This paper contrasts two influential theoretical accounts of language change and evolution – Iterated Learning and Social Coordination. The contrast is based on an experiment that compares drawings produced with Garrod et al’s ‘pictionary’ task with those produced in an Iterated Learning version of the same task. The main finding is that Iterated Learning does not lead to the systematic simplification and increased symbolicity of graphical signs produced in the standard interactive version of the (...) task. A second finding is that Iterated Learning leads to less conceptual and structural alignment between participants than observed for those in the interactive condition. The paper concludes with a comparison of the two accounts in relation to how each promotes signs that are effi cient, systematic and learnable. (shrink)

This paper contrasts two influential theoretical accounts of language change and evolution – Iterated Learning and Social Coordination. The contrast is based on an experiment that compares drawings produced with Garrod et al’s ‘pictionary’ task with those produced in an Iterated Learning version of the same task. The main finding is that Iterated Learning does not lead to the systematic simplification and increased symbolicity of graphical signs produced in the standard interactive version of the (...) task. A second finding is that Iterated Learning leads to less conceptual and structural alignment between participants than observed for those in the interactive condition. The paper concludes with a comparison of the two accounts in relation to how each promotes signs that are effi cient, systematic and learnable. (shrink)

A tracking control algorithm of nonlinear multiple agents with undirected communication is studied for each multiagent system affected by external interference and input saturation. A control design scheme combining iterative learning and adaptive control is proposed to perform parameter adaptive time-varying adjustment and prove the effectiveness of the control protocol by designing Lyapunov functions. Simulation results show that the high-precision tracking control problem of the nonlinear multiagent system based on adaptive iterative learning control can be well realized even (...) when the input is saturated. Finally, the validity of the proposed algorithm is verified by numerical analysis. (shrink)

Languages are transmitted from person to person and generation to generation via a process of iterated learning: people learn a language from other people who once learned that language themselves. We analyze the consequences of iterated learning for learning algorithms based on the principles of Bayesian inference, assuming that learners compute a posterior distribution over languages by combining a prior (representing their inductive biases) with the evidence provided by linguistic data. We show that when learners (...) sample languages from this posterior distribution, iterated learning converges to a distribution over languages that is determined entirely by the prior. Under these conditions, iterated learning is a form of Gibbs sampling, a widely-used Markov chain Monte Carlo algorithm. The consequences of iterated learning are more complicated when learners choose the language with maximum posterior probability, being affected by both the prior of the learners and the amount of information transmitted between generations. We show that in this case, iterated learning corresponds to another statistical inference algorithm, a variant of the expectation-maximization (EM) algorithm. These results clarify the role of iterated learning in explanations of linguistic universals and provide a formal connection between constraints on language acquisition and the languages that come to be spoken, suggesting that information transmitted via iterated learning will ultimately come to mirror the minds of the learners. (shrink)

This study first investigates robust iterative learning control issue for a class of two-dimensional linear discrete singular Fornasini–Marchesini systems under iteration-varying boundary states. Initially, using the singular value decomposition theory, an equivalent dynamical decomposition form of 2-D LDSFM is derived. A simple P-type ILC law is proposed such that the ILC tracking error can be driven into a residual range, the bound of which is relevant to the bound parameters of boundary states. Specially, while the boundary states of 2-D (...) LDSFM satisfy iteration-invariant boundary states, accurate tracking on 2-D desired surface trajectory can be accomplished by using 2-D linear inequality theory. In addition, extension to 2-D LDSFM without direct transmission from inputs to outputs is presented. A numerical example is used to illustrate the effectiveness and feasibility of the designed ILC law. (shrink)

This paper considers the consensus control problem of nonlinear spatial-temporal hyperbolic partial difference multiagent systems and parabolic partial difference multiagent systems with time delay. Based on the system’s own fixed topology and the method of generating the desired trajectory by introducing virtual leader, using the consensus tracking error between the agent and the virtual leader agent and neighbor agents in the last iteration, an iterative learning algorithm is proposed. The sufficient condition for the system consensus error to converge along (...) the iterative axis is given. When the iterative learning number k approaches infinity, the consensus error in the sense of the L 2 norm between all agents in the system will converge to zero. Furthermore, simulation results illustrate the effectiveness of the algorithm. (shrink)

For linear discrete-time systems with randomly variable input trail length, a proportional- type iterative learning control law is proposed. To tackle the randomly variable input trail length, a modified control input at the desirable trail length is introduced in the proposed ILC law. Under the assumption that the initial state fluctuates around the desired initial state with zero mean, the designed ILC scheme can drive the ILC tracking errors to zero at the desirable trail length in expectation sense. The (...) designed ILC algorithm allows the trail length of control input which is different from system state and output at a specific iteration. In addition, the identical initial condition widely used in conventional ILC design is also mitigated. An example manifests the validity of the proposed ILC algorithm. (shrink)

Natural languages exhibit manysemantic universals, that is, properties of meaning shared across all languages. In this paper, we develop an explanation of one very prominent semantic universal, the monotonicity universal. While the existing work has shown that quantifiers satisfying the monotonicity universal are easier to learn, we provide a more complete explanation by considering the emergence of quantifiers from the perspective of cultural evolution. In particular, we show that quantifiers satisfy the monotonicity universal evolve reliably in an iterated (...) class='Hi'>learning paradigm with neural networks as agents. (shrink)

Cognitive Science, Volume 46, Issue 4, April 2022.

An iterative learning control algorithm is presented for the force control circuit of a hydraulic cushion. A control scheme consisting of a PI controller, feed-forward and feedback-linearization is first derived. The uncertainties and nonlinearities of the proportional valve, the main system actuator, prevent the accurate tracking of the pressure reference signal. Therefore, an extra ILC FF signal is added to counteract the valve model uncertainties. The unknown valve dynamics are attenuated by adding a fourth-order low-pass filter to the iterative (...) learning control design, which is split up into two second-order low-pass filters to carry out forward and backward filtering and obtain zero-phase filtering. The addition of the ILC signal presents significant improvements in terms of settling time and overshoot of the pressure signal in the cylinder. (shrink)

In this paper we argue that ecological evolutionary developmental biology accounts of cognitive modernity are compatible with cultural evolution theories of language built upon iterated learning models. Cultural evolution models show that the emergence of near universal properties of language do not require the preexistence of strong specific constraints. Instead, the development of general abilities, unrelated to informational specificity, like the copying of complex signals and sharing of communicative intentions is required for cultural evolution to yield specific properties, (...) such as language structure. We argue that eco-evo-devo provides the appropriate conceptual background to ground an account for the many interconnected genetic, environmental and developmental factors that facilitated the emergence of an organic system able to develop language through the iterated transmission of information. We use the concept of niche construction to connect evolutionary developmental accounts for sensory guided motor capacities and cultural evolution guided by iterated learning models. This integrated theoretical model aims to build bridges between biological and cultural approaches. (shrink)

An iterative learning robust fault-tolerant control algorithm is proposed for a class of uncertain discrete systems with repeated action with nonlinear and actuator faults. First, by defining an actuator fault coefficient matrix, we convert the iterative learning control system into an equivalent unknown nonlinear repetitive process model. Then, based on the mixed Lyapunov function approach, we describe the stability of the nonlinear repetitive mechanism on time and trial indices and have appropriate conditions for the repeated control system’s stability (...) in terms of linear matrix inequality theory. Through LMI techniques, we have obtained satisfactory results and controller stability, and robustness against fault tolerance is also discussed in detail. Finally, the simulation results of the output tracking control of the two exemplary models verify the effectiveness of the proposed algorithm. (shrink)

Determining the knowledge that guides human judgments is fundamental to understanding how people reason, make decisions, and form predictions. We use an experimental procedure called ‘‘iterated learning,’’ in which the responses that people give on one trial are used to generate the data they see on the next, to pinpoint the knowledge that informs people's predictions about everyday events (e.g., predicting the total box office gross of a movie from its current take). In particular, we use this method (...) to discriminate between two models of human judgments: a simple Bayesian model (Griffiths & Tenenbaum, 2006) and a recently proposed alternative model that assumes people store only a few instances of each type of event in memory (MinK; Mozer, Pashler, & Homaei, 2008). Although testing these models using standard experimental procedures is difficult due to differences in the number of free parameters and the need to make assumptions about the knowledge of individual learners, we show that the two models make very different predictions about the outcome of iterated learning. The results of an experiment using this methodology provide a rich picture of how much people know about the distributions of everyday quantities, and they are inconsistent with the predictions of the MinK model. The results suggest that accurate predictions about everyday events reflect relatively sophisticated knowledge on the part of individuals. (shrink)

Many of the problems studied in cognitive science are inductive problems, requiring people to evaluate hypotheses in the light of data. The key to solving these problems successfully is having the right inductive biases—assumptions about the world that make it possible to choose between hypotheses that are equally consistent with the observed data. This article explores a novel experimental method for identifying the biases that guide human inductive inferences. The idea behind this method is simple: This article uses the responses (...) produced by a participant on one trial to generate the stimuli that either they or another participant will see on the next. A formal analysis of this “iterated learning” procedure, based on the assumption that the learners are Bayesian agents, predicts that it should reveal the inductive biases of these learners, as expressed in a prior probability distribution over hypotheses. This article presents a series of experiments using stimuli based on a well-studied set of category structures, demonstrating that iterated learning can be used to reveal the inductive biases of human learners. (shrink)

Studies of language evolution in the lab have used the iterated learning paradigm to show how linguistic structure emerges through cultural transmission—repeated cycles of learning and use across generations of speakers. However, agent-based simulations suggest that prior biases crucially impact the outcome of cultural transmission. Here, we explored this notion through an iterated learning study of English-French bilingual adults (mostly sequential bilinguals dominant in English). Each participant learned two unstructured artificial languages in a counterbalanced fashion, (...) one resembling English, another resembling French at the phono-orthographic level. The output of each participant was passed down to the next participant, forming diffusion chains of 10 generations per language. We hypothesized that artificial languages would become easier to learn and exhibit greater structure when they were aligned with participants’ bilingual experience (i.e., English languages being easier to learn overall), or as a function of practice (i.e., languages learned second being easier to learn overall). Instead, we found that English-like languages became more structured over generations, but only when they were learned first. In contrast, French-like languages became more structured regardless of the order of learning, suggesting the presence of an asymmetric switch cost during artificial language learning. Moreover, individual differences in language usage modulated the amount of structure produced by the participants. Overall, these data suggest that bilingual experience impacts how novel languages are learned at an individual level, which can then scale up to cultural transmission of novel language at a group level. (shrink)

This paper investigates a distributed coordination control for a class of high-order uncertain multiagent systems. Under the framework of iterative learning control, a novel fully distributed learning protocol is devised for the coordination problem of MASs including time-varying parameter uncertainties as well as actuator saturations. Meanwhile, the learning updating laws of various parameters are proposed. Utilizing Lyapunov theory and combining with Graph theory, the proposed algorithm can make each follower track a leader completely over a limited time (...) interval even though each follower is without knowing the dynamics of the leader. Moreover, the extension to formation control is made. The validity and feasibility of the algorithm are verified conclusively by two examples. (shrink)

At present, the motion control algorithms of lower limb exoskeleton robots have errors in tracking the desired trajectory of human hip and knee joints, which leads to poor follow-up performance of the human-machine system. Therefore, an iterative learning control algorithm is proposed to track the desired trajectory of human hip and knee joints. In this paper, the experimental platform of lower limb exoskeleton rehabilitation robot is built, and the control system software and hardware design and robot prototype function test (...) are carried out. On this basis, a series of experiments are carried out to verify the rationality of the robot structure and the feasibility of the control method. Firstly, the dynamic model of the lower limb exoskeleton robot is established based on the structure analysis of the human lower limb; secondly, the servo control model of the lower limb exoskeleton robot is established based on the iterative learning control algorithm; finally, the exponential gain closed-loop system is designed by using MATLAB software. The relationship between convergence speed and spectral radius is analyzed, and the expected trajectory of hip joint and knee joint is obtained. The simulation results show that the algorithm can effectively improve the gait tracking accuracy of the lower limb exoskeleton robot and improve the follow-up performance of the human-machine system. (shrink)

In this contribution we analyze how the actions of rational agents feed back on their beliefs. We present two agent-based computer simulations studying complex social interactions in which agents that follow utility maximizing strategies thereby deteriorate their own long-term quality of beliefs. We take these results as a starting point to discuss the complex relationship between rational action couched in terms of maximizing utility and the emergence of informational inequalities.

I discuss the ubiquity of power law distributions in language organisation (and elsewhere), and argue against Miller’s (The mating mind: How sexual choice shaped the evolution of human nature, William Heinemann, London, 2000) argument that large vocabulary size is a consequence of sexual selection. Instead I argue that power law distributions are evidence that languages are best modelled as dynamical systems but raise some issues for models of iterated language learning.

Belief revision theory concerns methods for reformulating an agent's epistemic state when the agent's beliefs are refuted by new information. The usual guiding principle in the design of such methods is to preserve as much of the agent's epistemic state as possible when the state is revised. Learning theoretic research focuses, instead, on a learning method's reliability or ability to converge to true, informative beliefs over a wide range of possible environments. This paper bridges the two perspectives by (...) assessing the reliability of several proposed belief revision operators. Stringent conceptions of minimal change are shown to occasion a limitation called inductive amnesia: they can predict the future only if they cannot remember the past. Avoidance of inductive amnesia can therefore function as a plausible and hitherto unrecognized constraint on the design of belief revision operators. (shrink)

Over the last decade, iterated learning studies have provided compelling evidence for the claim that linguistic structure can emerge from non‐structured input, through the process of transmission. However, it is unclear whether individuals differ in their tendency to add structure, an issue with implications for understanding who are the agents of change. Here, we identify and test two contrasting predictions: The first sees learning as a pre‐requisite for structure addition, and predicts a positive correlation between learning (...) accuracy and structure addition, whereas the second maintains that it is those learners who struggle with learning and reproducing their input who add structure to it. This prediction is hard to test in standard iterated learning paradigms since each learner is exposed to a different input, and since structure and accuracy are computed using the same test items. Here, we test these contrasting predictions in two experiments using a one‐generation artificial language learning paradigm designed to provide independent measures of learning accuracy and structure addition. Adults (N = 48 in each study) were exposed to a semi‐regular language (with probabilistic structure) and had to learn it: Learning was assessed using seen items, whereas structure addition was calculated over unseen items. In both studies, we find a strong positive correlation between individuals' ability to learn the language and their tendency to add structure to it: Better learners also produced more structured languages. These findings suggest a strong link between learning and generalization. We discuss the implications of these findings for iterated language models and theories of language change more generally. (shrink)

We explore multi-scale relations in artificial intelligence (AI) use in order to identify difficulties with coordinating relations between users, machine learning (ML) processes, and “sociobuilt contexts”—specifically in terms of their applications to medical technologies and decisions. We begin by analyzing a recent COVID-19 machine learning case study in order to present the difficulty of traversing the detailed causal topography of “sociobuilt contexts.” We propose that the adequate representation of the interactions between social and built processes that occur on (...) many scales ought to drive interdisciplinary approaches for ML modification. Next, we describe ML algorithm development as a process that is partly dependent on methodological stabilization for reliability and partly on coordinating relations. In the coordination between user, ML process, and sociobuilt contexts, we propose that new methods can be explored that promote the inclusion of patients and communities for the purpose of cross-checking portions of the ML process. Finally, we suggest that the advantages of responsible innovation emerge through the iterative entanglement of ethical, methodological, and ontological considerations within the broader conceptual infrastructure of epistemic responsibility. (shrink)

In referendum elections, voters are often required to register simultaneous votes on multiple proposals. The separability problem occurs when a voter’s preferred outcome on one proposal depends on the outcomes of other proposals. This type of interdependence can lead to unsatisfactory or even paradoxical election outcomes, such as a winning outcome that is the last choice of every voter. Here we propose an iterative voting scheme that allows voters to revise their voting strategies based on the outcomes of previous iterations. (...) Using a robust computer simulation, we investigate the potential of this approach to solve the separability problem. (shrink)

This paper discusses the possibility of modelling inductive inference (Gold 1967) in dynamic epistemic logic (see e.g. van Ditmarsch et al. 2007). The general purpose is to propose a semantic basis for designing a modal logic for learning in the limit. First, we analyze a variety of epistemological notions involved in identification in the limit and match it with traditional epistemic and doxastic logic approaches. Then, we provide a comparison of learning by erasing (Lange et al. 1996) and (...) iterated epistemic update (Baltag and Moss 2004) as analyzed in dynamic epistemic logic. We show that finite identification can be modelled in dynamic epistemic logic, and that the elimination process of learning by erasing can be seen as iterated belief-revision modelled in dynamic doxastic logic. Finally, we propose viewing hypothesis spaces as temporal frames and discuss possible advantages of that perspective. (shrink)

Maximizers in isolated Prisoner's Dilemmas are doomed to frustration. But in Braybrooke's view maximizers might do better in a series, securing Pareto-optimal arrangements if not from the very beginning, at least eventually. Given certain favourable special conditions, it can be shown according to Braybrooke and shown even without question-begging motivational or value assumptions, that in a series of Dilemmas maximizers could manage to communicate a readiness to reciprocate, generate thereby expectations of reciprocation, and so give rise to optimizing reciprocations which, (...) in due course, would reinforce expectations, the net result of all this being an increasingly stable practice to mutual benefit. In this way neighbours could learn to be good neighbours: they could learn to respect each other's property, to engage in reciprocal assistance, and perhaps even to make and keep promises covering a range of activities. So maximizers are, Braybrooke holds, capable of a society of sorts. Out of the ashes they might build something. But even under favourable conditions they could not build much and most conditions would defeat them almost entirely, for many-person Dilemmas whether isolated or in series would be quite beyond them, question-begging motivational assumptions aside. In settings at all like those in which we live, utility maximization assuming only ‘traditional’ motivation is self-defeating without remedy, and most certainly not an adequate basis for social life.' The probable inference from as a conception of individual rationality: for surely truly rational agents would be if not perfectly well designed at least better designed for communal living than utility maximizers would be assuming only ‘traditional’, non-question-begging values or motivation. Or, if this inference is denied and maximization is not challenged as a conception of rationality, we can conclude instead, Braybrooke has more recently suggested, that the advice, ‘Be rational,’ is incomplete, and when addressed to groups should be supplemented by, ‘And be somewhat ‘untraditional’ in your values; respect at least some rules that would make coordination and cooperation possible even in many-person interaction problems.’. (shrink)

Using a simple learning agent, we show that learning self-control in the primrose path experiment does parallel learning cooperation in the prisoner's dilemma. But Rachlin's claim that “there is no essential difference between self-control and altruism” is too strong. Only iterated prisoner's dilemmas played against reciprocators are reduced to self-control problems. There is more to cooperation than self-control and even altruism in a strong sense.

Kline argues for an expanded taxonomy of teaching focusing on the adaptive behaviors needed to solve learning problems. Absent from her analysis is an explicit definition of learning, or a discussion of the iterative nature of the relationship between teaching and learning. Including the learner in the discussion may help to distinguish among the adaptive values of different teaching behaviors.

Today, more people want to know how to make a meaningful difference to what they care about. But for that, traditional approaches to learning often fall short. In this book, we offer a theoretical and practical way forward. We introduce the concept of social learning spaces for developing both new capabilities and a sense of agency. We provide a rich framework for focusing on the value of social learning spaces: how to generate this value, monitor it, and (...) learn iteratively through the process. The book is a useful extension and refinement of 'communities of practice' for those familiar with the theory. For those who are not, the chapters will lay out a new way to approach learning. This volume is written to serve the needs of readers across fields, including researchers, educators, and leaders in business, government, healthcare, and international development. (shrink)

A symposium was held at the Centre for Research in the Arts, Social Sciences and Humanities at the University of Cambridge on June 12th 2019, ‘Rethinking Repetition in a Digital Age’, at which Geoff Stead, a leading mobile tech designer, was a keynote speaker. The focus of the Cambridge UK event was on how the potentials of digital technologies—whose harms have received widespread attention—could be redirected for the social good. For Stead, this is precisely what Babbel are doing in their (...) approach to commercial digital language learning. Stead spoke to the idea of reversing our personal relationships to mechanical affordances, and finding empowerment in understanding their designed logics. The transcript of the interview below, made in October 2021, revisits some of the main points he raised at that event. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore (...) overlapping causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

We apply an online optimization process based on machine learning to the production of Bose-Einstein condensates. BEC is typically created with an exponential evaporation ramp that is optimal for ergodic dynamics with two-body s-wave interactions and no other loss rates, but likely sub-optimal for real experiments. Through repeated machine-controlled scientific experimentation and observations our ’learner’ discovers an optimal evaporation ramp for BEC production. In contrast to previous work, our learner uses a Gaussian process to develop a statistical model of (...) the relationship between the parameters it controls and the quality of the BEC produced. We demonstrate that the Gaussian process machine learner is able to discover a ramp that produces high quality BECs in 10 times fewer iterations than a previously used online optimization technique. Furthermore, we show the internal model developed can be used to determine which parameters are essential in BEC creation and which are unimportant, providing insight into the optimization process of the system. (shrink)