Search

In many domains, learners extract recurring units from continuous sequences. For example, in unknown languages, fluent speech is perceived as a continuous signal. Learners need to extract the underlying words from this continuous signal and then memorize them. One prominent candidate mechanism is statistical learning, whereby learners track how predictive syllables (or other items) are of one another. Syllables within the same word predict each other better than syllables straddling word boundaries. But does statistical learning lead to memories of the underlying words—or just to pairwise associations among syllables? Electrophysiological results provide the strongest evidence for the memory view. Electrophysiological responses can be time‐locked to statistical word boundaries (e.g., N400s) and show rhythmic activity with a periodicity of word durations. Here, I reproduce such results with a simple Hebbian network. When exposed to statistically structured syllable sequences (and when the underlying words are not excessively long), the network activation is rhythmic with the periodicity of a word duration and activation maxima on word‐final syllables. This is because word‐final syllables receive more excitation from earlier syllables with which they are associated than less predictable syllables that occur earlier in words. The network is also sensitive to information whose electrophysiological correlates were used to support the encoding of ordinal positions within words. Hebbian learning can thus explain rhythmic neural activity in statistical learning tasks without any memory representations of words. Learners might thus need to rely on cues beyond statistical associations to learn the words of their native language. Research Highlights: Statistical learning may be utilized to identify recurring units in continuous sequences (e.g., words in fluent speech) but may not generate explicit memory for words.Exposure to statistically structured sequences leads to rhythmic activity with a period of the duration of the underlying units (e.g., words).I show that a memory‐less Hebbian network model can reproduce this rhythmic neural activity as well as putative encodings of ordinal positions observed in earlier research.Direct tests are needed to establish whether statistical learning leads to declarative memories for words. [ABSTRACT FROM AUTHOR]

In recent years, Bayesian learning models have been applied to an increasing variety of domains. While such models have been criticized on theoretical grounds, the underlying assumptions and predictions are rarely made concrete and tested experimentally. Here, I use Frank and Tenenbaum's (2011) Bayesian model of rule-learning as a case study to spell out the underlying assumptions, and to confront them with the empirical results Frank and Tenenbaum (2011) propose to simulate, as well as with novel experiments. While rule-learning is arguably well suited to rational Bayesian approaches, I show that their models are neither psychologically plausible nor ideal observer models. Further, I show that their central assumption is unfounded: humans do not always preferentially learn more specific rules, but, at least in some situations, those rules that happen to be more salient. Even when granting the unsupported assumptions, I show that all of the experiments modeled by Frank and Tenenbaum (2011) either contradict their models, or have a large number of more plausible interpretations. I provide an alternative account of the experimental data based on simple psychological mechanisms, and show that this account both describes the data better, and is easier to falsify. I conclude that, despite the recent surge in Bayesian models of cognitive phenomena, psychological phenomena are best understood by developing and testing psychological theories rather than models that can be fit to virtually any data. (Contains 1 table and 4 figures.)

Is formal simplicity a guide to learning in humans, as simplicity is said to be a guide to the acceptability of theories in science? Does simplicity determine the difficulty of various learning tasks? I argue that, similarly to how scientists sometimes preferred complex theories when this facilitated calculations, results from perception, learning and reasoning suggest that formal complexity is generally unrelated to what is easy to learn and process by humans, and depends on assumptions about available representational and processing primitives. "Simpler" hypotheses are preferred only when they are also easier to process. Historically, "simpler", easier‐to‐process, scientific theories might also be preferred if they are transmitted preferentially. Empirically viable complexity measures should build on the representational and processing primitives of actual learners, even if explanations of their behaviour become formally more complex. [ABSTRACT FROM AUTHOR]

When other individuals move, we interpret their movements as discrete, hierarchically-organized, goal-directed actions. However, the mechanisms that integrate visible movement features into actions are poorly understood. Here, we consider two sequence learning mechanisms--transitional probability-based (TP) and position-based encoding computations--that have been studied extensively in the domain of language learning, and investigate their potential for integrating movements into actions. If these learning mechanisms integrate movements into actions, then they should create memory units that contain (i) movement information, (ii) information about the order in which movements occurred, and (iii) information allowing actions to be recognized from different viewpoints. We show that both mechanisms retain movement information. However, only the position-based mechanism creates movement representations that are view-invariant and contain order information. The TP-based mechanism creates movement representations that are view-dependent and contain no order information. We therefore suggest that the TP-based mechanism is unlikely to play an important role for integrating movements into actions. In contrast, the position-based mechanism retains some of the types of information needed to represent goal-directed actions, which makes it an attractive target for further research to explore what, if any, role it plays in the perception of goal-directed actions. (Contains 12 figures and 2 tables.)

Rules, and exceptions to such rules, are ubiquitous in many domains, including language. Here we used simple artificial grammars to investigate the influence of 2 factors on the acquisition of rules and their exceptions, namely type frequency (the relative numbers of different exceptions to different regular items) and token frequency (the number of exception tokens relative to the number of regular tokens). We familiarized participants to either a prefixation pattern (where regulars started with /ZaI/ and exceptions ended with /ZaI/) or a suffixation pattern (where regulars ended with /ZaI/ and exceptions started with /ZaI/). We show that the type and the token frequency of regular items and exceptions influence in different ways what participants can learn. For the exceptions to be learned, they have to occur sufficiently often so that participants can memorize them; this can be achieved by a high token frequency. However, a high token frequency of the exceptions also impaired the acquisition of the "regular" pattern. In contrast, the type frequency of the patterns seemed to determine whether the regular pattern could be learned: When the type frequency of the regular items was sufficiently high, participants successfully learned the regular pattern even when the exceptions were played so often that 66% of the familiarization items were exceptions. We discuss these findings in the context of general learning mechanisms and the role they may play in language acquisition. (Contains 2 tables, 1 footnote and 12 figures.)

While content words (e.g., 'dog') tend to carry meaning, function words (e.g., 'the') mainly serve syntactic purposes. Here, we ask whether 17-month old infants can use one language-universal cue to identify function word candidates: their high frequency of occurrence. In Experiment 1, infants listened to a series of short, naturally recorded sentences in a foreign language (i.e., in French). In these sentences, two determiners appeared much more frequently than any content word. Following this, infants were presented with a visual object, and simultaneously with a word pair composed of a determiner and a noun. Results showed that infants associated the object more strongly with the infrequent noun than with the frequent determiner. That is, when presented with both the old object and a novel object, infants were more likely to orient towards the old object when hearing a label with a new determiner and the old noun compared to a label with a new noun and the old determiner. In Experiment 2, infants were tested using the same procedure as in Experiment 1, but without the initial exposure to French sentences. Under these conditions, infants did not preferentially associate the object with nouns, suggesting that the preferential association between nouns and objects does not result from specific acoustic or phonological properties. In line with various biases and heuristics involved in acquiring content words, we provide the first direct evidence that infants can use distributional cues, especially the high frequency of occurrence, to identify potential function words. (Contains 6 figures and 2 tables.)

Structural regularities in language have often been attributed to symbolic or statistical general purpose computations, whereas perceptual factors influencing such generalizations have received less interest. Here, we use phonotactic-like constraints as a case study to ask whether the structural properties of specific perceptual and memory mechanisms may facilitate the acquisition of grammatical-like regularities. Participants learned that the consonants C[subscript 1] and C[subscript 2] had to come from distinct sets in words of the form C[subscript 1]VccVC[subscript 2] (where the critical consonants were in word "edges") but not in words of the form cVC[subscript 1]C[subscript 2]Vc (where the critical consonants were in word middles). Control conditions ruled out attentional or psychophysical difficulties in word "middles." Participants did, however, learn such regularities in word middles when natural consonant classes were used instead of arbitrary consonant sets. We conclude that positional generalizations may be learned preferentially using edge-based positional codes, but that participants can also use other mechanisms when other linguistic cues are given. (Contains 4 footnotes and 4 figures.)

Word-segmentation, that is, the extraction of words from fluent speech, is one of the first problems language learners have to master. It is generally believed that statistical processes, in particular those tracking "transitional probabilities" (TPs), are important to word-segmentation. However, there is evidence that word forms are stored in memory formats differing from those that can be constructed from TPs, i.e. in terms of the positions of phonemes and syllables within words. In line with this view, we show that TP-based processes leave learners no more familiar with items heard 600 times than with "phantom-words" not heard at all if the phantom-words have the same statistical structure as the occurring items. Moreover, participants are more familiar with phantom-words than with frequent syllable combinations. In contrast, minimal prosody-like perceptual cues allow learners to recognize actual items. TPs may well signal co-occurring syllables; this, however, does not seem to lead to the extraction of word-like units. We review other, in particular prosodic, cues to word-boundaries which may allow the construction of positional memories while not requiring language-specific knowledge, and suggest that their contributions to word-segmentation need to be reassessed. (Contains 9 figures and 5 tables.)

Cognitive processes are often attributed to statistical or symbolic general-purpose mechanisms. Here we show that some spontaneous generalizations are driven by specialized, highly constrained symbolic operations. We explore how two types of artificial grammars are acquired, one based on repetitions and the other on characteristic relations between tones ("ordinal" grammars). Whereas participants readily acquire repetition-based grammars, displaying early electrophysiological responses to grammar violations, they perform poorly with ordinal grammars, displaying no such electrophysiological responses. This outcome is problematic for both general symbolic and statistical models, which predict that both types of grammars should be processed equally easily. This suggests that some simple grammars are acquired using "perceptual primitives" rather than general-purpose mechanisms; such primitives may be elements of a "toolbox" of specialized computational heuristics, which may ultimately allow constructing a psychological theory of symbol manipulation.

To learn a language, speakers must learn its words and rules from fluent speech; in particular, they must learn dependencies among linguistic classes. We show that when familiarized with a short artificial, subliminally bracketed stream, participants can learn relations about the structure of its words, which specify the classes of syllables occurring in first and last word positions. By studying the effect of familiarization length, we compared the general predictions of associative theories of learning and those of models postulating separate mechanisms for quickly extracting the word structure and for tracking the syllable distribution in the stream. As predicted by the dual-mechanism model, the preference for structurally correct items was negatively correlated with the familiarization length. This result is difficult to explain by purely associative schemes; an extensive set of neural network simulations confirmed this difficulty. Still, we show that powerful statistical computations operating on the stream are available to our participants, as they are sensitive to co-occurrence statistics among non-adjacent syllables. We suggest that different learning mechanisms analyze speech on-line: A rapid mechanism extracting structural information about the stream, and a slower mechanism detecting statistical regularities among the items occurring in it.

Recent research suggests that humans and other animals have sophisticated abilities to extract both statistical dependencies and rule-based regularities from sequences. Most of this research stresses the flexibility and generality of such processes. Here the authors take up an equally important project, namely, to explore the limits of such processes. As a case study for rule-based generalizations, the authors demonstrate that only repetition-based structures with repetitions at the edges of sequences (e.g., ABCDEFF but not ABCDDEF) can be reliably generalized, although token repetitions can easily be discriminated at both sequence edges and middles. This finding suggests limits on rule-based sequence learning and new interpretations of earlier work alleging rule learning in infants. Rather than implementing a computerlike, formal process that operates over all patterns equally well, rule-based learning may be a highly constrained and piecemeal process driven by perceptual primitives--specialized type operations that are highly sensitive to perceptual factors.

Previous research suggests that artificial-language learners exposed to quasi-continuous speech can learn that the first and the last syllables of words have to belong to distinct classes (e.g., Endress & Bonatti, 2007; Pea, Bonatti, Nespor, & Mehler, 2002). The mechanisms of these generalizations, however, are debated. Here we show that participants learn such generalizations only when the crucial syllables are in edge positions (i.e., the first and the last), but not when they are in medial positions (i.e., the second and the fourth in pentasyllabic items). In contrast to the generalizations, participants readily perform statistical analyses also in word middles. In analogy to sequential memory, we suggest that participants extract the generalizations using a simple but specific mechanism that encodes the positions of syllables that occur in edges. Simultaneously, they use another mechanism to track the syllable distribution in the speech streams. In contrast to previous accounts, this model explains why the generalizations are faster than the statistical computations, require additional cues, and break down under different conditions, and why they can be performed at all. We also show that that similar edge-based mechanisms may explain many results in artificial-grammar learning and also various linguistic observations.