Brad Wyble
@bwyble.bsky.social
Academic, cognitive & vision scientist, computational modeller, cofounder @neuromatch Academy, He/His. This is a personal account.
What I expected, of course, is that memory for these passages would improve systematically from 0th order to full text, but that's not quite what happened:
November 13, 2025 at 5:22 PM
What I expected, of course, is that memory for these passages would improve systematically from 0th order to full text, but that's not quite what happened:
Examples of complete gibberish (which they call a 0th order approximation to english):
November 13, 2025 at 5:15 PM
Examples of complete gibberish (which they call a 0th order approximation to english):
Examples of coherent prose of varying length:
November 13, 2025 at 5:13 PM
Examples of coherent prose of varying length:
I changed my profile picture, which they are welcome to use in perpetuity, worldwide, in any manner.
September 17, 2025 at 9:34 PM
I changed my profile picture, which they are welcome to use in perpetuity, worldwide, in any manner.
Prince of Persia was a groundbreaking game for its accurate character movement back in the late 80's. To do this, the creator videotaped his own brother running and jumping on the sidewalk and then did pixel mapping frame by frame as shown here at the Strong Museum of Play
September 7, 2025 at 11:31 PM
Prince of Persia was a groundbreaking game for its accurate character movement back in the late 80's. To do this, the creator videotaped his own brother running and jumping on the sidewalk and then did pixel mapping frame by frame as shown here at the Strong Museum of Play
In our two experiments, we found, then replicated, an accuracy boost for displays containing related pairs in visual working memory. These results suggest that our functional understanding of object interactions facilitates working memory compression. >
August 28, 2025 at 12:12 PM
In our two experiments, we found, then replicated, an accuracy boost for displays containing related pairs in visual working memory. These results suggest that our functional understanding of object interactions facilitates working memory compression. >
Our new study (Titled: Memory Loves Company) asks whether working memory hold more when objects belong together.
And yes, when everyday objects are paired meaningfully (Bow-Arrow), people remember them better than when they’re unrelated (Glass-Arrow). (mini thread)
And yes, when everyday objects are paired meaningfully (Bow-Arrow), people remember them better than when they’re unrelated (Glass-Arrow). (mini thread)
August 28, 2025 at 12:07 PM
Our new study (Titled: Memory Loves Company) asks whether working memory hold more when objects belong together.
And yes, when everyday objects are paired meaningfully (Bow-Arrow), people remember them better than when they’re unrelated (Glass-Arrow). (mini thread)
And yes, when everyday objects are paired meaningfully (Bow-Arrow), people remember them better than when they’re unrelated (Glass-Arrow). (mini thread)
Look at this interesting google trend graph for "attention". Those cycles peak in October and March of each year like clockwork. My guess is that these are Psych perception & cognition classes. Are there other explanations? @sbmost.bsky.social @bjbalas.bsky.social @benwolfevision.bsky.social
July 7, 2025 at 6:31 PM
Look at this interesting google trend graph for "attention". Those cycles peak in October and March of each year like clockwork. My guess is that these are Psych perception & cognition classes. Are there other explanations? @sbmost.bsky.social @bjbalas.bsky.social @benwolfevision.bsky.social
Drawers? Why not a whole dresser?
May 22, 2025 at 10:05 PM
Drawers? Why not a whole dresser?
But it's just not doing a great job, still. And yet we're being forced to integrate into our jobs. Being upset seems completely reasonable.
May 2, 2025 at 4:55 PM
But it's just not doing a great job, still. And yet we're being forced to integrate into our jobs. Being upset seems completely reasonable.
New Preprint! Interested in learning about how working memory is subserved by both compositional and generative mechanisms? Read on!
April 14, 2025 at 2:24 AM
New Preprint! Interested in learning about how working memory is subserved by both compositional and generative mechanisms? Read on!
I asked a friend with access to the new GPT image model to create a poster aggressively attacking my work on the attentional blink and it created this masterpiece. Perhaps we finally found a practical use for these models.
March 27, 2025 at 2:26 PM
I asked a friend with access to the new GPT image model to create a poster aggressively attacking my work on the attentional blink and it created this masterpiece. Perhaps we finally found a practical use for these models.
As a test of google, I searched for the attentional blink, which I know a lot about:
So much of this is wrong. It's not an inability to recognize, it's not that the T2 is "too soon" ( lag 1 sparing eliminates this).
Just really sad that simplistic and wrong takes are now presented as fact.
So much of this is wrong. It's not an inability to recognize, it's not that the T2 is "too soon" ( lag 1 sparing eliminates this).
Just really sad that simplistic and wrong takes are now presented as fact.
January 17, 2025 at 7:22 PM
As a test of google, I searched for the attentional blink, which I know a lot about:
So much of this is wrong. It's not an inability to recognize, it's not that the T2 is "too soon" ( lag 1 sparing eliminates this).
Just really sad that simplistic and wrong takes are now presented as fact.
So much of this is wrong. It's not an inability to recognize, it's not that the T2 is "too soon" ( lag 1 sparing eliminates this).
Just really sad that simplistic and wrong takes are now presented as fact.
I haven't gotten the Gemini upgrade to my gmail yet but I'm doing what I can to avoid it
January 17, 2025 at 2:37 AM
I haven't gotten the Gemini upgrade to my gmail yet but I'm doing what I can to avoid it
People keep rediscovering what is essentially the testing effect, which psychologists identified at least as early as the 1960's.
January 13, 2025 at 3:47 PM
People keep rediscovering what is essentially the testing effect, which psychologists identified at least as early as the 1960's.
These are not just from gotcha's either. E.g. check out this output from Gemini. Not only is the information dangerously wrong, it is inconsistent within the space of a few sentences.
December 31, 2024 at 5:27 PM
These are not just from gotcha's either. E.g. check out this output from Gemini. Not only is the information dangerously wrong, it is inconsistent within the space of a few sentences.
Still batting 100 Gemini. (Out of 1000)
December 31, 2024 at 3:44 PM
Still batting 100 Gemini. (Out of 1000)
Getting 80 young adults to answer 50 ambiguous MCQ questions the same way I would.
December 30, 2024 at 9:47 PM
Getting 80 young adults to answer 50 ambiguous MCQ questions the same way I would.
I feel it's like this though. Why does every place need to cater to every viewpoint?
December 14, 2024 at 1:04 PM
I feel it's like this though. Why does every place need to cater to every viewpoint?
Legendary game of hearts last night.
November 29, 2024 at 2:42 PM
Legendary game of hearts last night.
Slightly better but you can see why AI is no replacement for real art
November 12, 2024 at 11:01 PM
Slightly better but you can see why AI is no replacement for real art
Explaining to my grandkids that back in my day a picture like this required you to actually get 700 beds and put them on the beach instead of asking AI to draw it.
[Left picture is from Pink Floyd's 1987 Momentary Lapse of Reason album]
[Left picture is from Pink Floyd's 1987 Momentary Lapse of Reason album]
November 12, 2024 at 10:53 PM
Explaining to my grandkids that back in my day a picture like this required you to actually get 700 beds and put them on the beach instead of asking AI to draw it.
[Left picture is from Pink Floyd's 1987 Momentary Lapse of Reason album]
[Left picture is from Pink Floyd's 1987 Momentary Lapse of Reason album]