This theory of Social Tinkering can be thought of as a generalization of our idea of language-as-charades (detailed in our book, The Language Game) generalized to cultural evolution. If this piques your interest, keep an eye out for the call for commentaries

8/8
www.cambridge.org/core/journal...

We suggest that our framework has profound implications for theories of individual cognition; and diverges from existing traditions in cultural evolution (including those in evolutionary game theory), focusing on random variation and selection over individual, not joint social behavior.
7/8

We propose that adaptation, entrenchment, recombination, and interaction of these rules accumulate into complex cultural systems through processes of un-designed spontaneous order (from ethical and legal principles, to languages, to legal and political systems).
6/8

Second, we stress the critical importance of social tinkering: that is, the joint, and myopically purposive, adjustment of the rich and flexible norms and conventions that govern human social interactions.
5/8

First, our theory emphasizes that while cultural change is local, innovations are actively directed—shaped by the full power of human social intelligence to solve immediate problems: we draw on the metaphor of tinkering, from evolutionary biologist François Jacob.
4/8

We propose a theory of Social Tinkering: how active coordination in social interactions provides the driving force behind cultural evolution. Our account departs from much current cultural evolutionary thinking, which we believe leans too heavily on parallels with Darwinian natural selection.
3/8

Inspired by the Scottish Enlightenment philosopher Adam Ferguson, we explore how cultural and social organization arises from human action, but not human design: from myopic, moment-by-moment modifications to the rules by which we live—modifications intelligently chosen to solve current problems
2/8

This overview dovetails very nicely with the complementary, more historical and theoretical (albeit short) review of the study of language evolution in this entry in the Open Encyclopedia of Cognitive Science (open access for all): oecs.mit.edu/pub/18miikqb/

Or indeed my book, The Language Game, if they’re interested in the evolution, acquisition and processing of language 😉

We have a paper in preparation with more experiments, of which we discuss part of one of them in the Linguistics Vanguard paper

Indeed, though our experiments indicate that 'vanilla' versions without RLHF (and even 'instruct' ones) are not fully able to use this information to generate human-like language. Human feedback may be needed to "unlock" it -- at least when it comes to producing human-level grammatical language?

Ross D. Kristensen-McLachlan, Pablo Contreras Kallens and I suggest that the introduction of Reinforcement Learning with Human Feedback provides much need socially informed feedback and that this makes their linguistic behavior more human-like (including human errors) 2/2

Curious? The article is OA

Papers (cont):
Brown & Walasek
Trujillo, Zhang, Zhi-Xuan, Tenenbaum & Levine
Contreras Kallens & Christiansen
Chater

3/3

Learn more about Nick's groundbreaking work within cognitive science—including on simplicity, reasoning, similarity. decision-making, virtual bargaining, and language—along with personal musings by Mike and me.

onlinelibrary.wiley.com/doi/10.1111/...

Papers by
Oaksford
Hodgetts & Hahn

2/3