Minimal Requirements for the Emergence of Learned Signaling

Abstract

The emergence of signaling systems has been observed in numerous experimental and real‐world contexts, but there is no consensus on which (if any) shared mechanisms underlie such phenomena. A number of explanatory mechanisms have been proposed within several disciplines, all of which have been instantiated as credible working models. However, they are usually framed as being mutually incompatible. Using an exemplar‐based framework, we replicate these models in a minimal configuration which allows us to directly compare them. This reveals that the development of optimal signaling is driven by similar mechanisms in each model, which leads us to propose three requirements for the emergence of conventional signaling. These are the creation and transmission of referential information, a systemic bias against ambiguity, and finally some form of information loss. Considering this, we then discuss some implications for theoretical and experimental approaches to the emergence of learned communication.

Languages adapt to their contextual niche

“…(Millikan, 1998: 175) Aligning on shared communicative conventions draws upon two general resources. The first of these is feedback (De Ruiter et al, 2010; for overview, see: Spike et al, 2016): interlocutors are able to provide information to one another as to whether or not their interpretation is correct, allowing one, or both, of the parties to modify their behaviour in future interactions 6 . With feedback, speakers can modify their behaviour through trying to clarify the intended meaning, and use utterances which pick out relatively smaller sections of the context (Frank & Goodman, 2014: 85).…”

A Social Approach to Rule Dynamics Using an Agent‐Based Model

“…Colaiori et al (2015;see also, Cuskley et al, 2017) used a model of three-state dynamics wherein agents have either a regular, irregular, or mixed (optionally regular or irregular) rule for a given lemma. This approach examined both gradual population turnover with biased child learners and memory limitations (also known as, information loss; Spike, Stadler, Kirby, & Smith, 2016). These factors, combined with a skewed frequency distribution of lemmas, recovered patterns in regularity reminiscent of natural language data (Carrol, Svare, & Salmons, 2012;Cuskley et al, 2014;Lieberman, Michel, Jackson, Tang, & Nowak, 2007), wherein highly frequent forms are more likely to be irregular given certain starting conditions.…”

“…For each lemma-inflection pairing, the agent stores: 1 f i , the total number of interactions with that inflection and lemma; 2 w, a weight for the lemma-inflection pair defined as the number of successful interactions with the lemma-inflection pair divided by the total number of interactions with the lemma-inflection pair, f i ; 3 t last , the time step of the last interaction with that lemma-inflection combination Agents have temporally constrained memory for each lemma-inflection combination. Memory limitations are crucial to recovering realistic frequency-dependent rule dynamics found in natural language (e.g., in corpora; Cuskley et al, 2017) and have been identified as an essential factor in the emergence of learned signaling in agent-based models more generally (Spike et al, 2016). This memory constraint is implemented as a deterministic loss of a lemma-inflection pairing after a set period of time: If a pairing has not been encountered within a specific time window, d = 100 (regardless of the pairing's weight, w), the pairing will be forgotten.…”

“…2B), indicating that despite new learners constantly replacing more proficient speakers in the population, there is low interspeaker variation. This may be because turnover combines with memory constraints to create greater overall information loss, which can bolster the emergence of shared conventions (Spike et al, 2016).…”

Coordination of Multiple Autonomous Agents Using Naturally Generated Languages in Task Planning

“…We set out to answer the question by considering a particular situation in which communication is only available in task planning phase. In this case, the Our language generation problem has a connection with the work on the origin and evolution of natural languages, which has been studied earlier in evolutionary and computational linguistics [29,30]. Most of the prior works studied language evolution in the context of evolutionary games [31], for example, the talking heads experiment [32].…”