Samuel Hewitt, PhD
@samuelhewitt.bsky.social
920 followers 580 following 88 posts
Machine Learning Engineer at Limbic Building AI therapy Previously, Max Planck Centre for Computational Psychiatry @ UCL
Posts Media Videos Starter Packs
Pinned
samuelhewitt.bsky.social
A little bby from my PhD is published!

we did a complex study to answer a simple Q:
do day-to-day feelings impact choices?

A: Yes. When feeling motivated (independent of other stable or unstable feelings) choices changed because rewards seemed greater.

open access link:
doi.org/10.1073/pnas...
Day-to-day fluctuations in motivation drive effort-based decision-making | PNAS
Internal states like motivation fluctuate substantially over time. However, studies of the neurocomputational mechanims of motivated behavior have ...
doi.org
samuelhewitt.bsky.social
Here here !! Well done CPSY
xiaosigu.bsky.social
🔥💥📣 Excited to announce a new policy at the Computational Psychiatry (CPSY) journal @cpsyjournal.bsky.social. Starting from April 2025, CPSY will compensate reviewers for their time and effort invested in reviewing manuscripts. w/ @drrickadams.bsky.social
cpsyjournal.org/articles/10....
One Small Step Towards Fixing a Broken System | Computational Psychiatry
cpsyjournal.org
Reposted by Samuel Hewitt, PhD
samuelhewitt.bsky.social
I think that's already the case !! We can stop measuring the function of bored brains now
samuelhewitt.bsky.social
it wrote all the front-end code ! just need to connect your database and adjust the trials. I just gave it this image for the style
samuelhewitt.bsky.social
Hey cognitive scientists

I asked lovable.dev to make me a 2-arm bandit RL task

2 prompts and 3 minutes later
🤯🤯🤯🤯

Play it:
preview--cosmic-bandit-quest.lovable.app

Think of the hours and grad student tears saved 😅

Yes lovable I accept your sponsorship terms
samuelhewitt.bsky.social
Make a choice and get feedback
Make a choice and get reward feedback
samuelhewitt.bsky.social
When I began my PhD, there was a major disconnect between my own lived experience and the way researchers probed mental health-behaviour links.

I hope this work inspires others to use new designs to appreciate the complexity of human beings and their experiences.
link.growkudos.com/1f5ss3bn1ts
samuelhewitt.bsky.social
this is super cool and clever. Well done, really like it. Sth like this has been scratching at me for a while too - congrats on putting it together
Reposted by Samuel Hewitt, PhD
docqhuys.bsky.social
This is a great study, if I may say. We have had a lot of data showing correlations between subjective assessments and behaviour between individuals. With this study, we are starting to see how they interact over days within individuals. Congratulations @samuelhewitt.bsky.social!
samuelhewitt.bsky.social
A little bby from my PhD is published!

we did a complex study to answer a simple Q:
do day-to-day feelings impact choices?

A: Yes. When feeling motivated (independent of other stable or unstable feelings) choices changed because rewards seemed greater.

open access link:
doi.org/10.1073/pnas...
Day-to-day fluctuations in motivation drive effort-based decision-making | PNAS
Internal states like motivation fluctuate substantially over time. However, studies of the neurocomputational mechanims of motivated behavior have ...
doi.org
samuelhewitt.bsky.social
thank you to brilliant collaborators Prof @docqhuys.bsky.social and Dr Agnes Norbury (now Thymia) and my PhD supervisor Prof @tobiasuhauser.bsky.social
samuelhewitt.bsky.social
This suggests that specific transient feelings can drive computational decision mechanisms in the future.

We think this study design can allow many new questions in computational psychiatry, bridging long established lab theories with real, human experiences and potentially clinical translation
samuelhewitt.bsky.social
State motivation-choice coupling was driven by fluctuations in reward sensitivity. This is a fancy way of saying that when more motivated, (the same) rewards seemed more rewarding.

What I think is SUPER 😎 is that this was not only rewards NOW, but also rewards at the NEXT timepoint.
samuelhewitt.bsky.social
State motivation was a critical source of variability in choices over time (leading to greater willingness to make effort, duh).

But state motivation also interacted with trait-motivation, meaning that people with LOWER trait (higher apathy) had even stronger state-choice coupling.
samuelhewitt.bsky.social
N.B. our game could also reliably capture effort-based choices and the model-parameters which govern this value-based decision