192 Opinion TRENDS in Cognitive Sciences Vol.5 No.
5 May 2001
Implicit memory:
intention and
awareness revisited
Laurie T. Butler and Dianne C. Berry
One of the major developments in memory research in the past decade or so
has been a growing interest in implicit memory – task performance that is not
accompanied by conscious or intentional recollection. In this article, we
examine evidence for perceptual and conceptual implicit memory, using the
accepted definitions, and suggest that there is in fact a lack of strong evidence
for implicit memory in normal subjects. If more convincing evidence cannot
be obtained, one solution might be to modify the current terminology. We
suggest that the term implicit memory might have outgrown its usefulness as
an overall descriptor.
Over the past 15 years there has been a phenomenal
growth in the number of studies investigating implicit
memory. This interest can be traced to Schacter1
(see also Ref. 2) who in a seminal paper defined the
phenomenon as being ‘revealed when previous
experiences facilitate performance on a task that does
not require conscious or intentional recollection of
those experiences’ (Ref. 1, p. 501). Accordingly, to be
described as a demonstration of implicit memory,
two conditions should be satisfied: (1) performance
reflects an unintentional retrieval strategy, and
(2) performance is not accompanied by conscious
recollection. Ensuring that these conditions are met
has proved to be difficult, especially in studies using
normal participants. After outlining this problem in
greater detail, we review studies of implicit memory,
specifically repetition priming, that have met one or
both of these conditions. We consider evidence
pertaining to perceptual and well as conceptual
implicit memory tasks, and seek to address three
main questions: (1) are there convincing examples
of implicit memory performance in normal
participants; (2) should we be concerned that many
studies do not meet both conditions for demonstrating
implicit memory; and (3) how useful is the term
implicit memory as an overall descriptor?
Methods for measuring implicit memory
Numerous procedures exist for studying implicit
memory (e.g. word stem completion, perceptual
Laurie T. Butler* identification). Although many would be classified as
Dianne C. Berry
Dept of Psychology,
testing perceptual implicit memory there is a
University of Reading, growing number of conceptual implicit memory tasks
Earley Gate, (e.g. category exemplar production, word-cued
Whiteknights, Reading,
association). This perceptual/conceptual distinction is
UK RG6 6AL.
*e-mail: l.t.butler@ made on the basis of the task’s processing demands
reading.ac.uk and is an important means of classification. [It should
http://tics.trends.com 1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01636-3
be noted that some researchers have made the
classification on the basis of the obtained data
(e.g. Ref. 3), such that for a task to be described as
perceptual, higher data driven scores should be
shown, and vice versa for conceptual tasks.]
Briefly, a test engages perceptual processes when
performance involves the analysis of stimulus form
(e.g. identifying words), and conceptual processes
when performance involves the analysis of stimulus
meaning (e.g. producing category exemplars). Both
types of task can be contrasted with explicit memory
tasks (e.g. recall, recognition) in that no instruction is
given to think back intentionally to an earlier study
list. Rather participants carry out a seemingly
unrelated task that can be completed without
reference to the study episode. Thus, following
exposure to a study list (e.g. including the word
elephant), participants might be asked to respond
with the first word that comes to mind to a series of
word stems (e.g. ele-); or produce examples for various
categories (e.g. animals). In each case, priming is
observed; that is a greater proportion of studied words
is produced, relative to non-studied ones.
The contamination problem
In recent years the issue of contamination by explicit
memory processes has become increasingly
prominent in discussions of implicit memory4,5.
There are two distinct ways in which an implicit
memory task can be contaminated: firstly,
participants might adopt an intentional retrieval
strategy (i.e. deliberately retrieve study words).
Secondly, participants might be phenomenally aware
(i.e. possess knowledge that retrieved test words are
study list ones). [Schacter et al.6 provide a set of
scenarios to illustrate the subtle ways in which
intention and awareness might interact and
influence implicit task performance.]
Although many studies have used patient groups
where recourse to intentional retrieval strategies,
and presumably conscious recollection, is reduced or
absent, there is a real concern that the performance of
normal participants on implicit memory tasks might
be contaminated in one or both of these ways. Thus,
performance could reflect an intentional retrieval
strategy (normally accompanied by awareness), or
alternatively an unintentional retrieval strategy
accompanied by awareness. Distinguishing between
these and genuine cases of implicit memory
(i.e. unintentional and unconscious) presents a
considerable challenge.
This article is primarily concerned with
methodological issues. However, some orientating
background information on relevant theoretical
explanations of implicit memory is provided in Box 1.
Approaches to contamination
Researchers have developed a number of approaches
to address the problem of contamination (see Box 2).
In the following sections we review evidence for
 Opinion TRENDS in Cognitive Sciences Vol.5 No.5 May 2001 193

Box 1. Theoretical explanations of implicit memory

Although a number of theories have been
proposed to explain implicit memory
phenomenaa,b, two have been especially
prominent in explaining dissociations
between implicit and explicit memory tasks:
Transfer-appropriate processing
This account emphasizes the importance
of the match between processes at
encoding and retrievalc. Furthermore, the
framework distinguishes between two
classes of cognitive processes: perceptual
processes (i.e. analysis of stimulus form)
and conceptual processes (i.e. analysis of
stimulus meaning). Dissociations are
predicted between implicit and explicit
memory tasks to the extent that they
typically engage perceptual and
conceptual processes, respectively.
Although successful in accounting for
experimental data, for normal subjects at
least, recent evidence of dissociations
between conceptual implicit and explicit
memory tasks, and within the class of
conceptual implicit memory tasks, have
proven problematic for the framework.
Such results suggest that finer distinctions
between different forms of conceptual
processes are necessary. Recent
proposals include a distinction between
competitive and non-competitive
conceptual processesd and one between
relational and item specific processinge.
Perceptual representation system
This account proposes that priming
reflects the operation of a perceptual
representation system that functions
independently of the episodic memory
system believed to support explicit
memory performancef. The perceptual
representation system comprises three
domain-specific subsystems (visual word-
form system, auditory word-form system,
structural description system) dedicated
to the processing and representation of
information concerning the form and
structure of words and objects.
The approach has been particularly
useful for integrating findings from both
normal and memory impaired patients.
However, the systems that exist to date are
vague in terms of how conceptual priming
is supported (see Ref. g). Additionally, it is
not clear at what stage observed patterns
of findings require the specification of a
new memory system.
References
a Ratcliff, R. and McKoon, G. (1997) A counter
model for implicit priming in perceptual word
identification. Psychol. Rev. 104, 319–343
b Postle, B.R. and Corkin, S. (1999) Manipulation
of familiarity reveals a necessary lexical
component of the word-stem completion
priming effect. Mem. Cognit. 27, 12–25
c Roediger, H.L. et al. (1989) Explaining
dissociations between implicit and explicit
measures of retention: a processing account.
Varieties of Memory and Consciousness: Essays
in Honour of Endel Tulving (Roediger, H.L. and
Craik, F.I.M, eds), pp. 3–41, Erlbaum
d Vaidya, C.J. et al. (1997) Evidence for multiple
mechanisms of conceptual priming on implicit
memory tests. J. Exp. Psychol. Learn. Mem.
Cognit. 23, 1324–1343
e Mulligan, N.W. et al. (1999) The effects of levels
of processing and organization on conceptual
implicit memory in the category exemplar
production test. Mem. Cognit. 27, 633–647
f Tulving, E. and Schacter, D.L. (1990) Priming and
human memory systems. Science 247, 301–306
g Schacter, D.L. (1994) Priming and multiple
memory systems: perceptual mechanisms of
implicit memory. In Memory Systems (Schacter,
D.L. and Tulving, E., eds); pp. 233–268, MIT Press

implicit memory from studies which have used these
methods. At the outset we note that the ‘process
dissociation procedure’ does not allow for separate
estimates of intention and awareness (see Ref. 4 for
discussion). However, we cite evidence obtained using
this procedure because of its centrality in discussions
on contamination. Initially, we describe evidence
indicating that performance reflects unintentional
retrieval and then we examine evidence that
performance is both unintentional and unconscious.
A separate section on awareness is not included as it is
assumed that test unawareness is incompatible with
an intentional retrieval strategy. Whilst intentional
unconscious memory is definitionally possible there is
no evidence regarding its actual occurrence4.
Evidence for implicit memory
Intentionality condition
Perceptual implicit memory. Numerous studies have
shown performance on perceptual implicit memory
tasks to reflect unintentional retrieval. These have
usually taken the form of single dissociations between
perceptual implicit and explicit memory tasks
differing only in terms of test instructions. For
example, Mulligan and Hartman found that dividing
attention at study had no effect on performance on a
word fragment task, but impaired performance on a
matched word fragment cued recall task7. Other
well-replicated effects, in the form of single
dissociations, have been observed across a range of
different variables, for example, levels of processing8
and retinal size9. Java provides even stronger
evidence in the form of a crossed double dissociation
on implicit and explicit tasks matched on everything
bar test instructions10. In each of these cases we can
be confident that performance on the implicit
version of the task predominantly reflected an
unintentional retrieval strategy. Recently though,
Richardson-Klavehn et al. reported evidence showing
that performance on a word stem completion task was
entirely the result of an unintentional retrieval
strategy11. This conclusion was supported by a complex
pattern of results conjointly showing functional, double
and parallel (i.e. same effect of a variable on both
implicit and explicit versions of the task) dissociations.
Conceptual implicit memory. There is a relative paucity
of evidence regarding dissociations between
conceptual implicit and explicit memory tasks, and
what evidence there is tends to be based on the findings
of single studies. This might be because the systematic
study of conceptual tasks is relatively recent. However,
it could indicate that explicit memory tasks share more
similar processes in common with conceptual implicit
than with perceptual implicit memory tasks (see, for
example, Ref. 12). In terms of relevant evidence Weldon
and Coyote showed that explicit and conceptual implicit

http://tics.trends.com
194 Opinion TRENDS in Cognitive Sciences Vol.5 No.5 May 2001

Box 2. Approaches to contamination

The following are some of the proposals that have attempted to
address the problem of contamination of implicit memory tasks
by explicit (intentional or conscious) processes.
Retrieval intentionality criteriona
• Proposed to distinguish between unintentional and intentional
retrieval strategies.
• Requires that implicit and explicit memory tasks be matched on
all properties, bar test instructions. Consequently, differential
effects of experimental manipulations can be attributed to the
use of unintentional versus intentional retrieval strategies.
• Although now a feature of most experimental designs, it
does not address the more intractable issue of
phenomenological awareness.
Chance recognitionb
• Requires that the effects of a variable be shown to be greater on
an implicit rather than explicit memory task. If this is the case it
can be argued that performance on the implicit memory task
reflects the use of unconscious information as it would be used
on the explicit memory task if it were consciously available.
• Few studies have met this conservative criterion.
Awareness questionnairesc
• Consists of a sequence of questions designed to assess state of
awareness during testing.
• Easy to administer, but permits only a global assessment of
awareness. That is, it cannot distinguish between cases where
a few test responses are associated with awareness and cases
where most responses are.
Item by item analysisd,e
• Allows more fine grained analysis of individual responses.
Participants receive an extra recognition task in which they have
to indicate the status (previously studied or not) of each item.
From this, priming effects can be calculated based on items that
participants do and do not show awareness for.
• As with questionnaires, can be used with different implicit
memory tasks. However, in assessing awareness
retrospectively, all such methods are subject to criticisms such
as inability to recall the true nature of the mental state at test.
On-line awareness analysisf
• Awareness is measured ‘on line’, that is, at the time of testing.
• Method comprises standard stem-completion and cued-recall
tasks, plus an opposition task where participants are
encouraged to complete stems with the first word that comes
to mind but to omit previously studied words. Latter task
provides the crucial awareness data because participants are
required to suppress items associated with awareness.
• Although promising, its ability to generalize to other implicit
memory tasks has yet to be established.
Process dissociation procedureg
• Eschewing standard task-based methodology, the approach
endorses the idea that any task (implicit or explicit) can be
sensitive to both conscious and unconscious influences.
• The method comprises two test conditions (inclusion and
exclusion), which place conscious and unconscious processes
in concert and in opposition, respectively. Simple equations are
then used to derive separate estimates of conscious and
unconscious memory.
• Although an important development, intense methodological
and theoretical scrutiny renders its current status unclear.
References
a Schacter, D.L. et al. (1989) Intention, awareness, and implicit memory: the
retrieval intentionality criterion. In Implicit Memory: Theoretical Issues
(Lewandowsky, S. et al., eds), pp. 47–65, Erlbaum
b Reingold, E.M. and Merikle, P.M. (1990) On the inter-relatedness of theory
and measurement in the study of unconscious processes. Mind Lang. 5, 9–28
c Bowers, J. and Schacter, D.L. (1990) Implicit memory and test awareness.
J. Exp. Psychol. Learn. Mem. Cognit. 16, 404–416
d Java, R.I. (1994) States of awareness following word stem completion.
Eur. J. Cognit. Psychol. 6, 77–92
e Wippich, W. (1997) Input and output monitoring in implicit and explicit
memory tests. Sprache Kognition 3-4, 192–197
f Richardson-Klavehn, A. et al. (1994) Involuntary conscious memory and
the method of opposition. Memory 2, 1–29
g Jacoby, L.L. (1991) A process dissociation framework: separating automatic
from intentional uses of memory. J. Mem. Lang. 30, 513–541

tasks differing only in terms of test instructions were
dissociated by a picture-word manipulation13. Similar
effects have been obtained between matched tasks
following read/generate14, and perceptual interference
manipulations15. However, some manipulations have
not produced dissociations where test instructions have
been matched (e.g. Refs 7,14,16).
Intentionality and awareness conditions
Perceptual implicit memory. The strongest form of
evidence comes from studies where priming on the
implicit task is shown in the absence of explicit
memory performance, the logic being that if relevant
information was consciously available it would be
used on the latter test. A few studies do meet this
conservative criterion (e.g. Refs 17,18) including one
that obtained significant priming, but chance
recognition, on tasks matched on all characteristics
apart from test instructions18. Some evidence is also
provided by studies using the process dissociation
procedure (Box 2). Schmitter-Edgecombe showed, for
example, that although consciously controlled
estimates of memory were impaired by manipulating
divided attention, unconscious estimates remained
invariant19. Other studies have applied the approach
to a number of different variables including depth of
processing20 and modality21. Using an on-line
awareness method (see Box 2), Richardson-Klavehn
et al. demonstrated that priming following a
perceptual encoding task reflected unintentional
unconscious memory22. Finally, awareness has been
assessed retrospectively using item by item or
questionnaire measures. In these cases there is some

http://tics.trends.com
Opinion TRENDS in Cognitive Sciences Vol.5 No.5 May 2001
evidence that priming can be observed even when
participants claim to have no awareness of the
study–test relationship10,19,23.
Conceptual implicit memory. There is virtually no
evidence that priming on conceptual tasks reflects
unintentional and unconscious memory. To the best of
our knowledge only one study has obtained reliable
estimates of unconscious retrieval processes for a
conceptual task (category association) using the
process dissociation procedure19. These were found
under both full and divided attention conditions.
Other than this, two studies have reported significant
priming under conditions whereby participants
report no awareness16,19.
Relaxing the implicit memory definition
In the preceding sections we have shown that
considerably more studies using perceptual, rather
than conceptual, implicit memory tasks have met
the intentionality criterion. More importantly, of the
small fraction of studies that have met both
conditions, as defined by Schacter1, almost all have
used perceptual implicit memory tasks.
As the reader will probably agree, the lack of strong
evidence for implicit memory in normal participants is
surprising. Given the difficulty of meeting the second
criterion this apparent lack of methodological rigour
is to some degree understandable. Nevertheless, it is
certainly the case that the majority of studies make no
attempt to assess ‘conscious recollection’ when
purporting to demonstrate implicit memory.
Furthermore, some studies make no reference to this
second condition at all, particularly, it seems to us,
those measuring conceptual implicit memory14,24.
Thus although the term implicit memory still enjoys
widespread usage, we ask whether it is acceptable
that the originally proposed conditions for
demonstrating implicit memory have been relaxed
and in some cases ignored altogether?
Awareness on implicit memory tasks
We suggest that such a relaxation is not acceptable.
The reason for this is that when some measure of the
second criterion is included in implicit memory
studies, performance often seems to reflect an
unintentional but conscious use of memory. This
appears to be particularly evident when the implicit
memory task under consideration is a conceptual one.
For example, Mulligan and Hartman found that
93% participants claimed to be aware of what was
being tested on a category production task7. Also,
Thapar and Greene produced evidence showing, not
only that all their participants were test aware on a
general knowledge task, but also that participants
‘became’ aware very early on in the test25 (see also
Ref. 26). Furthermore, applying the process
dissociation procedure to a category-production task
Mecklenbräuker et al. showed that performance
almost entirely reflected a conscious use of memory27.
http://tics.trends.com
195
Whilst they argued that interpretation was difficult,
they concluded that performance on conceptual tasks
typically represents an unintentional, but conscious
use of memory.
‘There is virtually no evidence
that priming on conceptual tasks
reflects unintentional and
unconscious memory.’
Overall, despite occasional exceptions (e.g. Ref. 16),
these findings strike us as being incompatible with
those compelling instances of perceptual implicit
memory (e.g. Ref. 18) where performance is
characterized by a form of memory that is not only
unintentional but also unconscious. We should point
out that this issue is not exclusively restricted to
conceptual tasks. Both Richardson-Klavehn et al.28 and
Rybash and Osborne29 have produced evidence that
participants in their studies were routinely aware of
the study–test relationship. However, whereas there is
compelling evidence that perceptual implicit memory
performance for normal participants can operate in the
absence of accompanying awareness, the same cannot
be said for conceptual implicit memory tasks.
Confusion between unintentional conscious and
unconscious memory
Extrapolating from these studies, it seems likely that
a large percentage of purported demonstrations of
implicit memory, especially ones using conceptual
tasks, really reflect unintentional conscious memory.
So why is this a problem?
Firstly, if participants become aware during test
then they may switch to an intentional (and
conscious) retrieval strategy. Secondly, it makes
comparisons between perceptual and conceptual
implicit memory difficult because the latter appears
to be more routinely accompanied by awareness than
the former. However, the most fundamental problem
is that the majority of studies purporting to
demonstrate ‘implicit memory’ might not be
measuring what they claim to be. The need to
distinguish between unintentional unconscious and
unintentional conscious memory has been noted by a
number of researchers (e.g. Richardson-Klavehn
et al.4), including Schacter1 who explicitly warned
against confusing the two. Why then do researchers
persist in describing obtained effects as ‘implicit
memory’ whereas in all probability many are nothing
of the kind (i.e. they are consciously mediated)?
Are we being too harsh?
Some will of course claim that we are being harsh
when questioning evidence for implicit memory
obtained using normal populations. One can refer to
numerous examples of both perceptual and
conceptual implicit memory in a variety of patient
196 Opinion TRENDS in Cognitive Sciences Vol.5 No.5 May 2001

groups (see Ref. 30 for a review), whereby
contamination by explicit memory processes is
unlikely. These do provide compelling evidence for the
existence of both forms of implicit memory. However,
there are well-documented measurement and
interpretative problems with these forms of evidence
(see, for example, Ref. 31) and they may not have any
bearing on how normal participants will behave on
implicit memory tasks. We believe that the lack of
genuine evidence regarding implicit memory in
normal participants is a real concern. Thus, as we
enter the third decade of research into the
phenomenon, we wonder whether the term ‘implicit
memory’ has in fact outlived its usefulness.
Options for the future
At this point we believe that there are two main
options available. If the term ‘implicit memory’is to be
retained then it is imperative that more compelling
demonstrations are produced using both perceptual
and conceptual tasks. Roediger and McDermott outline
a number of precautions designed to reduce the
likelihood of participants becoming test aware32. We
suggest that researchers, particularly those using
conceptual tasks, incorporate such features, for
example, high ratios of filler to target test items, into
their designs (but see Ref. 4 for a different perspective).
‘…we wonder whether the term
implicit memory has in fact
outlived its usefulness.’
In meeting the second criterion for implicitness we
believe that some indicator of test awareness should
be routinely included in future studies (see also
Ref. 19). At the very least, this should take the form
of a retrospective questionnaire. A better approach
might be to incorporate the method of ‘on-line’
awareness testing devised by Richardson-Klavehn
and colleagues22. It could, for example, be used with a
conceptual task such as category exemplar production.
Finally, Beauregard et al. have developed an approach
based on the logic of demonstrating priming under
conditions where recognition is at chance33. Although
this approach precludes manipulating variables at
the time of study (e.g. levels of processing) we believe
it could be fruitfully applied to a variety of implicit
memory tasks.
Should more convincing evidence not be
forthcoming, then we believe the only other viable
option would be to drop the term ‘implicit memory’.
Recently, Gabrieli et al. have reframed the
explicit/implicit distinction in terms of ‘conscious
remembrance’ and occurring ‘incidentally’,
respectively24. This represents an attempt to distance
the term ‘implicit’ from ‘unconscious recollection’.
However, we believe that such a step is confusing
because the term ‘implicit’ has become so finely
ingrained with the concept of unconsciousness.
As an alternative, we suggest that it might be
appropriate to adopt the terminology used by
Richardson-Klavehn et al., who differentiate
between implicit and explicit memory tasks in terms
of retrieval instructions11. Thus, they describe a
word-stem completion task as an incidental
(i.e. unintentional) perceptual memory task and cued
recall as an intentional task. Again, this represents a
shift away from the troublesome issue of
consciousness. Although, in our opinion, it is precisely
the idea that memory can proceed without awareness
that makes implicit memory such an exciting field of
scientific inquiry, dropping the term might prove to be
the only realistic way forward.

References
1 Schacter, D.L. (1987) Implicit memory: history
and current status. J. Exp. Psychol. Learn. Mem.
Cognit. 13, 501–518
2 Graf, P. and Schacter, D.L. (1985) Implicit and
explicit memory for new associations in normal
and amnesic subjects. J. Exp. Psychol. Learn.
Mem. Cognit. 11, 501–518
3 Roediger, H.L. et al. (1989) Explaining
dissociations between implicit and explicit
measures of retention: a processing account. In
Varieties of Memory and Consciousness: Essays in
Honour of Endel Tulving (Roediger, H.L. and
Craik, F.I.M, eds), pp. 3–41, Erlbaum
4 Richardson-Klavehn, A. et al. (1996) Memory:
task dissociations, process dissociations and
dissociations of consciousness. In Implicit
Cognition (Underwood, G., ed.), pp. 85–158,
Oxford University Press
5 Jacoby, L.L. (1991) A process dissociation
framework: separating automatic from
intentional uses of memory. J. Mem. Lang.
30, 513–541
6 Schacter, D.L. et al. (1989) Intention, awareness,
and implicit memory: the retrieval intentionality
criterion. In Implicit Memory: Theoretical Issues
(Lewandowsky, S. et al., eds), pp. 47–65, Erlbaum
7 Mulligan, N.W. and Hartman, M. (1996) Divided
attention and indirect memory tests. Mem.
Cognit. 24, 453–465
8 Roediger, H.L. et al. (1992) Direct comparison of
two implicit memory tests: word fragment and
word stem completion. J. Exp. Psychol. Learn.
Mem. Cognit. 18, 1251–1269
9 Seamon, J.G. et al. (1997) A mere exposure effect
for transformed three-dimensional objects: effects
of reflection, size, or color changes on affect and
recognition. Mem. Cognit. 25, 367–374
10 Java, R.I. (1994) States of awareness following
word stem completion. Eur. J. Cognit. Psychol.
6, 77–92
11 Richardson-Klavehn, A. et al. (1999) Conjoint
dissociations reveal involuntary ‘perceptual’
priming from generating at study. Conscious.
Cognit. 8, 271–284
12 Mulligan, N.W. (1997) Attention and implicit
memory: the effects of varying attentional load
on conceptual priming. Mem. Cognit. 25, 11–17
13 Weldon, M.S. and Coyote, K.C. (1996) Failure to
find the picture superiority effect in implicit
conceptual memory tests. J. Exp. Psychol Learn.
Mem. Cognit. 22, 670–686
14 Vaidya, C.J. et al. (1997) Evidence for multiple
mechanisms of conceptual priming on implicit
memory tests. J. Exp. Psychol. Learn. Mem.
Cognit. 23, 1324–1343
15 Mulligan, N.W. (1996) The effects of perceptual
interference at encoding on implicit memory,
explicit memory, and memory for source. J. Exp.
Psychol. Learn. Mem. Cognit. 22, 1067–1087
16 Mulligan, N.W. et al. (1999) The effects of levels
of processing and organization on conceptual
implicit memory in the category exemplar
production test. Mem. Cognit. 27, 633–647
17 Kunst-Wilson, W.R. and Zajonc, R.B. (1980)
Affective discrimination of stimuli that cannot
be recognized. Science 207, 557–558
18 Seamon, J.G. et al. (1998) The mere exposure
effect is differentially sensitive to different
judgment tasks. Conscious. Cognit. 17, 85–105
19 Schmitter-Edgecombe, M. (1999) Effects of
divided attention on perceptual and conceptual
memory tests: an analysis using a process-
dissociation approach. Mem. Cognit. 27, 512–525
20 Toth, J.P. et al. (1994) Towards a redefinition of
implicit memory: process dissociations following
elaborative processing and self-generation.
J. Exp. Psychol. Learn. Mem. Cognit. 20, 290–303

http://tics.trends.com
 Opinion TRENDS in Cognitive Sciences Vol.5 No.5 May 2001 197

21 Jacoby, L.L. et al. (1993) Separating conscious and
unconscious influences of memory: measuring
recollection. J. Exp. Psychol. Gen. 122, 139–154
22 Richardson-Klavehn, A. et al. (1994) Involuntary
conscious memory and the method of opposition.
Memory 2, 1–29
23 Bowers, J.S. and Schacter, D.L. (1990) Implicit
memory and test awareness. J. Exp. Psychol.
Learn. Mem. Cognit. 16, 404–416
24 Gabrieli, J.D.E. et al. (1999) Convergent
behavioral and neuropsychological evidence for a
distinction between identification and production
forms of repetition priming. J. Exp. Psychol. Gen.
128, 479–498
25 Tharpar, A. and Greene, R.L. (1994) Effects of
levels of processing on implicit and explicit tasks.
J. Exp. Psychol. Learn. Mem. Cognit. 20, 671–679
26 Light, L.L. and Albertson, S.A. (1989) Direct and
indirect tests of memory for category exemplars in
young and older adults. Psychol. Aging 4, 487–492
27 Mecklenbräuker, S. et al. (1996) Conscious and
unconscious influences of memory in a conceptual
task: limitations of a process-dissociation
procedure. Swiss J. Psychol. 55, 34–48
28 Richardson-Klavehn, A. et al. (1994) Intention
and awareness in perceptual identification
priming. Mem. Cognit. 22, 293–312
29 Rybash, J.M. and Osborne, J.L. (1991) Implicit
memory, the serial effect, and test awareness.
Bull. Psychol. Soc. 29, 327–330
30 Gabrieli, J.D.E. (1998) Cognitive neuroscience of
human memory. Annu. Rev. Psychol. 49, 87–115
31 Ostergaard, A.L. and Jernigan, T.L. (1993) Are
word priming and explicit memory mediated by
different brain structures? In Implicit memory:
New Directions in Cognition, Development, and
Neuropsychology (Graf, P. and Masson, M.E.J., eds),
pp. 327–349, Erlbaum
32 Roediger, H.L. and McDermott, K.B. (1993)
Implicit memory in normal human subjects. In
Handbook of Neuropsychology (Vol. 8) (Boller, F.
and Grafman, J., eds), pp. 66–131, Elsevier
33 Beauregard, M. et al. (1999) Word priming without
awareness: a new approach to circumvent explicit
memory contamination. Brain Cognit. 39, 149–169

which we expect to see her. Although in strict

Surface perception geometrical terms any retinal image of granny is

ambiguous, because it is a 2-D projection of a 3-D
scene and therefore loses the spatial component

and the generic view along the line of sight, we seem to recognize her
effortlessly, without consciously considering any
alternative percepts.

principle Prior probabilities and Bayesian theories of perception

This suggests that the richness of our visual
experience depends upon the exploitation of
Marc K. Albert regularities of the environment to constrain image
interpretation. On this view, our prior beliefs about
our chances of encountering certain 3-D objects
The hypothesis that perceptual experience can be understood in terms of (and/or viewing them from certain vantage points)
rule-based processing has strongly influenced recent theories of visual surface are represented by biased prior probability
perception. However, many of the rules that these theories propose apply only distributions. Thus, we would give a relatively high
in relatively restricted situations. I suggest that more general and robust prior probability to our chances of encountering
principles for reducing perceptual ambiguity are available, such as the generic granny relative to any other object that could be
view principle (GVP) described here. According to the GVP, vision assumes that confused with her. This kind of emphasis on prior
qualitative (e.g. topological) image structure is stable with respect to small probabilities has been advocated by many
changes of viewpoint. Some consequences of the GVP for visual surfaces, investigators who favor a Bayesian analysis of
including illusory surfaces, are described. I also demonstrate the decisive role visual perception4.
of real and illusory background surfaces in specifying the 3-D shape and layout However, visual priors that are not based on
of visual objects and scenes. reliable regularities of the environment in which an
organism evolved must be learned by each organism
Surfaces are the building blocks of our visual from scratch through experience. Add to this the fact
experience1,2. Although images of visual surfaces that many priors are highly context-dependent, and
frequently do not contain depth cues in their we see that the Bayesian view entails that prior
interiors, such as perceptible texture or shading, probabilities are likely to include large amounts of
they are often perceived relatively unambiguously data that must be learned over an extended period
in 3-D (Ref. 3). What resolves the depth ambiguity of time before an organism would attain a reasonable
in sparse monocular images such as the silhouettes level of competence. For these reasons, among
in Fig. 1? As any given image is consistent with many others, it has been suggested that the use of detailed
possible 3-D interpretations, why is perception not priors (implicitly or explicitly) in a Bayesian analysis
equally ambiguous? is not a feasible strategy for a biological visual
One response to these questions is that the world system with limited resources that must get ‘up and
is not a random place, and our experiences are not running’ quickly, and consequently, that this strategy
Marc K. Albert
Dept of Psychology,
random sequences of events. They have a significant is not a sufficient theoretical basis for understanding
University of degree of ‘regularity’ to them. Certain visual images human visual surface perception .
Southampton, Highfield, are seemingly clear and unambiguous signs of the I argue instead for the central role played by more
Southampton,
presence of certain familiar objects in the world. For general principles of vision, specifically the ‘generic
UK SO17 1BJ.
e-mail: M.K.Albert@ example, we can easily identify our grandmother in view principle’ (GVP). This principle states that
soton.ac.uk a crowd of people, particularly in situations in visual images are likely to be qualitatively stable with

http://tics.trends.com 1364-6613/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1364-6613(00)01643-0