Departmental Studentships: Potential Projects and Supervisors

In this PhD project you will examine how differences in social structure, such as differences in the number of speakers of a language, can lead to cross-linguistic differences. Prior work in the lab has shown that language spoken by more people have a more systematic grammar, are more sound symbolic (that is, have words that sound like what they mean), and are more likely to have certain word orders.

You are welcome to suggest any linguistic feature that varies cross-linguistically and any social factor that you are interested in exploring. One suggestion is to look at how languages balance speakers’ and listeners’ conflicting needs. Both speakers and listeners want to expend as little effort as possible, but the less effort the speaker expends, the more effort the listener needs to expend to fill in the gaps. One potential hypothesis is that languages that are spoken by more people are more listener-oriented (rather than speaker-oriented), because speakers and listeners are more different from each other in large communities, and therefore, listeners will struggle to fill in the information on their own. Suggestions for other projects that explore how differences in social structure relate to linguistic differences are welcome.

All projects should include an experimental aspect. It is possible for some studies to focus on analyses of existing languages or computational simulations.

Every choice we make in life is accompanied by a sense of confidence – a subjective feeling about how likely it is that we are choosing the best possible course of action. Importantly, confidence is not only used to critically re-evaluate past decisions but serves a pivotal role in guiding future behaviour in many situations. For example, our sense of confidence can tell us if we need to acquire more information before committing to a decision, or whether we should blame ourselves or others for a negative outcome. In this and many other situations, an accurate, well-calibrated sense of confidence is critical for adaptive behaviour, and dysfunction of confidence (such as systematic biases of over/under-confidence) has been hypothesized to lie at the root of many psychiatric symptoms, from depression to schizophrenia. This project will use experiments, neuroimaging and formal models to investigate the computational and brain mechanisms of confidence biases. Whereas traditionally confidence has been investigated via self-report, here, crucially, we will leverage recent advances in model-based analyses of behaviour to infer biases in a principled manner by modelling how confidence is used to guide sequences of decisions. There is room for flexibility and the successful applicant will be able to shape the project to their interests.

The past few years have seen unbelievable advances in natural language applications such as Alexa and Siri. These applications are driven by powerful deep language models trained on vast text corpora to predict the next word in a sentence or passage.  The fact that these machines so closely mimic human language behaviour (at least on a superficial level) is intriguing.  However, thus far little research has asked whether and how the processes in these models map on to fine-grained human language behaviour.  This project will attempt to relate metrics derived from predictive natural language models to eye-movement behaviour as adults read.  Our goal is to understand how certainty within the models relate to behaviours such as fixation location, duration, and regressions in different types of text. Understanding this mapping will open vast possibilities for research and application.  This project requires strong quantitative, coding, and visualisation skills.

Bilingual children and adults are better at learning new words compared to monolingual children and adults. The precise cognitive mechanisms that support this bilingual advantage are not clear. The advantage could be due to pre-existing memory schemas: new information that is compatible with an existing memory schema can lead to accelerated memory consolidation. The aim of this project is to find out what role sleep-associated memory consolidation plays in this advantage. The project will use word learning and sleep paradigms in adults and children to discover whether bilinguals show accelerated consolidation of new words. Dr Tamminen will contribute expertise on sleep and memory research, and Professor Ricketts will contribute to the developmental aspects of the work.

The previous experimental literature robustly documented the tendency for individuals to be more cooperative with ingroup members than with outgroup members (i.e., ingroup favouritism). In addition, it has shown that people are not inclined to be more aggressive towards outgroup members than ingroup members. Existing social and evolutionary psychological work has delved into both proximate and ultimate explanations of those tendencies. However, past intergroup relations have been diverse and there have been many instances of peaceful and cooperative intergroup interactions as well as hostile intergroup conflicts. Thus, going beyond addressing why people display increased ingroup cooperation but not increased outgroup aggression, we are interested in understanding how psychological (e.g., reputation) and ecological (e.g., natural disasters) factors shape intergroup relations in general. More specifically, we would like to address questions, for instance, including a) under what circumstances and why individuals initiate/welcome and maintain cooperative/aggressive intergroup relations b) how different social structures (i.e., hierarchies and leadership) shape intergroup relations. Hiro is open to discussing ideas for student-led PhD projects. Please visit his website to find out more about his research and expertise.

Humans have the remarkable ability to learn concepts from a few experiences, an ability attributable to the hippocampus. However, inferior/anterior temporal cortex has also been implicated in representing long-term conceptual knowledge. How are concepts learnt and transformed over time into long-term knowledge in the brain? One possibility is that early learning is supported by the hippocampus, and knowledge is consolidated into cortex later on (standard systems consolidation theory). Alternatively, hippocampus and cortex may play different roles depending on conceptual content (multiple/trace transformation theory). 

Prior research focused on either early stages or semantic concepts after consolidation, and typically focussed on a particular domain (e.g., objects). In this project, you will investigate the cognitive, computational, and neural mechanisms of how we learn different kinds of concepts (objects, events, actions) through behavioural experiments and computational modelling (cognitive models, neural networks), and link these processes to interaction between brain regions through model-based fMRI and/or EEG. You will learn cutting-edge multivariate neuroimaging and modelling using artificial-intelligence tools (PyTorch). 

There is flexibility in the focus on modelling and neuroimaging (fMRI/EEG) and it will be possible to test how abilities change with age and neurodegeneration (brain structure changes), if of interest. 

Prosociality is a central, human quality allowing large-scale cooperation in modern societies. Previous theoretical work has suggested that prosociality requires at least two cognitive processes: 1) mentalizing (i.e., the ability to understand other’s intentions); 2) and empathy (particularly, empathic concern to motivate acts of prosociality).

However, recent work has cast doubt on the requirement of these cognitive processes for prosocial behaviours. This project aims to tackle the contributing role of mentalizing abilities in prosociality from a decision-theoretic perspective with focus on social learning.

The proposed project will develop interactive paradigms to measure prosocial behaviours by manipulating mentalizing abilities in which different levels of mentalization lead to different beliefs and, consequently, behaviours. It will be investigated how mentalizing abilities shape people’s decision-making processes underlying prosocial behaviours and help inferences on others’ prosocial behaviours and character traits. These abilities will be linked to different psychological traits (e.g., personality) and/or states (e.g., loneliness), and the interplay between mentalizing abilities and empathy will be further explored.

The project will require to combine advanced tools of experimental designing, like online and in-lab multiplayer games as well as interactions with model-based agents, with state-of-the-art computational modelling and neuroimaging techniques (i.e., functional magnetic resonance imaging) to dive into the different cognitive dimensions underlying prosocial acts.

It is well established that verbal serial short-term memory is highly vulnerable to disruption by the mere presence of irrelevant sound. According to Load Theory (Lavie, 2005), such distraction should be exacerbated under high cognitive load because such load is assumed to use up capacity that is otherwise available to protect short-term memory from task-irrelevant stimuli. However, there is evidence that at least certain kinds of load actually reduce or eliminate auditory distraction in this setting. This has been explained in terms of an alternative task-engagement account in which high load triggers a voluntary upward shift in task-engagement level, which acts to shield performance from the distracting effects of irrelevant material (Hughes, 2014). Proponents of Load Theory might counter, however, that past manipulations of load in the irrelevant sound/short-term memory setting have been ones of perceptual rather than cognitive load, which would indeed be expected to reduce, not increase, distraction according to Load Theory. The present PhD project would focus on whether increases in cognitive, as well as perceptual, load do indeed reduce distraction in this domain, thereby providing support for the task-engagement account and challenging Load Theory.  

Making good choices requires learning how our actions relate to specific events in the external world. Yet, this learning process can be biased by how we make, and evaluate, our decisions. While such biases can be generally adaptive, understanding them could help improve decision making. Moreover, such interactions can give rise to, and maintain, maladaptive patterns of (meta)cognition linked to psychiatric symptoms, like depression. Given the high economic and societal costs of mental illness, better treatment and prevention strategies are needed. Leveraging an existing dataset and combining neuroimaging (fMRI), computational and behavioural methods will improve our understanding of how learning and decision making interact with self-monitoring and attribution, in mental health and illness. Focusing on symptoms, over diagnostic categories, will serve to identify specific patterns of maladaptive cognition. This could inform the development of novel tools to refine differential diagnosis and improve treatment selection, as well as provide a foundation for the development of novel psychological interventions. 

See the lab website for more information on our work in this area. 

Many older adults experience the detrimental effects of age-related hearing loss (ARHL) on a daily basis, be it in a conversation with a friend or watching a movie. Beyond the immediate auditory challenges, ARHL can also lead to social isolation and cognitive decline. Critically, ARHL has been identified as a leading - and possibly modifiable - risk factor for dementia. Yet, the precise relationship between ARHL and cognitive decline is unknown.

The aim of the project is to better understand how ARHL relates to cognitive decline in both healthy ageing and in individuals with Mild Cognitive Impairment who are at high risk of developing dementia. This project will employ a combination of behavioural online testing and electrophysiology (EEG) to provide a comprehensive cognitive and neural profile of participants with varying degrees of hearing loss and cognitive abilities. The project will include a series of auditory perception and cognitive tasks as well as a attention and memory tasks to examine how cognitive performance relates to severities of hearing loss on a behavioural and neural level.  

By better understanding the connection between ARHL and cognitive decline, we can potentially develop strategies for early detection of cognitive decline in individuals with ARHL.