For more information, see: Research on Nonconsciousness.
Pawel Lewicki (University of Tulsa),
Thomas Hill (University of Tulsa), and
Maria Czyzewska (Southwest Texas State University)
We are reviewing and summarizing evidence for the processes of acquisition of information outside of conscious awareness (processing information about covariations, nonconscious indirect and interactive inferences, self-perpetuation of procedural knowledge). A considerable amount of data indicates that as compared to consciously controlled cognition, the nonconscious information-acquisition processes are not only much faster but also structurally more sophisticated in the sense that they are capable of efficient processing of multidimensional and interactive relations between variables. Those mechanisms of nonconscious acquisition of information provide a major channel for the development of procedural knowledge which is indispensable for such important aspects of cognitive functioning as encoding and interpretation of stimuli and the triggering emotional reactions.
In this article we are reviewing and summarizing empirical evidence for the processes of acquisition of information outside of conscious awareness. We are focusing on the most common in every-day life (and ubiquitous in human cognition) forms of nonconscious learning where the acquired information is not accessible to the perceiver's conscious control not because of the physical properties of the stimuli (such as subliminal exposure time) but due to the relative slowness and inefficiency of the human consciousness. A considerable amount of evidence indicates that as compared to consciously controlled cognition, the nonconscious information-acquisition processes are incomparably faster and structurally more sophisticated. They allow for the development of procedural knowledge that is "unknown" to conscious awareness not merely because it has been encoded (and entered the memory system) through channels that are independent from consciousness. This knowledge is fundamentally inaccessible to the consciousness because it involves a more advanced and structurally more complex organization than what could be handled by consciously controlled thinking.
UBIQUITY OF NONCONSCIOUS ACQUISITION
OF KNOWLEDGE IN HUMAN COGNITION
Although this statement might appear to some to be somewhat surprising, the ability of the human cognitive system to nonconsciously acquire information is a general metatheoretical assumption of almost all of contemporary cognitive psychology. Moreover, this assumption is so necessary that it is enthymemically present in almost every piece of research on human information processing published over the past two decades and it is indirectly reflected in most of the experimental paradigms developed by cognitive psychologists.
In hardly any experimental procedure do cognitive researchers assume that they can directly learn how humans process information by simply "asking" them to report the contents of the procedural knowledge they follow. No matter how cooperative and well trained our subjects are, they cannot tell us how they go about processing information (e.g., how they encode shapes of objects in three-dimensional space, or how they generate esthetic judgments). This is because subjects not only do not know how they do all those things but have never known it, and they do not have the slightest idea of how they learned all those information-processing algorithms and heuristics that are involved in the cognitive "software" that is indispensable for their psychological functioning. At the same time, there is no doubt that most of this procedural knowledge and skills result from experience and thus that they have been learned at some point.
Obviously, these trivial facts are not entirely new, and they have been observed and stated (usually in a somewhat implicit manner) by a number of researchers since Helmholtz. These facts became apparent especially to those who investigated (or tried to simulate) the processes of perception and realized the enormous complexity of those inferential encoding algorithms and heuristics that are necessary for every perception, even simple ones (e.g., Hochberg, 1978; Rock, 1975). These skills do not result from automatization through conscious experience. The complexity of those processes constitutes probably the most direct argument for that the perceivers' inability to tell us anything about how those processes work represents a fundamental lack of access to the nature of those algorithms and heuristics and not merely difficulties with articulating the knowledge about them. For example, the seemingly simple act of recognizing a shape and size of an object and its location in three-dimensional space requires a set of sophisticated geometrical transformations and calculations that go far beyond what most perceivers could articulate or even comprehend.
The conscious appraisal of the final "products" of perception (i.e., subjectively encoded meanings of stimuli) is functionally independent from the information-processing algorithms and heuristics responsible for generating those subjective meanings. This fundamental independence is evident in virtually all areas of human cognition. Moreover, this lack of access to the nature of these processes (which are essentially responsible for most of what we see, experience, and feel) is not limited to the so-called low-level processes that support only the consciously controlled cognition (e.g., pattern recognition). People have no access to processes as high-level as those involved in playing chess (deGroot, 1965), feeling love (Walster & Walster, 1978), forming impressions of people (Lewicki, 1986a), or problem solving and creative thinking (Sternberg, 1986), and when researchers attempt to learn directly from subjects anything about how such judgments or decisions are generated, subjects are usually as helpless as when they are asked to explain how they identify right angles or recognize patterns; all they know is that they "just do it."
In light of those arguments, it is important to learn about the processes leading to the acquisition of procedural knowledge outside of conscious awareness because they contribute to the very foundations of the human cognitive system.
ACQUISITION OF INFORMATION ABOUT COVARIATIONS
A considerable amount of evidence indicates that the human cognitive system is capable of nonconsciously detecting and processing of information about covariations between features or events in the outside world (Hill, Lewicki, & Sasaki, 1989; Lewicki, 1986a; Lewicki, Czyzewska, & Hoffman, 1987; for a review, see Hill & Lewicki, 1991). Moreover, subjects' nonconscious ability to detect and process covariations was found to be superior to their (relatively poor) ability to detect the same information in a consciously controlled manner (see also Nisbett & Ross, 1980). Nonconscious processing occurs even if the conditions that are necessary for the consciously controlled processing of covariations (Crocker, 1981) are not met, for example, when the covariation is "hidden" (e.g., pertains to peripheral aspects of the stimulus material).
The nonconscious processing of covariations results in the development of respective procedural knowledge (Winograd, 1975) that participates in the encoding of subsequently encountered, relevant stimuli. For example, the nonconscious processing of a covariation between a facial feature x and a personality characteristic y results in the development of a tendency to interpret (encode) behaviors of subsequently encountered people who posses this feature (x) as indicative of personality characteristic y. This kind of procedural knowledge is referred to as encoding algorithms. The encoding algorithms provide elementary "inferential rules" used by the individual in the process of translating stimuli into subjectively meaningful experiences and converting them into memory-compatible code. Therefore, the nonconscious processing of information about covariations results in the development of the elementary functional components of the cognitive system which determine the way in which individuals interpret information, think, make judgments, form preferences, etc.
HOW "NONCONSCIOUS" IS THE NONCONSCIOUS
ACQUISITION OF INFORMATION?
Results from a variety of tests provide convergent evidence that subjects in the nonconscious covariation processing experiments have no access to the newly acquired procedural knowledge and they even have no idea that they have learned anything (even though the newly acquired knowledge consistently guides their behavior).
In one of the studies that addressed the issue of the relation between conscious and nonconscious knowledge, an unusual sample of subjects was selected to assure that they would be sufficiently cooperative and intellectually capable to report any potential introspective experience (of "acquiring new information") they could have during the experiment: All subjects were faculty members of a Psychology department (Lewicki, Hill, & Bizot, 1988). In that study, subjects nonconsciously acquired a set of encoding algorithms allowing them to more efficiently (faster) encode locations of a target on a computer screen, and thus to perform better in a search task. When the crucial covariation that was built into the sequence of target locations (on the CRT) was changed (i.e., became inconsistent with the previous algorithm), subjects' performance deteriorated (as measured by reaction times and the accuracy of responses). The subjects (psychology professors) knew that the study investigated nonconscious cognition and tried hard to "figure out" the experiment. However, none of them came even close to discovering the real nature of the manipulation. Debriefing indicated that none of the subjects had any clue as to what kind of knowledge they had nonconsciously acquired in the experiment. Moreover, it was revealed that although subjects noticed the sudden decrease of their performance (at the point when the covariation changed) they attributed the decrease to factors that were entirely unrelated to the real cause of these changes, e.g., they suspected that some distracting (e.g., threatening) subliminal stimuli were flashed on the screen.
In another experiment, subjects (college students) were given an unlimited amount of time and offered a large monetary reward ($100) to uncover the "hidden" pattern in the stimulus material, which they had learned nonconsciously before, as indicated by the predicted pattern of changes in their performance (Lewicki, Czyzewska, & Hoffman, 1987). Some participants spent many hours trying to find the clue; however, none of them managed to come up with any ideas even remotely relevant to the real nature of the manipulation.
The results of those studies suggest that perceivers' access to the encoding algorithms that were acquired nonconsciously is limited to experiencing only the final outcomes of the nonconscious encoding processes (e.g., increased performance, liking or disliking something). Moreover, when a newly (and nonconsciously) acquired encoding algorithm involved some belief that had been consciously held by the subjects, the (consciously held) belief was never found to be affected by the nonconscious process. For example, after acquiring a nonconscious tendency to perceive people with facial feature x as also having the personality characteristic y, subjects' beliefs (i.e., declarative knowledge) concerning the relation between x and y was not found to be affected and subjects appeared to continue to be completely convinced that there is no relation between x and y (Hill, Lewicki, Czyzewska, & Schuller, 1990; Lewicki, 1986a).
The direct comparisons between conscious and nonconscious processing based on the same inferential rules were found to involve methodological problems that make the results difficult to interpret (Lewicki & Scepan, 1987). For example, the knowledge about the rules (and instructions to follow them) may induce stress, performance anxiety and other factors which are absent in situations when subjects are asked to follow their "intuitions" or simply "guess." However, the consistent results from the studies involving complex encoding algorithms suggest that, as compared to nonconscious processing, subjects require considerably more time to use the same encoding rules in a consciously controlled manner.
HOW SOPHISTICATED IS THE NONCONSCIOUS
ACQUISITION OF INFORMATION?
It has been shown in several studies that covariations of considerable complexity can be nonconsciously processed. Subjects in those experiments nonconsciously acquired procedural knowledge about formal structures of the material that were not only too complex and too confusing to be consciously noticed, but which even exceeded the complexity level of knowledge that one can use in any consciously controlled inferences. For example, in one series of studies, subjects nonconsciously acquired knowledge about a pattern of the stimulus material which involved a four-way interaction (Lewicki, Czyzewska, & Hoffman, 1987, see also the replication and extension of those experiments by Stadler, 1989; see also Cleeremans & McClelland, in press). Extensive post-experimental interviews and tests indicate that the participants in those studies not only had no consciously controlled knowledge about the nature of the pattern (which they had "learned"), but they even were not aware of the existence of any pattern, and unable to detect it when explicitly instructed to do so and promised a high cash award if they succeeded. Additional tests of subjects' ability to make consciously controlled inferences based on higher order interactions have revealed that the human cognitive system is not equipped to handle such tasks on the consciously controlled level. Our conscious thinking needs to rely on notes (with flowcharts or lists of "if-then" statements) or computers to do the same job which our nonconsciously operating processing algorithms can do without external help and instantly.
E Recent research with children, demonstrated that such encoding algorithms involving interactions between variables can be relatively easily acquired (nonconsciously) even by preschoolers (4-5 year olds), whose consciously controlled thinking is at this stage so undeveloped that they cannot "understand" concepts of conditional relations or transitivity (Czyzewska, Hill, & Lewicki, 1991). Those results are also consistent with common observations that all normally developed children at preschool age are capable of using complex semantic and syntactic rules (necessary to fluently use language) at the point when their (conscious) thinking skills do not allow them to articulate or even "understand" the simplest rules of language.
A series of studies on the process of acquisition of such conditional (or "interaction-based") encoding algorithms suggests that subjects learn those complex knowledge structures via a process of "conditional elimination": An encoding algorithm based on a simple covariation (between two features or events) can be abandoned (and replaced by a new one) when it does not fit the current stimuli well. However, the abandoned algorithm is not lost entirely but only deactivated (i.e., "conditionally eliminated"), and it can be reactivated and used again when stimulus material consistent with the old algorithm is encountered (Lewicki, 1986a). If the crucial feature of the material (that determines which encoding algorithm should be used) is detected, then a higher order encoding algorithm begins to develop.
ABSTRACTION AND GENERALIZATION
The evidence reviewed in the previous section indicates that the mechanisms of nonconscious acquisition of information about covariations support formally complex knowledge structures. They also support semantic abstraction, in other words, the nonconsciously acquired covariations affect the general meanings of concepts and not only their labels, symbols or other specific instantiations. For example, in one line of experiments (Hill, Lewicki, Czyzewska, & Boss, 1989), subjects nonconsciously learned covariations involving the feature of "sadness;" however, this specific label was never used in the training-phase stimulus material where subjects watched videotapes depicting their peers (some of whom expressed feelings of sadness or depression). In the testing phase, subjects showed the expected bias when they rated some of the new stimulus persons as more "pessimistic," "sad," "dissatisfied," and "lonely." These results indicate that in the process of acquisition of the manipulated encoding algorithm subjects nonconsciously abstracted and generalized the meaning of a general concept from its specific instantiations encountered in the stimulus material.
The process of semantic abstraction and generalization in the nonconscious development of encoding algorithms was also demonstrated in studies where verbal descriptions of activities of stimulus persons were used (in the training phase) to manipulate covariations between certain personality features (Lewicki, 1986a). No labels (adjectives) were used in those descriptions; instead, examples of specific behaviors instantiated the features. In the testing phase, subjects rated a sample of new stimulus persons on relevant dimensions anchored with labels which were not used in the learning phase. Subjects' responses were consistent with the manipulation, indicating that information about specific behaviors presented in the learning phase was nonconsciously abstracted and converted into general encoding concepts.
This process of nonconscious generalization was also observed in research with small children (Czyzewska, Hill, & Lewicki, 1991). In a recent experiment, four- and five-year olds nonconsciously learned a covariation between colors of clothes of children presented on posters and very general categories of their activities: "physically active" (e.g., riding a bike, jumping, playing ball, running) vs. "physically passive" (e.g., watching TV, waiting, drawing, reading).
NONCONSCIOUS DEVELOPMENT OF
KNOWLEDGE STRUCTURES THAT ARE
RELATIVELY INDEPENDENT FROM EXPERIENCE
If the process of nonconscious generalization of covariations (encountered in the "outside" world) were the only mechanism responsible for the nonconscious development of encoding algorithms and procedural knowledge, then all of them would have to mirror the actual covariations of events in reality, and this is obviously not the case. Common unreasonable biases, gradually developing irrational preferences for particular colors, places, people, as well as various common forms of disorders (neuroses, phobias, borderline personality dysfunctions, etc.) indicate that many encoding algorithms and other elements of procedural knowledge develop in the cognitive system relatively independent from, or at least not as a direct function of experience with the outside world.
Several mechanisms have been identified which can account for such instances of relative independence or even discrepancy between procedural knowledge and the environment.
Self-perpetuation of Encoding Algorithms
It has been demonstrated in a number of studies that when stimuli are ambiguous, then encoding algorithms may nonconsciously impose on them preexisting interpretive categories even if the stimuli "objectively" do not match those categories. The resulting biased interpretation of stimuli (as supportive of the preexisting encoding dispositions) has been shown to become a source of subjective experiences that are consistent with these dispositions. This way, the encoding bias may gradually develop in a self-perpetuating manner (Hill, Lewicki, Czyzewska, & Boss, 1989; Lewicki, Hill, & Sasaki, 1989). Considering the decisive role of encoding algorithms for generating the subjective meaning of what one is encountering, and given the ambiguity and openness to alternative interpretations of many (especially social) stimuli that one encounters in every-day life, the process of self-perpetuation of encoding algorithms may play an important role in the development of a variety of individual differences in how individuals encode (and react to) the environment (Hill, Lewicki, Czyzewska, & Schuller, 1990; Hill, Lewicki, & Neubauer, 1991; Lewicki, Hill, & Sasaki, 1989).
In the learning phase of a typical experiment, subjects participated in (ostensible) training in "the intuitive interpretation" of some stimuli. The stimulus material contained a hidden covariation between some subtle features or events. In the testing phase, subjects' task was to interpret a very long sequence of new, relevant stimuli "based on intuition." Consistent with the self-perpetuation hypothesis, over a prolonged testing phase, subjects' responses became (gradually) increasingly consistent with the nonconscious encoding algorithm acquired in the learning phase, despite the fact that this (testing phase) material did not include any supportive evidence. In other words, once initiated, the development of the new encoding algorithms continued in a self-perpetuating manner. The initial experiences capable of triggering such a self-perpetuating development of a bias (and starting the "snow-ball") may in real-life be conditions that are very difficult to identify because they may be incidental, nonsalient, and even not consciously remembered as meaningful events by a perceiver. There is evidence demonstrating that surprisingly little consistent evidence is sufficient to produce an initial encoding bias (Lewicki, 1986b), and in some circumstances even a single instance may be sufficient (Lewicki, 1985).
The self-perpetuation process was demonstrated using a variety of stimulus materials, such as videotaped social interactions, descriptions of stimulus persons, silhouettes of stimulus persons, kinematic traces of body movements, words of an artificial language, matrix scanning, digitized transformations of human faces, and others (Hill, Lewicki, Czyzewska, & Boss, 1989; Hill, Lewicki, & Neubauer, 1991; Lewicki, Hill, & Sasaki, 1989).
The process of self-perpetuation is probably the most clear example of a cognitive mechanism capable of nonconsciously generating ("making up") new knowledge structures which are independent of (or even inconsistent with) the objective nature of the person's environment. The other two identified mechanisms (see below) appear to operate in a somewhat closer relation to what the individual encounters in the environment, however, they also nonconsciously "create" new encoding algorithms which can be potentially completely "wrong" in the sense that they do not reflect accurately what the individual is actually encountering in the outside world.
Nonconscious Indirect Inferences
There is evidence indicating that the processes of nonconscious acquisition of information about covariations (reviewed in the previous sections) may prompt a "spontaneous" development of new relations between concepts or features. Specifically, the relations can emerge between variables which have not been found to be connected in the environment and whose relation could only be inferred "indirectly" (i.e., by applying the rule of transitivity). In other words, if an individual acquires information about a covariation between features A and B, and independently, between features B and C, then this may result in the "spontaneous" development of an "expectation" that A and C are also related (i.e., a new encoding algorithm would emerge representing the nonconscious knowledge that objects which are A, are also C).
This phenomenon of nonconscious indirect inferences--"generating" new encoding algorithms by applying the rule of transitivity to "connect" existing algorithms--was recently demonstrated with different stimulus materials (Lewicki, Hill, & Czyzewska, 1994). Most of the studies used modified versions of procedures tested in previous research (matrix scanning tasks, schematic pictures of stimulus persons, etc.)
One of the experiments used the "judging personalities from kinematics" paradigm (Hill, Lewicki, Czyzewska, & Boss, 1989), in which subjects were exposed to videotapes presenting "abstracted" movements of selected points on bodies of "invisible" stimulus persons engaged in various activities. Information about the personality of each stimulus person was provided. In the first phase of the study, subjects nonconsciously learned a covariation between the personality information (A) and a subtle variation of distances between the dots identifying the legs of the stimulus persons (B). In the second phase, dots on subjects' arms were introduced (C) and their distances covaried with distances between the dots on legs (B), however, no information about the personality of the stimulus persons was provided (A). In the testing phase, only dots on arms were shown (C) and subjects were asked to "make intuitive judgements of personality" of the persons "based on the dynamics of their body language." Consistent with expectations, subjects' judgements (of feature A) where found to be based on the distances between dots on arms (feature C), although in this arrangement of the stimulus material the relation between the two features (A and C) could be established by subjects only indirectly (i.e., "via" feature B). As usual, tests of participants' awareness revealed no trace of their knowledge about any relations between the crucial features manipulated in the stimulus material. Moreover, not a single subject noticed any variation of the (in fact systematically varied) distances between dots marking the limbs of the stimulus persons.
IS THE NONCONSCIOUS INFORMATION
PROCESSING SYSTEM "INTELLIGENT"?
The answer to this question (which appears to be at least implicitly present in most discussions on the role of the nonconscious in human cognition) depends on the meaning of "intelligence" in this context.
If "intelligent" means having its own goals or specific motivations and being able to pursue them by triggering particular actions (such as those proposed in the psychoanalytic literature), then the answer to this question would be: "no." There is no empirical evidence in the cognitive literature for any specific content "built into" (or otherwise involved in) nonconscious acquisition or processing of information. The mechanics of those processes may lead some researchers (especially clinicians) to the illusion of identifying their content-specific character because some of them may develop in a self-perpetuating manner and eventually lead to strong content-related dysfunctions (e.g., in one's emotional reactions towards some specific categories of situations or objects). However, despite their high efficiency and formal sophistication, those demonstrated processes (in the experiments on nonconscious acquisition of knowledge) appear to be at least initially unbiased towards any specific contents and "impartial" in the sense of being ready to process any type of information regardless of the level of its consistency with the perceiver's consciously held beliefs or motivations (Lewicki, 1986a).
The answer to the question about "intelligence" would be affirmative if "intelligent" is understood as "equipped to efficiently process complex information." In this sense, our nonconscious information processing system is incomparably more capable to process formally complex knowledge structures, faster, and "smarter" overall than our ability to think and identify meanings of stimuli in a consciously controlled manner.
In light of the evidence reviewed in this article, the "division of functions" between the nonconscious and consciously controlled aspects of human cognition appears to be quantitatively and qualitatively asymmetrical. Most of the "real work" (both in the acquisition of skills and the execution of cognitive operations such as encoding and interpretation of stimuli) is being done at the level to which our consciousness has no access. Moreover, even if the access to that level existed, it could not be used in any way because the formal sophistication of that level and its necessary speed of processing exceed considerably what can even be approached by our consciously controlled thinking. The "responsibilities" of this inaccessible level of our cognition are not limited to the "housekeeping" operations such as retrieving information from memory or adjusting the level of arousal; they are directly involved in the development of interpretive categories, drawing inferences, determining emotional reactions, and other "high-level" cognitive operations traditionally associated with consciously controlled "thinking."
Cleeremans, A. & McClelland, J. L. (in press). Learning the structure of event sequences. Journal of Experimental Psychology: General.
Crocker, J. (1981). Judgment of covariation by social perceivers. Psychological Bulletin, 90, 272-292.
Czyzewska, M., Hill, T., & Lewicki, P. (1991). Acquisition of information about conditional relations between variables in preschool children. Unpublished manuscript.
deGroot, A. D. (1965). Thought and mind in chess. The Hague: Mouton.
Fodor, J. A. (1983). The modularity of mind. Cambridge, MA: MIT/Bradford.
Hill, T., & Lewicki, P. (1991). Personality and the nonconscious. In V. Derlega and W. Jones (Eds.), Introduction to contemporary research in personality. New York: Nelson Hall.
Hill, T., Lewicki, P., Czyzewska, M., & Boss, A. (1989). Self-perpetuating development of encoding biases in person perception. Journal of Personality and Social Psychology, 57, 373-387.
Hill, T., Lewicki, P., Czyzewska, & Schuller, G. (1990). The role of learned inferential encoding rules in the perception of faces: Effects of nonconscious self-perpetuation of a bias. Journal of Experimental Social Psychology, 26, 350-371.
Hill, T., Lewicki, P., & Neubauer, R. M. (1991). The development of depressive dispositions: A case of self-perpetuation of encoding biases. Journal of Experimental Social Psychology, 27, 392-409.
Hochberg, J. (1978). Perception. Englewood Cliffs, NJ: Prentice Hall.
Jacoby, L. L., & Witherspoon, D. (1982). Remembering without awareness. Canadian Journal of Psychology, 36, 300-324.
Kihlstrom, J. (1987). The cognitive unconscious. Science, 237, 1445-1452.
Lewicki, P. (1985). Nonconscious biasing effects of single instances on subsequent judgments. Journal of Personality and Social Psychology, 48, 563-574.
Lewicki, P. (1986a). Nonconscious social information processing. New York: Academic Press.
Lewicki, P. (1986b). Processing information about covariations that cannot be articulated. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 135-146.
Lewicki, P., Czyzewska, M., & Hoffman, H. (1987). Unconscious acquisition of complex procedural knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 523-530.
Lewicki, P., & Hill, T. (1987). Unconscious processes as explanations of behavior in cognitive, personality, and social psychology. Personality and Social Psychology Bulletin, 13, 355-362.
Lewicki, P., Hill, T., & Bizot, E. (1988). Acquisition of procedural knowledge about a pattern of stimuli that cannot be articulated. Cognitive Psychology, 20, 24-37.
Lewicki, P., & Hill, T. (1989). On the status of nonconscious processes in human cognition: Comment on Reber. Journal of Experimental Psychology: General, 118, 239-241
Lewicki, P., Hill, T., & Sasaki, I. (1989). Self-perpetuating development of encoding biases. Journal of Experimental Psychology: General, 118, 323-337.
Lewicki, P., Hill, T., & Czyzewska, M. (1992). Nonconscious acquisition of information. American Psychologist, 47, 796-801.
Lewicki, P., Hill, T., & Czyzewska, M. (1994). Nonconscious indirect inferences in encoding. Journal of Experimental Psychology: General, 123, 257-263.
Lewicki, P., Hill, T., & Czyzewska, M. (in press). Cognitive mechanisms for acquiring "experience": The dissociation between conscious and nonconscious cognition. In J. D. Cohen and J. W. Schooler (Eds.). Carnegie Mellon Symposium on Consciousness. Hillsdale, NJ: Erlbaum.
Lewicki, P., Hill, T., & Czyzewska, M. (in press). Hidden covariation detection: A robust and ubiquitous phenomenon. Journal of Experimental Psychology: Learning, Memory, and Cognition.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Prentice-Hall.
Reber, A. S. (1989). Implicit learning and tacit knowledge. Journal of Experimental Psychology: General, 118, 219-235.
Rock, I. (1975). Introduction to perception. New York: Macmillan.
Stadler, M. (1989). On learning complex procedural knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 1061-1069.
Smith, E. E., & Medin, D. L. (1981). Categories and concepts. Cambridge, MA: Harvard University Press.
Sternberg, R. J. (1986). Intelligence applied. New York: Harcourt, Brace, Jovanowitch.
Walster, E., & Walster, G. W. (1978). Love. Reading, MA: Addison-Weseley.
Winograd, T. (1975). Computer memories: A metaphor for memory organization. In C. N. Cofer (Ed.), The structure of human memory. San Francisco: Freeman.
The preparation of this article was supported by the National Science Foundation Grant BNS-8920726 and the National Institute of Mental Health Grant MH-42715-05 to Pawel Lewicki and Thomas Hill.
address: Psychology Department, University of Tulsa, Tulsa, OK 74104, USA
phone: (918) 631-2248, fax: (918) 631-2094, e-mail: firstname.lastname@example.org
For more information, see: Research on Nonconsciousness.