A number of ways of taxonomizing human learning have been proposed. We examine the evidence for one such proposal, namely, that there exist independent explicit and implicit learning systems. This combines two further distinctions, (1) between learning that takes place with versus without concurrent awareness, and (2) between learning that involves the encoding of instances (or fragments) versus the induction of abstract rules or hypotheses. Implicit learning is assumed to involve unconscious rule learning. We examine the evidence for implicit learning (...) derived from subliminal learning, conditioning, artificial grammar learning, instrumental learning, and reaction times in sequence learning. We conclude that unconscious learning has not been satisfactorily established in any of these areas. The assumption that learning in some of these tasks (e.g., artificial grammar learning) is predominantly based on rule abstraction is questionable. When subjects cannot report the rules that govern stimulus selection, this is often because their knowledge consists of instances or fragments of the training stimuli rather than rules. In contrast to the distinction between conscious and unconscious learning, the distinction between instance and rule learning is a sound and meaningful way of taxonomizing human learning. We discuss various computational models of these two forms of learning. (shrink)

A number of ways of taxonomizing human learning have been proposed. We examine the evidence for one such proposal, namely, that there exist independent explicit and implicit learning systems. This combines two further distinctions, between learning that takes place with versus without concurrent awareness, and between learning that involves the encoding of instances versus the induction of abstract rules or hypotheses. Implicit learning is assumed to involve unconscious rule learning. We examine the evidence for implicit learning derived from subliminal learning, (...) conditioning, artificial grammar learning, instrumental learning, and reaction times in sequence learning. We conclude that unconscious learning has not been satisfactorily established in any of these areas. The assumption that learning in some of these tasks is predominantly based on rule abstraction is questionable. When subjects cannot report the “implicitly learned” rules that govern stimulus selection, this is often because their knowledge consists of instances or fragments of the training stimuli rather than rules. In contrast to the distinction between conscious and unconscious learning, the distinction between instance and rule learning is a sound and meaningful way of taxonomizing human learning. We discuss various computational models of these two forms of learning. (shrink)

I examine the phenomenon of implicit learning, the process by which knowledge about the rule-governed complexities of the stimulus environment is acquired independently of conscious attempts to do so. Our research with the two seemingly disparate experimental paradigms of synthetic grammar learning and probability learning, is reviewed and integrated with other approaches to the general problem of unconscious cognition. The conclusions reached are as follows: Implicit learning produces a tacit knowledge base that is abstract and representative of the structure of (...) the environment; such knowledge is optimally acquired independently of conscious efforts to learn; and it can be used implicitly to solve problems and make accurate decisions about novel stimulus circumstances. Various epistemological issues and related problems such as intuition, neuroclinical disorders of learning and memory, and the relationship of evolutionary processes to cognitive science are also discussed. (shrink)

Reviews evidence which suggests that there may be little or no direct introspective access to higher order cognitive processes. Ss are sometimes unaware of the existence of a stimulus that importantly influenced a response, unaware of the existence of the response, and unaware that the stimulus has affected the response. It is proposed that when people attempt to report on their cognitive processes, that is, on the processes mediating the effects of a stimulus on a response, they do not do (...) so on the basis of any true introspection. Instead, their reports are based on a priori, implicit causal theories, or judgments about the extent to which a particular stimulus is a plausible cause of a given response. This suggests that though people may not be able to observe directly their cognitive processes, they will sometimes be able to report accurately about them. Accurate reports will occur when influential stimuli are salient and are plausible causes of the responses they produce, and will not occur when stimuli are not salient or are not plausible causes. (shrink)

Time underlies many interesting human behaviors. Thus, the question of how to represent time in connectionist models is very important. One approach is to represent time implicitly by its effects on processing rather than explicitly (as in a spatial representation). The current report develops a proposal along these lines first described by Jordan (1986) which involves the use of recurrent links in order to provide networks with a dynamic memory. In this approach, hidden unit patterns are fed back to themselves: (...) the internal representations which develop thus reflect task demands in the context of prior internal states. A set of simulations is reported which range from relatively simple problems (temporal version of XOR) to discovering syntactic/semantic features for words. The networks are able to learn interesting internal representations which incorporate task demands with memory demands: indeed, in this approach the notion of memory is inextricably bound up with task processing. These representations reveal a rich structure, which allows them to be highly context‐dependent, while also expressing generalizations across classes of items. These representations suggest a method for representing lexical categories and the type/token distinction. (shrink)

Subjects exposed to strings of letters generated by a finite state grammar can later classify grammatical and nongrammatical test strings, even though they cannot adequately say what the rules of the grammar are (e.g., Reber, 1989). The MINERVA 2 (Hintzman, 1986) and Medin and Schaffer (1978) memory‐array models and a number of connectionist outoassociator models are tested against experimental data by deriving mainly parameter‐free predictions from the models of the rank order of classification difficulty of test strings. The importance of (...) different assumptions regarding the coding of features (How should the absence of a feature be coded? Should single letters or digrams be coded?), the learning rule used (Hebb rule vs. delta rule), and the connectivity (Should features be predicted only by previous features in the string, or by all features simultaneously?) is investigated by determining the performance of the models with and without each assumption. Only one class of connectionist model (the simultaneous delta rule) passes all the tests. It is shown that this class of model can be regarded by abstracting a set of representative but incomplete rules of the grammar. (shrink)

Implicit learning is said to occur when a person learns about a complex stimulus without necessarily intending to do so, and in such a way that the resulting knowledge is difficult to express. Over the last 30 years, a number of studies have claimed to show evidence of implicit learning. In more recent years, however, considerable debate has arisen over the extent to which cognitive tasks can in fact be learned implicitly. Much of the debate has centred on the questions (...) of how unconscious, and how abstract, is implicitly acquired knowledge? The aim of this book is to provide students and researchers with a self-contained and balanced summary of the various theoretical and empirical positions that are currently shaping this exciting area of research. (shrink)

Comparing the relative sensitivity of direct and indirect measures of learning is proposed as the best way to provide evidence for unconscious learning when both conceptual and operative definitions of awareness are lacking. This approach was first proposed by Reingold & Merikle (1988) in the context of subliminal perception. In this paper, we apply it to a choice reaction time task in which the material is generated based on a probabilistic finite-state grammar (Cleeremans, 1993). We show (1) that participants progressively (...) learn about the statistical structure of the stimulus material over training with the choice reaction time task, and (2) that they can use some of this knowledge to predict the location of the next stimulus in a subsequent “generation” task. However, detailed partial correlational analyses of the correspondence between performance during the reaction time task and the statistical structure of the training material showed that large effects remained even when controlling for explicit knowledge as assessed by the generation task. Hence we conclude (1) that at least some of the knowledge expressed through reaction time performance can not be characterized as conscious, and (2) that even when associations are found at a global level of analysis, dissociations can still be obtained when more detailed analyses are conducted. Finally, we also show that participants are limited in the depth of the contingencies they can learn about, and that these limitations are shared by the Simple Recurrent Network model of Cleeremans & McClelland (1991). (shrink)

Comparing the relative sensitivity of direct and indirect measures of learning is proposed as the best way to provide evidence for unconscious learning when both conceptual and operative definitions of awareness are lacking. This approach was first proposed by Reingold & Merikle (1988) in the context of subliminal perception. In this paper, we apply it to a choice reaction time task in which the material is generated based on a probabilistic finite-state grammar (Cleeremans, 1993). We show (1) that participants progressively (...) learn about the statistical structure of the stimulus material over training with the choice reaction time task, and (2) that they can use some of this knowledge to predict the location of the next stimulus in a subsequent “generation” task. However, detailed partial correlational analyses of the correspondence between performance during the reaction time task and the statistical structure of the training material showed that large effects remained even when controlling for explicit knowledge as assessed by the generation task. Hence we conclude (1) that at least some of the knowledge expressed through reaction time performance can not be characterized as conscious, and (2) that even when associations are found at a global level of analysis, dissociations can still be obtained when more detailed analyses are conducted. Finally, we also show that participants are limited in the depth of the contingencies they can learn about, and that these limitations are shared by the Simple Recurrent Network model of Cleeremans & McClelland (1991). (shrink)

Je lui ai associÉ un court extrait d'une revue de questions portant sur le même thème. Implicit memory is revealed when previous experiences facilitate perf on a task that does not require conscious or intentional recollection of those expces. Explicit memory is revealed when perf on a task requires conscious recolelction of previous expces. Il s'agit de defs descriptives qui n'impliquent pas l'existence de deux systs de mÉmo sÉparÉs. Historiquement Descartes est le premier ˆ faire mention de phÉnomènes de mÉmo (...) implicites, Leibniz, Maine de Biran, Bergson en ont Également parlÉ. Korsakoff (1889) avec ses travaux sur les amnÉsiques montre qu'ils sont capables d'utiliser de l'info qu'ils ont acquises dans des Épisodes rÉcents sans pour autant manifester le moindre souvenir de l'Épisode. L'interprÉtation de ces phÉnomènes Était alors que ça renvoyait ˆ des traces en mÉmoire trop tÉnues pour accèder jusqu'au champ de la conscience. Cette interprÉtation va pouvoir être ÉcartÉe par les travaux rÉcents. Thorndike a Également conduit un nbre important de travaux sur MI. Les trauvaux un peu plus rÉcents ont montrÉ qu'un apprentissage antÉrieur pouvait faciliter une perfce sans qu'il y ait ref explicite à l'Épisode d'apprentissage. On a montrÉ aussi l'effet de stimuli subliminaux, qui en tant que tels ne peuvent pas être rappelÉs consciemment, mais qui influencent la conduite ou les jugts ultÉrieurs. Par ex, la prÉsentation de formes sur des temps très courts qui excluent la possibilitÉ d'une perception consciente influence significativement lechoix de formes dans une pÉreuve ultÉrieure de choix forcÉs où les sjts doivent indiquer la forme qu'il prÉfère entre deux. Ils choisissent prÉfÉrentiellement celles qui leur ont ÉtÉ prÉsentÉes de manière subliminale. De même Bargh et al. ont montrÉ que des sjts soumis de manière subliminale ˆ des mots hostiles portent ultÉrieurement des jugts plus nÉgatifs sur une personne target. Il ya Également eu des travaux sur l'apprsge ou le condt without awareness en particulier apprsge implicite de règles de grammaire (ie le sjt peut construire de nouvelles phrases conformes ˆ la syntaxe de celles qui lui ont ÉtÉ prÉsentÉes sans pour autant petre capables d'noncer les règles de grammaire qu'il utilise) Rem: N'est ce pas un pbe de typemÉtacognition??? Enfin, tous les travaux sur le priming sont des elts en faveur de MI: Facilitation in the processing of a stimulus as a function of a recent encounter with the same stimulus. Type de perfce sur lesquels les effets de priming ont ÉtÉ dÉmontrÉs: -T‰ches de dÉcision lexicale: Le sjt doit dire si la cha"ne de caractères qu'on lui prÉsente est un mot ou pas: Le temps de latence est beaucoup plus court si la cha"ne a dÉjˆ ÉtÉ prÉsentÉe une première fois. -Mesures de temps d'identification de mots en rpÉsentation tachitoscopiques -ComplÉmentation de mots: les sjts ont tendance ˆ complÉter pour crÉer les mots qui leur ont ÉtÉ prÉsentÉs antÉrieurement. L'intÉrêt pour le paradigme de priming se retouve chez les gens qui s'intÉressent ˆ: -reconnaissance de mots et organisation lexicale. les effets de priming permettent de faire des hyps sur l'accès lexical et la reprÉsentation. -M. Épisodique: travux sur les amnÉsiques qui ne se souviennent pas qu'on vient de leur prÉsenter des mots mais qui manifestent une mÉmoire implicite lorsqu'on leur fait faire de la complÉmentation de mots. DISSOCIATION ENTRE M.I. ET M EXPLICITE 1-Si de nombreux travaux montrent que la M explicite est influencÉe par le travail d'Élaboration (profondeur de traitement) en jeu pendant la première phase d'Étude des stimuli, il semble que la MI ne le soit pas. 2-Le changt de modalitÉ (par ex auditif -> visuel) entre la phase d'Étude et la phase de test provoque une attÉnuation importante de l'effet de priming si le test est un test de MI mais pas si c'est un test de ME. 3-Les effets de primings persistent assez largement sur plusieurs jours ce qui n'est pas le cas des perfces ˆ une Épreuve de reconnaissance. 4-Les perfces sur des tests de MI ne semblent pas influencÉes par des manipulations d'interfÉrences proactives ou rÉtroactives dont on sait qu'elles influencent les perfces en ME. 5-Absce de corrÉlation entre perfces ˆ des tests de reconnaissance et perfces ˆ une Épreuve de complÉmentation de mots. SIMILARITES ENTRE PRIMING ET REMEMBERING 1- A certaines conditions les durÉes de rÉtention on des effets parallèles sur les effets de priming et les mesures de ME (REm: en partie contradictoire avec 3 ci dessus) 2-La manipulation du contexte de la liste lors de la phase d'Étude affecte ˆ la fois les perfces en reconnaissance et en identification de mots. 3- ME et MI sont influencÉes par des associations nouvellement acquises entre des paires de mots non reliÉes (Rem: contradictoires avec 4 ci dessus ??) 4- Rem: Le contraire de 1 du pt "dissociation" !!!! dans une manip particulière 5-Johnston et al. ont montrÉ que des mots identifÉs plus rapidement Étaient plus svt jugÉs comme 'old' dans une t‰che de reconnaissance que les mots identifiÉs plus lentement. Il n'est pas exclu que les phÉnomènes que l'on interprètent comme relevant de MI soient parfois de la ME involontaire cad des cas o les cues fournis dans le test conduisent ˆ un souvenir involontaire mais totalement conscient. LES INTERPRETATIONS THEORIQUES DE MI 1-Les effets de priming sur des test de MI sont attribuables ˆ l'activation temporaires de reprÉsentations ou structures de K prÉ-existantes . Cette activation est supposÉe avoir lieu de manière automatique, indÉpendamment des processus d'Élaboration nÉcessaires pour Établir de nouvelles traces Épisodiques. 2- les difÉrences entre ME et MI sont dues ˆ la nature et aux relations entre les processus d'encodage et de rÉcupÉration. Une version de cette approche est celle qui s'appuie sur la distinction entre processus dirigÉs par les donnÉes et processus dirigÉs par les concepts. Ces derniers reflètent des activitÉs initiÉes par le sjt comme l'Élaboration, l'organisation et la reconstruction, c'est sur ce type de processus que reposerait la ME alors que la MI reposerait sur des processus data-driven. 3-ME et MI renvoient ˆ des systs sous jacents diffÉrents. Pour Cohen & Squire (1984), la ME est supportÉe par une M DECLARATIVE , syst qui est en jeu dans la formation de nouvelles reprÉsentations. La MI repose sur un syst PROCEDURAL dans lequel la M se manifeste par des modifications on line de procÉdures ou opÉrations de traitement. Le syst de M Épisodique est vu comme la base pour le souvenir explicite d'evts rÉcents tandis que la M sÉmantique est responsable des perfces sur des tests de MI. Chacune de ces approches est conforme ˆ certaines des donnÉes recueillies mais pas ˆ d'autres. (shrink)