Stimulus set size and statistical coverage of the grammar in artificial grammar learning

Applying different levels of practice variability for motor learning: More is not better

Background

Variable practice is a broadly used tool to improve motor learning processes. However, controversial results can be found in literature about the success of this type of practice compared to constant practice. This study explored one potential reason for this controversy: the manipulation of variable practice load applied during practice and its effects according to the initial performance level and the initial intrinsic variability of the learner.

Method

Sixty-five participants were grouped into four practice schedules to learn a serial throwing task, in which the training load of variable practice was manipulated: one constant practice group and three groups with different variable practice loads applied. After a pre-test, participants trained for 2 weeks. A post-test and three retests (96 h, 2 weeks and 1 month) were carried out after training. The participants' throwing accuracy was assessed through error parameters and their initial intrinsic motor variability was assessed by the autocorrelation coefficient of the error.

Results

The four groups improved their throwing performance. Pairwise comparisons and effect sizes showed larger error reduction in the low variability group. Different loads of variable practice seem to induce different performance improvements in a throwing task. The modulation of the variable practice load seems to be a step forward to clarify the controversy about its benefits, but it has to be guided by the individuals' features, mainly by the initial intrinsic variability of the learner.

Feature variability determines specificity and transfer in multiorientation feature detection learning

Historically, in many perceptual learning experiments, only a single stimulus is practiced, and learning is often specific to the trained feature. Our prior work has demonstrated that multi-stimulus learning (e.g., training-plus-exposure procedure) has the potential to achieve generalization. Here, we investigated two important characteristics of multi-stimulus learning, namely, roving and feature variability, and their impacts on multi-stimulus learning and generalization. We adopted a feature detection task in which an oddly oriented target bar differed by 16° from the background bars. The stimulus onset asynchrony threshold between the target and the mask was measured with a staircase procedure. Observers were trained with four target orientation search stimuli, either with a 5° deviation (30°-35°-40°-45°) or with a 45° deviation (30°-75°-120°-165°), and the four reference stimuli were presented in a roving manner. The transfer of learning to the swapped target-background orientations was evaluated after training. We found that multi-stimulus training with a 5° deviation resulted in significant learning improvement, but learning failed to transfer to the swapped target-background orientations. In contrast, training with a 45° deviation slowed learning but produced a significant generalization to swapped orientations. Furthermore, a modified training-plus-exposure procedure, in which observers were trained with four orientation search stimuli with a 5° deviation and simultaneously passively exposed to orientations with high feature variability (45° deviation), led to significant orientation learning generalization. Learning transfer also occurred when the four orientation search stimuli with a 5° deviation were presented in separate blocks. These results help us to specify the condition under which multistimuli learning produces generalization, which holds potential for real-world applications of perceptual learning, such as vision rehabilitation and expert training.

Stimulus variation-based training enhances artificial grammar learning

The current study was designed to explore whether statistical learning ability is affected by the diversity of the stimulus set used in the training phase. The effect of stimulus diversity was assessed by controlling and manipulating the number of exposures to a given set and the number of unique strings presented to the learner during the training phase. 147 students participated in two studies. In the unvaried stimulus study, 71 participants learned the same basic set of 15 exemplars, once(15 × 1 exposure), twice (15 × 2 exposures = 30 total strings) and 3 times (15 × 3 exposures = 45 total strings). In the varied stimulus study, 75 participants learned 15, 30 and 45, all of which were unique, unrepeated exemplars. All groups were asked to classify test strings for their grammaticality following training. Results of the d' measures in the unvaried stimulus study indicate similar performance across the groups. Conversely, the results of the varied stimulus study show that the group presented with 45 unique strings performed significantly better than the baseline group (15 strings). Analysis of the differences across the equivalent groups in the two studies (15 × 2 exposures vs. 30 unique strings and 15 × 3 exposures vs. 45 unique strings) indicates differences in performance only between the group who was presented with the same 15 strings three times and the group presented with 45 unrepeated strings. Taken together, our results shed additional light on the central role of stimulus variation in Artificial Grammar Learning.

Implicit learning of two artificial grammars

This study investigated the implicit learning of two artificial systems. Two finite-state grammars were implemented with the same tone set (leading to short melodies) and played by the same timbre in exposure and test phases. The grammars were presented in separate exposure phases, and potentially acquired knowledge was tested with two experimental tasks: a grammar categorization task (Experiment 1) and a grammatical error detection task (Experiment 2). Results showed that participants were able to categorize new items as belonging to one or the other grammar (Experiment 1) and detect grammatical errors in new sequences of each grammar (Experiment 2). Our findings suggest the capacity of intra-modal learning of regularities in the auditory modality and based on stimuli that share the same perceptual properties.

Under What Conditions Can Recursion Be Learned? Effects of Starting Small in Artificial Grammar Learning of Center-Embedded Structure

It has been suggested that external and/or internal limitations paradoxically may lead to superior learning, that is, the concepts of starting small and less is more (Elman, 1993; Newport, 1990). In this paper, we explore the type of incremental ordering during training that might help learning, and what mechanism explains this facilitation. We report four artificial grammar learning experiments with human participants. In Experiments 1a and 1b we found a beneficial effect of starting small using two types of simple recursive grammars: right‐branching and center‐embedding, with recursive embedded clauses in fixed positions and fixed length. This effect was replicated in Experiment 2 (N = 100). In Experiment 3 and 4, we used a more complex center‐embedded grammar with recursive loops in variable positions, producing strings of variable length. When participants were presented an incremental ordering of training stimuli, as in natural language, they were better able to generalize their knowledge of simple units to more complex units when the training input “grew” according to structural complexity, compared to when it “grew” according to string length. Overall, the results suggest that starting small confers an advantage for learning complex center‐embedded structures when the input is organized according to structural complexity.

Implicit learning is order dependent

We report two experiments using the artificial-grammar task that demonstrate order dependence in implicit learning. Studying grammatical training strings in different orders did not affect participants' discrimination of grammatical from ungrammatical test strings, but it did affect their judgments about specific test strings. Current accounts of learning in the artificial-grammar task focus on category-level discrimination and largely ignore item-level discrimination. Hence, the results highlight the importance of moving theory from a category- to an item-level of analysis and point to a new way to evaluate and to refine accounts of implicit learning.

Input Variability Facilitates Unguided Subcategory Learning in Adults

PURPOSE

This experiment investigated whether input variability would affect initial learning of noun gender subcategories in an unfamiliar, natural language (Russian), as it is known to assist learning of other grammatical forms.

METHOD

Forty adults (20 men, 20 women) were familiarized with examples of masculine and feminine Russian words. Half of the participants were familiarized with 32 different root words in a high-variability condition. The other half were familiarized with 16 different root words, each repeated twice for a total of 32 presentations in a high-repetition condition. Participants were tested on untrained members of the category to assess generalization. Familiarization and testing was completed 2 additional times.

RESULTS

Only participants in the high-variability group showed evidence of learning after an initial period of familiarization. Participants in the high-repetition group were able to learn after additional input. Both groups benefited when words included 2 cues to gender compared to a single cue.

CONCLUSIONS

The results demonstrate that the degree of input variability can influence learners' ability to generalize a grammatical subcategory (noun gender) from a natural language. In addition, the presence of multiple cues to linguistic subcategory facilitated learning independent of variability condition.

Information theory and artificial grammar learning: inferring grammaticality from redundancy

In artificial grammar learning experiments, participants study strings of letters constructed using a grammar and then sort novel grammatical test exemplars from novel ungrammatical ones. The ability to distinguish grammatical from ungrammatical strings is often taken as evidence that the participants have induced the rules of the grammar. We show that judgements of grammaticality are predicted by the local redundancy of the test strings, not by grammaticality itself. The prediction holds in a transfer test in which test strings involve different letters than the training strings. Local redundancy is usually confounded with grammaticality in stimuli widely used in the literature. The confounding explains why the ability to distinguish grammatical from ungrammatical strings has popularized the idea that participants have induced the rules of the grammar, when they have not. We discuss the judgement of grammaticality task in terms of attribute substitution and pattern goodness. When asked to judge grammaticality (an inaccessible attribute), participants answer an easier question about pattern goodness (an accessible attribute).

Learning simple and complex artificial grammars in the presence of a semantic reference field: effects on performance and awareness

This study investigated whether the negative effect of complexity on artificial grammar learning could be compensated by adding semantics. Participants were exposed to exemplars from a simple or a complex finite state grammar presented with or without a semantic reference field. As expected, performance on a grammaticality judgment test was higher for the simple grammar than for the complex grammar. For the simple grammar, the results also showed that participants presented with a reference field and instructed to decode the meaning of each exemplar (decoding condition) did better than participants who memorized the exemplars without semantic referents (memorize condition). Contrary to expectations, however, there was no significant difference between the decoding condition and the memorize condition for the complex grammar. These findings indicated that the negative effect of complexity remained, despite the addition of semantics. To clarify how the presence of a reference field influenced the learning process, its effects on the acquisition of two types of knowledge (first- and second-order dependencies) and on participants' awareness of their knowledge were examined. The results tentatively suggested that the reference field enhanced the learning of second-order dependencies. In addition, participants in the decoding condition realized when they had knowledge relevant to making a grammaticality judgment, whereas participants in the memorize condition demonstrated some knowledge of which they were unaware. These results are in line with the view that the reference field enhanced structure learning by making certain dependencies more salient. Moreover, our findings stress the influence of complexity on artificial grammar learning.

Exemplar variability facilitates rapid learning of an otherwise unlearnable grammar by individuals with language-based learning disability

PURPOSE

Even without explicit instruction, learners are able to extract information about the form of a language simply by attending to input that reflects the underlying grammar. In this study, the authors explored the role of variability in this learning by asking whether varying the number of unique exemplars heard by the learner affects learning of an artificial syntactic form.

METHOD

Learners with normal language (n = 16) and language-based learning disability (LLD; n = 16) were exposed to strings of nonwords that represented an underlying grammar. Half of the learners heard 3 exemplars 16 times each (low variability group), and the other half of the learners heard 24 exemplars twice each (high variability group). Learners were then tested for recognition of items heard and generalization of the grammar with new nonword strings.

RESULTS

Only those learners with LLD who were in the high variability group were able to demonstrate generalization of the underlying grammar. For learners with normal language, both those in the high and the low variability groups showed generalization of the grammar, but relative effect sizes suggested a larger learning effect in the high variability group.

CONCLUSION

The results demonstrate that the structure of the learning context can determine the ability to generalize from specific training items to novel cases.

How semantic biases in simple adjacencies affect learning a complex structure with non-adjacencies in AGL: a statistical account

A major theoretical debate in language acquisition research regards the learnability of hierarchical structures. The artificial grammar learning methodology is increasingly influential in approaching this question. Studies using an artificial centre-embedded A(n)B(n) grammar without semantics draw conflicting conclusions. This study investigates the facilitating effect of distributional biases in simple AB adjacencies in the input sample--caused in natural languages, among others, by semantic biases-on learning a centre-embedded structure. A mathematical simulation of the linguistic input and the learning, comparing various distributional biases in AB pairs, suggests that strong distributional biases might help us to grasp the complex A(n)B(n) hierarchical structure in a later stage. This theoretical investigation might contribute to our understanding of how distributional features of the input--including those caused by semantic variation--help learning complex structures in natural languages.

An entropy model for artificial grammar learning

A model is proposed to characterize the type of knowledge acquired in artificial grammar learning (AGL). In particular, Shannon entropy is employed to compute the complexity of different test items in an AGL task, relative to the training items. According to this model, the more predictable a test item is from the training items, the more likely it is that this item should be selected as compatible with the training items. The predictions of the entropy model are explored in relation to the results from several previous AGL datasets and compared to other AGL measures. This particular approach in AGL resonates well with similar models in categorization and reasoning which also postulate that cognitive processing is geared towards the reduction of entropy.