Longer description of your proposed project

Social scientists have become increasingly concerned about increases in issue polarisation, i.e. a growing divergence in issue positions. Though issue polarisation has traditionally been seen as occurring between right and leftwing issues, findings have shown that polarisation may occur across both a social and economic dimension - and that citizens are more divided over socio-cultural issues than economic ones. While ‘motivated reasoning’ and ‘rational updating’ accounts have been suggested, based on findings that other forms of polarisation can be modelled among biassed Bayesians without requiring motivated reasoning, I aim to take a Bayesian approach. I propose that there may be differences in the belief updating processes between social and economic issues. I suggest that using Bayesian networks, with additional variables like trust and expertise, may account for divergences in belief revision: for instance, as suggested by findings on ‘morality policies’, individuals could give lower weight to expertise when updating on social issues, because they view their moral values as sufficient evidence. My proposed research will have three stages: first, I will conduct a literature review, to provide more examples and nuance for what constitutes a social, compared to an economic, belief. Following this, I will develop a model, where variables can explain how belief revision diverges for different issues. In the third stage, I will test the proposed model experimentally: I will elicit parameters from participants to determine whether there are belief updating differences. Then, if distributions differ, I will vary components of the Bayesian network, and compare the observed data to the model’s predictions. If I have enough time, I hope to integrate these findings into my existing Agent-Based Model, which I am currently developing for my dissertation.

Based on my findings, I hope to be able to propose interventions for minimising issue polarisation. For instance, identifying that expertise is discounted in belief updating on social issues due to perceptions of simplicity could motivate interventions that make the complexity of policy issues more salient. This could lead to increased information-seeking from experts and, in turn, potentially reduce issue polarisation.

My project will also help us better understand how certain policies become more salient than others, what makes people more averse to seeking out evidence/ updating their beliefs/ listening to expertise. As I aim to work in government/ a think tank following my MPhil/PhD, these findings will be useful for learning how to frame policies on important causes that matter to me (eg increasing pandemic preparedness, decreasing AI risks). For instance, by understanding how people update based on information, I could learn how best to frame AI policy - i.e. as a social or economic issue - to increase the receptiveness to certain AI policies being passed in government.

Describe why you think you're qualified to work on this

I am qualified to undertake this project because of a long-standing interest in political polarisation and psychology, and recent skill development in computational psychology. Aged 17, I carried out a dissertation length study (‘extended project qualification’) to evaluate the efficacy of accuracy nudges in pre-empting the spread of fake news, and received an A*. At LSE, for the last 2 years, I have worked as a Research Assistant to Jens Koed Madsen and to George Melios. Both of these projects are on political polarisation, though the former interrogates how algorithms interact with open-minded users to push polarisation, while the latter focuses on meta-analyses for interventions to reduce political polarisation. Additionally, for my dissertation, I am developing an agent-based model to explore the interactions between user differences, such as novelty preferences, and varying recommendation algorithms, to determine how issue polarisation can emerge.

I am also qualified to undertake this project because I have sourced the supervisors best placed to assist me with my work. My prospective supervisor at Cambridge, Lee De Wit, has recently been given a grant by the Templeton World Charity, along with my current dissertation supervisor Jens Koed Madsen, to study whether computational models can be used to understand when and why societies polarise on issues like climate change or social questions. Similarly, my prospective supervisor at UCL, David Lagnado, has significant experience in Bayesian networks, and building computational models.

Finally, though I am aware that MPhil and MPhil/PhDs are difficult, and have high drop-out rates, I am highly confident that this will be a good path for me. Not only have I had an interest in studying psychology at university since the age of 13, but I have enjoyed the independent work at university (i.e. dissertation) the most, as I’ve found it intellectually challenging in a way that taught aspects of my degree have not been. I’ve excelled in my degree thus far, having achieved a First in every module (including those not in my department, like Philosophy and Political Science), but have also enjoyed being proactive, rather than merely passive in my studies: I’ve been President of the Psychological and Behavioural Science society for two years, where I have coordinated large-scale projects with other universities and employers including a Careers Fair and a behavioural science hackathon (Nudgeathon), as well as founding a Psychology journal in collaboration with the department, and starting a behavioural science consulting division, which is currently advising One for the World on increasing uptake of pledges among students.

Other ways I can learn about you

https://uk.linkedin.com/in/sonja-belkin-3562651b9

https://twitter.com/SonjaBelkin

How much money do you need?

I’m currently deciding between both offers: if I choose the MPhil programme, £28,483 would cover the maintenance (18,625) and tuition (9,858), as per website estimates Finance | Postgraduate Study (cam.ac.uk). While being covered for the full expenses would be the best case, I consider being covered for tuition a very good case. If I choose the UCL MPhil/PhD programme, £28,483 would cover 4 years of tuition (£24,140 - one year is £6,035) and leave me £4343 for living costs, which would be minimal, as I would be living at home and would not need to pay rent. I would likely use it for commuting, and attending conferences/summer schools/ funding an internship during the PhD (which my supervisor is happy to be the case). In the good case scenario, I would be able to fund 1.5 years worth of the MPhil/PhD: I would try to make up the remainder of the money through doing the programme part-time, with a job, or through demonstratorships/teaching/odd job work in the university.

Links to any supporting documents or information

No response.

Estimate your probability of succeeding if you get the amount of money you asked for

75%. I’m fairly confident this is the right path for me, and that this is the project I’d be keen to work on, as I’ve been interested in political psychology & online interactions for the last six years, but know that postgraduate work can be difficult, and the findings may not align with what I would hope to achieve