Ben Ambridge

This really winds me up - vaccines, masks, air purification? Nah don’t bother - just get covid first then we’ll give you counselling

tl;dr Collinearity is a form of lack of information that is appropriately reflected in the output of your statistical model. When collinearity is associated with interpretational difficulties, these difficulties aren’t caused by the collinearity itself. Rather, they reveal that the model was poorly specified (in that it answers a question different to the one of interest), that the analyst overly focuses on significance rather than estimates and the uncertainty about them or that the analyst took a mental shortcut in interpreting the model that could’ve also led them astray in the absence of collinearity. If you do decide to “deal with” collinearity, make sure you can still answer the question of interest.

Was asked about collinearity again, so here's Vahove's 2019 post on why it isn't a problem that needs a solution. Design the model(s) to answer a formal question and free your mind janhove.github.io/posts/2019-0...

Sigmoid function. Non-linearities in neural network allow it to behave in distributed and near-symbolic fashions.

New paper! 🚨 I argue that LLMs represent a synthesis between distributed and symbolic approaches to language, because, when exposed to language, they develop highly symbolic representations and processing mechanisms in addition to distributed ones.
arxiv.org/abs/2502.11856

My research group ❤️. Reposts welcome - we’re looking for little scientists to come and help us research what makes them tick

pluralistic.net/2025/09/27/e... When you think about it, there a quite a few technologies that are kinda useful but just can’t be made to work economically- makes sense that “AI” would be in that category

This post prompted me to finally try to understand the difference between an odds ratio and a risk ratio! Finding a good explainer was hard (especially now, with too much AI-generated slop) but I finally understood it after reading this! freerangestats.info/blog/2018/08...

email screenshot: Dear XX, thanks for thinking of me. Unfortunately I need to decline — I am one of those people who feels that Frontiers and big-money-OA is corrupting the scientific enterprise beyond repair, so I don't submit, review, or edit for their journals. I feel the same for journals like Scientific Reports, Communications Psychology, MDPI journals, and the like. In my opinion the only way to improve scientific publishing is to vote with our feet and our grant dollars. Sorry that I don't have a more positive response....! best from NZ Sam

I screen out Frontiers emails so hadn't seen a colleague's invite to their editorial board. so the colleague emailed me directly to ask again (we're friendly w/ each other)

I don't like saying 'no' to service for good colleagues, but one must act on principle against bad publishers. here's my reply

For me, what LLMs tell us is that any abstractions that are good enough to represent language will be too distributed, abstract and complex for satisfying/intuitive/humanly-understandable explanations. I.e that these questions are more or less unanswerable

Mike Tomasello would almost certainly agree with you! But for some of us UB acquisition people the part about building complex syntactic abstractions by compressing stored exemplars is at least as interesting. For all their many faults, LLMs do that bit well!

In the meantime, I'm no expert, but presumably it doesn't have to be this way. I can't see any technical reason why journals that use Editorial Manger HAVE to use a login page that starts with "editorialmanager.com" - @ariessystems.bsky.social please work with them to sort it out! 5/5

And this is ironic (if only in an Alanis Morissette sense) for us Open Source/Open Access fans because the need to keep paying for Editorial Manager is one of the most frequently cited objections to ditching subscription journals/paid Open Access and setting up our own totally free journals. 4/5

The irony is...journals that use free Open Source alternatives (like Janeway) don't have this problem - because the login page (sensibly!) starts with the name of the journal (e.g., ldr.lps.library.cmu.edu/login/) NOT with "editiorialmanager.com" 3/5

As a result, I can count on the fingers of one hand the number of times that I've successfully logged into to any Editorial-Manager-using journal without having to reset my password. For us Open Source/Open Access fans, there's also a huge irony here... 2/5

Can I please waste a few moments of your time to complain about @ariessystems.bsky.social Editorial Manager. Because of the way it interacts with journals' own pages (it's always editorialmanager.com/nameofjournal), password managers assume the same username+password FOR EVERY BLOODY JOURNAL 1/5

📣@futrell.bsky.social and I have a BBS target article with an optimistic take on LLMs + linguistics. Commentary proposals (just need a few hundred words) are OPEN until Oct 8. If we are too optimistic for you (or not optimistic enough!) or you have anything to say: www.cambridge.org/core/journal...

This is so infuriating- the Guardian article even says that the study can’t prove causation, but then just carries on in the rest of the article as if it has. And it’s not like this is a one off - this happens for just about every correlational health study!

Sweeteners can harm cognitive health equivalent to 1.6 years of ageing, study finds Researchers say low- and no-calorie sweeteners appear to affect thinking and memory in middle age

So, this study linking sweeteners to cognitive decline:

🚩 confuses correlation + causation
🚩 has no theory
🚩 shows evidence of p-hacking
🚩 reports inconsistent CIs / p-values
🚩 measures diet only once
🚩 has a big missing data problem
🚩 adjusts outcome opaquely

www.theguardian.com/food/2025/se...

We did include all the adult papers we found that had a forced choice behavioural measure, but we could well have missed some! Interestingly even adults are not quite at 100% on case marked sentences with noncanonical word order

In fairness, the rest of the summary is spot on, but that’s what makes it so dangerous - you never know which completely-plausible bit is wrong!

What a surprise! The AI overview completely misinterprets the findings of the study! Word order is LESS reliable than case-marking, not more. Read the correct version at psycnet.apa.org/fulltext/202...

A good argument for using Effect Size and Confidence Intervals over p values. In some sense we already know as a field this is the right thing to do as it’s what we already (mainly) do for meta analysis

www.facebook.com/share/p/19Sz...

Psycholinguistic Databases, Stimuli, Utilities — Concepts & Cognition Laboratory

Many updates to our psycholinguistic database page. You terrific people have been busy distributing open source data including lexical databases for BSL, French, German sign. LMK if I've missed anything www.reilly-coglab.com/data

Postdoctoral Research Scholar, The Cognition, Affect, and Temperament Lab, College of the Liberal Arts, Department of Psychology-2 APPLICATION INSTRUCTIONS: CURRENT PENN STATE EMPLOYEE (faculty, staff, technical service, or student), please login to Workday to complete the internal application process. Please do not apply here, a...

My lab has an open post-doc position!

We will begin reviewing applications as they come in.

The post-doc will work with a vibrant research team at Penn State in the Department of Psychology and the Child Study Center.

psu.wd1.myworkdayjobs.com/PSU_Academic...

1/3

The cover of the book « babel » by RF Kuang

The cover of Empire of AI by Karen Hao

Recently finished @rfkuang.bsky.social Babel, and Karen Hao’s @karenhao.bsky.social Empire of AI.

Incredibly interesting parallels on the colonization of language/text. The people who make these technologies possible are considered disposable and get considerably less benefit than the creators.

An abbreviation (ABB) in a journal article (JA) or Grant Application (GA) is rarely worth the words it saves. Every ABB requires cognitive resources (CR) and at my age by the time I'm halfway through a JA or GA I no longer have the CR to remember what your ABB stood for.

Screenshot of paper title

Our position paper is now out: "Learning Variability Network Exchange (LEVANTE): A Global Framework for Measuring Children's Learning Variability Through Collaborative Data Sharing".

srcd.onlinelibrary.wiley.com/doi/10.1111/...
(preprint: osf.io/preprints/ps...)

What is LEVANTE? 🧵

On top of everything else, it’s just so massively unnecessary- If sites want these things, actual experts on this stuff would be queuing up to write them for free!

What Is Evil? Explaining away even the most horrific acts of violence by saying some people are just wicked is understandable—but it won’t help us build a safer society.

Writing someone off as “evil” is an excuse to ignore the causes of human dysfunction, Amanda Knox writes. “That’s why I keep trying, even when I fail, to feel some degree of compassion for those labeled as ‘evil’”:

How Does Speaking A Free Word Order Language Influence Sentence Planning and Production? Evidence From Pitjantjatjara (Pama‐Nyungan, Australia) Sentence production is a stage-like process of mapping a conceptual representation to the linear speech signal via grammatical rules. While the typological diversity of languages is vast and thus mus...

How does speaking a free word order language influence sentence planning and production? Evidence from Pitjantjatjara (Pama‐Nyungan, Australia). New paper by Evan Kidd & al. with Gabriela Garrido Rodríguez
doi.org/10.1111/cogs.70087

it's wild that R, the ubiquitous statistical computing language, was co-created by a Māori prof (Ross Ihaka) — and yet the vast majority of scientists who use R don't know

this is like inventing the toaster. possibly the largest impact of a single member of an indigenous community on modern science