Hi friends! I'm finally feeling well enough to share some news that I've been very excited about: earlier this week, my kidney was removed through a small cut in my stomach, tucked into a box, flown down to johns hopkins, and implanted into a stranger where its working great

Thrilled this paper is finally out (I ran the first pilots in 2022), and writing it with Brian Scholl was such a joy (his work on the El Greco fallacy was a clear inspiration). Happy to chat with those who want to use these measures (or disagree!).

Though we tested just one claim about one group difference in one emotion, the problem is much broader. Group differences in emotional intensity very well may exist (I have no horse in this race), but if they do, then existing measures probably couldn't find them. (13/14).

Though we replicated the purported sex difference in anger using the standard 10-point scale (twice, actually), it reliably disappeared when we used the gLMS (even though we had a sample so large that it could detect an effect of only d=.14 with 99% power). (12/14)

So if only measures like the gLMS can find group differences in perceptual sensations, what happens if you use those same measures in the context of emotion? That's exactly what we did in our paper (preprint below for those w/o access). (11/14).

Even though these real differences are massive (much larger than any claimed difference in emotion), they're still reliably hidden by 10-point scales. Thus, taste researchers had to develop entirely new measures (like the general Labeled Magnitude Scale, or gLMS). (10/14)

What's so cool about supertasters is that they differ in more than just their experience of bitterness, they experience ALL tastes more intensely. Salt is saltier, sugar is sweeter, lemons are lemonier, and the meaning of "very strong" *itself* is very. stronger. (9/14)

This problem has been long-recognized by researchers studying differences in perceptual sensations. For example, a chemical used in thyroid medicine (6-n-propylthiouracil) tastes extremely bitter to so-called "supertasters" and entirely neutral to "nontasters." (8/14).

This mistake (called the El Greco fallacy) is just as much a problem for our 10-point scales. If a person (or a group) feels anger more intensely, then our scales would never show as much, since that difference in anger changes what "very intense" even means. (7/14)

If you think about it, though, that can't possibly be why. The world would be elongated, sure, but so would the canvas and his brush strokes, meaning that the distortion would cancel out (psivision on TikTok has a great and funny explainer). (6/14) .

This is what El Greco's painting *actually* looks likes: unusually elongated. Art historians puzzled over why El Greco painted this way, and one art historian provided a simple answer: astigmatism. El Greco saw the world as elongated, and he simply painted what he saw. (5/14)

But there's a problem. That can't work, and to understand why, consider the following painting of Saint John the Baptist by the Spanish Renaissance painter El Greco. On first glance, this painting seems totally normal, but that's only because I lied to you. (4/14)

Because of this, researchers typically use self-report, e.g. a 10-point scale ranging from "not at all intense" to "very intense." If you want to know whether two groups differ in the intensity of some emotion, you'd just average the ratings each gives on that scale. (3/14).

Like any other sensation, having an emotion is an inherently subjective and necessarily private experience, which means that (in the words of Natasha Bedingfield’s 2004 chart-topping multi-platinum hit, "Unwritten") no one else can feel it for you. (2/14).

Do women feel some emotions more strongly than men? Out today in Affective Science, I argue that claims like this make a notoriously subtle mistake. What is it? And what does it have to do with an astigmatic painter and thyroid medicine? A short thread.

The whole reply is pretty short (~1400 words), and feedback greatly appreciated (I spent just about all of February reading up on this stuff, but its still pretty far outside my wheelhouse). Attaching the link one more time:

So what seems like a speed limit might actually be something like a solution to an optimization problem, e.g. the tension between between seeking new information (exploring) and pursuing a known reward (exploiting). This is of course speculative, but it's straightforward to test.

We also see something surprising: the information throughput of AI isn't just in the ballpark of 10 bits/s, it's converging toward it. Language models with the highest bitrate had the worst performance on standard benchmarks, and models playing StarCraft do best at 10 bits/s.

So if we can't use our most familiar point of reference, how should we interpret this finding? An obvious candidate for an apples-to-apples comparison would be artificial intelligence. What does their throughput look like if we calculate it the same way? Actually very similar!

In the context of Shannon's entropy, an information throughput of 10 bits/s is equivalent to differentiating 1024 equally likely alternatives each second. And the bitrate of human language (39 bits/s) is almost comically (or cosmically) huge.

A quick summary: we're all pretty familiar with the word "bit" in the context of everyday data transmission over something like WiFi, but this isn't necessarily a good point of comparison for a bit in the information theoretic sense (i.e. Shannon's entropy).