1st post on bsky - about bsky! I was fascinated with academic starter packs and made an interactive network to see academic communities and how they connect - a map of knowledge! link to an interactive & searchable network: ketikagarg.github.io/blueSkyAcade...
New paper out today! When baby zebra finches are tutored with manipulated songs with shuffled syllables, they transform them to be consistent with linguistic laws 🐥 Reanalysis of data from James and Sakata (2017) link.springer.com/article/10.1...
Come work with us! And get in touch with any questions you might have about the position, our labs or living/working in Germany #PostdocWanted
Great thread by @apvelilla.bsky.social about our new paper: "A Demographic Theory of Similarity-Biased Social Learning."
We use mathematical modeling to explore the functional role played by identity markers in how we learn from others, and in the evolution of social learning more generally.
This work is part of the first half of my dissertation, which deals with the theme of social learning under uncertainty and risk.
Only two more projects to go! Feedback is very welcome!
In other words, density-dependent costs lead the population to evolve past the point of maximum efficiency. While this leads to the evolution of mixed social learning, it ends up granting the same fitness benefits that pure individual learners obtain with no social learning at all.
This rescue can be so effective as to lead social learning to fixation. However, if we assume a density-dependent cost to social learning, social learning can stabilize at an intermediate frequency, recovering "Rogers' paradox".
This "rescue" depends on how informative tags are: when the correlation between tags and behavioral correctness is strong, optimal bias becomes useful enough that, at high enough frequencies of social learning, its benefits offset the costs of social learning with respect to individual learning.
Our setup also shows that bias can actually "rescue" social learning, even when unbiased social learning is costlier than individual learning. However, this can only happen when social learning is common enough to build a cultural pool rich enough to bootstrap the benefits of social learning.
We extend this idea to the possibility of anti-bias, where agents preferentially learn from targets whose tags are associated with out-group membership. Doing so allows us to get a richer picture of how demographic mixing and ecological structure lead to different strength and direction of bias.
In our previous paper we found that a similarity bias---which exploits group identity information contained in visible tags---is favored when identity tags are correlated with behavioral correctness. In diverse pops with group-structured ecological niches, parochialism evolves to avoid mismatch.
New preprint with @psmaldino.bsky.social on the evolution of similarity-biased learning: osf.io/preprints/so...
This mathematical treatment both simplifies and extends our previous simulation treatment of the subject: www.cambridge.org/core/journal...
A highlight of our approach is that, by parameterizing the possible underlying communication networks and their role in how agents decide to share, we can potentially use observed network data to infer properties of the unobserved communication networks driving them. Comments welcome!
This tells us the mechanism is likely a strong contributor to observable sharing network structure, but since real networks are the product of many different mechanisms (such as kin-based sharing and risk pooling) it should not be leveraged in isolation.
We then fit the simulation model to data from a community in the Canadian Arctic, showing that our "network signaling" mechanism can explain several aspects of the empirical sharing network structure (while missing the mark on others, such as reciprocity).
Our model reveals a neat insight: agents will find it advantageous to share widely when the underlying communication networks through which reputation travels are connected but sparse. As comm. networks become denser, agents find that few sharing partners are needed to spread signals of generosity.
Here we explore how signaling on a (unobserved) communication network can lead to the (observed) sharing networks in forager communities. We formalize the mechanism through a game-theoretic model which we then turn into a strategic network formation model that simulates network data.
New preprint out with @elspethready.bsky.social : "The emergence of sharing networks through indirect signaling".
osf.io/preprints/so...
Super proud of this paper with @apvelilla.bsky.social and @babeheim.bsky.social, now out in Psych Review.
Non-paywalled version (preprint) here: osf.io/preprints/so...
Full comic here: www.smbc-comics.com/comic/capita...
#smbc
This week we talk to @babeheim.bsky.social about culture, change, modeling, running red lights, and the game of go.
youtu.be/nqTkSK-qtJM
www.podbean.com/eas/pb-hrn9v...
Happy to see this work published in Psych Review. It's an impressive and important bit of theory/modeling about how we learn about decision-making under risk. Here's a slide with the super-coarse-grained summary of our results. Read the paper for (much) more. osf.io/preprints/so...
We believe these ideas can help explain behavioral variation in many domains by specifying how risk landscapes shape beliefs and behavior through learning strategies. See our (in press) commentary to the (great) target article by @sheinalew.bsky.social & @dorsaamir.bsky.social
osf.io/preprints/so...
@psmaldino.bsky.social and I also made a contribution to the Oxford Handbook of Evolutionary Psychology on how an integrative view spanning evolutionary, developmental and cultural influences on risk behavior can help us understand decisions under risk—a good companion piece. osf.io/preprints/so...
I am happy to announce that our project on risk and social learning is now in press at Psychological Review. Several new additions and revisions thanks to detailed feedback from colleagues and anonymous reviewers. osf.io/preprints/so...
@psmaldino.bsky.social @babeheim.bsky.social
@psmaldino.bsky.social and I also made a contribution to the Oxford Handbook of Evolutionary Psychology on how an integrative view spanning evolutionary, developmental and cultural influences on risk behavior can help us understand decisions under risk—a good companion piece. osf.io/preprints/so...
Miyagawa Shuntei's 1898 painting, "Playing Go (Japanese Chess)"
How to quantify the impact of AI on long-run cultural evolution? Published today, I give it a go!
400+ years of strategic dynamics in the game of Go (Baduk/Weiqi), from feudalism to AlphaGo!
📣 Job alert! *Assistant Prof in Computational Social Science*. We're a friendly department, with sharp students, at a great institution, in a lovely city. We have real strengths in computational social science & are looking for a colleague to build this further. Share and reach out with quesions!
📢 Apply to our (2-year) research fellowships at @iast.fr
Join a multidisciplinary, interdisciplinary, transdisciplinary, and indisciplinary group of scholars in Toulouse, walkable/cyclable pink city of chocolatines in the South of France.
Deadline: November 15, 2025.
www.iast.fr/research-fel...