$ An appreciation of
=laparede and 0ecognition et Moiiete. Consciousness 9 Cognition+ n
%nternational 8ournal' )(A), C'4<C:>.
#ihlstrom, ?. 7. (344>). Memor* research$ The conver)ence of theor*
and practice. 5n .. 9ermann & =. McEvo* & =. 9ert1o) & @. 9ertel & M.
#. ?ohnson (Eds.), &asic and applied memory research+ Theory in
conte#t (Vol. 3, pp. ;<%;). Mahwah, I.?.$ Erl+a,m.
#ihlstrom, ?. 7. (3444). =onscio,s and ,nconscio,s co)nition. 5n 2. ?.
Stern+er) (Ed.), The nature of cognition (pp. 3'C<%&A). =am+rid)e,
Ma.$ M5T @ress.
Mandler, B. (34:&). 2eco)ni1in)$ The /,d)ment of previo,s
occ,rrence. Psychological 0eview' <;(C), %;%<%'3.
2a1ran, B. (34>3). The o+serva+le ,nconscio,s and the infera+le
conscio,s in c,rrent Soviet ps*choph*siolo)*$ 5nteroceptive
conditionin), semantic conditionin), and the orientin) refle!.
Psychological 0eview' =<, :3<3A'.
2e+er, A. S. (344C). %mplicit learning and tacit knowledge+ n essay
on the cognitive unconscious. !ford, ".#.$ !ford "niversit* @ress.
2ichardson<#lavehn, A., Bardiner, ?. M., & ?ava, 2. 5. (344>). Memor*$
Task dissociations, process dissociations, and dissociations of
conscio,sness. 5n B. "nderwood (Ed.), %mplicit cognition (pp. :;<3;:).
!ford$ !ford "niversit* @ress.
2oedi)er, 9. (., & Mc.ermott, #. B. (344C). 5mplicit memor* in normal
h,man s,+/ects. 5n 7. Boller & ?. Brafman (Eds.), 7andbook of
>europsychology (pp. >C<3C3). Amsterdam$ Elsevier Science
@,+lishers.
Salaman, E. (34'&). collection of moments+ study of involuntary
memories. (ondon$ (on)man.
Schacter, .. (. (34:'). 5mplicit memor*$ 9istor* and c,rrent stat,s.
8ournal of E#perimental Psychology+ *earning' Memory' and Cognition'
2:, ;&3<;3:.
Schacter, .. (., & T,lvin), E. (Eds.). (344A). Memory systems 2??).
=am+rid)e, MA, "SA$ The Mit @ress.
Schwart1, B., & 2eis+er), .. (3443). *earning and memory (3st ed.).
Iew Gork$ Iorton.
Se)er, =. A. (344A). 5mplicit learnin). Psychological &ulletin' 22(, 3>C<
34>.
S-,ire, (. 2., #nowlton, B., & M,sen, B. (344C). The str,ct,re and
or)ani1ation of memor*. nnual 0eview of Psychology' )), A;C<A4;.
S-,ire, (. 2., & Eola<Mor)an, S. (3443). The medial temporal lo+e
memor* s*stem. 1cience' -(:, 3C:&<3C:>.
Stadler, M. A., & 7rensch, @. A. (Eds.). (344:). 7andbook of implicit
learning. Tho,sand aks, =a.$ SABE.
T,lvin), E., & =raik, 7. 5. M. (Eds.). (%&&&). The @#ford handbook of
memory. Iew Gork, IG, "S$ !ford "niversit* @ress.
T,lvin), E., & Schacter, .. (. (344&). @rimin) and h,man memor*
s*stems. 1cience' -);(A4A&), C&3<C&>.

This pa)e last modified .