Mastodon @Mastodon@mastodon.social

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · 3d

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Yet another paper showing dual-minded #LLMs (intuitive + reflective) can improve accuracy-cost tradeoffs: https://doi.org/10.48550/arXiv.2504.12329

As I argue in #StrategicReflectivism, pragmatic switching between the two modes is key to intelligent systems: https://www.researchgate.net/publication/390166382

$Figure 4: Overview of speculative thinking. A small model generates most output but selectively delegates challenging segments—marked by structural cues such as paragraph breaks (“\n\n”) followed by reflective phrases like “wait,” “alternatively,” or “hold on”—to a stronger model. Small models often produce verbose or incoherent outputs at these points, while larger models handle them concisely. The proposed speculative thinking preserves efficiency while leveraging the large model’s strength when most needed.$

#AI #cogSci

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · 4d

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Do people diagnosed with #autism respond differently to moral dilemmas?

In MINORS, sacrificial harm waned with age, more slowly in the ASD group: https://doi.org/10.1007/s10803-022-05795-6

In ADULTS, decisions were similar: https://doi.org/10.1016/j.paid.2024.112889

Labusch, M., Perea, M., Sahuquillo-Leal, R., Bofill-Moscardó, I., Carrasco-Tornero, Á., Cañada-Pérez, A., & García-Blanco, A. (2022). Development of Moral Judgments in Impersonal and Personal Dilemmas in Autistic Spectrum Disorders from Childhood to Late Adolescence. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-022-05795-6

Mantchala, S., Gosling, C. J., Trémolière, B., & Moutier, S. (2025). Relationship between reasoning, autistic and alexithymic traits in moral judgments. Personality and Individual Differences, 233, 112889. https://doi.org/10.1016/j.paid.2024.112889

#philosophy #cogSci #psychiatry

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · 4d

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Does thinking aloud disrupt reasoning?

We didn't find effects on a verbal reflection test (https://pubmed.ncbi.nlm.nih.gov/37103261), but Shealy et al. found effects on word count, completion time, and DLPFC activity during a design task (N = 50).

https://doi.org/10.1017/pds.2023.87

Thinking aloud did not impact performance. Figure 1 affirms our pre-registered hypothesis and prior meta-analytic work (Fox et al. 2011): we did not detect an interference effect of thinking aloud on the number of lured or correct responses on the vCRT.

Figure 1. The effect of thinking aloud on (A) the number of lured responses and (B) the number of correct responses on the verbal cognitive reflection test (vCRT) in Study 1 (N = 99). Error bars represent a standard error.

The control group spent more time designing ...(t = 2.94, p = 0.005) compared to the think-aloud group. ...design sketches produced by the control group included significantly more words (statistic = 203, p = 0.05) than the think-aloud group. The total number of combined words verbalized and written to describe their design ideas for the think-aloud group was significantly more (statistic = 532.5, p < 0.001) compared to ...the control group. [...] Differences in the right dorsolateral prefrontal cortex (PFC) were observed in Channel 2 (statistic=303, p = 0.016) and Channel 3 (statistic=316, p = 0.006), and in the left dorsolateral PFC, in Channel 35 (statistic=300, p = 0.02).

#cogSci #neuroscience

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · 5d *

5d *

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Does asking people to justify their answer improve subsequent answers?

In two experiments (N > 300), students asked to justify their answers performed better on multiple choices immediately afterward (and maybe even 48 hours later).

https://doi.org/10.3390/bs15040477

#edu #cogSci #psychology

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · 5d

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Does awareness of intuition's fallibility help people avoid faulty intuitions?

#Teaching students #DualProcessTheory didn't help them avoid faulty intuitions about #physics problems.

https://doi.org/10.1103/PhysRevPhysEducRes.21.010135

#edu #cogSci #bias

**Anne Fausto Sterling** @faustosterling@mastodon.world · Apr 16

Apr 16

Anne Fausto Sterling @faustosterling@mastodon.world

why am I not surprised? #cogsci #devpsy #psychscisky #blackinstem
https://wordinblack.com/2025/04/psych-association-under-pressure-ends-diversity-requirements/?utm_source=ActiveCampaign&utm_medium=email&utm_content=Crisis%20to%20Care%3A%205%20Charts%20on%20Black%20Maternal%20Health%20Progress&utm_campaign=Health%20newsletter%204%2016
#AmericanPsychologicalAssociation #DiversityInRecruitment #DiversityEquityInclusion #PostgraduatePrograms #PoliticalPressure
#InclusionMatters

Word In Black · Apr 9Psychological Assn. Pauses Diversity in Recruitment RequirementsBy Jennifer Porter Gore

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Apr 16

Apr 16

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Wow! #QualiService could be a great resource!

It wasn't obvious to me how to find the transcripts for these doctor-patient interaction data from 4 countries, but if such transcripts are accessible, that's GREAT!

https://www.qualiservice.org/en/qsearch.html?q=diagnosis

A Qualiservice search for "diagnosis" returned a list of qualitative datasets from video recordings of doctor-patient interactions in four countries (China, Germany, Netherlands, and Turkey).

#medicine #openData #cogSci

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Apr 13

Apr 13

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Jon Baron shared Peter Wakker's annotated bibliography of #decisionTheory

> 9000 entries!

DocX http://personal.eur.nl/wakker/refs/webrfrncs.docx

PDF http://personal.eur.nl/wakker/refs/webrfrncs.pdf

BibTeX (no annotations, I merged redundancies) https://www.dropbox.com/scl/fi/vbkki82h62ydq0fol1g8d/Decision-Theory.bib?rlkey=84m8zx3tyaa4uy6p0zptkybnx&st=qx9myebb&dl=0

#cogSci #decisionScience #epistemology

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Apr 2 *

Apr 2 *

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

In Religious Studies: Analytic #Atheism & Analytic Apostasy...

Steve Stich, Justin Sytsma, and I studied >70k people across the globe.

Apostates were more reflective thinkers.

That explained links between reflection and #religion.

#Preprint + audio: https://byrdnick.com/archives/28471/upon-reflection-ep-14-analytic-atheism-analytic-apostasy-across-cultures

The final typeset version is also freely available (#openAccess) via #CambridgeUP; https://doi.org/10.1017/S0034412525000198

Pages from the accepted manuscript (Byrd, Stich, and Sytsma 2025).

Nick Byrd, Ph.D. · Apr 2Why do atheists tend to do better on critical thinking tests?Apostates seem to score better than others. And they tend to be atheists. So "analytic atheism" seems to be explained by analytic APOSTASY.

#psychology #psychometrics #cogSci

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Mar 31

Mar 31

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Can #AI improve #healthcare?

#ambientListening apps that take notes during appointments seemed to improve efficiency and make clinicians' jobs easier.

But it didn't seem to increase clinicians' confidence in the #LLM tool.

https://doi.org/10.1001/jamanetworkopen.2024.60637

"This study included 46 clinicians from 17 different medical specialties, with a mean (SD) of 11.1 (8.7) years in practice. From baseline to post intervention, use of the ambient scribing tool was associated with 20.4% less time in notes per appointment (from 10.3 to 8.2 minutes; P < .001), 9.3% greater same-day appointment closure (from 66.2% to 72.4%; P <.001), and 30.0% less after-hours work time per workday (from 50.6 to 35.4 minutes per workday; P = .02). Targeted perspective questions showed more favorable scores. Open-ended qualitative feedback from clinicians showed a range of positive, negative, and mixed feedback regarding their use of ambient scribing technology."

#medicine #bioethics #cogSci

**Gabriel Belmar | El área de Broca** @elareadebroca.bsky.social@bsky.brid.gy · Mar 24

Mar 24

Gabriel Belmar | El área de Broca @elareadebroca.bsky.social@bsky.brid.gy

Cálculos espaciotemporales en la brújula celeste de los insectos. Cómo los insectos se orientan utilizando "neuronas reloj" #Scicomm #CogSci

Spatiotemporal computations in...

Bluesky SocialBluesky

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Mar 18 *

Mar 18 *

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Some arguments and evidence suggest that intuitions about some philosophical thought experiments are remarkably stable.

Wu et al replicate this for one case (Exp 1), but find arguments changed people's conclusions about 10 cases (Exp 2).

https://doi.org/10.1080/09515089.2025.2456570

#xPhi #CogSci

**Gabriel Belmar | El área de Broca** @elareadebroca.bsky.social@bsky.brid.gy · Mar 15

Mar 15

Gabriel Belmar | El área de Broca @elareadebroca.bsky.social@bsky.brid.gy

Gracias a este efecto puedes asociar una palabra con una idea, una canción con un autor o un color con una emoción. En el hilo de esta semana les cuento sobre el efecto priming y su relación con las teorías de conspiración #SocialPsychology #Cogsci

RE: https://bsky.app/profile/did:plc:qegujouoycjebqli45kvsihh/post/3lkfisknakc2c

Bluesky SocialBluesky

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Mar 12

Mar 12

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

#socialMedia posts about #ADHD seemed to cause #availabilityBias about post quality and even people's self #diagnosis.

#Chatbot #symptom checkers (CSCs) that prompted #reflection on evidence or counterfactuals mitigated the #bias.

https://doi.org/10.48550/arXiv.2501.15028

Figure 4: Examples of symptom checking questions asked by the designed CSCs. The blue sections represent controlled explanations in responses. The red and yellow sections are uniquely crafted cognitive interventions, aimed at encouraging deeper reflection.

Figure 5: Overview of the experimental procedure.

Figure 6: Comparison of Social Media influence scores across different conditions, including CSC with Evidence Reflection (ER), CSC with Counterfactual Thinking (CT), and the control group. Statistical significance is denoted by p-value annotations, where *: p < 0.05, **: p < 0.01.

Figure 8: Mean evaluation scores on mental effort across the four conditions for study 2. Error bars show .95 confidence intervals. We report the results of the one-way ANOVA test and pairwise comparisons, where * : p < 0.05, ** : p < 0.01.

#AI #medicine #cogSci

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Mar 10

Mar 10

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Judges' asylum decisions vary wildly, violating a "prerequisite for ...legal coherence—to 'treat like cases alike."

This paper shows how #AI and #algorithms can "instantiate a reasonable approximation of a coherence theory of #truth":

https://osf.io/4tjz5_v1

Title, authors, abstract, and teal-world variation in asylum decisions (pages 1 and 2).

How "LLMs can accurately determine local consistency of propositions" (pages 5 and 6).

Upgrading from "Thagard’s classical model of coherence" to "sheaf cohomology" (Pages 7 and 8).

How propositional logic can computational instantiate sheaf cohomology, providing the reflective (system 2) against which intuitive (System 1) output from language models can be revised and the two major challenges in implementing this system (pages 9 and 10).

#law #xJur #CogSci

**Nick Byrd, Ph.D.** @ByrdNick@nerdculture.de · Mar 7

Mar 7

Nick Byrd, Ph.D. @ByrdNick@nerdculture.de

Recognizing the fallibilities of #ChatGPT was most reliably predicted by

(a) better performance on a reflection test

(b) more self-reported prioritization of arguments and evidence (over myside political #bias)

in a study of 999 students: https://doi.org/10.1080/10494820.2025.2454448

#cogSci #edu

**Lobsters** @lobsters@mastodon.social · Mar 6

Mar 6

Lobsters @lobsters@mastodon.social

The Hitchhiker's Guide to Digital Colour https://lobste.rs/s/ibfmsg #art #cogsci #graphics
https://hg2dc.com/

lobste.rsThe Hitchhiker's Guide to Digital Colour | Lobsters

**Gary McGraw** @cigitalgem@sigmoid.social · Mar 5

Mar 5

Gary McGraw @cigitalgem@sigmoid.social

Thirty years ago, I published my PhD dissertation on Letter Spirit --- all about creativity, AI, and cognitive science. My advisor was Doug Hofstadter. Last week I was asked to give a keynote talk about that work in France. The talk was captured in video.
#AI #cogsci #creativity #typography

https://berryvilleiml.com/2025/02/26/revisiting-letter-spirit-thirty-years-later/

Berryville Institute of Machine LearningRevisiting Letter Spirit Thirty Years Later | BIML(crossposted to apothecaryshed) I was honored to be asked to present a talk on my thesis work in Nancy at the Automatic