Computational challenges in explaining communication: How deep the rabbit hole goes

Braak, Laura D. van de; Dingemanse, Mark; Toni, Ivan; Rooij, Iris van; Blokpoel, Mark

doi:10.31234/osf.io/3wh5g

Cited by 5 publications

(4 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In principle, we expect that D k should reflect all relevant sources of information that may expose an agent’s state of understanding or misunderstanding. Not just ostensive signals like pointing (van de Braak et al, 2021), but verbal and nonverbal backchannels (e.g., mmhmm , nods or quizzical looks), forms of self-initiated and other-initiated repair (Arkel et al, 2020; Dingemanse et al, 2015; Schegloff et al, 1977), and downstream actions taken in the world (e.g., attempts to follow instructions).…”

Section: Convention Formation As Hierarchical Bayesian Inferencementioning

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

Section: Convention Formation As Hierarchical Bayesian Inferencementioning

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

show abstract

“…While this can sometimes be done analytically, more often computer simulations prove useful for these types of (complex and dynamic) models. Using computer simulations, for example, one can assess claims about possible functioning under network damage (Guest, Caso, & Cooper, 2020), claims of explanatory scope and adequacy (Adolfi, Bowers, & Poeppel, 2023;van de Braak, Dingemanse, Toni, van Rooij, & Blokpoel, 2021), claims of approximation (Blokpoel & van Rooij, 2021, Chapter 8), claims of ruling out possible socalled neural codes (Guest & Love, 2017), and claims of mechanistic possibilities (Bartlett et al, 2023;ten Oever & Martin, 2021).…”

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

“…This is because our cognitive explanations of how humans do such things as resolve misunderstandings contain large gaps (we go into case studies in sections §6.3 and §7.1). And indeed researchers invest time in locating and delimiting the source of these theoretical problems (e.g., van de Braak et al, 2021;Woensdregt et al, 2021). It is through these interlocking cycles of observation and theoretical activity that the capacity for communication is carved out as a scientific problem for cognitive science at all (see van Rooij & Baggio, 2021).…”

Section: The Provenance Of Cognitive-scientific Problemsmentioning

confidence: 99%

“…This theoretical problem-finding work led the authors to the following conclusions: "This shows that state-of-the-art computational explanations have difficulty explaining how people solve the puzzle of underdetermination, and that doing so will require a fundamental leap forward." (van de Braak et al, 2021). This change to the theory must be fundamental if we are to reach a plausible explanation.…”

Section: Theory Revision Driven By Cognitive Scope Violationsmentioning

confidence: 99%

From empirical problem-solving to theoretical problem-finding perspectives on the cognitive sciences

Adolfi,

van de Braak,

Woensdregt

2023

Preprint

View full text Add to dashboard Cite

We attempt to restore an integrative perspective on what the scientific problems of the cognitive sciences are and how they arise. As a point of departure, we draw attention to an increasingly impoverished view of science as empirical problem-solving. This frame, while occasionally useful, is typically fraught with misconceptions. We explicate some of its undesirable consequences, such as the neglect of plausibility constraints on cognitive explanations. Briefly, we challenge the implied autonomy of so-called empirical problems, the overemphasis on solving them versus finding them, and an overly narrow construal of what it means to solve them. To lift this epistemic fog, we bring theoretical problem-finding out of obscurity as a core activity in the cognitive scientist's toolbox. We illustrate various issues through hypothetical and factual case studies; among them, the inadequacies of portraying empirical problems as separate from theoretical problems; how phenomena in need of explanation depend critically on theoretical problems and plausibility constraints; and how theoretical problem-finding can be productive even when no immediate solutions are possible. Along the way, we give concrete examples of what can go wrong if we neglect the role of theoretical problems in shaping scientific questions. We arrive at an integrative account of how phenomena of interest, and cognitive-scientific problems more broadly, are carved out through theoretical problem-finding.

show abstract

Computational challenges in explaining communication: How deep the rabbit hole goes

Cited by 5 publications

References 18 publications

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

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

Reclaiming AI as a theoretical tool for cognitive science

From empirical problem-solving to theoretical problem-finding perspectives on the cognitive sciences

Contact Info

Product

Resources

About