Edwin Dalmaijer
LightNews LightNews
banner
dalmaijer.bsky.social
Edwin Dalmaijer
@dalmaijer.bsky.social
760 followers 650 following 410 posts
Cognitive neuroscientist with many interests, including why our stomachs churn when we feel disgust. I also write books on programming; teach Python, statistics, and machine learning; and develop open-source software.

https://www.dalmaijer.org
Posts Replies Media Videos
dalmaijer.bsky.social
Are you curious about electrogastrography, but keep getting chicken-related results when googling "EGG"? We have the preprint for you!

In this tutorial, we describe how to acquire and analyse gastric data from human participants. Plus FREE software! Read it here: arxiv.org/abs/2509.17260
Image illustrating sensor placement. The reference sensor is placed on the xiphoid process (just under the sternum), the ground is placed on the left costal margin (just under the bottom rib), and sensors one through four are place over where the stomach would be. Sensor 3 is midway through the umbilicus (navel) and the reference sensor. Sensor 4 is roughly three centimetres from sensor 3 to participant-right. Sensors 2 is also roughly three centimetres away from sensor 3, but along a 45 degree angle upwards and to participant-left. Sensor 1 lies 3 centimetres further from sensor 2, along the same 45 degree line. Two-panel image showing a processed electrogastrogram. The left panel shows five minutes of raw data from four different leads (measured from the four sensors in the previous image and their common reference). The right panel shows the same signal in frequency space, illustrating the typical normogastric peak at 3 cycles per minute.
December 18, 2025 at 2:49 PM
dalmaijer.bsky.social
This sent me down a spiral, leading to the discovery that the lolcats I wasted my poor parents’ precious dial-up minutes for are from 19 years ago. How is such recent internet history already two decades old?
Meme of a cat with what looks like a smile on its slightly tilted head, captioned “I CAN HAS CHEEZBURGER?”
December 16, 2025 at 12:45 PM
dalmaijer.bsky.social
Quick post on @jonaspotthoff.bsky.social new paper! We found people looked more at novel foods at an experimental buffet, including insects that they find gross! People suppress disgust avoidance when making foraging decisions.

Blog: www.dalmaijer.org/2025/12/fora...
Paper: doi.org/10.1016/j.fo...
A three by four grid of example stimuli found in the experimental buffet. The first column shows edible insects: meal worms, crickets, and buffalo worms. The second column shows novel savoury snacks for participants in the UK and Austria: Haldiram's Mullu Murukku, roasted broad beans, and Wah-Yuen fried dough. The third column shows common savoury snacks for the same participants: tortilla chips, crisps, and peanuts. The last column shows visually matched non-foods: Lego, wooden buttons, and wooden clothes pegs. Figure with four panels. In each panel, the x-axis depicts the four types of stimuli: insect snacks, novel snacks, familiar snacks, and non-food. The top two panels show total dwell time across the experiment and average fixation duration. Overall, participants looked at the insects and novel snacks more than at the familiar snacks and non-food. The bottom panels show disgust and desire-to-eat ratings. Insects are scored as much higher on disgust and much lower on desire to eat compared to novel and familiar foods (both of which are rated low on disgust and relatively high on desire to eat).
December 12, 2025 at 11:17 AM
dalmaijer.bsky.social
Would you eat insects, or is that disgusting? @jonaspotthoff.bsky.social used mobile eye-tracking as people browsed a buffet. They looked longer at insect and unfamiliar snacks, but only rated insects as disgusting. This suggests disgust isn't always avoided when foraging decisions have to be made!
A three by four grid of example stimuli found in the experimental buffet. The first column shows edible insects: meal worms, crickets, and buffalo worms. The second column shows novel savoury snacks for participants in the UK and Austria: Haldiram's Mullu Murukku, roasted broad beans, and Wah-Yuen fried dough. The third column shows common savoury snacks for the same participants: tortilla chips, crisps, and peanuts. The last column shows visually matched non-foods: Lego, wooden buttons, and wooden clothes pegs. Figure with four panels. In each panel, the x-axis depicts the four types of stimuli: insect snacks, novel snacks, familiar snacks, and non-food. The top two panels show total dwell time across the experiment and average fixation duration. Overall, participants looked at the insects and novel snacks more than at the familiar snacks and non-food. The bottom panels show disgust and desire-to-eat ratings. Insects are scored as much higher on disgust and much lower on desire to eat compared to novel and familiar foods (both of which are rated low on disgust and relatively high on desire to eat).
December 11, 2025 at 1:44 PM
dalmaijer.bsky.social
The perceived embodiment of proper disgust is entirely different than that of “political disgust”, which actually closely overlaps with anger. Excellently timed preprint, with @disgust-nerds.bsky.social starting tomorrow. Hope someone brings this up in the “Is moral disgust really disgust?” debate!
Figure from the preprint with two rows of visualised perceived embodiment, one for the non-political and one for the political emotion. Disgust is perceived by participants to be embodied in the stomach, whereas political disgust was in the chest. This overlaps very closely with political anger, suggesting the two overlap. (And maybe that “political disgust” is not necessarily actual disgust!)
December 3, 2025 at 1:20 PM
dalmaijer.bsky.social
Reviewer asked something similar for a recent paper on disgust (doi.org/10.1371/jour... ), and we wrote what’s in the image. TL;DR: behaviour can be as good as self-report, and is substantially more robust against demand effects. (CC @tahnee-engelen.bsky.social and @nicolecrust.bsky.social)
Paragraph from Edgar et al. (2024, PLOS ONE): “The summarised studies mostly rely on self-report measures of disgust elicited by written or picture stimuli. Such self-report measures can have excellent psychometric properties, for example the internal consistency [6] and external agreement with others’ ratings [20] of the Three-Domain Disgust Scale, or the high test-retest reliability of self-reported disgust for images [12]. While experimental tasks are often less reliable than self-report measures [21], behavioural disgust avoidance measured in preferential looking tasks has excellent internal consistency, Spearman-Brown ρ = 0.81 to 0.94 and Cronbach’s α = 0.78 to 0.91 [22], and test-retest reliability equal to self-reported disgust ratings [12]. Behavioural disgust avoidance correlates with self-reported disgust sensitivity, R = 0.28 to R = 0.34 [12], and with self-reported disgust ratings for stimuli, R = 0.19 to 0.50 [22]. However, a major difference between self-report and behavioural measures of disgust avoidance is that self-reported disgust shows a very large (Cohen’s d = 1.25) placebo effect, whereas oculomotor avoidance (gaze dwell time) in preferential looking tasks only shows a small to medium (Cohen’s d = 0.44) effect [23].”
November 28, 2025 at 8:25 PM
dalmaijer.bsky.social
Obligatory reference to “Falsehoods Programmers Believe About Names”: www.kalzumeus.com/2010/06/17/f...
People have exactly one canonical full name. People have exactly one full name which they go by. People have, at this point in time, exactly one canonical full name. People have, at this point in time, one full name which they go by. People have exactly N names, for any value of N. People’s names fit within a certain defined amount of space. People’s names do not change. People’s names change, but only at a certain enumerated set of events. People’s names are written in ASCII. People’s names are written in any single character set. People’s names are all mapped in Unicode code points. People’s names are case sensitive. People’s names are case insensitive. People’s names sometimes have prefixes or suffixes, but you can safely ignore those. People’s names do not contain numbers. People’s names are not written in ALL CAPS. People’s names are not written in all lower case letters. People’s names have an order to them. Picking any ordering scheme will automatically result in consistent ordering among all systems, as long as both use the same ordering scheme for the same name. People’s first names and last names are, by necessity, different. People have last names, family names, or anything else which is shared by folks recognized as their relatives. People’s names are globally unique. People’s names are almost globally unique. Alright alright but surely people’s names are diverse enough such that no million people share the same name. My system will never have to deal with names from China. Or Japan. Or Korea. Or Ireland, the United Kingdom, the United States, Spain, Mexico, Brazil, Peru, Russia, Sweden, Botswana, South Africa, Trinidad, Haiti, France, or the Klingon Empire, all of which have “weird” naming schemes in common use. That Klingon Empire thing was a joke, right? Confound your cultural relativism! People in my society, at least, agree on one commonly accepted standard for names. There exists an algorithm which transforms names and can be reversed losslessly. (Yes, yes, you can do it if your algorithm returns the input. You get a gold star.) I can safely assume that this dictionary of bad words contains no people’s names in it. People’s names are assigned at birth. OK, maybe not at birth, but at least pretty close to birth. Alright, alright, within a year or so of birth. Five years? You’re kidding me, right? Two different systems containing data about the same person will use the same name for that person. Two different data entry operators, given a person’s name, will by necessity enter bitwise equivalent strings on any single system, if the system is well-designed. People whose names break my system are weird outliers. They should have had solid, acceptable names, like 田中太郎. People have names.
November 10, 2025 at 3:38 PM
dalmaijer.bsky.social
Delighted to see another validation of MouseView.js, this time to test its usability for biases to alcohol-related stimuli. It was reliable, and captured overt attentional bias in those with high problematic drinking scores. For more info, read the paper by Thulin et al.: doi.org/10.1016/j.ps...
The image is a screenshot of the title, authors, and abstract of the paper. Title: "Testing the utility of Mouseview.js for measuring associations between alcohol related attentional bias and problematic alcohol use". Abstract: "Objectives: This study examined the utility of Mouseview.js, an alternative to webcam-based eye-tracking, to assess associations between alcohol-related attentional bias and problematic alcohol use. Methods: Canadians recruited through Academic Prolific (N = 526) completed an online free-viewing task using Mouseview.js to measure biased attention to alcohol-related images. Eighteen alcohol-related images were paired with neutral images matched on visual characteristics. Attentional bias was estimated as the differences in trial-level dwell times for the image pairs. The sample consisted of 132 non-drinkers, 241 recreational drinkers, and 153 problematic drinkers. Participants’ mean age was 33.5 years, 50.8% were men, and 40.4 % identified as a racialized minority. Multi-level modelling was used for statistical analyses. Results: The difference in dwell times between alcohol-related and neutral images was larger for problematic drinkers compared to non-drinkers, B = 3.75, z = 2.61, p = .009. There were no significant differences between recreational drinkers and non-drinkers, or between recreational and problematic drinkers. Reliability for dwell times and dwell time differences ranged from acceptable (≥0.70) to good (≥0.80). Conclusions: There was an attentional bias for alcohol-related images among problematic drinkers relative to non-drinkers and the Mouseview.js indices of bias demonstrated adequate reliability. These findings provide preliminary support for the utility of Mouseview.js as a measure of attentional bias in problematic drinkers. With further validation studies, Mouseview.js may prove to be a powerful tool for conducting high-powered online studies of attentional biases associated with addictive behaviors."
October 17, 2025 at 7:28 AM
dalmaijer.bsky.social
Such a lovely talk, and I’m glad our admin upgraded the venue as there was standing room only! I particularly enjoyed Aidan’s perspective on hippocampal and non-hippocampal encoding as independent processes, and flexibility of memories within that space. (I also enjoyed the brief Mario Kart break.)
Photo of a projector screen showing the results of a Mario Kart race. Two players are shown, one as Wario in first place and another as Peach in third place.
September 24, 2025 at 2:49 PM
dalmaijer.bsky.social
I've seen people use "clanker" in the wild, and others commenting how wild it is to live through the invention of the first anti-AI slur.

Few seem to realise "clanker" is from a fairly niche Star Wars game and show, where battle droids ("roger, roger!") are called this by clone troopers.
This is a version of the "The More You Know" meme: a screenshot from NBC public service announcement showing a shooting star with a rainbow trail that underscores blue text that says "The More You Know". In this version, the shooting star has the Death Star from Star Wars superimposed, and a Battle Droid seems to be heroically jumping into the frame (from the left) while holding their blaster ready for shooting.
August 4, 2025 at 3:30 PM
dalmaijer.bsky.social
New gender just dropped!
Text from a paper by Flohr-Devaud and Neumann (2025, Translational Issues in Psychological Science, doi: https://doi.org/10.1037/tps0000469 ). The screenshot offers a larger extract from the Methods section to show the "Sample" heading for context, but the crucial bit is "Paramedics (40 male, 24 female, one diver) from a German emergency service organization (M age = 33.25, SD = 12.87) participated." The section "one diver" is highlighted for the purpose of this post (but not in the original paper).
July 29, 2025 at 12:34 PM
dalmaijer.bsky.social
To check if our data was just weird, we ran a control analysis to test the effect of participants' ratings of food healthiness on their stomach response. Those awarding higher healthiness ratings increased showed higher normogastric power for healthy and lower for unhealthy foods. Sensible! (5/7)
Figure comprised of three subplots. Each subplot represents a third of the sample, here ordered by the healthiness ratings that participants gave to foods in each category. Each panel shows three lines, each depicting gastric power along its typical frequencies (x-axis, ranging from 1 to 10 cycles per minute). Each line represents a food type: "healthy" (in green), "unhealthy" (in yellow), and "disgusting" (in brown). The important point is that normogastric power is higher for those who gave low healthiness ratings, gastric power is high for unhealthy foods and low for healthy foods. However, those who gave high healthiness ratings show higher gastric power for healthy foods. (Note that the presentation in three groups is just for visualisation; the actual statistical analysis treats orthorexia as continuous variable and shows an interaction between it and food conditions.)
July 4, 2025 at 10:34 AM
dalmaijer.bsky.social
We expected reduced normogastric power (proto-nausea) for unhealthy foods, but found the opposite! Those higher on orthorexia had similar stomach responses to healthy and unhealthy foods. Do self-denied foods become more appetitive? Or is orthorexia cognitive control to overrule the stomach? (4/7)
Figure comprised of three subplots. Each subplot represents a third of the sample, here ordered by orthorexia scores. Each panel shows three lines, each depicting gastric power along its typical frequencies (x-axis, ranging from 1 to 10 cycles per minute). Each line represents a food type: "healthy" (in green), "unhealthy" (in yellow), and "disgusting" (in brown). The important point is that normogastric power is higher for healthy compared to unhealthy foods for those who score lower on orthorexic traits. However, for those high on orthorexic traits, this difference disappears. (Note that the presentation in three groups is just for visualisation; the actual statistical analysis treats orthorexia as continuous variable and shows an interaction between it and food conditions.)
July 4, 2025 at 10:34 AM
dalmaijer.bsky.social
We also found that people higher on orthorexia literally turned up their noses at unhealthy and disgusting foods! The coefficient vector maps align with facial expressions of disgust. At this point, orthorexia is associated with a preference for healthy foods and disgust for anything else. (3/7)
Perfect figure for an album cover, depicting nine faces in a 3x3 grid. The middle face looks directly at the viewer, and each other face is looking away in the direction they are in from the central face (e.g. top-left is looking towards top-left). The faces are made up of facial landmarks (grey disks), from each of which originates a coefficient vector. This line is longer and brighter for stronger effects. In this case, that effect is an interaction between orthorexic traits and facial responses to disgusting foods (compared to healthy ones). The largest effects are around the nose, indicating pulling up of the levator labii. There is also a bit of squinting and brow furrowing.
July 4, 2025 at 10:34 AM
dalmaijer.bsky.social
We used three categories of foods: "healthy" (minimally processed), "unhealthy" (highly processed), and "disgusting" (uncommon in local culture). Participants' ratings confirmed we chose our stimuli well, and people higher on orthorexia reported higher desire to eat healthy foods. (2/7)
This is a figure depicting a series of 15 box plots with density estimates. There are three ratings along the x-axis (healthiness, disgust, familiarity, pleasantness, and desire to eat). Each rating is given for 45 stimuli, 15 each in three food categories: "healthy" (in green), "unhealthy" (in yellow), and "disgusting" (in brown). Healthy foods received higher healthiness ratings; disgusting foods received higher disgust ratings; and familiarity, pleasantness, and desire to eat show very similar distributions between healthy and unhealthy foods (but all are lower for disgusting foods!).
July 4, 2025 at 10:34 AM
dalmaijer.bsky.social
Ha! I either missed this at the time, or so thoroughly blocked it from memory that it no longer rings familiar! What I do recall is the meeting where you, Kia (@brognition.bsky.social), and several uni officials called us all in to tell us the building would be closing. I still have the photos!
Photo taken from a seat in the Tinbergen’s main lecture theatre, looking out onto the speaker area. Five people, including the Heads of Departments in the Tinbergen and several university officials, are at the front. The speakers are informing staff of the Tinbergen closure, and answering questions.
June 26, 2025 at 6:34 PM
dalmaijer.bsky.social
If you're thinking "hey, that looks a bit like a sausage fest, and what's with all the references to masculinity in the photo captions?" I have good news! The hospital hired its first women in the two decades after this photo was taken, e.g. Dr Pearl Jane Sproule, chief of ENT 1924-1940.
Yet another image from "Doing Good: The Life of Toronto's General Hospital" by J.T.H. Connor. It shows an official photograph of two rows of physicians: nine standing in the back row, and five seated in the front row. The caption reads: "Although medicine remained an exquisitely male domain, in the twentieth century women started to breach its ranks. Dr Pearl Jane Sproule (seated left) was among the first to be appointed to the hospital's house staff; from 1924 to 1940, she was chief of the ear, nose, and throat department of Toronto's Women's College Hospital. Dr J.N.E. Brown, then medical superintendent, is seated centre."
May 12, 2025 at 10:26 AM
dalmaijer.bsky.social
I ended up looking up the book, and found the photo with the cited caption but without further context. I assume the newborn babies are indeed borrowed from the maternity ward. The book has pages and pages of photos with brief captions, some of which raise further questions (e.g. attached).
The image is again from the book "Doing good : the life of Toronto's General Hospital" by J.T.H. Connor. It shows ten physicians sat at a table. They are wearing black suits, most with their white cotes still on. Some appear to be smoking or eating, and one is draped over another. The caption reads "Members of the house staff had their own quarters in the hospital. As this 1898 glimpse of 'off-duty' behaviour illustrates, the manly pursuit of boxing, the consumption of beer, and the enjoyment of a good cigar contributed to 'male bonding'."
May 12, 2025 at 10:20 AM
dalmaijer.bsky.social
Older journal articles are typically uploaded as pages from the print copy, which means you often get the start and/or end of unrelated papers as a little bonus. This is the little bonus I got from @jama.com this morning. There is zero further context.
The image is a screenshot from a 2001 edition of the Journal of the American Medical Association (JAMA), volume 285, number 17. It shows fourteen physicians, posing for a group photo in their white coats. What makes this photo unique, is that all fourteen men are holding newborn babies. The photo caption reads: "“Even while ‘on duty,’ residents might indulge themselves to show the tender side of their masculine nature. Here members of house staff in 1904-5 pose with babies born in the Burnside Lying-In wing.” From: Doing Good: The Life of Toronto’s General Hospital, by J. T. H. Connor, 342 pp, with illus, $60, ISBN 0-8020-4774-2, Toronto, Ontario, Canada, University of Toronto Press, 2001. Reprinted by permission of University Health Network, Toronto, Ontario, Canada."
May 12, 2025 at 9:08 AM
dalmaijer.bsky.social
A real strength of this study is that we asked participants for their individual ratings for each stimulus and emotion. It allowed us to account for 1) individual differences in stimulus perception, and 2) stimuli often evoking a combination of core emotions.
The image is a graph with six sub-plots, each in a new row, top-to-bottom: Pupil size (arbitrary units), anger, disgust, fear, happy, and sad (all emotions rated 1-9). The x-axis shows the stimulus numbers, each coding for a different 30-second sound-stimulus from Experiment 2 in the paper. A box-plot with distribution estimate is visible for each stimulus-rating combination. Low ratings almost always showing strong agreement within each stimulus, but higher ratings show variability. For example, stimulus 19 was almost exclusively rated low on anger, disgust, fear, and sadness; but while its happiness rating was reasonably high on average (about 6), there was still substantial variability (especially in the 5-9 range). Key take-aways from the plot are that ratings are variable but have some agreement, that stimuli can be rated high for more than one emotion, and that high disgust ratings mostly occured for stimuli that are also higher in pupil size.
May 6, 2025 at 11:35 AM
dalmaijer.bsky.social
NEW! Using #pupillometry during emotional stimuli, we found if/when individual stimulus ratings are associated with pupil size. In short, disgust was the strongest pupil dilator, whereas anger constricts it (a bit).

Paper by fantastic first-author @kmccul.bsky.social et al.: doi.org/10.1016/j.bi...
The image is a graph that shows five lines (one for each emotion), punctuated with small dots for uncorrected statistical significance and bigger dots for statistical significance after multiple-comparisons correction. The x-axis is labelled "Time (seconds)" and runs from just before 0 to 30 seconds. The y-axis reads "Standardised beta coefficient" and runs from just under -0.15 to 0.3. A dotted line spans across y=0, and labels indicate that positive values mean that "Emotion predicts larger pupil" and negative values that "Emotion predicts smaller pupil". Anger (in red) shows a negative trajectory from ~5 seconds after stimulus onset, with some bits statistically different from 0. Fear (purple) is indistinguishable from 0, until it is associated with larger pupils in the last 2-3 seconds. Sadness is somewhat positively associated with pupil size, but not after multiple-comparisons correction. Happiness is associated with relative pupil dilation from ~3 seconds until the end of the stimulus. Disgust is by far the most prominent predictor of larger pupil size, with standardised coefficients about twice as high as for happiness.
May 6, 2025 at 11:03 AM
dalmaijer.bsky.social
One of my favourite parts of supervising undergraduate student projects is seeing some of them present at the Southwest undergrad conference. It feels like a proper conference, but is organised by and for final-year students. Lovely to see so many keen beans talking about their work! #SWUC2025
Image is a photo with a view from the back of a conference room, looking at students speaking in front of a projector slide that says “BPS Southwest Undergraduate Conference 2025”.
April 26, 2025 at 9:16 AM
dalmaijer.bsky.social
Was joined by a cheeky little robin for lunch today. Does this count as #UKbirding, or was I the subject of its human-ing?
This photo shows a pair of shoes, seemingly at the end of crossed legs (not visible). On the tip of the toes is a robin, stood straight up, with its head turned so that one eye looks into the camera.
April 11, 2025 at 11:30 AM
dalmaijer.bsky.social
OK, but:
Histogram showing a roughly normal distribution that is centred on 0.5. The x-label is "N(top&bottom) / N(top)", the y-label is "N observations". The face colour of the bars is hot pink.
March 4, 2025 at 11:19 AM
dalmaijer.bsky.social
If, for whatever reason, you're looking for a novel with a fresh take on AI tech dystopia: Adrian Tchaikovsky's Service Model is excellent.

It's a clever take on robot cognition (but is it really cognition?), human nature, and societal collapse through perfectly normal function.
The cover of Service Model by Adrian Tchaikovsky. The background shows a modern skyline against a sunset-coloured sky. The foreground shows a robot hand that is shaped like a human's. The hand is delicately holding a cup that is slightly cracked and a has some thick red liquid along the rim.
February 8, 2025 at 9:51 AM