Mental models, computational explanation and Bayesian cognitive science: Commentary on Knauff and Gazzo Castañeda (2023)

Abstract

Knauff and Gazzo Castañeda (2022) object to using the term “new paradigm” to describe recent developments in the psychology of reasoning. This paper concedes that the Kuhnian term “paradigm” may be queried. What cannot is that the work subsumed under this heading is part of a new, progressive movement that spans the brain and cognitive sciences: Bayesian cognitive science. Sampling algorithms and Bayes nets used to explain biases in JDM can implement the Bayesian new paradigm approach belying any advantages of mental models theory (MMT) at the algorithmic level. Moreover, this paper argues that new versions of MMT lack a computational level theory and questions the grounds for MMTs much-vaunted generality. The paper then examines common ground on the importance of small-scale models/simulations of the world and the importance of argumentation in the social domain rather than individual reasoning. Finally, the paper concludes that although there may be prospects for moving reasoning research forward in a more collective, collaborative manner, many disagreements remain to be resolved.

Keywords:

• Mental models
• Bayesian cognitive science
• sampling algorithms

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Nick Chater and I were recently invited to contribute an Annual Review of Psychology article where the editors explicitly gave us the title, “New paradigms in the psychology of reasoning”. Consequently, this nomenclature would now appear to be well-embedded in contemporary psychological culture, and there is probably little this exchange of views will do about that.

2 Mental probability logic hypothesizes an algorithmic implementation given by the relevant proof theory for which soundness and completeness results have been established. How does this indicate a failure to apply Bayesian principles as a descriptive theory? To my knowledge there is no experimental evidence that discriminates Bayesian probability theory from Ranking theory as approaches to the dynamics of belief change. One paper by Neils Skovgaard-Olsen does not make Ranking theory a part of the “Bayesian new paradigm”. However, Ranking theory seems like a very promising approach to uncertainty, more promising than the non-monotonic logics critiqued by Oaksford and Chater (Oaksford & Chater 1991, 2014, 2016). Moreover, Spohn’s Baconian probability theory may prove to be a more empirically adequate account, but the jury is well and truly out on this question.

3 In a previous brief career in the Royal Navy, at my one and only Trafalgar night dinner at Dartmouth as a Midshipman in 1974, I fell foul of the rule of etiquette that, if, as the decanters are passed from right to left, someone passes the madeira before the port, then they must buy everyone present in the wardroom a drink. The Captain let me off because it was my first time (and my last, as the RN and I parted company a few months later), but I have since wondered how MMT would represent this conditional?

4 In a Bayes net, certain configurations of evidence (instantiations of its variables to their states) will be (epistemically) impossible in the model (inconsistent evidence). Configurations can also conflict to a lesser or greater degree with the model (Flesch et al. 2007).

5 That is, deduction and any machinery required for it, is unlikely to provide the foundational scaffolding of human reasoning abilities. Our ability to model the casual structure of the world is a more likely candidate to play this role, a view borne out by the inability to reduce causation to logical (Stalnaker, 1984) or probabilistic (Cartwright, 1983) concepts.