Steven

Action repetition biases choice in context-dependent decision-making - Communications Psychology This study shows that decision biases previously attributed to value normalization (e.g. relative value learning or range normalization) are better explained by action repetition. Repeating an action ...

Very happy that this is out www.nature.com/articles/s44.... Together with @stefankiebel.bsky.social we show that decision biases in context-dependent decision making, previously attributed to different forms of value normalization, are very well explained by habit-like action repetition.

A habit and working memory model as an alternative account of human reward-based learning Nature Human Behaviour - In this study, Collins proposes an alternative dual-process (working memory and habit) model of reinforcement learning in humans.

My paper is out!
Computational modeling of error patterns during reward-based learning show evidence that habit learning (value free!) supplements working memory in 7 human data sets.
rdcu.be/eQjLN

Participants in study 2 were instructed to not eat after their standardised dinner in both sleep conditions. Ghrelin was increased by both hunger and sleep deprivation in this study.

Those are great questions. I agree that the inclusion of acyl ghrelin would strengthen our claims. However, we did not have acyl ghrelin data for both studies, so to allow for the comparison we focused on des-acyl ghrelin.

Ghrelin and risky decision-making: No credible evidence for homeostatic state modulation of neural or behavioural effects Risk-taking is often thought to depend on homeostatic systems. However, evidence remains mixed, and the underlying mechanisms remain debated. One candidate that might drive changes in risky decision-m...

More details are in the preprint ( doi.org/10.1101/2025... )
Let me know if you have any questions

Our results suggest that state-dependent influences, via ghrelin in particular, on risky decision-making are weaker than previously thought.

The extracted 𝛽‐coefficients of subjective value in the right PCC (A; 𝑋=2,𝑌=−32,𝑍=38) and VS (B; 12,11,−14) for study 1, and the right PCC (C; 2,−32,34) for study 2. The 𝛽‐coefficients support the finding of a main effect of subjective value but no effect of condition.

When looking at the neural activity, we found again no effect of ghrelin. Not for subjective value (figure) or choices.

We compared several computational models to understand the underlying processes. It is possible that the same behaviour comes from different parameter values. We found that the prospect theory provided the best fit to the data and that none of its parameters differed between conditions.

Fig.A) Blue boxplot with probability of risky choice in sated condition, red boxplot with probability of risky choice in fasted condition. Fig.B) Blue boxplot with probability of risky choice after one night of normal sleep, green boxplot with probability of risky choice after one night of total sleep deprivation

This is not what we found. While we saw that our manipulations increased ghrelin levels, we did not see an increase in risky choices in either study.

In study 2, another group of participants did the same. However, they did it after one night of normal sleep (NNS) and after one night of total sleep deprivation (TSD). Both manipulations increase the ghrelin levels so we expected increased risk taking in the fasting condition and the TSD condition.

We set out to investigate this effect with two separate studies. Participants had to choose between €20 guaranteed or a larger amount with a smaller reward probability. In study 1 they did this while sated (SAT) and after a short fasting period (FAS).

Blood levels of ghrelin increase when people get hungry. And due to its interaction with the dopaminergic reward system, it is a prime candidate to explain why we become more risk-seeking when lacking food.

Previous studies found that people make more risky decisions when they are hungry or sleep deprived. One of the suspects behind this effect is “the hunger hormone” ghrelin.

‼️New preprint‼️
There does not seem to be an effect of ghrelin on risky decision-making in probability discounting. Not in behaviour, underlying computational processes, or neural activity.
More details ⬇️

Congrats on the new work, it looks great! (I think the repo is still private because I get a 404 when I try to open it)

A short summary of the study is here:
2025.ccneuro.org/poster/?id=j...
The actual paper should follow soon.
🧠📈🧠💻

Today I will present some of our lab's work at #CCN2025
In two separate studies, we investigated the effect ghrelin, the hunger hormone, on probability discounting. We manipulated ghrelin levels by fasting (study 1) and 1 night if sleep deprivation (study 2)
More information and figures on poster 80

ScriptsThesisUncertainty/fns/behavioural_functions.py at main · StevenGeysen/ScriptsThesisUncertainty Scripts for my master thesis "Orienting in an Uncertain World" - StevenGeysen/ScriptsThesisUncertainty

I used RL for my Master thesis and the scripts are all available github.com/StevenGeysen.... They are not a great example for organization though. For that I have since then learned to use the data science cookiecutter cookiecutter-data-science.drivendata.org

Beyond Focal Lesions: Dynamical Network Effects of White Matter Hyperintensities By combining whole-brain neural mass modeling with a virtual-lesioning approach informed by empirical WMH lesion masks, we provide evidence that WMH induce changes in brain dynamics by reducing the g...

2 years ago I built a fMRI preprocessing pipeline for ADNI data. Now I can share the work that was done with it.
Riccardo Leone, in the lab of Xenia Kobeleva, used it to model the effects of white matter hyperintensities using whole-brain modelling 🧠💻🧠📈
onlinelibrary.wiley.com/doi/10.1002/...

From our lab!