Psychological Review Copyright 1981 by the American Psychological Association, Inc.

198!, Vol. 88, No. 1,67-85 0033-295X/81 /8801 -0067S00.75

Reality Monitoring
Marcia K. Johnson Carol L. Raye
State University of New York at Stony Brook Barnard College

People remember information from two basic sources: that derived from external
sources (obtained through perceptual processes) and that generated by internal
processes such as reasoning, imagination, and thought. Of particular interest to
us are the processes people use in deciding whether information initially had an
external or an internal source, which we call "reality monitoring." We propose
a working model of reality monitoring to account for both discrimination and
confusion between memories for thoughts and memories for perceptions. Ex-
amples of questions the model addresses are, What types of information are more
likely to be represented in memories of external events than in memories of
internal events? What cues allow people to decide the origin of a memory? What
is the nature of the decision processes involved? Which processes, and under
what conditions, are likely to break down and lead to unreliable memory? What
assumptions do individuals have about their memory for their thoughts com-
pared to their memory for their perceptions? How accurate are these assump-
tions? We summarize some research that is encouraging as far as the tractability
of some of these problems is concerned and that demonstrates the usefulness of
the particular working model proposed here.

Both perception of external stimuli and something may make it seem as though it
thought produce memories. We have re- was perceived more often than it actually
ferred to the processes by which a person was (e.g., Johnson, Taylor, & Raye, 1977).
attributes a memory to an external or an However, real commonly refers to things
internal source as reality monitoring (John- existing outside of oneself, and it is this sense
son, 1977; Johnson & Raye, Note 1). Of that is implied by reality monitoring. The
course, in one sense, the memories created term is intended to suggest similarities with
by thoughts are no less "real" than those the concepts of reality testing and memory
created by perceptual experiences, and the monitoring. Reality testing generally refers
former can be shown to have important con- to the process of distinguishing a present
sequences; for example, thinking about perception from a present act of imagination
or act of remembering (Cameron, 1963;
Freud, 1895/1966, p. 325, 1925/1961, p.
This research was supported by Grant BNS-7813054
from the National Science Foundation. We would like 237; Horowitz, 1978).
to thank Mark Aronoff, Ed Casey, Don Fry, Julie Hoch- Reality testing has received some experi-
berg, Emil Menzel, and Tom Taylor for their partici- mental attention from investigators inter-
pation in discussions that clarified some of the ideas in ested in perception (Perky, 1910; Segal &
this article and for their helpful comments on an earlier
draft of the manuscript.
Fusella, 1970) and from those interested in
The general issue addressed here was raised at the clinical problems, especially because of the
annual meeting of the American Psychological Asso- central role confusion between the real and
ciation, Chicago, 1975. The proposed model and some the imagined plays in diagnoses of schizo-
of the present data were initially presented at the annual phrenia (McGuigan, 1966; Mellor, 1970;
meeting of the Psychonomic Society, San Antonio,
Texas, 1978, and at the annual meeting of the Rocky Mintz & Alpert, 1972). In contrast, reality
Mountain Psychological Association, Las Vegas, Ne- monitoring refers to the process of distin-
vada, 1979. guishing a past perception from a past act
Carol Raye is now at Bell Telephone Laboratories, of imagination, both of which resulted in
Piscataway, New Jersey 08854.
Requests for reprints should be sent to Marcia K. memories. Reality monitoring thus also
Johnson, Department of Psychology, State University shares with Hart's (1967) memory monitor-
of New York, Stony Brook, New York 11794. ing concept the idea of making judgments
67
68 MARCIA K. JOHNSON AND CAROL L. RAYE

about information in memory. Reality mon- hood experiences such as sexual encounters
itoring is particularly interesting in that may suggest different psychological con-
fairly extreme errors about the origin of structs depending on whether they represent
memories are probably more common than early imaginative processes or actual per-
reality testing failures such as hallucina- ceptual experiences. Freud's decision to treat
tions. Reality monitoring failures may also such memories as largely the product of fan-
generally be more difficult to correct than tasy greatly influenced the development of
misperceptions. You cannot, for example, psychoanalytic theory, especially of the idea
move your head to look for changes in oc- of infantile sexuality (Heidbreder, 1933, pp.
clusion or reach out and try to touch an er- 385-386). In short, the fact that the criteria
roneous memory. of veridicality may not be entirely clear to
One way to regard the events we call fail- memory theorists does not argue against the
ures in reality monitoring is as the extreme central role veridicality plays in the concept
consequences of elaborative processes that of memory.
have consistently interested cognitive psy- Our position is not really in conflict with
chologists. Cognitive approaches to memory a general cognitive view emphasizing the
emphasize memories as a joint product of contribution of thought to comprehension
external and internal events (e.g., Bartlett, and memory (e.g., Bransford & Johnson,
1932; Bruner, 1957; James, 1890/1950; 1972, 1973). However, in attempting to
Krech, 1949; Kohler, 1947/1959; Lashley, demonstrate that people do think while per-
1967; Postman, 1951; Tolman, 1948; and, ceiving and remembering, there has been a
more recently, Bower, 1972a; Bransford & tendency to overemphasize the hopeless en-
Johnson, 1973; Cofer, 1973; Craik & Lock- tanglement of memories derived from per-
hart, 1972; Dooling & Lachman, 1971; ception and thought. Many investigators
Hochberg, 1979; Kintsch, 1976; Mandler, have assumed not only that a memory is a
1967; Neisser, 1967; Paivio, 1969; Tulving, joint product of the external and internal
1968). With rare exceptions (e.g., Gibson, events but also that this joint memory exists
1966, 1979), few modern investigators pro- in place of what might be called the external
pose that there is a straightforward rela- and the internal components. For example,
tionship between physical stimuli and the "We have strong evidence that the list-as-
mind's impression of those stimuli, much less presented is not the list-as-encoded, just as
our memory for them. Rather, the more gen- a CVC presented may not lead to a CVC
eral view is that most, if not all, stimuli are stored, but to an NLM [natural language
subject to alternative interpretations and mediator] instead" (Klatzky, 1980, p. 222,
elaboration and that what is remembered is emphasis added). The notions that exter-
based on the way that events are interpreted, nally generated information is discarded
organized, and thought about. Given this once it is used to generate a construction
theoretical framework, is there any reason (Sachs, 1967) or that externally and inter-
to make a distinction between memories of nally generated traces are equivalent in ev-
perceptions and memories of thoughts? ery respect have led to a number of expec-
The idea that there can be no memories tations—for example, that ideas and their
of pure sensory experience without some de- paraphrases should be indistinguishable
gree of cognitive elaboration is certainly (Bransford & Franks, 1971; Jenkins, 1974),
compelling. Yet equally compelling is the that both perceived and inferred components
idea that memories do vary in the extent to of a mental map should be equally available
which they are tied to external events (i.e., (e.g., Levine & Jankovic, Note 2), and that
publicly available stimulus conditions). In if inferences do not intrude in recall, they
ordinary usage, memory is by definition dis- must not have been made previously (e.g.,
tinguished from imagination by virtue of its Corbett & Dosher, 1978).
veridicality. Furthermore, the importance of Thus, cognitive theorizing has sometimes
the distinction between externally and in- assumed that the inability to disentangle in-
ternally generated information is not simply ternally and externally generated informa-
definitional. For example, memories of child- tion is the defining characteristic of a "cog-
 REALITY MONITORING 69

nitive" system. However, it is not logically able in the having, it is likely that the copy
necessary to prove that specific memory (i.e., of the sensation should be distinguishable
memory faithful to perceptual events) does from the revival of the idea" (p. 334). Al-
not exist in order to show that organized though little attention has been given to the
mental representations do exist. Yet it is issue of discriminating memories produced
precisely the tendency to equate evidence for by perceptions from those produced by
constructive or imaginative processes with thought, a number of philosophers have ad-
evidence against accurate memory (and vice dressed the issue of how sensations and ideas
versa) that limits the questions cognitive psy- are distinguishable initially ("in the hav-
chologists have posed. In addition, in em- ing"). Two general lines of thought have
phasizing the many transformations a stim- been that perception and imagination pri-
ulus might undergo, we run the danger of marily differ in that percepts are stronger
forgetting the dire functional implications or more vivid (e.g., Hume, 1739/1978;
of a memory system that is assumed to be James, 1890/1950) and the opposing idea
so loosely tied to external events (although that "sensation . . . and imagination, even
the problem is beginning to be recognized; where they have the same object, are oper-
Neisser, 1976). ations of a quite different nature, and per-
If thought is assumed to produce memory fectly distinguishable by those who are
representations but not necessarily to sup- sound and sober" (Reid, 1764/1975, chap.
plant those yielded by perception, a different 2, section 5, p. 18). The model presented
set of questions for research is suggested. below incorporates aspects of both of these
What are the characteristics of both mem- hypotheses and further suggests that some
ories derived primarily from perceptual ex- of the same cues that allow judgments about
perience and memories more remotely re- the nature (present perception or present
lated to perceptual experience? How are the imagination) of conscious experience are
representations of externally and internally preserved and become the bases for judg-
generated events distinguished, and when ments about the nature of past experience.
and why are they confused? Understanding
the differences and similarities in the nature The Present Approach
of internally and externally generated mem-
ories is important for any theory of the way Some Preliminary Assumptions
information is represented in memory, and
an adequate model must be able to account Although we assume that perceptual events
for both confusion and discrimination be- produce persistent memory traces (perhaps
tween past imaginations and past percep- in a kind of continuous record of experience;
tions (Johnson, Note 3). J. S. Mill (1869/ Landauer, 1975), we also assume that in-
1967) suggested that "the only difficulty ternally generated events produce persistent
about Memory, when once the laws of As- memory traces. Not all, but many of the
sociation are understood, is the difference errors we see in memory may be the con-
between it and imagination" (p. 339). To sequence of a failure to discriminate the or-
paraphrase this notion to fit the present igin of a memory trace. The ideas that mem-
problem, once we have acknowledged the ory traces for perceptual experiences are
role of imagination in memory, we still have accurate, in that they reflect the character-
the problem of accounting for the condi- istics of stimulus-determined processes and
tions under which we know the difference that they exist side by side or intertwined
between memories of external events and with memories produced by imaginative pro-
memories of imaginations. What are our cesses have particular implications for the-
"cues to reality?" ories of representation that are based on er-
For the most part, philosophers and psy- ror data such as false recognitions. Insofar
chologists have not been concerned with the as false recognitions reflect a failure to dis-
problem of reality monitoring, although criminate the origin of a trace, their appro-
James Mill (1829/1967) did assert that be- priateness as an indication of the nature of
cause sensations and ideas "are distinguish- the memory representation of external events
70 MARCIA K. JOHNSON AND CAROL L. RAYE

is somewhat questionable. That is, errors monitoring as a way of referring to the ac-
may reflect coexistence of information from tivity of discriminating between memories
multiple sources as well as integration, and primarily derived from external events and
this distinction is beginning to be investi- those primarily derived from internal events.
gated (e.g., Cole & Loftus, 1979; Hasher However, this term is not meant to imply
& Griffin, 1978). Although this argument that the processes involved in discriminating
may at first seem to be quibbling over ter- the origin (internal or external) of infor-
minology, these two metaphors certainly im- mation in memory are always the same.
ply different directions for research. They Rather, the details of the processing (acti-
might also lead to different expectations vating information, applying decision strat-
about the potential success of introspective egies) might reasonably be expected to vary
effort (and more extreme measures such as depending on the nature of the remembered
hypnosis) in separating the real from the information, the conditions under which the
imagined. monitoring occurs, the cost of mistakes, and
so forth. Finally, reality monitoring can be
Types of Self-Generated Information thought of as one of the general class of
metamemory processes (e.g., see Brown,
There are many types of thoughts and 1975; Kreutzer, Leonard, & Flavell, 1975)
imaginal events that would be relevant for by which people reflect varying degrees of
reality monitoring. These types can tenta- understanding about the nature of their own
tively be separated into three major cate- memories.
gories: (a) re-representation, (b) cotemporal
thoughts, and (c) fantasy. The first category A Working Model
refers to the re-representation of perceptual
experience or remembering something pre- As a vehicle for interpreting available data
viously experienced. In re-representation, in- and as a stimulus for further research, we
formation that has dropped out of conscious- have developed the following working model
ness or working memory is reactivated at a of reality monitoring. Basically, we propose
later time in the absence of the original ex- that reality monitoring is affected by two
ternal stimulus (in contrast to, for example, major factors: the nature of the traces being
rehearsing recently perceived information evaluated and the types of decision processes
several times in succession). Cotemporal applied. Either of these should influence the
thought refers to the sort of elaborative and amount of confusion a person displays about
associative processes that augment, bridge, the origin of information. Our first assump-
or embellish ongoing perceptual experience tion is that any memory potentially consists
but that are not neccessarily part of the ve- of many types of information or attributes
ridical representation of perceptual experi- (e.g., Bower, 1967; Craik & Lockhart, 1972;
ence. Most work on imaginal processes in Posner & Warren, 1972; Underwood, 1969;
memory has been directed at this category Wickens, 1970). Thus, for example, if a
(e.g., Bartlett, 1932; Bower, 1972b; Neisser, word is presented or imagined, information
1967). Fantasy involves novel combinations may be stored with respect to its physical
of information that produce imaginary events properties (e.g., pitch) and its semantic prop-
that take place only in our imagination. erties. Types of potential attributes that
Making up a story or dreaming would fall should be particularly important for reality
primarily in this category. While recognizing monitoring include information about the
that there are some problems with this clas- sensory characteristics of the stimulus pre-
sification scheme (where does "re-represen- sentation (e.g., auditory information), the
tation" end and "fantasy" begin?), we have type of cognitive processing engaged (e.g.,
found it useful as a starting point. To date, imagery), semantic content, and contextual
our research has primarily consisted of lab- information (e.g., spatial and temporal in-
oratory investigations of the first two cate- formation). Various subcategories could be
gories. defined within each of these four. For ex-
As stated above, we use the term reality ample, semantic content for a spoken word,
 REALITY MONITORING 71

as used here, would include physical aspects Although for some cases these character-
of its referent such as size and color, func- istics might not hold, if the two classes of
tional characteristics of the referent, emo- representations generally differ in these ways,
tional connotation, and so on. An issue that then these dimensions could be used to de-
could plague us, as well as most recent the- cide the origin of a trace regardless of the
ories of memory based on "attributes," "lev- particular content of the memory involved.
els," or "codes" (e.g., Craik & Lockhart, Assuming that the populations of external
1972; Paivio, 1971; Underwood, 1969), is, and internal memories form distributions on
How is it possible to separate sensory from each of these four dimensions, a reasonable
semantic dimensions of memory represen- model of the process involved in decisions
tations? The fact that something is seen and about the origin (internal or external) of a
not heard (or said by a low frequency voice memory might be similar to some models of
rather than a voice higher in frequency) old-new recognition decisions (e.g., Atkin-
might have significance in the semantic son & Juola, 1973; Mandler, 1972; Raye,
sense, as might the spatial or temporal po- 1976). That is, criteria may be set on each
sition of an event. These categories are not dimension, one for identifying probable in-
necessarily proposed as hard and fast divi- ternal events and one for probable external
sions (any more than one can make a clear events, and any event registering between
distinction between internal and external) the two cutoffs would be of uncertain origin
but represent sensible and useful analytic and require further processing. Figure 1 il-
categories for characterizing and under- lustrates such a decision structure for one
standing the nature of information repre- of the proposed dimensions.
sented in memory. Information about these four dimensions
We propose that, as a class, internally could of course be combined in different
generated memories may differ from the ways. Decisions could be based on a strategy
class of externally generated memories along that always or usually weighted the four in
specific dimensions. First, externally gener- a constant fashion, for example, sensory at-
ated memories in general may have more tributes might always be weighted more
spatial and temporal contextual attributes heavily than amount of detail, and amount
coded in the representation of the event than of detail more heavily than contextual at-
internally generated memories do. Second, tributes, and so on. On the other hand, dif-
they should also have more sensory attri- ferent dimensions might be given more
butes, although imaginal processes presum- weight in varying instances depending on
ably also generate some sensory information what other information the individual has
(e.g., see Atwood, 1971; Brooks, 1968; Finke, about the situation. In general, then, it
1979; Kosslyn, 1976). Third, we propose that should be possible to produce confusion
externally generated representations are
more semantically detailed—that is, contain PROBABLE UNCERTAIN ORIGIN PROBABLE
more information or more specific infor- INTERNAL EXTERNAL
mation—than internally generated represen-
tations. Thought tends to be more schematic. A
Finally, we propose that internally generated 0 A AA
memories may typically have more opera-
tional attributes associated with them (coded
in the trace). This latter notion is based
o O OA
partly on the assumption that perception is OAO AOA A A A A
usually somewhat more "automatic" (Brown,
STRENGTH OR AMOUNT OF CONTEXTUAL INFORMATION
1975; Hasher & Zacks, 1979; Posner & Sny-
der, 1975) than imaginal processes and that Figure I. Representation of a set of decision rules for
attention (e.g., voluntarily engaging in cre- judging the origin of a memory on the basis of the
amount of contextual (time and place) information it
ating images) increases the chances that includes. (In this sample, externally derived memories
coded information about operations will be are represented by triangles and internally derived ones
available later (see Posner & Warren, 1972). by circles.)
72 MARCIA K. JOHNSON AND CAROL L. RAYE

about the origin of an event by manipula- memory assumptions or beliefs held by the
tions that cause it to have abnormally high individual. For example, subjects might have
or low values on one or more dimensions, certain "biases" about memories derived
compared to typical memories of its class; from their own thoughts (e.g., "I'd remem-
an interesting question is whether some di- ber if it had been my idea"). These latter
mensions are more important in this regard processes need not always wait for the out-
than others. come of the more general decision based on
In addition to atypical class characteris- class characteristics discussed above but
tics, a major vehicle for confusion should be may be initiated simultaneously. However,
specific similarity (with respect to any at- since they are assumed often to involve more
tribute) between particular memories from specific information, which may require ad-
the two classes. We have discussed "a mem- ditional retrieval from memory, they are
ory" as if it could be isolated, but this is just presumed normally to take longer and for
for convenience. Clearly, when a memory is this reason might be considered a second
activated, so are other memories, and the stage. Which processes play the predomi-
characteristics of these other memories may nant role in a particular decision should de-
influence or dominate the decision process. pend on such factors as the amount of time
Thus in considering the origin of a memory a person has, availability of additional in-
that was actually generated internally, one formation in memory, and the cost of mis-
might be influenced by extensive sensory takes. Thus, if the consequence of asserting
characteristics of semantically similar traces that something was perceived when in fact
that are also activated while the target is it might only have been inferred is serious,
being evaluated. The extreme case would be a person should be particularly likely to at-
one in which a retrieval cue activates a sim- tempt to retrieve additional information that
ilar trace from the opposite class but not the would tend to support or confirm one or the
target, causing an apparent error in identi- other decision.
fication of origin when, in fact, the target In general, both perception and thought
trace is not being considered at all. result in memories that persist. Confusion
As mentioned earlier, decisions about the of the two does not necessarily happen via
origin of a memory based on the proposed degradation of traces but rather can happen
typical class differences may be inconclusive, via other mechanisms.
and the person may access further infor-
mation. The person then may use this further Empirical Evidence
information to reason toward a decision
about the origin of the memory. This further The model developed above (see Table 1)
information may consist of qualitative char- represents a set of interrelated hypotheses,
acteristics of the trace being evaluated in and we have tried in a number of ways to
combination with the person's prior knowl- translate some of these hypotheses into ex-
edge. For example, a person may remember periments.
an extremely vivid dream about a money tree
but reason that it could not be an externally Evidence That the Classes of Externally
generated memory because money does not and Internally Generated Memories Differ
grow on trees. Or this further information
may consist of characteristics of related A fundamental assumption of the model
traces. For example, suppose an architect is that the classes of externally and internally
was not sure whether a design she had in derived memories differ. Two lines of evi-
mind was initially hers or her associate's. dence suggest this is a reasonable starting
She then remembers not only the idea but point.
also her associate's starting to draw a rough Superiority effects. Slamecka and Graf
sketch and breaking his pencil. Therefore, (1978) reported the results of several exper-
she reasons, the idea must have been his. iments specifically directed at determining
Finally, this type of reasoning also prob- whether there is a memory advantage for
ably uses strategies based on certain meta- self-generated information. Their general
 REALITY MONITORING 73

Table 1
Summary of the Model
Types of attributes potentially comprising memories
Contextual
Sensory
Semantic
Cognitive operations
Dimensions on which the classes of externally generated and internally generated memories typically differ
External have more contextual attributes
External have more sensory attributes
External have more semantic detail
Internal include more information about cognitive operations
Processing characteristics of reality monitoring
Decisions about the origin of a specific trace may be based on a weighted combination of the results of
comparing the target trace attributes indicated above to criteria defining the general classes of external and
internal representations
Decisions may be based on a reasoning process involving
Qualitative characteristics of the target trace
Characteristics of related traces
Metamemory assumptions
Which processes take place should depend on such factors as time, availability of different types of
information, cost of mistakes, and so forth
Sources of errors in reality monitoring
Target trace not typical of its class
Characteristics of similar incorrect traces
Failure in a reasoning process (e.g., failure to retrieve additional information, incorrect metamemory
assumptions)

procedure was to present subjects with a Raye, Foley, & Foley, in press, Experiment
stimulus and either to present a response 1). This outcome supports the idea that the
(E items) or to require the subject to gen- locus of the advantage is in the generation
erate a response (S items) according to a process itself and does not, for example, de-
rule and a specified first letter. The rule and pend on overt expression of the generated
first letter constrained the generated re- item and consequent sensory components
sponses (e.g., synonym-rapid-f almost al- (e.g., hearing one's own voice or seeing one's
ways produced the response fast), allowing own handwriting) or on some combination
Slamecka and Graf to counterbalance the of generation and sensory components.
assignment of items to conditions such that In several studies, we compared frequency
E and S items were the same across subjects. judgments of externally generated events
Across a number of types of items and in- with frequency judgments of self-generated
structional conditions, recognition and recall events (Johnson, Taylor, & Raye, 1977;
were better for the subject-generated items Raye, Johnson, & Taylor, 1980; Taylor,
(the "generation effect"). Johnson, Birnbaum, & Raye, Note 4). The
Using a similar procedure, we have also basic paradigm involves two kinds of trials,
found a generation effect after retention in- presentation and generation. On presenta-
tervals as long as 10 days (Johnson, Raye, tion trials, subjects see a number of cues and
Foley, & Foley, in press, Experiment 2). In to-be-remembered words; on generation
addition, in one study we manipulated trials, the cues are shown, and subjects are
whether subjects overtly responded by writ- asked to generate the items. The cues are
ing down E and S words as they occurred preselected so that they will elicit the ap-
or whether they were simply required to re- propriate response with almost 100% cer-
spond covertly by listening to E words and tainty (e.g., color-blue). Over the course of
thinking about S words. The generation ef- the procedure, each item is seen either two,
fect was as great in both free and cued recall five, or eight times and is generated by sub-
when the subjects thought about but did not jects either two, five, or eight times. Finally,
overtly express their responses (Johnson, subjects are given a surprise frequency judg-
74 MARCIA K. JOHNSON AND CAROL L. RAYE

ment task. For example, they are asked to manipulation to affect the criteria subjects
estimate the number of times they actually used for including memories in an "event
saw each item, and we explain to them that count." Of special interest was whether the
we are not interested in the number of times two types of judgments would be affected
they generated it but only the number of similarly. If they were, it would emphasize
times we presented it. Or, they are asked to the similarity between externally and inter-
indicate the number of times they generated nally derived memories. However, if the two
an item, ignoring the number of times they were affected differently, the results would
perceived it. point to differences in the memories or judg-
These studies had two major outcomes. ment processes in the two cases. For ex-
The first was that subjects were more sen- ample, if memories for one type of infor-
sitive to the relative frequency of self-gen- mation are more stable and faithful, they
erated events than to the relative frequency might be less susceptible to variations in
of externally generated events. The second judgment criteria. In fact, the limit did not
major finding concerned confusion between affect the magnitude of confusion when peo-
the two types of events. If there were no con- ple were judging the frequency of internally
fusion between presentations and genera- generated events. This finding suggests that
tions, it would not matter, for example, how even without the limit, people were using a
often an item had been generated when sub- fairly stringent criterion of trace appropri-
jects judged how often it was presented. ateness. (Perhaps specific would be more
However, generating events systematically accurate than stringent, since whatever they
increased the frequency with which those used allowed for good discrimination of rel-
events seemed to have been perceived. Sim- ative frequency of self-generated events.)
ilarly, perceiving items systematically in- When external event frequency was judged,
creased the frequency with which those however, the limit reduced confusion, ap-
items seemed to have been generated. How- parently because subjects selectively ex-
ever, the two types of events did not interfere cluded more inappropriate, self-generated
with each other equally. The extent to which event memories. Thus, these two types of
subjects were confused by internal events traces must differ in some way that allows
when they judged external events was greater the subject to discriminate between them.
than vice versa. Thus both the steeper fre- Raye et al. argued that these results were
quency discrimination function for self-gen- consistent with the present working model
erated information and the greater confusion if it were assumed that when judging the
in judging external events indicated that in- frequency of internally generated events,
ternally generated memories were superior subjects were relying heavily on the presence
to or had some advantage over externally of information about cognitive operations in
generated memories in this situation, as well making decisions about whether to include
as in recall and recognition tests. A number memories in an event count (see Raye et al.,
of related findings also suggest there is some 1980, for an expanded discussion).
advantage associated with self-generated in- Comparing between- and within-class
formation (e.g., Bobrow & Bower, 1969; discriminations. Another line of research
Davies, Milnee, & Glennie, 1973; Erdelyi, is based on the logic that if two classes of
Buschke, & Finkelstein, 1977; Jacoby, 1978; events are different, discriminating between
Jarvella & Collas, 1974; Ross & Sicoly, instances from within a class should be more
1979). difficult than discriminating between in-
Another indication that external and in- stances drawn from separate classes. For
ternal memories differ comes from Raye et example, is reality monitoring subject to
al. (1980, Experiment 2). Half the people more or less error than monitoring the
judging each type of event were given a limit origins of information derived from two ex-
for their estimates—they were asked to use ternal sources? Our model proposes that,
numbers between 0 and 10. The other half along with more specific information, there
of the subjects were not given any limit or are a number of general dimensions on which
range for their estimates. We expected this externally and internally derived memories
 REALITY MONITORING 75

may characteristically differ and which may were discriminating between internal and
be used in the process of reality monitoring. external sources, were better at identifica-
That is, decisions could be based simply on tion of origin than the listeners and re-
the amount of certain types of information corders, who were discriminating between
(e.g., if it has a high sensory component, two external sources of information.
classify it as external). In contrast, two This result was taken as support for the
memories derived from external sources idea that external events as a class differ in
could not be distinguished along these gen- some ways from the class of internal events.
eral dimensions. Thus memories from exter- These differences potentially provide addi-
nal sources would always have to be discrim- tional cues that allow the origin of memories
inated on the basis of specific content (specific for internally generated events and external
sensory information rather than amount, events to be distinguished more easily than
etc.). It follows that, given similar situations, the memories for two comparable external
it should be easier to identify the origin of events. Such cues would operate in the first
internally versus externally derived memo- stage of the model. For example, greater sa-
ries than to identify the sources of memories lience of the cognitive operations that pro-
derived from two external sources. duced internally generated traces may serve
In one experiment testing this prediction, as a discriminative cue available in external-
subjects participated in small groups during internal discriminations that is not available
the first phase of the experiment (Raye & in external-external discriminations. An-
Johnson, 1980). Within a group they were other possibility is that the semantic contents
randomly assigned to play different roles: of the memories differ in some way. Intu-
speakers, recorders, or listeners. All subjects itively, internally generated responses should
thought that the speakers and recorders were be more idiosyncratically determined than
serving as "experimenters," that the listeners external events. Internally generated events
were the "subjects," and that the listeners are not necessarily more meaningful than
would later be tested for their memory of external events, but they are more likely to
the "conversation." The group was given a refer to biographical information or infor-
topic of conversation to start them off, for mation relevant to the self. Internal events
example, snow. Speaker A then gave a reflect idiosyncratic probabilities, whereas
highly related word such as white. Speaker external events reflect normative probabili-
B gave a word related to Speaker A's word, ties. The presence of a difference in average
for example, black, then Speaker A replied, idiosyncracy value for external and internal
for example, coal, and so on, alternating. items could provide a discriminative cue:
One recorder wrote down all of the words Would I have said this? Is this a high fre-
spoken by Speaker A; the other recorder quency response for me? Does this response
wrote down all the words given by Speaker have some special meaning for me? This fac-
B. The listeners were instructed to listen tor might operate in the reasoning stage of
carefully to both speakers in preparation for reality monitoring.
a memory test later. After a 1-hour, filled, In a second conversation study (Raye &
retention interval, all subjects received a rec- Johnson, 1980, Experiment 2), the content
ognition test which had the standard re- (as well as the idiosyncracies) of the con-
quirement that subjects discriminate be- versation was controlled. A subject, the
tween old and new items and the additional "director," asked the speakers questions
requirement that, for old items, they identify such as, "What is a word that is the opposite
which speaker said each. of fast, beginning with s and ending with
The measure of primary interest was the wl" The director indicated which of the two
percent correct identification of origin: the speakers was to answer, and the speakers'
total number of words correctly attributed responses were essentially determined by the
to each of the two speakers divided by the questions, which, obviously, left little room
total number of words correctly identified as for idiosyncratic contributions to the content
from the conversation (multiplied by 100). of the conversations. Again, there were re-
As predicted by the model, the speakers, who corders and listeners, and only the listeners
76 MARCIA K. JOHNSON AND CAROL L. RAYE

thought they would get a memory test. All & Chromiak, 1979). In contrast, compared
subjects, including the directors, were later to older subjects, 6-year-olds did have a par-
given a recognition and identification-of-or- ticularly difficult time discriminating be-
igin test. tween words they had said aloud and words
Again, the speakers were significantly bet- they had only thought. Within the present
ter at identifying the origin of an item they framework, these data suggest two interest-
recognized as old than were the recorders ing possibilities. First, children as young as
and listeners, and the recorders and listeners 6 years may have learned to take advantage
did not differ from one another. The direc- of cues in memory that differentiate the self
tors had the lowest discrimination scores. from others but not those that differentiate
This finding rules out, we think, the possi- various self-initiated acts such as speech and
bility that the entire difference in the pre- thought. Thus the special confusion of young
vious experiment could be attributed to dif- children could be a consequence of a less
ferences across conditions in the idiosyncratic developed reasoning process and not a con-
value of the content of the items. In fact, the sequence of trace characteristics. Alterna-
magnitude of the difference was almost ex- tively, 6-year-olds, compared to older sub-
actly the same as it was in the previous ex- jects, may produce memories when thinking
periment, suggesting that idiosyncratic cues that are, in fact, more like the memories
did not add much. Of course, we would not produced from talking. Similarity between
want to offer this as a general conclusion, thought and speech would be increased, for
since we would expect idiosyncratic value example, by subvocalizing during thought,
typically to be one of the cues used in the and there is some evidence that children of
reasoning processes involved in reality mon- this age do show relatively high levels of sub-
itoring. However, the present results are con- vocalization (Garrity, 1977).
sistent with the possibility that, indepen-
dently of the personal significance of what Specific Dimensions Proposed as
an individual says, the cognitive operations Discriminators
that go into generating information persist
in memory and become potential cues as to Cognitive operators. According to the
the origin of that information. reality monitoring model, increasing auto-
A parallel logic suggests that it should be maticity in responding ought to reduce in-
easier to discriminate between memories of formation regarding cognitive operations.
external and internal events than between The stronger or more specific the cue for a
memories of two internally generated events response, the more automatically the re-
because, for example, the latter would both sponse should be generated, reducing search
include large amounts of cognitive opera- and decision processes that later could be
tions information, minimizing cognitive op- used as cues to the origin of information.
erations information as a discriminative cue. Conversely, if a response requires somewhat
In two developmental studies, Foley, John- more effort or search, stored information
son, and Raye (Note 5) had subjects engage about these operations should become a po-
in two types of self-generated acts, either tential cue for identifying the source of a
saying items aloud or only thinking about memory. In order to manipulate this search
them. Subjects then attempted to identify process, we used a procedure in which sub-
which items they said and which they jects heard category cues followed by cate-
thought. Consistent with expectations from gory instances on half of the trials and gen-
our model, this condition was more difficult erated their own category instances for the
than one in which the subjects discriminated other half of the trials. Half of the subjects
between items they had said and items some- were cued with the first letter of the re-
one else had said. sponse, and half were not cued with the first
Compared to older subjects, 6-year-old letter. Our reasoning was that providing the
children were at no disadvantage in discrim- first-letter cue should more completely de-
inating what they said from what someone termine the response, making it more au-
else said (see also Johnson, Raye, Hasher, tomatic, than not providing the first letter.
 REALITY MONITORING 77

Later, subjects were asked to discriminate likely to be confused with external events.
between category instances that were pre- Intrusions seem to occur more often with
sented and instances they generated. Gen- prose. It is not that more thought necessarily
erating instances of categories to fit first-let- goes into processing prose compared to word
ter cues resulted later in lower correct lists; it is perhaps that more conscious (in
identification of origin than generating in- the sense of purposeful) thought may be ex-
stances of categories without first-letter cues pended on the word lists.
(Johnson, Raye, Foley, & Foley, in press, Similarly, daydreams often seem less real
Experiment 1). In Johnson et al.'s Experi- than dreams and rarely lead to the momen-
ment 2, increasing the difficulty of gener- tary confusion that dreams produce. Day-
ating a response and thus presumably in- dreams are typically more controlled by con-
creasing cognitive operations information by sciously selected schemata than are dreams
requiring a less typical response (e.g., ani- during sleep. In general, those self-generated
mal-p-, rather than animal-d-) increased memories that do not have a voluntary qual-
correct identification of origin. These find- ity but rather have been elicited by external
ings support the notion that cognitive oper- events or generated when voluntary pro-
ations information is important in reality cesses are suspended should be susceptible
monitoring decisions. to greater confusion. Creating a thought un-
The cue value of remembered cognitive der voluntary control (or bringing a previ-
operations may explain why mediators gen- ously involuntary idea under voluntary con-
erated by subjects in list learning situations trol) should provide a powerful cue for
are rarely intruded. That is, there is ample localizing it as self-generated rather than as
evidence that in paired-associate learning, existing externally. Consistent with this
for example, subjects generate verbal or pic- speculation is at least one finding: Whereas
torial elaborators (e.g., Martin, Boersma, repeated re-representations of perceived
& Cox, 1965; Yuille, 1973) and that specific items increase confusion (e.g., Johnson, Raye,
qualitative characteristics of these elabora- Wang, & Taylor, 1979; Johnson, Taylor,
tors influence recall (Hasher, Griffin, & & Raye, 1977), generating an initially self-
Johnson, 1977; Hasher & Johnson, 1975). generated item for a second time increased
In addition, self-generated information has correct identification of origin (Johnson,
a recall advantage (Slamecka & Graf, 1978), Raye, & Durso, 1980).
even when the self-generated information Contextual information. In one experi-
occurs only covertly (Johnson, Raye, Foley, ment (Johnson, Raye, Foley, & Kim, Note
& Foley, in press, Experiment 1). Why is 6), we compared the spatial location infor-
it then that intrusions of the mediator or self- mation associated with perceived pictures
generated components of an image or other with the spatial location information asso-
types of elaborators are so rare? From the ciated with imagined pictures. Subjects saw
present perspective, it is important that when a label followed by an appropriate line draw-
subjects create verbal or imaginal elabora- ing projected on a screen to the left or a
tors, the strategy is usually purposefully em- screen to the right, or they saw a label and
ployed and not the result of automatic or then themselves created an appropriate im-
incidental processes (Brown, 1975; Hasher age projected to the left or to the right. As
& Zacks, 1979; Posner & Snyder, 1975; a cover task, subjects rated the pictures and
Schneider & Shiffrin, 1977). The best ar- images with respect to how easy they would
gument for the strategic as opposed to au- be to draw. This task produced equal rec-
tomatic nature of elaborators is that it is ognition of items that had been presented as
quite easy to improve people's performance pictures and items that had been imagined
by instructing them to make elaborators (one indication of the effectiveness of this
(e.g., Bower, 1972b; Paivio, 1971). cover task). Of primary interest was the
On the other hand, those elaborations that finding that, as predicted by the model, sub-
more automatically arise on the basis of past jects were significantly more accurate in des-
experience, such as those involved in pro- ignating the location for perceived as com-
cessing meaningful prose, should be more pared to imagined items. In addition, for
78 MARCIA K. JOHNSON AND CAROL L. RAYE

those items whose location was correctly des- the harshness of the perceived or imagined
ignated, subjects were more confident about sound.
the location of the perceived pictures. We asked a similar question with a pro-
We adapted a similar procedure to ad- cedure in which subjects saw or heard items
dress the question of whether temporal in- on some trials or imagined seeing or imag-
formation is better for internally or exter- ined hearing them on other trials. The cover
nally generated information. Subjects heard task required subjects to rate the angularity
labels and then either saw an appropriate of the visual display of a perceived or imag-
line drawing or imagined one. The cover task ined word or the harshness of the sound of
was the rating task just described. Following an auditorially perceived or imagined item.
this, each subject received a booklet, and on Across four experiments, two requiring a
each page, in random order, were labels for discrimination between male and female
eight items, each drawn from a different voices and two between visual and auditory
eighth of the list. The eight items either all information, we have found a consistent, but
referred to pictures or all referred to images. not always significant, advantage for per-
The subject's task was to indicate the tem- ceived information. We think these cover
poral order of occurrence of the eight items. tasks very closely equated the information
We used this procedure to eliminate differ- produced during perception and thought,
ential response bias across positions of the primarily by restricting it to a single type of
list (Toglia & Kimble, 1976). This task sensory information, and were thus a con-
turned out to be quite difficult, but the sub- servative test of the hypothesis. The subject-
jects were able to identify the temporal po- report data described below suggest that
sition of very early and very late items. For when semantic processing is manipulated
items presented early, performance did not and sensory information is left free to vary,
differ on imagined as compared to perceived sensory information is more frequently re-
items, whereas the subjects were signifi- ported for perceived than for subject-gen-
cantly less accurate in identifying temporal erated items.
order of imagined items for the most recent Of course, the model does not demand
portion of the list. We are currently follow- that sensory information always be better for
ing this up with an easier task; however, this memories created by perception than for
initial finding is particularly interesting be- those created by thought; it only indicates
cause some investigators (Bjork & Whitten, that this is the more typical case. The very
1974; Crowder, 1976, pp. 461-464) have capacity for some thoughts to have more
proposed that the recency portion of the se- sensory detail than some perceptions creates
rial position curve especially reflects the con- one of the conditions for confusion. Fur-
tribution of temporal information. thermore, with the right orienting task, spa-
Sensory information. The model also tial location and temporal information might
predicts that sensory information should be be equated for imagined and perceived items.
greater for perceived compared to imagined One problem for future work would be to
items. In one experiment (Johnson, Raye, attempt to characterize those situations in
Foley, & Foley, Note 7), subjects saw a word which contextual, semantic, or sensory in-
that was then pronounced by a male or fe- formation from imagination is likely to be
male experimenter or saw a word and then unusually high, or from perception, unusu-
imagined it as pronounced by the male or ally low.
female experimenter. Craik and Kirsner For example, the amount of sensory in-
(1974) had previously shown significant formation in an imaginary experience should
memory for voice of perceived items, and vary among individuals. Good imagers, com-
Geiselman and Glenny (1977) had shown pared to poor imagers, should be more able
memory for the "voice" of an imagined item. to regenerate accurately the details of their
Our interest was in comparing the infor- experiences. Although it may be difficult to
mation about voice for perceived and imag- specify exactly the differences in the nature
ined items. As a cover task, subjects rated of the representations of good and poor im-
 REALITY MONITORING 79

agers (Pylyshyn, 1973), a likely difference the subjects were asked to indicate the orig-
is in the amount of sensory detail that is inal mode of presentation of the items they
represented in the images they create. We identified as old. The orienting task at ac-
might then expect that good imagers would quisition was manipulated for different
more readily confuse re-representations of groups of subjects. Some subjects were asked
experiences with similar perceptions. In one to perform tasks that required that they at-
experiment (Johnson, Raye, Wang, & Tay- tend to the formal properties of the word
lor, 1979, Experiment 1; Johnson & Raye, symbol (what is the last letter of the name
Note 8), we separated subjects into groups of this concept?). Other subjects were asked
of good and poor imagers on the basis of an to perform tasks that explicitly required that
imagery test. Using our re-representation they create an image of the concept (How
paradigm, subjects saw pictures of common long would it take for an artist to make a
objects taken from children's books on pre- line drawing of this concept?). And some
sentation trials and were asked to generate subjects were asked to perform tasks that
images of the pictures on generation trials. required them to consult semantic infor-
On a later test of presentation frequency, the mation regarding the referent of the concept
judgments of good imagers were more af- (e.g., What is this thing used for?). The most
fected by the number of times they had interesting data for the present purposes are
imagined a picture than were the judgments the confusions between the two modes of
of poor imagers. (Good and poor imagers did representations, that is, the mistaken judg-
not differ in their ability to recall the set of ments that words were presented as pictures
pictures). or that pictures were presented as words.
Although it simplifies the discussion to The relative frequency of these two types of
look at one type of information at a time, confusions did not differ for either the ori-
we should keep in mind that all memories enting tasks involving the formal properties
include several types of information and that of words or those involving explicit imagery.
any reality monitoring decision is the result However, for the semantic conditions, sub-
of processes that weight and integrate in- jects were significantly more likely to say a
formation across these dimensions. The pre- word had been presented as a picture than
viously discussed Raye et al. study (1980, to say a picture had been presented as a
Experiment 2), which compared frequency word. This finding is all the more striking
judgments with and without a limit, dem- because there was clearly a significant bias
onstrated that subjects can change the on false positives (calling a completely new
weighting given to various dimensions. Which item old) to say they were presented as words
components of memories will weigh more rather than as pictures. The high rate of
heavily in reality monitoring judgments may misidentifications of words as pictures by
depend on the nature of the events being those subjects who did semantic processing
judged. For example, cognitive operations suggests they had been incidentally creating
may be particularly important in decisions images or consulting sensory information
about language, especially words and/or during the orienting tasks and that they later
sentences given in isolation. Sensory infor- mistook this self-generated pictorial infor-
mation may prove to be the more dominant mation for externally presented pictorial in-
component in decisions about the origin of formation.
visual information such as pictures. Memories for words created in the explicit
Furthermore, that decisions are not simply imagery conditions should also have had
made on the basis of one type of information, considerable sensory information; however,
without regard for other attributes, is illus- the memories also very likely included cues
trated by some results reported by Durso and indicating that the image was purposefully
Johnson (1980). Subjects saw a concept rep- created. On the other hand, the incidentally
resented by a word or a line drawing, and created images, which had less salient cues
on a later recognition test (where both tar- that they were generated by cognitive op-
gets and distractors were given auditorially), erations yet which had high sensory infer-
80 MARCIA K. JOHNSON AND CAROL L. RAYE

mation, should be more likely to be mistaken to perceptual events, the likelihood of later
for externally presented pictures. Thus, here confusing their memory representations in-
is a case where it makes sense to assume that creases. However, even though the related
decisions about the origin of memories took subjects showed the least ability to identify
into account both sensory and cognitive op- the origin of an item, they were still signif-
erations information. icantly above a chance score of 50%.
Following the identification-of-origin test,
Metamemory Assumptions subjects in this study were asked to describe
how they were able to identify the source of
An important point incorporated in the various items. Subjects tended to mention
working model of reality monitoring that sensory cues more often with respect to ex-
appears to be born out by our experiments perimenter-generated items ("I differen-
is the idea that subjects' assumptions about tiated words which you said by remembering
how their memories work will play a critical your pronunciation"; "I could visualize your
role in decision strategies and biases oper- saying it"; or "The words which the exper-
ating during reality monitoring. imenter stated were remembered in her
What subjects report. The good-poor voice"). Cognitive processing, additional in-
imager study above attempted to manipulate formation, and a consideration of specific
the similarity between perceptions and re- semantic content were mentioned more in
representations of those perceptions. We conjunction with subject-generated items
have also attempted to vary the relationship ("When I was very sure [about my words]
between perceptions and cotemporal, self- I could remember I had a very specific rea-
generated information (Johnson, Raye, son for making the association. If the word
Foley, & Foley, in press, Experiment 3). An [only] seemed familiar, I would say that it
experimenter read a list of common, unre- was the experimenter's word"; "I made the
lated words. Subjects in the related condi- decision by knowing what my train of thought
tion were told to generate a word for each was during the exercise"; "Sometimes the
stimulus word that had some meaningful words I chose went together with a certain
relationship to the stimulus word. Subjects scene, i.e., pond, cloud, tree.'And when I saw
in the unrelated condition were told to give the words again I tried to remember if they
a word that was not in any way related to fit in any of the images I had").
the stimulus item. Subjects in the first-letter Biases in judgments. If you ask them,
condition were told to give a word that many subjects expect self-generated infor-
started with the same first letter as the word mation to have an advantage in memory
they heard. Thus in the related and unre- (Johnson, Raye, Foley, & Foley, in press,
lated conditions subjects gave some attention Experiment 2). This expectation is consis-
to the meaning of each externally presented tent with a notable bias we found: When
item in order to perform the task, but the subjects felt that a completely new item was
internally generated product should have familiar (a false positive), they displayed a
been different in the two cases. The related much greater willingness to attribute it to
and the first-letter conditions should both an external source than to say they gener-
produce related items, but these relation- ated it. We found this "it-had-to-be-you"
ships should primarily be semantic in the ("I'd remember if it were me") effect with
related group and orthographic in the first- the three different instructions for generat-
letter condition. All subjects wrote down ing responses described above (related, un-
both presented and generated words. related, first letter), for high and low fre-
A week later, subjects received an iden- quency category instances and opposites
tification-of-origin test. For those words cor- (Johnson, Raye, Foley, & Foley, in press
rectly recognized as old, the mean percents Experiment 2) and for identification of ori-
correct identification of the source of an item gin of words from sentences (Johnson, Raye,
were 62, 68, and 74, for related, first-letter, & Durso, 1980). The pervasiveness of this
and unrelated, respectively, indicating that bias suggests that it is probably an important
as cotemporal thoughts become more related component of performance in memory tasks,
 REALITY MONITORING 81

contributing, perhaps, to false positives in affected similarly. For example, directors,

certain recognition paradigms (e.g., Cramer, who asked the questions, were as good as
1965; Johnson, Bransford, & Solomon, 1973; speakers in later identifying answers that
Kimble, 1968; Sulin & Dooling, 1974; Un- had been given but far worse at identifying
derwood, 1965) and unedited intrusions in who said what (Raye & Johnson, 1980, Ex-
recall (Brewer, 1977; Deese, 1959). Under periment 2). And 6-year-olds who said some
some circumstances, subjects can be induced words and thought others were as good at
with a stricter criterion to perform in a way identifying which words had occurred as
that more closely matches real external were 6-year-olds who had listened to two
events (Raye et al., 1980). An investigation speakers say the words; however, the chil-
of the circumstances that might affect this dren found it much more difficult to dis-
bias could prove interesting. For example, criminate what they had said from what they
we have recently found that alcohol intoxi- had thought than to discriminate between
cation reduces (but does not eliminate) this what the two speakers had said (Foley et al.,
bias, reflecting perhaps people's belief that Note 5).
one consequence of drinking is that it di- In some cases, memory for occurrence and
minishes one's ability to remember what one identification of origin appear actually to be
did (Taylor et al., Note 4). Of course, the affected in opposite ways by the same vari-
bias might be completely reversed if the so- able. For example, recall and recognition
cial desirability of having been the source were both better when subjects generated
of an idea were markedly increased ("I don't semantically related responses than when
remember generating it, but that idea is so they generated responses sharing the same
good it has to be mine"). first letter as the stimuli. However, discrim-
inating between items heard and items gen-
Remembering Occurrence Versus erated was more difficult in the related case
Remembering Origin (Johnson, Raye, Foley, & Foley, in press).
In the experiments in which Durso and
Do we need a model of reality monitoring Johnson (1980) compared memory for pic-
that is distinct from the theories of memory tures and words under different orienting
that have been proposed to account for rec- tasks, attending to the referent for a word
ognition and recall in general? For example, rather than properties of the verbal name of
could we simply say that "deep" processing the concept resulted in higher recognition,
of information leads to better memory, in- but it also resulted in more words that were
cluding better identification of origin? A later identified as having been pictures. With
strength theorist might propose that identi- the explicit imagery tasks, there was also an
fication of origin simply requires more of the increase in recognition relative to verbal
same sort of information that recall and rec- tasks, but this was not accompanied by an
ognition depend on. However, attempts to increase in identifying words as pictures. In
characterize memory functioning in terms summary, it does not appear to us that the
of a single dimension have always run into information and processes that are impor-
difficulties. For example, in the present con- tant in discriminating the origin of infor-
text, both strength and levels theories would mation can be equated with those that pro-
expect a consistent correlation between duce voluntary recall or a feeling of
memory for occurrence measures (recall familiarity.
or recognition) and identification-of-origin The factor that probably most distin-
measures. Memory for occurrence and mem- guishes the process of voluntary recall from
ory for origin sometimes do respond to the recognition is the dependence of recall on
same variables. For example, both recogni- organizational information (Mandler, 1967;
tion and identification of origin were better Tulving, 1968). In contrast, recognition is
for low as compared to high frequency cat- notable for the variety of types of informa-
egory instances (Johnson, Raye, Foley, & tion that can produce a "familiarity" re-
Foley, in press, Experiment 2). However, sponse (e.g., rapid activation of semantic or
these two measures are by no means always sensory information, cognitive operations in-
82 MARCIA K. JOHNSON AND CAROL L. RAYE

formation, context cues). Reality monitor- knowledge available to the person, as well
ing, as a process, has its own properties, al- as metamemory assumptions of the person.
though to a certain extent it of course draws Overall, our results support these ideas:
on information that is operative in recall and Memories from external and internal sources
recognition. (For example, the potential use- appear to differ in class-characteristic ways;
fulness in recognition of cognitive opera- confusion is increased by semantic and sen-
tions, especially, has received recent atten- sory similarity between memories from the
tion, e.g., Kolers, 1975; Russo & Wisher, two sources; confusion is reduced with in-
1976.) Although a model of reality moni- creases in the information about cognitive
toring does not offer a theory of recall or operations associated with internally gen-
recognition, neither do theories of recall or erated memories; memories based in percep-
recognition constitute theories of reality tion have better spatial, temporal, and
monitoring. Furthermore, understanding sensory information; and people's tacit
reality monitoring can contribute to our un- assumptions about these characteristic dif-
derstanding of recall and recognition pro- ferences are reflected in metamemory as-
cesses, for example, by providing mecha- sumptions that influence reality monitoring
nisms for editing out inappropriate responses judgments.
and by accounting for failures in editing. For the most part, the materials and pro-
cedures we have used seem to be extremely
conducive to confusing external and internal
Summary and Conclusions sources of memories. The words, sentences,
and pictures were not, after all, embedded
We have intended to focus attention on in social or emotional contexts that might
the processes involved in distinguishing be- provide cues from related information stored
tween the perceived and the self-generated in memory about the origin of a particular
in memory (reality monitoring). We started trace. Given the seemingly meager cues, it
with the assumption that perceiving differs is perhaps remarkable that significant reality
from self-generated events such as thinking monitoring was obtained in some conditions
and imagining. That is, they are either dif- with retention intervals as long as 7 and 10
ferent operations of the mind or operations days. This sensitivity to the origin of infor-
of the mind resulting in characteristically mation, when a discrimination is called for,
different amounts of various types of infor- further recommends the usefulness of mak-
mation. Furthermore, we believe that the ing a distinction between external and in-
processes involved in reality monitoring, ternal sources of information and the use-
when they are functioning optimally, capi- fulness of attempting to incorporate such a
talize on these characteristic differences. We distinction into models of memory.
have outlined a working model of reality The overall picture suggested by the evi-
monitoring and have investigated some of dence reviewed here is one of a memory sys-
the questions it raises. The present model tem that preserves information about the
proposes that reality monitoring can be origin of information remarkably well; the
based on some general dimensions—rather decision criteria through which this infor-
independent of the particular content of a mation is filtered, however, allow for some
memory—that serve as distinguishers be- error in attributing memories to sources. We
tween the two classes of memories. It is pro- feel that the working model proposed here
posed that dimensions or attributes partic- is a viable framework for generating ques-
ularly central to reality monitoring include tions and integrating results of studies di-
amount of contextual information (time and rected at the processes involved in reality
place), semantic detail, sensory information, monitoring.
and cognitive operations. The model addi-
tionally assumes that reality monitoring may
involve more extended reasoning processes Reference Notes
that take into account the content of partic- 1. Johnson, M. K., & Raye, C. L. A working model of
ular memories and their relation to other reality monitoring. Paper presented at the annual
 REALITY MONITORING 83

meeting of the Psychonomic Society, San Antonio, uisites for understanding: Some investigations of com-
Texas, November 1978. prehension and recall. Journal of Verbal Learning
2. Levine, M., & Jankovic, I. N. An introduction to and Verbal Behavior, 1972, 11, 717-726.
spatial problem solving. Manuscript submitted for Bransford, J. D., & Johnson, M. K. Considerations of
publication. some problems of comprehension. In W. Chase (Ed.),
3. Johnson, M. K. Constructive aspects of memory: Visual information processing. New York: Academic
Historical antecedents. Paper presented at the sym- Press, 1973.
posium on constructive processes in memory, annual Brewer, W. F. Memory for pragmatic implications of
meeting of the American Psychological Association, sentences. Memory & Cognition, 1977, 5, 673-678.
Chicago, September 1975. Brooks, L. R. Spatial and verbal components of the act
4. Taylor, T. H., Johnson, M. K., Birnbaum, I. M., of recall. Canadian Journal of Psychology, 1968, 22,
& Raye, C. L. Alcohol and reality monitoring. Man- 349-368.
uscript submitted for publication. Brown, A. L. The development of memory: Knowing
5. Foley, M. A., Johnson, M. K., & Raye, C. L. A about knowing, and knowing how to know. In H. W.
developmental study of reality monitoring, compar- Reese (Ed.), Advances in child development and be-
ing two types of judgments about memories. Paper havior (Vol. 10). New York: Academic Press, 1975.
presented at the annual meeting of the Eastern Psy- Bruner, J. S. Going beyond the information given. In
chological Association, Hartford, Connecticut, April Contemporary approaches to cognition. Cambridge,
1980. Mass.: Harvard University Press, 1957.
6. Johnson, M. K., Raye, C. L., Foley, M. A., & Kim,
J. Pictures and images: Spatial and temporal infor- Cameron, N. Personality development and psychopa-
mation compared. Manuscript in preparation, 1980. thology: A dynamic approach. Boston: Houghton
Mifflin, 1963.
7. Johnson, M. K., Raye, C. L., Foley, M. A., & Foley,
H. J. Perceptions and imaginations: sensory infor- Cofer, C. N. Constructive processes in memory. Amer-
mation compared. Manuscript in preparation, 1980. ican Scientist, 1973, 61, 537-543.
8. Johnson, M. K., & Raye, C. L. Fact and fantasy: Cole, W. G., & Loftus, E. F. Incorporating new infor-
Confusion between perceptual experience and imag- mation into memory. American Journal of Psychol-
inations. Paper presented at the annual meeting of ogy, 1979, 92, 413-425.
the Psychonomic Society, St. Louis, Missouri, No- Corbett, A. T., & Dosher, B. A. Instrument inferences
vember 1976. in sentence encoding. Journal of Verbal Learning and
Verbal Behavior, 1978, 17, 479-491.
Craik, F. I. M., & Kirsner, K. The effect of speaker's
References voice on word recognition. Quarterly Journal of Ex-
perimental Psychology, 1974, 26, 274-284.
Atkinson, R. C., & Juola, J. F. Factors influencing speed Craik, F. I. M., & Lockhart, R. S. Levels of processing:
and accuracy of word recognition. In S. Kornblum A framework for memory research. Journal of Verbal
(Ed.), Attention and performance, IV. New York: Learning and Verbal Behavior, 1972, / / , 671-684.
Academic Press, 1973. Cramer, P. Recovery of a discrete memory. Journal of
Atwood, G. An experimental study of visual imagination Personality and Social Psychology, 1965, /, 326-332.
and memory. Cognitive Psychology, 1971, 2, 290- Crowder, R. G. Principles of Learning and memory.
299. Hillsdale, N.J.: Erlbaum, 1976.
Bartlett, F. C. Remembering: A study in experimental
Davies, G. M., Milnee, J. E., & Glennie, B. S. On the
and social psychology. Cambridge, Eng.: Cambridge
significance of "double encoding" for the superior re-
University Press, 1932. call of pictures to names. Quarterly Journal of Ex-
Bjork, R. A., & Whitten, W. B. Recency-sensitive re- perimental Psychology, 1973, 25, 413-423.
trieval processes in long-term free recall. Cognitive
Psychology, 1974, 6, 173-189. Deese, J. On the prediction of occurrence of particular
Bobrow, S. A., & Bower, G. H. Comprehension and verbal intrusions in immediate recall. Journal of Ex-
recall of sentences. Journal of Experimental Psy- perimental Psychology, 1959, 58, 17-22.
chology, 1969, 80, 455-461. Dooling, D. J., & Lachman, R. Effects of comprehen-
Bower, G. H. A multi-component theory of the memory sion on retention of prose. Journal of Experimental
trace. In K. W. Spence & J. T. Spence (Eds.), The Psychology, 1971, 88, 216-222.
psychology of learning and motivation (Vol. 1). New Durso, F. T., & Johnson, M. K. The effects of orienting
York: Academic Press, 1967. tasks on recognition, recall, and modality confusion
Bower, G. H. A selective review of organizational factors of pictures and words. Journal of Verbal Learning
in memory. In E. Tulving & W. Donaldson, (Eds.), and Verbal Behavior, 1980, 19, 416-429.
Organization of memory. New York: Academic Erdelyi, M., Buschke, H., & Finkelstein, S. Hyper-
Press, 1972. (a) amnesia for socratic stimuli: The growth of recall for
Bower, G. H. Mental imagery and associative learning. internally generated memory list abstracted from a
In L. W. Gregg (Ed.), Cognition in learning and series of riddles. Memory & Cognition, 1977, 5, 283-
memory. New York: Wiley, 1972. (b) 286.
Bransford, J. D., & Franks, J. J. The abstraction of Finke, R. A. The functional equivalence of mental im-
linguistic ideas. Cognitive Psychology, 1971, 2, 331- ages and errors of movement. Cognitive Psychology,
350. 1979, //, 235-264.
Bransford, J. D., & Johnson, M. K. Contextual prereq- Freud, S. Negation (Standard Ed., Vol. 19. James
84 MARCIA K. JOHNSON AND CAROL L. RAYE

Strachey, Ed.). Toronto, Canada: Clarke, Irwin, W. Are there developmental differences in reality
1961. (Originally published, 1925.) monitoring? Journal of Experimental Child Psy-
Freud, S. Project for a scientific psychology (Standard chology, 1979, 27, 120-128.
Ed., Vol. 1. James Strachey, Ed.). Toronto, Canada: Johnson, M. K., Raye, C. L., Wang, A. Y., & Taylor,
Clarke, Irwin, 1966. (Originally published, 1895.) T. H. Fact and fantasy: The roles of accuracy and
Garrity, L. I. Electromyography: A review of the current variability in confusing imaginations with perceptual
status of subvocal speech research. Memory & Cog- experiences. Journal of Experimental Psychology:
nition, 1977, 5, 615-622. Human Learning and Memory, 1979, 5, 229-240.
Geiselman, R. E., & Glenny, J. Effects of imagining Johnson, M. K., Taylor, T. H., & Raye, C. L. Fact and
speakers' voices on the retention of words presented fantasy: The effects of internally generated events on
visually. Memory & Cognition, 1977, 5, 499-504. the apparent frequency of externally generated events.
Gibson, J. J. The senses considered as perceptual sys- Memory & Cognition, 1977, 5, 116-122.
tems. Boston: Houghton Mifflin, 1966. Kimble, G. A. Mediating associations. Journal of Ex-
Gibson, J. J. The ecological approach to visual percep- perimental Psychology, 1968, 76, 263-266.
tion. Boston: Houghton Mifflin, 1979. Kintsch, W. Memory for prose. In C. N. Cofer (Ed.),
Hart, J. T. Memory and the memory-monitoring pro- The structure of human memory. San Francisco:
cess. Journal of Verbal Learning and Verbal Behav- Freeman, 1976.
ior, 1967, 6, 685-691. Klatzky, R. L. Human memory: Structure and pro-
Hasher, L., & Griffin, M. Reconstructive and repro- cesses. San Francisco: Freeman, 1980.
ductive processes in memory. Journal of Experimen- Kohler, W. Gestalt psychology. New York: Mentor
tal Psychology, 1978, 4, 318-330. Books, 1959. (Originally published, 1947.)
Hasher, L., Griffin, M., & Johnson, M. K. More on Kolers, P. A. Specificity of operations in sentence rec-
interpretive factors in forgetting. Memory & Cogni- ognition. Cognitive Psychology, 1975, 7, 289-306.
tion, 1977, 5, 41-45. Kosslyn, S. M. Can imagery be distinguished from other
Hasher, L., & Johnson, M. K. Interpretive factors in forms of internal representation? Evidence from stud-
forgetting. Journal of Experimental Psychology: ies of information retrieval times. Memory & Cog-
Human Learning and Memory, 1975, /, 567-575. nition, 1976, 4, 291-297.
Hasher, L., & Zacks, R. T. Automatic and effortful Krech, D. Notes toward a psychological theory. Journal
processes in memory. Journal of Experimental Psy- of Personality, 1949, 18, 66-87.
chology: General, 1979, 108, 356-388. Kreutzer, M. A., Leonard, C., & Flavell, J. H. An in-
Heidbreder, E. Seven psychologies. New York: Apple- terview study of children's knowledge about memory.
ton-Century-Crofts, 1933. Monographs of the Society for Research in Child
Hochberg, J. Perception. Englewood Cliffs, N.J.: Pren- Development, 1975, 40 (1, Serial No. 159).
tice-Hall, 1979. Landauer, T. K. Memory without organization: Prop-
Horowitz, M. J. Image formation and cognition. New erties of a model with random storage and undirected
York: Appleton-Century-Crofts, 1978. retrieval. Cognitive Psychology, 1975, 7, 495-531.
Hume, D. A treatise of human nature. Oxford, Eng.: Lashley, K. S. The problem of serial order in behavior.
Clarendon Press, 1978. (Originally published, 1739.) In N. J. Slamecka (Ed.), Human learning and mem-
Jacoby, L. L. On interpreting the effects of repetition: ory. New York: Oxford University Press, 1967.
Solving a problem versus remembering a solution. Mandler, G. Organization and memory. In K. W.
Journal of Verbal Learning and Verbal Behavior, Spence & J. T. Spence (Eds.), The psychology of
1978, 17, 649-667. learning and motivation (Vol. 1). New York: Aca-
James, W. The principles of psychology. New York: demic Press, 1967.
Dover, 1950. (Originally published, 1890.) Mandler, G. Organization and recognition. In E. Tulv-
Jarvella, R. J., & Collas, J. G. Memory for the inten- ing & W. Donaldson (Eds.), Organization of memory.
tions of sentences. Memory & Cognition, 1974, 2, New York: Academic Press, 1972.
185-188. Martin, C. J., Boersma, F. J., & Cox, D. L. A classi-
Jenkins, J. J. Remember that old theory of memory? fication of associative strategies in paired-associate
Well, forget it! American Psychologist, November learning. Psychonomic Science, 1965, 3, 455-456.
1974, pp. 785-795. McGuigan, F. J. Covert oral behavior and auditory hal-
Johnson, M. K. What is being counted none the less? lucinations. Psychophysiology, 1966, 3(1), 73-80.
In I. M. Birnbaum & E. S. Parker (Eds.), Alcohol Mellor, C. S. First rank symptoms of schizophrenia. I:
and human memory. Hillsdale, N.J.: Erlbaum, 1977. The frequency in schizophrenics on admission to hos-
Johnson, M. K., Bransford, J. D., & Solomon, S. K. pital. II: Differences between individual first rank
Memory for tacit implications of sentences. Journal symptoms. British Journal of Psychiatry, 1970, 117,
of Experimental Psychology, 1973, 98, 203-205. 15-23.
Johnson, M. K., Raye, C. L., & Durso, F. T. Reality Mill, J. Analysis of the phenomena of the human mind,
monitoring: Second perceptions and thoughts. Bul- by James Mill (edited with additional notes by J. S.
letin of the Psychonomic Society, 1980, 15, 402-404. Mill). New York: Kelley, 1967. (Originally published,
Johnson, M. K., Raye, C. L., Foley, H. J., & Foley, 1829.)
M. A. Cognitive operations and decision bias in reality Mill, J. S. Analysis of the phenomena of the human
monitoring. American Journal of Psychology, in mind, by James Mill (edited with additional notes by
press. J. S. Mill). New York: Kelley, 1967. (Originally pub-
Johnson, M. K., Raye, C. L., Hasher, L., & Chromiak, lished, 1869.)
 REALITY MONITORING 85

Mintz, S., & Alpert, M. Imagery vividness, reality test- Russo, J. E., & Wisher, R. A. Reprocessing as a rec-
ing, and schizophrenic hallucinations. Journal of Ab- ognition cue. Memory & Cognition, 1976, 4, 683-
normal Psychology, 1972, 79, 310-316. 689.
Neisser, U. Cognitive Psychology. New York: Apple- Sachs, J. S. Recognition memory for syntactic and se-
ton-Century-Crofts, 1967. mantic aspects of connected discourse. Perception
Neisser, U. Cognition and reality. San Francisco: Free- & Psychophysics, 1967, 2, 437-442.
man, 1976. Schneider, W., & Shiffrin, R. M. Controlled and au-
Paivio, A. Mental imagery in associative learning and tomatic human information processing: I. Detection,
memory. Psychological Review, 1969, 76, 241-263. search, and attention. Psychological Review, 1977,
Paivio, A. Imagery and verbal processes. New York: 84, 1-66.
Holt, Rinehart & Winston, 1971. Segal, S. J., & Fusella, V. Influence of imaged pictures
Perky, C. W. An experimental study of imagination. and sounds on detection of visual and auditory signals.
American Journal of Psychology, 1910, 21, 422-452. Journal of Experimental Psychology, 1970, 83, 458-
Posner, M. I., & Snyder, C. R. R. Attention and cog- 464.
nitive control. In R. L. Solso (Ed.), Information pro- Slamecka, N. J., & Graf, P. The generation effect:
cessing and cognition: The Loyola Symposium. Hills- Delineation of a phenomenon. Journal of Experi-
dale, N.J.: Erlbaum, 1975. mental Psychology: Human Learning and Memory,
Posner, M. I., & Warren, R. E. Traces, concepts and 1978,4,592-604.
conscious construction. In A. W. Melton & E. Martin Sulin, R. A., & Dooling, D. J. Intrusion of a thematic
(Eds.), Coding processes in human memory. Wash- idea in retention of prose. Journal of Experimental
ington, D.C.: Winston, 1972. Psychology, 1974, 103, 255-262.
Postman, L. Toward a general theory of cognition. In Toglia, M. P., & Kimble, G. A. Recall and use of serial
J. H. Rohrer & M. Sherif (Eds.), Social psychology position information. Journal of Experimental Psy-
at the crossroads. New York: Harper, 1951. chology: Human Learning and Memory, 1976, 2,
Pylyshyn, Z. W. What the mind's eye tells the mind's 431-445.
brain: A critique of mental imagery. Psychological Tolman, E. C. Cognitive maps in rats and men. Psy-
Bulletin, 1973, 80, 1-23. chological Review, 1948, 55, 189-208.
Raye, C. L. Recognition: Frequency or organization? Tulving, E. Theoretical issues in free recall. In T. R.
American Journal of Psychology, 1976, 89, 645-658. Dixon & D. L. Horton (Eds.), Verbal behavior and
general behavior theory. Englewood Cliffs, N.J.:
Raye, C. L., & Johnson, M. K. Reality monitoring vs.
Prentice-Hall, 1968.
discrimination between external sources. Bulletin of Underwood, B. J. False recognition produced by implicit
the Psychonomic Society, 1980, 15, 405-408.
verbal responses. Journal of Experimental Psychol-
Raye, C. L., Johnson, M. K., & Taylor, T. H. Is there ogy, 1965,70, 122-129.
something special about memory for internally-gen- Underwood, B. J. Attributes of memory. Psychological
erated information? Memory & Cognition, 1980, 8, Review, 1969, 76, 559-573.
141-148. Wickens, D. D. Encoding categories of words: An em-
Reid, T. An inquiry into the human mind, on the prin- pirical approach to meaning. Psychological Review,
ciples of common sense. In K. Lehrer & R. E. Bean- 1970, 77, 1-15.
blossom (Eds.), Thomas Reid's inquiry and essays. Yuille, J. C. A detailed examination of mediation in
Indianapolis, Ind.: Bobbs-Merrill, 1975. (Originally paired-associate learning. Memory & Cognition, 1973,
published, 1764.) /, 333-342.
Ross, M., & Sicoly, F. Egocentric biases in availability
and attribution. Journal of Personality and Social
Psychology, 1979, 37, 322-336. Received February 29, 1980 •