Our group within Connected Minds Lab studies how people update their mood and choices in algorithmic social feeds. We have investigated how received rewards from others (da Pinho et al., 2024), content reward and your own behaviour co-vary with short-term mood. This project will try to bring light to the mechanisms behind these effects combining (i) large-scale, real-world platform data, (ii) formal reinforcement-learning models of trial-by-trial mood/value updates, and (iii) tightly controlled human experiments. The aim is to discover which computations best explain everyday behavior online and which manipulations causally shift mood and engagement.

Depending on your preference and skills, we have a series of projects available:

1) Computational reinforcement learning: Help specify and compare RL accounts of user learning and decision-making under feed algorithms. You’ll learn how to derive and fit models (e.g., Rescorla-Wagner, bandit models, etc).
2) Behavioural experiments: Design and run controlled behavioral studies (web-based) to test causal predictions from our models (e.g., how content quality and feedback shape mood and subsequent choices).
3) Real-world data: Work with large-scale, ecologically valid datasets to study learning dynamics in the wild, coming from real platforms such as TikTok or Instagram. For this, we use Data Donation Packages from participants volunteering their social media data. We are also collaborating with the platform OneSec to develop real-world interventions to measure mood during real-time of use.

Our projects are valid for both RP1 and RP2 internships.

Profile: We are looking for several MSc students with an interest in learning and mood dynamics in social media feeds. For all projects, you should be willing to develop solid computational skills (data handling, basic statistics, model fitting, computational modelling). Prior experience with Python or other coding frameworks  will be very useful. For our multiple positions, we will consider the specific match between student and project as well as a first-come, first-served basis.

–          Starting date: February 2026 (exact starting date could be discussed)
–          Location: Roeterseiland
–          You will be working with Violeta Céspedes (PhD student), Ana Pinho (Postdoc), Wouter van den Bos (PI).

If you are interested in any of these projects, please apply with your CV and a brief statement of motivation to v.cespedesizquierdo@uva.nl & W.vandenBos@uva.nl. In this statement of motivation, please specify whether you have a preference for doing either an RP1 or RP2 with us.

For a good example on the type of work we do, please read:

da Silva Pinho, A., Céspedes Izquierdo, V., Lindström, B., & van den Bos, W. (2024). Youths’ sensitivity to social media feedback: A computational account. Science Advances, 10(43), eadp8775.

We look forward to hearing from you.