Social norms are often characterized as a system of rules that guide behavior. However, social norms also allow for flexibility; not entirely restricting individuals to one possible behavior. Here, we put forward an agent-based cognitive model that captures social norms as processes that are socially constructed through interactions between individuals. In this modelling work, we focus on the role of norm acquisition and conformity bias in both action production and inference-making. This computational cognitive model allows us to think about social norms along three dimensions: individual vs. collective, behavior vs. belief, and subjective vs. objective. Our simulation results show that increased conformity bias can induce misjudgments about the true desires of others and misalignment between different agents' perceptions of the social norm. However, if agents do not assume that others also conform in their behavior, this increased conformity bias does not necessarily lead to excessive misperceptions of the social norm.

Demonstratives---simple referential devices like this and that---are linguistic universals, but their meaning varies cross-linguistically. In languages like English and Italian, demonstratives are thought to encode the referent's distance from the producer (e.g., that one means ``the one far away from me"), while in others, like Portuguese and Spanish, they encode relative distance from both producer and receiver (e.g., aquel means ``the one far away from both of us"). Here we propose that demonstratives are also sensitive to the receiver's focus of attention, hence requiring a deeper form of social cognition than previously thought. We provide initial empirical and computational evidence for this idea, suggesting that producers use demonstratives to redirect the receiver's attention towards the intended referent, rather than only to indicate its physical distance.

We identify theoretical challenges for developing a computational explanation of flexible linguistic inference. Specifically, the human ability to interpret a novel linguistic expression (like “mask-shaming”), where inferring plausible meanings requires integrating relevant background knowledge (e.g., COVID-19 pandemic). We lay out (i) the core properties of the phenomenon that together make up our construal of the explanandum, (ii) explanatory desiderata to help make sure a theory explains the explanandum, and (iii) cognitive constraints to ensure a theory can be plausibly realised by human cognition and the brain. By doing so, we lay bare the ‘force field' that theories of this explanandum have to navigate, and we give examples of tensions that arise between different components of this force field. This is an important step in theory-development because it allows researchers who aim to solve one part of the puzzle of flexible linguistic inference to keep in clear view the other parts.

Language requires mindreading for entertaining communicative intentions, and mindreading in turn profits fromlanguage as a means for sharing mental states. Hence it has been hypothesised that the two skills have co-evolved.We present a Bayesian agent-based model to formalise this hypothesis. This model combines referential signalling withmental states, such that a speaker’s topic choice is probabilistically dependent on their perspective on the world. In order tolearn the language, a learner has to simultaneously infer the speaker’s lexicon and perspective. Learners can solve this task bybootstrapping one with the other, but only if the speaker uses an informative language.We will present results of an iterated learning version of this model, showing that selection on communication results inthe emergence of a fully informative lexicon from scratch. However, selection on perspective-taking alone also results in theemergence of partially-informative lexicons, which is sufficient for inferring others’ perspectives.

Word learning involves mapping observable words to unob-servable speaker intentions. The ability to infer referential in-tentions in turn has been shown to depend in part on accessto language. Thus, word learning and intention-reading co-develop. To explore this interaction, we present an agent-basedmodel in which an individual simultaneously learns a lexiconand learns about the speaker’s perspective, given a shared con-text and the speaker’s utterances, by performing Bayesian in-ference. Simulations with this model show that (i) lexicon-learning and perspective-learning are strongly interdependent:learning one is impossible without some knowledge of theother, (ii) lexicon- and perspective-learning can bootstrap eachother, resulting in successful inference of both even when thelearner starts with no knowledge of the lexicon and unhelpfulassumptions about the minds of others, and (iii) receiving ini-tial input from a ‘helpful’ speaker (who adopts the learner’sperspective on the world) paves the way for later learning fromspeakers with perspectives which diverge from the learner’s.This approach represents a first attempt to model the hypoth-esis that language and mindreading co-develop, and a first ex-ploration of the implications for theories of word learning andmindreading development.

Computational model simulations have been very fruitful for gaining insight into how the systematic structure we observe in the world's natural languages could have emerged through cultural evolution. However, these model simulations operate on a toy scale compared to the size of actual human vocabularies, due to the prohibitive computational resource demands that simulations with larger lexicons would pose. Using computational complexity analysis, we show that this is not an implementational artifact, but instead it reflects a deeper theoretical issue: these models are (in their current formulation) computationally intractable. This has important theoretical implications, because it means that there is no way of knowing whether or not the properties and regularities observed for the toy models would scale up. All is not lost however, because awareness of intractability allows us to face the issue of scaling head-on, and can guide the development of our theories.