Contributions of declarative and procedural memory to second language development: An examination of accuracy and automatization during practice
Diana Pili-Moss, Kara Morgan-Short; Lancaster University, Lancaster, United Kingdom; University of Illinois, Chicago, United States of America
Behavioral and event-related potential studies have examined the effects of declarative and procedural memory (DM and PM) on second language (L2) and have shown that they respectively predict accuracy at early and later stages (Faretta-Stutenberg & Morgan-Short, 2017; Hamrick, 2015; Morgan-Short et al., 2014). However, these studies do not examine how these memory systems contribute over a more fine-grained time course to the development of accuracy and automatization during extended L2 practice. Thus, we aimed to look at how these memory systems contributed to practice itself investigating their relationship longitudinally with measures of accuracy, of reaction times and of automatization, as reflected by a measure of processing stability (Segalowitz, 2010), in the artificial language learning task employed in Morgan-Short et al. (2014). Regression analyses found that a composite measure of PM significantly predicted processing stability from relatively early in practice, although it did not predict accuracy. In contrast, DM did not contribute significantly to processing stability but significantly predicted accuracy from early in practice. Taken together, these results confirm the complementarity of the contributions of PM and DM to the process of L2 learning (Ullman, 2005) providing evidence that newly elucidates the relationship between memory and L2 accuracy and automatization.
Keywords: Declarative Memory, Procedural Memory, Automatization, Artificial Language Learning
Implicit knowledge transfer in problem-solving
Ivan Ivanchei; Saint Petersburg State University, Saint Petersburg, Russia; Russian Academy of National Economy and Public Administration, Moscow, Russia
For decades, researchers in the problem-solving field were unable to increase the ability to solve so-called insight problems with explicit hints (see e.g. Weisberg, 2015). The aim of the present study was to examine the unconscious transfer of implicitly learned regularity on the new task. We were also interested in factors modulating such transfer. In the first experiment, we showed that the hidden regularity of the first task (“16 dots”) can be transferred to a new task (“Labyrinth”). In the second experiment, we asked participants to verbalize every decision in the Labyrinth task, which leads to a decreased performance (see Berry & Broadbent, 1984; Dickson, McLennan & Omodei, 2000). In the third experiment, we tried to make the relationships between two tasks as implicit as possible, and participants did not reveal any transfer, even with the detailed analysis of their movements. We conclude that despite we found some evidence for implicit nature of transferred skill (decreased performance due to verbalization of decisions), it can’t be transferred if participants are not aware of the relationships between two problem-solving tasks. This result supports the approach postulating different levels of knowledge used by subjects after learning (Dienes, 2012).
Keywords: Transfer, Problem-Solving, Srt, Verbalization
Pre-attentuive representation of transitional probabilities in sound sequences
István Winkler, Maria Mittag, Rika Takegata; Hungarian Academy of Sciences, Budapest, Hungary; University of Washington, Seattle, United States of America; University of Helsinki, Helsinki, Finland
Predictions of upcoming sounds in a sequence can be most easily made if the auditory system encoded the probabilities with which sounds follows one another (transitional probability). Although current theories of pre-attentive auditory deviance detection (indexed by the mismatch negativity [MMN] event-related brain potential) are based on predictive processing principles, to now no study has tested whether this function involves representations of stimulus/pattern or transitional probabilities. In separate passive and active detection conditions, we presented healthy adults with three types of rare tone-triplets among frequent standard triplets of High-Low-High (HLH) or LHL pitch structure: proximity deviant (HHH/LLL), reversal deviant (LHL/HLH), and first-tone deviant (LLH/HHL). If deviance detection was based on stimulus probabilities, then no MMN can be expected to be elicited by any of the deviants. If the underlying representation stores pattern probabilities, the reversal and first-tone deviants should be detected with similar latency because both differ from the standard at the first pattern position. If deviance detection was based on transitional probabilities, then reversal deviants should be the most difficult to detect, because, unlike the other two deviants, they do not contain low-probability pitch transitions. The data clearly showed that both behavioral deviance detection and MMN utilize transitional probabilities.
Keywords: Sound Sequence, Transitional Probability, Memory Representation, Pre-Attentive Processing
Does implicit knowledge facilitate rule discovery?
John Williams, Dimitrios Alikaniotis; University of Cambridge, Cambridge, U.K.
What is the relationship between implicit and explicit knowledge – are these two distinct non-interfacing knowledge types, or can explicit knowledge emerge from implicit knowledge under the right conditions? We address this issue in the context of learning a simple (but difficult to spot) linguistic regularity using an SRT-inspired RT procedure. The ‘Structured’ group (n = 46) were exposed to the system for sufficient time for evidence of learning the regularity to appear in their RTs. A ‘Random’ group (n = 40) performed the same task but on material that lacked any such regularity. Both groups then received structured material in a prediction task with feedback so as to encourage rule discovery. There was no difference in awareness rate between the Structured and Random groups (39% and 42% respectively), nor on level of rule recognition, or subjective estimates of when awareness emerged. The same result was obtained in an experiment where trial repetition was used to induce insight. Further work is on going, but the results so far suggest that prior exposure to regularities, sufficient to produce demonstrable implicit learning, does not necessarily facilitate the emergence of explicit knowledge. Methodological problems involved in addressing this question will be discussed.
Keywords: Insight, Awareness, Language Learning, Rule Learning
Cross cultural differences in implicit learning of chunks and non-adjacent dependencies
Xiaoli Ling1, Li Zheng, Xiuyan Guo, Shouxin Li1, Shiyu Song, Lining Sun, Zoltan Dienes; Shandong Normal University, Jinan, China; East China Normal University, Shanghai, China; University of Connecticut, Storrs, United States of America; University of Sussex, Brighton, United Kingdom
Three experiments explore whether knowledge of grammars defining global vs local regularities has an advantage in implicit acquisition and whether this advantage is affected by cultural differences. Participants were asked to listen to and memorize a number of strings of 10 syllables instantiating a mirror symmetry (an inversion, i.e., a global pattern); after the training phase, they were required to judge whether new strings were well formed. In Experiment1, Western people implicitly acquired the inversion rule defined over the Chinese tones in a similar way as Chinese participants when alternative structures (specifically, chunking and repetition structures) were controlled. In Experiment 2 and 3, we directly pitted knowledge of the inversion (global) against chunk (local) knowledge, and found that Chinese participants striking global advantage in implicit learning, which was greater than that of Western participants. Taken together, we show for the first time cross cultural differences in the type of regularities implicitly acquired.
Keywords: Artificial Grammar Learning; Supra-Finite State State; Subjective Measures; Cross Cultural Differences
Event-related brain activity discloses an implicit memory system in vision
István Czigler; Hungarian Academy of Sciences, Budapest, Hungary
As the emergence of auditory and visual mismatch negativity (aMMN and vMMN) event-related potential components demonstrate, sensory memory systems are sensitive to regular appearance of consecutive events. The acquisition of this memory representation is automatic (i.e. it does not require attentional processes). Dealing with basically successive information, it is not surprising in the auditory modality. However, in vision at the level of perceptual experience sensitivity to successive (temporal) changes is poor, as it is demonstrated by such obvious effects like the change blindness. On the contrary, as our results shows, the implicit memory system, as tested by its reaction to the violated regularities (i.e. the vMMN) is capable of representing elementary visual features (e.g. color, orientation), conjunctions of features (objects), perceptual categories (symmetry), image statistics, laterality of body parts (left vs. right hand), facial emotions and gender-like characteristics (female vs. male faces).
Keywords: Sensory Memory, Change Detection, Vmmm
Statistical learning in adults: an EEG study on the units of speech perception
Sarah von Grebmer zu Wolfsthurn, Nina Kazanina, Josie Briscoe; University of Bristol, Bristol, England
Statistical learning (SL) reflects an ability to extract regularities from the environment, evidenced in a range of domains (Conway & Christiansen, 2005) including speech perception (Johnson & Jusczyk, 2001). Our study uses SL to investigate the units of speech perception using implicit and explicit measures of SL combined with EEG (Batterink & Paller, 2016; unpublished). In the training phase, participants listen to artificial speech streams devoid of intonation cues (e.g. peadoosabezogufootameanevuko...) while EEG is recorded. The streams contain either syllable-based or phoneme-based statistical regularities (Bonatti, Peña, Nespor & Mehler, 2005). In the testing phase, participants complete a rating task and a forced-choice task (Bonatti et al., 2005) testing explicit learning and a target-detection task testing implicit learning of words (Franco et al., 2015). Furthermore, we expect an increase in EEG power at the word frequency (4Hz) in addition to the syllable level (1.3Hz) in both conditions as a signature of word learning. Whereas the EEG results are still being analysed, behavioural results to-date (n=6) show the listeners’ ability to extract both statistical regularities as reflected in implicit and explicit tasks.Results have implications for theories of speech perception, multilingual speakers and for diagnosis of clinical populations (Obeid et al., 2016).
Keywords: Statistical Learning, Units Of Speech Perception, Frequency-Tagging Analysis, Phonemes
Implicit learning: Cognitive consequences of human neuroplasticity
Paul J. Reber; Northwestern University, Evaston, United States of America
Memory systems theory provides a valuable framework for understanding types of memory in the brain including both explicit, conscious memory and implicit learning that occurs without awareness and emerges automatically from experience. These two different forms of learning depend on different neural mechanisms but both contribute to our ability to learn complex cognitive abilities. A brain-based theory of implicit learning as emergent from widespread neuroplasticity mechanisms throughout cortex will be described. The theory bridges findings from behavior-based research, neuropsychology and neuroimaging, and connects to phenomena associated with implicit learning such as subjective lack of awareness, gradual learning and relative inflexibility. A computational model grounded in systems neuroscience will be used to illustrate how the memory systems framework can guide research on applications of implicit learning to studies of skill learning and the development of intuition from experience.
Keywords: Memory Systems, Neuroscience, Computational Models
Implicit learning of emotional sequences
Razvan Jurchis, Andrei Costea, Adrian Opre; Babeș-Bolyai University, Cluj-Napoca, Romania
The majority of studies on implicit learning use emotionally neutral stimuli. In this study we investigate if sequences of neutral and emotional stimuli can be learned implicitly. In the acquisition phase, participants are exposed to a second-order conditional sequence of neutral, positive, and negative words (e.g., neutral – positive – negative –positive – neutral - negative). On each iteration of the sequence, the specific word that instantiates the emotion changes, but the abstract sequence remains constant. In the test phase, we expose participants with word triplets and ask them to judge, for each triplet, whether it appeared in the previous phase or not. Half of the triplets are incompatible with the abstract acquisition sequence (non-grammatical; e.g., neutral – positive – neutral). The others are compatible with the acquisition sequence (grammatical; e.g., neutral – positive – negative). However, all triplets are new: even though they obey the abstract sequence, the specific words that compose them have not appeared consecutively in the acquisition phase. We expect that if participants learn the acquisition sequence, they will consider that a larger proportion of grammatical triplets were exposed in the first phase, compared to nongrammatical ones. We will check the awareness of learning using knowledge attributions, confidence ratings, and a post-experimental questionnaire. The study is now in the data collection phase.
Keywords: Implicit Learning; Abstract Emotional Sequences
Exploiting statistical regularities in implicit sequence learning
Beat Meier; University of Bern, Bern, Switzerland
Despite the long research tradition it is not clear what is learned in implicit sequence learning. Is learning based simply on the sensitivity to the frequencies of sequence element transitions (i.e., statistical learning of sequence transitions) or is it based on the built up of a more comprehensive sequence representation (i.e., a kind of “melody”)? The purpose of this study was to distinguish between these two possibilities. In one condition the target sequence (a 12-elements SOC) was presented repeatedly without interruption and surrounded by other 12-element SOCs at the beginning and the end of each learning block. In the other condition, the same target sequence was presented twice a time, alternating with the other 12-element SOCs. Importantly, across the two conditions, frequencies of sequence element transitions were kept constant across eight learning blocks. If only the transition probabilities are learned one would expect the same learning effect across conditions. In contrast, if a more comprehensive sequence representation is built up one would expect a learning advantage for the former condition. The results are consistent with the second hypothesis. This can be interpreted as evidence that the human cognitive system can exploit regularities in the environment beyond simple sequence element transitions.
Keywords: Srtt, Sequence Represenation, Transition Probabilities
Statistical learning and the effect of Starting Small in Specific Language Impairment
Ferenc Kemény, Ágnes Lukács; University of Graz, Graz, Austria; Budapest University of Technology and Economics, Budapest, Hungary
Children with Specific Language Impairment (SLI) lag behind their typically developing (TD) peers in tasks requiring statistical learning, and some argue that this deficit in statistical learning may account for difficulties in different domains of language as well. To gain more insight to the nature of this deficit, we compared the statistical learning performance of children with SLI (n = 40, mean age = 9.3, Sd = 1.2) and age-matched TD children on an Artificial Grammar Learning (AGL) task to test whether children with SLI show a statistical learning deficit on an AGL task with auditory sequences of nonsense syllables. Motivated by the Starting Small hypothesis assuming that incremental presentation of stimuli of different length has a facilitating effect on learning complex structures, we also aimed to test the effect of Starting Small to see weather ordered presentation of the auditory sequences helps AGL in SLI. Our results show that children with SLI have difficulties in extracting regularities from acoustic sequences of syllables organized by an artificial grammar. Moreover, while the learning performance of TD children is enhanced by presenting shorter strings before longer ones during training, no Starting Small benefit was observed in SLI.
Keywords: Statistical Learning, Starting Small, Specific Language Impairment
Statistical learning in children: exploring syllable-detection latencies as an implicit measure
Josie Briscoe, Rebecca Prosser, Sarah Von Grebmer Zu Wolfsthurn , Nina Kazanina; University of Bristol, Bristol, United Kingdom
Statistical learning (SL) captures regularities from the environment across multiple domains, including speech (Conway & Christiansen, 2005). Measuring SL from speech offers a useful insight to language acquisition, especially for atypical language development. Traditional methods rely on explicit recognition of tri-syllabic items at test, as evidence of SL, in adults and children. Learning is indexed through generalization from concatenated nonsense syllable streams that vary in probability during exposure, to word-like units at test. For children, recognition performance is typically above chance (e.g. Saffran et al., 1997), but not reliably so, suggesting that recognition performance varies with developmental factors (e.g. phonological memory). Our study combined implicit and explicit measures of SL to assess school-age children. Using a target-detection paradigm for syllables (following Franco et al., 2015) we sought to establish children’s sensitivity to SL on an implicit measure. We predicted that children, like adults, would identify targets from a presented stream at test, and that latencies would vary with probabilistic structure of exposure stream. With n=36 children aged 6-9 years tested, results are currently under analysis. Preliminary findings indicate variation in children’s ability to detect the target syllables, subsequent validation of latency data has implications for identifying children at language risk.
Keywords: Statistical Learning, Children, Target-Detection, Implicit
Fine-grained visual statistical learning in infancy: Extending Kirkham et al. (2002)
Arnaud Destrebecqz, Esti San Anton, Julie Bertels; Université libre de Bruxelles, Brussels, Belgium
Infants’ ability to detect statistical regularities between visual objects has been demonstrated in several studies (e.g., Kirkham et al., 2002). The extent to which infants learn the actual values of the transitional probabilities (TPs) between these objects nevertheless remains an open question. In order to obtain a fine-grained measure of the infants’ visual statistical learning ability, we examined in two experiments 8-month-old infants’ ability to discriminate between highly familiar and fairly familiar sequences (high and medium TPs, respectively, as in previous studies)high and medium values of TPs, on the one hand, and between to differentiate between these two types of sequences and new sequences that involved completely unfamiliar ones (i.e., null TPs) on the other hand. Results showed that infants discriminated between these three types of sequences, hence supporting that 8-month-olds extract subtlefine-grained statistical information from a visual stream of visual stimuli. Statistical learning processes would thus not only involve chunking the stream in smaller units reflecting the associations between visual elements, but also learning the transitional probabilities between these elements.
Keywords: Statistical Learning, Infant Learning, Transitional Probabilities, Chunking
Structural and Judgment Knowledge in Multiple-Cue Learning Task
Roman Tikhonov, Nadezda Moroshkina; Saint Petersburg University, Saint Petersburg, Russia
The study investigates the relation between structural and judgment knowledge in a multiple-cue learning task with 6 cues (4 relevant and 2 irrelevant) (N = 20). There were 2 phases of 60 trials each — learning phase (with feedback) and test phase (without feedback). Conscious judgment knowledge was measured with trial-by-trial 4-scale Confidence Ratings (CR). Post-experimental Cue Categorization (CC), and a modified Process Dissociation Procedure (PDP) were used as a measure of structural knowledge. The percentage of correct responses ranged from 33% to 82%, only 9 out of 20 participants reached above chance overall performance level. PDP score was positively correlated with average overall performance (r = .62, p = .004) and with CC results (r = .57, p = .009), but there was no statistically significant correlation with CR (r = .11, p = .657). PDP score was more correlated with results in the test phase (r = .70, p = .001), than in learning phase (r = .40, p = .08). CC was also positively correlated with performance, but it has not reached statistical significance. CR was not significantly correlated with performance (r = .04, p = .877), and the difference between mean CR for wrong and correct responses also was not significant (t = 0.923, df = 19, p = .368). The results suggest that participants obtained conscious structural knowledge about the cues, but has no conscious judgment knowledge. The study was supported by the Russian Foundation for Humanities, project 15-36-013-55.
Keywords: Multiple-Cue Learning, Structural Knowledge, Judgment Knowledge, Process Dissociation Procedure, Confidence Rating
Time to Learn: why my Sense of Agency grows?
Almara Kulieva, Maria Kuvaldina; St. Petersburg State University, St. Petersburg, Russia; SUNY Farmingdale State College, New York, United States of America
Sense of Agency (SoA) refers to the feeling of voluntarily control over our actions (Chambon & Haggard, 2012). Participants responded to arrow targets that appeared after prime arrows. Then a circle which color depended on the pressed key and its compatibility with the prime appeared. Participants judged their subjective control over the circle color. According to comparator model (Sato & Yasuda, 2005), we suggested that the ability to predict the color and SoA ratings will grow with practice and the perceptual availability of primes. We conducted three experiments: 288 trials each, liminal and subliminal prime trials in 2 or 4 blocks or mixed together. In blocked design participants could form a response bias based on prime visibility (Kulieva & Kuvaldina, 2016). SoA gradually increased in the mixed design experiment. This illustrates the possibility to learn the association rule in the current paradigm (Spengler et al., 2009). When a response bias was possible (blocked design, 2 blocks), SoA increased only in liminal priming trials, but decreased in subliminal. When response bias was formed within a shorter time period (4 blocks), SoA decreased independently of priming. Results are explained by the insufficient rule learning time and SoA “Judgement of Agency“ level (Synofzik et al., 2008).
Keywords: Voluntary Action, Sense Of Agency, Implicit Learning
Lack of frequency-tagged magnetic responses suggests statistical regularities remain undetected during sleep
Juliane Farthouat, Anne Atas, Xavier de Tiège, Philippe Peigneux; Université libre de Bruxelles, Brussels, Belgium
Recently, it was demonstrated that acquisition of reflex stimulus-response associations is possible during sleep. Whether sleep allows more complex forms of learning, such as stimulus-stimulus contingencies, remains an open question. In the current study, we recorded during diurnal sleep auditory magnetoencephalographic (MEG) frequency-tagged responses mirroring ongoing statistical implicit learning. Sleeping participants were exposed at non-awakenings thresholds to fast auditory streams of tones either randomly organized or structured in such a way that the stream could be statistically segmented in which tones set of 3 elements (tritones). Although unambiguous tone-related frequency-tagged MEG responses were found in all participants during sleep, there was no evidence of segmentation, i.e. no tritone frequency-tagged responses. In the ensuing wake period however, all participants exhibited robust tritone-related responses during exposure to statistical streams. Finally, the temporal evolution of segmentation-related MEG responses during exposure at wake was not different between participants exposed vs. not to statistical streams during prior sleep. Consequently, our results suggest that complex stimulus-stimulus associations embedded in statistical regularities are not detected during sleep, and that learning capabilities during sleep restrict to simpler elementary associations.
Keywords: Sleep, Learning, Meg, Statistical Learning
Focusing on goal relevance – is it crucial for artificial grammar learning?
Agnieszka Popławska-Boruc, Radosław Sterczyński, Marta Roczniewska; University of Social Sciences and Humanities, Sopot, Poland
Eitam, Schul and Hassin (2009) demonstrated that focusing on relevant dimension of sequence was necessary for artificial grammar learning to occur. Participants were exposed to sequences of stimuli that varied in two dimensions (colours and letters) i.e. letters were presented on coloured backgrounds. The order of background colours and the order of letters were determined by different grammars. In one group the attention was oriented to learning letters, in second - to colours. In both groups the artificial grammar learning process was observed; however, participants learned only the grammar that their attention was directed to. The aim of our studies was to demonstrate which of the grammars (letters or colours) is acquired spontaneously, without explicit instruction. In study 1 we used a very similar paradigm in which we presented coloured letters. We observed a spontaneous learning of letters-order rule and no learning of colours-order rule. To verify if difference between grammar rules complexities affected the above results, we used exactly the same grammar for letters and colours in study 2. We replicated results from study 1. We interpreted these results as spontaneous attention oriented towards the aspect of stimuli that is processed more automatically. Namely, reading is more automatic than naming colours. In our studies the colours belonged to letters, so they may have been perceived as a secondary feature and hence their learning could have been impaired. To avoid this effect, we conducted a third study where colours were presented in the background of black letters. We replicated the pattern of results from both studies. Overall, the studies demonstrate that implicit learning process can be observed in both conditions - with and without explicit instruction to focus on the specific aspect of the stimuli. In case of spontaneous learning people acquire knowledge connected to more automatic processing.
Keywords: Implicit Learning, Artificial Grammar Learning, Motivation
How do we know we are conscious? In search of an integrative model
Michal Wierzchon; Jagiellonian University, Krakow, Poland
Identifying the neurobiological and cognitive mechanisms of consciousness is one of the main challenges of modern science. This task is difficult because we need to simultaneously describe the mechanisms that enable access to conscious content and cause the subjective nature of the experience of this content. Previously proposed scientific models of consciousness are often limited to visual perception and focus on investigating the threshold of (most often visual) awareness rather than the problem of the subjective character of conscious experience. There is no widely accepted, wide-ranging theory that describes these mechanisms’ underlying access not only to visual, but also to other types of conscious content (i.e. related to other senses and memory). It seems that a theoretical model is needed that will compromise the philosophical and neurobiological approaches that are trying to address both issues. Here, I propose that this could be achieved with a unification of theories and methods proposed in the context of (neurobiological) cognitive studies of consciousness and I offer a new hierarchical model: a dynamic level interaction model that offers an integrative view on the subject.
Keywords: Consciousness, Higher-Order Theories Of Awareness, Perception, Memory
Implicit grammar learning from social context
Albertyna Paciorek , Małgorzata Foryś-Nogala; Pedagogical University of Krakow, Krakow, Poland; University of Warsaw, Warsaw, Poland
Recent advances in implicit learning show that it is not limited to form-level regularities but can also lead to meaning-based generalizations. One case where implicit learning is notoriously difficult to demonstrate is the learning of definiteness, i.e., the distinction between ‘the’ and ‘a’ in English. Crucially, this distinction is discourse based and requires evaluation of shared knowledge between speaker and listener. Previous attempts, which mainly relied on written text stimuli, may have failed to show implicit learning of such a regularity either because context-dependent form-meaning associations are unlearnable at the implicit level, or because shared knowledge was not salient enough from the contexts provided. We propose a learning adaptation of the false-belief “Sally Ann” task typically used to investigate social cognition. In this task monitoring what belongs to shared and non-shared knowledge occurs naturally and without intention. Four novel articles are inserted into English dialogues, ostensibly encoding distance relations, but underlyingly correlated with definiteness. Here we investigate whether the definiteness correlation can be learned implicitly by advanced Polish speakers of English using false-belief scenarios, as measured by performance in a forced choice task with subjective measures, hence demonstrating implicit learning of a discourse-based form-meaning mapping.
Keywords: Implicit Generalisation, Semantic Implicit Learning, Form-Meaning Mapping, Language, False-Belief
Regularity extraction in a simple letter naming task
Arnaud Rey, Laure Tosatto, Louisa Bogaerts; CNRS & Aix-Marseille University, Marseille, France; Hebrew University of Jerusalem, Jerusalem, Israel
In a simple letter naming task, participants are instructed to name single letters presented on a computer screen. In the present set of experiments, participants were not informed that regularities were inserted in the sequence of presented letters. These regularities corresponded to a triplet of letters (e.g., A-B-C) that occurred systematically in that order. The target triplet was repeated several times and random letters were inserted between two repetitions of the triplet. By varying the nature of the triplet (i.e., a triplet of vowels among random consonants or a triplet of consonants among other random consonants) and the amount of interfering information (i.e., the number of random letters between two repetitions of the triplet), we explored the conditions under which regularity extraction was possible, both implicitly (by observing, over several repetitions of the triplet, the decrease of naming onset latencies on the second and third letters of the triplet, i.e., B and C) and explicitly (by asking participants if they noticed the repeated triplet and if they could produce it). The results indicate that this simple task provide an additional online measure of regularity extraction that will help us testing models of statistical and implicit learning.
Keywords: Regularity Extraction, Online Measure, Letter Naming
Not sleep per se, but sleep-specific oscillations are associated with the consolidation of higher order sequence learning
Péter Simor, Zsófia Zavecz, Karolina Janacsek, Ferenc Gombos, Dezső Németh; Budapest University of Technology and Economics, Budapest, Hungary; Eötvös Loránd University, Budapest, Hungary; Pázmány Péter Catholic University, Budapest, Hungary
Post-learning sleep and sleep-specific neural oscillations can facilitate off-line memory consolidation. Some of these oscillatory patterns might also be functional during a wakeful quiet rest state, however, the influence of wakeful rest on memory consolidation was only scarcely investigated. Furthermore, the beneficial impact of sleep on non-declarative, procedural skills, especially sequence learning is less conclusive. Here, we applied a complex perceptual-motor probabilistic sequence learning task in order to investigate the consolidation of two learning processes: 1) implicit statistical learning, a fundamental mechanism of the brain, which extracts and represents regularities, and 2) explicit sequence learning which is a higher-order type of learning with explicit access to the represented regularities. Young adults (N = 60) after performing the task were randomly allocated into one of three different groups to spend a one-hour off-line period in 1) a relaxed resting state, 2) an active wakeful state, or 3) asleep. EEG power was analyzed throughout the off-line period. On a behavioral level, we have found no differences between these groups: statistical learning and explicit sequence learning was preserved in all groups after the off-line period. Interestingly, however, within the nap group, in frontal electrode sites, spectral power comprising the delta and theta frequency range was positively associated with the consolidation (gain) of explicit sequence learning, but not of statistical learning. Our findings indicate, that sleep-specific cortical oscillations might facilitate the consolidation of sequence learning only if an explicit representation of the sequence structure can be acquired.
Keywords: Sleep, Eeg, Statistical Learning, Sequence Learning, Implicit Learning
Learning and applying two rules in SRT
Mikael Ringstad Hedne, Elisabeth Norman, Mark C. Price; University of Bergen, Bergen, Norway
We present the results from an experiment that builds on and contributes to existing research on the extent to which knowledge acquired in an implicit learning experiment can be strategically controlled. This was explored in an SRT-experiment in which all participants learned two different second order conditional (SOC) sequences. Mong and colleagues (2012) argued that participants displayed strategic control over two sequences using a recognition task with inclusion and exclusion instructions. However, a more robust measure of strategic control would be one which required participants to have trial-by-trial control over the two rules. One example of this is the mixed-block procedure applied in AGL (Norman et al., 2011a, 2011b, 2016), in which the target grammar varies randomly between different classification trials. In the present study, participants were trained on two SOC sequences on alternating blocks. Strategic control was measured on a cued generation task, using a mixed-block procedure where the target sequence varied randomly between trials. The nature of the rule was made ambiguous by introducing distractor dimensions to the sequences, and awareness of the rule was measured by self-report. Results will be reported at the conference.
Keywords: Serial Reaction Time Task, Strategic Control, Consciousness, Rule Awareness
Confidence ratings in implicit learning
Elisabeth Norman, Mikael Ringstad Hedne, Mark C. Price; University of Bergen, Bergen, Norway
In implicit learning, participants’ metacognitive awareness of acquired knowledge is often used to indicate the extent to which knowledge is consciously available. For example, the relationship between confidence ratings and classification accuracy can be used to infer whether knowledge is conscious or unconscious (Dienes & Berry, 1997; Dienes et al., 1995; Norman & Price, 2015). It has been demonstrated that the relationship between confidence and classification accuracy can be influenced by the complexity of the confidence scale used. For example, Tunney and Shanks (2003) found a simple, binary scale to be more sensitive to conscious knowledge than a continuous scale where response alternatives ranged from 50 to 100%. However, a more recent study comparing a broader set of subjective measures of awareness found no systematic differences between awareness scales that included a small versus large number of response alternatives (Wierzchon et al., 2012). We report the results from an experiment in which participants took part in a simple AGL task and each person rated their subjective confidence in their classification responses on a 3-alternative or a continuous visual analog scale. Results will be discussed in relationship to previous findings.
Keywords: Artificial Grammar Learning, Metacognition, Confidence, Consciousness
A matter of conditions: An alternative perspective on procedural memory consolidation processes
Avi Karni; University of Haifa, Haifa, Israel
I argue for the notion that, in line with a current neurobiological perspective, procedural (skill, ‘how to” knowledge) memory consolidation processes (but also the proceduralization of knowledge initially retained by the declarative – MLT dependent - long-term memory system; e.g., systems’ consolidation) necessarily means that long-lasting changes are allowed to occur in the neural substrates that support previously acquired knowledge and skills in cortex. Moreover, mnemonic selectivity is mandatory, because LTM is realized through structural changes, which are difficult to reverse, even in critical low-level neural ensembles relevant to the performance of multiple tasks. Thus, as we develop and mature, procedural memory consolidation (and perhaps proceduralization) processes are more stringently controlled (gated, conditional) but this does not correspond to a decrease in the potential for plasticity per se. More stringent gating is implemented in part through changes in the time-windows in which new knowledge can be tested for relevancy and consistency before it is allowed to consolidate. In some learning conditions, stringent gating may lead to less-than-expected performance gains. Thus learning conditions optimized to generate LTM in typical children or young adults may not be suitable for others (older age-groups, developmental or acquired impairments) for expressing their full memory consolidation potential. However, one would expect to find learning conditions wherein individuals with what we tend to assume are ‘deficits in procedural memory’ (elderly, ADHD, SLI) may have a mnemonic advantage over typical children and young adults.