Regularization of languages by adults and children: A mathematical framework

Predictability and Variation in Language Are Differentially Affected by Learning and Production

General principles of human cognition can help to explain why languages are more likely to have certain characteristics than others: structures that are difficult to process or produce will tend to be lost over time. One aspect of cognition that is implicated in language use is working memory-the component of short-term memory used for temporary storage and manipulation of information. In this study, we consider the relationship between working memory and regularization of linguistic variation. Regularization is a well-documented process whereby languages become less variable (on some dimension) over time. This process has been argued to be driven by the behavior of individual language users, but the specific mechanism is not agreed upon. Here, we use an artificial language learning experiment to investigate whether limitations in working memory during either language learning or language production drive regularization behavior. We find that taxing working memory during production results in the loss of all types of variation, but the process by which random variation becomes more predictable is better explained by learning biases. A computational model offers a potential explanation for the production effect using a simple self-priming mechanism.

Taking the chance!-Interindividual differences in rule-breaking

While some individuals tend to follow norms, others, in the face of tempting but forbidden options, tend to commit rule-breaking when this action is beneficial for themselves. Previous studies have neglected such interindividual differences in rule-breaking. The present study fills this gap by investigating cognitive characteristics of individuals who commit spontaneous deliberative rule-breaking (rule-breakers) versus rule-followers. We developed a computerised task, in which 133 participants were incentivised to sometimes violate set rules which would-if followed-lead to a loss. While 52% of participants tended to break rules to obtain a benefit, 48% tended to follow rules even if this behaviour led to loss. Although rule-breakers experienced significantly more cognitive conflict (measured via response times and mouse movement trajectories) than rule-followers, they also obtained higher payoffs. In rule-breakers, cognitive conflict was more pronounced when violating the rules than when following them, and mainly during action planning. This conflict increased with frequent, recurrent, and early rule-breaking. Our results were in line with the Decision-Implementation-Mandatory switch-Inhibition model and thus extend the application of this model to the interindividual differences in rule-breaking. Furthermore, personality traits such as extroversion, disagreeableness, risk propensity, high impulsiveness seem to play a role in the appreciation of behaviours and cognitive characteristics of rule-followers and rule-breakers. This study opens the path towards the understanding of the cognitive characteristics of the interindividual differences in responses towards rules, and especially in spontaneous deliberative rule-breaking.

How children learn to communicate discriminatively

How do children learn to communicate, and what do they learn? Traditionally, most theories have taken an associative, compositional approach to these questions, supposing children acquire an inventory of form-meaning associations, and procedures for composing / decomposing them; into / from messages in production and comprehension. This paper presents an alternative account of human communication and its acquisition based on the systematic, discriminative approach embodied in psychological and computational models of learning, and formally described by communication theory. It describes how discriminative learning theory offers an alternative perspective on the way that systems of semantic cues are conditioned onto communicative codes, while information theory provides a very different view of the nature of the codes themselves. It shows how the distributional properties of languages satisfy the communicative requirements described in information theory, enabling language learners to align their expectations despite the vastly different levels of experience among language users, and to master communication systems far more abstract than linguistic intuitions traditionally assume. Topics reviewed include morphological development, the acquisition of verb argument structures, and the functions of linguistic systems that have proven to be stumbling blocks for compositional theories: grammatical gender and personal names.

Object-Label-Order Effect When Learning From an Inconsistent Source

Learning in natural environments is often characterized by a degree of inconsistency from an input. These inconsistencies occur, for example, when learning from more than one source, or when the presence of environmental noise distorts incoming information; as a result, the task faced by the learner becomes ambiguous. In this study, we investigate how learners handle such situations. We focus on the setting where a learner receives and processes a sequence of utterances to master associations between objects and their labels, where the source is inconsistent by design: It uses both "correct" and "incorrect" object-label pairings. We hypothesize that depending on the order of presentation, the result of the learning may be different. To this end, we consider two types of symbolic learning procedures: the Object-Label (OL) and the Label-Object (LO) process. In the OL process, the learner is first exposed to the object, and then the label. In the LO process, this order is reversed. We perform experiments with human subjects, and also construct a computational model that is based on a nonlinear stochastic reinforcement learning algorithm. It is observed experimentally that OL learners are generally better at processing inconsistent input compared to LO learners. We show that the patterns observed in the learning experiments can be reproduced in the simulations if the model includes (a) an ability to regularize the input (and also to do the opposite, i.e., undermatch) and (b) an ability to take account of implicit negative evidence (i.e., interactions among different objects/labels). The model suggests that while both types of learners utilize implicit negative evidence in a similar way, there is a difference in regularization patterns: OL learners regularize the input, whereas LO learners undermatch. As a result, OL learners are able to form a more consistent system of image-utterance associations, despite the ambiguous learning task.

Reconsidering retrieval effects on adult regularization of inconsistent variation in language

The phenomenon of regularization - learners imposing systematicity on inconsistent variation in language input - is complex. Studies show that children are more likely to regularize than adults, but adults will also regularize under certain circumstances. Exactly why we see the pattern of behaviour that we do is not well understood, however. This paper reports on an experiment investigating whether it is possible to induce regularization in adults by varying the conditions of learning and/or testing in ways that made retrieval more difficult, something predicted by Hudson Kam and Newport (2009). The data show that interfering with learning does not lead to regularization, in accord with the findings of Perfors (2012), but that interfering with retrieval at test does, although only to a small degree.

Cognitive influences in language evolution: Psycholinguistic predictors of loan word borrowing

Languages change due to social, cultural, and cognitive influences. In this paper, we provide an assessment of these cognitive influences on diachronic change in the vocabulary. Previously, tests of stability and change of vocabulary items have been conducted on small sets of words where diachronic change is imputed from cladistics studies. Here, we show for a substantially larger set of words that stability and change in terms of documented borrowings of words into English and into Dutch can be predicted by psycholinguistic properties of words that reflect their representational fidelity. We found that grammatical category, word length, age of acquisition, and frequency predict borrowing rates, but frequency has a non-linear relationship. Frequency correlates negatively with probability of borrowing for high-frequency words, but positively for low-frequency words. This borrowing evidence documents recent, observable diachronic change in the vocabulary enabling us to distinguish between change associated with transmission during language acquisition and change due to innovations by proficient speakers.

Spatial evolution of regularization in learned behavior of animals

Stochastic population dynamics of learned traits are studied, where individual learners behave according to a reinforcement learner model, which is a nonlinear version of the Bush-Mosteller model. Depending on a regularization parameter (parameter a), the learners may possess different degrees of overmatching (regularization behavior, 0 ≤ a < 1), frequency matching (corresponding to a=1), or undermatching behavior (a > 1). Both non-spatial and spatial models are considered, to study the interplay of individual heterogeneity of behavior, spatial and temporal effects of learning, and the possibility of emergence of regional culture. In non-spatial models, we observe that populations of individuals learning from each other converge to a universally shared, deterministic rule (either rule "1" or rule "0"), only if they to some extent possess the ability to generalize (a < 1). Otherwise, a low-coherence solution where both rules are used intermittently by everyone, is achieved. If the evolution of the regularization ability is included, then we find that a initially evolves toward lower values, and a shared solution is established when everyone reliably uses the same rule. The spatial (2D) model has two well known limiting cases: if a=0 (the strongest degree of regularization), the model converges to a threshold voter model, and if a=1 (frequency matching), it is equivalent to the discrete diffusion equation. If 0 < a < 1 (the case where individuals regularize), spatial patterns emerge, where patches of different usage of the rule are formed. Smaller values of a lead to sharper and longer lived patches. Values of a < 1 close to unity result in probabilistic outcomes where patches only survive if they are attached to the boundary. Analytical treatment of the 1D case reveals the existence of approximate equilibria that have front structure, where spatially intermittent deterministic usage of one and the other rule are separated by interfaces whose analytical form is derived.

Language learning, language use and the evolution of linguistic variation

Linguistic universals arise from the interaction between the processes of language learning and language use. A test case for the relationship between these factors is linguistic variation, which tends to be conditioned on linguistic or sociolinguistic criteria. How can we explain the scarcity of unpredictable variation in natural language, and to what extent is this property of language a straightforward reflection of biases in statistical learning? We review three strands of experimental work exploring these questions, and introduce a Bayesian model of the learning and transmission of linguistic variation along with a closely matched artificial language learning experiment with adult participants. Our results show that while the biases of language learners can potentially play a role in shaping linguistic systems, the relationship between biases of learners and the structure of languages is not straightforward. Weak biases can have strong effects on language structure as they accumulate over repeated transmission. But the opposite can also be true: strong biases can have weak or no effects. Furthermore, the use of language during interaction can reshape linguistic systems. Combining data and insights from studies of learning, transmission and use is therefore essential if we are to understand how biases in statistical learning interact with language transmission and language use to shape the structural properties of language.

This article is part of the themed issue ‘New frontiers for statistical learning in the cognitive sciences’.

Acquiring variation in an artificial language: Children and adults are sensitive to socially conditioned linguistic variation

Languages exhibit sociolinguistic variation, such that adult native speakers condition the usage of linguistic variants on social context, gender, and ethnicity, among other cues. While the existence of this kind of socially conditioned variation is well-established, less is known about how it is acquired. Studies of naturalistic language use by children provide various examples where children's production of sociolinguistic variants appears to be conditioned on similar factors to adults' production, but it is difficult to determine whether this reflects knowledge of sociolinguistic conditioning or systematic differences in the input to children from different social groups. Furthermore, artificial language learning experiments have shown that children have a tendency to eliminate variation, a process which could potentially work against their acquisition of sociolinguistic variation. The current study used a semi-artificial language learning paradigm to investigate learning of the sociolinguistic cue of speaker identity in 6-year-olds and adults. Participants were trained and tested on an artificial language where nouns were obligatorily followed by one of two meaningless particles and were produced by one of two speakers (one male, one female). Particle usage was conditioned deterministically on speaker identity (Experiment 1), probabilistically (Experiment 2), or not at all (Experiment 3). Participants were given tests of production and comprehension. In Experiments 1 and 2, both children and adults successfully acquired the speaker identity cue, although the effect was stronger for adults and in Experiment 1. In addition, in all three experiments, there was evidence of regularization in participants' productions, although the type of regularization differed with age: children showed regularization by boosting the frequency of one particle at the expense of the other, while adults regularized by conditioning particle usage on lexical items. Overall, results demonstrate that children and adults are sensitive to speaker identity cues, an ability which is fundamental to tracking sociolinguistic variation, and that children's well-established tendency to regularize does not prevent them from learning sociolinguistically conditioned variation.

Mathematical Modeling of Learning from an Inconsistent Source: A Nonlinear Approach

Continuing the discussion of how children can modify and regularize linguistic inputs from adults, we present a new interpretation of existing algorithms to model and investigate the process of a learner learning from an inconsistent source. On the basis of this approach is a (possibly nonlinear) function (the update function) that relates the current state of the learner with an increment that it receives upon processing the source's input, in a sequence of updates. The model can be considered a nonlinear generalization of the classic Bush-Mosteller algorithm. Our model allows us to analyze and present a theoretical explanation of a frequency boosting property, whereby the learner surpasses the fluency of the source by increasing the frequency of the most common input. We derive analytical expressions for the frequency of the learner, and also identify a class of update functions that exhibit frequency boosting. Applications to the Feature-Label-Order effect in learning are presented.

Personalized learning: From neurogenetics of behaviors to designing optimal language training

Variability in drug responsivity has prompted the development of Personalized Medicine, which has shown great promise in utilizing genotypic information to develop safer and more effective drug regimens for patients. Similarly, individual variability in learning outcomes has puzzled researchers who seek to create optimal learning environments for students. "Personalized Learning" seeks to identify genetic, neural and behavioral predictors of individual differences in learning and aims to use predictors to help create optimal teaching paradigms. Evidence for Personalized Learning can be observed by connecting research in pharmacogenomics, cognitive genetics and behavioral experiments across domains of learning, which provides a framework for conducting empirical studies from the laboratory to the classroom and holds promise for addressing learning effectiveness in the individual learners. Evidence can also be seen in the subdomain of speech learning, thus providing initial support for the applicability of Personalized Learning to language.

Reinforcement Learning Explains Conditional Cooperation and Its Moody Cousin

Direct reciprocity, or repeated interaction, is a main mechanism to sustain cooperation under social dilemmas involving two individuals. For larger groups and networks, which are probably more relevant to understanding and engineering our society, experiments employing repeated multiplayer social dilemma games have suggested that humans often show conditional cooperation behavior and its moody variant. Mechanisms underlying these behaviors largely remain unclear. Here we provide a proximate account for this behavior by showing that individuals adopting a type of reinforcement learning, called aspiration learning, phenomenologically behave as conditional cooperator. By definition, individuals are satisfied if and only if the obtained payoff is larger than a fixed aspiration level. They reinforce actions that have resulted in satisfactory outcomes and anti-reinforce those yielding unsatisfactory outcomes. The results obtained in the present study are general in that they explain extant experimental results obtained for both so-called moody and non-moody conditional cooperation, prisoner’s dilemma and public goods games, and well-mixed groups and networks. Different from the previous theory, individuals are assumed to have no access to information about what other individuals are doing such that they cannot explicitly use conditional cooperation rules. In this sense, myopic aspiration learning in which the unconditional propensity of cooperation is modulated in every discrete time step explains conditional behavior of humans. Aspiration learners showing (moody) conditional cooperation obeyed a noisy GRIM-like strategy. This is different from the Pavlov, a reinforcement learning strategy promoting mutual cooperation in two-player situations.

Laboratory experiments using human participants have shown that, in groups or contact networks, humans often behave as conditional cooperator or its moody variant. Although conditional cooperation in dyadic interaction is well understood, mechanisms underlying these behaviors in group or networks beyond a pair of individuals largely remain unclear. In this study, we show that players adopting a type of reinforcement learning exhibit these conditional cooperation behaviors. The results are general in the sense that the model explains experimental results to date obtained in various situations. It explains moody conditional cooperation, which is a recently discovered behavioral trait of humans, in addition to traditional conditional cooperation. It also explains experimental results obtained with both the prisoner’s dilemma and public goods games and with different population structure. Crucially, our model assumes that individuals do not have access to information about what other individuals are doing such that they cannot explicitly condition their behavior on how many others have previously cooperated. Thus, our results provide a proximate and unified understanding of these experimentally observed patterns.