What's statistical about learning? Insights from modelling statistical learning as a set of memory processes

181

199

One contribution of 13 to a theme issue 'New frontiers for statistical learning in the cognitive sciences'. In recent years, statistical learning (SL) research has seen a growing interest in tracking individual performance in SL tasks, mainly as a predictor of linguistic abilities. We review studies from this line of research and outline three presuppositions underlying the experimental approach they employ: (i) that SL is a unified theoretical construct; (ii) that current SL tasks are interchangeable, and equally valid for assessing SL ability; and (iii) that performance in the standard forced-choice test in the task is a good proxy of SL ability. We argue that these three critical presuppositions are subject to a number of theoretical and empirical issues. First, SL shows patterns of modality-and informationalspecificity, suggesting that SL cannot be treated as a unified construct. Second, different SL tasks may tap into separate sub-components of SL that are not necessarily interchangeable. Third, the commonly used forced-choice tests in most SL tasks are subject to inherent limitations and confounds. As a first step, we offer a methodological approach that explicitly spells out a potential set of different SL dimensions, allowing for better transparency in choosing a specific SL task as a predictor of a given linguistic outcome. We then offer possible methodological solutions for better tracking and measuring SL ability. Taken together, these discussions provide a novel theoretical and methodological approach for assessing individual differences in SL, with clear testable predictions.This article is part of the themed issue 'New frontiers for statistical learning in the cognitive sciences'.

Section: Endnotesmentioning

confidence: 99%

Towards a theory of individual differences in statistical learning

Siegelman

Bogaerts

Christiansen

et al. 2017

181

199

“…Although adults acquire a second language with sufficient exposure [87], I argue here that adults are likely to rely on different mechanisms than infants for memory formation, contrary to views proposing continuity in SL mechanisms. With that said, Thiessen [88] points to factors that explain differences in infant and adult SL that do not require different memory mechanisms (also see [89]). For one, infants may encode statistical information in a noisier representation than adults, and thus, may have more difficulty updating older representations with new information.…”

Section: Development Of Memory Systemsmentioning

confidence: 99%

Do infants retain the statistics of a statistical learning experience? Insights from a developmental cognitive neuroscience perspective

Gómez

2017

One contribution of 13 to a theme issue 'New frontiers for statistical learning in the cognitive sciences'. Statistical structure abounds in language. Human infants show a striking capacity for using statistical learning (SL) to extract regularities in their linguistic environments, a process thought to bootstrap their knowledge of language. Critically, studies of SL test infants in the minutes immediately following familiarization, but long-term retention unfolds over hours and days, with almost no work investigating retention of SL. This creates a critical gap in the literature given that we know little about how single or multiple SL experiences translate into permanent knowledge. Furthermore, different memory systems with vastly different encoding and retention profiles emerge at different points in development, with the underlying memory system dictating the fidelity of the memory trace hours later. I describe the scant literature on retention of SL, the learning and retention properties of memory systems as they apply to SL, and the development of these memory systems. I propose that different memory systems support retention of SL in infant and adult learners, suggesting an explanation for the slow pace of natural language acquisition in infancy. I discuss the implications of developing memory systems for SL and suggest that we exercise caution in extrapolating from adult to infant properties of SL.This article is part of the themed issue 'New frontiers for statistical learning in the cognitive sciences'.

“…the episodic experience of the intact onion is bound through spatio-temporal continuity to the episodic experience of the chopped onion and, subsequently, to that of the onion being fried; e.g. [53]) 5 .…”

Section: From Semantic Memory To Episodic Realizationmentioning

confidence: 99%

“…Theirs is a model of visual attention, (visual) working memory, and the attentional blink [52], and is not intended to address the semantic type-episodic token distinction as used here. 5 In fact, spatio-temporal continuity is not required to explain object persistence across change: if the transition from intact to chopped is occluded, the onion in its chopped state will activate semantic knowledge of onions, in general, which will re-activate the episodic memory of the previously seen intact onion (its recency gives it prepotency in respect of its activation state). This latter representation will, by virtue of its co-activation with the currently seen chopped onion, become associated through time with the chopped onion.…”

Section: Endnotesmentioning

confidence: 99%

Abstraction and generalization in statistical learning: implications for the relationship between semantic types and episodic tokens

Altmann¹

2017

Statistical approaches to emergent knowledge have tended to focus on the process by which experience of individual episodes accumulates into generalizable experience across episodes. However, there is a seemingly opposite, but equally critical, process that such experience affords: the process by which, from a space of types (e.g. onions-a semantic class that develops through exposure to individual episodes involving individual onions), we can perceive or create, on-the-fly, a specific token (a specific onion, perhaps one that is chopped) in the absence of any prior perceptual experience with that specific token. This article reviews a selection of statistical learning studies that lead to the speculation that this process-the generation, on the basis of semantic memory, of a novel episodic representation-is itself an instance of a statistical, in fact associative, process. The article concludes that the same processes that enable statistical abstraction across individual episodes to form semantic memories also enable the generation, from those semantic memories, of representations that correspond to individual tokens, and of novel episodic facts about those tokens. Statistical learning is a window onto these deeper processes that underpin cognition.This article is part of the themed issue 'New frontiers for statistical learning in the cognitive sciences'.