Why is scaling up models of language evolution hard?

Woensdregt, Marieke; Spike, Matthew; Haan, Ronald de; Wareham, Todd; Rooij, Iris van; Blokpoel, Mark

doi:10.31234/osf.io/d2h5c

Cited by 6 publications

(9 citation statements)

References 28 publications

(49 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Together, the results proven here caution against intuitive notions about the complexity properties of computational problems driving empirical programs, and demonstrate the need and benefits of critically assessing their soundness (something that is too rarely done explicitly; but see, e.g., van de Braak, de Haan, van Rooij, and Blokpoel, 2022;Woensdregt et al, 2021;van de Pol, van Rooij, and Szymanik, 2018;Rich, Blokpoel, de Haan, and van Rooij, 2020;Zeppi and Blokpoel, 2017 for notable exceptions). Whenever intuitions are challenged, this enables researchers to reevaluate the current meta-theoretical calculus (cf.…”

Section: Discussionmentioning

confidence: 74%

A Computational Complexity Perspective on Segmentation as a Cognitive Subcomputation

Adolfi

Wareham²,

Rooij

2022

Topics in Cognitive Science

Self Cite

View full text Add to dashboard Cite

Computational feasibility is a widespread concern that guides the framing and modeling of natural and artificial intelligence. The specification of cognitive system capacities is often shaped by unexamined intuitive assumptions about the search space and complexity of a subcomputation. However, a mistaken intuition might make such initial conceptualizations misleading for what empirical questions appear relevant later on. We undertake here computational‐level modeling and complexity analyses of segmentation — a widely hypothesized subcomputation that plays a requisite role in explanations of capacities across domains, such as speech recognition, music cognition, active sensing, event memory, action parsing, and statistical learning — as a case study to show how crucial it is to formally assess these assumptions. We mathematically prove two sets of results regarding computational hardness and search space size that may run counter to intuition, and position their implications with respect to existing views on the subcapacity.

show abstract

Section: Discussionmentioning

confidence: 74%

A Computational Complexity Perspective on Segmentation as a Cognitive Subcomputation

Adolfi

Wareham²,

Rooij

2022

Topics in Cognitive Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…While a fully Bayesian formulation elegantly formalizes the computational-level inference problem at the core of the CHAI account, this formulation faces a number of limitations. For one, it is clearly intractable (Van Rooij, 2008; Van Rooij et al, 2019): The posterior update step in Equation 5 grows increasingly intensive as the space of possible utterances and meanings grows (Woensdregt et al, 2021). The intractability problem also raises a scalability problem: Does CHAI provide any guidance toward building artificial agents that are actually able to adapt to human partners as humans do with one another?…”

Section: Discussionmentioning

confidence: 99%

From partners to populations: A hierarchical Bayesian account of coordination and convention.

Hawkins¹,

Franke²,

Frank³

et al. 2023

Psychological Review

View full text Add to dashboard Cite

Languages are powerful solutions to coordination problems: They provide stable, shared expectations about how the words we say correspond to the beliefs and intentions in our heads. Yet, language use in a variable and nonstationary social environment requires linguistic representations to be flexible: Old words acquire new ad hoc or partner-specific meanings on the fly. In this article, we introduce continual hierarchical adaptation through inference (CHAI), a hierarchical Bayesian theory of coordination and convention formation that aims to reconcile the long-standing tension between these two basic observations. We argue that the central computational problem of communication is not simply transmission, as in classical formulations, but continual learning and adaptation over multiple timescales. Partner-specific common ground quickly emerges from social inferences within dyadic interactions, while community-wide social conventions are stable priors that have been abstracted away from interactions with multiple partners. We present new empirical data alongside simulations showing how our model provides a computational foundation for several phenomena that have posed a challenge for previous accounts: (a) the convergence to more efficient referring expressions across repeated interaction with the same partner, (b) the gradual transfer of partner-specific common ground to strangers, and (c) the influence of communicative context on which conventions eventually form.

show abstract

“…Computational modelling of capacities can help us to make our assumptions precise and explicit, and to draw out their consequences, without the need to simulate the postulated computations (though simulations have their uses; more on that next). For instance, with formal computationallevel models and mathematical proof techniques at hand, one can critically assess claims of explanatory adequacy Egan, 2017;van Rooij & Baggio, 2021), claims of intractability (Adolfi, Wareham, & van Rooij, 2023), claims of tractability van Rooij, Evans, Muller, Gedge, & Wareham, 2008), claims of competing theories , claims of evolvability (Rich, Blokpoel, de Haan, & van Rooij, 2020;Woensdregt et al, 2021), and claims of approximability (Kwisthout & Van Rooij, 2013;Kwisthout, Wareham, & Van Rooij, 2011). 16 We acknowledge that computational modelling can also contribute to productive theory development without committing to computationalism (Guest & Martin, 2021;Morgan & Morrison, 1999).…”

Section: Theory Without Makeingmentioning

confidence: 99%

Reclaiming AI as a theoretical tool for cognitive science

Rooij¹,

Guest²,

Adolfi³

et al. 2023

Preprint

View full text Add to dashboard Cite

The idea that human cognition is, or can be understood as, a form of computation is a useful conceptual tool for cognitive science. It was a foundational assumption during the birth of cognitive science as a multidisciplinary field, with Artificial Intelligence (AI) as one of its contributing fields. One conception of AI in this context is as a provider of computational tools (frameworks, concepts, formalisms, models, proofs, simulations, etc.) that support theory building in cognitive science. The contemporary field of AI, however, has taken the theoretical possibility of explaining human cognition as a form of computation to imply the practical feasibility of realising human(-like or -level) cognition in factual computational systems; and, the field frames this realisation as a short-term inevitability. Yet, as we formally prove herein, creating systems with human(-like or -level) cognition is intrinsically computationally intractable. This means that any factual AI systems created in the short-run are at best decoys. When we think these systems capture something deep about ourselves and our thinking, we induce distorted and impoverished images of ourselves and our cognition. In other words, AI in current practice is deteriorating our theoretical understanding of cognition rather than advancing and enhancing it. The situation could be remediated by releasing the grip of the currently dominant view on AI and by returning to the idea of AI as a theoretical tool for cognitive science. In reclaiming this older idea of AI, however, it is important not to repeat conceptual mistakes of the past (and present) that brought us to where we are today.

show abstract

Why is scaling up models of language evolution hard?

Cited by 6 publications

References 28 publications

A Computational Complexity Perspective on Segmentation as a Cognitive Subcomputation

A Computational Complexity Perspective on Segmentation as a Cognitive Subcomputation

From partners to populations: A hierarchical Bayesian account of coordination and convention.

Reclaiming AI as a theoretical tool for cognitive science

Contact Info

Product

Resources

About