Emergent communication models support interaction-based language learning, benefiting both Natural Language Processing (NLP) applications and simulations of language evolution, but they are prone to destabilizing language drift.Inspired by developmental trajectories in human language acquisition, this paper investigates whether age-based plasticity, where younger agents learn quickly and older agents maintain stable representations, can reduce language drift.In our set-up, static populations first reliably develop shared languages, followed by a phase in which population turnover gradually replaces older agents with new learners.Age-based plasticity significantly reduces drift in this setting, maintaining high accuracy and language similarity.In contrast, in populations with uniformly low plasticity agents cannot adapt quickly enough to integrate newcomers and in those with uniformly high plasticity the language changes faster than stable conventions can form.These findings demonstrate that developmental trajectories in individual learners substantially reduce overall language drift in dynamic populations.

Human languages have evolved to be structured through repeated language learning and use. These processes introduce biases that operate during language acquisition and shape linguistic systems toward communicative efficiency. In this paper, we investigate whether the same happens if artificial languages are optimised for implicit biases of Large Language Models (LLMs). To this end, we simulate a classical referential game in which LLMs learn and use artificial languages. Our results show that initially unstructured holistic languages are indeed shaped to have some structural properties that allow two LLM agents to communicate successfully. Similar to observations in human experiments, generational transmission increases the learnability of languages, but can at the same time result in non-humanlike degenerate vocabularies. Taken together, this work extends experimental findings, shows that LLMs can be used as tools in simulations of language evolution, and opens possibilities for future human-machine experiments in this field.

Languages are shaped by the inductive biases of their users. Using a classical referential game, we investigate how artificial languages evolve when optimised for inductive biases in humans and large language models (LLMs) via Human-Human, LLM-LLM and Human-LLM experiments. We show that referentially grounded vocabularies emerge that enable reliable communication in all conditions, even when humans and LLMs collaborate. Comparisons between conditions reveal that languages optimised for LLMs subtly differ from those optimised for humans. Interestingly, interactions between humans and LLMs alleviate these differences and result in vocabularies more human-like than LLM-like. These findings advance our understanding of the role inductive biases in LLMs play in the dynamic nature of human language and contribute to maintaining alignment in human and machine communication. In particular, our work underscores the need to think of new LLM training methods that include human interaction and shows that using communicative success as a reward signal can be a fruitful, novel direction.

Recent advances in multimodal models have raised questions about whether vision-and-language models (VLMs) integrate cross-modal information in ways that reflect human cognition. One well-studied test case in this domain is the bouba-kiki effect, where humans reliably associate pseudowords like ‘bouba’ with round shapes and ‘kiki’ with jagged ones. Given the mixed evidence found in prior studies for this effect in VLMs, we present a comprehensive re-evaluation focused on two variants of CLIP, ResNet and Vision Transformer (ViT), given their centrality in many state-of-the-art VLMs. We apply two complementary methods closely modelled after human experiments: a prompt-based evaluation that uses probabilities as a measure of model preference, and we use Grad-CAM as a novel approach to interpret visual attention in shape-word matching tasks. Our findings show that these model variants do not consistently exhibit the bouba-kiki effect. While ResNet shows a preference for round shapes, overall performance across both model variants lacks the expected associations. Moreover, direct comparison with prior human data on the same task shows that the models’ responses fall markedly short of the robust, modality-integrated behaviour characteristic of human cognition. These results contribute to the ongoing debate about the extent to which VLMs truly understand cross-modal concepts, highlighting limitations in their internal representations and alignment with human intuitions.

Natural languages show a tendency to minimize the linear distance between heads and their dependents in a sentence, known as dependency length minimization (DLM). Such a preference, however, has not been consistently replicated with neural agent simulations. Comparing the behavior of models with that of human learners can reveal which aspects affect the emergence of this phenomenon. In this work, we investigate the minimal conditions that may lead neural learners to develop a DLM preference. We add three factors to the standard neural-agent language learning and communication framework to make the simulation more realistic, namely: (i) the presence of noise during listening, (ii) context-sensitivity of word use through non-uniform conditional word distributions, and (iii) incremental sentence processing, or the extent to which an utterance’s meaning can be guessed before hearing it entirely. While no preference appears in production, we show that the proposed factors can contribute to a small but significant learning advantage of DLM for listeners of verb-initial languages.

Missing diversity, equity, and inclusion elements in affective computing datasets directly affect the accuracy and fairness of emotion recognition algorithms across different groups. A literature review reveals how affective computing systems may work differently for different groups due to, for instance, mental health conditions impacting facial expressions and speech or age-related changes in facial appearance and health. Our work analyzes existing affective computing datasets and highlights a disconcerting lack of diversity in current affective computing datasets regarding race, sex/gender, age, and (mental) health representation. By emphasizing the need for more inclusive sampling strategies and standardized documentation of demographic factors in datasets, this paper provides recommendations and calls for greater attention to inclusivity and consideration of societal consequences in affective computing research to promote ethical and accurate outcomes in this emerging field.

Abstract Artificial learners often behave differently from human learners in the context of neural agent-based simulations of language emergence and change. A common explanation is the lack of appropriate cognitive biases in these learners. However, it has also been proposed that more naturalistic settings of language learning and use could lead to more human-like results. We investigate this latter account, focusing on the word-order/case-marking trade-off, a widely attested language universal that has proven particularly hard to simulate. We propose a new Neural-agent Language Learning and Communication framework (NeLLCom) where pairs of speaking and listening agents first learn a miniature language via supervised learning, and then optimize it for communication via reinforcement learning. Following closely the setup of earlier human experiments, we succeed in replicating the trade-off with the new framework without hard-coding specific biases in the agents. We see this as an essential step towards the investigation of language universals with neural learners.

In healthcare, gender and sex considerations are crucial because they affect individuals' health and disease differences. Yet, most algorithms deployed in the healthcare context do not consider these aspects and do not account for bias detection. Missing these dimensions in algorithms used in medicine is a huge point of concern, as neglecting these aspects will inevitably produce far from optimal results and generate errors that may lead to misdiagnosis and potential discrimination. This paper explores how current algorithmic-based systems may reinforce gender biases and affect marginalized communities in healthcare-related applications. To do so, we bring together notions and reflections from computer science, queer media studies, and legal insights to better understand the magnitude of failing to consider gender and sex difference in the use of algorithms for medical purposes. Our goal is to illustrate the potential impact that algorithmic bias may have on inadvertent discriminatory, safety, and privacy-related concerns for patients in increasingly automated medicine. This is necessary because by rushing the deployment of AI technologies that do not account for diversity, we risk having an even more unsafe and inadequate healthcare delivery. By promoting the account for privacy, safety, diversity, and inclusion in algorithmic developments with health-related outcomes, we ultimately aim to inform the Artificial Intelligence (AI) global governance landscape and practice on the importance of integrating gender and sex considerations in the development of algorithms to avoid exacerbating existing or new prejudices.

Natural languages display a trade-off among different strategies to convey syntactic structure, such as word order or inflection.This trade-off, however, has not appeared in recent simulations of iterated language learning with neural network agents (Chaabouni et al., 2019b).We re-evaluate this result in light of three factors that play an important role in comparable experiments from the Language Evolution field: (i) speaker bias towards efficient messaging, (ii) non systematic input languages, and (iii) learning bottleneck.Our simulations show that neural agents mainly strive to maintain the utterance type distribution observed during learning, instead of developing a more efficient or systematic language.

Developing successful sign language recognition, generation, and translation systems requires expertise in a wide range of fields, including computer vision, computer graphics, natural language processing, human-computer interaction, linguistics, and Deaf culture. Despite the need for deep interdisciplinary knowledge, existing research occurs in separate disciplinary silos, and tackles separate portions of the sign language processing pipeline. This leads to three key questions: 1) What does an interdisciplinary view of the current landscape reveal? 2) What are the biggest challenges facing the field? and 3) What are the calls to action for people working in the field? To help answer these questions, we brought together a diverse group of experts for a two-day workshop. This paper presents the results of that interdisciplinary workshop, providing key background that is often overlooked by computer scientists, a review of the state-of-the-art, a set of pressing challenges, and a call to action for the research community.

Music is a peculiar human behavior, yet we still know little as to why and how music emerged. For centuries, the study of music has been the sole prerogative of the humanities. Lately, however, music is being increasingly investigated by psychologists, neuroscientists, biologists, and computer scientists. One approach to studying the origins of music is to empirically test hypotheses about the mechanisms behind this structured behavior. Recent lab experiments show how musical rhythm and melody can emerge via the process of cultural transmission. In particular, Lumaca and Baggio (2017) tested the emergence of a sound system at the boundary between music and language. In this study, participants were given random pairs of signal-meanings; when participants negotiated their meaning and played a "game of telephone" with them, these pairs became more structured and systematic. Over time, the small biases introduced in each artificial transmission step accumulated, displaying quantitative trends, including the emergence, over the course of artificial human generations, of features resembling properties of language and music. In this Note, we highlight the importance of Lumaca and Baggio's experiment, place it in the broader literature on the evolution of language and music, and suggest refinements for future experiments. We conclude that, while psychological evidence for the emergence of proto-musical features is accumulating, complementary work is needed: Mathematical modeling and computer simulations should be used to test the internal consistency of experimentally generated hypotheses and to make new predictions.