Fig 4 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Numerical World Color Survey with human and neutral JNDs: structure of categories. The two solid curves represent the histograms of position of category-centroids obtained in simulations with the human (black curve) and neutral JND (blue curve). As a reference, the inverse of the human JND, rescaled by a constant factor, is displayed (red dashed curve). In the central region strong correlation is seen between the centroid distribution from “human-like” simulations and the inverse of the human JND, while the outcome of unbiased simulations is flat in the same region. Strong oscillations near the two extrema (0 and 1) are appreciated in both models, typical of “hard boundaries”. The solid circles displayed at the bottom of the graph represent the “average pattern” of the ~ 150 populations which display 14 categories at the end of “human-like” simulations.
Source publication
Language universals have long been attributed to an innate Universal Grammar. An alternative explanation states that linguistic universals emerged independently in every language in response to shared cognitive or perceptual biases. A computational model has recently shown how this could be the case, focusing on the paradigmatic example of the univ...
Context in source publication
Context 1
... computed the average, or "typi- cal", pattern of categories for this sub-group, s ¼ f x 1 ; x 2 ; :::; x 14 g. The 14 average positions of typical centroids are displayed at the bottom of Fig 4. Having established the typical pattern, we went further, asking how far the pattern s j of the j-th population is from the typical one s. The simplest tool to this end is the Euclidean We stress that a power-law tail is a typical signal of fluctuations larger than normal. ...
Citations
... Indeed, there has been accumulating evidence that in some languages the basic nomenclature can exceed the ceiling of 11, whereby an incipient BCC may develop as a cross-language universal, or emerge as a "culturally basic" language-specific, as argued by Paramei (2005). This empirical observation is buttressed by a computational model of Baronchelli et al. (2015), who demonstrated that the universal properties of colour naming patterns emerge independently in every language in response to universal communicative, cognitive or perceptual biases. Along with these, at play is the cultural transmission bias: the specific cultural history of a language is effective in shaping non-universal properties of colour naming due to the need to communicate the prominent colour shades of speakers' natural and artefactual environment, as well as cross-language contacts. ...
We investigated the Tuscan Italian colour inventory, with the aim of establishing the cognitive salience of the basic colour terms (BCTs) and most frequent non-BCTs. Native speakers from Tuscany ( N = 89) completed a colour-term elicitation task lasting for 5 min. In total, 337 unique terms were elicited, with an average list length of 30.06. The frequency of each term, its mean list position and cognitive salience index ( S ) were calculated. The CTs with the highest S (ranked 1–13) included 10 counterparts of the Berlin and Kay BCTs listed in their 1969 seminal work and three basic ‘blue’ terms, blu, azzurro, celeste , estimated for Tuscan respondents by Del Viva et al. in 2022. S -index and Zipf-function (the terms’ “popularity”) indicated that fucsia (rank 14) is conceivably an emerging BCT in (Tuscan) Italian. Other cognitively salient non-BCTs are lilla , magenta , ocra and beige . The terms’ 3D semantic map (conceptual closeness), assessed using multidimensional scaling and cluster analysis, revealed that in the lists, closely associated CTs were arranged along three competing criteria: the term’s salience gradient; word length; and clustering of fully chromatic concepts with those defined primarily by lightness or desaturation. We also consider salient Italian non-BCTs as indicators of the ongoing process of lexical refinement in certain areas of the colour space. In conclusion, measures of elicitation productivity, as well as the augmented BCT inventory, including the Tuscan ‘triple blues’, and abundant hyponyms and derived forms all indicate (Tuscan) Italian speakers’ “cultural competence” in the colour domain and the need to communicate nuanced information about colour shades.
... In the field of cognitive psychology, the emphasis is on the connection between communication and language and other cognitive processes, mainly internal, such as memory, categorisation, and reasoning. For example, emergent language, stabilised in language games, has been shown to be a factor determining the shared structure of individual categories within populations (Baronchelli, Loreto, & Puglisi, 2015;Steels & Belpaeme, 2005). Some models in this field do concern the emergence of language on the developmental timescale, although, curiously and symptomatically, these rarely take the form of computational simulation. ...
Computational simulations are a popular method for testing hypotheses about the emergence of communication. This kind of research is performed in a variety of traditions including language evolution, developmental psychology, cognitive science, machine learning, robotics, etc. The motivations for the models are different, but the operationalizations and methods used are often similar. We identify the assumptions and explanatory targets of several most representative models and summarise the known results. We claim that some of the assumptions -- such as portraying meaning in terms of mapping, focusing on the descriptive function of communication, modelling signals with amodal tokens -- may hinder the success of modelling. Relaxing these assumptions and foregrounding the interactions of embodied and situated agents allows one to systematise the multiplicity of pressures under which symbolic systems evolve. In line with this perspective, we sketch the road towards modelling the emergence of meaningful symbolic communication, where symbols are simultaneously grounded in action and perception and form an abstract system.
... A well-researched classic case is the perception of color: faced with a continuum of wavelengths in the visible, we categorize and name colors according to our language and cultural biases (Baronchelli et al., 2015;Hardin et al., 1997). In the color labels, as in the phonetic inventory or in the use of words, our brain uses mechanisms of categorical perception (Goldstone & Hendrickson, 2010;Harnad, 1987). ...
The dichotomy between the qualitative and the quantitative has been a classic throughout the history of science. As will be seen, this dichotomy permeates all ontological levels of reality. In this work, phenomenological examples potentially related to semiosis are presented at the different levels established by Mario Bunge and Josep Ferrater Mora, contrasting the qualitative categorizations with the quantifiable physical reality. Likewise, the need to continue in the quantification of the biosemiotic and linguistic studies will be presented, while, in contrast, the need to establish a qualitative framework in the little-addressed study of technosemiotics will be raised, of potential interest given the notable advances that are expected in communication systems for inert artifacts in the next years. In short, in the thesis defended here the qualitative precedes the quantitative in the defining path of science.
... Computational simulation 18,19 has recently joined the endeavour to tackle these limitations, by tracing the emergence and spread of linguistic categories among individuals during socio-cultural transmissions [20][21][22][23][24] . Some models used predefined focal colours 22,23 or pseudo stimuli from the hue space 12,[24][25][26] . For example, the category game model 24 incorporated a nonlinear curve of Just Noticeable Difference (JND, the lowest perceptual distance between colour stimuli that normal human eyes could discriminate) in the hue space. ...
... The comparison of the cross-language dispersions between the categories emergent under the human JND and a uniform JND resembled the one between the real and randomised WCS data 12 . Follow-up studies using the same model attempted to ascribe the emerging sequence of focal colours to the non-uniformity of the human JND 25 , and the cross-language categorical variations to the cultural histories of communities 26 . ...
The universal linguistic colour categorisation pattern as evident in the World Colour Survey (WCS) has been a principal focus of investigations on the relationship between language and cognition, yet most existing studies have failed to clarify whether this universality resulted primarily from individual perceptual constraints and/or socio-cultural transmissions. This paper designed an agent-based, unsupervised learning model to address the relative importance of these two aspects to linguistic colour categorisation. By directly comparing with the empirical data in the WCS, our study demonstrated that: the physical colour stimuli that reflect human perceptual constraints on colours trigger a categorisation pattern quantitatively resembling the WCS data, the randomised stimuli that distort such constraints lead to distinct categorisation patterns, and the processes of linguistic categorisation in both cases follow similar dynamics. These results reveal how perceptual and socio-cultural factors interact with each other to trigger linguistic universality, and serve as decisive evidence that human perceptual constraints induce the universality in linguistic categorisation, yet socio-cultural transmissions, though imperative, play an auxiliary role of transcribing perceptual constraints into common linguistic categories with slight variations.
... For example, if the first-person singular pronoun (e.g., "I" in English) is included in a sentence, the speaker is the focus of attention (Kashima and Kashima 2003). By contrast, dropping a 2 See, e.g., Christiansen and Kirby (2003), Levinson (2003), Johansson (2005), Kirby et al. (2007), Evans and Levinson (2009), Baronchelli et al. (2015), Thompson et al. (2016). 3 Evans (2013) labels the impact of culture on language "Vico-Herder effects", after two early proponents of the view that languages express unique aspects of the history and worldview of their cultures (or peoples or nations). ...
This paper empirically studies the human capital effects of grammatical rules that permit speakers to drop a personal pronoun when used as a subject of a sentence. By de‐emphasizing the significance of the individual, such languages may perpetuate ancient values and norms that give primacy to the collective, inducing governments and families to invest relatively little in education because education usually increases the individual's independence from both the state and the family and may thus reduce the individual's commitment to these institutions. Carrying out both an individual‐level and a country‐level analysis, the paper indeed finds negative effects of pronoun‐drop languages. The individual‐level analysis uses data on 114,894 individuals from 75 countries over 1999‐2014. It establishes that speakers of such languages have a lower probability of having completed secondary or tertiary education, compared with speakers of languages that do not allow pronoun drop. The country‐level analysis uses data from 101 countries over 1972‐2012. Consistent with the individual‐level analysis, it finds that countries where the dominant languages permit pronoun drop have lower secondary school enrollment rates. In both cases, the magnitude of the effect is substantial, particularly among females.
... For NG to better describe consensus dynamics in the real world, it is necessary to incorporate catagorization through learning and knowledge growth into the model. While there were previous works on categorization games that deal with feature recognition and analysis [19][20][21][22][23] , the process of categorization itself remains to be an outstanding topic of research, due to the complexity of the underlying process. It is thus a challenging task to incorporate perception and categorization into NGs. ...
To reach consensus among interacting agents is a problem of interest for social, economical, and political systems. A computational and mathematical framework to investigate consensus dynamics on complex networks is naming games. In general, naming is not an independent process but relies on perception and categorization. Existing works focus on consensus process of vocabulary evolution in a population of agents. However, in order to name an object, agents must first be able to distinguish objects according to their features. We articulate a likelihood category game model (LCGM) to integrate feature learning and the naming process. In the LCGM, self-organized agents can define category based on acquired knowledge through learning and use likelihood estimation to distinguish objects. The information communicated among the agents is no longer simply in some form of absolute answer, but involves one's perception. Extensive simulations with LCGM reveal that a more complex knowledge makes it harder to reach consensus. We also find that agents with larger degree contribute more to the knowledge formation and are more likely to be intelligent. The proposed LCGM and the findings provide new insights into the emergence and evolution of consensus in complex systems in general.
... Many approaches have been adopted to address this issue, such as psychological experimentation of color acquisition [208][209][210][211][212] and perception [213][214][215][216], neuroimaging studies of categorical perception of color stimuli [217][218][219][220][221], and agent-based simulations of color category development and spread [222][223][224][225][226][227]. These studies report contrasting results in support of either of the two hypotheses. ...
... interest in the domain of colour. Most of the models proposed have focused on the emergence of single colour terms [140,15,11,12], although there have been some attempts to model more complex colour descriptions. ...
... 12: Resulting dynamics of the experiment for a population of 10 agents averaged over 10 runs of 100000 interactions (Figure 6.12a). By the end of the simulation all agents in the population reach a steady communicative success value above 85% and high confidence values for all the communicative challenges. ...
This thesis studies the role of intrinsic motivation in the emergence and development of communicative systems in populations of artificial agents. To be more specific, our goal is to explore how populations of agents can use a particular motivation system called autotelic principle to regulate their language development and the resulting dynamics at the population level.To achieve this, we first propose a concrete implementation of the autotelic principle. The core of this system is based on the balance between challenges, tasks to be done to achieve a goal, and skills, the abilities the system can employ to accomplish the different tasks. The relation between the two elements is not steady but regularly becomes destabilised when new skills are learned, which allows the system to attempt challenges of increasing complexity. Then, we test the usefulness of the autotelic principle in a series of language evolution experiments. In the first set of experiments, a population of artificial agents should develop a language to refer to objects with discrete values. These experiments focus on how unambiguous communicative systems can emerge when the autotelic principle is employed to scaffold language development into stages of increasing difficulty. In the second set of experiments, agents should agree on a language to communicate with about colour samples. In this part, we explore how the motivation system can regulate the linguistic complexity of interactions for a continuous domain and examine the value of the autotelic principle as a mechanism to control several language strategies simultaneously.To summarise, we have shown through our work that the autotelic principle can be used as a general mechanism to regulate complexity in language emergence in an autonomous way for discrete and continuous domains.
... Then, the speaker assigns one of these objects to teach the hearer, which is identical to guessing game in [25]. The CGM has been further studied in [27,28]. However, a "God" is always required to provide two different objects to the agents. ...
Naming game simulates the evolution of vocabulary in a population of agents. Through pairwise interactions in the games, agents acquire a set of vocabulary in their memory for object naming. The existing model confines to a one-to-one mapping between a name and an object. Focus is usually put onto name consensus in the population rather than knowledge learning in agents, and hence simple learning model is usually adopted. However, the cognition system of human being is much more complex and knowledge is usually presented in a complicated form. Therefore, in this work, we extend the agent learning model and design a new game to incorporate domain learning, which is essential for more complicated form of knowledge. In particular, we demonstrate the evolution of color categorization and naming in a population of agents. We incorporate the human perceptive model into the agents and introduce two new concepts, namely subjective perception and subliminal stimulation, in domain learning. Simulation results show that, even without any supervision or pre-requisition, a consensus of a color naming system can be reached in a population solely via the interactions. Our work confirms the importance of society interactions in color categorization, which is a long debate topic in human cognition. Moreover, our work also demonstrates the possibility of cognitive system development in autonomous intelligent agents.
... This method has also been profitably applied to more complex issues, such as categorization. In this context, the Category Game model [44], which is literally built on top of the minimal NG, has proven to be able to reproduce experimental data concerning color naming systems [45,67,68]. Different research avenues remain open for the future, ranging from addressing more complex problems such as the emergence compositionality [69,70] to un-derstanding the nature of language change [71,72], and there is consensus among the researches in different disciplines on the substantial contribution that the complex systems approach will continue to provide [11]. ...
Social conventions govern countless behaviors all of us engage in every day, from how we greet each other to the languages we speak. But how can shared conventions emerge spontaneously in the absence of a central coordinating authority? The Naming Game model shows that networks of locally interacting individuals can spontaneously self-organize to produce global coordination. Here, we provide a gentle introduction to the main features of the model, from the dynamics observed in homogeneously mixing populations to the role played by more complex social networks, and to how slight modifications of the basic interaction rules give origin to a richer phenomenology in which more conventions can co-exist indefinitely.
